diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,26785 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8025343189017952,
+  "eval_steps": 200,
+  "global_step": 3800,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00021119324181626187,
+      "grad_norm": 61.50020217895508,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 4.488,
+      "step": 1
+    },
+    {
+      "epoch": 0.00042238648363252375,
+      "grad_norm": 61.16289520263672,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 4.4894,
+      "step": 2
+    },
+    {
+      "epoch": 0.0006335797254487857,
+      "grad_norm": 60.783447265625,
+      "learning_rate": 3e-06,
+      "loss": 4.4733,
+      "step": 3
+    },
+    {
+      "epoch": 0.0008447729672650475,
+      "grad_norm": 58.85840606689453,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 4.3823,
+      "step": 4
+    },
+    {
+      "epoch": 0.0010559662090813093,
+      "grad_norm": 52.98855972290039,
+      "learning_rate": 5e-06,
+      "loss": 4.0793,
+      "step": 5
+    },
+    {
+      "epoch": 0.0012671594508975714,
+      "grad_norm": 40.7520637512207,
+      "learning_rate": 6e-06,
+      "loss": 3.3185,
+      "step": 6
+    },
+    {
+      "epoch": 0.0014783526927138332,
+      "grad_norm": 38.59539031982422,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 3.0455,
+      "step": 7
+    },
+    {
+      "epoch": 0.001689545934530095,
+      "grad_norm": 31.923809051513672,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.8366,
+      "step": 8
+    },
+    {
+      "epoch": 0.0019007391763463568,
+      "grad_norm": 26.97309112548828,
+      "learning_rate": 9e-06,
+      "loss": 1.4803,
+      "step": 9
+    },
+    {
+      "epoch": 0.0021119324181626186,
+      "grad_norm": 5.7373433113098145,
+      "learning_rate": 1e-05,
+      "loss": 0.8852,
+      "step": 10
+    },
+    {
+      "epoch": 0.0023231256599788807,
+      "grad_norm": 2.6410648822784424,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.8242,
+      "step": 11
+    },
+    {
+      "epoch": 0.0025343189017951427,
+      "grad_norm": 1.0567806959152222,
+      "learning_rate": 1.2e-05,
+      "loss": 0.7871,
+      "step": 12
+    },
+    {
+      "epoch": 0.0027455121436114043,
+      "grad_norm": 1.338511347770691,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.7721,
+      "step": 13
+    },
+    {
+      "epoch": 0.0029567053854276664,
+      "grad_norm": 2.0022473335266113,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.7758,
+      "step": 14
+    },
+    {
+      "epoch": 0.0031678986272439284,
+      "grad_norm": 2.5779531002044678,
+      "learning_rate": 1.5e-05,
+      "loss": 0.7641,
+      "step": 15
+    },
+    {
+      "epoch": 0.00337909186906019,
+      "grad_norm": 1.3967171907424927,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.76,
+      "step": 16
+    },
+    {
+      "epoch": 0.003590285110876452,
+      "grad_norm": 1.0678988695144653,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 0.7526,
+      "step": 17
+    },
+    {
+      "epoch": 0.0038014783526927136,
+      "grad_norm": 0.6299343109130859,
+      "learning_rate": 1.8e-05,
+      "loss": 0.7405,
+      "step": 18
+    },
+    {
+      "epoch": 0.004012671594508976,
+      "grad_norm": 0.7110719680786133,
+      "learning_rate": 1.9e-05,
+      "loss": 0.7327,
+      "step": 19
+    },
+    {
+      "epoch": 0.004223864836325237,
+      "grad_norm": 0.5829989910125732,
+      "learning_rate": 2e-05,
+      "loss": 0.7218,
+      "step": 20
+    },
+    {
+      "epoch": 0.0044350580781415,
+      "grad_norm": 1.2814234495162964,
+      "learning_rate": 2.1e-05,
+      "loss": 0.7211,
+      "step": 21
+    },
+    {
+      "epoch": 0.004646251319957761,
+      "grad_norm": 0.7550607323646545,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.7115,
+      "step": 22
+    },
+    {
+      "epoch": 0.004857444561774023,
+      "grad_norm": 2.179175853729248,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.7234,
+      "step": 23
+    },
+    {
+      "epoch": 0.005068637803590285,
+      "grad_norm": 0.7029999494552612,
+      "learning_rate": 2.4e-05,
+      "loss": 0.7006,
+      "step": 24
+    },
+    {
+      "epoch": 0.005279831045406547,
+      "grad_norm": 0.9172762036323547,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6996,
+      "step": 25
+    },
+    {
+      "epoch": 0.005491024287222809,
+      "grad_norm": 0.6699315309524536,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.6929,
+      "step": 26
+    },
+    {
+      "epoch": 0.005702217529039071,
+      "grad_norm": 2.9052278995513916,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.7199,
+      "step": 27
+    },
+    {
+      "epoch": 0.005913410770855333,
+      "grad_norm": 1.3325793743133545,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.6954,
+      "step": 28
+    },
+    {
+      "epoch": 0.006124604012671594,
+      "grad_norm": 1.1330795288085938,
+      "learning_rate": 2.9e-05,
+      "loss": 0.6871,
+      "step": 29
+    },
+    {
+      "epoch": 0.006335797254487857,
+      "grad_norm": 2.175365447998047,
+      "learning_rate": 3e-05,
+      "loss": 0.7211,
+      "step": 30
+    },
+    {
+      "epoch": 0.006546990496304118,
+      "grad_norm": 0.8306956887245178,
+      "learning_rate": 3.1e-05,
+      "loss": 0.6909,
+      "step": 31
+    },
+    {
+      "epoch": 0.00675818373812038,
+      "grad_norm": 1.1500612497329712,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.7035,
+      "step": 32
+    },
+    {
+      "epoch": 0.006969376979936642,
+      "grad_norm": 0.7306738495826721,
+      "learning_rate": 3.3e-05,
+      "loss": 0.6999,
+      "step": 33
+    },
+    {
+      "epoch": 0.007180570221752904,
+      "grad_norm": 0.8608762621879578,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.6909,
+      "step": 34
+    },
+    {
+      "epoch": 0.007391763463569166,
+      "grad_norm": 1.1340913772583008,
+      "learning_rate": 3.5e-05,
+      "loss": 0.6911,
+      "step": 35
+    },
+    {
+      "epoch": 0.007602956705385427,
+      "grad_norm": 0.8462384343147278,
+      "learning_rate": 3.6e-05,
+      "loss": 0.6772,
+      "step": 36
+    },
+    {
+      "epoch": 0.00781414994720169,
+      "grad_norm": 0.6609564423561096,
+      "learning_rate": 3.7e-05,
+      "loss": 0.6719,
+      "step": 37
+    },
+    {
+      "epoch": 0.008025343189017951,
+      "grad_norm": 0.7771788239479065,
+      "learning_rate": 3.8e-05,
+      "loss": 0.6714,
+      "step": 38
+    },
+    {
+      "epoch": 0.008236536430834213,
+      "grad_norm": 1.105125904083252,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 0.6642,
+      "step": 39
+    },
+    {
+      "epoch": 0.008447729672650475,
+      "grad_norm": 1.272701621055603,
+      "learning_rate": 4e-05,
+      "loss": 0.6648,
+      "step": 40
+    },
+    {
+      "epoch": 0.008658922914466738,
+      "grad_norm": 1.3464202880859375,
+      "learning_rate": 4.1e-05,
+      "loss": 0.6774,
+      "step": 41
+    },
+    {
+      "epoch": 0.008870116156283,
+      "grad_norm": 0.6994144916534424,
+      "learning_rate": 4.2e-05,
+      "loss": 0.6593,
+      "step": 42
+    },
+    {
+      "epoch": 0.009081309398099261,
+      "grad_norm": 1.6229274272918701,
+      "learning_rate": 4.3e-05,
+      "loss": 0.6847,
+      "step": 43
+    },
+    {
+      "epoch": 0.009292502639915523,
+      "grad_norm": 0.8361057043075562,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.6661,
+      "step": 44
+    },
+    {
+      "epoch": 0.009503695881731784,
+      "grad_norm": 0.9926273822784424,
+      "learning_rate": 4.5e-05,
+      "loss": 0.6739,
+      "step": 45
+    },
+    {
+      "epoch": 0.009714889123548046,
+      "grad_norm": 0.5843594074249268,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.6684,
+      "step": 46
+    },
+    {
+      "epoch": 0.009926082365364308,
+      "grad_norm": 0.9820445775985718,
+      "learning_rate": 4.7e-05,
+      "loss": 0.6694,
+      "step": 47
+    },
+    {
+      "epoch": 0.01013727560718057,
+      "grad_norm": 0.8192408680915833,
+      "learning_rate": 4.8e-05,
+      "loss": 0.655,
+      "step": 48
+    },
+    {
+      "epoch": 0.010348468848996832,
+      "grad_norm": 1.1418250799179077,
+      "learning_rate": 4.9e-05,
+      "loss": 0.6624,
+      "step": 49
+    },
+    {
+      "epoch": 0.010559662090813094,
+      "grad_norm": 0.9435445070266724,
+      "learning_rate": 5e-05,
+      "loss": 0.6588,
+      "step": 50
+    },
+    {
+      "epoch": 0.010770855332629356,
+      "grad_norm": 0.7310433983802795,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 0.6506,
+      "step": 51
+    },
+    {
+      "epoch": 0.010982048574445617,
+      "grad_norm": 0.9302086234092712,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.6564,
+      "step": 52
+    },
+    {
+      "epoch": 0.011193241816261879,
+      "grad_norm": 0.6242333054542542,
+      "learning_rate": 5.300000000000001e-05,
+      "loss": 0.6489,
+      "step": 53
+    },
+    {
+      "epoch": 0.011404435058078142,
+      "grad_norm": 0.9954692721366882,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 0.6512,
+      "step": 54
+    },
+    {
+      "epoch": 0.011615628299894404,
+      "grad_norm": 1.1416066884994507,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 0.6447,
+      "step": 55
+    },
+    {
+      "epoch": 0.011826821541710665,
+      "grad_norm": 0.7708131074905396,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 0.6413,
+      "step": 56
+    },
+    {
+      "epoch": 0.012038014783526927,
+      "grad_norm": 0.9874745011329651,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 0.643,
+      "step": 57
+    },
+    {
+      "epoch": 0.012249208025343189,
+      "grad_norm": 1.146091103553772,
+      "learning_rate": 5.8e-05,
+      "loss": 0.6439,
+      "step": 58
+    },
+    {
+      "epoch": 0.01246040126715945,
+      "grad_norm": 0.619839608669281,
+      "learning_rate": 5.9e-05,
+      "loss": 0.6387,
+      "step": 59
+    },
+    {
+      "epoch": 0.012671594508975714,
+      "grad_norm": 1.0728814601898193,
+      "learning_rate": 6e-05,
+      "loss": 0.644,
+      "step": 60
+    },
+    {
+      "epoch": 0.012882787750791975,
+      "grad_norm": 0.8369746804237366,
+      "learning_rate": 6.1e-05,
+      "loss": 0.6325,
+      "step": 61
+    },
+    {
+      "epoch": 0.013093980992608237,
+      "grad_norm": 0.8612938523292542,
+      "learning_rate": 6.2e-05,
+      "loss": 0.6289,
+      "step": 62
+    },
+    {
+      "epoch": 0.013305174234424498,
+      "grad_norm": 0.614213228225708,
+      "learning_rate": 6.3e-05,
+      "loss": 0.6198,
+      "step": 63
+    },
+    {
+      "epoch": 0.01351636747624076,
+      "grad_norm": 0.8491672277450562,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.6235,
+      "step": 64
+    },
+    {
+      "epoch": 0.013727560718057022,
+      "grad_norm": 1.5606390237808228,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 0.6506,
+      "step": 65
+    },
+    {
+      "epoch": 0.013938753959873283,
+      "grad_norm": 0.8846147060394287,
+      "learning_rate": 6.6e-05,
+      "loss": 0.6272,
+      "step": 66
+    },
+    {
+      "epoch": 0.014149947201689547,
+      "grad_norm": 0.6623694896697998,
+      "learning_rate": 6.7e-05,
+      "loss": 0.6175,
+      "step": 67
+    },
+    {
+      "epoch": 0.014361140443505808,
+      "grad_norm": 1.3997465372085571,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 0.6302,
+      "step": 68
+    },
+    {
+      "epoch": 0.01457233368532207,
+      "grad_norm": 1.3659074306488037,
+      "learning_rate": 6.9e-05,
+      "loss": 0.6509,
+      "step": 69
+    },
+    {
+      "epoch": 0.014783526927138331,
+      "grad_norm": 0.6625523567199707,
+      "learning_rate": 7e-05,
+      "loss": 0.6195,
+      "step": 70
+    },
+    {
+      "epoch": 0.014994720168954593,
+      "grad_norm": 0.73038649559021,
+      "learning_rate": 7.1e-05,
+      "loss": 0.6361,
+      "step": 71
+    },
+    {
+      "epoch": 0.015205913410770855,
+      "grad_norm": 0.5066683292388916,
+      "learning_rate": 7.2e-05,
+      "loss": 0.6244,
+      "step": 72
+    },
+    {
+      "epoch": 0.015417106652587118,
+      "grad_norm": 0.8289284706115723,
+      "learning_rate": 7.3e-05,
+      "loss": 0.6319,
+      "step": 73
+    },
+    {
+      "epoch": 0.01562829989440338,
+      "grad_norm": 0.8158730864524841,
+      "learning_rate": 7.4e-05,
+      "loss": 0.6273,
+      "step": 74
+    },
+    {
+      "epoch": 0.01583949313621964,
+      "grad_norm": 1.013464093208313,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.6154,
+      "step": 75
+    },
+    {
+      "epoch": 0.016050686378035903,
+      "grad_norm": 0.596935510635376,
+      "learning_rate": 7.6e-05,
+      "loss": 0.62,
+      "step": 76
+    },
+    {
+      "epoch": 0.016261879619852166,
+      "grad_norm": 0.911916971206665,
+      "learning_rate": 7.7e-05,
+      "loss": 0.6177,
+      "step": 77
+    },
+    {
+      "epoch": 0.016473072861668426,
+      "grad_norm": 0.8805202841758728,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 0.6183,
+      "step": 78
+    },
+    {
+      "epoch": 0.01668426610348469,
+      "grad_norm": 0.7835665941238403,
+      "learning_rate": 7.900000000000001e-05,
+      "loss": 0.6169,
+      "step": 79
+    },
+    {
+      "epoch": 0.01689545934530095,
+      "grad_norm": 0.8202574253082275,
+      "learning_rate": 8e-05,
+      "loss": 0.6097,
+      "step": 80
+    },
+    {
+      "epoch": 0.017106652587117212,
+      "grad_norm": 0.8718443512916565,
+      "learning_rate": 8.1e-05,
+      "loss": 0.6177,
+      "step": 81
+    },
+    {
+      "epoch": 0.017317845828933476,
+      "grad_norm": 0.6485271453857422,
+      "learning_rate": 8.2e-05,
+      "loss": 0.6016,
+      "step": 82
+    },
+    {
+      "epoch": 0.017529039070749736,
+      "grad_norm": 0.9156302213668823,
+      "learning_rate": 8.3e-05,
+      "loss": 0.616,
+      "step": 83
+    },
+    {
+      "epoch": 0.017740232312566,
+      "grad_norm": 0.7427339553833008,
+      "learning_rate": 8.4e-05,
+      "loss": 0.6039,
+      "step": 84
+    },
+    {
+      "epoch": 0.01795142555438226,
+      "grad_norm": 0.6160241961479187,
+      "learning_rate": 8.5e-05,
+      "loss": 0.6024,
+      "step": 85
+    },
+    {
+      "epoch": 0.018162618796198522,
+      "grad_norm": 0.8504828810691833,
+      "learning_rate": 8.6e-05,
+      "loss": 0.6062,
+      "step": 86
+    },
+    {
+      "epoch": 0.018373812038014782,
+      "grad_norm": 0.791434109210968,
+      "learning_rate": 8.7e-05,
+      "loss": 0.6034,
+      "step": 87
+    },
+    {
+      "epoch": 0.018585005279831045,
+      "grad_norm": 0.8009198307991028,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.6156,
+      "step": 88
+    },
+    {
+      "epoch": 0.01879619852164731,
+      "grad_norm": 0.4931645393371582,
+      "learning_rate": 8.900000000000001e-05,
+      "loss": 0.5941,
+      "step": 89
+    },
+    {
+      "epoch": 0.01900739176346357,
+      "grad_norm": 0.8181516528129578,
+      "learning_rate": 9e-05,
+      "loss": 0.5968,
+      "step": 90
+    },
+    {
+      "epoch": 0.019218585005279832,
+      "grad_norm": 0.9011086821556091,
+      "learning_rate": 9.1e-05,
+      "loss": 0.6096,
+      "step": 91
+    },
+    {
+      "epoch": 0.019429778247096092,
+      "grad_norm": 0.6743105053901672,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 0.5979,
+      "step": 92
+    },
+    {
+      "epoch": 0.019640971488912355,
+      "grad_norm": 0.7464610934257507,
+      "learning_rate": 9.300000000000001e-05,
+      "loss": 0.6081,
+      "step": 93
+    },
+    {
+      "epoch": 0.019852164730728615,
+      "grad_norm": 0.5392186045646667,
+      "learning_rate": 9.4e-05,
+      "loss": 0.5936,
+      "step": 94
+    },
+    {
+      "epoch": 0.02006335797254488,
+      "grad_norm": 0.8590493202209473,
+      "learning_rate": 9.5e-05,
+      "loss": 0.6156,
+      "step": 95
+    },
+    {
+      "epoch": 0.02027455121436114,
+      "grad_norm": 0.6266841292381287,
+      "learning_rate": 9.6e-05,
+      "loss": 0.5991,
+      "step": 96
+    },
+    {
+      "epoch": 0.0204857444561774,
+      "grad_norm": 0.7061191201210022,
+      "learning_rate": 9.7e-05,
+      "loss": 0.6007,
+      "step": 97
+    },
+    {
+      "epoch": 0.020696937697993665,
+      "grad_norm": 0.5466405749320984,
+      "learning_rate": 9.8e-05,
+      "loss": 0.5938,
+      "step": 98
+    },
+    {
+      "epoch": 0.020908130939809925,
+      "grad_norm": 0.596760630607605,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 0.5878,
+      "step": 99
+    },
+    {
+      "epoch": 0.021119324181626188,
+      "grad_norm": 0.7114918231964111,
+      "learning_rate": 0.0001,
+      "loss": 0.6001,
+      "step": 100
+    },
+    {
+      "epoch": 0.02133051742344245,
+      "grad_norm": 0.6783373355865479,
+      "learning_rate": 9.99999885147551e-05,
+      "loss": 0.5836,
+      "step": 101
+    },
+    {
+      "epoch": 0.02154171066525871,
+      "grad_norm": 0.683100163936615,
+      "learning_rate": 9.999995405902565e-05,
+      "loss": 0.5941,
+      "step": 102
+    },
+    {
+      "epoch": 0.021752903907074975,
+      "grad_norm": 0.5566619634628296,
+      "learning_rate": 9.999989663282748e-05,
+      "loss": 0.5934,
+      "step": 103
+    },
+    {
+      "epoch": 0.021964097148891235,
+      "grad_norm": 0.5563331246376038,
+      "learning_rate": 9.999981623618697e-05,
+      "loss": 0.5822,
+      "step": 104
+    },
+    {
+      "epoch": 0.022175290390707498,
+      "grad_norm": 0.9614548087120056,
+      "learning_rate": 9.999971286914107e-05,
+      "loss": 0.5825,
+      "step": 105
+    },
+    {
+      "epoch": 0.022386483632523758,
+      "grad_norm": 0.6896506547927856,
+      "learning_rate": 9.999958653173727e-05,
+      "loss": 0.579,
+      "step": 106
+    },
+    {
+      "epoch": 0.02259767687434002,
+      "grad_norm": 0.6285958290100098,
+      "learning_rate": 9.999943722403359e-05,
+      "loss": 0.5796,
+      "step": 107
+    },
+    {
+      "epoch": 0.022808870116156284,
+      "grad_norm": 0.5965277552604675,
+      "learning_rate": 9.999926494609865e-05,
+      "loss": 0.5723,
+      "step": 108
+    },
+    {
+      "epoch": 0.023020063357972544,
+      "grad_norm": 0.7423815131187439,
+      "learning_rate": 9.999906969801157e-05,
+      "loss": 0.5767,
+      "step": 109
+    },
+    {
+      "epoch": 0.023231256599788808,
+      "grad_norm": 0.6487621665000916,
+      "learning_rate": 9.999885147986206e-05,
+      "loss": 0.5768,
+      "step": 110
+    },
+    {
+      "epoch": 0.023442449841605068,
+      "grad_norm": 1.091518759727478,
+      "learning_rate": 9.999861029175038e-05,
+      "loss": 0.5783,
+      "step": 111
+    },
+    {
+      "epoch": 0.02365364308342133,
+      "grad_norm": 0.7326909303665161,
+      "learning_rate": 9.999834613378731e-05,
+      "loss": 0.5809,
+      "step": 112
+    },
+    {
+      "epoch": 0.02386483632523759,
+      "grad_norm": 0.5578820109367371,
+      "learning_rate": 9.999805900609423e-05,
+      "loss": 0.5734,
+      "step": 113
+    },
+    {
+      "epoch": 0.024076029567053854,
+      "grad_norm": 0.6447977423667908,
+      "learning_rate": 9.999774890880304e-05,
+      "loss": 0.5741,
+      "step": 114
+    },
+    {
+      "epoch": 0.024287222808870117,
+      "grad_norm": 0.47397392988204956,
+      "learning_rate": 9.999741584205621e-05,
+      "loss": 0.5602,
+      "step": 115
+    },
+    {
+      "epoch": 0.024498416050686377,
+      "grad_norm": 0.7054709792137146,
+      "learning_rate": 9.999705980600674e-05,
+      "loss": 0.5723,
+      "step": 116
+    },
+    {
+      "epoch": 0.02470960929250264,
+      "grad_norm": 0.7567771077156067,
+      "learning_rate": 9.999668080081821e-05,
+      "loss": 0.5709,
+      "step": 117
+    },
+    {
+      "epoch": 0.0249208025343189,
+      "grad_norm": 0.5105625987052917,
+      "learning_rate": 9.999627882666473e-05,
+      "loss": 0.5636,
+      "step": 118
+    },
+    {
+      "epoch": 0.025131995776135164,
+      "grad_norm": 0.5971348285675049,
+      "learning_rate": 9.999585388373097e-05,
+      "loss": 0.5731,
+      "step": 119
+    },
+    {
+      "epoch": 0.025343189017951427,
+      "grad_norm": 0.5111599564552307,
+      "learning_rate": 9.999540597221217e-05,
+      "loss": 0.5667,
+      "step": 120
+    },
+    {
+      "epoch": 0.025554382259767687,
+      "grad_norm": 0.6556239724159241,
+      "learning_rate": 9.999493509231408e-05,
+      "loss": 0.565,
+      "step": 121
+    },
+    {
+      "epoch": 0.02576557550158395,
+      "grad_norm": 0.45394471287727356,
+      "learning_rate": 9.999444124425303e-05,
+      "loss": 0.56,
+      "step": 122
+    },
+    {
+      "epoch": 0.02597676874340021,
+      "grad_norm": 0.6118949055671692,
+      "learning_rate": 9.999392442825593e-05,
+      "loss": 0.5573,
+      "step": 123
+    },
+    {
+      "epoch": 0.026187961985216474,
+      "grad_norm": 0.5693320035934448,
+      "learning_rate": 9.999338464456016e-05,
+      "loss": 0.5621,
+      "step": 124
+    },
+    {
+      "epoch": 0.026399155227032733,
+      "grad_norm": 0.511299192905426,
+      "learning_rate": 9.999282189341375e-05,
+      "loss": 0.556,
+      "step": 125
+    },
+    {
+      "epoch": 0.026610348468848997,
+      "grad_norm": 0.5127890110015869,
+      "learning_rate": 9.99922361750752e-05,
+      "loss": 0.5528,
+      "step": 126
+    },
+    {
+      "epoch": 0.02682154171066526,
+      "grad_norm": 0.3647984266281128,
+      "learning_rate": 9.999162748981361e-05,
+      "loss": 0.5391,
+      "step": 127
+    },
+    {
+      "epoch": 0.02703273495248152,
+      "grad_norm": 0.5274839401245117,
+      "learning_rate": 9.999099583790863e-05,
+      "loss": 0.5476,
+      "step": 128
+    },
+    {
+      "epoch": 0.027243928194297783,
+      "grad_norm": 0.5898672342300415,
+      "learning_rate": 9.99903412196504e-05,
+      "loss": 0.5427,
+      "step": 129
+    },
+    {
+      "epoch": 0.027455121436114043,
+      "grad_norm": 0.4763058125972748,
+      "learning_rate": 9.998966363533971e-05,
+      "loss": 0.5446,
+      "step": 130
+    },
+    {
+      "epoch": 0.027666314677930307,
+      "grad_norm": 0.5160780549049377,
+      "learning_rate": 9.998896308528782e-05,
+      "loss": 0.5419,
+      "step": 131
+    },
+    {
+      "epoch": 0.027877507919746566,
+      "grad_norm": 0.6170157790184021,
+      "learning_rate": 9.998823956981657e-05,
+      "loss": 0.5497,
+      "step": 132
+    },
+    {
+      "epoch": 0.02808870116156283,
+      "grad_norm": 0.6491498351097107,
+      "learning_rate": 9.998749308925839e-05,
+      "loss": 0.5524,
+      "step": 133
+    },
+    {
+      "epoch": 0.028299894403379093,
+      "grad_norm": 0.695838451385498,
+      "learning_rate": 9.998672364395616e-05,
+      "loss": 0.5564,
+      "step": 134
+    },
+    {
+      "epoch": 0.028511087645195353,
+      "grad_norm": 0.4686194956302643,
+      "learning_rate": 9.998593123426339e-05,
+      "loss": 0.549,
+      "step": 135
+    },
+    {
+      "epoch": 0.028722280887011616,
+      "grad_norm": 0.46365734934806824,
+      "learning_rate": 9.998511586054414e-05,
+      "loss": 0.5373,
+      "step": 136
+    },
+    {
+      "epoch": 0.028933474128827876,
+      "grad_norm": 0.4870682954788208,
+      "learning_rate": 9.998427752317299e-05,
+      "loss": 0.5327,
+      "step": 137
+    },
+    {
+      "epoch": 0.02914466737064414,
+      "grad_norm": 0.45998021960258484,
+      "learning_rate": 9.998341622253508e-05,
+      "loss": 0.5341,
+      "step": 138
+    },
+    {
+      "epoch": 0.029355860612460403,
+      "grad_norm": 0.4955160319805145,
+      "learning_rate": 9.998253195902611e-05,
+      "loss": 0.5407,
+      "step": 139
+    },
+    {
+      "epoch": 0.029567053854276663,
+      "grad_norm": 0.7111489176750183,
+      "learning_rate": 9.99816247330523e-05,
+      "loss": 0.5426,
+      "step": 140
+    },
+    {
+      "epoch": 0.029778247096092926,
+      "grad_norm": 0.5503473281860352,
+      "learning_rate": 9.998069454503044e-05,
+      "loss": 0.5319,
+      "step": 141
+    },
+    {
+      "epoch": 0.029989440337909186,
+      "grad_norm": 0.5920684337615967,
+      "learning_rate": 9.997974139538788e-05,
+      "loss": 0.5317,
+      "step": 142
+    },
+    {
+      "epoch": 0.03020063357972545,
+      "grad_norm": 0.5110545754432678,
+      "learning_rate": 9.99787652845625e-05,
+      "loss": 0.5279,
+      "step": 143
+    },
+    {
+      "epoch": 0.03041182682154171,
+      "grad_norm": 0.749497652053833,
+      "learning_rate": 9.997776621300275e-05,
+      "loss": 0.533,
+      "step": 144
+    },
+    {
+      "epoch": 0.030623020063357972,
+      "grad_norm": 0.5508199334144592,
+      "learning_rate": 9.997674418116758e-05,
+      "loss": 0.5283,
+      "step": 145
+    },
+    {
+      "epoch": 0.030834213305174236,
+      "grad_norm": 0.40619024634361267,
+      "learning_rate": 9.997569918952655e-05,
+      "loss": 0.5205,
+      "step": 146
+    },
+    {
+      "epoch": 0.031045406546990496,
+      "grad_norm": 0.6415976881980896,
+      "learning_rate": 9.997463123855973e-05,
+      "loss": 0.5215,
+      "step": 147
+    },
+    {
+      "epoch": 0.03125659978880676,
+      "grad_norm": 0.715039849281311,
+      "learning_rate": 9.997354032875776e-05,
+      "loss": 0.5214,
+      "step": 148
+    },
+    {
+      "epoch": 0.03146779303062302,
+      "grad_norm": 0.6289534568786621,
+      "learning_rate": 9.997242646062178e-05,
+      "loss": 0.5289,
+      "step": 149
+    },
+    {
+      "epoch": 0.03167898627243928,
+      "grad_norm": 0.5409113168716431,
+      "learning_rate": 9.997128963466355e-05,
+      "loss": 0.5314,
+      "step": 150
+    },
+    {
+      "epoch": 0.031890179514255546,
+      "grad_norm": 0.6138347387313843,
+      "learning_rate": 9.997012985140531e-05,
+      "loss": 0.5286,
+      "step": 151
+    },
+    {
+      "epoch": 0.032101372756071805,
+      "grad_norm": 0.5598776936531067,
+      "learning_rate": 9.99689471113799e-05,
+      "loss": 0.5162,
+      "step": 152
+    },
+    {
+      "epoch": 0.032312565997888065,
+      "grad_norm": 0.609798789024353,
+      "learning_rate": 9.996774141513067e-05,
+      "loss": 0.5202,
+      "step": 153
+    },
+    {
+      "epoch": 0.03252375923970433,
+      "grad_norm": 0.6075906753540039,
+      "learning_rate": 9.996651276321153e-05,
+      "loss": 0.5151,
+      "step": 154
+    },
+    {
+      "epoch": 0.03273495248152059,
+      "grad_norm": 0.4902147054672241,
+      "learning_rate": 9.996526115618692e-05,
+      "loss": 0.5119,
+      "step": 155
+    },
+    {
+      "epoch": 0.03294614572333685,
+      "grad_norm": 0.5438671112060547,
+      "learning_rate": 9.996398659463187e-05,
+      "loss": 0.5122,
+      "step": 156
+    },
+    {
+      "epoch": 0.03315733896515311,
+      "grad_norm": 0.5398592948913574,
+      "learning_rate": 9.99626890791319e-05,
+      "loss": 0.5067,
+      "step": 157
+    },
+    {
+      "epoch": 0.03336853220696938,
+      "grad_norm": 0.5693660378456116,
+      "learning_rate": 9.996136861028314e-05,
+      "loss": 0.5094,
+      "step": 158
+    },
+    {
+      "epoch": 0.03357972544878564,
+      "grad_norm": 0.5610343217849731,
+      "learning_rate": 9.996002518869218e-05,
+      "loss": 0.5172,
+      "step": 159
+    },
+    {
+      "epoch": 0.0337909186906019,
+      "grad_norm": 0.6559011340141296,
+      "learning_rate": 9.995865881497621e-05,
+      "loss": 0.5199,
+      "step": 160
+    },
+    {
+      "epoch": 0.034002111932418165,
+      "grad_norm": 0.700296938419342,
+      "learning_rate": 9.995726948976299e-05,
+      "loss": 0.5256,
+      "step": 161
+    },
+    {
+      "epoch": 0.034213305174234425,
+      "grad_norm": 0.4303273856639862,
+      "learning_rate": 9.995585721369074e-05,
+      "loss": 0.5101,
+      "step": 162
+    },
+    {
+      "epoch": 0.034424498416050685,
+      "grad_norm": 0.6189026832580566,
+      "learning_rate": 9.995442198740833e-05,
+      "loss": 0.5176,
+      "step": 163
+    },
+    {
+      "epoch": 0.03463569165786695,
+      "grad_norm": 0.8110437989234924,
+      "learning_rate": 9.995296381157505e-05,
+      "loss": 0.5126,
+      "step": 164
+    },
+    {
+      "epoch": 0.03484688489968321,
+      "grad_norm": 0.4078983962535858,
+      "learning_rate": 9.995148268686085e-05,
+      "loss": 0.5067,
+      "step": 165
+    },
+    {
+      "epoch": 0.03505807814149947,
+      "grad_norm": 0.49522921442985535,
+      "learning_rate": 9.994997861394616e-05,
+      "loss": 0.5091,
+      "step": 166
+    },
+    {
+      "epoch": 0.03526927138331573,
+      "grad_norm": 0.6856568455696106,
+      "learning_rate": 9.994845159352198e-05,
+      "loss": 0.5157,
+      "step": 167
+    },
+    {
+      "epoch": 0.035480464625132,
+      "grad_norm": 0.492906779050827,
+      "learning_rate": 9.994690162628979e-05,
+      "loss": 0.5063,
+      "step": 168
+    },
+    {
+      "epoch": 0.03569165786694826,
+      "grad_norm": 0.7875059247016907,
+      "learning_rate": 9.994532871296172e-05,
+      "loss": 0.5063,
+      "step": 169
+    },
+    {
+      "epoch": 0.03590285110876452,
+      "grad_norm": 0.5107460021972656,
+      "learning_rate": 9.994373285426034e-05,
+      "loss": 0.5055,
+      "step": 170
+    },
+    {
+      "epoch": 0.036114044350580785,
+      "grad_norm": 0.6555144190788269,
+      "learning_rate": 9.994211405091883e-05,
+      "loss": 0.5024,
+      "step": 171
+    },
+    {
+      "epoch": 0.036325237592397044,
+      "grad_norm": 0.5787709355354309,
+      "learning_rate": 9.994047230368086e-05,
+      "loss": 0.5066,
+      "step": 172
+    },
+    {
+      "epoch": 0.036536430834213304,
+      "grad_norm": 0.40571317076683044,
+      "learning_rate": 9.993880761330068e-05,
+      "loss": 0.499,
+      "step": 173
+    },
+    {
+      "epoch": 0.036747624076029564,
+      "grad_norm": 0.5026586651802063,
+      "learning_rate": 9.993711998054307e-05,
+      "loss": 0.4989,
+      "step": 174
+    },
+    {
+      "epoch": 0.03695881731784583,
+      "grad_norm": 0.3971845209598541,
+      "learning_rate": 9.993540940618334e-05,
+      "loss": 0.4908,
+      "step": 175
+    },
+    {
+      "epoch": 0.03717001055966209,
+      "grad_norm": 1.0329375267028809,
+      "learning_rate": 9.993367589100733e-05,
+      "loss": 0.4986,
+      "step": 176
+    },
+    {
+      "epoch": 0.03738120380147835,
+      "grad_norm": 0.40550753474235535,
+      "learning_rate": 9.993191943581144e-05,
+      "loss": 0.4973,
+      "step": 177
+    },
+    {
+      "epoch": 0.03759239704329462,
+      "grad_norm": 0.7783941030502319,
+      "learning_rate": 9.993014004140261e-05,
+      "loss": 0.5102,
+      "step": 178
+    },
+    {
+      "epoch": 0.03780359028511088,
+      "grad_norm": 0.5721800327301025,
+      "learning_rate": 9.992833770859833e-05,
+      "loss": 0.5081,
+      "step": 179
+    },
+    {
+      "epoch": 0.03801478352692714,
+      "grad_norm": 0.4971166253089905,
+      "learning_rate": 9.992651243822657e-05,
+      "loss": 0.5053,
+      "step": 180
+    },
+    {
+      "epoch": 0.0382259767687434,
+      "grad_norm": 0.44722849130630493,
+      "learning_rate": 9.99246642311259e-05,
+      "loss": 0.4988,
+      "step": 181
+    },
+    {
+      "epoch": 0.038437170010559664,
+      "grad_norm": 0.558010995388031,
+      "learning_rate": 9.99227930881454e-05,
+      "loss": 0.4964,
+      "step": 182
+    },
+    {
+      "epoch": 0.038648363252375924,
+      "grad_norm": 0.5738832354545593,
+      "learning_rate": 9.99208990101447e-05,
+      "loss": 0.4906,
+      "step": 183
+    },
+    {
+      "epoch": 0.038859556494192184,
+      "grad_norm": 0.7163248062133789,
+      "learning_rate": 9.991898199799395e-05,
+      "loss": 0.5064,
+      "step": 184
+    },
+    {
+      "epoch": 0.03907074973600845,
+      "grad_norm": 0.6203807592391968,
+      "learning_rate": 9.991704205257382e-05,
+      "loss": 0.4985,
+      "step": 185
+    },
+    {
+      "epoch": 0.03928194297782471,
+      "grad_norm": 0.3731309175491333,
+      "learning_rate": 9.991507917477559e-05,
+      "loss": 0.488,
+      "step": 186
+    },
+    {
+      "epoch": 0.03949313621964097,
+      "grad_norm": 0.6444423794746399,
+      "learning_rate": 9.991309336550099e-05,
+      "loss": 0.4954,
+      "step": 187
+    },
+    {
+      "epoch": 0.03970432946145723,
+      "grad_norm": 0.655950665473938,
+      "learning_rate": 9.991108462566233e-05,
+      "loss": 0.4959,
+      "step": 188
+    },
+    {
+      "epoch": 0.0399155227032735,
+      "grad_norm": 0.3854362964630127,
+      "learning_rate": 9.990905295618244e-05,
+      "loss": 0.4833,
+      "step": 189
+    },
+    {
+      "epoch": 0.04012671594508976,
+      "grad_norm": 0.6350207924842834,
+      "learning_rate": 9.990699835799469e-05,
+      "loss": 0.4905,
+      "step": 190
+    },
+    {
+      "epoch": 0.04033790918690602,
+      "grad_norm": 0.5363175868988037,
+      "learning_rate": 9.990492083204299e-05,
+      "loss": 0.4925,
+      "step": 191
+    },
+    {
+      "epoch": 0.04054910242872228,
+      "grad_norm": 0.5717889666557312,
+      "learning_rate": 9.990282037928178e-05,
+      "loss": 0.4934,
+      "step": 192
+    },
+    {
+      "epoch": 0.04076029567053854,
+      "grad_norm": 0.43231910467147827,
+      "learning_rate": 9.990069700067601e-05,
+      "loss": 0.4791,
+      "step": 193
+    },
+    {
+      "epoch": 0.0409714889123548,
+      "grad_norm": 0.6627949476242065,
+      "learning_rate": 9.98985506972012e-05,
+      "loss": 0.4853,
+      "step": 194
+    },
+    {
+      "epoch": 0.04118268215417107,
+      "grad_norm": 0.6908307671546936,
+      "learning_rate": 9.989638146984336e-05,
+      "loss": 0.4896,
+      "step": 195
+    },
+    {
+      "epoch": 0.04139387539598733,
+      "grad_norm": 0.6175587177276611,
+      "learning_rate": 9.989418931959909e-05,
+      "loss": 0.4772,
+      "step": 196
+    },
+    {
+      "epoch": 0.04160506863780359,
+      "grad_norm": 0.49213936924934387,
+      "learning_rate": 9.989197424747544e-05,
+      "loss": 0.4788,
+      "step": 197
+    },
+    {
+      "epoch": 0.04181626187961985,
+      "grad_norm": 0.6289134621620178,
+      "learning_rate": 9.988973625449007e-05,
+      "loss": 0.4848,
+      "step": 198
+    },
+    {
+      "epoch": 0.042027455121436116,
+      "grad_norm": 0.43408048152923584,
+      "learning_rate": 9.988747534167111e-05,
+      "loss": 0.4743,
+      "step": 199
+    },
+    {
+      "epoch": 0.042238648363252376,
+      "grad_norm": 0.69822758436203,
+      "learning_rate": 9.988519151005728e-05,
+      "loss": 0.4831,
+      "step": 200
+    },
+    {
+      "epoch": 0.042238648363252376,
+      "eval_train_split_loss": 0.533808708190918,
+      "eval_train_split_runtime": 6.3908,
+      "eval_train_split_samples_per_second": 1.565,
+      "eval_train_split_steps_per_second": 0.156,
+      "step": 200
+    },
+    {
+      "epoch": 0.042449841605068636,
+      "grad_norm": 0.4250997304916382,
+      "learning_rate": 9.988288476069776e-05,
+      "loss": 0.4749,
+      "step": 201
+    },
+    {
+      "epoch": 0.0426610348468849,
+      "grad_norm": 0.6909369230270386,
+      "learning_rate": 9.988055509465233e-05,
+      "loss": 0.49,
+      "step": 202
+    },
+    {
+      "epoch": 0.04287222808870116,
+      "grad_norm": 0.5814520716667175,
+      "learning_rate": 9.987820251299122e-05,
+      "loss": 0.4772,
+      "step": 203
+    },
+    {
+      "epoch": 0.04308342133051742,
+      "grad_norm": 0.6437883973121643,
+      "learning_rate": 9.987582701679526e-05,
+      "loss": 0.4859,
+      "step": 204
+    },
+    {
+      "epoch": 0.04329461457233368,
+      "grad_norm": 0.4975798428058624,
+      "learning_rate": 9.987342860715575e-05,
+      "loss": 0.4837,
+      "step": 205
+    },
+    {
+      "epoch": 0.04350580781414995,
+      "grad_norm": 0.37458425760269165,
+      "learning_rate": 9.987100728517457e-05,
+      "loss": 0.4759,
+      "step": 206
+    },
+    {
+      "epoch": 0.04371700105596621,
+      "grad_norm": 0.6102822422981262,
+      "learning_rate": 9.986856305196408e-05,
+      "loss": 0.4759,
+      "step": 207
+    },
+    {
+      "epoch": 0.04392819429778247,
+      "grad_norm": 0.5576267838478088,
+      "learning_rate": 9.986609590864719e-05,
+      "loss": 0.4768,
+      "step": 208
+    },
+    {
+      "epoch": 0.044139387539598736,
+      "grad_norm": 0.5000496506690979,
+      "learning_rate": 9.986360585635734e-05,
+      "loss": 0.4728,
+      "step": 209
+    },
+    {
+      "epoch": 0.044350580781414996,
+      "grad_norm": 0.4135527014732361,
+      "learning_rate": 9.986109289623848e-05,
+      "loss": 0.4769,
+      "step": 210
+    },
+    {
+      "epoch": 0.044561774023231256,
+      "grad_norm": 0.3956877887248993,
+      "learning_rate": 9.985855702944508e-05,
+      "loss": 0.4736,
+      "step": 211
+    },
+    {
+      "epoch": 0.044772967265047516,
+      "grad_norm": 0.585966169834137,
+      "learning_rate": 9.985599825714214e-05,
+      "loss": 0.472,
+      "step": 212
+    },
+    {
+      "epoch": 0.04498416050686378,
+      "grad_norm": 0.6574434638023376,
+      "learning_rate": 9.985341658050519e-05,
+      "loss": 0.475,
+      "step": 213
+    },
+    {
+      "epoch": 0.04519535374868004,
+      "grad_norm": 0.42700281739234924,
+      "learning_rate": 9.985081200072029e-05,
+      "loss": 0.4695,
+      "step": 214
+    },
+    {
+      "epoch": 0.0454065469904963,
+      "grad_norm": 0.47480159997940063,
+      "learning_rate": 9.984818451898399e-05,
+      "loss": 0.4737,
+      "step": 215
+    },
+    {
+      "epoch": 0.04561774023231257,
+      "grad_norm": 0.47683823108673096,
+      "learning_rate": 9.98455341365034e-05,
+      "loss": 0.4626,
+      "step": 216
+    },
+    {
+      "epoch": 0.04582893347412883,
+      "grad_norm": 0.43239492177963257,
+      "learning_rate": 9.98428608544961e-05,
+      "loss": 0.4729,
+      "step": 217
+    },
+    {
+      "epoch": 0.04604012671594509,
+      "grad_norm": 0.4528360962867737,
+      "learning_rate": 9.984016467419027e-05,
+      "loss": 0.4711,
+      "step": 218
+    },
+    {
+      "epoch": 0.04625131995776135,
+      "grad_norm": 0.6283021569252014,
+      "learning_rate": 9.983744559682451e-05,
+      "loss": 0.476,
+      "step": 219
+    },
+    {
+      "epoch": 0.046462513199577615,
+      "grad_norm": 0.40431347489356995,
+      "learning_rate": 9.983470362364801e-05,
+      "loss": 0.4583,
+      "step": 220
+    },
+    {
+      "epoch": 0.046673706441393875,
+      "grad_norm": 0.49223989248275757,
+      "learning_rate": 9.983193875592049e-05,
+      "loss": 0.4731,
+      "step": 221
+    },
+    {
+      "epoch": 0.046884899683210135,
+      "grad_norm": 0.45038843154907227,
+      "learning_rate": 9.982915099491213e-05,
+      "loss": 0.4736,
+      "step": 222
+    },
+    {
+      "epoch": 0.0470960929250264,
+      "grad_norm": 0.4200994074344635,
+      "learning_rate": 9.982634034190364e-05,
+      "loss": 0.4695,
+      "step": 223
+    },
+    {
+      "epoch": 0.04730728616684266,
+      "grad_norm": 0.382929265499115,
+      "learning_rate": 9.98235067981863e-05,
+      "loss": 0.4573,
+      "step": 224
+    },
+    {
+      "epoch": 0.04751847940865892,
+      "grad_norm": 0.5341088175773621,
+      "learning_rate": 9.982065036506183e-05,
+      "loss": 0.4633,
+      "step": 225
+    },
+    {
+      "epoch": 0.04772967265047518,
+      "grad_norm": 0.5148834586143494,
+      "learning_rate": 9.981777104384251e-05,
+      "loss": 0.461,
+      "step": 226
+    },
+    {
+      "epoch": 0.04794086589229145,
+      "grad_norm": 0.6391489505767822,
+      "learning_rate": 9.981486883585115e-05,
+      "loss": 0.4687,
+      "step": 227
+    },
+    {
+      "epoch": 0.04815205913410771,
+      "grad_norm": 0.3890036344528198,
+      "learning_rate": 9.981194374242104e-05,
+      "loss": 0.4666,
+      "step": 228
+    },
+    {
+      "epoch": 0.04836325237592397,
+      "grad_norm": 0.4448256492614746,
+      "learning_rate": 9.980899576489601e-05,
+      "loss": 0.4553,
+      "step": 229
+    },
+    {
+      "epoch": 0.048574445617740235,
+      "grad_norm": 0.5338238477706909,
+      "learning_rate": 9.980602490463036e-05,
+      "loss": 0.4584,
+      "step": 230
+    },
+    {
+      "epoch": 0.048785638859556495,
+      "grad_norm": 0.5028356909751892,
+      "learning_rate": 9.980303116298894e-05,
+      "loss": 0.4562,
+      "step": 231
+    },
+    {
+      "epoch": 0.048996832101372755,
+      "grad_norm": 0.4594384431838989,
+      "learning_rate": 9.980001454134714e-05,
+      "loss": 0.462,
+      "step": 232
+    },
+    {
+      "epoch": 0.04920802534318902,
+      "grad_norm": 0.4114096164703369,
+      "learning_rate": 9.979697504109077e-05,
+      "loss": 0.4529,
+      "step": 233
+    },
+    {
+      "epoch": 0.04941921858500528,
+      "grad_norm": 0.39983123540878296,
+      "learning_rate": 9.979391266361626e-05,
+      "loss": 0.45,
+      "step": 234
+    },
+    {
+      "epoch": 0.04963041182682154,
+      "grad_norm": 0.6592827439308167,
+      "learning_rate": 9.979082741033047e-05,
+      "loss": 0.4604,
+      "step": 235
+    },
+    {
+      "epoch": 0.0498416050686378,
+      "grad_norm": 0.6422247290611267,
+      "learning_rate": 9.978771928265079e-05,
+      "loss": 0.4625,
+      "step": 236
+    },
+    {
+      "epoch": 0.05005279831045407,
+      "grad_norm": 0.5138159990310669,
+      "learning_rate": 9.978458828200513e-05,
+      "loss": 0.4575,
+      "step": 237
+    },
+    {
+      "epoch": 0.05026399155227033,
+      "grad_norm": 0.39905717968940735,
+      "learning_rate": 9.97814344098319e-05,
+      "loss": 0.4502,
+      "step": 238
+    },
+    {
+      "epoch": 0.05047518479408659,
+      "grad_norm": 0.47037822008132935,
+      "learning_rate": 9.977825766758003e-05,
+      "loss": 0.4582,
+      "step": 239
+    },
+    {
+      "epoch": 0.050686378035902854,
+      "grad_norm": 0.4472580552101135,
+      "learning_rate": 9.977505805670894e-05,
+      "loss": 0.4505,
+      "step": 240
+    },
+    {
+      "epoch": 0.050897571277719114,
+      "grad_norm": 0.4599186182022095,
+      "learning_rate": 9.977183557868856e-05,
+      "loss": 0.4496,
+      "step": 241
+    },
+    {
+      "epoch": 0.051108764519535374,
+      "grad_norm": 0.3974071443080902,
+      "learning_rate": 9.976859023499933e-05,
+      "loss": 0.4449,
+      "step": 242
+    },
+    {
+      "epoch": 0.051319957761351634,
+      "grad_norm": 0.5136964917182922,
+      "learning_rate": 9.976532202713219e-05,
+      "loss": 0.4447,
+      "step": 243
+    },
+    {
+      "epoch": 0.0515311510031679,
+      "grad_norm": 0.41473063826560974,
+      "learning_rate": 9.97620309565886e-05,
+      "loss": 0.4431,
+      "step": 244
+    },
+    {
+      "epoch": 0.05174234424498416,
+      "grad_norm": 0.47496432065963745,
+      "learning_rate": 9.975871702488049e-05,
+      "loss": 0.4492,
+      "step": 245
+    },
+    {
+      "epoch": 0.05195353748680042,
+      "grad_norm": 0.5250230431556702,
+      "learning_rate": 9.975538023353033e-05,
+      "loss": 0.4485,
+      "step": 246
+    },
+    {
+      "epoch": 0.05216473072861669,
+      "grad_norm": 0.4929940104484558,
+      "learning_rate": 9.975202058407104e-05,
+      "loss": 0.4513,
+      "step": 247
+    },
+    {
+      "epoch": 0.05237592397043295,
+      "grad_norm": 0.44747957587242126,
+      "learning_rate": 9.974863807804614e-05,
+      "loss": 0.4502,
+      "step": 248
+    },
+    {
+      "epoch": 0.05258711721224921,
+      "grad_norm": 0.48234352469444275,
+      "learning_rate": 9.974523271700952e-05,
+      "loss": 0.451,
+      "step": 249
+    },
+    {
+      "epoch": 0.05279831045406547,
+      "grad_norm": 0.620921790599823,
+      "learning_rate": 9.974180450252569e-05,
+      "loss": 0.4506,
+      "step": 250
+    },
+    {
+      "epoch": 0.053009503695881734,
+      "grad_norm": 0.442747563123703,
+      "learning_rate": 9.973835343616957e-05,
+      "loss": 0.4399,
+      "step": 251
+    },
+    {
+      "epoch": 0.053220696937697994,
+      "grad_norm": 0.5102415084838867,
+      "learning_rate": 9.973487951952663e-05,
+      "loss": 0.4405,
+      "step": 252
+    },
+    {
+      "epoch": 0.05343189017951425,
+      "grad_norm": 0.7556713223457336,
+      "learning_rate": 9.97313827541928e-05,
+      "loss": 0.4541,
+      "step": 253
+    },
+    {
+      "epoch": 0.05364308342133052,
+      "grad_norm": 0.3216419219970703,
+      "learning_rate": 9.972786314177457e-05,
+      "loss": 0.4411,
+      "step": 254
+    },
+    {
+      "epoch": 0.05385427666314678,
+      "grad_norm": 0.468943327665329,
+      "learning_rate": 9.972432068388884e-05,
+      "loss": 0.4442,
+      "step": 255
+    },
+    {
+      "epoch": 0.05406546990496304,
+      "grad_norm": 0.7243496775627136,
+      "learning_rate": 9.972075538216308e-05,
+      "loss": 0.4499,
+      "step": 256
+    },
+    {
+      "epoch": 0.0542766631467793,
+      "grad_norm": 0.40651193261146545,
+      "learning_rate": 9.97171672382352e-05,
+      "loss": 0.4375,
+      "step": 257
+    },
+    {
+      "epoch": 0.05448785638859557,
+      "grad_norm": 0.43810659646987915,
+      "learning_rate": 9.971355625375364e-05,
+      "loss": 0.4355,
+      "step": 258
+    },
+    {
+      "epoch": 0.054699049630411826,
+      "grad_norm": 0.4257538914680481,
+      "learning_rate": 9.970992243037735e-05,
+      "loss": 0.4307,
+      "step": 259
+    },
+    {
+      "epoch": 0.054910242872228086,
+      "grad_norm": 0.589087724685669,
+      "learning_rate": 9.970626576977569e-05,
+      "loss": 0.4446,
+      "step": 260
+    },
+    {
+      "epoch": 0.05512143611404435,
+      "grad_norm": 0.7838018536567688,
+      "learning_rate": 9.97025862736286e-05,
+      "loss": 0.4521,
+      "step": 261
+    },
+    {
+      "epoch": 0.05533262935586061,
+      "grad_norm": 0.32075434923171997,
+      "learning_rate": 9.969888394362647e-05,
+      "loss": 0.4372,
+      "step": 262
+    },
+    {
+      "epoch": 0.05554382259767687,
+      "grad_norm": 0.5080913305282593,
+      "learning_rate": 9.96951587814702e-05,
+      "loss": 0.4461,
+      "step": 263
+    },
+    {
+      "epoch": 0.05575501583949313,
+      "grad_norm": 0.705872118473053,
+      "learning_rate": 9.969141078887113e-05,
+      "loss": 0.442,
+      "step": 264
+    },
+    {
+      "epoch": 0.0559662090813094,
+      "grad_norm": 0.5380907654762268,
+      "learning_rate": 9.968763996755114e-05,
+      "loss": 0.439,
+      "step": 265
+    },
+    {
+      "epoch": 0.05617740232312566,
+      "grad_norm": 0.49181580543518066,
+      "learning_rate": 9.968384631924262e-05,
+      "loss": 0.4465,
+      "step": 266
+    },
+    {
+      "epoch": 0.05638859556494192,
+      "grad_norm": 0.3819003105163574,
+      "learning_rate": 9.968002984568836e-05,
+      "loss": 0.4341,
+      "step": 267
+    },
+    {
+      "epoch": 0.056599788806758186,
+      "grad_norm": 0.6388276815414429,
+      "learning_rate": 9.967619054864171e-05,
+      "loss": 0.4415,
+      "step": 268
+    },
+    {
+      "epoch": 0.056810982048574446,
+      "grad_norm": 0.5784807801246643,
+      "learning_rate": 9.967232842986646e-05,
+      "loss": 0.4391,
+      "step": 269
+    },
+    {
+      "epoch": 0.057022175290390706,
+      "grad_norm": 0.4376797080039978,
+      "learning_rate": 9.966844349113694e-05,
+      "loss": 0.444,
+      "step": 270
+    },
+    {
+      "epoch": 0.05723336853220697,
+      "grad_norm": 0.49134719371795654,
+      "learning_rate": 9.966453573423791e-05,
+      "loss": 0.4315,
+      "step": 271
+    },
+    {
+      "epoch": 0.05744456177402323,
+      "grad_norm": 0.5387450456619263,
+      "learning_rate": 9.966060516096461e-05,
+      "loss": 0.4341,
+      "step": 272
+    },
+    {
+      "epoch": 0.05765575501583949,
+      "grad_norm": 0.5205551981925964,
+      "learning_rate": 9.965665177312283e-05,
+      "loss": 0.4359,
+      "step": 273
+    },
+    {
+      "epoch": 0.05786694825765575,
+      "grad_norm": 0.43937525153160095,
+      "learning_rate": 9.965267557252873e-05,
+      "loss": 0.433,
+      "step": 274
+    },
+    {
+      "epoch": 0.05807814149947202,
+      "grad_norm": 0.5116094946861267,
+      "learning_rate": 9.964867656100909e-05,
+      "loss": 0.4407,
+      "step": 275
+    },
+    {
+      "epoch": 0.05828933474128828,
+      "grad_norm": 0.3165574371814728,
+      "learning_rate": 9.964465474040104e-05,
+      "loss": 0.4275,
+      "step": 276
+    },
+    {
+      "epoch": 0.05850052798310454,
+      "grad_norm": 0.5547077059745789,
+      "learning_rate": 9.964061011255226e-05,
+      "loss": 0.4351,
+      "step": 277
+    },
+    {
+      "epoch": 0.058711721224920806,
+      "grad_norm": 0.6870663166046143,
+      "learning_rate": 9.963654267932092e-05,
+      "loss": 0.4482,
+      "step": 278
+    },
+    {
+      "epoch": 0.058922914466737066,
+      "grad_norm": 0.4742313325405121,
+      "learning_rate": 9.96324524425756e-05,
+      "loss": 0.4382,
+      "step": 279
+    },
+    {
+      "epoch": 0.059134107708553325,
+      "grad_norm": 0.6057625412940979,
+      "learning_rate": 9.96283394041954e-05,
+      "loss": 0.4446,
+      "step": 280
+    },
+    {
+      "epoch": 0.059345300950369585,
+      "grad_norm": 0.48816588521003723,
+      "learning_rate": 9.96242035660699e-05,
+      "loss": 0.4367,
+      "step": 281
+    },
+    {
+      "epoch": 0.05955649419218585,
+      "grad_norm": 0.4100516438484192,
+      "learning_rate": 9.962004493009915e-05,
+      "loss": 0.4332,
+      "step": 282
+    },
+    {
+      "epoch": 0.05976768743400211,
+      "grad_norm": 0.5070685744285583,
+      "learning_rate": 9.961586349819366e-05,
+      "loss": 0.4345,
+      "step": 283
+    },
+    {
+      "epoch": 0.05997888067581837,
+      "grad_norm": 0.704001784324646,
+      "learning_rate": 9.961165927227443e-05,
+      "loss": 0.4434,
+      "step": 284
+    },
+    {
+      "epoch": 0.06019007391763464,
+      "grad_norm": 0.46731334924697876,
+      "learning_rate": 9.96074322542729e-05,
+      "loss": 0.4355,
+      "step": 285
+    },
+    {
+      "epoch": 0.0604012671594509,
+      "grad_norm": 0.4700104892253876,
+      "learning_rate": 9.960318244613103e-05,
+      "loss": 0.4372,
+      "step": 286
+    },
+    {
+      "epoch": 0.06061246040126716,
+      "grad_norm": 0.4634333848953247,
+      "learning_rate": 9.95989098498012e-05,
+      "loss": 0.4394,
+      "step": 287
+    },
+    {
+      "epoch": 0.06082365364308342,
+      "grad_norm": 0.34873420000076294,
+      "learning_rate": 9.95946144672463e-05,
+      "loss": 0.4299,
+      "step": 288
+    },
+    {
+      "epoch": 0.061034846884899685,
+      "grad_norm": 0.575365424156189,
+      "learning_rate": 9.959029630043968e-05,
+      "loss": 0.4419,
+      "step": 289
+    },
+    {
+      "epoch": 0.061246040126715945,
+      "grad_norm": 0.36025410890579224,
+      "learning_rate": 9.958595535136512e-05,
+      "loss": 0.4274,
+      "step": 290
+    },
+    {
+      "epoch": 0.061457233368532205,
+      "grad_norm": 0.44464626908302307,
+      "learning_rate": 9.95815916220169e-05,
+      "loss": 0.4305,
+      "step": 291
+    },
+    {
+      "epoch": 0.06166842661034847,
+      "grad_norm": 0.473283588886261,
+      "learning_rate": 9.957720511439977e-05,
+      "loss": 0.4292,
+      "step": 292
+    },
+    {
+      "epoch": 0.06187961985216473,
+      "grad_norm": 0.4417226314544678,
+      "learning_rate": 9.957279583052895e-05,
+      "loss": 0.4302,
+      "step": 293
+    },
+    {
+      "epoch": 0.06209081309398099,
+      "grad_norm": 0.5806403756141663,
+      "learning_rate": 9.956836377243007e-05,
+      "loss": 0.4353,
+      "step": 294
+    },
+    {
+      "epoch": 0.06230200633579725,
+      "grad_norm": 0.4068332016468048,
+      "learning_rate": 9.95639089421393e-05,
+      "loss": 0.4284,
+      "step": 295
+    },
+    {
+      "epoch": 0.06251319957761352,
+      "grad_norm": 0.3271664083003998,
+      "learning_rate": 9.95594313417032e-05,
+      "loss": 0.4226,
+      "step": 296
+    },
+    {
+      "epoch": 0.06272439281942978,
+      "grad_norm": 0.4257061779499054,
+      "learning_rate": 9.955493097317884e-05,
+      "loss": 0.4212,
+      "step": 297
+    },
+    {
+      "epoch": 0.06293558606124604,
+      "grad_norm": 0.6846235990524292,
+      "learning_rate": 9.955040783863374e-05,
+      "loss": 0.4234,
+      "step": 298
+    },
+    {
+      "epoch": 0.0631467793030623,
+      "grad_norm": 0.49378421902656555,
+      "learning_rate": 9.954586194014584e-05,
+      "loss": 0.4246,
+      "step": 299
+    },
+    {
+      "epoch": 0.06335797254487856,
+      "grad_norm": 0.4490949511528015,
+      "learning_rate": 9.954129327980361e-05,
+      "loss": 0.4253,
+      "step": 300
+    },
+    {
+      "epoch": 0.06356916578669483,
+      "grad_norm": 0.5967462658882141,
+      "learning_rate": 9.953670185970592e-05,
+      "loss": 0.4319,
+      "step": 301
+    },
+    {
+      "epoch": 0.06378035902851109,
+      "grad_norm": 0.4124678075313568,
+      "learning_rate": 9.953208768196212e-05,
+      "loss": 0.4228,
+      "step": 302
+    },
+    {
+      "epoch": 0.06399155227032735,
+      "grad_norm": 0.2861862778663635,
+      "learning_rate": 9.952745074869199e-05,
+      "loss": 0.4217,
+      "step": 303
+    },
+    {
+      "epoch": 0.06420274551214361,
+      "grad_norm": 0.34011000394821167,
+      "learning_rate": 9.952279106202581e-05,
+      "loss": 0.4212,
+      "step": 304
+    },
+    {
+      "epoch": 0.06441393875395987,
+      "grad_norm": 0.6858258843421936,
+      "learning_rate": 9.951810862410426e-05,
+      "loss": 0.43,
+      "step": 305
+    },
+    {
+      "epoch": 0.06462513199577613,
+      "grad_norm": 0.5112482905387878,
+      "learning_rate": 9.951340343707852e-05,
+      "loss": 0.4201,
+      "step": 306
+    },
+    {
+      "epoch": 0.06483632523759239,
+      "grad_norm": 0.4969583749771118,
+      "learning_rate": 9.950867550311018e-05,
+      "loss": 0.4251,
+      "step": 307
+    },
+    {
+      "epoch": 0.06504751847940866,
+      "grad_norm": 0.4808504581451416,
+      "learning_rate": 9.950392482437131e-05,
+      "loss": 0.426,
+      "step": 308
+    },
+    {
+      "epoch": 0.06525871172122492,
+      "grad_norm": 0.563737154006958,
+      "learning_rate": 9.949915140304442e-05,
+      "loss": 0.4295,
+      "step": 309
+    },
+    {
+      "epoch": 0.06546990496304118,
+      "grad_norm": 0.38801857829093933,
+      "learning_rate": 9.949435524132245e-05,
+      "loss": 0.4255,
+      "step": 310
+    },
+    {
+      "epoch": 0.06568109820485744,
+      "grad_norm": 0.5231031179428101,
+      "learning_rate": 9.948953634140884e-05,
+      "loss": 0.4273,
+      "step": 311
+    },
+    {
+      "epoch": 0.0658922914466737,
+      "grad_norm": 0.3490457236766815,
+      "learning_rate": 9.948469470551738e-05,
+      "loss": 0.4167,
+      "step": 312
+    },
+    {
+      "epoch": 0.06610348468848996,
+      "grad_norm": 0.35016798973083496,
+      "learning_rate": 9.947983033587242e-05,
+      "loss": 0.4138,
+      "step": 313
+    },
+    {
+      "epoch": 0.06631467793030622,
+      "grad_norm": 0.594514012336731,
+      "learning_rate": 9.947494323470867e-05,
+      "loss": 0.4209,
+      "step": 314
+    },
+    {
+      "epoch": 0.0665258711721225,
+      "grad_norm": 0.5466874241828918,
+      "learning_rate": 9.947003340427133e-05,
+      "loss": 0.4169,
+      "step": 315
+    },
+    {
+      "epoch": 0.06673706441393876,
+      "grad_norm": 0.43836620450019836,
+      "learning_rate": 9.946510084681602e-05,
+      "loss": 0.418,
+      "step": 316
+    },
+    {
+      "epoch": 0.06694825765575502,
+      "grad_norm": 0.49365633726119995,
+      "learning_rate": 9.946014556460879e-05,
+      "loss": 0.4178,
+      "step": 317
+    },
+    {
+      "epoch": 0.06715945089757128,
+      "grad_norm": 0.4619889259338379,
+      "learning_rate": 9.945516755992617e-05,
+      "loss": 0.4229,
+      "step": 318
+    },
+    {
+      "epoch": 0.06737064413938754,
+      "grad_norm": 0.4487181007862091,
+      "learning_rate": 9.945016683505508e-05,
+      "loss": 0.4139,
+      "step": 319
+    },
+    {
+      "epoch": 0.0675818373812038,
+      "grad_norm": 0.46683815121650696,
+      "learning_rate": 9.944514339229292e-05,
+      "loss": 0.4178,
+      "step": 320
+    },
+    {
+      "epoch": 0.06779303062302007,
+      "grad_norm": 0.5421741604804993,
+      "learning_rate": 9.94400972339475e-05,
+      "loss": 0.4256,
+      "step": 321
+    },
+    {
+      "epoch": 0.06800422386483633,
+      "grad_norm": 0.33267587423324585,
+      "learning_rate": 9.943502836233707e-05,
+      "loss": 0.4153,
+      "step": 322
+    },
+    {
+      "epoch": 0.06821541710665259,
+      "grad_norm": 0.5326710939407349,
+      "learning_rate": 9.942993677979031e-05,
+      "loss": 0.4251,
+      "step": 323
+    },
+    {
+      "epoch": 0.06842661034846885,
+      "grad_norm": 0.46235206723213196,
+      "learning_rate": 9.942482248864638e-05,
+      "loss": 0.4211,
+      "step": 324
+    },
+    {
+      "epoch": 0.06863780359028511,
+      "grad_norm": 0.43010082840919495,
+      "learning_rate": 9.941968549125481e-05,
+      "loss": 0.4184,
+      "step": 325
+    },
+    {
+      "epoch": 0.06884899683210137,
+      "grad_norm": 0.3705753982067108,
+      "learning_rate": 9.941452578997558e-05,
+      "loss": 0.4127,
+      "step": 326
+    },
+    {
+      "epoch": 0.06906019007391763,
+      "grad_norm": 0.4769006371498108,
+      "learning_rate": 9.940934338717913e-05,
+      "loss": 0.4151,
+      "step": 327
+    },
+    {
+      "epoch": 0.0692713833157339,
+      "grad_norm": 0.4973369836807251,
+      "learning_rate": 9.940413828524629e-05,
+      "loss": 0.4156,
+      "step": 328
+    },
+    {
+      "epoch": 0.06948257655755016,
+      "grad_norm": 0.36741238832473755,
+      "learning_rate": 9.939891048656835e-05,
+      "loss": 0.4119,
+      "step": 329
+    },
+    {
+      "epoch": 0.06969376979936642,
+      "grad_norm": 0.5299829840660095,
+      "learning_rate": 9.9393659993547e-05,
+      "loss": 0.4101,
+      "step": 330
+    },
+    {
+      "epoch": 0.06990496304118268,
+      "grad_norm": 0.5758510828018188,
+      "learning_rate": 9.938838680859434e-05,
+      "loss": 0.4175,
+      "step": 331
+    },
+    {
+      "epoch": 0.07011615628299894,
+      "grad_norm": 0.34021636843681335,
+      "learning_rate": 9.938309093413298e-05,
+      "loss": 0.4125,
+      "step": 332
+    },
+    {
+      "epoch": 0.0703273495248152,
+      "grad_norm": 0.37205636501312256,
+      "learning_rate": 9.937777237259587e-05,
+      "loss": 0.4076,
+      "step": 333
+    },
+    {
+      "epoch": 0.07053854276663146,
+      "grad_norm": 0.44349250197410583,
+      "learning_rate": 9.937243112642638e-05,
+      "loss": 0.4094,
+      "step": 334
+    },
+    {
+      "epoch": 0.07074973600844774,
+      "grad_norm": 0.6392989158630371,
+      "learning_rate": 9.936706719807837e-05,
+      "loss": 0.4204,
+      "step": 335
+    },
+    {
+      "epoch": 0.070960929250264,
+      "grad_norm": 0.3534295856952667,
+      "learning_rate": 9.936168059001609e-05,
+      "loss": 0.4066,
+      "step": 336
+    },
+    {
+      "epoch": 0.07117212249208026,
+      "grad_norm": 0.5454416275024414,
+      "learning_rate": 9.935627130471417e-05,
+      "loss": 0.4202,
+      "step": 337
+    },
+    {
+      "epoch": 0.07138331573389652,
+      "grad_norm": 0.5341911911964417,
+      "learning_rate": 9.93508393446577e-05,
+      "loss": 0.4114,
+      "step": 338
+    },
+    {
+      "epoch": 0.07159450897571278,
+      "grad_norm": 0.374310702085495,
+      "learning_rate": 9.934538471234216e-05,
+      "loss": 0.4125,
+      "step": 339
+    },
+    {
+      "epoch": 0.07180570221752904,
+      "grad_norm": 0.4413489103317261,
+      "learning_rate": 9.933990741027349e-05,
+      "loss": 0.4024,
+      "step": 340
+    },
+    {
+      "epoch": 0.0720168954593453,
+      "grad_norm": 0.4629468619823456,
+      "learning_rate": 9.9334407440968e-05,
+      "loss": 0.4099,
+      "step": 341
+    },
+    {
+      "epoch": 0.07222808870116157,
+      "grad_norm": 0.5960588455200195,
+      "learning_rate": 9.932888480695243e-05,
+      "loss": 0.4109,
+      "step": 342
+    },
+    {
+      "epoch": 0.07243928194297783,
+      "grad_norm": 0.37413856387138367,
+      "learning_rate": 9.932333951076394e-05,
+      "loss": 0.3988,
+      "step": 343
+    },
+    {
+      "epoch": 0.07265047518479409,
+      "grad_norm": 0.4418419301509857,
+      "learning_rate": 9.931777155495008e-05,
+      "loss": 0.4082,
+      "step": 344
+    },
+    {
+      "epoch": 0.07286166842661035,
+      "grad_norm": 0.6489225029945374,
+      "learning_rate": 9.931218094206882e-05,
+      "loss": 0.417,
+      "step": 345
+    },
+    {
+      "epoch": 0.07307286166842661,
+      "grad_norm": 0.4457958936691284,
+      "learning_rate": 9.930656767468858e-05,
+      "loss": 0.412,
+      "step": 346
+    },
+    {
+      "epoch": 0.07328405491024287,
+      "grad_norm": 0.41500693559646606,
+      "learning_rate": 9.930093175538811e-05,
+      "loss": 0.4045,
+      "step": 347
+    },
+    {
+      "epoch": 0.07349524815205913,
+      "grad_norm": 0.35880276560783386,
+      "learning_rate": 9.929527318675663e-05,
+      "loss": 0.4,
+      "step": 348
+    },
+    {
+      "epoch": 0.0737064413938754,
+      "grad_norm": 0.4673283100128174,
+      "learning_rate": 9.928959197139372e-05,
+      "loss": 0.4027,
+      "step": 349
+    },
+    {
+      "epoch": 0.07391763463569166,
+      "grad_norm": 0.4399472773075104,
+      "learning_rate": 9.928388811190939e-05,
+      "loss": 0.42,
+      "step": 350
+    },
+    {
+      "epoch": 0.07412882787750792,
+      "grad_norm": 0.5320001244544983,
+      "learning_rate": 9.927816161092407e-05,
+      "loss": 0.4074,
+      "step": 351
+    },
+    {
+      "epoch": 0.07434002111932418,
+      "grad_norm": 0.4711773097515106,
+      "learning_rate": 9.927241247106855e-05,
+      "loss": 0.4061,
+      "step": 352
+    },
+    {
+      "epoch": 0.07455121436114044,
+      "grad_norm": 0.6437652111053467,
+      "learning_rate": 9.926664069498405e-05,
+      "loss": 0.4174,
+      "step": 353
+    },
+    {
+      "epoch": 0.0747624076029567,
+      "grad_norm": 0.4807835519313812,
+      "learning_rate": 9.926084628532218e-05,
+      "loss": 0.4057,
+      "step": 354
+    },
+    {
+      "epoch": 0.07497360084477296,
+      "grad_norm": 0.415286123752594,
+      "learning_rate": 9.925502924474494e-05,
+      "loss": 0.4008,
+      "step": 355
+    },
+    {
+      "epoch": 0.07518479408658924,
+      "grad_norm": 0.41239967942237854,
+      "learning_rate": 9.924918957592477e-05,
+      "loss": 0.4035,
+      "step": 356
+    },
+    {
+      "epoch": 0.0753959873284055,
+      "grad_norm": 0.7052282094955444,
+      "learning_rate": 9.924332728154442e-05,
+      "loss": 0.4139,
+      "step": 357
+    },
+    {
+      "epoch": 0.07560718057022175,
+      "grad_norm": 0.37168359756469727,
+      "learning_rate": 9.923744236429711e-05,
+      "loss": 0.4051,
+      "step": 358
+    },
+    {
+      "epoch": 0.07581837381203801,
+      "grad_norm": 0.40557220578193665,
+      "learning_rate": 9.923153482688644e-05,
+      "loss": 0.4091,
+      "step": 359
+    },
+    {
+      "epoch": 0.07602956705385427,
+      "grad_norm": 0.45359525084495544,
+      "learning_rate": 9.922560467202637e-05,
+      "loss": 0.4049,
+      "step": 360
+    },
+    {
+      "epoch": 0.07624076029567053,
+      "grad_norm": 0.46855127811431885,
+      "learning_rate": 9.921965190244129e-05,
+      "loss": 0.4001,
+      "step": 361
+    },
+    {
+      "epoch": 0.0764519535374868,
+      "grad_norm": 0.37576156854629517,
+      "learning_rate": 9.921367652086594e-05,
+      "loss": 0.4066,
+      "step": 362
+    },
+    {
+      "epoch": 0.07666314677930307,
+      "grad_norm": 0.36967527866363525,
+      "learning_rate": 9.920767853004549e-05,
+      "loss": 0.3999,
+      "step": 363
+    },
+    {
+      "epoch": 0.07687434002111933,
+      "grad_norm": 0.5333089232444763,
+      "learning_rate": 9.920165793273546e-05,
+      "loss": 0.4124,
+      "step": 364
+    },
+    {
+      "epoch": 0.07708553326293559,
+      "grad_norm": 0.38473832607269287,
+      "learning_rate": 9.919561473170177e-05,
+      "loss": 0.3945,
+      "step": 365
+    },
+    {
+      "epoch": 0.07729672650475185,
+      "grad_norm": 0.6910049915313721,
+      "learning_rate": 9.918954892972074e-05,
+      "loss": 0.4088,
+      "step": 366
+    },
+    {
+      "epoch": 0.07750791974656811,
+      "grad_norm": 0.5292808413505554,
+      "learning_rate": 9.918346052957905e-05,
+      "loss": 0.4105,
+      "step": 367
+    },
+    {
+      "epoch": 0.07771911298838437,
+      "grad_norm": 0.40103912353515625,
+      "learning_rate": 9.917734953407377e-05,
+      "loss": 0.3982,
+      "step": 368
+    },
+    {
+      "epoch": 0.07793030623020063,
+      "grad_norm": 0.42740222811698914,
+      "learning_rate": 9.917121594601235e-05,
+      "loss": 0.4028,
+      "step": 369
+    },
+    {
+      "epoch": 0.0781414994720169,
+      "grad_norm": 0.42264145612716675,
+      "learning_rate": 9.916505976821263e-05,
+      "loss": 0.411,
+      "step": 370
+    },
+    {
+      "epoch": 0.07835269271383316,
+      "grad_norm": 0.48413199186325073,
+      "learning_rate": 9.915888100350282e-05,
+      "loss": 0.406,
+      "step": 371
+    },
+    {
+      "epoch": 0.07856388595564942,
+      "grad_norm": 0.42317599058151245,
+      "learning_rate": 9.915267965472148e-05,
+      "loss": 0.3957,
+      "step": 372
+    },
+    {
+      "epoch": 0.07877507919746568,
+      "grad_norm": 0.42475858330726624,
+      "learning_rate": 9.91464557247176e-05,
+      "loss": 0.4011,
+      "step": 373
+    },
+    {
+      "epoch": 0.07898627243928194,
+      "grad_norm": 0.36108893156051636,
+      "learning_rate": 9.914020921635048e-05,
+      "loss": 0.3956,
+      "step": 374
+    },
+    {
+      "epoch": 0.0791974656810982,
+      "grad_norm": 0.39691439270973206,
+      "learning_rate": 9.913394013248987e-05,
+      "loss": 0.3977,
+      "step": 375
+    },
+    {
+      "epoch": 0.07940865892291446,
+      "grad_norm": 0.4440408945083618,
+      "learning_rate": 9.912764847601581e-05,
+      "loss": 0.3965,
+      "step": 376
+    },
+    {
+      "epoch": 0.07961985216473073,
+      "grad_norm": 0.4564206898212433,
+      "learning_rate": 9.912133424981878e-05,
+      "loss": 0.3985,
+      "step": 377
+    },
+    {
+      "epoch": 0.079831045406547,
+      "grad_norm": 0.3990284502506256,
+      "learning_rate": 9.911499745679958e-05,
+      "loss": 0.3992,
+      "step": 378
+    },
+    {
+      "epoch": 0.08004223864836325,
+      "grad_norm": 0.4210834205150604,
+      "learning_rate": 9.910863809986941e-05,
+      "loss": 0.3931,
+      "step": 379
+    },
+    {
+      "epoch": 0.08025343189017951,
+      "grad_norm": 0.7663113474845886,
+      "learning_rate": 9.910225618194979e-05,
+      "loss": 0.412,
+      "step": 380
+    },
+    {
+      "epoch": 0.08046462513199577,
+      "grad_norm": 0.38950127363204956,
+      "learning_rate": 9.909585170597266e-05,
+      "loss": 0.4077,
+      "step": 381
+    },
+    {
+      "epoch": 0.08067581837381203,
+      "grad_norm": 0.44707247614860535,
+      "learning_rate": 9.90894246748803e-05,
+      "loss": 0.4038,
+      "step": 382
+    },
+    {
+      "epoch": 0.0808870116156283,
+      "grad_norm": 0.5945481061935425,
+      "learning_rate": 9.908297509162533e-05,
+      "loss": 0.4202,
+      "step": 383
+    },
+    {
+      "epoch": 0.08109820485744457,
+      "grad_norm": 0.35499122738838196,
+      "learning_rate": 9.90765029591708e-05,
+      "loss": 0.3972,
+      "step": 384
+    },
+    {
+      "epoch": 0.08130939809926083,
+      "grad_norm": 0.5596731305122375,
+      "learning_rate": 9.907000828049001e-05,
+      "loss": 0.4131,
+      "step": 385
+    },
+    {
+      "epoch": 0.08152059134107709,
+      "grad_norm": 0.3295954167842865,
+      "learning_rate": 9.906349105856672e-05,
+      "loss": 0.3939,
+      "step": 386
+    },
+    {
+      "epoch": 0.08173178458289335,
+      "grad_norm": 0.7758201956748962,
+      "learning_rate": 9.905695129639498e-05,
+      "loss": 0.4164,
+      "step": 387
+    },
+    {
+      "epoch": 0.0819429778247096,
+      "grad_norm": 0.46655789017677307,
+      "learning_rate": 9.905038899697924e-05,
+      "loss": 0.4008,
+      "step": 388
+    },
+    {
+      "epoch": 0.08215417106652587,
+      "grad_norm": 0.6977465152740479,
+      "learning_rate": 9.904380416333428e-05,
+      "loss": 0.4212,
+      "step": 389
+    },
+    {
+      "epoch": 0.08236536430834214,
+      "grad_norm": 0.4163126051425934,
+      "learning_rate": 9.903719679848522e-05,
+      "loss": 0.4134,
+      "step": 390
+    },
+    {
+      "epoch": 0.0825765575501584,
+      "grad_norm": 0.4952920079231262,
+      "learning_rate": 9.903056690546758e-05,
+      "loss": 0.4075,
+      "step": 391
+    },
+    {
+      "epoch": 0.08278775079197466,
+      "grad_norm": 0.42831680178642273,
+      "learning_rate": 9.902391448732718e-05,
+      "loss": 0.3988,
+      "step": 392
+    },
+    {
+      "epoch": 0.08299894403379092,
+      "grad_norm": 0.5524213910102844,
+      "learning_rate": 9.90172395471202e-05,
+      "loss": 0.4158,
+      "step": 393
+    },
+    {
+      "epoch": 0.08321013727560718,
+      "grad_norm": 0.5064900517463684,
+      "learning_rate": 9.901054208791318e-05,
+      "loss": 0.4063,
+      "step": 394
+    },
+    {
+      "epoch": 0.08342133051742344,
+      "grad_norm": 0.3617955148220062,
+      "learning_rate": 9.900382211278299e-05,
+      "loss": 0.402,
+      "step": 395
+    },
+    {
+      "epoch": 0.0836325237592397,
+      "grad_norm": 0.40193456411361694,
+      "learning_rate": 9.899707962481687e-05,
+      "loss": 0.4081,
+      "step": 396
+    },
+    {
+      "epoch": 0.08384371700105597,
+      "grad_norm": 0.39680448174476624,
+      "learning_rate": 9.899031462711237e-05,
+      "loss": 0.4082,
+      "step": 397
+    },
+    {
+      "epoch": 0.08405491024287223,
+      "grad_norm": 0.4096421003341675,
+      "learning_rate": 9.89835271227774e-05,
+      "loss": 0.3988,
+      "step": 398
+    },
+    {
+      "epoch": 0.08426610348468849,
+      "grad_norm": 0.49320128560066223,
+      "learning_rate": 9.897671711493022e-05,
+      "loss": 0.3993,
+      "step": 399
+    },
+    {
+      "epoch": 0.08447729672650475,
+      "grad_norm": 0.4275283217430115,
+      "learning_rate": 9.89698846066994e-05,
+      "loss": 0.3927,
+      "step": 400
+    },
+    {
+      "epoch": 0.08447729672650475,
+      "eval_train_split_loss": 0.526931881904602,
+      "eval_train_split_runtime": 6.6481,
+      "eval_train_split_samples_per_second": 1.504,
+      "eval_train_split_steps_per_second": 0.15,
+      "step": 400
+    },
+    {
+      "epoch": 0.08468848996832101,
+      "grad_norm": 0.4381150007247925,
+      "learning_rate": 9.896302960122385e-05,
+      "loss": 0.3995,
+      "step": 401
+    },
+    {
+      "epoch": 0.08489968321013727,
+      "grad_norm": 0.3630185127258301,
+      "learning_rate": 9.895615210165285e-05,
+      "loss": 0.3934,
+      "step": 402
+    },
+    {
+      "epoch": 0.08511087645195353,
+      "grad_norm": 0.40775397419929504,
+      "learning_rate": 9.894925211114598e-05,
+      "loss": 0.3941,
+      "step": 403
+    },
+    {
+      "epoch": 0.0853220696937698,
+      "grad_norm": 0.5196607708930969,
+      "learning_rate": 9.894232963287315e-05,
+      "loss": 0.3972,
+      "step": 404
+    },
+    {
+      "epoch": 0.08553326293558607,
+      "grad_norm": 0.4806980788707733,
+      "learning_rate": 9.893538467001465e-05,
+      "loss": 0.3882,
+      "step": 405
+    },
+    {
+      "epoch": 0.08574445617740233,
+      "grad_norm": 0.38502663373947144,
+      "learning_rate": 9.892841722576103e-05,
+      "loss": 0.3874,
+      "step": 406
+    },
+    {
+      "epoch": 0.08595564941921859,
+      "grad_norm": 0.3797892928123474,
+      "learning_rate": 9.892142730331321e-05,
+      "loss": 0.3867,
+      "step": 407
+    },
+    {
+      "epoch": 0.08616684266103485,
+      "grad_norm": 0.3686940371990204,
+      "learning_rate": 9.891441490588245e-05,
+      "loss": 0.3814,
+      "step": 408
+    },
+    {
+      "epoch": 0.0863780359028511,
+      "grad_norm": 0.5003431439399719,
+      "learning_rate": 9.890738003669029e-05,
+      "loss": 0.3875,
+      "step": 409
+    },
+    {
+      "epoch": 0.08658922914466737,
+      "grad_norm": 0.5106502175331116,
+      "learning_rate": 9.890032269896861e-05,
+      "loss": 0.3934,
+      "step": 410
+    },
+    {
+      "epoch": 0.08680042238648364,
+      "grad_norm": 0.43301957845687866,
+      "learning_rate": 9.889324289595964e-05,
+      "loss": 0.3912,
+      "step": 411
+    },
+    {
+      "epoch": 0.0870116156282999,
+      "grad_norm": 0.3765442967414856,
+      "learning_rate": 9.888614063091593e-05,
+      "loss": 0.3822,
+      "step": 412
+    },
+    {
+      "epoch": 0.08722280887011616,
+      "grad_norm": 0.3632931113243103,
+      "learning_rate": 9.887901590710029e-05,
+      "loss": 0.3868,
+      "step": 413
+    },
+    {
+      "epoch": 0.08743400211193242,
+      "grad_norm": 0.4401683807373047,
+      "learning_rate": 9.88718687277859e-05,
+      "loss": 0.3902,
+      "step": 414
+    },
+    {
+      "epoch": 0.08764519535374868,
+      "grad_norm": 0.40098610520362854,
+      "learning_rate": 9.886469909625625e-05,
+      "loss": 0.3877,
+      "step": 415
+    },
+    {
+      "epoch": 0.08785638859556494,
+      "grad_norm": 0.3971259891986847,
+      "learning_rate": 9.885750701580515e-05,
+      "loss": 0.3891,
+      "step": 416
+    },
+    {
+      "epoch": 0.0880675818373812,
+      "grad_norm": 0.5239662528038025,
+      "learning_rate": 9.885029248973667e-05,
+      "loss": 0.3925,
+      "step": 417
+    },
+    {
+      "epoch": 0.08827877507919747,
+      "grad_norm": 0.39574363827705383,
+      "learning_rate": 9.884305552136529e-05,
+      "loss": 0.3829,
+      "step": 418
+    },
+    {
+      "epoch": 0.08848996832101373,
+      "grad_norm": 0.33224740624427795,
+      "learning_rate": 9.88357961140157e-05,
+      "loss": 0.3803,
+      "step": 419
+    },
+    {
+      "epoch": 0.08870116156282999,
+      "grad_norm": 0.4350009262561798,
+      "learning_rate": 9.882851427102298e-05,
+      "loss": 0.3854,
+      "step": 420
+    },
+    {
+      "epoch": 0.08891235480464625,
+      "grad_norm": 0.4803755581378937,
+      "learning_rate": 9.882120999573243e-05,
+      "loss": 0.3865,
+      "step": 421
+    },
+    {
+      "epoch": 0.08912354804646251,
+      "grad_norm": 0.5031775832176208,
+      "learning_rate": 9.881388329149976e-05,
+      "loss": 0.3916,
+      "step": 422
+    },
+    {
+      "epoch": 0.08933474128827877,
+      "grad_norm": 0.454530268907547,
+      "learning_rate": 9.88065341616909e-05,
+      "loss": 0.3822,
+      "step": 423
+    },
+    {
+      "epoch": 0.08954593453009503,
+      "grad_norm": 0.4043891429901123,
+      "learning_rate": 9.879916260968212e-05,
+      "loss": 0.3797,
+      "step": 424
+    },
+    {
+      "epoch": 0.0897571277719113,
+      "grad_norm": 0.3948330879211426,
+      "learning_rate": 9.879176863885997e-05,
+      "loss": 0.3827,
+      "step": 425
+    },
+    {
+      "epoch": 0.08996832101372756,
+      "grad_norm": 0.3381126821041107,
+      "learning_rate": 9.878435225262135e-05,
+      "loss": 0.3836,
+      "step": 426
+    },
+    {
+      "epoch": 0.09017951425554382,
+      "grad_norm": 0.4839237332344055,
+      "learning_rate": 9.877691345437337e-05,
+      "loss": 0.3868,
+      "step": 427
+    },
+    {
+      "epoch": 0.09039070749736008,
+      "grad_norm": 0.37446367740631104,
+      "learning_rate": 9.87694522475335e-05,
+      "loss": 0.374,
+      "step": 428
+    },
+    {
+      "epoch": 0.09060190073917634,
+      "grad_norm": 0.5827283263206482,
+      "learning_rate": 9.876196863552953e-05,
+      "loss": 0.3883,
+      "step": 429
+    },
+    {
+      "epoch": 0.0908130939809926,
+      "grad_norm": 0.40927037596702576,
+      "learning_rate": 9.875446262179948e-05,
+      "loss": 0.3855,
+      "step": 430
+    },
+    {
+      "epoch": 0.09102428722280886,
+      "grad_norm": 0.5010876059532166,
+      "learning_rate": 9.874693420979166e-05,
+      "loss": 0.3822,
+      "step": 431
+    },
+    {
+      "epoch": 0.09123548046462514,
+      "grad_norm": 0.3707132935523987,
+      "learning_rate": 9.873938340296474e-05,
+      "loss": 0.3849,
+      "step": 432
+    },
+    {
+      "epoch": 0.0914466737064414,
+      "grad_norm": 0.4502716362476349,
+      "learning_rate": 9.87318102047876e-05,
+      "loss": 0.3842,
+      "step": 433
+    },
+    {
+      "epoch": 0.09165786694825766,
+      "grad_norm": 0.36961624026298523,
+      "learning_rate": 9.872421461873946e-05,
+      "loss": 0.3783,
+      "step": 434
+    },
+    {
+      "epoch": 0.09186906019007392,
+      "grad_norm": 0.451084166765213,
+      "learning_rate": 9.87165966483098e-05,
+      "loss": 0.3755,
+      "step": 435
+    },
+    {
+      "epoch": 0.09208025343189018,
+      "grad_norm": 0.5110785961151123,
+      "learning_rate": 9.87089562969984e-05,
+      "loss": 0.3835,
+      "step": 436
+    },
+    {
+      "epoch": 0.09229144667370644,
+      "grad_norm": 0.3381638824939728,
+      "learning_rate": 9.87012935683153e-05,
+      "loss": 0.3816,
+      "step": 437
+    },
+    {
+      "epoch": 0.0925026399155227,
+      "grad_norm": 0.3330974578857422,
+      "learning_rate": 9.869360846578084e-05,
+      "loss": 0.3907,
+      "step": 438
+    },
+    {
+      "epoch": 0.09271383315733897,
+      "grad_norm": 0.3130118250846863,
+      "learning_rate": 9.868590099292563e-05,
+      "loss": 0.3776,
+      "step": 439
+    },
+    {
+      "epoch": 0.09292502639915523,
+      "grad_norm": 0.34392625093460083,
+      "learning_rate": 9.867817115329055e-05,
+      "loss": 0.3727,
+      "step": 440
+    },
+    {
+      "epoch": 0.09313621964097149,
+      "grad_norm": 0.45021942257881165,
+      "learning_rate": 9.867041895042678e-05,
+      "loss": 0.3777,
+      "step": 441
+    },
+    {
+      "epoch": 0.09334741288278775,
+      "grad_norm": 0.4014663100242615,
+      "learning_rate": 9.866264438789573e-05,
+      "loss": 0.3754,
+      "step": 442
+    },
+    {
+      "epoch": 0.09355860612460401,
+      "grad_norm": 0.440481960773468,
+      "learning_rate": 9.865484746926914e-05,
+      "loss": 0.3803,
+      "step": 443
+    },
+    {
+      "epoch": 0.09376979936642027,
+      "grad_norm": 0.3959442675113678,
+      "learning_rate": 9.864702819812899e-05,
+      "loss": 0.3757,
+      "step": 444
+    },
+    {
+      "epoch": 0.09398099260823653,
+      "grad_norm": 0.8664902448654175,
+      "learning_rate": 9.863918657806752e-05,
+      "loss": 0.3789,
+      "step": 445
+    },
+    {
+      "epoch": 0.0941921858500528,
+      "grad_norm": 0.39863425493240356,
+      "learning_rate": 9.863132261268724e-05,
+      "loss": 0.3883,
+      "step": 446
+    },
+    {
+      "epoch": 0.09440337909186906,
+      "grad_norm": 0.4829177260398865,
+      "learning_rate": 9.862343630560094e-05,
+      "loss": 0.393,
+      "step": 447
+    },
+    {
+      "epoch": 0.09461457233368532,
+      "grad_norm": 0.5166580080986023,
+      "learning_rate": 9.861552766043167e-05,
+      "loss": 0.3793,
+      "step": 448
+    },
+    {
+      "epoch": 0.09482576557550158,
+      "grad_norm": 0.49370276927948,
+      "learning_rate": 9.860759668081274e-05,
+      "loss": 0.382,
+      "step": 449
+    },
+    {
+      "epoch": 0.09503695881731784,
+      "grad_norm": 0.4161822199821472,
+      "learning_rate": 9.85996433703877e-05,
+      "loss": 0.3798,
+      "step": 450
+    },
+    {
+      "epoch": 0.0952481520591341,
+      "grad_norm": 0.3794650733470917,
+      "learning_rate": 9.85916677328104e-05,
+      "loss": 0.3758,
+      "step": 451
+    },
+    {
+      "epoch": 0.09545934530095036,
+      "grad_norm": 1.3567155599594116,
+      "learning_rate": 9.858366977174493e-05,
+      "loss": 0.3845,
+      "step": 452
+    },
+    {
+      "epoch": 0.09567053854276664,
+      "grad_norm": 0.6081620454788208,
+      "learning_rate": 9.85756494908656e-05,
+      "loss": 0.3895,
+      "step": 453
+    },
+    {
+      "epoch": 0.0958817317845829,
+      "grad_norm": 0.6530495882034302,
+      "learning_rate": 9.856760689385705e-05,
+      "loss": 0.3899,
+      "step": 454
+    },
+    {
+      "epoch": 0.09609292502639916,
+      "grad_norm": 0.3539671003818512,
+      "learning_rate": 9.855954198441411e-05,
+      "loss": 0.3809,
+      "step": 455
+    },
+    {
+      "epoch": 0.09630411826821542,
+      "grad_norm": 0.7074968814849854,
+      "learning_rate": 9.855145476624186e-05,
+      "loss": 0.3984,
+      "step": 456
+    },
+    {
+      "epoch": 0.09651531151003168,
+      "grad_norm": 0.5208008289337158,
+      "learning_rate": 9.854334524305567e-05,
+      "loss": 0.3953,
+      "step": 457
+    },
+    {
+      "epoch": 0.09672650475184794,
+      "grad_norm": 0.5858681201934814,
+      "learning_rate": 9.85352134185811e-05,
+      "loss": 0.3839,
+      "step": 458
+    },
+    {
+      "epoch": 0.0969376979936642,
+      "grad_norm": 0.45296531915664673,
+      "learning_rate": 9.852705929655404e-05,
+      "loss": 0.3827,
+      "step": 459
+    },
+    {
+      "epoch": 0.09714889123548047,
+      "grad_norm": 0.4775009751319885,
+      "learning_rate": 9.851888288072053e-05,
+      "loss": 0.3808,
+      "step": 460
+    },
+    {
+      "epoch": 0.09736008447729673,
+      "grad_norm": 0.4343079924583435,
+      "learning_rate": 9.851068417483693e-05,
+      "loss": 0.3863,
+      "step": 461
+    },
+    {
+      "epoch": 0.09757127771911299,
+      "grad_norm": 0.4545423686504364,
+      "learning_rate": 9.850246318266979e-05,
+      "loss": 0.3814,
+      "step": 462
+    },
+    {
+      "epoch": 0.09778247096092925,
+      "grad_norm": 0.5713446140289307,
+      "learning_rate": 9.849421990799592e-05,
+      "loss": 0.3861,
+      "step": 463
+    },
+    {
+      "epoch": 0.09799366420274551,
+      "grad_norm": 0.3935050964355469,
+      "learning_rate": 9.848595435460233e-05,
+      "loss": 0.38,
+      "step": 464
+    },
+    {
+      "epoch": 0.09820485744456177,
+      "grad_norm": 0.4800340533256531,
+      "learning_rate": 9.847766652628634e-05,
+      "loss": 0.3888,
+      "step": 465
+    },
+    {
+      "epoch": 0.09841605068637804,
+      "grad_norm": 0.3701709508895874,
+      "learning_rate": 9.846935642685545e-05,
+      "loss": 0.3829,
+      "step": 466
+    },
+    {
+      "epoch": 0.0986272439281943,
+      "grad_norm": 0.49356237053871155,
+      "learning_rate": 9.846102406012738e-05,
+      "loss": 0.3743,
+      "step": 467
+    },
+    {
+      "epoch": 0.09883843717001056,
+      "grad_norm": 0.4767773449420929,
+      "learning_rate": 9.845266942993011e-05,
+      "loss": 0.3773,
+      "step": 468
+    },
+    {
+      "epoch": 0.09904963041182682,
+      "grad_norm": 0.4471038579940796,
+      "learning_rate": 9.844429254010185e-05,
+      "loss": 0.366,
+      "step": 469
+    },
+    {
+      "epoch": 0.09926082365364308,
+      "grad_norm": 0.43721315264701843,
+      "learning_rate": 9.843589339449102e-05,
+      "loss": 0.37,
+      "step": 470
+    },
+    {
+      "epoch": 0.09947201689545934,
+      "grad_norm": 0.36612990498542786,
+      "learning_rate": 9.842747199695626e-05,
+      "loss": 0.37,
+      "step": 471
+    },
+    {
+      "epoch": 0.0996832101372756,
+      "grad_norm": 0.40150582790374756,
+      "learning_rate": 9.841902835136646e-05,
+      "loss": 0.3736,
+      "step": 472
+    },
+    {
+      "epoch": 0.09989440337909188,
+      "grad_norm": 0.49380064010620117,
+      "learning_rate": 9.841056246160068e-05,
+      "loss": 0.374,
+      "step": 473
+    },
+    {
+      "epoch": 0.10010559662090814,
+      "grad_norm": 0.4522191882133484,
+      "learning_rate": 9.840207433154827e-05,
+      "loss": 0.3679,
+      "step": 474
+    },
+    {
+      "epoch": 0.1003167898627244,
+      "grad_norm": 0.34154659509658813,
+      "learning_rate": 9.839356396510875e-05,
+      "loss": 0.3704,
+      "step": 475
+    },
+    {
+      "epoch": 0.10052798310454066,
+      "grad_norm": 0.35681986808776855,
+      "learning_rate": 9.838503136619186e-05,
+      "loss": 0.3679,
+      "step": 476
+    },
+    {
+      "epoch": 0.10073917634635692,
+      "grad_norm": 0.3400050401687622,
+      "learning_rate": 9.837647653871755e-05,
+      "loss": 0.3612,
+      "step": 477
+    },
+    {
+      "epoch": 0.10095036958817317,
+      "grad_norm": 0.44964343309402466,
+      "learning_rate": 9.836789948661602e-05,
+      "loss": 0.3616,
+      "step": 478
+    },
+    {
+      "epoch": 0.10116156282998943,
+      "grad_norm": 0.41153883934020996,
+      "learning_rate": 9.835930021382762e-05,
+      "loss": 0.362,
+      "step": 479
+    },
+    {
+      "epoch": 0.10137275607180571,
+      "grad_norm": 0.35465046763420105,
+      "learning_rate": 9.835067872430298e-05,
+      "loss": 0.3642,
+      "step": 480
+    },
+    {
+      "epoch": 0.10158394931362197,
+      "grad_norm": 0.4018512964248657,
+      "learning_rate": 9.834203502200284e-05,
+      "loss": 0.3621,
+      "step": 481
+    },
+    {
+      "epoch": 0.10179514255543823,
+      "grad_norm": 0.47260621190071106,
+      "learning_rate": 9.833336911089825e-05,
+      "loss": 0.3659,
+      "step": 482
+    },
+    {
+      "epoch": 0.10200633579725449,
+      "grad_norm": 0.48547399044036865,
+      "learning_rate": 9.832468099497039e-05,
+      "loss": 0.366,
+      "step": 483
+    },
+    {
+      "epoch": 0.10221752903907075,
+      "grad_norm": 0.42939066886901855,
+      "learning_rate": 9.831597067821067e-05,
+      "loss": 0.3672,
+      "step": 484
+    },
+    {
+      "epoch": 0.10242872228088701,
+      "grad_norm": 0.38982000946998596,
+      "learning_rate": 9.830723816462072e-05,
+      "loss": 0.365,
+      "step": 485
+    },
+    {
+      "epoch": 0.10263991552270327,
+      "grad_norm": 0.4386598467826843,
+      "learning_rate": 9.82984834582123e-05,
+      "loss": 0.3706,
+      "step": 486
+    },
+    {
+      "epoch": 0.10285110876451954,
+      "grad_norm": 0.5117161273956299,
+      "learning_rate": 9.828970656300744e-05,
+      "loss": 0.3671,
+      "step": 487
+    },
+    {
+      "epoch": 0.1030623020063358,
+      "grad_norm": 0.7300944328308105,
+      "learning_rate": 9.828090748303831e-05,
+      "loss": 0.3715,
+      "step": 488
+    },
+    {
+      "epoch": 0.10327349524815206,
+      "grad_norm": 0.5145391821861267,
+      "learning_rate": 9.82720862223473e-05,
+      "loss": 0.3664,
+      "step": 489
+    },
+    {
+      "epoch": 0.10348468848996832,
+      "grad_norm": 0.47213923931121826,
+      "learning_rate": 9.8263242784987e-05,
+      "loss": 0.3679,
+      "step": 490
+    },
+    {
+      "epoch": 0.10369588173178458,
+      "grad_norm": 0.4861038625240326,
+      "learning_rate": 9.825437717502015e-05,
+      "loss": 0.3662,
+      "step": 491
+    },
+    {
+      "epoch": 0.10390707497360084,
+      "grad_norm": 0.46394339203834534,
+      "learning_rate": 9.824548939651971e-05,
+      "loss": 0.3703,
+      "step": 492
+    },
+    {
+      "epoch": 0.1041182682154171,
+      "grad_norm": 0.48191115260124207,
+      "learning_rate": 9.823657945356881e-05,
+      "loss": 0.3648,
+      "step": 493
+    },
+    {
+      "epoch": 0.10432946145723337,
+      "grad_norm": 0.5292214751243591,
+      "learning_rate": 9.822764735026075e-05,
+      "loss": 0.3623,
+      "step": 494
+    },
+    {
+      "epoch": 0.10454065469904963,
+      "grad_norm": 0.3605903685092926,
+      "learning_rate": 9.821869309069906e-05,
+      "loss": 0.3689,
+      "step": 495
+    },
+    {
+      "epoch": 0.1047518479408659,
+      "grad_norm": 0.4905593991279602,
+      "learning_rate": 9.820971667899739e-05,
+      "loss": 0.3657,
+      "step": 496
+    },
+    {
+      "epoch": 0.10496304118268215,
+      "grad_norm": 0.4230380654335022,
+      "learning_rate": 9.820071811927959e-05,
+      "loss": 0.3583,
+      "step": 497
+    },
+    {
+      "epoch": 0.10517423442449841,
+      "grad_norm": 0.35652440786361694,
+      "learning_rate": 9.81916974156797e-05,
+      "loss": 0.3659,
+      "step": 498
+    },
+    {
+      "epoch": 0.10538542766631467,
+      "grad_norm": 0.5630844831466675,
+      "learning_rate": 9.81826545723419e-05,
+      "loss": 0.3787,
+      "step": 499
+    },
+    {
+      "epoch": 0.10559662090813093,
+      "grad_norm": 0.37567421793937683,
+      "learning_rate": 9.817358959342058e-05,
+      "loss": 0.3594,
+      "step": 500
+    },
+    {
+      "epoch": 0.10580781414994721,
+      "grad_norm": 0.5392574667930603,
+      "learning_rate": 9.816450248308028e-05,
+      "loss": 0.3678,
+      "step": 501
+    },
+    {
+      "epoch": 0.10601900739176347,
+      "grad_norm": 0.3202340006828308,
+      "learning_rate": 9.81553932454957e-05,
+      "loss": 0.3635,
+      "step": 502
+    },
+    {
+      "epoch": 0.10623020063357973,
+      "grad_norm": 0.4233471751213074,
+      "learning_rate": 9.81462618848517e-05,
+      "loss": 0.3673,
+      "step": 503
+    },
+    {
+      "epoch": 0.10644139387539599,
+      "grad_norm": 0.433623343706131,
+      "learning_rate": 9.813710840534334e-05,
+      "loss": 0.3629,
+      "step": 504
+    },
+    {
+      "epoch": 0.10665258711721225,
+      "grad_norm": 0.4622725248336792,
+      "learning_rate": 9.81279328111758e-05,
+      "loss": 0.3645,
+      "step": 505
+    },
+    {
+      "epoch": 0.1068637803590285,
+      "grad_norm": 0.4192069172859192,
+      "learning_rate": 9.811873510656446e-05,
+      "loss": 0.3639,
+      "step": 506
+    },
+    {
+      "epoch": 0.10707497360084477,
+      "grad_norm": 0.43452927470207214,
+      "learning_rate": 9.810951529573481e-05,
+      "loss": 0.359,
+      "step": 507
+    },
+    {
+      "epoch": 0.10728616684266104,
+      "grad_norm": 0.38890939950942993,
+      "learning_rate": 9.810027338292254e-05,
+      "loss": 0.3606,
+      "step": 508
+    },
+    {
+      "epoch": 0.1074973600844773,
+      "grad_norm": 0.44490891695022583,
+      "learning_rate": 9.809100937237345e-05,
+      "loss": 0.3632,
+      "step": 509
+    },
+    {
+      "epoch": 0.10770855332629356,
+      "grad_norm": 0.4774106740951538,
+      "learning_rate": 9.808172326834356e-05,
+      "loss": 0.3665,
+      "step": 510
+    },
+    {
+      "epoch": 0.10791974656810982,
+      "grad_norm": 0.34790369868278503,
+      "learning_rate": 9.807241507509897e-05,
+      "loss": 0.3595,
+      "step": 511
+    },
+    {
+      "epoch": 0.10813093980992608,
+      "grad_norm": 0.3966074287891388,
+      "learning_rate": 9.806308479691595e-05,
+      "loss": 0.3605,
+      "step": 512
+    },
+    {
+      "epoch": 0.10834213305174234,
+      "grad_norm": 0.39080914855003357,
+      "learning_rate": 9.805373243808093e-05,
+      "loss": 0.3589,
+      "step": 513
+    },
+    {
+      "epoch": 0.1085533262935586,
+      "grad_norm": 0.5701823234558105,
+      "learning_rate": 9.804435800289047e-05,
+      "loss": 0.3655,
+      "step": 514
+    },
+    {
+      "epoch": 0.10876451953537487,
+      "grad_norm": 0.43490010499954224,
+      "learning_rate": 9.803496149565129e-05,
+      "loss": 0.359,
+      "step": 515
+    },
+    {
+      "epoch": 0.10897571277719113,
+      "grad_norm": 0.3392639458179474,
+      "learning_rate": 9.802554292068023e-05,
+      "loss": 0.3612,
+      "step": 516
+    },
+    {
+      "epoch": 0.1091869060190074,
+      "grad_norm": 0.41610240936279297,
+      "learning_rate": 9.801610228230427e-05,
+      "loss": 0.3547,
+      "step": 517
+    },
+    {
+      "epoch": 0.10939809926082365,
+      "grad_norm": 0.382011353969574,
+      "learning_rate": 9.800663958486053e-05,
+      "loss": 0.3572,
+      "step": 518
+    },
+    {
+      "epoch": 0.10960929250263991,
+      "grad_norm": 0.4388468861579895,
+      "learning_rate": 9.79971548326963e-05,
+      "loss": 0.3585,
+      "step": 519
+    },
+    {
+      "epoch": 0.10982048574445617,
+      "grad_norm": 0.5489420890808105,
+      "learning_rate": 9.798764803016891e-05,
+      "loss": 0.3587,
+      "step": 520
+    },
+    {
+      "epoch": 0.11003167898627243,
+      "grad_norm": 0.3725614547729492,
+      "learning_rate": 9.797811918164592e-05,
+      "loss": 0.3581,
+      "step": 521
+    },
+    {
+      "epoch": 0.1102428722280887,
+      "grad_norm": 0.47695085406303406,
+      "learning_rate": 9.796856829150495e-05,
+      "loss": 0.3594,
+      "step": 522
+    },
+    {
+      "epoch": 0.11045406546990497,
+      "grad_norm": 0.6474586129188538,
+      "learning_rate": 9.795899536413382e-05,
+      "loss": 0.3609,
+      "step": 523
+    },
+    {
+      "epoch": 0.11066525871172123,
+      "grad_norm": 0.3713063597679138,
+      "learning_rate": 9.794940040393037e-05,
+      "loss": 0.3558,
+      "step": 524
+    },
+    {
+      "epoch": 0.11087645195353749,
+      "grad_norm": 0.42622917890548706,
+      "learning_rate": 9.793978341530265e-05,
+      "loss": 0.364,
+      "step": 525
+    },
+    {
+      "epoch": 0.11108764519535375,
+      "grad_norm": 0.5361518263816833,
+      "learning_rate": 9.793014440266878e-05,
+      "loss": 0.3601,
+      "step": 526
+    },
+    {
+      "epoch": 0.11129883843717,
+      "grad_norm": 0.5129774808883667,
+      "learning_rate": 9.792048337045705e-05,
+      "loss": 0.3636,
+      "step": 527
+    },
+    {
+      "epoch": 0.11151003167898627,
+      "grad_norm": 0.5159335136413574,
+      "learning_rate": 9.79108003231058e-05,
+      "loss": 0.3652,
+      "step": 528
+    },
+    {
+      "epoch": 0.11172122492080254,
+      "grad_norm": 0.32603776454925537,
+      "learning_rate": 9.790109526506353e-05,
+      "loss": 0.3628,
+      "step": 529
+    },
+    {
+      "epoch": 0.1119324181626188,
+      "grad_norm": 0.422537624835968,
+      "learning_rate": 9.789136820078883e-05,
+      "loss": 0.3655,
+      "step": 530
+    },
+    {
+      "epoch": 0.11214361140443506,
+      "grad_norm": 0.5601354837417603,
+      "learning_rate": 9.788161913475043e-05,
+      "loss": 0.3638,
+      "step": 531
+    },
+    {
+      "epoch": 0.11235480464625132,
+      "grad_norm": 0.37436768412590027,
+      "learning_rate": 9.787184807142713e-05,
+      "loss": 0.3571,
+      "step": 532
+    },
+    {
+      "epoch": 0.11256599788806758,
+      "grad_norm": 0.34980684518814087,
+      "learning_rate": 9.786205501530785e-05,
+      "loss": 0.3572,
+      "step": 533
+    },
+    {
+      "epoch": 0.11277719112988384,
+      "grad_norm": 0.46807289123535156,
+      "learning_rate": 9.785223997089161e-05,
+      "loss": 0.357,
+      "step": 534
+    },
+    {
+      "epoch": 0.11298838437170011,
+      "grad_norm": 0.4457637667655945,
+      "learning_rate": 9.784240294268756e-05,
+      "loss": 0.3592,
+      "step": 535
+    },
+    {
+      "epoch": 0.11319957761351637,
+      "grad_norm": 0.34789973497390747,
+      "learning_rate": 9.783254393521491e-05,
+      "loss": 0.3604,
+      "step": 536
+    },
+    {
+      "epoch": 0.11341077085533263,
+      "grad_norm": 0.48582032322883606,
+      "learning_rate": 9.7822662953003e-05,
+      "loss": 0.3577,
+      "step": 537
+    },
+    {
+      "epoch": 0.11362196409714889,
+      "grad_norm": 0.4699117839336395,
+      "learning_rate": 9.781276000059123e-05,
+      "loss": 0.3533,
+      "step": 538
+    },
+    {
+      "epoch": 0.11383315733896515,
+      "grad_norm": 0.4898450970649719,
+      "learning_rate": 9.780283508252912e-05,
+      "loss": 0.3522,
+      "step": 539
+    },
+    {
+      "epoch": 0.11404435058078141,
+      "grad_norm": 0.4105646312236786,
+      "learning_rate": 9.779288820337629e-05,
+      "loss": 0.3531,
+      "step": 540
+    },
+    {
+      "epoch": 0.11425554382259767,
+      "grad_norm": 0.38872838020324707,
+      "learning_rate": 9.77829193677024e-05,
+      "loss": 0.3586,
+      "step": 541
+    },
+    {
+      "epoch": 0.11446673706441395,
+      "grad_norm": 0.4091218411922455,
+      "learning_rate": 9.777292858008726e-05,
+      "loss": 0.3564,
+      "step": 542
+    },
+    {
+      "epoch": 0.1146779303062302,
+      "grad_norm": 0.375906378030777,
+      "learning_rate": 9.776291584512073e-05,
+      "loss": 0.355,
+      "step": 543
+    },
+    {
+      "epoch": 0.11488912354804647,
+      "grad_norm": 0.44066986441612244,
+      "learning_rate": 9.775288116740278e-05,
+      "loss": 0.3511,
+      "step": 544
+    },
+    {
+      "epoch": 0.11510031678986272,
+      "grad_norm": 0.6004911065101624,
+      "learning_rate": 9.774282455154338e-05,
+      "loss": 0.3612,
+      "step": 545
+    },
+    {
+      "epoch": 0.11531151003167898,
+      "grad_norm": 0.45257073640823364,
+      "learning_rate": 9.773274600216269e-05,
+      "loss": 0.3514,
+      "step": 546
+    },
+    {
+      "epoch": 0.11552270327349524,
+      "grad_norm": 0.46931493282318115,
+      "learning_rate": 9.772264552389088e-05,
+      "loss": 0.3513,
+      "step": 547
+    },
+    {
+      "epoch": 0.1157338965153115,
+      "grad_norm": 0.4743399918079376,
+      "learning_rate": 9.771252312136819e-05,
+      "loss": 0.3531,
+      "step": 548
+    },
+    {
+      "epoch": 0.11594508975712778,
+      "grad_norm": 0.5340331792831421,
+      "learning_rate": 9.770237879924498e-05,
+      "loss": 0.3527,
+      "step": 549
+    },
+    {
+      "epoch": 0.11615628299894404,
+      "grad_norm": 0.36304306983947754,
+      "learning_rate": 9.769221256218164e-05,
+      "loss": 0.3582,
+      "step": 550
+    },
+    {
+      "epoch": 0.1163674762407603,
+      "grad_norm": 0.3978298604488373,
+      "learning_rate": 9.768202441484863e-05,
+      "loss": 0.3572,
+      "step": 551
+    },
+    {
+      "epoch": 0.11657866948257656,
+      "grad_norm": 0.4838181734085083,
+      "learning_rate": 9.767181436192649e-05,
+      "loss": 0.3574,
+      "step": 552
+    },
+    {
+      "epoch": 0.11678986272439282,
+      "grad_norm": 0.45764172077178955,
+      "learning_rate": 9.766158240810583e-05,
+      "loss": 0.3584,
+      "step": 553
+    },
+    {
+      "epoch": 0.11700105596620908,
+      "grad_norm": 0.4697604179382324,
+      "learning_rate": 9.765132855808729e-05,
+      "loss": 0.3563,
+      "step": 554
+    },
+    {
+      "epoch": 0.11721224920802534,
+      "grad_norm": 0.5870717763900757,
+      "learning_rate": 9.76410528165816e-05,
+      "loss": 0.3534,
+      "step": 555
+    },
+    {
+      "epoch": 0.11742344244984161,
+      "grad_norm": 0.4838230609893799,
+      "learning_rate": 9.763075518830955e-05,
+      "loss": 0.3532,
+      "step": 556
+    },
+    {
+      "epoch": 0.11763463569165787,
+      "grad_norm": 0.3610585629940033,
+      "learning_rate": 9.762043567800194e-05,
+      "loss": 0.3475,
+      "step": 557
+    },
+    {
+      "epoch": 0.11784582893347413,
+      "grad_norm": 0.39439016580581665,
+      "learning_rate": 9.761009429039968e-05,
+      "loss": 0.3457,
+      "step": 558
+    },
+    {
+      "epoch": 0.11805702217529039,
+      "grad_norm": 0.413786917924881,
+      "learning_rate": 9.759973103025368e-05,
+      "loss": 0.3507,
+      "step": 559
+    },
+    {
+      "epoch": 0.11826821541710665,
+      "grad_norm": 0.5340571999549866,
+      "learning_rate": 9.758934590232496e-05,
+      "loss": 0.359,
+      "step": 560
+    },
+    {
+      "epoch": 0.11847940865892291,
+      "grad_norm": 0.4098823666572571,
+      "learning_rate": 9.757893891138452e-05,
+      "loss": 0.3572,
+      "step": 561
+    },
+    {
+      "epoch": 0.11869060190073917,
+      "grad_norm": 0.43419209122657776,
+      "learning_rate": 9.756851006221344e-05,
+      "loss": 0.3576,
+      "step": 562
+    },
+    {
+      "epoch": 0.11890179514255544,
+      "grad_norm": 0.6430238485336304,
+      "learning_rate": 9.755805935960283e-05,
+      "loss": 0.3536,
+      "step": 563
+    },
+    {
+      "epoch": 0.1191129883843717,
+      "grad_norm": 0.4067290425300598,
+      "learning_rate": 9.754758680835387e-05,
+      "loss": 0.3532,
+      "step": 564
+    },
+    {
+      "epoch": 0.11932418162618796,
+      "grad_norm": 0.4863293468952179,
+      "learning_rate": 9.753709241327773e-05,
+      "loss": 0.3516,
+      "step": 565
+    },
+    {
+      "epoch": 0.11953537486800422,
+      "grad_norm": 0.4200882911682129,
+      "learning_rate": 9.752657617919562e-05,
+      "loss": 0.3548,
+      "step": 566
+    },
+    {
+      "epoch": 0.11974656810982048,
+      "grad_norm": 0.4104411005973816,
+      "learning_rate": 9.751603811093886e-05,
+      "loss": 0.3529,
+      "step": 567
+    },
+    {
+      "epoch": 0.11995776135163674,
+      "grad_norm": 0.4035990238189697,
+      "learning_rate": 9.750547821334868e-05,
+      "loss": 0.3519,
+      "step": 568
+    },
+    {
+      "epoch": 0.120168954593453,
+      "grad_norm": 0.4402064085006714,
+      "learning_rate": 9.749489649127644e-05,
+      "loss": 0.3444,
+      "step": 569
+    },
+    {
+      "epoch": 0.12038014783526928,
+      "grad_norm": 0.4356549382209778,
+      "learning_rate": 9.748429294958346e-05,
+      "loss": 0.3529,
+      "step": 570
+    },
+    {
+      "epoch": 0.12059134107708554,
+      "grad_norm": 0.3548963963985443,
+      "learning_rate": 9.747366759314113e-05,
+      "loss": 0.3537,
+      "step": 571
+    },
+    {
+      "epoch": 0.1208025343189018,
+      "grad_norm": 0.4516805410385132,
+      "learning_rate": 9.746302042683084e-05,
+      "loss": 0.3557,
+      "step": 572
+    },
+    {
+      "epoch": 0.12101372756071806,
+      "grad_norm": 0.48298823833465576,
+      "learning_rate": 9.7452351455544e-05,
+      "loss": 0.3484,
+      "step": 573
+    },
+    {
+      "epoch": 0.12122492080253432,
+      "grad_norm": 0.3595397472381592,
+      "learning_rate": 9.744166068418204e-05,
+      "loss": 0.3466,
+      "step": 574
+    },
+    {
+      "epoch": 0.12143611404435058,
+      "grad_norm": 0.44566258788108826,
+      "learning_rate": 9.74309481176564e-05,
+      "loss": 0.3572,
+      "step": 575
+    },
+    {
+      "epoch": 0.12164730728616684,
+      "grad_norm": 0.38086140155792236,
+      "learning_rate": 9.742021376088855e-05,
+      "loss": 0.3475,
+      "step": 576
+    },
+    {
+      "epoch": 0.12185850052798311,
+      "grad_norm": 0.4275955557823181,
+      "learning_rate": 9.740945761880993e-05,
+      "loss": 0.3477,
+      "step": 577
+    },
+    {
+      "epoch": 0.12206969376979937,
+      "grad_norm": 0.46991103887557983,
+      "learning_rate": 9.739867969636205e-05,
+      "loss": 0.3403,
+      "step": 578
+    },
+    {
+      "epoch": 0.12228088701161563,
+      "grad_norm": 0.4576331377029419,
+      "learning_rate": 9.738787999849639e-05,
+      "loss": 0.3442,
+      "step": 579
+    },
+    {
+      "epoch": 0.12249208025343189,
+      "grad_norm": 0.43520596623420715,
+      "learning_rate": 9.737705853017441e-05,
+      "loss": 0.3487,
+      "step": 580
+    },
+    {
+      "epoch": 0.12270327349524815,
+      "grad_norm": 0.4478459358215332,
+      "learning_rate": 9.736621529636764e-05,
+      "loss": 0.3513,
+      "step": 581
+    },
+    {
+      "epoch": 0.12291446673706441,
+      "grad_norm": 0.490109920501709,
+      "learning_rate": 9.735535030205752e-05,
+      "loss": 0.3514,
+      "step": 582
+    },
+    {
+      "epoch": 0.12312565997888067,
+      "grad_norm": 0.5468922853469849,
+      "learning_rate": 9.734446355223557e-05,
+      "loss": 0.353,
+      "step": 583
+    },
+    {
+      "epoch": 0.12333685322069694,
+      "grad_norm": 0.501846969127655,
+      "learning_rate": 9.733355505190327e-05,
+      "loss": 0.3489,
+      "step": 584
+    },
+    {
+      "epoch": 0.1235480464625132,
+      "grad_norm": 0.447126179933548,
+      "learning_rate": 9.732262480607208e-05,
+      "loss": 0.3452,
+      "step": 585
+    },
+    {
+      "epoch": 0.12375923970432946,
+      "grad_norm": 0.44398441910743713,
+      "learning_rate": 9.731167281976345e-05,
+      "loss": 0.3445,
+      "step": 586
+    },
+    {
+      "epoch": 0.12397043294614572,
+      "grad_norm": 0.4827435612678528,
+      "learning_rate": 9.730069909800885e-05,
+      "loss": 0.345,
+      "step": 587
+    },
+    {
+      "epoch": 0.12418162618796198,
+      "grad_norm": 0.4498518407344818,
+      "learning_rate": 9.72897036458497e-05,
+      "loss": 0.3431,
+      "step": 588
+    },
+    {
+      "epoch": 0.12439281942977824,
+      "grad_norm": 0.36961629986763,
+      "learning_rate": 9.727868646833741e-05,
+      "loss": 0.3413,
+      "step": 589
+    },
+    {
+      "epoch": 0.1246040126715945,
+      "grad_norm": 0.3624129295349121,
+      "learning_rate": 9.726764757053344e-05,
+      "loss": 0.347,
+      "step": 590
+    },
+    {
+      "epoch": 0.12481520591341078,
+      "grad_norm": 0.36161142587661743,
+      "learning_rate": 9.725658695750909e-05,
+      "loss": 0.3486,
+      "step": 591
+    },
+    {
+      "epoch": 0.12502639915522704,
+      "grad_norm": 0.39048582315444946,
+      "learning_rate": 9.724550463434577e-05,
+      "loss": 0.3479,
+      "step": 592
+    },
+    {
+      "epoch": 0.1252375923970433,
+      "grad_norm": 0.4342297911643982,
+      "learning_rate": 9.723440060613476e-05,
+      "loss": 0.3506,
+      "step": 593
+    },
+    {
+      "epoch": 0.12544878563885956,
+      "grad_norm": 0.4358193576335907,
+      "learning_rate": 9.72232748779774e-05,
+      "loss": 0.3583,
+      "step": 594
+    },
+    {
+      "epoch": 0.12565997888067582,
+      "grad_norm": 0.37452349066734314,
+      "learning_rate": 9.721212745498494e-05,
+      "loss": 0.3541,
+      "step": 595
+    },
+    {
+      "epoch": 0.12587117212249208,
+      "grad_norm": 0.38477423787117004,
+      "learning_rate": 9.720095834227861e-05,
+      "loss": 0.3497,
+      "step": 596
+    },
+    {
+      "epoch": 0.12608236536430834,
+      "grad_norm": 0.5149921178817749,
+      "learning_rate": 9.718976754498962e-05,
+      "loss": 0.3471,
+      "step": 597
+    },
+    {
+      "epoch": 0.1262935586061246,
+      "grad_norm": 0.4617641568183899,
+      "learning_rate": 9.717855506825915e-05,
+      "loss": 0.3449,
+      "step": 598
+    },
+    {
+      "epoch": 0.12650475184794086,
+      "grad_norm": 0.6561935544013977,
+      "learning_rate": 9.716732091723829e-05,
+      "loss": 0.3467,
+      "step": 599
+    },
+    {
+      "epoch": 0.12671594508975711,
+      "grad_norm": 0.35332342982292175,
+      "learning_rate": 9.715606509708812e-05,
+      "loss": 0.3485,
+      "step": 600
+    },
+    {
+      "epoch": 0.12671594508975711,
+      "eval_train_split_loss": 0.41040199995040894,
+      "eval_train_split_runtime": 6.5314,
+      "eval_train_split_samples_per_second": 1.531,
+      "eval_train_split_steps_per_second": 0.153,
+      "step": 600
+    },
+    {
+      "epoch": 0.1269271383315734,
+      "grad_norm": 0.491060733795166,
+      "learning_rate": 9.71447876129797e-05,
+      "loss": 0.3547,
+      "step": 601
+    },
+    {
+      "epoch": 0.12713833157338966,
+      "grad_norm": 0.42359644174575806,
+      "learning_rate": 9.7133488470094e-05,
+      "loss": 0.3579,
+      "step": 602
+    },
+    {
+      "epoch": 0.12734952481520592,
+      "grad_norm": 0.5692341923713684,
+      "learning_rate": 9.712216767362194e-05,
+      "loss": 0.3496,
+      "step": 603
+    },
+    {
+      "epoch": 0.12756071805702218,
+      "grad_norm": 0.6686945557594299,
+      "learning_rate": 9.711082522876444e-05,
+      "loss": 0.3533,
+      "step": 604
+    },
+    {
+      "epoch": 0.12777191129883844,
+      "grad_norm": 0.48921629786491394,
+      "learning_rate": 9.709946114073232e-05,
+      "loss": 0.3526,
+      "step": 605
+    },
+    {
+      "epoch": 0.1279831045406547,
+      "grad_norm": 0.5002605319023132,
+      "learning_rate": 9.708807541474632e-05,
+      "loss": 0.3515,
+      "step": 606
+    },
+    {
+      "epoch": 0.12819429778247096,
+      "grad_norm": 0.4124280512332916,
+      "learning_rate": 9.70766680560372e-05,
+      "loss": 0.3594,
+      "step": 607
+    },
+    {
+      "epoch": 0.12840549102428722,
+      "grad_norm": 0.4404142498970032,
+      "learning_rate": 9.706523906984559e-05,
+      "loss": 0.3483,
+      "step": 608
+    },
+    {
+      "epoch": 0.12861668426610348,
+      "grad_norm": 0.705427885055542,
+      "learning_rate": 9.705378846142208e-05,
+      "loss": 0.353,
+      "step": 609
+    },
+    {
+      "epoch": 0.12882787750791974,
+      "grad_norm": 0.3854082226753235,
+      "learning_rate": 9.70423162360272e-05,
+      "loss": 0.3443,
+      "step": 610
+    },
+    {
+      "epoch": 0.129039070749736,
+      "grad_norm": 0.5718660950660706,
+      "learning_rate": 9.70308223989314e-05,
+      "loss": 0.3591,
+      "step": 611
+    },
+    {
+      "epoch": 0.12925026399155226,
+      "grad_norm": 0.3569101393222809,
+      "learning_rate": 9.701930695541504e-05,
+      "loss": 0.3521,
+      "step": 612
+    },
+    {
+      "epoch": 0.12946145723336852,
+      "grad_norm": 0.44557350873947144,
+      "learning_rate": 9.700776991076845e-05,
+      "loss": 0.3548,
+      "step": 613
+    },
+    {
+      "epoch": 0.12967265047518478,
+      "grad_norm": 0.41888096928596497,
+      "learning_rate": 9.699621127029186e-05,
+      "loss": 0.3508,
+      "step": 614
+    },
+    {
+      "epoch": 0.12988384371700107,
+      "grad_norm": 0.3901000916957855,
+      "learning_rate": 9.698463103929542e-05,
+      "loss": 0.3454,
+      "step": 615
+    },
+    {
+      "epoch": 0.13009503695881733,
+      "grad_norm": 0.4447787404060364,
+      "learning_rate": 9.697302922309921e-05,
+      "loss": 0.3506,
+      "step": 616
+    },
+    {
+      "epoch": 0.1303062302006336,
+      "grad_norm": 0.37344270944595337,
+      "learning_rate": 9.696140582703319e-05,
+      "loss": 0.3426,
+      "step": 617
+    },
+    {
+      "epoch": 0.13051742344244985,
+      "grad_norm": 0.43621841073036194,
+      "learning_rate": 9.694976085643728e-05,
+      "loss": 0.3507,
+      "step": 618
+    },
+    {
+      "epoch": 0.1307286166842661,
+      "grad_norm": 0.42425382137298584,
+      "learning_rate": 9.69380943166613e-05,
+      "loss": 0.3411,
+      "step": 619
+    },
+    {
+      "epoch": 0.13093980992608237,
+      "grad_norm": 0.44694778323173523,
+      "learning_rate": 9.692640621306496e-05,
+      "loss": 0.3451,
+      "step": 620
+    },
+    {
+      "epoch": 0.13115100316789863,
+      "grad_norm": 0.42875832319259644,
+      "learning_rate": 9.691469655101793e-05,
+      "loss": 0.342,
+      "step": 621
+    },
+    {
+      "epoch": 0.1313621964097149,
+      "grad_norm": 0.39556992053985596,
+      "learning_rate": 9.690296533589968e-05,
+      "loss": 0.3511,
+      "step": 622
+    },
+    {
+      "epoch": 0.13157338965153115,
+      "grad_norm": 0.4137139916419983,
+      "learning_rate": 9.689121257309967e-05,
+      "loss": 0.3527,
+      "step": 623
+    },
+    {
+      "epoch": 0.1317845828933474,
+      "grad_norm": 0.5551238656044006,
+      "learning_rate": 9.687943826801725e-05,
+      "loss": 0.3408,
+      "step": 624
+    },
+    {
+      "epoch": 0.13199577613516367,
+      "grad_norm": 0.406699001789093,
+      "learning_rate": 9.686764242606163e-05,
+      "loss": 0.338,
+      "step": 625
+    },
+    {
+      "epoch": 0.13220696937697993,
+      "grad_norm": 0.6248592734336853,
+      "learning_rate": 9.685582505265196e-05,
+      "loss": 0.3472,
+      "step": 626
+    },
+    {
+      "epoch": 0.1324181626187962,
+      "grad_norm": 0.5051349997520447,
+      "learning_rate": 9.684398615321724e-05,
+      "loss": 0.3451,
+      "step": 627
+    },
+    {
+      "epoch": 0.13262935586061245,
+      "grad_norm": 0.47564437985420227,
+      "learning_rate": 9.683212573319637e-05,
+      "loss": 0.3464,
+      "step": 628
+    },
+    {
+      "epoch": 0.13284054910242873,
+      "grad_norm": 0.4299936890602112,
+      "learning_rate": 9.682024379803815e-05,
+      "loss": 0.3512,
+      "step": 629
+    },
+    {
+      "epoch": 0.133051742344245,
+      "grad_norm": 0.3649164140224457,
+      "learning_rate": 9.680834035320127e-05,
+      "loss": 0.3493,
+      "step": 630
+    },
+    {
+      "epoch": 0.13326293558606125,
+      "grad_norm": 0.4412953853607178,
+      "learning_rate": 9.679641540415427e-05,
+      "loss": 0.3501,
+      "step": 631
+    },
+    {
+      "epoch": 0.13347412882787751,
+      "grad_norm": 0.5034704804420471,
+      "learning_rate": 9.67844689563756e-05,
+      "loss": 0.3494,
+      "step": 632
+    },
+    {
+      "epoch": 0.13368532206969377,
+      "grad_norm": 0.393743097782135,
+      "learning_rate": 9.677250101535358e-05,
+      "loss": 0.3451,
+      "step": 633
+    },
+    {
+      "epoch": 0.13389651531151003,
+      "grad_norm": 0.38411155343055725,
+      "learning_rate": 9.676051158658638e-05,
+      "loss": 0.3408,
+      "step": 634
+    },
+    {
+      "epoch": 0.1341077085533263,
+      "grad_norm": 0.4106031656265259,
+      "learning_rate": 9.674850067558208e-05,
+      "loss": 0.3382,
+      "step": 635
+    },
+    {
+      "epoch": 0.13431890179514255,
+      "grad_norm": 0.4469754099845886,
+      "learning_rate": 9.67364682878586e-05,
+      "loss": 0.3406,
+      "step": 636
+    },
+    {
+      "epoch": 0.1345300950369588,
+      "grad_norm": 0.43816012144088745,
+      "learning_rate": 9.672441442894374e-05,
+      "loss": 0.3449,
+      "step": 637
+    },
+    {
+      "epoch": 0.13474128827877507,
+      "grad_norm": 0.3395436108112335,
+      "learning_rate": 9.671233910437515e-05,
+      "loss": 0.3394,
+      "step": 638
+    },
+    {
+      "epoch": 0.13495248152059133,
+      "grad_norm": 0.3956664502620697,
+      "learning_rate": 9.670024231970038e-05,
+      "loss": 0.3442,
+      "step": 639
+    },
+    {
+      "epoch": 0.1351636747624076,
+      "grad_norm": 0.48493459820747375,
+      "learning_rate": 9.668812408047679e-05,
+      "loss": 0.3494,
+      "step": 640
+    },
+    {
+      "epoch": 0.13537486800422385,
+      "grad_norm": 0.3504188656806946,
+      "learning_rate": 9.667598439227161e-05,
+      "loss": 0.3464,
+      "step": 641
+    },
+    {
+      "epoch": 0.13558606124604014,
+      "grad_norm": 0.36579054594039917,
+      "learning_rate": 9.666382326066195e-05,
+      "loss": 0.342,
+      "step": 642
+    },
+    {
+      "epoch": 0.1357972544878564,
+      "grad_norm": 0.4615013301372528,
+      "learning_rate": 9.665164069123475e-05,
+      "loss": 0.3433,
+      "step": 643
+    },
+    {
+      "epoch": 0.13600844772967266,
+      "grad_norm": 0.5685523748397827,
+      "learning_rate": 9.663943668958679e-05,
+      "loss": 0.3455,
+      "step": 644
+    },
+    {
+      "epoch": 0.13621964097148892,
+      "grad_norm": 0.37745457887649536,
+      "learning_rate": 9.662721126132473e-05,
+      "loss": 0.3503,
+      "step": 645
+    },
+    {
+      "epoch": 0.13643083421330518,
+      "grad_norm": 0.4073110520839691,
+      "learning_rate": 9.661496441206503e-05,
+      "loss": 0.3525,
+      "step": 646
+    },
+    {
+      "epoch": 0.13664202745512144,
+      "grad_norm": 0.44020023941993713,
+      "learning_rate": 9.660269614743402e-05,
+      "loss": 0.3337,
+      "step": 647
+    },
+    {
+      "epoch": 0.1368532206969377,
+      "grad_norm": 0.4848185181617737,
+      "learning_rate": 9.659040647306785e-05,
+      "loss": 0.3441,
+      "step": 648
+    },
+    {
+      "epoch": 0.13706441393875396,
+      "grad_norm": 0.5470314621925354,
+      "learning_rate": 9.657809539461256e-05,
+      "loss": 0.3412,
+      "step": 649
+    },
+    {
+      "epoch": 0.13727560718057022,
+      "grad_norm": 0.3907366096973419,
+      "learning_rate": 9.656576291772392e-05,
+      "loss": 0.3421,
+      "step": 650
+    },
+    {
+      "epoch": 0.13748680042238648,
+      "grad_norm": 0.5687809586524963,
+      "learning_rate": 9.655340904806762e-05,
+      "loss": 0.3401,
+      "step": 651
+    },
+    {
+      "epoch": 0.13769799366420274,
+      "grad_norm": 0.39392709732055664,
+      "learning_rate": 9.654103379131916e-05,
+      "loss": 0.3368,
+      "step": 652
+    },
+    {
+      "epoch": 0.137909186906019,
+      "grad_norm": 0.4936131238937378,
+      "learning_rate": 9.652863715316384e-05,
+      "loss": 0.35,
+      "step": 653
+    },
+    {
+      "epoch": 0.13812038014783526,
+      "grad_norm": 0.3379993438720703,
+      "learning_rate": 9.651621913929678e-05,
+      "loss": 0.3358,
+      "step": 654
+    },
+    {
+      "epoch": 0.13833157338965152,
+      "grad_norm": 0.5267112255096436,
+      "learning_rate": 9.650377975542297e-05,
+      "loss": 0.3485,
+      "step": 655
+    },
+    {
+      "epoch": 0.1385427666314678,
+      "grad_norm": 0.47686767578125,
+      "learning_rate": 9.649131900725718e-05,
+      "loss": 0.3408,
+      "step": 656
+    },
+    {
+      "epoch": 0.13875395987328407,
+      "grad_norm": 0.5162385702133179,
+      "learning_rate": 9.647883690052399e-05,
+      "loss": 0.348,
+      "step": 657
+    },
+    {
+      "epoch": 0.13896515311510033,
+      "grad_norm": 0.3816085755825043,
+      "learning_rate": 9.64663334409578e-05,
+      "loss": 0.3387,
+      "step": 658
+    },
+    {
+      "epoch": 0.13917634635691659,
+      "grad_norm": 0.5173656344413757,
+      "learning_rate": 9.645380863430281e-05,
+      "loss": 0.3504,
+      "step": 659
+    },
+    {
+      "epoch": 0.13938753959873285,
+      "grad_norm": 0.3803335726261139,
+      "learning_rate": 9.644126248631306e-05,
+      "loss": 0.3487,
+      "step": 660
+    },
+    {
+      "epoch": 0.1395987328405491,
+      "grad_norm": 0.48671549558639526,
+      "learning_rate": 9.642869500275238e-05,
+      "loss": 0.3614,
+      "step": 661
+    },
+    {
+      "epoch": 0.13980992608236537,
+      "grad_norm": 0.470328688621521,
+      "learning_rate": 9.641610618939436e-05,
+      "loss": 0.3422,
+      "step": 662
+    },
+    {
+      "epoch": 0.14002111932418163,
+      "grad_norm": 0.6042503714561462,
+      "learning_rate": 9.640349605202246e-05,
+      "loss": 0.3476,
+      "step": 663
+    },
+    {
+      "epoch": 0.14023231256599789,
+      "grad_norm": 0.5421839952468872,
+      "learning_rate": 9.639086459642988e-05,
+      "loss": 0.3429,
+      "step": 664
+    },
+    {
+      "epoch": 0.14044350580781415,
+      "grad_norm": 0.3998701870441437,
+      "learning_rate": 9.637821182841965e-05,
+      "loss": 0.3361,
+      "step": 665
+    },
+    {
+      "epoch": 0.1406546990496304,
+      "grad_norm": 0.5468639135360718,
+      "learning_rate": 9.636553775380457e-05,
+      "loss": 0.3365,
+      "step": 666
+    },
+    {
+      "epoch": 0.14086589229144666,
+      "grad_norm": 0.5405282378196716,
+      "learning_rate": 9.635284237840721e-05,
+      "loss": 0.3444,
+      "step": 667
+    },
+    {
+      "epoch": 0.14107708553326292,
+      "grad_norm": 0.6655111312866211,
+      "learning_rate": 9.634012570805999e-05,
+      "loss": 0.3395,
+      "step": 668
+    },
+    {
+      "epoch": 0.14128827877507918,
+      "grad_norm": 0.46089744567871094,
+      "learning_rate": 9.632738774860504e-05,
+      "loss": 0.3383,
+      "step": 669
+    },
+    {
+      "epoch": 0.14149947201689547,
+      "grad_norm": 0.443733811378479,
+      "learning_rate": 9.631462850589433e-05,
+      "loss": 0.3384,
+      "step": 670
+    },
+    {
+      "epoch": 0.14171066525871173,
+      "grad_norm": 0.5310661792755127,
+      "learning_rate": 9.630184798578956e-05,
+      "loss": 0.3306,
+      "step": 671
+    },
+    {
+      "epoch": 0.141921858500528,
+      "grad_norm": 0.4663335680961609,
+      "learning_rate": 9.628904619416224e-05,
+      "loss": 0.3356,
+      "step": 672
+    },
+    {
+      "epoch": 0.14213305174234425,
+      "grad_norm": 0.43902599811553955,
+      "learning_rate": 9.627622313689362e-05,
+      "loss": 0.3382,
+      "step": 673
+    },
+    {
+      "epoch": 0.1423442449841605,
+      "grad_norm": 0.42671558260917664,
+      "learning_rate": 9.626337881987474e-05,
+      "loss": 0.3376,
+      "step": 674
+    },
+    {
+      "epoch": 0.14255543822597677,
+      "grad_norm": 0.5909038186073303,
+      "learning_rate": 9.625051324900644e-05,
+      "loss": 0.3471,
+      "step": 675
+    },
+    {
+      "epoch": 0.14276663146779303,
+      "grad_norm": 0.4703707695007324,
+      "learning_rate": 9.623762643019926e-05,
+      "loss": 0.3325,
+      "step": 676
+    },
+    {
+      "epoch": 0.1429778247096093,
+      "grad_norm": 0.3852248787879944,
+      "learning_rate": 9.622471836937354e-05,
+      "loss": 0.3308,
+      "step": 677
+    },
+    {
+      "epoch": 0.14318901795142555,
+      "grad_norm": 0.5060172080993652,
+      "learning_rate": 9.621178907245935e-05,
+      "loss": 0.3339,
+      "step": 678
+    },
+    {
+      "epoch": 0.1434002111932418,
+      "grad_norm": 0.42436233162879944,
+      "learning_rate": 9.619883854539656e-05,
+      "loss": 0.3296,
+      "step": 679
+    },
+    {
+      "epoch": 0.14361140443505807,
+      "grad_norm": 0.39677804708480835,
+      "learning_rate": 9.618586679413477e-05,
+      "loss": 0.3286,
+      "step": 680
+    },
+    {
+      "epoch": 0.14382259767687433,
+      "grad_norm": 0.41080427169799805,
+      "learning_rate": 9.617287382463331e-05,
+      "loss": 0.3382,
+      "step": 681
+    },
+    {
+      "epoch": 0.1440337909186906,
+      "grad_norm": 0.3663625717163086,
+      "learning_rate": 9.615985964286129e-05,
+      "loss": 0.3403,
+      "step": 682
+    },
+    {
+      "epoch": 0.14424498416050685,
+      "grad_norm": 0.46445101499557495,
+      "learning_rate": 9.614682425479754e-05,
+      "loss": 0.3375,
+      "step": 683
+    },
+    {
+      "epoch": 0.14445617740232314,
+      "grad_norm": 0.4081145226955414,
+      "learning_rate": 9.613376766643065e-05,
+      "loss": 0.3359,
+      "step": 684
+    },
+    {
+      "epoch": 0.1446673706441394,
+      "grad_norm": 0.37794867157936096,
+      "learning_rate": 9.612068988375897e-05,
+      "loss": 0.3394,
+      "step": 685
+    },
+    {
+      "epoch": 0.14487856388595566,
+      "grad_norm": 0.3830733001232147,
+      "learning_rate": 9.610759091279053e-05,
+      "loss": 0.3372,
+      "step": 686
+    },
+    {
+      "epoch": 0.14508975712777192,
+      "grad_norm": 0.39936771988868713,
+      "learning_rate": 9.609447075954314e-05,
+      "loss": 0.3363,
+      "step": 687
+    },
+    {
+      "epoch": 0.14530095036958818,
+      "grad_norm": 0.744636595249176,
+      "learning_rate": 9.608132943004432e-05,
+      "loss": 0.3403,
+      "step": 688
+    },
+    {
+      "epoch": 0.14551214361140444,
+      "grad_norm": 0.4297600984573364,
+      "learning_rate": 9.606816693033132e-05,
+      "loss": 0.341,
+      "step": 689
+    },
+    {
+      "epoch": 0.1457233368532207,
+      "grad_norm": 0.4593130648136139,
+      "learning_rate": 9.605498326645114e-05,
+      "loss": 0.3382,
+      "step": 690
+    },
+    {
+      "epoch": 0.14593453009503696,
+      "grad_norm": 0.5597952604293823,
+      "learning_rate": 9.604177844446047e-05,
+      "loss": 0.3371,
+      "step": 691
+    },
+    {
+      "epoch": 0.14614572333685322,
+      "grad_norm": 0.4945879876613617,
+      "learning_rate": 9.602855247042575e-05,
+      "loss": 0.3371,
+      "step": 692
+    },
+    {
+      "epoch": 0.14635691657866948,
+      "grad_norm": 0.40828970074653625,
+      "learning_rate": 9.601530535042311e-05,
+      "loss": 0.3372,
+      "step": 693
+    },
+    {
+      "epoch": 0.14656810982048574,
+      "grad_norm": 0.3982973098754883,
+      "learning_rate": 9.60020370905384e-05,
+      "loss": 0.3337,
+      "step": 694
+    },
+    {
+      "epoch": 0.146779303062302,
+      "grad_norm": 0.5124569535255432,
+      "learning_rate": 9.598874769686719e-05,
+      "loss": 0.3361,
+      "step": 695
+    },
+    {
+      "epoch": 0.14699049630411826,
+      "grad_norm": 0.5415783524513245,
+      "learning_rate": 9.597543717551479e-05,
+      "loss": 0.3393,
+      "step": 696
+    },
+    {
+      "epoch": 0.14720168954593452,
+      "grad_norm": 0.4658135771751404,
+      "learning_rate": 9.596210553259613e-05,
+      "loss": 0.3388,
+      "step": 697
+    },
+    {
+      "epoch": 0.1474128827877508,
+      "grad_norm": 1.401531457901001,
+      "learning_rate": 9.594875277423594e-05,
+      "loss": 0.3344,
+      "step": 698
+    },
+    {
+      "epoch": 0.14762407602956706,
+      "grad_norm": 0.47507861256599426,
+      "learning_rate": 9.593537890656858e-05,
+      "loss": 0.3373,
+      "step": 699
+    },
+    {
+      "epoch": 0.14783526927138332,
+      "grad_norm": 0.5604132413864136,
+      "learning_rate": 9.592198393573816e-05,
+      "loss": 0.3378,
+      "step": 700
+    },
+    {
+      "epoch": 0.14804646251319958,
+      "grad_norm": 0.48407474160194397,
+      "learning_rate": 9.590856786789845e-05,
+      "loss": 0.3384,
+      "step": 701
+    },
+    {
+      "epoch": 0.14825765575501584,
+      "grad_norm": 0.6295081377029419,
+      "learning_rate": 9.589513070921292e-05,
+      "loss": 0.3342,
+      "step": 702
+    },
+    {
+      "epoch": 0.1484688489968321,
+      "grad_norm": 0.45591798424720764,
+      "learning_rate": 9.588167246585473e-05,
+      "loss": 0.3303,
+      "step": 703
+    },
+    {
+      "epoch": 0.14868004223864836,
+      "grad_norm": 0.6077401041984558,
+      "learning_rate": 9.586819314400674e-05,
+      "loss": 0.3297,
+      "step": 704
+    },
+    {
+      "epoch": 0.14889123548046462,
+      "grad_norm": 0.4915049076080322,
+      "learning_rate": 9.585469274986147e-05,
+      "loss": 0.3337,
+      "step": 705
+    },
+    {
+      "epoch": 0.14910242872228088,
+      "grad_norm": 0.535550057888031,
+      "learning_rate": 9.584117128962115e-05,
+      "loss": 0.333,
+      "step": 706
+    },
+    {
+      "epoch": 0.14931362196409714,
+      "grad_norm": 0.5761807560920715,
+      "learning_rate": 9.582762876949765e-05,
+      "loss": 0.3355,
+      "step": 707
+    },
+    {
+      "epoch": 0.1495248152059134,
+      "grad_norm": 0.46275466680526733,
+      "learning_rate": 9.581406519571257e-05,
+      "loss": 0.3344,
+      "step": 708
+    },
+    {
+      "epoch": 0.14973600844772966,
+      "grad_norm": 0.4903484582901001,
+      "learning_rate": 9.580048057449709e-05,
+      "loss": 0.3384,
+      "step": 709
+    },
+    {
+      "epoch": 0.14994720168954592,
+      "grad_norm": 0.40944942831993103,
+      "learning_rate": 9.578687491209218e-05,
+      "loss": 0.3296,
+      "step": 710
+    },
+    {
+      "epoch": 0.1501583949313622,
+      "grad_norm": 0.4731316566467285,
+      "learning_rate": 9.577324821474839e-05,
+      "loss": 0.332,
+      "step": 711
+    },
+    {
+      "epoch": 0.15036958817317847,
+      "grad_norm": 0.476868599653244,
+      "learning_rate": 9.575960048872594e-05,
+      "loss": 0.3258,
+      "step": 712
+    },
+    {
+      "epoch": 0.15058078141499473,
+      "grad_norm": 0.4580700993537903,
+      "learning_rate": 9.574593174029475e-05,
+      "loss": 0.3378,
+      "step": 713
+    },
+    {
+      "epoch": 0.150791974656811,
+      "grad_norm": 0.4308335483074188,
+      "learning_rate": 9.57322419757344e-05,
+      "loss": 0.3349,
+      "step": 714
+    },
+    {
+      "epoch": 0.15100316789862725,
+      "grad_norm": 0.5005003809928894,
+      "learning_rate": 9.571853120133405e-05,
+      "loss": 0.3359,
+      "step": 715
+    },
+    {
+      "epoch": 0.1512143611404435,
+      "grad_norm": 0.4359145760536194,
+      "learning_rate": 9.570479942339259e-05,
+      "loss": 0.3367,
+      "step": 716
+    },
+    {
+      "epoch": 0.15142555438225977,
+      "grad_norm": 0.43031927943229675,
+      "learning_rate": 9.569104664821853e-05,
+      "loss": 0.3315,
+      "step": 717
+    },
+    {
+      "epoch": 0.15163674762407603,
+      "grad_norm": 0.4488793909549713,
+      "learning_rate": 9.567727288213005e-05,
+      "loss": 0.3262,
+      "step": 718
+    },
+    {
+      "epoch": 0.1518479408658923,
+      "grad_norm": 0.433355450630188,
+      "learning_rate": 9.566347813145493e-05,
+      "loss": 0.3342,
+      "step": 719
+    },
+    {
+      "epoch": 0.15205913410770855,
+      "grad_norm": 0.4327280521392822,
+      "learning_rate": 9.56496624025306e-05,
+      "loss": 0.3396,
+      "step": 720
+    },
+    {
+      "epoch": 0.1522703273495248,
+      "grad_norm": 0.48347118496894836,
+      "learning_rate": 9.563582570170417e-05,
+      "loss": 0.3395,
+      "step": 721
+    },
+    {
+      "epoch": 0.15248152059134107,
+      "grad_norm": 0.5038233399391174,
+      "learning_rate": 9.562196803533235e-05,
+      "loss": 0.3417,
+      "step": 722
+    },
+    {
+      "epoch": 0.15269271383315733,
+      "grad_norm": 0.4363105595111847,
+      "learning_rate": 9.560808940978148e-05,
+      "loss": 0.3355,
+      "step": 723
+    },
+    {
+      "epoch": 0.1529039070749736,
+      "grad_norm": 0.4099661111831665,
+      "learning_rate": 9.559418983142755e-05,
+      "loss": 0.3306,
+      "step": 724
+    },
+    {
+      "epoch": 0.15311510031678988,
+      "grad_norm": 0.48323655128479004,
+      "learning_rate": 9.558026930665615e-05,
+      "loss": 0.3361,
+      "step": 725
+    },
+    {
+      "epoch": 0.15332629355860614,
+      "grad_norm": 0.36575576663017273,
+      "learning_rate": 9.55663278418625e-05,
+      "loss": 0.3309,
+      "step": 726
+    },
+    {
+      "epoch": 0.1535374868004224,
+      "grad_norm": 0.38799694180488586,
+      "learning_rate": 9.555236544345146e-05,
+      "loss": 0.3358,
+      "step": 727
+    },
+    {
+      "epoch": 0.15374868004223866,
+      "grad_norm": 0.4442598819732666,
+      "learning_rate": 9.553838211783747e-05,
+      "loss": 0.3362,
+      "step": 728
+    },
+    {
+      "epoch": 0.15395987328405492,
+      "grad_norm": 0.4882688522338867,
+      "learning_rate": 9.552437787144464e-05,
+      "loss": 0.3424,
+      "step": 729
+    },
+    {
+      "epoch": 0.15417106652587118,
+      "grad_norm": 0.3939456045627594,
+      "learning_rate": 9.551035271070664e-05,
+      "loss": 0.3374,
+      "step": 730
+    },
+    {
+      "epoch": 0.15438225976768744,
+      "grad_norm": 0.4823892116546631,
+      "learning_rate": 9.549630664206677e-05,
+      "loss": 0.3327,
+      "step": 731
+    },
+    {
+      "epoch": 0.1545934530095037,
+      "grad_norm": 0.3858034610748291,
+      "learning_rate": 9.548223967197791e-05,
+      "loss": 0.3348,
+      "step": 732
+    },
+    {
+      "epoch": 0.15480464625131996,
+      "grad_norm": 0.4620276093482971,
+      "learning_rate": 9.546815180690261e-05,
+      "loss": 0.3358,
+      "step": 733
+    },
+    {
+      "epoch": 0.15501583949313621,
+      "grad_norm": 0.45092567801475525,
+      "learning_rate": 9.545404305331292e-05,
+      "loss": 0.3317,
+      "step": 734
+    },
+    {
+      "epoch": 0.15522703273495247,
+      "grad_norm": 0.331874281167984,
+      "learning_rate": 9.543991341769057e-05,
+      "loss": 0.3339,
+      "step": 735
+    },
+    {
+      "epoch": 0.15543822597676873,
+      "grad_norm": 0.5371387004852295,
+      "learning_rate": 9.542576290652687e-05,
+      "loss": 0.3455,
+      "step": 736
+    },
+    {
+      "epoch": 0.155649419218585,
+      "grad_norm": 0.4991358518600464,
+      "learning_rate": 9.541159152632266e-05,
+      "loss": 0.3288,
+      "step": 737
+    },
+    {
+      "epoch": 0.15586061246040125,
+      "grad_norm": 0.3737116754055023,
+      "learning_rate": 9.539739928358845e-05,
+      "loss": 0.3268,
+      "step": 738
+    },
+    {
+      "epoch": 0.15607180570221754,
+      "grad_norm": 0.5014138221740723,
+      "learning_rate": 9.538318618484425e-05,
+      "loss": 0.3356,
+      "step": 739
+    },
+    {
+      "epoch": 0.1562829989440338,
+      "grad_norm": 0.4366362988948822,
+      "learning_rate": 9.536895223661974e-05,
+      "loss": 0.33,
+      "step": 740
+    },
+    {
+      "epoch": 0.15649419218585006,
+      "grad_norm": 0.4134378731250763,
+      "learning_rate": 9.535469744545413e-05,
+      "loss": 0.3334,
+      "step": 741
+    },
+    {
+      "epoch": 0.15670538542766632,
+      "grad_norm": 0.4393095374107361,
+      "learning_rate": 9.534042181789618e-05,
+      "loss": 0.3332,
+      "step": 742
+    },
+    {
+      "epoch": 0.15691657866948258,
+      "grad_norm": 0.4475516974925995,
+      "learning_rate": 9.532612536050429e-05,
+      "loss": 0.3334,
+      "step": 743
+    },
+    {
+      "epoch": 0.15712777191129884,
+      "grad_norm": 0.5534899234771729,
+      "learning_rate": 9.531180807984637e-05,
+      "loss": 0.3319,
+      "step": 744
+    },
+    {
+      "epoch": 0.1573389651531151,
+      "grad_norm": 0.37943577766418457,
+      "learning_rate": 9.529746998249993e-05,
+      "loss": 0.3302,
+      "step": 745
+    },
+    {
+      "epoch": 0.15755015839493136,
+      "grad_norm": 0.40101131796836853,
+      "learning_rate": 9.528311107505203e-05,
+      "loss": 0.3342,
+      "step": 746
+    },
+    {
+      "epoch": 0.15776135163674762,
+      "grad_norm": 0.4060884416103363,
+      "learning_rate": 9.526873136409928e-05,
+      "loss": 0.3354,
+      "step": 747
+    },
+    {
+      "epoch": 0.15797254487856388,
+      "grad_norm": 0.3312695324420929,
+      "learning_rate": 9.525433085624788e-05,
+      "loss": 0.3312,
+      "step": 748
+    },
+    {
+      "epoch": 0.15818373812038014,
+      "grad_norm": 0.4205334782600403,
+      "learning_rate": 9.523990955811356e-05,
+      "loss": 0.3291,
+      "step": 749
+    },
+    {
+      "epoch": 0.1583949313621964,
+      "grad_norm": 0.39660462737083435,
+      "learning_rate": 9.522546747632159e-05,
+      "loss": 0.3329,
+      "step": 750
+    },
+    {
+      "epoch": 0.15860612460401266,
+      "grad_norm": 0.4399871230125427,
+      "learning_rate": 9.521100461750682e-05,
+      "loss": 0.3255,
+      "step": 751
+    },
+    {
+      "epoch": 0.15881731784582892,
+      "grad_norm": 0.3940870463848114,
+      "learning_rate": 9.519652098831364e-05,
+      "loss": 0.3313,
+      "step": 752
+    },
+    {
+      "epoch": 0.1590285110876452,
+      "grad_norm": 0.4242289960384369,
+      "learning_rate": 9.518201659539593e-05,
+      "loss": 0.341,
+      "step": 753
+    },
+    {
+      "epoch": 0.15923970432946147,
+      "grad_norm": 0.4816989004611969,
+      "learning_rate": 9.516749144541718e-05,
+      "loss": 0.333,
+      "step": 754
+    },
+    {
+      "epoch": 0.15945089757127773,
+      "grad_norm": 0.3451484441757202,
+      "learning_rate": 9.515294554505039e-05,
+      "loss": 0.3304,
+      "step": 755
+    },
+    {
+      "epoch": 0.159662090813094,
+      "grad_norm": 0.36413100361824036,
+      "learning_rate": 9.513837890097807e-05,
+      "loss": 0.33,
+      "step": 756
+    },
+    {
+      "epoch": 0.15987328405491025,
+      "grad_norm": 0.4149964153766632,
+      "learning_rate": 9.512379151989229e-05,
+      "loss": 0.3181,
+      "step": 757
+    },
+    {
+      "epoch": 0.1600844772967265,
+      "grad_norm": 0.47271728515625,
+      "learning_rate": 9.510918340849464e-05,
+      "loss": 0.3275,
+      "step": 758
+    },
+    {
+      "epoch": 0.16029567053854277,
+      "grad_norm": 0.4869644045829773,
+      "learning_rate": 9.509455457349622e-05,
+      "loss": 0.3305,
+      "step": 759
+    },
+    {
+      "epoch": 0.16050686378035903,
+      "grad_norm": 0.4085192382335663,
+      "learning_rate": 9.507990502161768e-05,
+      "loss": 0.3224,
+      "step": 760
+    },
+    {
+      "epoch": 0.1607180570221753,
+      "grad_norm": 0.4083760976791382,
+      "learning_rate": 9.506523475958913e-05,
+      "loss": 0.3313,
+      "step": 761
+    },
+    {
+      "epoch": 0.16092925026399155,
+      "grad_norm": 0.49541711807250977,
+      "learning_rate": 9.505054379415026e-05,
+      "loss": 0.3258,
+      "step": 762
+    },
+    {
+      "epoch": 0.1611404435058078,
+      "grad_norm": 0.40919989347457886,
+      "learning_rate": 9.503583213205025e-05,
+      "loss": 0.3249,
+      "step": 763
+    },
+    {
+      "epoch": 0.16135163674762407,
+      "grad_norm": 0.41402333974838257,
+      "learning_rate": 9.502109978004776e-05,
+      "loss": 0.3378,
+      "step": 764
+    },
+    {
+      "epoch": 0.16156282998944033,
+      "grad_norm": 0.41789570450782776,
+      "learning_rate": 9.500634674491097e-05,
+      "loss": 0.3296,
+      "step": 765
+    },
+    {
+      "epoch": 0.1617740232312566,
+      "grad_norm": 0.476440966129303,
+      "learning_rate": 9.499157303341762e-05,
+      "loss": 0.3253,
+      "step": 766
+    },
+    {
+      "epoch": 0.16198521647307287,
+      "grad_norm": 0.40855103731155396,
+      "learning_rate": 9.497677865235485e-05,
+      "loss": 0.3255,
+      "step": 767
+    },
+    {
+      "epoch": 0.16219640971488913,
+      "grad_norm": 0.6218886375427246,
+      "learning_rate": 9.496196360851934e-05,
+      "loss": 0.3275,
+      "step": 768
+    },
+    {
+      "epoch": 0.1624076029567054,
+      "grad_norm": 0.4886719882488251,
+      "learning_rate": 9.49471279087173e-05,
+      "loss": 0.3241,
+      "step": 769
+    },
+    {
+      "epoch": 0.16261879619852165,
+      "grad_norm": 0.40743178129196167,
+      "learning_rate": 9.493227155976439e-05,
+      "loss": 0.3314,
+      "step": 770
+    },
+    {
+      "epoch": 0.1628299894403379,
+      "grad_norm": 0.41739529371261597,
+      "learning_rate": 9.491739456848573e-05,
+      "loss": 0.3271,
+      "step": 771
+    },
+    {
+      "epoch": 0.16304118268215417,
+      "grad_norm": 0.45175400376319885,
+      "learning_rate": 9.490249694171597e-05,
+      "loss": 0.3228,
+      "step": 772
+    },
+    {
+      "epoch": 0.16325237592397043,
+      "grad_norm": 0.4218711256980896,
+      "learning_rate": 9.488757868629925e-05,
+      "loss": 0.3286,
+      "step": 773
+    },
+    {
+      "epoch": 0.1634635691657867,
+      "grad_norm": 0.48408418893814087,
+      "learning_rate": 9.487263980908914e-05,
+      "loss": 0.3345,
+      "step": 774
+    },
+    {
+      "epoch": 0.16367476240760295,
+      "grad_norm": 0.3920229375362396,
+      "learning_rate": 9.485768031694872e-05,
+      "loss": 0.3244,
+      "step": 775
+    },
+    {
+      "epoch": 0.1638859556494192,
+      "grad_norm": 0.4762434661388397,
+      "learning_rate": 9.48427002167505e-05,
+      "loss": 0.3139,
+      "step": 776
+    },
+    {
+      "epoch": 0.16409714889123547,
+      "grad_norm": 0.4162623882293701,
+      "learning_rate": 9.482769951537651e-05,
+      "loss": 0.3217,
+      "step": 777
+    },
+    {
+      "epoch": 0.16430834213305173,
+      "grad_norm": 0.48831644654273987,
+      "learning_rate": 9.481267821971822e-05,
+      "loss": 0.3274,
+      "step": 778
+    },
+    {
+      "epoch": 0.164519535374868,
+      "grad_norm": 0.3706333041191101,
+      "learning_rate": 9.479763633667656e-05,
+      "loss": 0.3187,
+      "step": 779
+    },
+    {
+      "epoch": 0.16473072861668428,
+      "grad_norm": 0.4122789204120636,
+      "learning_rate": 9.47825738731619e-05,
+      "loss": 0.3207,
+      "step": 780
+    },
+    {
+      "epoch": 0.16494192185850054,
+      "grad_norm": 0.4652899503707886,
+      "learning_rate": 9.476749083609408e-05,
+      "loss": 0.3225,
+      "step": 781
+    },
+    {
+      "epoch": 0.1651531151003168,
+      "grad_norm": 0.43659114837646484,
+      "learning_rate": 9.475238723240242e-05,
+      "loss": 0.3207,
+      "step": 782
+    },
+    {
+      "epoch": 0.16536430834213306,
+      "grad_norm": 0.38358327746391296,
+      "learning_rate": 9.473726306902567e-05,
+      "loss": 0.3263,
+      "step": 783
+    },
+    {
+      "epoch": 0.16557550158394932,
+      "grad_norm": 0.4371603727340698,
+      "learning_rate": 9.472211835291199e-05,
+      "loss": 0.3239,
+      "step": 784
+    },
+    {
+      "epoch": 0.16578669482576558,
+      "grad_norm": 0.6550732254981995,
+      "learning_rate": 9.470695309101903e-05,
+      "loss": 0.3321,
+      "step": 785
+    },
+    {
+      "epoch": 0.16599788806758184,
+      "grad_norm": 0.4869711995124817,
+      "learning_rate": 9.469176729031384e-05,
+      "loss": 0.3224,
+      "step": 786
+    },
+    {
+      "epoch": 0.1662090813093981,
+      "grad_norm": 0.4528375267982483,
+      "learning_rate": 9.467656095777294e-05,
+      "loss": 0.3278,
+      "step": 787
+    },
+    {
+      "epoch": 0.16642027455121436,
+      "grad_norm": 0.5814886093139648,
+      "learning_rate": 9.466133410038227e-05,
+      "loss": 0.3239,
+      "step": 788
+    },
+    {
+      "epoch": 0.16663146779303062,
+      "grad_norm": 0.46977880597114563,
+      "learning_rate": 9.464608672513719e-05,
+      "loss": 0.32,
+      "step": 789
+    },
+    {
+      "epoch": 0.16684266103484688,
+      "grad_norm": 0.4674695134162903,
+      "learning_rate": 9.463081883904252e-05,
+      "loss": 0.3231,
+      "step": 790
+    },
+    {
+      "epoch": 0.16705385427666314,
+      "grad_norm": 0.49594321846961975,
+      "learning_rate": 9.461553044911242e-05,
+      "loss": 0.3269,
+      "step": 791
+    },
+    {
+      "epoch": 0.1672650475184794,
+      "grad_norm": 0.4163094460964203,
+      "learning_rate": 9.460022156237058e-05,
+      "loss": 0.3252,
+      "step": 792
+    },
+    {
+      "epoch": 0.16747624076029566,
+      "grad_norm": 0.4525318145751953,
+      "learning_rate": 9.458489218585002e-05,
+      "loss": 0.3271,
+      "step": 793
+    },
+    {
+      "epoch": 0.16768743400211195,
+      "grad_norm": 0.4324900805950165,
+      "learning_rate": 9.456954232659324e-05,
+      "loss": 0.3348,
+      "step": 794
+    },
+    {
+      "epoch": 0.1678986272439282,
+      "grad_norm": 0.381202757358551,
+      "learning_rate": 9.455417199165208e-05,
+      "loss": 0.3245,
+      "step": 795
+    },
+    {
+      "epoch": 0.16810982048574447,
+      "grad_norm": 0.41851064562797546,
+      "learning_rate": 9.453878118808783e-05,
+      "loss": 0.3275,
+      "step": 796
+    },
+    {
+      "epoch": 0.16832101372756073,
+      "grad_norm": 0.535701334476471,
+      "learning_rate": 9.452336992297119e-05,
+      "loss": 0.3249,
+      "step": 797
+    },
+    {
+      "epoch": 0.16853220696937699,
+      "grad_norm": 0.39901670813560486,
+      "learning_rate": 9.450793820338225e-05,
+      "loss": 0.3186,
+      "step": 798
+    },
+    {
+      "epoch": 0.16874340021119325,
+      "grad_norm": 0.39781031012535095,
+      "learning_rate": 9.449248603641048e-05,
+      "loss": 0.317,
+      "step": 799
+    },
+    {
+      "epoch": 0.1689545934530095,
+      "grad_norm": 0.4598289728164673,
+      "learning_rate": 9.447701342915473e-05,
+      "loss": 0.3186,
+      "step": 800
+    },
+    {
+      "epoch": 0.1689545934530095,
+      "eval_train_split_loss": 0.3939213454723358,
+      "eval_train_split_runtime": 6.4543,
+      "eval_train_split_samples_per_second": 1.549,
+      "eval_train_split_steps_per_second": 0.155,
+      "step": 800
+    },
+    {
+      "epoch": 0.16916578669482576,
+      "grad_norm": 0.4896026849746704,
+      "learning_rate": 9.446152038872334e-05,
+      "loss": 0.327,
+      "step": 801
+    },
+    {
+      "epoch": 0.16937697993664202,
+      "grad_norm": 0.37723612785339355,
+      "learning_rate": 9.444600692223389e-05,
+      "loss": 0.33,
+      "step": 802
+    },
+    {
+      "epoch": 0.16958817317845828,
+      "grad_norm": 0.4697837829589844,
+      "learning_rate": 9.443047303681344e-05,
+      "loss": 0.3202,
+      "step": 803
+    },
+    {
+      "epoch": 0.16979936642027454,
+      "grad_norm": 0.7565957903862,
+      "learning_rate": 9.441491873959844e-05,
+      "loss": 0.3207,
+      "step": 804
+    },
+    {
+      "epoch": 0.1700105596620908,
+      "grad_norm": 0.5566945672035217,
+      "learning_rate": 9.439934403773468e-05,
+      "loss": 0.3206,
+      "step": 805
+    },
+    {
+      "epoch": 0.17022175290390706,
+      "grad_norm": 0.5193995237350464,
+      "learning_rate": 9.438374893837728e-05,
+      "loss": 0.3219,
+      "step": 806
+    },
+    {
+      "epoch": 0.17043294614572332,
+      "grad_norm": 0.5711631178855896,
+      "learning_rate": 9.436813344869083e-05,
+      "loss": 0.3259,
+      "step": 807
+    },
+    {
+      "epoch": 0.1706441393875396,
+      "grad_norm": 0.4157835841178894,
+      "learning_rate": 9.435249757584922e-05,
+      "loss": 0.33,
+      "step": 808
+    },
+    {
+      "epoch": 0.17085533262935587,
+      "grad_norm": 0.4902370870113373,
+      "learning_rate": 9.433684132703575e-05,
+      "loss": 0.3288,
+      "step": 809
+    },
+    {
+      "epoch": 0.17106652587117213,
+      "grad_norm": 0.4441160261631012,
+      "learning_rate": 9.4321164709443e-05,
+      "loss": 0.3285,
+      "step": 810
+    },
+    {
+      "epoch": 0.1712777191129884,
+      "grad_norm": 0.43177324533462524,
+      "learning_rate": 9.430546773027302e-05,
+      "loss": 0.3246,
+      "step": 811
+    },
+    {
+      "epoch": 0.17148891235480465,
+      "grad_norm": 0.5109300017356873,
+      "learning_rate": 9.428975039673713e-05,
+      "loss": 0.321,
+      "step": 812
+    },
+    {
+      "epoch": 0.1717001055966209,
+      "grad_norm": 1.077674388885498,
+      "learning_rate": 9.427401271605603e-05,
+      "loss": 0.3224,
+      "step": 813
+    },
+    {
+      "epoch": 0.17191129883843717,
+      "grad_norm": 0.5562537908554077,
+      "learning_rate": 9.425825469545974e-05,
+      "loss": 0.3368,
+      "step": 814
+    },
+    {
+      "epoch": 0.17212249208025343,
+      "grad_norm": 0.6010218262672424,
+      "learning_rate": 9.42424763421877e-05,
+      "loss": 0.3387,
+      "step": 815
+    },
+    {
+      "epoch": 0.1723336853220697,
+      "grad_norm": 0.47284823656082153,
+      "learning_rate": 9.422667766348859e-05,
+      "loss": 0.3223,
+      "step": 816
+    },
+    {
+      "epoch": 0.17254487856388595,
+      "grad_norm": 0.7043172717094421,
+      "learning_rate": 9.421085866662051e-05,
+      "loss": 0.3296,
+      "step": 817
+    },
+    {
+      "epoch": 0.1727560718057022,
+      "grad_norm": 0.550128698348999,
+      "learning_rate": 9.419501935885084e-05,
+      "loss": 0.3281,
+      "step": 818
+    },
+    {
+      "epoch": 0.17296726504751847,
+      "grad_norm": 0.7333943843841553,
+      "learning_rate": 9.417915974745633e-05,
+      "loss": 0.3213,
+      "step": 819
+    },
+    {
+      "epoch": 0.17317845828933473,
+      "grad_norm": 0.4838084876537323,
+      "learning_rate": 9.416327983972304e-05,
+      "loss": 0.3327,
+      "step": 820
+    },
+    {
+      "epoch": 0.173389651531151,
+      "grad_norm": 0.6104544997215271,
+      "learning_rate": 9.414737964294636e-05,
+      "loss": 0.3332,
+      "step": 821
+    },
+    {
+      "epoch": 0.17360084477296728,
+      "grad_norm": 2.8598692417144775,
+      "learning_rate": 9.413145916443098e-05,
+      "loss": 0.3311,
+      "step": 822
+    },
+    {
+      "epoch": 0.17381203801478354,
+      "grad_norm": 0.701901912689209,
+      "learning_rate": 9.411551841149092e-05,
+      "loss": 0.3362,
+      "step": 823
+    },
+    {
+      "epoch": 0.1740232312565998,
+      "grad_norm": 1.1731001138687134,
+      "learning_rate": 9.409955739144953e-05,
+      "loss": 0.3204,
+      "step": 824
+    },
+    {
+      "epoch": 0.17423442449841606,
+      "grad_norm": 0.8659184575080872,
+      "learning_rate": 9.408357611163945e-05,
+      "loss": 0.3308,
+      "step": 825
+    },
+    {
+      "epoch": 0.17444561774023232,
+      "grad_norm": 0.9301297664642334,
+      "learning_rate": 9.406757457940266e-05,
+      "loss": 0.3316,
+      "step": 826
+    },
+    {
+      "epoch": 0.17465681098204858,
+      "grad_norm": 0.9139550924301147,
+      "learning_rate": 9.40515528020904e-05,
+      "loss": 0.3263,
+      "step": 827
+    },
+    {
+      "epoch": 0.17486800422386484,
+      "grad_norm": 1.0153284072875977,
+      "learning_rate": 9.403551078706324e-05,
+      "loss": 0.3273,
+      "step": 828
+    },
+    {
+      "epoch": 0.1750791974656811,
+      "grad_norm": 0.5354955196380615,
+      "learning_rate": 9.401944854169102e-05,
+      "loss": 0.3277,
+      "step": 829
+    },
+    {
+      "epoch": 0.17529039070749736,
+      "grad_norm": 0.54505455493927,
+      "learning_rate": 9.400336607335293e-05,
+      "loss": 0.3252,
+      "step": 830
+    },
+    {
+      "epoch": 0.17550158394931362,
+      "grad_norm": 0.47695547342300415,
+      "learning_rate": 9.398726338943737e-05,
+      "loss": 0.3233,
+      "step": 831
+    },
+    {
+      "epoch": 0.17571277719112988,
+      "grad_norm": 0.40267011523246765,
+      "learning_rate": 9.39711404973421e-05,
+      "loss": 0.3166,
+      "step": 832
+    },
+    {
+      "epoch": 0.17592397043294614,
+      "grad_norm": 0.7231528759002686,
+      "learning_rate": 9.395499740447415e-05,
+      "loss": 0.3237,
+      "step": 833
+    },
+    {
+      "epoch": 0.1761351636747624,
+      "grad_norm": 0.4179030656814575,
+      "learning_rate": 9.393883411824977e-05,
+      "loss": 0.3204,
+      "step": 834
+    },
+    {
+      "epoch": 0.17634635691657866,
+      "grad_norm": 0.7190957069396973,
+      "learning_rate": 9.392265064609455e-05,
+      "loss": 0.3253,
+      "step": 835
+    },
+    {
+      "epoch": 0.17655755015839494,
+      "grad_norm": 0.7434033751487732,
+      "learning_rate": 9.390644699544334e-05,
+      "loss": 0.3228,
+      "step": 836
+    },
+    {
+      "epoch": 0.1767687434002112,
+      "grad_norm": 0.3990352153778076,
+      "learning_rate": 9.389022317374025e-05,
+      "loss": 0.3265,
+      "step": 837
+    },
+    {
+      "epoch": 0.17697993664202746,
+      "grad_norm": 0.6141509413719177,
+      "learning_rate": 9.38739791884387e-05,
+      "loss": 0.3295,
+      "step": 838
+    },
+    {
+      "epoch": 0.17719112988384372,
+      "grad_norm": 0.4323461055755615,
+      "learning_rate": 9.385771504700127e-05,
+      "loss": 0.3191,
+      "step": 839
+    },
+    {
+      "epoch": 0.17740232312565998,
+      "grad_norm": 0.3937886655330658,
+      "learning_rate": 9.38414307568999e-05,
+      "loss": 0.3233,
+      "step": 840
+    },
+    {
+      "epoch": 0.17761351636747624,
+      "grad_norm": 0.770662248134613,
+      "learning_rate": 9.382512632561577e-05,
+      "loss": 0.3241,
+      "step": 841
+    },
+    {
+      "epoch": 0.1778247096092925,
+      "grad_norm": 0.5537351965904236,
+      "learning_rate": 9.380880176063926e-05,
+      "loss": 0.3272,
+      "step": 842
+    },
+    {
+      "epoch": 0.17803590285110876,
+      "grad_norm": 0.4853312075138092,
+      "learning_rate": 9.379245706947008e-05,
+      "loss": 0.326,
+      "step": 843
+    },
+    {
+      "epoch": 0.17824709609292502,
+      "grad_norm": 0.44467559456825256,
+      "learning_rate": 9.377609225961709e-05,
+      "loss": 0.3246,
+      "step": 844
+    },
+    {
+      "epoch": 0.17845828933474128,
+      "grad_norm": 0.4518132507801056,
+      "learning_rate": 9.375970733859848e-05,
+      "loss": 0.3219,
+      "step": 845
+    },
+    {
+      "epoch": 0.17866948257655754,
+      "grad_norm": 0.39297959208488464,
+      "learning_rate": 9.374330231394162e-05,
+      "loss": 0.3247,
+      "step": 846
+    },
+    {
+      "epoch": 0.1788806758183738,
+      "grad_norm": 0.6081179976463318,
+      "learning_rate": 9.372687719318317e-05,
+      "loss": 0.3248,
+      "step": 847
+    },
+    {
+      "epoch": 0.17909186906019006,
+      "grad_norm": 0.4550902843475342,
+      "learning_rate": 9.371043198386895e-05,
+      "loss": 0.3205,
+      "step": 848
+    },
+    {
+      "epoch": 0.17930306230200635,
+      "grad_norm": 0.4423544704914093,
+      "learning_rate": 9.369396669355408e-05,
+      "loss": 0.3208,
+      "step": 849
+    },
+    {
+      "epoch": 0.1795142555438226,
+      "grad_norm": 0.3300669193267822,
+      "learning_rate": 9.367748132980287e-05,
+      "loss": 0.3205,
+      "step": 850
+    },
+    {
+      "epoch": 0.17972544878563887,
+      "grad_norm": 0.5292059779167175,
+      "learning_rate": 9.366097590018885e-05,
+      "loss": 0.3225,
+      "step": 851
+    },
+    {
+      "epoch": 0.17993664202745513,
+      "grad_norm": 0.48071083426475525,
+      "learning_rate": 9.364445041229477e-05,
+      "loss": 0.3184,
+      "step": 852
+    },
+    {
+      "epoch": 0.1801478352692714,
+      "grad_norm": 0.5632315874099731,
+      "learning_rate": 9.362790487371262e-05,
+      "loss": 0.3148,
+      "step": 853
+    },
+    {
+      "epoch": 0.18035902851108765,
+      "grad_norm": 0.433458536863327,
+      "learning_rate": 9.36113392920436e-05,
+      "loss": 0.3211,
+      "step": 854
+    },
+    {
+      "epoch": 0.1805702217529039,
+      "grad_norm": 0.43643510341644287,
+      "learning_rate": 9.359475367489806e-05,
+      "loss": 0.3123,
+      "step": 855
+    },
+    {
+      "epoch": 0.18078141499472017,
+      "grad_norm": 0.43392258882522583,
+      "learning_rate": 9.35781480298956e-05,
+      "loss": 0.3149,
+      "step": 856
+    },
+    {
+      "epoch": 0.18099260823653643,
+      "grad_norm": 0.389758437871933,
+      "learning_rate": 9.356152236466503e-05,
+      "loss": 0.3183,
+      "step": 857
+    },
+    {
+      "epoch": 0.1812038014783527,
+      "grad_norm": 0.45618849992752075,
+      "learning_rate": 9.354487668684433e-05,
+      "loss": 0.3201,
+      "step": 858
+    },
+    {
+      "epoch": 0.18141499472016895,
+      "grad_norm": 0.36153000593185425,
+      "learning_rate": 9.352821100408071e-05,
+      "loss": 0.3117,
+      "step": 859
+    },
+    {
+      "epoch": 0.1816261879619852,
+      "grad_norm": 0.4454471170902252,
+      "learning_rate": 9.351152532403054e-05,
+      "loss": 0.3183,
+      "step": 860
+    },
+    {
+      "epoch": 0.18183738120380147,
+      "grad_norm": 0.512215793132782,
+      "learning_rate": 9.349481965435938e-05,
+      "loss": 0.3234,
+      "step": 861
+    },
+    {
+      "epoch": 0.18204857444561773,
+      "grad_norm": 0.5209541916847229,
+      "learning_rate": 9.347809400274197e-05,
+      "loss": 0.3227,
+      "step": 862
+    },
+    {
+      "epoch": 0.18225976768743402,
+      "grad_norm": 1.0272666215896606,
+      "learning_rate": 9.346134837686225e-05,
+      "loss": 0.3202,
+      "step": 863
+    },
+    {
+      "epoch": 0.18247096092925028,
+      "grad_norm": 0.37011298537254333,
+      "learning_rate": 9.34445827844133e-05,
+      "loss": 0.3201,
+      "step": 864
+    },
+    {
+      "epoch": 0.18268215417106654,
+      "grad_norm": 0.39306700229644775,
+      "learning_rate": 9.342779723309745e-05,
+      "loss": 0.316,
+      "step": 865
+    },
+    {
+      "epoch": 0.1828933474128828,
+      "grad_norm": 0.454194575548172,
+      "learning_rate": 9.341099173062609e-05,
+      "loss": 0.3176,
+      "step": 866
+    },
+    {
+      "epoch": 0.18310454065469906,
+      "grad_norm": 0.3529050946235657,
+      "learning_rate": 9.339416628471989e-05,
+      "loss": 0.313,
+      "step": 867
+    },
+    {
+      "epoch": 0.18331573389651531,
+      "grad_norm": 0.3878239393234253,
+      "learning_rate": 9.337732090310855e-05,
+      "loss": 0.3146,
+      "step": 868
+    },
+    {
+      "epoch": 0.18352692713833157,
+      "grad_norm": 0.4624103903770447,
+      "learning_rate": 9.336045559353108e-05,
+      "loss": 0.3148,
+      "step": 869
+    },
+    {
+      "epoch": 0.18373812038014783,
+      "grad_norm": 0.48445650935173035,
+      "learning_rate": 9.334357036373551e-05,
+      "loss": 0.313,
+      "step": 870
+    },
+    {
+      "epoch": 0.1839493136219641,
+      "grad_norm": 0.35929012298583984,
+      "learning_rate": 9.332666522147913e-05,
+      "loss": 0.3145,
+      "step": 871
+    },
+    {
+      "epoch": 0.18416050686378035,
+      "grad_norm": 0.3527107536792755,
+      "learning_rate": 9.330974017452827e-05,
+      "loss": 0.3184,
+      "step": 872
+    },
+    {
+      "epoch": 0.18437170010559661,
+      "grad_norm": 0.4347122609615326,
+      "learning_rate": 9.329279523065852e-05,
+      "loss": 0.3211,
+      "step": 873
+    },
+    {
+      "epoch": 0.18458289334741287,
+      "grad_norm": 0.4034436345100403,
+      "learning_rate": 9.327583039765453e-05,
+      "loss": 0.3143,
+      "step": 874
+    },
+    {
+      "epoch": 0.18479408658922913,
+      "grad_norm": 0.4989505410194397,
+      "learning_rate": 9.325884568331009e-05,
+      "loss": 0.3226,
+      "step": 875
+    },
+    {
+      "epoch": 0.1850052798310454,
+      "grad_norm": 0.4032548666000366,
+      "learning_rate": 9.324184109542818e-05,
+      "loss": 0.3169,
+      "step": 876
+    },
+    {
+      "epoch": 0.18521647307286168,
+      "grad_norm": 0.400626003742218,
+      "learning_rate": 9.322481664182083e-05,
+      "loss": 0.3121,
+      "step": 877
+    },
+    {
+      "epoch": 0.18542766631467794,
+      "grad_norm": 0.3362039029598236,
+      "learning_rate": 9.320777233030929e-05,
+      "loss": 0.3129,
+      "step": 878
+    },
+    {
+      "epoch": 0.1856388595564942,
+      "grad_norm": 0.4295433759689331,
+      "learning_rate": 9.319070816872385e-05,
+      "loss": 0.3148,
+      "step": 879
+    },
+    {
+      "epoch": 0.18585005279831046,
+      "grad_norm": 0.42669329047203064,
+      "learning_rate": 9.317362416490396e-05,
+      "loss": 0.3098,
+      "step": 880
+    },
+    {
+      "epoch": 0.18606124604012672,
+      "grad_norm": 0.339597225189209,
+      "learning_rate": 9.315652032669819e-05,
+      "loss": 0.3134,
+      "step": 881
+    },
+    {
+      "epoch": 0.18627243928194298,
+      "grad_norm": 0.4802435338497162,
+      "learning_rate": 9.313939666196421e-05,
+      "loss": 0.3096,
+      "step": 882
+    },
+    {
+      "epoch": 0.18648363252375924,
+      "grad_norm": 0.5464080572128296,
+      "learning_rate": 9.31222531785688e-05,
+      "loss": 0.3118,
+      "step": 883
+    },
+    {
+      "epoch": 0.1866948257655755,
+      "grad_norm": 0.492174357175827,
+      "learning_rate": 9.310508988438783e-05,
+      "loss": 0.3183,
+      "step": 884
+    },
+    {
+      "epoch": 0.18690601900739176,
+      "grad_norm": 0.4431619346141815,
+      "learning_rate": 9.308790678730627e-05,
+      "loss": 0.3136,
+      "step": 885
+    },
+    {
+      "epoch": 0.18711721224920802,
+      "grad_norm": 0.33839544653892517,
+      "learning_rate": 9.307070389521824e-05,
+      "loss": 0.3119,
+      "step": 886
+    },
+    {
+      "epoch": 0.18732840549102428,
+      "grad_norm": 0.3547189235687256,
+      "learning_rate": 9.305348121602691e-05,
+      "loss": 0.3164,
+      "step": 887
+    },
+    {
+      "epoch": 0.18753959873284054,
+      "grad_norm": 0.40625622868537903,
+      "learning_rate": 9.303623875764454e-05,
+      "loss": 0.3119,
+      "step": 888
+    },
+    {
+      "epoch": 0.1877507919746568,
+      "grad_norm": 0.305142879486084,
+      "learning_rate": 9.301897652799247e-05,
+      "loss": 0.3182,
+      "step": 889
+    },
+    {
+      "epoch": 0.18796198521647306,
+      "grad_norm": 0.3166196644306183,
+      "learning_rate": 9.300169453500117e-05,
+      "loss": 0.323,
+      "step": 890
+    },
+    {
+      "epoch": 0.18817317845828935,
+      "grad_norm": 0.3596815764904022,
+      "learning_rate": 9.298439278661012e-05,
+      "loss": 0.3165,
+      "step": 891
+    },
+    {
+      "epoch": 0.1883843717001056,
+      "grad_norm": 0.4279139041900635,
+      "learning_rate": 9.296707129076793e-05,
+      "loss": 0.3127,
+      "step": 892
+    },
+    {
+      "epoch": 0.18859556494192187,
+      "grad_norm": 0.4031926393508911,
+      "learning_rate": 9.294973005543228e-05,
+      "loss": 0.3079,
+      "step": 893
+    },
+    {
+      "epoch": 0.18880675818373813,
+      "grad_norm": 0.44852691888809204,
+      "learning_rate": 9.29323690885699e-05,
+      "loss": 0.3155,
+      "step": 894
+    },
+    {
+      "epoch": 0.1890179514255544,
+      "grad_norm": 0.43203306198120117,
+      "learning_rate": 9.291498839815657e-05,
+      "loss": 0.3178,
+      "step": 895
+    },
+    {
+      "epoch": 0.18922914466737065,
+      "grad_norm": 0.44379857182502747,
+      "learning_rate": 9.289758799217715e-05,
+      "loss": 0.3137,
+      "step": 896
+    },
+    {
+      "epoch": 0.1894403379091869,
+      "grad_norm": 0.4251607060432434,
+      "learning_rate": 9.288016787862557e-05,
+      "loss": 0.3129,
+      "step": 897
+    },
+    {
+      "epoch": 0.18965153115100317,
+      "grad_norm": 0.41411468386650085,
+      "learning_rate": 9.28627280655048e-05,
+      "loss": 0.3136,
+      "step": 898
+    },
+    {
+      "epoch": 0.18986272439281943,
+      "grad_norm": 0.49519532918930054,
+      "learning_rate": 9.284526856082685e-05,
+      "loss": 0.3203,
+      "step": 899
+    },
+    {
+      "epoch": 0.1900739176346357,
+      "grad_norm": 0.4332972466945648,
+      "learning_rate": 9.282778937261278e-05,
+      "loss": 0.312,
+      "step": 900
+    },
+    {
+      "epoch": 0.19028511087645195,
+      "grad_norm": 0.4271685779094696,
+      "learning_rate": 9.281029050889274e-05,
+      "loss": 0.3079,
+      "step": 901
+    },
+    {
+      "epoch": 0.1904963041182682,
+      "grad_norm": 0.48702356219291687,
+      "learning_rate": 9.279277197770584e-05,
+      "loss": 0.3125,
+      "step": 902
+    },
+    {
+      "epoch": 0.19070749736008447,
+      "grad_norm": 0.4745144248008728,
+      "learning_rate": 9.277523378710027e-05,
+      "loss": 0.3202,
+      "step": 903
+    },
+    {
+      "epoch": 0.19091869060190073,
+      "grad_norm": 0.42556750774383545,
+      "learning_rate": 9.275767594513327e-05,
+      "loss": 0.3162,
+      "step": 904
+    },
+    {
+      "epoch": 0.191129883843717,
+      "grad_norm": 0.41050592064857483,
+      "learning_rate": 9.274009845987106e-05,
+      "loss": 0.3097,
+      "step": 905
+    },
+    {
+      "epoch": 0.19134107708553327,
+      "grad_norm": 0.5205094218254089,
+      "learning_rate": 9.272250133938892e-05,
+      "loss": 0.3141,
+      "step": 906
+    },
+    {
+      "epoch": 0.19155227032734953,
+      "grad_norm": 0.46741610765457153,
+      "learning_rate": 9.270488459177114e-05,
+      "loss": 0.3112,
+      "step": 907
+    },
+    {
+      "epoch": 0.1917634635691658,
+      "grad_norm": 0.4121737778186798,
+      "learning_rate": 9.268724822511104e-05,
+      "loss": 0.316,
+      "step": 908
+    },
+    {
+      "epoch": 0.19197465681098205,
+      "grad_norm": 0.3711234927177429,
+      "learning_rate": 9.266959224751091e-05,
+      "loss": 0.3241,
+      "step": 909
+    },
+    {
+      "epoch": 0.1921858500527983,
+      "grad_norm": 0.5043278932571411,
+      "learning_rate": 9.265191666708209e-05,
+      "loss": 0.317,
+      "step": 910
+    },
+    {
+      "epoch": 0.19239704329461457,
+      "grad_norm": 0.5299025177955627,
+      "learning_rate": 9.263422149194492e-05,
+      "loss": 0.3211,
+      "step": 911
+    },
+    {
+      "epoch": 0.19260823653643083,
+      "grad_norm": 0.31985560059547424,
+      "learning_rate": 9.261650673022872e-05,
+      "loss": 0.3063,
+      "step": 912
+    },
+    {
+      "epoch": 0.1928194297782471,
+      "grad_norm": 0.5549888610839844,
+      "learning_rate": 9.259877239007186e-05,
+      "loss": 0.3105,
+      "step": 913
+    },
+    {
+      "epoch": 0.19303062302006335,
+      "grad_norm": 0.3656262159347534,
+      "learning_rate": 9.258101847962166e-05,
+      "loss": 0.3083,
+      "step": 914
+    },
+    {
+      "epoch": 0.1932418162618796,
+      "grad_norm": 0.42893877625465393,
+      "learning_rate": 9.25632450070344e-05,
+      "loss": 0.3088,
+      "step": 915
+    },
+    {
+      "epoch": 0.19345300950369587,
+      "grad_norm": 0.46078476309776306,
+      "learning_rate": 9.254545198047544e-05,
+      "loss": 0.3111,
+      "step": 916
+    },
+    {
+      "epoch": 0.19366420274551213,
+      "grad_norm": 0.3873847424983978,
+      "learning_rate": 9.252763940811904e-05,
+      "loss": 0.3071,
+      "step": 917
+    },
+    {
+      "epoch": 0.1938753959873284,
+      "grad_norm": 0.513229250907898,
+      "learning_rate": 9.250980729814846e-05,
+      "loss": 0.3112,
+      "step": 918
+    },
+    {
+      "epoch": 0.19408658922914468,
+      "grad_norm": 0.4906502366065979,
+      "learning_rate": 9.2491955658756e-05,
+      "loss": 0.3133,
+      "step": 919
+    },
+    {
+      "epoch": 0.19429778247096094,
+      "grad_norm": 0.40044915676116943,
+      "learning_rate": 9.247408449814281e-05,
+      "loss": 0.31,
+      "step": 920
+    },
+    {
+      "epoch": 0.1945089757127772,
+      "grad_norm": 0.47862112522125244,
+      "learning_rate": 9.245619382451914e-05,
+      "loss": 0.3087,
+      "step": 921
+    },
+    {
+      "epoch": 0.19472016895459346,
+      "grad_norm": 0.4993024170398712,
+      "learning_rate": 9.243828364610408e-05,
+      "loss": 0.3072,
+      "step": 922
+    },
+    {
+      "epoch": 0.19493136219640972,
+      "grad_norm": 0.5243088006973267,
+      "learning_rate": 9.242035397112578e-05,
+      "loss": 0.3019,
+      "step": 923
+    },
+    {
+      "epoch": 0.19514255543822598,
+      "grad_norm": 0.537132978439331,
+      "learning_rate": 9.24024048078213e-05,
+      "loss": 0.3171,
+      "step": 924
+    },
+    {
+      "epoch": 0.19535374868004224,
+      "grad_norm": 0.634045422077179,
+      "learning_rate": 9.238443616443666e-05,
+      "loss": 0.3125,
+      "step": 925
+    },
+    {
+      "epoch": 0.1955649419218585,
+      "grad_norm": 0.396308958530426,
+      "learning_rate": 9.236644804922683e-05,
+      "loss": 0.3108,
+      "step": 926
+    },
+    {
+      "epoch": 0.19577613516367476,
+      "grad_norm": 0.5975167155265808,
+      "learning_rate": 9.234844047045574e-05,
+      "loss": 0.3144,
+      "step": 927
+    },
+    {
+      "epoch": 0.19598732840549102,
+      "grad_norm": 0.40371406078338623,
+      "learning_rate": 9.233041343639622e-05,
+      "loss": 0.3098,
+      "step": 928
+    },
+    {
+      "epoch": 0.19619852164730728,
+      "grad_norm": 0.4257807731628418,
+      "learning_rate": 9.23123669553301e-05,
+      "loss": 0.3133,
+      "step": 929
+    },
+    {
+      "epoch": 0.19640971488912354,
+      "grad_norm": 0.5098201632499695,
+      "learning_rate": 9.229430103554809e-05,
+      "loss": 0.314,
+      "step": 930
+    },
+    {
+      "epoch": 0.1966209081309398,
+      "grad_norm": 0.4288916289806366,
+      "learning_rate": 9.227621568534984e-05,
+      "loss": 0.3085,
+      "step": 931
+    },
+    {
+      "epoch": 0.19683210137275609,
+      "grad_norm": 0.4888262450695038,
+      "learning_rate": 9.225811091304394e-05,
+      "loss": 0.3143,
+      "step": 932
+    },
+    {
+      "epoch": 0.19704329461457235,
+      "grad_norm": 0.33979111909866333,
+      "learning_rate": 9.22399867269479e-05,
+      "loss": 0.3033,
+      "step": 933
+    },
+    {
+      "epoch": 0.1972544878563886,
+      "grad_norm": 0.4262025058269501,
+      "learning_rate": 9.222184313538819e-05,
+      "loss": 0.3125,
+      "step": 934
+    },
+    {
+      "epoch": 0.19746568109820486,
+      "grad_norm": 0.46783286333084106,
+      "learning_rate": 9.220368014670011e-05,
+      "loss": 0.3117,
+      "step": 935
+    },
+    {
+      "epoch": 0.19767687434002112,
+      "grad_norm": 0.36377832293510437,
+      "learning_rate": 9.218549776922793e-05,
+      "loss": 0.3173,
+      "step": 936
+    },
+    {
+      "epoch": 0.19788806758183738,
+      "grad_norm": 0.44558021426200867,
+      "learning_rate": 9.216729601132481e-05,
+      "loss": 0.3174,
+      "step": 937
+    },
+    {
+      "epoch": 0.19809926082365364,
+      "grad_norm": 0.40159881114959717,
+      "learning_rate": 9.21490748813528e-05,
+      "loss": 0.3064,
+      "step": 938
+    },
+    {
+      "epoch": 0.1983104540654699,
+      "grad_norm": 0.3726361095905304,
+      "learning_rate": 9.21308343876829e-05,
+      "loss": 0.3036,
+      "step": 939
+    },
+    {
+      "epoch": 0.19852164730728616,
+      "grad_norm": 0.38802623748779297,
+      "learning_rate": 9.211257453869494e-05,
+      "loss": 0.3152,
+      "step": 940
+    },
+    {
+      "epoch": 0.19873284054910242,
+      "grad_norm": 0.48056280612945557,
+      "learning_rate": 9.209429534277769e-05,
+      "loss": 0.3138,
+      "step": 941
+    },
+    {
+      "epoch": 0.19894403379091868,
+      "grad_norm": 0.47986841201782227,
+      "learning_rate": 9.207599680832879e-05,
+      "loss": 0.3081,
+      "step": 942
+    },
+    {
+      "epoch": 0.19915522703273494,
+      "grad_norm": 0.4544437825679779,
+      "learning_rate": 9.205767894375475e-05,
+      "loss": 0.3054,
+      "step": 943
+    },
+    {
+      "epoch": 0.1993664202745512,
+      "grad_norm": 0.6503584980964661,
+      "learning_rate": 9.2039341757471e-05,
+      "loss": 0.3057,
+      "step": 944
+    },
+    {
+      "epoch": 0.19957761351636746,
+      "grad_norm": 0.6316682696342468,
+      "learning_rate": 9.202098525790181e-05,
+      "loss": 0.3098,
+      "step": 945
+    },
+    {
+      "epoch": 0.19978880675818375,
+      "grad_norm": 0.5716366767883301,
+      "learning_rate": 9.200260945348034e-05,
+      "loss": 0.3049,
+      "step": 946
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.42491409182548523,
+      "learning_rate": 9.19842143526486e-05,
+      "loss": 0.3091,
+      "step": 947
+    },
+    {
+      "epoch": 0.20021119324181627,
+      "grad_norm": 0.5326544046401978,
+      "learning_rate": 9.196579996385752e-05,
+      "loss": 0.3027,
+      "step": 948
+    },
+    {
+      "epoch": 0.20042238648363253,
+      "grad_norm": 0.4682331383228302,
+      "learning_rate": 9.19473662955668e-05,
+      "loss": 0.3067,
+      "step": 949
+    },
+    {
+      "epoch": 0.2006335797254488,
+      "grad_norm": 0.5115171074867249,
+      "learning_rate": 9.192891335624508e-05,
+      "loss": 0.3038,
+      "step": 950
+    },
+    {
+      "epoch": 0.20084477296726505,
+      "grad_norm": 0.8378593325614929,
+      "learning_rate": 9.191044115436981e-05,
+      "loss": 0.3052,
+      "step": 951
+    },
+    {
+      "epoch": 0.2010559662090813,
+      "grad_norm": 0.5452477931976318,
+      "learning_rate": 9.18919496984273e-05,
+      "loss": 0.3156,
+      "step": 952
+    },
+    {
+      "epoch": 0.20126715945089757,
+      "grad_norm": 0.45884060859680176,
+      "learning_rate": 9.187343899691271e-05,
+      "loss": 0.3149,
+      "step": 953
+    },
+    {
+      "epoch": 0.20147835269271383,
+      "grad_norm": 0.4604043662548065,
+      "learning_rate": 9.185490905833003e-05,
+      "loss": 0.3079,
+      "step": 954
+    },
+    {
+      "epoch": 0.2016895459345301,
+      "grad_norm": 0.4370904862880707,
+      "learning_rate": 9.18363598911921e-05,
+      "loss": 0.3099,
+      "step": 955
+    },
+    {
+      "epoch": 0.20190073917634635,
+      "grad_norm": 0.43059855699539185,
+      "learning_rate": 9.181779150402062e-05,
+      "loss": 0.3103,
+      "step": 956
+    },
+    {
+      "epoch": 0.2021119324181626,
+      "grad_norm": 0.5378050804138184,
+      "learning_rate": 9.179920390534602e-05,
+      "loss": 0.3111,
+      "step": 957
+    },
+    {
+      "epoch": 0.20232312565997887,
+      "grad_norm": 0.4257240295410156,
+      "learning_rate": 9.178059710370768e-05,
+      "loss": 0.3156,
+      "step": 958
+    },
+    {
+      "epoch": 0.20253431890179513,
+      "grad_norm": 0.3500191569328308,
+      "learning_rate": 9.176197110765373e-05,
+      "loss": 0.3091,
+      "step": 959
+    },
+    {
+      "epoch": 0.20274551214361142,
+      "grad_norm": 0.4893301725387573,
+      "learning_rate": 9.174332592574114e-05,
+      "loss": 0.3069,
+      "step": 960
+    },
+    {
+      "epoch": 0.20295670538542768,
+      "grad_norm": 0.7117564082145691,
+      "learning_rate": 9.172466156653569e-05,
+      "loss": 0.3055,
+      "step": 961
+    },
+    {
+      "epoch": 0.20316789862724394,
+      "grad_norm": 0.5905559062957764,
+      "learning_rate": 9.170597803861195e-05,
+      "loss": 0.3049,
+      "step": 962
+    },
+    {
+      "epoch": 0.2033790918690602,
+      "grad_norm": 0.3803396224975586,
+      "learning_rate": 9.168727535055334e-05,
+      "loss": 0.3053,
+      "step": 963
+    },
+    {
+      "epoch": 0.20359028511087646,
+      "grad_norm": 0.533340334892273,
+      "learning_rate": 9.166855351095204e-05,
+      "loss": 0.313,
+      "step": 964
+    },
+    {
+      "epoch": 0.20380147835269272,
+      "grad_norm": 0.45417195558547974,
+      "learning_rate": 9.164981252840907e-05,
+      "loss": 0.3087,
+      "step": 965
+    },
+    {
+      "epoch": 0.20401267159450898,
+      "grad_norm": 0.38196906447410583,
+      "learning_rate": 9.16310524115342e-05,
+      "loss": 0.2973,
+      "step": 966
+    },
+    {
+      "epoch": 0.20422386483632524,
+      "grad_norm": 0.5206608176231384,
+      "learning_rate": 9.161227316894602e-05,
+      "loss": 0.3126,
+      "step": 967
+    },
+    {
+      "epoch": 0.2044350580781415,
+      "grad_norm": 0.3364376723766327,
+      "learning_rate": 9.15934748092719e-05,
+      "loss": 0.3022,
+      "step": 968
+    },
+    {
+      "epoch": 0.20464625131995776,
+      "grad_norm": 0.5518354773521423,
+      "learning_rate": 9.157465734114796e-05,
+      "loss": 0.2985,
+      "step": 969
+    },
+    {
+      "epoch": 0.20485744456177402,
+      "grad_norm": 0.38300836086273193,
+      "learning_rate": 9.155582077321918e-05,
+      "loss": 0.2993,
+      "step": 970
+    },
+    {
+      "epoch": 0.20506863780359028,
+      "grad_norm": 0.3528032898902893,
+      "learning_rate": 9.153696511413923e-05,
+      "loss": 0.3034,
+      "step": 971
+    },
+    {
+      "epoch": 0.20527983104540654,
+      "grad_norm": 0.44244614243507385,
+      "learning_rate": 9.151809037257059e-05,
+      "loss": 0.3117,
+      "step": 972
+    },
+    {
+      "epoch": 0.2054910242872228,
+      "grad_norm": 0.4263085126876831,
+      "learning_rate": 9.149919655718454e-05,
+      "loss": 0.308,
+      "step": 973
+    },
+    {
+      "epoch": 0.20570221752903908,
+      "grad_norm": 0.40139538049697876,
+      "learning_rate": 9.148028367666102e-05,
+      "loss": 0.3033,
+      "step": 974
+    },
+    {
+      "epoch": 0.20591341077085534,
+      "grad_norm": 0.5020092129707336,
+      "learning_rate": 9.146135173968882e-05,
+      "loss": 0.3038,
+      "step": 975
+    },
+    {
+      "epoch": 0.2061246040126716,
+      "grad_norm": 0.44325128197669983,
+      "learning_rate": 9.144240075496545e-05,
+      "loss": 0.3137,
+      "step": 976
+    },
+    {
+      "epoch": 0.20633579725448786,
+      "grad_norm": 0.3613310158252716,
+      "learning_rate": 9.14234307311972e-05,
+      "loss": 0.3108,
+      "step": 977
+    },
+    {
+      "epoch": 0.20654699049630412,
+      "grad_norm": 0.5693209767341614,
+      "learning_rate": 9.140444167709908e-05,
+      "loss": 0.3163,
+      "step": 978
+    },
+    {
+      "epoch": 0.20675818373812038,
+      "grad_norm": 0.44777923822402954,
+      "learning_rate": 9.138543360139483e-05,
+      "loss": 0.3188,
+      "step": 979
+    },
+    {
+      "epoch": 0.20696937697993664,
+      "grad_norm": 0.42752325534820557,
+      "learning_rate": 9.136640651281694e-05,
+      "loss": 0.3235,
+      "step": 980
+    },
+    {
+      "epoch": 0.2071805702217529,
+      "grad_norm": 0.6453577876091003,
+      "learning_rate": 9.134736042010667e-05,
+      "loss": 0.3076,
+      "step": 981
+    },
+    {
+      "epoch": 0.20739176346356916,
+      "grad_norm": 0.34251394867897034,
+      "learning_rate": 9.132829533201397e-05,
+      "loss": 0.3079,
+      "step": 982
+    },
+    {
+      "epoch": 0.20760295670538542,
+      "grad_norm": 0.5311204791069031,
+      "learning_rate": 9.130921125729753e-05,
+      "loss": 0.3074,
+      "step": 983
+    },
+    {
+      "epoch": 0.20781414994720168,
+      "grad_norm": 0.521913468837738,
+      "learning_rate": 9.129010820472475e-05,
+      "loss": 0.3095,
+      "step": 984
+    },
+    {
+      "epoch": 0.20802534318901794,
+      "grad_norm": 0.4575255215167999,
+      "learning_rate": 9.127098618307177e-05,
+      "loss": 0.3068,
+      "step": 985
+    },
+    {
+      "epoch": 0.2082365364308342,
+      "grad_norm": 0.4767809808254242,
+      "learning_rate": 9.125184520112343e-05,
+      "loss": 0.3085,
+      "step": 986
+    },
+    {
+      "epoch": 0.20844772967265046,
+      "grad_norm": 0.4565068781375885,
+      "learning_rate": 9.123268526767328e-05,
+      "loss": 0.3019,
+      "step": 987
+    },
+    {
+      "epoch": 0.20865892291446675,
+      "grad_norm": 0.3910835385322571,
+      "learning_rate": 9.12135063915236e-05,
+      "loss": 0.2998,
+      "step": 988
+    },
+    {
+      "epoch": 0.208870116156283,
+      "grad_norm": 0.45932385325431824,
+      "learning_rate": 9.119430858148532e-05,
+      "loss": 0.306,
+      "step": 989
+    },
+    {
+      "epoch": 0.20908130939809927,
+      "grad_norm": 0.3926941752433777,
+      "learning_rate": 9.117509184637814e-05,
+      "loss": 0.3014,
+      "step": 990
+    },
+    {
+      "epoch": 0.20929250263991553,
+      "grad_norm": 0.4854736328125,
+      "learning_rate": 9.115585619503039e-05,
+      "loss": 0.2983,
+      "step": 991
+    },
+    {
+      "epoch": 0.2095036958817318,
+      "grad_norm": 0.3221428096294403,
+      "learning_rate": 9.113660163627912e-05,
+      "loss": 0.3119,
+      "step": 992
+    },
+    {
+      "epoch": 0.20971488912354805,
+      "grad_norm": 0.4460935592651367,
+      "learning_rate": 9.111732817897007e-05,
+      "loss": 0.3094,
+      "step": 993
+    },
+    {
+      "epoch": 0.2099260823653643,
+      "grad_norm": 0.5592827200889587,
+      "learning_rate": 9.109803583195767e-05,
+      "loss": 0.2982,
+      "step": 994
+    },
+    {
+      "epoch": 0.21013727560718057,
+      "grad_norm": 0.38764703273773193,
+      "learning_rate": 9.107872460410496e-05,
+      "loss": 0.303,
+      "step": 995
+    },
+    {
+      "epoch": 0.21034846884899683,
+      "grad_norm": 0.48559486865997314,
+      "learning_rate": 9.105939450428378e-05,
+      "loss": 0.3039,
+      "step": 996
+    },
+    {
+      "epoch": 0.2105596620908131,
+      "grad_norm": 0.3938203454017639,
+      "learning_rate": 9.104004554137451e-05,
+      "loss": 0.3104,
+      "step": 997
+    },
+    {
+      "epoch": 0.21077085533262935,
+      "grad_norm": 0.4309868812561035,
+      "learning_rate": 9.102067772426628e-05,
+      "loss": 0.3137,
+      "step": 998
+    },
+    {
+      "epoch": 0.2109820485744456,
+      "grad_norm": 0.3919241726398468,
+      "learning_rate": 9.100129106185685e-05,
+      "loss": 0.3015,
+      "step": 999
+    },
+    {
+      "epoch": 0.21119324181626187,
+      "grad_norm": 0.37178322672843933,
+      "learning_rate": 9.098188556305263e-05,
+      "loss": 0.2986,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21119324181626187,
+      "eval_train_split_loss": 0.3661482036113739,
+      "eval_train_split_runtime": 6.4391,
+      "eval_train_split_samples_per_second": 1.553,
+      "eval_train_split_steps_per_second": 0.155,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21140443505807816,
+      "grad_norm": 0.5438383221626282,
+      "learning_rate": 9.096246123676871e-05,
+      "loss": 0.3062,
+      "step": 1001
+    },
+    {
+      "epoch": 0.21161562829989441,
+      "grad_norm": 0.387076199054718,
+      "learning_rate": 9.094301809192882e-05,
+      "loss": 0.2989,
+      "step": 1002
+    },
+    {
+      "epoch": 0.21182682154171067,
+      "grad_norm": 0.541725754737854,
+      "learning_rate": 9.092355613746533e-05,
+      "loss": 0.3075,
+      "step": 1003
+    },
+    {
+      "epoch": 0.21203801478352693,
+      "grad_norm": 0.4548662304878235,
+      "learning_rate": 9.090407538231924e-05,
+      "loss": 0.3086,
+      "step": 1004
+    },
+    {
+      "epoch": 0.2122492080253432,
+      "grad_norm": 0.4895817041397095,
+      "learning_rate": 9.088457583544021e-05,
+      "loss": 0.3108,
+      "step": 1005
+    },
+    {
+      "epoch": 0.21246040126715945,
+      "grad_norm": 0.46272626519203186,
+      "learning_rate": 9.086505750578651e-05,
+      "loss": 0.3091,
+      "step": 1006
+    },
+    {
+      "epoch": 0.21267159450897571,
+      "grad_norm": 0.4470599591732025,
+      "learning_rate": 9.084552040232508e-05,
+      "loss": 0.3072,
+      "step": 1007
+    },
+    {
+      "epoch": 0.21288278775079197,
+      "grad_norm": 0.39361894130706787,
+      "learning_rate": 9.082596453403143e-05,
+      "loss": 0.3063,
+      "step": 1008
+    },
+    {
+      "epoch": 0.21309398099260823,
+      "grad_norm": 0.3631224036216736,
+      "learning_rate": 9.08063899098897e-05,
+      "loss": 0.3041,
+      "step": 1009
+    },
+    {
+      "epoch": 0.2133051742344245,
+      "grad_norm": 0.36913612484931946,
+      "learning_rate": 9.078679653889273e-05,
+      "loss": 0.3038,
+      "step": 1010
+    },
+    {
+      "epoch": 0.21351636747624075,
+      "grad_norm": 0.6241827607154846,
+      "learning_rate": 9.076718443004184e-05,
+      "loss": 0.3113,
+      "step": 1011
+    },
+    {
+      "epoch": 0.213727560718057,
+      "grad_norm": 0.3821466565132141,
+      "learning_rate": 9.074755359234706e-05,
+      "loss": 0.2967,
+      "step": 1012
+    },
+    {
+      "epoch": 0.21393875395987327,
+      "grad_norm": 0.4781816005706787,
+      "learning_rate": 9.072790403482698e-05,
+      "loss": 0.3055,
+      "step": 1013
+    },
+    {
+      "epoch": 0.21414994720168953,
+      "grad_norm": 0.4603241980075836,
+      "learning_rate": 9.07082357665088e-05,
+      "loss": 0.3086,
+      "step": 1014
+    },
+    {
+      "epoch": 0.21436114044350582,
+      "grad_norm": 0.3561108410358429,
+      "learning_rate": 9.068854879642831e-05,
+      "loss": 0.3038,
+      "step": 1015
+    },
+    {
+      "epoch": 0.21457233368532208,
+      "grad_norm": 0.4482893943786621,
+      "learning_rate": 9.066884313362991e-05,
+      "loss": 0.3104,
+      "step": 1016
+    },
+    {
+      "epoch": 0.21478352692713834,
+      "grad_norm": 0.4323103725910187,
+      "learning_rate": 9.064911878716655e-05,
+      "loss": 0.3101,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2149947201689546,
+      "grad_norm": 0.47991809248924255,
+      "learning_rate": 9.062937576609982e-05,
+      "loss": 0.3073,
+      "step": 1018
+    },
+    {
+      "epoch": 0.21520591341077086,
+      "grad_norm": 0.6043739318847656,
+      "learning_rate": 9.060961407949984e-05,
+      "loss": 0.312,
+      "step": 1019
+    },
+    {
+      "epoch": 0.21541710665258712,
+      "grad_norm": 0.41732579469680786,
+      "learning_rate": 9.058983373644532e-05,
+      "loss": 0.3039,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21562829989440338,
+      "grad_norm": 0.42411288619041443,
+      "learning_rate": 9.057003474602356e-05,
+      "loss": 0.3028,
+      "step": 1021
+    },
+    {
+      "epoch": 0.21583949313621964,
+      "grad_norm": 0.5190529227256775,
+      "learning_rate": 9.055021711733039e-05,
+      "loss": 0.3063,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2160506863780359,
+      "grad_norm": 0.45785045623779297,
+      "learning_rate": 9.053038085947021e-05,
+      "loss": 0.3004,
+      "step": 1023
+    },
+    {
+      "epoch": 0.21626187961985216,
+      "grad_norm": 0.47266775369644165,
+      "learning_rate": 9.051052598155604e-05,
+      "loss": 0.3077,
+      "step": 1024
+    },
+    {
+      "epoch": 0.21647307286166842,
+      "grad_norm": 0.4383353292942047,
+      "learning_rate": 9.049065249270935e-05,
+      "loss": 0.3093,
+      "step": 1025
+    },
+    {
+      "epoch": 0.21668426610348468,
+      "grad_norm": 0.42178237438201904,
+      "learning_rate": 9.047076040206027e-05,
+      "loss": 0.3141,
+      "step": 1026
+    },
+    {
+      "epoch": 0.21689545934530094,
+      "grad_norm": 0.4098151922225952,
+      "learning_rate": 9.045084971874738e-05,
+      "loss": 0.3096,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2171066525871172,
+      "grad_norm": 0.507820725440979,
+      "learning_rate": 9.043092045191785e-05,
+      "loss": 0.307,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2173178458289335,
+      "grad_norm": 0.5636327862739563,
+      "learning_rate": 9.04109726107274e-05,
+      "loss": 0.3049,
+      "step": 1029
+    },
+    {
+      "epoch": 0.21752903907074975,
+      "grad_norm": 0.46158289909362793,
+      "learning_rate": 9.039100620434025e-05,
+      "loss": 0.3032,
+      "step": 1030
+    },
+    {
+      "epoch": 0.217740232312566,
+      "grad_norm": 0.42042306065559387,
+      "learning_rate": 9.037102124192916e-05,
+      "loss": 0.3111,
+      "step": 1031
+    },
+    {
+      "epoch": 0.21795142555438227,
+      "grad_norm": 0.44014430046081543,
+      "learning_rate": 9.03510177326754e-05,
+      "loss": 0.3026,
+      "step": 1032
+    },
+    {
+      "epoch": 0.21816261879619853,
+      "grad_norm": 0.330649733543396,
+      "learning_rate": 9.033099568576883e-05,
+      "loss": 0.301,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2183738120380148,
+      "grad_norm": 0.4007191061973572,
+      "learning_rate": 9.031095511040772e-05,
+      "loss": 0.3093,
+      "step": 1034
+    },
+    {
+      "epoch": 0.21858500527983105,
+      "grad_norm": 0.4331982433795929,
+      "learning_rate": 9.029089601579894e-05,
+      "loss": 0.3034,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2187961985216473,
+      "grad_norm": 0.3965606391429901,
+      "learning_rate": 9.027081841115783e-05,
+      "loss": 0.2986,
+      "step": 1036
+    },
+    {
+      "epoch": 0.21900739176346357,
+      "grad_norm": 0.41323423385620117,
+      "learning_rate": 9.025072230570823e-05,
+      "loss": 0.3013,
+      "step": 1037
+    },
+    {
+      "epoch": 0.21921858500527983,
+      "grad_norm": 0.46276599168777466,
+      "learning_rate": 9.023060770868248e-05,
+      "loss": 0.3009,
+      "step": 1038
+    },
+    {
+      "epoch": 0.21942977824709609,
+      "grad_norm": 0.3755812644958496,
+      "learning_rate": 9.021047462932145e-05,
+      "loss": 0.2975,
+      "step": 1039
+    },
+    {
+      "epoch": 0.21964097148891235,
+      "grad_norm": 0.38216447830200195,
+      "learning_rate": 9.019032307687445e-05,
+      "loss": 0.2974,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2198521647307286,
+      "grad_norm": 0.46699321269989014,
+      "learning_rate": 9.017015306059931e-05,
+      "loss": 0.3059,
+      "step": 1041
+    },
+    {
+      "epoch": 0.22006335797254487,
+      "grad_norm": 0.35265669226646423,
+      "learning_rate": 9.014996458976232e-05,
+      "loss": 0.3039,
+      "step": 1042
+    },
+    {
+      "epoch": 0.22027455121436115,
+      "grad_norm": 0.40066736936569214,
+      "learning_rate": 9.01297576736383e-05,
+      "loss": 0.2993,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2204857444561774,
+      "grad_norm": 0.43739432096481323,
+      "learning_rate": 9.010953232151045e-05,
+      "loss": 0.3064,
+      "step": 1044
+    },
+    {
+      "epoch": 0.22069693769799367,
+      "grad_norm": 0.345170795917511,
+      "learning_rate": 9.008928854267054e-05,
+      "loss": 0.3064,
+      "step": 1045
+    },
+    {
+      "epoch": 0.22090813093980993,
+      "grad_norm": 0.34310269355773926,
+      "learning_rate": 9.006902634641873e-05,
+      "loss": 0.3011,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2211193241816262,
+      "grad_norm": 0.5596431493759155,
+      "learning_rate": 9.00487457420637e-05,
+      "loss": 0.3038,
+      "step": 1047
+    },
+    {
+      "epoch": 0.22133051742344245,
+      "grad_norm": 0.3156079947948456,
+      "learning_rate": 9.002844673892252e-05,
+      "loss": 0.3042,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2215417106652587,
+      "grad_norm": 0.5299674868583679,
+      "learning_rate": 9.000812934632079e-05,
+      "loss": 0.3062,
+      "step": 1049
+    },
+    {
+      "epoch": 0.22175290390707497,
+      "grad_norm": 0.4766310751438141,
+      "learning_rate": 8.99877935735925e-05,
+      "loss": 0.2995,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22196409714889123,
+      "grad_norm": 0.40622851252555847,
+      "learning_rate": 8.996743943008011e-05,
+      "loss": 0.3035,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2221752903907075,
+      "grad_norm": 0.44834914803504944,
+      "learning_rate": 8.99470669251345e-05,
+      "loss": 0.3034,
+      "step": 1052
+    },
+    {
+      "epoch": 0.22238648363252375,
+      "grad_norm": 0.4130789637565613,
+      "learning_rate": 8.9926676068115e-05,
+      "loss": 0.3022,
+      "step": 1053
+    },
+    {
+      "epoch": 0.22259767687434,
+      "grad_norm": 0.3828883469104767,
+      "learning_rate": 8.990626686838938e-05,
+      "loss": 0.3054,
+      "step": 1054
+    },
+    {
+      "epoch": 0.22280887011615627,
+      "grad_norm": 0.4701351225376129,
+      "learning_rate": 8.988583933533383e-05,
+      "loss": 0.3032,
+      "step": 1055
+    },
+    {
+      "epoch": 0.22302006335797253,
+      "grad_norm": 0.3285447657108307,
+      "learning_rate": 8.986539347833296e-05,
+      "loss": 0.2979,
+      "step": 1056
+    },
+    {
+      "epoch": 0.22323125659978882,
+      "grad_norm": 0.38936424255371094,
+      "learning_rate": 8.984492930677979e-05,
+      "loss": 0.3046,
+      "step": 1057
+    },
+    {
+      "epoch": 0.22344244984160508,
+      "grad_norm": 0.4038061499595642,
+      "learning_rate": 8.982444683007573e-05,
+      "loss": 0.3075,
+      "step": 1058
+    },
+    {
+      "epoch": 0.22365364308342134,
+      "grad_norm": 0.42253637313842773,
+      "learning_rate": 8.98039460576307e-05,
+      "loss": 0.3041,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2238648363252376,
+      "grad_norm": 0.36475831270217896,
+      "learning_rate": 8.978342699886288e-05,
+      "loss": 0.3034,
+      "step": 1060
+    },
+    {
+      "epoch": 0.22407602956705386,
+      "grad_norm": 0.4150843918323517,
+      "learning_rate": 8.976288966319899e-05,
+      "loss": 0.3057,
+      "step": 1061
+    },
+    {
+      "epoch": 0.22428722280887012,
+      "grad_norm": 0.4708026945590973,
+      "learning_rate": 8.974233406007405e-05,
+      "loss": 0.3042,
+      "step": 1062
+    },
+    {
+      "epoch": 0.22449841605068638,
+      "grad_norm": 0.37599995732307434,
+      "learning_rate": 8.972176019893149e-05,
+      "loss": 0.3029,
+      "step": 1063
+    },
+    {
+      "epoch": 0.22470960929250264,
+      "grad_norm": 0.3879624307155609,
+      "learning_rate": 8.970116808922318e-05,
+      "loss": 0.3024,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2249208025343189,
+      "grad_norm": 0.38679739832878113,
+      "learning_rate": 8.968055774040933e-05,
+      "loss": 0.2985,
+      "step": 1065
+    },
+    {
+      "epoch": 0.22513199577613516,
+      "grad_norm": 0.3342354893684387,
+      "learning_rate": 8.96599291619585e-05,
+      "loss": 0.3,
+      "step": 1066
+    },
+    {
+      "epoch": 0.22534318901795142,
+      "grad_norm": 0.4106636643409729,
+      "learning_rate": 8.963928236334771e-05,
+      "loss": 0.2973,
+      "step": 1067
+    },
+    {
+      "epoch": 0.22555438225976768,
+      "grad_norm": 0.3167661130428314,
+      "learning_rate": 8.961861735406229e-05,
+      "loss": 0.3011,
+      "step": 1068
+    },
+    {
+      "epoch": 0.22576557550158394,
+      "grad_norm": 0.4498090147972107,
+      "learning_rate": 8.959793414359592e-05,
+      "loss": 0.3051,
+      "step": 1069
+    },
+    {
+      "epoch": 0.22597676874340022,
+      "grad_norm": 0.5876569747924805,
+      "learning_rate": 8.957723274145069e-05,
+      "loss": 0.3018,
+      "step": 1070
+    },
+    {
+      "epoch": 0.22618796198521648,
+      "grad_norm": 0.5624187588691711,
+      "learning_rate": 8.955651315713702e-05,
+      "loss": 0.2991,
+      "step": 1071
+    },
+    {
+      "epoch": 0.22639915522703274,
+      "grad_norm": 0.3975844979286194,
+      "learning_rate": 8.95357754001737e-05,
+      "loss": 0.3025,
+      "step": 1072
+    },
+    {
+      "epoch": 0.226610348468849,
+      "grad_norm": 0.361043244600296,
+      "learning_rate": 8.951501948008786e-05,
+      "loss": 0.2992,
+      "step": 1073
+    },
+    {
+      "epoch": 0.22682154171066526,
+      "grad_norm": 0.3400260806083679,
+      "learning_rate": 8.949424540641495e-05,
+      "loss": 0.2914,
+      "step": 1074
+    },
+    {
+      "epoch": 0.22703273495248152,
+      "grad_norm": 0.4212849736213684,
+      "learning_rate": 8.947345318869882e-05,
+      "loss": 0.3043,
+      "step": 1075
+    },
+    {
+      "epoch": 0.22724392819429778,
+      "grad_norm": 0.3398955464363098,
+      "learning_rate": 8.945264283649158e-05,
+      "loss": 0.3043,
+      "step": 1076
+    },
+    {
+      "epoch": 0.22745512143611404,
+      "grad_norm": 0.36498767137527466,
+      "learning_rate": 8.943181435935373e-05,
+      "loss": 0.298,
+      "step": 1077
+    },
+    {
+      "epoch": 0.2276663146779303,
+      "grad_norm": 0.3562159240245819,
+      "learning_rate": 8.941096776685408e-05,
+      "loss": 0.2946,
+      "step": 1078
+    },
+    {
+      "epoch": 0.22787750791974656,
+      "grad_norm": 0.33870917558670044,
+      "learning_rate": 8.939010306856975e-05,
+      "loss": 0.2991,
+      "step": 1079
+    },
+    {
+      "epoch": 0.22808870116156282,
+      "grad_norm": 0.334480345249176,
+      "learning_rate": 8.936922027408618e-05,
+      "loss": 0.3002,
+      "step": 1080
+    },
+    {
+      "epoch": 0.22829989440337908,
+      "grad_norm": 0.36234697699546814,
+      "learning_rate": 8.934831939299714e-05,
+      "loss": 0.2941,
+      "step": 1081
+    },
+    {
+      "epoch": 0.22851108764519534,
+      "grad_norm": 0.40647614002227783,
+      "learning_rate": 8.932740043490472e-05,
+      "loss": 0.3015,
+      "step": 1082
+    },
+    {
+      "epoch": 0.2287222808870116,
+      "grad_norm": 0.3645108938217163,
+      "learning_rate": 8.930646340941927e-05,
+      "loss": 0.3004,
+      "step": 1083
+    },
+    {
+      "epoch": 0.2289334741288279,
+      "grad_norm": 0.4460848569869995,
+      "learning_rate": 8.928550832615945e-05,
+      "loss": 0.296,
+      "step": 1084
+    },
+    {
+      "epoch": 0.22914466737064415,
+      "grad_norm": 0.48058050870895386,
+      "learning_rate": 8.926453519475225e-05,
+      "loss": 0.3045,
+      "step": 1085
+    },
+    {
+      "epoch": 0.2293558606124604,
+      "grad_norm": 0.4843015968799591,
+      "learning_rate": 8.924354402483293e-05,
+      "loss": 0.3005,
+      "step": 1086
+    },
+    {
+      "epoch": 0.22956705385427667,
+      "grad_norm": 0.5331267714500427,
+      "learning_rate": 8.922253482604506e-05,
+      "loss": 0.3017,
+      "step": 1087
+    },
+    {
+      "epoch": 0.22977824709609293,
+      "grad_norm": 0.4366138279438019,
+      "learning_rate": 8.920150760804043e-05,
+      "loss": 0.3035,
+      "step": 1088
+    },
+    {
+      "epoch": 0.2299894403379092,
+      "grad_norm": 0.4204513132572174,
+      "learning_rate": 8.918046238047917e-05,
+      "loss": 0.2988,
+      "step": 1089
+    },
+    {
+      "epoch": 0.23020063357972545,
+      "grad_norm": 0.5362744927406311,
+      "learning_rate": 8.915939915302968e-05,
+      "loss": 0.2974,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2304118268215417,
+      "grad_norm": 0.4959648549556732,
+      "learning_rate": 8.91383179353686e-05,
+      "loss": 0.2908,
+      "step": 1091
+    },
+    {
+      "epoch": 0.23062302006335797,
+      "grad_norm": 0.6338536739349365,
+      "learning_rate": 8.911721873718083e-05,
+      "loss": 0.3004,
+      "step": 1092
+    },
+    {
+      "epoch": 0.23083421330517423,
+      "grad_norm": 0.4491744935512543,
+      "learning_rate": 8.909610156815958e-05,
+      "loss": 0.2935,
+      "step": 1093
+    },
+    {
+      "epoch": 0.2310454065469905,
+      "grad_norm": 0.6652294397354126,
+      "learning_rate": 8.907496643800626e-05,
+      "loss": 0.3024,
+      "step": 1094
+    },
+    {
+      "epoch": 0.23125659978880675,
+      "grad_norm": 0.39966198801994324,
+      "learning_rate": 8.905381335643055e-05,
+      "loss": 0.2964,
+      "step": 1095
+    },
+    {
+      "epoch": 0.231467793030623,
+      "grad_norm": 0.6586857438087463,
+      "learning_rate": 8.903264233315041e-05,
+      "loss": 0.3018,
+      "step": 1096
+    },
+    {
+      "epoch": 0.23167898627243927,
+      "grad_norm": 0.49340805411338806,
+      "learning_rate": 8.901145337789201e-05,
+      "loss": 0.2965,
+      "step": 1097
+    },
+    {
+      "epoch": 0.23189017951425556,
+      "grad_norm": 0.5592809319496155,
+      "learning_rate": 8.899024650038975e-05,
+      "loss": 0.3029,
+      "step": 1098
+    },
+    {
+      "epoch": 0.23210137275607182,
+      "grad_norm": 0.6285479068756104,
+      "learning_rate": 8.896902171038628e-05,
+      "loss": 0.3023,
+      "step": 1099
+    },
+    {
+      "epoch": 0.23231256599788808,
+      "grad_norm": 0.6789811253547668,
+      "learning_rate": 8.89477790176325e-05,
+      "loss": 0.3018,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23252375923970434,
+      "grad_norm": 0.4868791699409485,
+      "learning_rate": 8.892651843188746e-05,
+      "loss": 0.2907,
+      "step": 1101
+    },
+    {
+      "epoch": 0.2327349524815206,
+      "grad_norm": 0.4807937443256378,
+      "learning_rate": 8.890523996291853e-05,
+      "loss": 0.3016,
+      "step": 1102
+    },
+    {
+      "epoch": 0.23294614572333686,
+      "grad_norm": 0.3784332871437073,
+      "learning_rate": 8.888394362050123e-05,
+      "loss": 0.2934,
+      "step": 1103
+    },
+    {
+      "epoch": 0.23315733896515312,
+      "grad_norm": 0.4823589324951172,
+      "learning_rate": 8.886262941441932e-05,
+      "loss": 0.3008,
+      "step": 1104
+    },
+    {
+      "epoch": 0.23336853220696938,
+      "grad_norm": 0.477171391248703,
+      "learning_rate": 8.884129735446471e-05,
+      "loss": 0.3004,
+      "step": 1105
+    },
+    {
+      "epoch": 0.23357972544878564,
+      "grad_norm": 0.5387280583381653,
+      "learning_rate": 8.881994745043762e-05,
+      "loss": 0.308,
+      "step": 1106
+    },
+    {
+      "epoch": 0.2337909186906019,
+      "grad_norm": 0.4201565086841583,
+      "learning_rate": 8.879857971214637e-05,
+      "loss": 0.3011,
+      "step": 1107
+    },
+    {
+      "epoch": 0.23400211193241816,
+      "grad_norm": 0.48487338423728943,
+      "learning_rate": 8.877719414940751e-05,
+      "loss": 0.3024,
+      "step": 1108
+    },
+    {
+      "epoch": 0.23421330517423442,
+      "grad_norm": 0.4532690942287445,
+      "learning_rate": 8.875579077204577e-05,
+      "loss": 0.3063,
+      "step": 1109
+    },
+    {
+      "epoch": 0.23442449841605068,
+      "grad_norm": 0.42182812094688416,
+      "learning_rate": 8.87343695898941e-05,
+      "loss": 0.2997,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23463569165786693,
+      "grad_norm": 0.3504948616027832,
+      "learning_rate": 8.871293061279357e-05,
+      "loss": 0.3016,
+      "step": 1111
+    },
+    {
+      "epoch": 0.23484688489968322,
+      "grad_norm": 0.41274547576904297,
+      "learning_rate": 8.869147385059348e-05,
+      "loss": 0.3016,
+      "step": 1112
+    },
+    {
+      "epoch": 0.23505807814149948,
+      "grad_norm": 1.416022539138794,
+      "learning_rate": 8.866999931315127e-05,
+      "loss": 0.3007,
+      "step": 1113
+    },
+    {
+      "epoch": 0.23526927138331574,
+      "grad_norm": 0.5758169293403625,
+      "learning_rate": 8.864850701033254e-05,
+      "loss": 0.312,
+      "step": 1114
+    },
+    {
+      "epoch": 0.235480464625132,
+      "grad_norm": 0.4731518030166626,
+      "learning_rate": 8.862699695201106e-05,
+      "loss": 0.3015,
+      "step": 1115
+    },
+    {
+      "epoch": 0.23569165786694826,
+      "grad_norm": 0.34466761350631714,
+      "learning_rate": 8.860546914806879e-05,
+      "loss": 0.304,
+      "step": 1116
+    },
+    {
+      "epoch": 0.23590285110876452,
+      "grad_norm": 0.49389195442199707,
+      "learning_rate": 8.858392360839581e-05,
+      "loss": 0.3055,
+      "step": 1117
+    },
+    {
+      "epoch": 0.23611404435058078,
+      "grad_norm": 0.41955575346946716,
+      "learning_rate": 8.856236034289032e-05,
+      "loss": 0.3027,
+      "step": 1118
+    },
+    {
+      "epoch": 0.23632523759239704,
+      "grad_norm": 0.4470333456993103,
+      "learning_rate": 8.854077936145873e-05,
+      "loss": 0.3097,
+      "step": 1119
+    },
+    {
+      "epoch": 0.2365364308342133,
+      "grad_norm": 0.4475184679031372,
+      "learning_rate": 8.851918067401553e-05,
+      "loss": 0.3024,
+      "step": 1120
+    },
+    {
+      "epoch": 0.23674762407602956,
+      "grad_norm": 0.39979448914527893,
+      "learning_rate": 8.849756429048339e-05,
+      "loss": 0.3084,
+      "step": 1121
+    },
+    {
+      "epoch": 0.23695881731784582,
+      "grad_norm": 0.4675282835960388,
+      "learning_rate": 8.847593022079306e-05,
+      "loss": 0.3116,
+      "step": 1122
+    },
+    {
+      "epoch": 0.23717001055966208,
+      "grad_norm": 0.40124183893203735,
+      "learning_rate": 8.845427847488348e-05,
+      "loss": 0.3071,
+      "step": 1123
+    },
+    {
+      "epoch": 0.23738120380147834,
+      "grad_norm": 0.37363532185554504,
+      "learning_rate": 8.843260906270165e-05,
+      "loss": 0.2947,
+      "step": 1124
+    },
+    {
+      "epoch": 0.2375923970432946,
+      "grad_norm": 0.42004406452178955,
+      "learning_rate": 8.841092199420271e-05,
+      "loss": 0.2942,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2378035902851109,
+      "grad_norm": 0.3261277377605438,
+      "learning_rate": 8.838921727934992e-05,
+      "loss": 0.2942,
+      "step": 1126
+    },
+    {
+      "epoch": 0.23801478352692715,
+      "grad_norm": 0.4796265959739685,
+      "learning_rate": 8.836749492811462e-05,
+      "loss": 0.2993,
+      "step": 1127
+    },
+    {
+      "epoch": 0.2382259767687434,
+      "grad_norm": 0.3944534659385681,
+      "learning_rate": 8.834575495047632e-05,
+      "loss": 0.2963,
+      "step": 1128
+    },
+    {
+      "epoch": 0.23843717001055967,
+      "grad_norm": 0.808988630771637,
+      "learning_rate": 8.832399735642252e-05,
+      "loss": 0.3106,
+      "step": 1129
+    },
+    {
+      "epoch": 0.23864836325237593,
+      "grad_norm": 0.43221452832221985,
+      "learning_rate": 8.83022221559489e-05,
+      "loss": 0.3096,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2388595564941922,
+      "grad_norm": 0.3803726136684418,
+      "learning_rate": 8.82804293590592e-05,
+      "loss": 0.297,
+      "step": 1131
+    },
+    {
+      "epoch": 0.23907074973600845,
+      "grad_norm": 0.37804630398750305,
+      "learning_rate": 8.825861897576526e-05,
+      "loss": 0.2974,
+      "step": 1132
+    },
+    {
+      "epoch": 0.2392819429778247,
+      "grad_norm": 0.33442267775535583,
+      "learning_rate": 8.823679101608696e-05,
+      "loss": 0.2994,
+      "step": 1133
+    },
+    {
+      "epoch": 0.23949313621964097,
+      "grad_norm": 0.46652498841285706,
+      "learning_rate": 8.821494549005226e-05,
+      "loss": 0.3016,
+      "step": 1134
+    },
+    {
+      "epoch": 0.23970432946145723,
+      "grad_norm": 0.34999969601631165,
+      "learning_rate": 8.819308240769724e-05,
+      "loss": 0.292,
+      "step": 1135
+    },
+    {
+      "epoch": 0.2399155227032735,
+      "grad_norm": 0.3796578645706177,
+      "learning_rate": 8.817120177906601e-05,
+      "loss": 0.2986,
+      "step": 1136
+    },
+    {
+      "epoch": 0.24012671594508975,
+      "grad_norm": 0.3523526191711426,
+      "learning_rate": 8.814930361421077e-05,
+      "loss": 0.2941,
+      "step": 1137
+    },
+    {
+      "epoch": 0.240337909186906,
+      "grad_norm": 0.38143569231033325,
+      "learning_rate": 8.812738792319169e-05,
+      "loss": 0.2974,
+      "step": 1138
+    },
+    {
+      "epoch": 0.2405491024287223,
+      "grad_norm": 0.4256843626499176,
+      "learning_rate": 8.81054547160771e-05,
+      "loss": 0.2995,
+      "step": 1139
+    },
+    {
+      "epoch": 0.24076029567053855,
+      "grad_norm": 0.3317326605319977,
+      "learning_rate": 8.808350400294332e-05,
+      "loss": 0.294,
+      "step": 1140
+    },
+    {
+      "epoch": 0.24097148891235481,
+      "grad_norm": 0.320777952671051,
+      "learning_rate": 8.80615357938747e-05,
+      "loss": 0.2922,
+      "step": 1141
+    },
+    {
+      "epoch": 0.24118268215417107,
+      "grad_norm": 0.4781387150287628,
+      "learning_rate": 8.80395500989637e-05,
+      "loss": 0.2846,
+      "step": 1142
+    },
+    {
+      "epoch": 0.24139387539598733,
+      "grad_norm": 0.3685510754585266,
+      "learning_rate": 8.801754692831073e-05,
+      "loss": 0.3023,
+      "step": 1143
+    },
+    {
+      "epoch": 0.2416050686378036,
+      "grad_norm": 0.36746513843536377,
+      "learning_rate": 8.799552629202424e-05,
+      "loss": 0.298,
+      "step": 1144
+    },
+    {
+      "epoch": 0.24181626187961985,
+      "grad_norm": 0.4832763671875,
+      "learning_rate": 8.797348820022078e-05,
+      "loss": 0.2983,
+      "step": 1145
+    },
+    {
+      "epoch": 0.2420274551214361,
+      "grad_norm": 0.368178129196167,
+      "learning_rate": 8.795143266302482e-05,
+      "loss": 0.2976,
+      "step": 1146
+    },
+    {
+      "epoch": 0.24223864836325237,
+      "grad_norm": 0.36382707953453064,
+      "learning_rate": 8.79293596905689e-05,
+      "loss": 0.2942,
+      "step": 1147
+    },
+    {
+      "epoch": 0.24244984160506863,
+      "grad_norm": 0.3757737874984741,
+      "learning_rate": 8.790726929299358e-05,
+      "loss": 0.2973,
+      "step": 1148
+    },
+    {
+      "epoch": 0.2426610348468849,
+      "grad_norm": 0.36132925748825073,
+      "learning_rate": 8.78851614804474e-05,
+      "loss": 0.3002,
+      "step": 1149
+    },
+    {
+      "epoch": 0.24287222808870115,
+      "grad_norm": 0.3548225462436676,
+      "learning_rate": 8.786303626308689e-05,
+      "loss": 0.3001,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2430834213305174,
+      "grad_norm": 0.42585113644599915,
+      "learning_rate": 8.78408936510766e-05,
+      "loss": 0.302,
+      "step": 1151
+    },
+    {
+      "epoch": 0.24329461457233367,
+      "grad_norm": 0.40330666303634644,
+      "learning_rate": 8.781873365458906e-05,
+      "loss": 0.2997,
+      "step": 1152
+    },
+    {
+      "epoch": 0.24350580781414996,
+      "grad_norm": 0.33278578519821167,
+      "learning_rate": 8.779655628380478e-05,
+      "loss": 0.2974,
+      "step": 1153
+    },
+    {
+      "epoch": 0.24371700105596622,
+      "grad_norm": 0.44139373302459717,
+      "learning_rate": 8.77743615489123e-05,
+      "loss": 0.2979,
+      "step": 1154
+    },
+    {
+      "epoch": 0.24392819429778248,
+      "grad_norm": 0.36429286003112793,
+      "learning_rate": 8.775214946010806e-05,
+      "loss": 0.2912,
+      "step": 1155
+    },
+    {
+      "epoch": 0.24413938753959874,
+      "grad_norm": 0.28980696201324463,
+      "learning_rate": 8.772992002759652e-05,
+      "loss": 0.2949,
+      "step": 1156
+    },
+    {
+      "epoch": 0.244350580781415,
+      "grad_norm": 0.34710177779197693,
+      "learning_rate": 8.77076732615901e-05,
+      "loss": 0.2971,
+      "step": 1157
+    },
+    {
+      "epoch": 0.24456177402323126,
+      "grad_norm": 0.4039250314235687,
+      "learning_rate": 8.768540917230918e-05,
+      "loss": 0.2921,
+      "step": 1158
+    },
+    {
+      "epoch": 0.24477296726504752,
+      "grad_norm": 0.3574005365371704,
+      "learning_rate": 8.766312776998211e-05,
+      "loss": 0.2911,
+      "step": 1159
+    },
+    {
+      "epoch": 0.24498416050686378,
+      "grad_norm": 0.3589351773262024,
+      "learning_rate": 8.764082906484519e-05,
+      "loss": 0.2964,
+      "step": 1160
+    },
+    {
+      "epoch": 0.24519535374868004,
+      "grad_norm": 0.36039379239082336,
+      "learning_rate": 8.761851306714263e-05,
+      "loss": 0.3006,
+      "step": 1161
+    },
+    {
+      "epoch": 0.2454065469904963,
+      "grad_norm": 0.3660932183265686,
+      "learning_rate": 8.759617978712667e-05,
+      "loss": 0.2906,
+      "step": 1162
+    },
+    {
+      "epoch": 0.24561774023231256,
+      "grad_norm": 0.6778573989868164,
+      "learning_rate": 8.75738292350574e-05,
+      "loss": 0.2974,
+      "step": 1163
+    },
+    {
+      "epoch": 0.24582893347412882,
+      "grad_norm": 0.5603726506233215,
+      "learning_rate": 8.755146142120287e-05,
+      "loss": 0.3055,
+      "step": 1164
+    },
+    {
+      "epoch": 0.24604012671594508,
+      "grad_norm": 0.44827595353126526,
+      "learning_rate": 8.752907635583911e-05,
+      "loss": 0.3022,
+      "step": 1165
+    },
+    {
+      "epoch": 0.24625131995776134,
+      "grad_norm": 0.48088130354881287,
+      "learning_rate": 8.750667404925002e-05,
+      "loss": 0.2977,
+      "step": 1166
+    },
+    {
+      "epoch": 0.24646251319957763,
+      "grad_norm": 0.40855151414871216,
+      "learning_rate": 8.748425451172745e-05,
+      "loss": 0.3003,
+      "step": 1167
+    },
+    {
+      "epoch": 0.2466737064413939,
+      "grad_norm": 0.5142477750778198,
+      "learning_rate": 8.746181775357114e-05,
+      "loss": 0.2983,
+      "step": 1168
+    },
+    {
+      "epoch": 0.24688489968321015,
+      "grad_norm": 0.3854672312736511,
+      "learning_rate": 8.743936378508876e-05,
+      "loss": 0.3019,
+      "step": 1169
+    },
+    {
+      "epoch": 0.2470960929250264,
+      "grad_norm": 0.5618686676025391,
+      "learning_rate": 8.741689261659589e-05,
+      "loss": 0.2989,
+      "step": 1170
+    },
+    {
+      "epoch": 0.24730728616684267,
+      "grad_norm": 0.3572016954421997,
+      "learning_rate": 8.739440425841599e-05,
+      "loss": 0.2981,
+      "step": 1171
+    },
+    {
+      "epoch": 0.24751847940865893,
+      "grad_norm": 0.4637525975704193,
+      "learning_rate": 8.737189872088046e-05,
+      "loss": 0.3027,
+      "step": 1172
+    },
+    {
+      "epoch": 0.24772967265047519,
+      "grad_norm": 0.39551347494125366,
+      "learning_rate": 8.734937601432853e-05,
+      "loss": 0.2959,
+      "step": 1173
+    },
+    {
+      "epoch": 0.24794086589229145,
+      "grad_norm": 0.46402719616889954,
+      "learning_rate": 8.732683614910739e-05,
+      "loss": 0.2979,
+      "step": 1174
+    },
+    {
+      "epoch": 0.2481520591341077,
+      "grad_norm": 0.41231870651245117,
+      "learning_rate": 8.730427913557204e-05,
+      "loss": 0.2934,
+      "step": 1175
+    },
+    {
+      "epoch": 0.24836325237592397,
+      "grad_norm": 0.36750203371047974,
+      "learning_rate": 8.728170498408541e-05,
+      "loss": 0.2935,
+      "step": 1176
+    },
+    {
+      "epoch": 0.24857444561774023,
+      "grad_norm": 0.5125410556793213,
+      "learning_rate": 8.725911370501829e-05,
+      "loss": 0.2955,
+      "step": 1177
+    },
+    {
+      "epoch": 0.24878563885955648,
+      "grad_norm": 0.48806118965148926,
+      "learning_rate": 8.723650530874933e-05,
+      "loss": 0.2954,
+      "step": 1178
+    },
+    {
+      "epoch": 0.24899683210137274,
+      "grad_norm": 0.5123559832572937,
+      "learning_rate": 8.721387980566504e-05,
+      "loss": 0.2932,
+      "step": 1179
+    },
+    {
+      "epoch": 0.249208025343189,
+      "grad_norm": 0.6568512916564941,
+      "learning_rate": 8.71912372061598e-05,
+      "loss": 0.3028,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2494192185850053,
+      "grad_norm": 0.48580092191696167,
+      "learning_rate": 8.716857752063586e-05,
+      "loss": 0.2985,
+      "step": 1181
+    },
+    {
+      "epoch": 0.24963041182682155,
+      "grad_norm": 0.3478964865207672,
+      "learning_rate": 8.714590075950329e-05,
+      "loss": 0.3002,
+      "step": 1182
+    },
+    {
+      "epoch": 0.2498416050686378,
+      "grad_norm": 0.43428584933280945,
+      "learning_rate": 8.712320693318001e-05,
+      "loss": 0.2918,
+      "step": 1183
+    },
+    {
+      "epoch": 0.25005279831045407,
+      "grad_norm": 0.4001418948173523,
+      "learning_rate": 8.710049605209178e-05,
+      "loss": 0.2894,
+      "step": 1184
+    },
+    {
+      "epoch": 0.2502639915522703,
+      "grad_norm": 0.4972737729549408,
+      "learning_rate": 8.707776812667223e-05,
+      "loss": 0.2924,
+      "step": 1185
+    },
+    {
+      "epoch": 0.2504751847940866,
+      "grad_norm": 0.3703025281429291,
+      "learning_rate": 8.705502316736276e-05,
+      "loss": 0.2964,
+      "step": 1186
+    },
+    {
+      "epoch": 0.2506863780359028,
+      "grad_norm": 0.3990064263343811,
+      "learning_rate": 8.703226118461265e-05,
+      "loss": 0.2947,
+      "step": 1187
+    },
+    {
+      "epoch": 0.2508975712777191,
+      "grad_norm": 0.36676833033561707,
+      "learning_rate": 8.700948218887896e-05,
+      "loss": 0.2942,
+      "step": 1188
+    },
+    {
+      "epoch": 0.2511087645195354,
+      "grad_norm": 0.4371195137500763,
+      "learning_rate": 8.69866861906266e-05,
+      "loss": 0.2976,
+      "step": 1189
+    },
+    {
+      "epoch": 0.25131995776135163,
+      "grad_norm": 0.4770589768886566,
+      "learning_rate": 8.696387320032827e-05,
+      "loss": 0.2977,
+      "step": 1190
+    },
+    {
+      "epoch": 0.2515311510031679,
+      "grad_norm": 0.40124502778053284,
+      "learning_rate": 8.694104322846448e-05,
+      "loss": 0.2943,
+      "step": 1191
+    },
+    {
+      "epoch": 0.25174234424498415,
+      "grad_norm": 0.4722228944301605,
+      "learning_rate": 8.691819628552353e-05,
+      "loss": 0.2993,
+      "step": 1192
+    },
+    {
+      "epoch": 0.25195353748680044,
+      "grad_norm": 0.43731167912483215,
+      "learning_rate": 8.689533238200156e-05,
+      "loss": 0.2947,
+      "step": 1193
+    },
+    {
+      "epoch": 0.25216473072861667,
+      "grad_norm": 0.3831590414047241,
+      "learning_rate": 8.687245152840244e-05,
+      "loss": 0.2907,
+      "step": 1194
+    },
+    {
+      "epoch": 0.25237592397043296,
+      "grad_norm": 0.4787651300430298,
+      "learning_rate": 8.684955373523786e-05,
+      "loss": 0.2951,
+      "step": 1195
+    },
+    {
+      "epoch": 0.2525871172122492,
+      "grad_norm": 0.32605835795402527,
+      "learning_rate": 8.682663901302732e-05,
+      "loss": 0.2932,
+      "step": 1196
+    },
+    {
+      "epoch": 0.2527983104540655,
+      "grad_norm": 0.3382115066051483,
+      "learning_rate": 8.680370737229804e-05,
+      "loss": 0.2919,
+      "step": 1197
+    },
+    {
+      "epoch": 0.2530095036958817,
+      "grad_norm": 0.4737050533294678,
+      "learning_rate": 8.678075882358506e-05,
+      "loss": 0.2984,
+      "step": 1198
+    },
+    {
+      "epoch": 0.253220696937698,
+      "grad_norm": 0.40661999583244324,
+      "learning_rate": 8.675779337743114e-05,
+      "loss": 0.3027,
+      "step": 1199
+    },
+    {
+      "epoch": 0.25343189017951423,
+      "grad_norm": 0.3520958125591278,
+      "learning_rate": 8.673481104438686e-05,
+      "loss": 0.2933,
+      "step": 1200
+    },
+    {
+      "epoch": 0.25343189017951423,
+      "eval_train_split_loss": 0.3787453770637512,
+      "eval_train_split_runtime": 6.4955,
+      "eval_train_split_samples_per_second": 1.54,
+      "eval_train_split_steps_per_second": 0.154,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2536430834213305,
+      "grad_norm": 0.37416312098503113,
+      "learning_rate": 8.671181183501052e-05,
+      "loss": 0.28,
+      "step": 1201
+    },
+    {
+      "epoch": 0.2538542766631468,
+      "grad_norm": 0.3563540279865265,
+      "learning_rate": 8.668879575986816e-05,
+      "loss": 0.283,
+      "step": 1202
+    },
+    {
+      "epoch": 0.25406546990496304,
+      "grad_norm": 0.31723958253860474,
+      "learning_rate": 8.666576282953363e-05,
+      "loss": 0.2901,
+      "step": 1203
+    },
+    {
+      "epoch": 0.2542766631467793,
+      "grad_norm": 0.4513072371482849,
+      "learning_rate": 8.664271305458845e-05,
+      "loss": 0.2932,
+      "step": 1204
+    },
+    {
+      "epoch": 0.25448785638859556,
+      "grad_norm": 0.37472087144851685,
+      "learning_rate": 8.661964644562193e-05,
+      "loss": 0.2857,
+      "step": 1205
+    },
+    {
+      "epoch": 0.25469904963041184,
+      "grad_norm": 0.36192786693573,
+      "learning_rate": 8.65965630132311e-05,
+      "loss": 0.2999,
+      "step": 1206
+    },
+    {
+      "epoch": 0.2549102428722281,
+      "grad_norm": 0.39012131094932556,
+      "learning_rate": 8.657346276802071e-05,
+      "loss": 0.2999,
+      "step": 1207
+    },
+    {
+      "epoch": 0.25512143611404436,
+      "grad_norm": 0.3847809135913849,
+      "learning_rate": 8.655034572060322e-05,
+      "loss": 0.29,
+      "step": 1208
+    },
+    {
+      "epoch": 0.2553326293558606,
+      "grad_norm": 0.8860164880752563,
+      "learning_rate": 8.652721188159886e-05,
+      "loss": 0.2868,
+      "step": 1209
+    },
+    {
+      "epoch": 0.2555438225976769,
+      "grad_norm": 0.40524816513061523,
+      "learning_rate": 8.650406126163552e-05,
+      "loss": 0.2929,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2557550158394931,
+      "grad_norm": 0.39258337020874023,
+      "learning_rate": 8.648089387134882e-05,
+      "loss": 0.2952,
+      "step": 1211
+    },
+    {
+      "epoch": 0.2559662090813094,
+      "grad_norm": 0.37229761481285095,
+      "learning_rate": 8.645770972138209e-05,
+      "loss": 0.2881,
+      "step": 1212
+    },
+    {
+      "epoch": 0.25617740232312564,
+      "grad_norm": 0.399451345205307,
+      "learning_rate": 8.643450882238638e-05,
+      "loss": 0.2948,
+      "step": 1213
+    },
+    {
+      "epoch": 0.2563885955649419,
+      "grad_norm": 0.4417630732059479,
+      "learning_rate": 8.641129118502039e-05,
+      "loss": 0.2972,
+      "step": 1214
+    },
+    {
+      "epoch": 0.2565997888067582,
+      "grad_norm": 0.3017156720161438,
+      "learning_rate": 8.638805681995052e-05,
+      "loss": 0.2891,
+      "step": 1215
+    },
+    {
+      "epoch": 0.25681098204857444,
+      "grad_norm": 0.747345507144928,
+      "learning_rate": 8.636480573785089e-05,
+      "loss": 0.2889,
+      "step": 1216
+    },
+    {
+      "epoch": 0.25702217529039073,
+      "grad_norm": 0.33316177129745483,
+      "learning_rate": 8.634153794940322e-05,
+      "loss": 0.2955,
+      "step": 1217
+    },
+    {
+      "epoch": 0.25723336853220696,
+      "grad_norm": 0.4526560306549072,
+      "learning_rate": 8.631825346529703e-05,
+      "loss": 0.2974,
+      "step": 1218
+    },
+    {
+      "epoch": 0.25744456177402325,
+      "grad_norm": 0.34297114610671997,
+      "learning_rate": 8.629495229622942e-05,
+      "loss": 0.2914,
+      "step": 1219
+    },
+    {
+      "epoch": 0.2576557550158395,
+      "grad_norm": 0.5042100548744202,
+      "learning_rate": 8.627163445290513e-05,
+      "loss": 0.299,
+      "step": 1220
+    },
+    {
+      "epoch": 0.25786694825765577,
+      "grad_norm": 0.32757219672203064,
+      "learning_rate": 8.624829994603667e-05,
+      "loss": 0.2909,
+      "step": 1221
+    },
+    {
+      "epoch": 0.258078141499472,
+      "grad_norm": 0.42026495933532715,
+      "learning_rate": 8.622494878634411e-05,
+      "loss": 0.2995,
+      "step": 1222
+    },
+    {
+      "epoch": 0.2582893347412883,
+      "grad_norm": 0.3336133062839508,
+      "learning_rate": 8.62015809845552e-05,
+      "loss": 0.2933,
+      "step": 1223
+    },
+    {
+      "epoch": 0.2585005279831045,
+      "grad_norm": 0.3481805622577667,
+      "learning_rate": 8.617819655140533e-05,
+      "loss": 0.2877,
+      "step": 1224
+    },
+    {
+      "epoch": 0.2587117212249208,
+      "grad_norm": 0.36924806237220764,
+      "learning_rate": 8.615479549763756e-05,
+      "loss": 0.2884,
+      "step": 1225
+    },
+    {
+      "epoch": 0.25892291446673704,
+      "grad_norm": 0.3114871680736542,
+      "learning_rate": 8.613137783400256e-05,
+      "loss": 0.2885,
+      "step": 1226
+    },
+    {
+      "epoch": 0.25913410770855333,
+      "grad_norm": 0.35585883259773254,
+      "learning_rate": 8.61079435712586e-05,
+      "loss": 0.2921,
+      "step": 1227
+    },
+    {
+      "epoch": 0.25934530095036956,
+      "grad_norm": 0.33406704664230347,
+      "learning_rate": 8.608449272017165e-05,
+      "loss": 0.2866,
+      "step": 1228
+    },
+    {
+      "epoch": 0.25955649419218585,
+      "grad_norm": 0.4188520312309265,
+      "learning_rate": 8.606102529151525e-05,
+      "loss": 0.2874,
+      "step": 1229
+    },
+    {
+      "epoch": 0.25976768743400214,
+      "grad_norm": 0.400134801864624,
+      "learning_rate": 8.603754129607055e-05,
+      "loss": 0.2959,
+      "step": 1230
+    },
+    {
+      "epoch": 0.25997888067581837,
+      "grad_norm": 0.4098871350288391,
+      "learning_rate": 8.601404074462636e-05,
+      "loss": 0.2933,
+      "step": 1231
+    },
+    {
+      "epoch": 0.26019007391763466,
+      "grad_norm": 0.2871926426887512,
+      "learning_rate": 8.599052364797903e-05,
+      "loss": 0.2901,
+      "step": 1232
+    },
+    {
+      "epoch": 0.2604012671594509,
+      "grad_norm": 0.4753386974334717,
+      "learning_rate": 8.596699001693255e-05,
+      "loss": 0.2922,
+      "step": 1233
+    },
+    {
+      "epoch": 0.2606124604012672,
+      "grad_norm": 0.3135407269001007,
+      "learning_rate": 8.594343986229854e-05,
+      "loss": 0.2848,
+      "step": 1234
+    },
+    {
+      "epoch": 0.2608236536430834,
+      "grad_norm": 0.32311269640922546,
+      "learning_rate": 8.591987319489611e-05,
+      "loss": 0.2871,
+      "step": 1235
+    },
+    {
+      "epoch": 0.2610348468848997,
+      "grad_norm": 0.3962644040584564,
+      "learning_rate": 8.589629002555206e-05,
+      "loss": 0.294,
+      "step": 1236
+    },
+    {
+      "epoch": 0.26124604012671593,
+      "grad_norm": 0.3401992917060852,
+      "learning_rate": 8.587269036510073e-05,
+      "loss": 0.2885,
+      "step": 1237
+    },
+    {
+      "epoch": 0.2614572333685322,
+      "grad_norm": 0.3294556140899658,
+      "learning_rate": 8.5849074224384e-05,
+      "loss": 0.2923,
+      "step": 1238
+    },
+    {
+      "epoch": 0.26166842661034845,
+      "grad_norm": 0.33915790915489197,
+      "learning_rate": 8.582544161425139e-05,
+      "loss": 0.2956,
+      "step": 1239
+    },
+    {
+      "epoch": 0.26187961985216474,
+      "grad_norm": 0.3442264199256897,
+      "learning_rate": 8.580179254555996e-05,
+      "loss": 0.286,
+      "step": 1240
+    },
+    {
+      "epoch": 0.26209081309398097,
+      "grad_norm": 0.3817264139652252,
+      "learning_rate": 8.577812702917427e-05,
+      "loss": 0.2852,
+      "step": 1241
+    },
+    {
+      "epoch": 0.26230200633579726,
+      "grad_norm": 0.40634483098983765,
+      "learning_rate": 8.575444507596655e-05,
+      "loss": 0.2949,
+      "step": 1242
+    },
+    {
+      "epoch": 0.26251319957761354,
+      "grad_norm": 0.5194107890129089,
+      "learning_rate": 8.573074669681649e-05,
+      "loss": 0.2876,
+      "step": 1243
+    },
+    {
+      "epoch": 0.2627243928194298,
+      "grad_norm": 0.3602850139141083,
+      "learning_rate": 8.570703190261138e-05,
+      "loss": 0.2902,
+      "step": 1244
+    },
+    {
+      "epoch": 0.26293558606124606,
+      "grad_norm": 0.45110800862312317,
+      "learning_rate": 8.5683300704246e-05,
+      "loss": 0.2884,
+      "step": 1245
+    },
+    {
+      "epoch": 0.2631467793030623,
+      "grad_norm": 0.40936121344566345,
+      "learning_rate": 8.565955311262271e-05,
+      "loss": 0.2996,
+      "step": 1246
+    },
+    {
+      "epoch": 0.2633579725448786,
+      "grad_norm": 0.37665265798568726,
+      "learning_rate": 8.563578913865137e-05,
+      "loss": 0.287,
+      "step": 1247
+    },
+    {
+      "epoch": 0.2635691657866948,
+      "grad_norm": 0.4111715257167816,
+      "learning_rate": 8.561200879324942e-05,
+      "loss": 0.2919,
+      "step": 1248
+    },
+    {
+      "epoch": 0.2637803590285111,
+      "grad_norm": 0.4066551923751831,
+      "learning_rate": 8.558821208734176e-05,
+      "loss": 0.2908,
+      "step": 1249
+    },
+    {
+      "epoch": 0.26399155227032733,
+      "grad_norm": 0.4552525579929352,
+      "learning_rate": 8.556439903186083e-05,
+      "loss": 0.2913,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2642027455121436,
+      "grad_norm": 0.36655759811401367,
+      "learning_rate": 8.554056963774657e-05,
+      "loss": 0.2921,
+      "step": 1251
+    },
+    {
+      "epoch": 0.26441393875395985,
+      "grad_norm": 0.3455119729042053,
+      "learning_rate": 8.551672391594645e-05,
+      "loss": 0.2886,
+      "step": 1252
+    },
+    {
+      "epoch": 0.26462513199577614,
+      "grad_norm": 0.34361720085144043,
+      "learning_rate": 8.549286187741543e-05,
+      "loss": 0.288,
+      "step": 1253
+    },
+    {
+      "epoch": 0.2648363252375924,
+      "grad_norm": 0.30754244327545166,
+      "learning_rate": 8.546898353311597e-05,
+      "loss": 0.2864,
+      "step": 1254
+    },
+    {
+      "epoch": 0.26504751847940866,
+      "grad_norm": 0.42501863837242126,
+      "learning_rate": 8.5445088894018e-05,
+      "loss": 0.2858,
+      "step": 1255
+    },
+    {
+      "epoch": 0.2652587117212249,
+      "grad_norm": 0.44186148047447205,
+      "learning_rate": 8.542117797109894e-05,
+      "loss": 0.288,
+      "step": 1256
+    },
+    {
+      "epoch": 0.2654699049630412,
+      "grad_norm": 0.33038267493247986,
+      "learning_rate": 8.539725077534374e-05,
+      "loss": 0.2881,
+      "step": 1257
+    },
+    {
+      "epoch": 0.26568109820485747,
+      "grad_norm": 0.31095200777053833,
+      "learning_rate": 8.537330731774476e-05,
+      "loss": 0.2866,
+      "step": 1258
+    },
+    {
+      "epoch": 0.2658922914466737,
+      "grad_norm": 0.5105559229850769,
+      "learning_rate": 8.534934760930188e-05,
+      "loss": 0.2868,
+      "step": 1259
+    },
+    {
+      "epoch": 0.26610348468849,
+      "grad_norm": 0.5872840881347656,
+      "learning_rate": 8.53253716610224e-05,
+      "loss": 0.2902,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2663146779303062,
+      "grad_norm": 0.45522835850715637,
+      "learning_rate": 8.530137948392112e-05,
+      "loss": 0.2938,
+      "step": 1261
+    },
+    {
+      "epoch": 0.2665258711721225,
+      "grad_norm": 0.4069291055202484,
+      "learning_rate": 8.527737108902029e-05,
+      "loss": 0.29,
+      "step": 1262
+    },
+    {
+      "epoch": 0.26673706441393874,
+      "grad_norm": 0.7565174698829651,
+      "learning_rate": 8.525334648734957e-05,
+      "loss": 0.2922,
+      "step": 1263
+    },
+    {
+      "epoch": 0.26694825765575503,
+      "grad_norm": 0.4196932911872864,
+      "learning_rate": 8.522930568994614e-05,
+      "loss": 0.2925,
+      "step": 1264
+    },
+    {
+      "epoch": 0.26715945089757126,
+      "grad_norm": 0.4165521562099457,
+      "learning_rate": 8.520524870785453e-05,
+      "loss": 0.2922,
+      "step": 1265
+    },
+    {
+      "epoch": 0.26737064413938755,
+      "grad_norm": 0.372698038816452,
+      "learning_rate": 8.51811755521268e-05,
+      "loss": 0.288,
+      "step": 1266
+    },
+    {
+      "epoch": 0.2675818373812038,
+      "grad_norm": 0.48085272312164307,
+      "learning_rate": 8.515708623382237e-05,
+      "loss": 0.2898,
+      "step": 1267
+    },
+    {
+      "epoch": 0.26779303062302007,
+      "grad_norm": 0.40782758593559265,
+      "learning_rate": 8.51329807640081e-05,
+      "loss": 0.2905,
+      "step": 1268
+    },
+    {
+      "epoch": 0.2680042238648363,
+      "grad_norm": 0.4860495328903198,
+      "learning_rate": 8.510885915375831e-05,
+      "loss": 0.2913,
+      "step": 1269
+    },
+    {
+      "epoch": 0.2682154171066526,
+      "grad_norm": 0.3825613558292389,
+      "learning_rate": 8.508472141415467e-05,
+      "loss": 0.2862,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2684266103484689,
+      "grad_norm": 0.41222870349884033,
+      "learning_rate": 8.506056755628631e-05,
+      "loss": 0.2882,
+      "step": 1271
+    },
+    {
+      "epoch": 0.2686378035902851,
+      "grad_norm": 0.37624865770339966,
+      "learning_rate": 8.503639759124975e-05,
+      "loss": 0.2918,
+      "step": 1272
+    },
+    {
+      "epoch": 0.2688489968321014,
+      "grad_norm": 0.34061068296432495,
+      "learning_rate": 8.50122115301489e-05,
+      "loss": 0.2886,
+      "step": 1273
+    },
+    {
+      "epoch": 0.2690601900739176,
+      "grad_norm": 0.3435481786727905,
+      "learning_rate": 8.498800938409508e-05,
+      "loss": 0.2863,
+      "step": 1274
+    },
+    {
+      "epoch": 0.2692713833157339,
+      "grad_norm": 0.33692166209220886,
+      "learning_rate": 8.496379116420699e-05,
+      "loss": 0.2835,
+      "step": 1275
+    },
+    {
+      "epoch": 0.26948257655755015,
+      "grad_norm": 0.36858054995536804,
+      "learning_rate": 8.493955688161072e-05,
+      "loss": 0.2873,
+      "step": 1276
+    },
+    {
+      "epoch": 0.26969376979936643,
+      "grad_norm": 0.40617939829826355,
+      "learning_rate": 8.491530654743976e-05,
+      "loss": 0.2919,
+      "step": 1277
+    },
+    {
+      "epoch": 0.26990496304118267,
+      "grad_norm": 0.3340827524662018,
+      "learning_rate": 8.48910401728349e-05,
+      "loss": 0.2815,
+      "step": 1278
+    },
+    {
+      "epoch": 0.27011615628299895,
+      "grad_norm": 0.34798315167427063,
+      "learning_rate": 8.486675776894439e-05,
+      "loss": 0.2833,
+      "step": 1279
+    },
+    {
+      "epoch": 0.2703273495248152,
+      "grad_norm": 0.45338308811187744,
+      "learning_rate": 8.484245934692379e-05,
+      "loss": 0.2922,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2705385427666315,
+      "grad_norm": 0.3489081561565399,
+      "learning_rate": 8.481814491793604e-05,
+      "loss": 0.2852,
+      "step": 1281
+    },
+    {
+      "epoch": 0.2707497360084477,
+      "grad_norm": 0.49687275290489197,
+      "learning_rate": 8.479381449315142e-05,
+      "loss": 0.2885,
+      "step": 1282
+    },
+    {
+      "epoch": 0.270960929250264,
+      "grad_norm": 0.35449495911598206,
+      "learning_rate": 8.476946808374757e-05,
+      "loss": 0.2877,
+      "step": 1283
+    },
+    {
+      "epoch": 0.2711721224920803,
+      "grad_norm": 0.39048194885253906,
+      "learning_rate": 8.474510570090945e-05,
+      "loss": 0.2862,
+      "step": 1284
+    },
+    {
+      "epoch": 0.2713833157338965,
+      "grad_norm": 0.5302790999412537,
+      "learning_rate": 8.47207273558294e-05,
+      "loss": 0.2887,
+      "step": 1285
+    },
+    {
+      "epoch": 0.2715945089757128,
+      "grad_norm": 0.42626941204071045,
+      "learning_rate": 8.469633305970708e-05,
+      "loss": 0.2845,
+      "step": 1286
+    },
+    {
+      "epoch": 0.27180570221752903,
+      "grad_norm": 0.49231231212615967,
+      "learning_rate": 8.467192282374945e-05,
+      "loss": 0.2856,
+      "step": 1287
+    },
+    {
+      "epoch": 0.2720168954593453,
+      "grad_norm": 0.37713485956192017,
+      "learning_rate": 8.464749665917081e-05,
+      "loss": 0.2867,
+      "step": 1288
+    },
+    {
+      "epoch": 0.27222808870116155,
+      "grad_norm": 0.4466076195240021,
+      "learning_rate": 8.462305457719277e-05,
+      "loss": 0.2891,
+      "step": 1289
+    },
+    {
+      "epoch": 0.27243928194297784,
+      "grad_norm": 0.5410495400428772,
+      "learning_rate": 8.45985965890443e-05,
+      "loss": 0.29,
+      "step": 1290
+    },
+    {
+      "epoch": 0.27265047518479407,
+      "grad_norm": 0.38790881633758545,
+      "learning_rate": 8.457412270596161e-05,
+      "loss": 0.2829,
+      "step": 1291
+    },
+    {
+      "epoch": 0.27286166842661036,
+      "grad_norm": 0.42601290345191956,
+      "learning_rate": 8.454963293918822e-05,
+      "loss": 0.2872,
+      "step": 1292
+    },
+    {
+      "epoch": 0.2730728616684266,
+      "grad_norm": 0.42196938395500183,
+      "learning_rate": 8.452512729997504e-05,
+      "loss": 0.2858,
+      "step": 1293
+    },
+    {
+      "epoch": 0.2732840549102429,
+      "grad_norm": 0.3712303340435028,
+      "learning_rate": 8.450060579958012e-05,
+      "loss": 0.2852,
+      "step": 1294
+    },
+    {
+      "epoch": 0.2734952481520591,
+      "grad_norm": 0.46956533193588257,
+      "learning_rate": 8.447606844926894e-05,
+      "loss": 0.2881,
+      "step": 1295
+    },
+    {
+      "epoch": 0.2737064413938754,
+      "grad_norm": 0.42075303196907043,
+      "learning_rate": 8.445151526031416e-05,
+      "loss": 0.2909,
+      "step": 1296
+    },
+    {
+      "epoch": 0.27391763463569163,
+      "grad_norm": 0.4201053977012634,
+      "learning_rate": 8.442694624399576e-05,
+      "loss": 0.2903,
+      "step": 1297
+    },
+    {
+      "epoch": 0.2741288278775079,
+      "grad_norm": 0.31366413831710815,
+      "learning_rate": 8.440236141160101e-05,
+      "loss": 0.2883,
+      "step": 1298
+    },
+    {
+      "epoch": 0.2743400211193242,
+      "grad_norm": 0.41620567440986633,
+      "learning_rate": 8.437776077442441e-05,
+      "loss": 0.2848,
+      "step": 1299
+    },
+    {
+      "epoch": 0.27455121436114044,
+      "grad_norm": 0.3817262053489685,
+      "learning_rate": 8.435314434376773e-05,
+      "loss": 0.2855,
+      "step": 1300
+    },
+    {
+      "epoch": 0.2747624076029567,
+      "grad_norm": 0.4744551479816437,
+      "learning_rate": 8.432851213094e-05,
+      "loss": 0.2977,
+      "step": 1301
+    },
+    {
+      "epoch": 0.27497360084477296,
+      "grad_norm": 0.31848856806755066,
+      "learning_rate": 8.430386414725748e-05,
+      "loss": 0.2902,
+      "step": 1302
+    },
+    {
+      "epoch": 0.27518479408658925,
+      "grad_norm": 0.4387161135673523,
+      "learning_rate": 8.427920040404374e-05,
+      "loss": 0.2903,
+      "step": 1303
+    },
+    {
+      "epoch": 0.2753959873284055,
+      "grad_norm": 0.3203023374080658,
+      "learning_rate": 8.425452091262952e-05,
+      "loss": 0.2915,
+      "step": 1304
+    },
+    {
+      "epoch": 0.27560718057022177,
+      "grad_norm": 0.44606611132621765,
+      "learning_rate": 8.422982568435281e-05,
+      "loss": 0.2986,
+      "step": 1305
+    },
+    {
+      "epoch": 0.275818373812038,
+      "grad_norm": 0.39073047041893005,
+      "learning_rate": 8.420511473055886e-05,
+      "loss": 0.2943,
+      "step": 1306
+    },
+    {
+      "epoch": 0.2760295670538543,
+      "grad_norm": 0.3794722557067871,
+      "learning_rate": 8.418038806260012e-05,
+      "loss": 0.2792,
+      "step": 1307
+    },
+    {
+      "epoch": 0.2762407602956705,
+      "grad_norm": 0.3715200424194336,
+      "learning_rate": 8.415564569183626e-05,
+      "loss": 0.2861,
+      "step": 1308
+    },
+    {
+      "epoch": 0.2764519535374868,
+      "grad_norm": 0.5410581827163696,
+      "learning_rate": 8.413088762963415e-05,
+      "loss": 0.2861,
+      "step": 1309
+    },
+    {
+      "epoch": 0.27666314677930304,
+      "grad_norm": 0.5651510953903198,
+      "learning_rate": 8.410611388736792e-05,
+      "loss": 0.2881,
+      "step": 1310
+    },
+    {
+      "epoch": 0.2768743400211193,
+      "grad_norm": 0.37738484144210815,
+      "learning_rate": 8.408132447641885e-05,
+      "loss": 0.2906,
+      "step": 1311
+    },
+    {
+      "epoch": 0.2770855332629356,
+      "grad_norm": 0.4375596344470978,
+      "learning_rate": 8.405651940817544e-05,
+      "loss": 0.2857,
+      "step": 1312
+    },
+    {
+      "epoch": 0.27729672650475184,
+      "grad_norm": 0.32695794105529785,
+      "learning_rate": 8.403169869403338e-05,
+      "loss": 0.2831,
+      "step": 1313
+    },
+    {
+      "epoch": 0.27750791974656813,
+      "grad_norm": 0.4250601530075073,
+      "learning_rate": 8.400686234539557e-05,
+      "loss": 0.2809,
+      "step": 1314
+    },
+    {
+      "epoch": 0.27771911298838436,
+      "grad_norm": 0.39116808772087097,
+      "learning_rate": 8.398201037367201e-05,
+      "loss": 0.2903,
+      "step": 1315
+    },
+    {
+      "epoch": 0.27793030623020065,
+      "grad_norm": 0.41614648699760437,
+      "learning_rate": 8.395714279028002e-05,
+      "loss": 0.2886,
+      "step": 1316
+    },
+    {
+      "epoch": 0.2781414994720169,
+      "grad_norm": 0.33232125639915466,
+      "learning_rate": 8.393225960664397e-05,
+      "loss": 0.2894,
+      "step": 1317
+    },
+    {
+      "epoch": 0.27835269271383317,
+      "grad_norm": 0.30973201990127563,
+      "learning_rate": 8.390736083419543e-05,
+      "loss": 0.2771,
+      "step": 1318
+    },
+    {
+      "epoch": 0.2785638859556494,
+      "grad_norm": 0.3771383762359619,
+      "learning_rate": 8.388244648437316e-05,
+      "loss": 0.2859,
+      "step": 1319
+    },
+    {
+      "epoch": 0.2787750791974657,
+      "grad_norm": 0.3191227614879608,
+      "learning_rate": 8.385751656862305e-05,
+      "loss": 0.2817,
+      "step": 1320
+    },
+    {
+      "epoch": 0.2789862724392819,
+      "grad_norm": 0.4345984160900116,
+      "learning_rate": 8.383257109839815e-05,
+      "loss": 0.2778,
+      "step": 1321
+    },
+    {
+      "epoch": 0.2791974656810982,
+      "grad_norm": 0.30248335003852844,
+      "learning_rate": 8.380761008515865e-05,
+      "loss": 0.2799,
+      "step": 1322
+    },
+    {
+      "epoch": 0.27940865892291444,
+      "grad_norm": 0.3206419050693512,
+      "learning_rate": 8.378263354037188e-05,
+      "loss": 0.2838,
+      "step": 1323
+    },
+    {
+      "epoch": 0.27961985216473073,
+      "grad_norm": 0.3429322838783264,
+      "learning_rate": 8.375764147551231e-05,
+      "loss": 0.2791,
+      "step": 1324
+    },
+    {
+      "epoch": 0.27983104540654696,
+      "grad_norm": 0.48219940066337585,
+      "learning_rate": 8.373263390206154e-05,
+      "loss": 0.2892,
+      "step": 1325
+    },
+    {
+      "epoch": 0.28004223864836325,
+      "grad_norm": 0.33997657895088196,
+      "learning_rate": 8.370761083150832e-05,
+      "loss": 0.2819,
+      "step": 1326
+    },
+    {
+      "epoch": 0.28025343189017954,
+      "grad_norm": 0.31234824657440186,
+      "learning_rate": 8.368257227534844e-05,
+      "loss": 0.2814,
+      "step": 1327
+    },
+    {
+      "epoch": 0.28046462513199577,
+      "grad_norm": 0.4097597002983093,
+      "learning_rate": 8.365751824508491e-05,
+      "loss": 0.2791,
+      "step": 1328
+    },
+    {
+      "epoch": 0.28067581837381206,
+      "grad_norm": 0.3415398895740509,
+      "learning_rate": 8.363244875222776e-05,
+      "loss": 0.2848,
+      "step": 1329
+    },
+    {
+      "epoch": 0.2808870116156283,
+      "grad_norm": 0.46665292978286743,
+      "learning_rate": 8.360736380829419e-05,
+      "loss": 0.282,
+      "step": 1330
+    },
+    {
+      "epoch": 0.2810982048574446,
+      "grad_norm": 0.32126355171203613,
+      "learning_rate": 8.358226342480844e-05,
+      "loss": 0.2764,
+      "step": 1331
+    },
+    {
+      "epoch": 0.2813093980992608,
+      "grad_norm": 0.3728634715080261,
+      "learning_rate": 8.355714761330191e-05,
+      "loss": 0.2811,
+      "step": 1332
+    },
+    {
+      "epoch": 0.2815205913410771,
+      "grad_norm": 0.37381964921951294,
+      "learning_rate": 8.3532016385313e-05,
+      "loss": 0.2866,
+      "step": 1333
+    },
+    {
+      "epoch": 0.28173178458289333,
+      "grad_norm": 0.3681824207305908,
+      "learning_rate": 8.35068697523873e-05,
+      "loss": 0.2836,
+      "step": 1334
+    },
+    {
+      "epoch": 0.2819429778247096,
+      "grad_norm": 0.3088197112083435,
+      "learning_rate": 8.348170772607738e-05,
+      "loss": 0.2856,
+      "step": 1335
+    },
+    {
+      "epoch": 0.28215417106652585,
+      "grad_norm": 0.3462510406970978,
+      "learning_rate": 8.345653031794292e-05,
+      "loss": 0.283,
+      "step": 1336
+    },
+    {
+      "epoch": 0.28236536430834214,
+      "grad_norm": 0.39286041259765625,
+      "learning_rate": 8.343133753955068e-05,
+      "loss": 0.2843,
+      "step": 1337
+    },
+    {
+      "epoch": 0.28257655755015837,
+      "grad_norm": 0.3647727370262146,
+      "learning_rate": 8.340612940247447e-05,
+      "loss": 0.2788,
+      "step": 1338
+    },
+    {
+      "epoch": 0.28278775079197466,
+      "grad_norm": 0.4464494287967682,
+      "learning_rate": 8.338090591829515e-05,
+      "loss": 0.2819,
+      "step": 1339
+    },
+    {
+      "epoch": 0.28299894403379094,
+      "grad_norm": 0.3262149691581726,
+      "learning_rate": 8.335566709860065e-05,
+      "loss": 0.2866,
+      "step": 1340
+    },
+    {
+      "epoch": 0.2832101372756072,
+      "grad_norm": 0.3643028438091278,
+      "learning_rate": 8.33304129549859e-05,
+      "loss": 0.2823,
+      "step": 1341
+    },
+    {
+      "epoch": 0.28342133051742346,
+      "grad_norm": 0.382761687040329,
+      "learning_rate": 8.330514349905293e-05,
+      "loss": 0.2894,
+      "step": 1342
+    },
+    {
+      "epoch": 0.2836325237592397,
+      "grad_norm": 0.3442258834838867,
+      "learning_rate": 8.327985874241077e-05,
+      "loss": 0.2871,
+      "step": 1343
+    },
+    {
+      "epoch": 0.283843717001056,
+      "grad_norm": 0.3331126868724823,
+      "learning_rate": 8.325455869667548e-05,
+      "loss": 0.2839,
+      "step": 1344
+    },
+    {
+      "epoch": 0.2840549102428722,
+      "grad_norm": 0.4125712811946869,
+      "learning_rate": 8.322924337347015e-05,
+      "loss": 0.2787,
+      "step": 1345
+    },
+    {
+      "epoch": 0.2842661034846885,
+      "grad_norm": 0.30621564388275146,
+      "learning_rate": 8.320391278442489e-05,
+      "loss": 0.2837,
+      "step": 1346
+    },
+    {
+      "epoch": 0.28447729672650474,
+      "grad_norm": 0.4869130849838257,
+      "learning_rate": 8.317856694117681e-05,
+      "loss": 0.2826,
+      "step": 1347
+    },
+    {
+      "epoch": 0.284688489968321,
+      "grad_norm": 0.31983137130737305,
+      "learning_rate": 8.315320585537006e-05,
+      "loss": 0.2829,
+      "step": 1348
+    },
+    {
+      "epoch": 0.28489968321013726,
+      "grad_norm": 0.5172858238220215,
+      "learning_rate": 8.312782953865575e-05,
+      "loss": 0.2901,
+      "step": 1349
+    },
+    {
+      "epoch": 0.28511087645195354,
+      "grad_norm": 0.43172696232795715,
+      "learning_rate": 8.3102438002692e-05,
+      "loss": 0.2897,
+      "step": 1350
+    },
+    {
+      "epoch": 0.2853220696937698,
+      "grad_norm": 0.5306637287139893,
+      "learning_rate": 8.307703125914397e-05,
+      "loss": 0.2865,
+      "step": 1351
+    },
+    {
+      "epoch": 0.28553326293558606,
+      "grad_norm": 0.38214248418807983,
+      "learning_rate": 8.305160931968373e-05,
+      "loss": 0.2841,
+      "step": 1352
+    },
+    {
+      "epoch": 0.28574445617740235,
+      "grad_norm": 0.43097570538520813,
+      "learning_rate": 8.302617219599038e-05,
+      "loss": 0.2927,
+      "step": 1353
+    },
+    {
+      "epoch": 0.2859556494192186,
+      "grad_norm": 0.43339377641677856,
+      "learning_rate": 8.300071989975e-05,
+      "loss": 0.2832,
+      "step": 1354
+    },
+    {
+      "epoch": 0.28616684266103487,
+      "grad_norm": 0.37728792428970337,
+      "learning_rate": 8.29752524426556e-05,
+      "loss": 0.2793,
+      "step": 1355
+    },
+    {
+      "epoch": 0.2863780359028511,
+      "grad_norm": 0.8370750546455383,
+      "learning_rate": 8.29497698364072e-05,
+      "loss": 0.2802,
+      "step": 1356
+    },
+    {
+      "epoch": 0.2865892291446674,
+      "grad_norm": 0.42214515805244446,
+      "learning_rate": 8.292427209271173e-05,
+      "loss": 0.283,
+      "step": 1357
+    },
+    {
+      "epoch": 0.2868004223864836,
+      "grad_norm": 0.44586893916130066,
+      "learning_rate": 8.289875922328314e-05,
+      "loss": 0.2832,
+      "step": 1358
+    },
+    {
+      "epoch": 0.2870116156282999,
+      "grad_norm": 0.4203036427497864,
+      "learning_rate": 8.287323123984226e-05,
+      "loss": 0.2747,
+      "step": 1359
+    },
+    {
+      "epoch": 0.28722280887011614,
+      "grad_norm": 0.42375409603118896,
+      "learning_rate": 8.284768815411692e-05,
+      "loss": 0.2808,
+      "step": 1360
+    },
+    {
+      "epoch": 0.28743400211193243,
+      "grad_norm": 0.3957616090774536,
+      "learning_rate": 8.282212997784184e-05,
+      "loss": 0.2828,
+      "step": 1361
+    },
+    {
+      "epoch": 0.28764519535374866,
+      "grad_norm": 0.41193485260009766,
+      "learning_rate": 8.279655672275871e-05,
+      "loss": 0.2848,
+      "step": 1362
+    },
+    {
+      "epoch": 0.28785638859556495,
+      "grad_norm": 0.5645737051963806,
+      "learning_rate": 8.277096840061616e-05,
+      "loss": 0.2839,
+      "step": 1363
+    },
+    {
+      "epoch": 0.2880675818373812,
+      "grad_norm": 0.3895312547683716,
+      "learning_rate": 8.274536502316966e-05,
+      "loss": 0.2838,
+      "step": 1364
+    },
+    {
+      "epoch": 0.28827877507919747,
+      "grad_norm": 0.41736871004104614,
+      "learning_rate": 8.27197466021817e-05,
+      "loss": 0.2871,
+      "step": 1365
+    },
+    {
+      "epoch": 0.2884899683210137,
+      "grad_norm": 0.42396506667137146,
+      "learning_rate": 8.269411314942159e-05,
+      "loss": 0.2819,
+      "step": 1366
+    },
+    {
+      "epoch": 0.28870116156283,
+      "grad_norm": 0.37646934390068054,
+      "learning_rate": 8.266846467666564e-05,
+      "loss": 0.2766,
+      "step": 1367
+    },
+    {
+      "epoch": 0.2889123548046463,
+      "grad_norm": 0.40967217087745667,
+      "learning_rate": 8.264280119569696e-05,
+      "loss": 0.2857,
+      "step": 1368
+    },
+    {
+      "epoch": 0.2891235480464625,
+      "grad_norm": 0.43097782135009766,
+      "learning_rate": 8.261712271830564e-05,
+      "loss": 0.2795,
+      "step": 1369
+    },
+    {
+      "epoch": 0.2893347412882788,
+      "grad_norm": 0.3831927478313446,
+      "learning_rate": 8.259142925628862e-05,
+      "loss": 0.2825,
+      "step": 1370
+    },
+    {
+      "epoch": 0.28954593453009503,
+      "grad_norm": 0.41198793053627014,
+      "learning_rate": 8.256572082144974e-05,
+      "loss": 0.2862,
+      "step": 1371
+    },
+    {
+      "epoch": 0.2897571277719113,
+      "grad_norm": 0.33437374234199524,
+      "learning_rate": 8.253999742559965e-05,
+      "loss": 0.2802,
+      "step": 1372
+    },
+    {
+      "epoch": 0.28996832101372755,
+      "grad_norm": 0.4741155505180359,
+      "learning_rate": 8.251425908055599e-05,
+      "loss": 0.2811,
+      "step": 1373
+    },
+    {
+      "epoch": 0.29017951425554384,
+      "grad_norm": 0.4141404330730438,
+      "learning_rate": 8.248850579814318e-05,
+      "loss": 0.2818,
+      "step": 1374
+    },
+    {
+      "epoch": 0.29039070749736007,
+      "grad_norm": 0.47356119751930237,
+      "learning_rate": 8.246273759019252e-05,
+      "loss": 0.2779,
+      "step": 1375
+    },
+    {
+      "epoch": 0.29060190073917636,
+      "grad_norm": 0.3216777443885803,
+      "learning_rate": 8.243695446854222e-05,
+      "loss": 0.2806,
+      "step": 1376
+    },
+    {
+      "epoch": 0.2908130939809926,
+      "grad_norm": 0.3380235731601715,
+      "learning_rate": 8.241115644503726e-05,
+      "loss": 0.2837,
+      "step": 1377
+    },
+    {
+      "epoch": 0.2910242872228089,
+      "grad_norm": 0.4187667667865753,
+      "learning_rate": 8.238534353152951e-05,
+      "loss": 0.2923,
+      "step": 1378
+    },
+    {
+      "epoch": 0.2912354804646251,
+      "grad_norm": 0.37491580843925476,
+      "learning_rate": 8.235951573987769e-05,
+      "loss": 0.2784,
+      "step": 1379
+    },
+    {
+      "epoch": 0.2914466737064414,
+      "grad_norm": 0.3649769127368927,
+      "learning_rate": 8.233367308194734e-05,
+      "loss": 0.2799,
+      "step": 1380
+    },
+    {
+      "epoch": 0.2916578669482577,
+      "grad_norm": 0.3530876636505127,
+      "learning_rate": 8.23078155696108e-05,
+      "loss": 0.2826,
+      "step": 1381
+    },
+    {
+      "epoch": 0.2918690601900739,
+      "grad_norm": 0.37215369939804077,
+      "learning_rate": 8.228194321474731e-05,
+      "loss": 0.2844,
+      "step": 1382
+    },
+    {
+      "epoch": 0.2920802534318902,
+      "grad_norm": 0.3498789966106415,
+      "learning_rate": 8.225605602924284e-05,
+      "loss": 0.2812,
+      "step": 1383
+    },
+    {
+      "epoch": 0.29229144667370643,
+      "grad_norm": 0.3470887839794159,
+      "learning_rate": 8.223015402499026e-05,
+      "loss": 0.2831,
+      "step": 1384
+    },
+    {
+      "epoch": 0.2925026399155227,
+      "grad_norm": 0.3238767385482788,
+      "learning_rate": 8.220423721388917e-05,
+      "loss": 0.2813,
+      "step": 1385
+    },
+    {
+      "epoch": 0.29271383315733895,
+      "grad_norm": 0.3761839270591736,
+      "learning_rate": 8.217830560784601e-05,
+      "loss": 0.2785,
+      "step": 1386
+    },
+    {
+      "epoch": 0.29292502639915524,
+      "grad_norm": 0.32137349247932434,
+      "learning_rate": 8.215235921877404e-05,
+      "loss": 0.2806,
+      "step": 1387
+    },
+    {
+      "epoch": 0.2931362196409715,
+      "grad_norm": 0.36933663487434387,
+      "learning_rate": 8.212639805859324e-05,
+      "loss": 0.2861,
+      "step": 1388
+    },
+    {
+      "epoch": 0.29334741288278776,
+      "grad_norm": 0.2983377277851105,
+      "learning_rate": 8.210042213923048e-05,
+      "loss": 0.2787,
+      "step": 1389
+    },
+    {
+      "epoch": 0.293558606124604,
+      "grad_norm": 0.3261181116104126,
+      "learning_rate": 8.20744314726193e-05,
+      "loss": 0.2786,
+      "step": 1390
+    },
+    {
+      "epoch": 0.2937697993664203,
+      "grad_norm": 0.3497909605503082,
+      "learning_rate": 8.20484260707001e-05,
+      "loss": 0.2839,
+      "step": 1391
+    },
+    {
+      "epoch": 0.2939809926082365,
+      "grad_norm": 0.322889119386673,
+      "learning_rate": 8.202240594541999e-05,
+      "loss": 0.2791,
+      "step": 1392
+    },
+    {
+      "epoch": 0.2941921858500528,
+      "grad_norm": 0.32643362879753113,
+      "learning_rate": 8.199637110873289e-05,
+      "loss": 0.2767,
+      "step": 1393
+    },
+    {
+      "epoch": 0.29440337909186903,
+      "grad_norm": 0.4049574136734009,
+      "learning_rate": 8.197032157259946e-05,
+      "loss": 0.2776,
+      "step": 1394
+    },
+    {
+      "epoch": 0.2946145723336853,
+      "grad_norm": 0.3508993983268738,
+      "learning_rate": 8.194425734898709e-05,
+      "loss": 0.2741,
+      "step": 1395
+    },
+    {
+      "epoch": 0.2948257655755016,
+      "grad_norm": 0.36577939987182617,
+      "learning_rate": 8.191817844986998e-05,
+      "loss": 0.2766,
+      "step": 1396
+    },
+    {
+      "epoch": 0.29503695881731784,
+      "grad_norm": 0.40553370118141174,
+      "learning_rate": 8.1892084887229e-05,
+      "loss": 0.2775,
+      "step": 1397
+    },
+    {
+      "epoch": 0.29524815205913413,
+      "grad_norm": 0.37327200174331665,
+      "learning_rate": 8.186597667305178e-05,
+      "loss": 0.2865,
+      "step": 1398
+    },
+    {
+      "epoch": 0.29545934530095036,
+      "grad_norm": 0.3162059485912323,
+      "learning_rate": 8.183985381933273e-05,
+      "loss": 0.281,
+      "step": 1399
+    },
+    {
+      "epoch": 0.29567053854276665,
+      "grad_norm": 0.32042810320854187,
+      "learning_rate": 8.181371633807288e-05,
+      "loss": 0.2797,
+      "step": 1400
+    },
+    {
+      "epoch": 0.29567053854276665,
+      "eval_train_split_loss": 0.39149004220962524,
+      "eval_train_split_runtime": 6.2771,
+      "eval_train_split_samples_per_second": 1.593,
+      "eval_train_split_steps_per_second": 0.159,
+      "step": 1400
+    },
+    {
+      "epoch": 0.2958817317845829,
+      "grad_norm": 0.42449113726615906,
+      "learning_rate": 8.178756424128012e-05,
+      "loss": 0.2878,
+      "step": 1401
+    },
+    {
+      "epoch": 0.29609292502639917,
+      "grad_norm": 0.28539809584617615,
+      "learning_rate": 8.176139754096893e-05,
+      "loss": 0.2859,
+      "step": 1402
+    },
+    {
+      "epoch": 0.2963041182682154,
+      "grad_norm": 0.36146843433380127,
+      "learning_rate": 8.173521624916056e-05,
+      "loss": 0.2854,
+      "step": 1403
+    },
+    {
+      "epoch": 0.2965153115100317,
+      "grad_norm": 0.30738380551338196,
+      "learning_rate": 8.170902037788296e-05,
+      "loss": 0.2859,
+      "step": 1404
+    },
+    {
+      "epoch": 0.2967265047518479,
+      "grad_norm": 0.34454530477523804,
+      "learning_rate": 8.168280993917077e-05,
+      "loss": 0.2785,
+      "step": 1405
+    },
+    {
+      "epoch": 0.2969376979936642,
+      "grad_norm": 0.43022286891937256,
+      "learning_rate": 8.165658494506529e-05,
+      "loss": 0.2795,
+      "step": 1406
+    },
+    {
+      "epoch": 0.29714889123548044,
+      "grad_norm": 0.35184264183044434,
+      "learning_rate": 8.163034540761461e-05,
+      "loss": 0.284,
+      "step": 1407
+    },
+    {
+      "epoch": 0.2973600844772967,
+      "grad_norm": 0.415888249874115,
+      "learning_rate": 8.160409133887334e-05,
+      "loss": 0.2808,
+      "step": 1408
+    },
+    {
+      "epoch": 0.297571277719113,
+      "grad_norm": 1.0405657291412354,
+      "learning_rate": 8.157782275090294e-05,
+      "loss": 0.2794,
+      "step": 1409
+    },
+    {
+      "epoch": 0.29778247096092925,
+      "grad_norm": 0.4436485469341278,
+      "learning_rate": 8.15515396557714e-05,
+      "loss": 0.2828,
+      "step": 1410
+    },
+    {
+      "epoch": 0.29799366420274553,
+      "grad_norm": 0.363044410943985,
+      "learning_rate": 8.152524206555345e-05,
+      "loss": 0.2766,
+      "step": 1411
+    },
+    {
+      "epoch": 0.29820485744456177,
+      "grad_norm": 0.46662256121635437,
+      "learning_rate": 8.149892999233046e-05,
+      "loss": 0.2816,
+      "step": 1412
+    },
+    {
+      "epoch": 0.29841605068637805,
+      "grad_norm": 0.3794650733470917,
+      "learning_rate": 8.147260344819046e-05,
+      "loss": 0.2851,
+      "step": 1413
+    },
+    {
+      "epoch": 0.2986272439281943,
+      "grad_norm": 0.37556973099708557,
+      "learning_rate": 8.144626244522813e-05,
+      "loss": 0.2801,
+      "step": 1414
+    },
+    {
+      "epoch": 0.2988384371700106,
+      "grad_norm": 0.4210280776023865,
+      "learning_rate": 8.141990699554476e-05,
+      "loss": 0.2774,
+      "step": 1415
+    },
+    {
+      "epoch": 0.2990496304118268,
+      "grad_norm": 0.39584624767303467,
+      "learning_rate": 8.139353711124831e-05,
+      "loss": 0.2745,
+      "step": 1416
+    },
+    {
+      "epoch": 0.2992608236536431,
+      "grad_norm": 0.3636949360370636,
+      "learning_rate": 8.136715280445337e-05,
+      "loss": 0.281,
+      "step": 1417
+    },
+    {
+      "epoch": 0.2994720168954593,
+      "grad_norm": 0.34028229117393494,
+      "learning_rate": 8.134075408728115e-05,
+      "loss": 0.2805,
+      "step": 1418
+    },
+    {
+      "epoch": 0.2996832101372756,
+      "grad_norm": 0.34924423694610596,
+      "learning_rate": 8.131434097185947e-05,
+      "loss": 0.2796,
+      "step": 1419
+    },
+    {
+      "epoch": 0.29989440337909185,
+      "grad_norm": 0.34077712893486023,
+      "learning_rate": 8.128791347032279e-05,
+      "loss": 0.2783,
+      "step": 1420
+    },
+    {
+      "epoch": 0.30010559662090813,
+      "grad_norm": 0.3574024438858032,
+      "learning_rate": 8.126147159481214e-05,
+      "loss": 0.2764,
+      "step": 1421
+    },
+    {
+      "epoch": 0.3003167898627244,
+      "grad_norm": 0.34380096197128296,
+      "learning_rate": 8.12350153574752e-05,
+      "loss": 0.2779,
+      "step": 1422
+    },
+    {
+      "epoch": 0.30052798310454065,
+      "grad_norm": 0.35979071259498596,
+      "learning_rate": 8.12085447704662e-05,
+      "loss": 0.2771,
+      "step": 1423
+    },
+    {
+      "epoch": 0.30073917634635694,
+      "grad_norm": 0.3103550970554352,
+      "learning_rate": 8.118205984594603e-05,
+      "loss": 0.2743,
+      "step": 1424
+    },
+    {
+      "epoch": 0.30095036958817317,
+      "grad_norm": 0.4158700406551361,
+      "learning_rate": 8.115556059608207e-05,
+      "loss": 0.2746,
+      "step": 1425
+    },
+    {
+      "epoch": 0.30116156282998946,
+      "grad_norm": 0.31426742672920227,
+      "learning_rate": 8.112904703304837e-05,
+      "loss": 0.2726,
+      "step": 1426
+    },
+    {
+      "epoch": 0.3013727560718057,
+      "grad_norm": 0.3494434058666229,
+      "learning_rate": 8.11025191690255e-05,
+      "loss": 0.2826,
+      "step": 1427
+    },
+    {
+      "epoch": 0.301583949313622,
+      "grad_norm": 0.3594941198825836,
+      "learning_rate": 8.107597701620064e-05,
+      "loss": 0.2795,
+      "step": 1428
+    },
+    {
+      "epoch": 0.3017951425554382,
+      "grad_norm": 0.32080164551734924,
+      "learning_rate": 8.10494205867675e-05,
+      "loss": 0.2786,
+      "step": 1429
+    },
+    {
+      "epoch": 0.3020063357972545,
+      "grad_norm": 0.38638702034950256,
+      "learning_rate": 8.102284989292638e-05,
+      "loss": 0.2824,
+      "step": 1430
+    },
+    {
+      "epoch": 0.30221752903907073,
+      "grad_norm": 0.3699018359184265,
+      "learning_rate": 8.09962649468841e-05,
+      "loss": 0.2795,
+      "step": 1431
+    },
+    {
+      "epoch": 0.302428722280887,
+      "grad_norm": 0.5072017908096313,
+      "learning_rate": 8.096966576085406e-05,
+      "loss": 0.2787,
+      "step": 1432
+    },
+    {
+      "epoch": 0.30263991552270325,
+      "grad_norm": 0.5698899030685425,
+      "learning_rate": 8.094305234705617e-05,
+      "loss": 0.2791,
+      "step": 1433
+    },
+    {
+      "epoch": 0.30285110876451954,
+      "grad_norm": 0.4896374046802521,
+      "learning_rate": 8.09164247177169e-05,
+      "loss": 0.2798,
+      "step": 1434
+    },
+    {
+      "epoch": 0.30306230200633577,
+      "grad_norm": 0.3744699954986572,
+      "learning_rate": 8.088978288506924e-05,
+      "loss": 0.2746,
+      "step": 1435
+    },
+    {
+      "epoch": 0.30327349524815206,
+      "grad_norm": 0.4496450126171112,
+      "learning_rate": 8.086312686135272e-05,
+      "loss": 0.2774,
+      "step": 1436
+    },
+    {
+      "epoch": 0.30348468848996835,
+      "grad_norm": 0.4216213822364807,
+      "learning_rate": 8.083645665881338e-05,
+      "loss": 0.2738,
+      "step": 1437
+    },
+    {
+      "epoch": 0.3036958817317846,
+      "grad_norm": 0.35429900884628296,
+      "learning_rate": 8.080977228970375e-05,
+      "loss": 0.2901,
+      "step": 1438
+    },
+    {
+      "epoch": 0.30390707497360087,
+      "grad_norm": 0.47646719217300415,
+      "learning_rate": 8.07830737662829e-05,
+      "loss": 0.2783,
+      "step": 1439
+    },
+    {
+      "epoch": 0.3041182682154171,
+      "grad_norm": 0.4150383174419403,
+      "learning_rate": 8.075636110081644e-05,
+      "loss": 0.2766,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3043294614572334,
+      "grad_norm": 0.39623433351516724,
+      "learning_rate": 8.072963430557636e-05,
+      "loss": 0.2781,
+      "step": 1441
+    },
+    {
+      "epoch": 0.3045406546990496,
+      "grad_norm": 0.46262648701667786,
+      "learning_rate": 8.070289339284123e-05,
+      "loss": 0.2749,
+      "step": 1442
+    },
+    {
+      "epoch": 0.3047518479408659,
+      "grad_norm": 0.3374291956424713,
+      "learning_rate": 8.067613837489612e-05,
+      "loss": 0.2808,
+      "step": 1443
+    },
+    {
+      "epoch": 0.30496304118268214,
+      "grad_norm": 0.3909549415111542,
+      "learning_rate": 8.06493692640325e-05,
+      "loss": 0.2878,
+      "step": 1444
+    },
+    {
+      "epoch": 0.3051742344244984,
+      "grad_norm": 0.3823314905166626,
+      "learning_rate": 8.06225860725484e-05,
+      "loss": 0.2738,
+      "step": 1445
+    },
+    {
+      "epoch": 0.30538542766631466,
+      "grad_norm": 0.4025537967681885,
+      "learning_rate": 8.059578881274827e-05,
+      "loss": 0.2755,
+      "step": 1446
+    },
+    {
+      "epoch": 0.30559662090813094,
+      "grad_norm": 1.5882189273834229,
+      "learning_rate": 8.056897749694303e-05,
+      "loss": 0.2759,
+      "step": 1447
+    },
+    {
+      "epoch": 0.3058078141499472,
+      "grad_norm": 0.4432904124259949,
+      "learning_rate": 8.054215213745006e-05,
+      "loss": 0.2737,
+      "step": 1448
+    },
+    {
+      "epoch": 0.30601900739176346,
+      "grad_norm": 0.6036256551742554,
+      "learning_rate": 8.05153127465932e-05,
+      "loss": 0.2899,
+      "step": 1449
+    },
+    {
+      "epoch": 0.30623020063357975,
+      "grad_norm": 0.41523632407188416,
+      "learning_rate": 8.048845933670273e-05,
+      "loss": 0.2814,
+      "step": 1450
+    },
+    {
+      "epoch": 0.306441393875396,
+      "grad_norm": 0.5620731115341187,
+      "learning_rate": 8.046159192011533e-05,
+      "loss": 0.2884,
+      "step": 1451
+    },
+    {
+      "epoch": 0.30665258711721227,
+      "grad_norm": 0.32544565200805664,
+      "learning_rate": 8.043471050917421e-05,
+      "loss": 0.2835,
+      "step": 1452
+    },
+    {
+      "epoch": 0.3068637803590285,
+      "grad_norm": 0.5412189364433289,
+      "learning_rate": 8.040781511622894e-05,
+      "loss": 0.2937,
+      "step": 1453
+    },
+    {
+      "epoch": 0.3070749736008448,
+      "grad_norm": 0.3798193633556366,
+      "learning_rate": 8.038090575363552e-05,
+      "loss": 0.2826,
+      "step": 1454
+    },
+    {
+      "epoch": 0.307286166842661,
+      "grad_norm": 0.528525173664093,
+      "learning_rate": 8.035398243375636e-05,
+      "loss": 0.2928,
+      "step": 1455
+    },
+    {
+      "epoch": 0.3074973600844773,
+      "grad_norm": 0.4857248067855835,
+      "learning_rate": 8.03270451689603e-05,
+      "loss": 0.2935,
+      "step": 1456
+    },
+    {
+      "epoch": 0.30770855332629354,
+      "grad_norm": 0.7164490222930908,
+      "learning_rate": 8.030009397162261e-05,
+      "loss": 0.281,
+      "step": 1457
+    },
+    {
+      "epoch": 0.30791974656810983,
+      "grad_norm": 0.46731099486351013,
+      "learning_rate": 8.027312885412493e-05,
+      "loss": 0.2976,
+      "step": 1458
+    },
+    {
+      "epoch": 0.30813093980992606,
+      "grad_norm": 0.3495425581932068,
+      "learning_rate": 8.024614982885527e-05,
+      "loss": 0.2841,
+      "step": 1459
+    },
+    {
+      "epoch": 0.30834213305174235,
+      "grad_norm": 0.41502735018730164,
+      "learning_rate": 8.021915690820808e-05,
+      "loss": 0.2852,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3085533262935586,
+      "grad_norm": 0.4496575593948364,
+      "learning_rate": 8.019215010458417e-05,
+      "loss": 0.2832,
+      "step": 1461
+    },
+    {
+      "epoch": 0.30876451953537487,
+      "grad_norm": 0.6821022033691406,
+      "learning_rate": 8.016512943039072e-05,
+      "loss": 0.2831,
+      "step": 1462
+    },
+    {
+      "epoch": 0.3089757127771911,
+      "grad_norm": 0.4880288243293762,
+      "learning_rate": 8.01380948980413e-05,
+      "loss": 0.2903,
+      "step": 1463
+    },
+    {
+      "epoch": 0.3091869060190074,
+      "grad_norm": 0.38595178723335266,
+      "learning_rate": 8.011104651995585e-05,
+      "loss": 0.2778,
+      "step": 1464
+    },
+    {
+      "epoch": 0.3093980992608237,
+      "grad_norm": 0.43524447083473206,
+      "learning_rate": 8.008398430856063e-05,
+      "loss": 0.2895,
+      "step": 1465
+    },
+    {
+      "epoch": 0.3096092925026399,
+      "grad_norm": 0.40644416213035583,
+      "learning_rate": 8.005690827628831e-05,
+      "loss": 0.2867,
+      "step": 1466
+    },
+    {
+      "epoch": 0.3098204857444562,
+      "grad_norm": 0.3350018858909607,
+      "learning_rate": 8.002981843557788e-05,
+      "loss": 0.2812,
+      "step": 1467
+    },
+    {
+      "epoch": 0.31003167898627243,
+      "grad_norm": 0.3972829580307007,
+      "learning_rate": 8.000271479887468e-05,
+      "loss": 0.2838,
+      "step": 1468
+    },
+    {
+      "epoch": 0.3102428722280887,
+      "grad_norm": 0.31908780336380005,
+      "learning_rate": 7.997559737863039e-05,
+      "loss": 0.2756,
+      "step": 1469
+    },
+    {
+      "epoch": 0.31045406546990495,
+      "grad_norm": 0.542262852191925,
+      "learning_rate": 7.994846618730301e-05,
+      "loss": 0.2826,
+      "step": 1470
+    },
+    {
+      "epoch": 0.31066525871172124,
+      "grad_norm": 0.38031938672065735,
+      "learning_rate": 7.992132123735687e-05,
+      "loss": 0.2819,
+      "step": 1471
+    },
+    {
+      "epoch": 0.31087645195353747,
+      "grad_norm": 0.42320671677589417,
+      "learning_rate": 7.989416254126263e-05,
+      "loss": 0.2814,
+      "step": 1472
+    },
+    {
+      "epoch": 0.31108764519535376,
+      "grad_norm": 0.3276204764842987,
+      "learning_rate": 7.986699011149725e-05,
+      "loss": 0.2808,
+      "step": 1473
+    },
+    {
+      "epoch": 0.31129883843717,
+      "grad_norm": 0.36513686180114746,
+      "learning_rate": 7.983980396054405e-05,
+      "loss": 0.2847,
+      "step": 1474
+    },
+    {
+      "epoch": 0.3115100316789863,
+      "grad_norm": 0.29972290992736816,
+      "learning_rate": 7.981260410089257e-05,
+      "loss": 0.2787,
+      "step": 1475
+    },
+    {
+      "epoch": 0.3117212249208025,
+      "grad_norm": 0.3682856559753418,
+      "learning_rate": 7.978539054503873e-05,
+      "loss": 0.2806,
+      "step": 1476
+    },
+    {
+      "epoch": 0.3119324181626188,
+      "grad_norm": 0.4157532751560211,
+      "learning_rate": 7.975816330548467e-05,
+      "loss": 0.2742,
+      "step": 1477
+    },
+    {
+      "epoch": 0.3121436114044351,
+      "grad_norm": 0.3274967074394226,
+      "learning_rate": 7.973092239473887e-05,
+      "loss": 0.2758,
+      "step": 1478
+    },
+    {
+      "epoch": 0.3123548046462513,
+      "grad_norm": 0.3120342493057251,
+      "learning_rate": 7.970366782531606e-05,
+      "loss": 0.276,
+      "step": 1479
+    },
+    {
+      "epoch": 0.3125659978880676,
+      "grad_norm": 0.34169700741767883,
+      "learning_rate": 7.967639960973726e-05,
+      "loss": 0.2733,
+      "step": 1480
+    },
+    {
+      "epoch": 0.31277719112988384,
+      "grad_norm": 0.32815298438072205,
+      "learning_rate": 7.964911776052977e-05,
+      "loss": 0.2755,
+      "step": 1481
+    },
+    {
+      "epoch": 0.3129883843717001,
+      "grad_norm": 0.360624223947525,
+      "learning_rate": 7.962182229022712e-05,
+      "loss": 0.269,
+      "step": 1482
+    },
+    {
+      "epoch": 0.31319957761351636,
+      "grad_norm": 0.33829376101493835,
+      "learning_rate": 7.959451321136911e-05,
+      "loss": 0.2776,
+      "step": 1483
+    },
+    {
+      "epoch": 0.31341077085533264,
+      "grad_norm": 0.31415605545043945,
+      "learning_rate": 7.956719053650182e-05,
+      "loss": 0.2705,
+      "step": 1484
+    },
+    {
+      "epoch": 0.3136219640971489,
+      "grad_norm": 0.36182916164398193,
+      "learning_rate": 7.953985427817756e-05,
+      "loss": 0.2758,
+      "step": 1485
+    },
+    {
+      "epoch": 0.31383315733896516,
+      "grad_norm": 0.3341808021068573,
+      "learning_rate": 7.951250444895484e-05,
+      "loss": 0.2711,
+      "step": 1486
+    },
+    {
+      "epoch": 0.3140443505807814,
+      "grad_norm": 0.43559330701828003,
+      "learning_rate": 7.948514106139846e-05,
+      "loss": 0.2723,
+      "step": 1487
+    },
+    {
+      "epoch": 0.3142555438225977,
+      "grad_norm": 0.36032235622406006,
+      "learning_rate": 7.945776412807945e-05,
+      "loss": 0.282,
+      "step": 1488
+    },
+    {
+      "epoch": 0.3144667370644139,
+      "grad_norm": 0.37380170822143555,
+      "learning_rate": 7.943037366157499e-05,
+      "loss": 0.2703,
+      "step": 1489
+    },
+    {
+      "epoch": 0.3146779303062302,
+      "grad_norm": 0.3804195523262024,
+      "learning_rate": 7.940296967446859e-05,
+      "loss": 0.2703,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3148891235480465,
+      "grad_norm": 0.3498181104660034,
+      "learning_rate": 7.937555217934984e-05,
+      "loss": 0.2735,
+      "step": 1491
+    },
+    {
+      "epoch": 0.3151003167898627,
+      "grad_norm": 0.3623705506324768,
+      "learning_rate": 7.934812118881468e-05,
+      "loss": 0.2702,
+      "step": 1492
+    },
+    {
+      "epoch": 0.315311510031679,
+      "grad_norm": 0.3450925648212433,
+      "learning_rate": 7.932067671546512e-05,
+      "loss": 0.2791,
+      "step": 1493
+    },
+    {
+      "epoch": 0.31552270327349524,
+      "grad_norm": 0.5411653518676758,
+      "learning_rate": 7.929321877190944e-05,
+      "loss": 0.2743,
+      "step": 1494
+    },
+    {
+      "epoch": 0.31573389651531153,
+      "grad_norm": 0.4627842307090759,
+      "learning_rate": 7.926574737076211e-05,
+      "loss": 0.2698,
+      "step": 1495
+    },
+    {
+      "epoch": 0.31594508975712776,
+      "grad_norm": 0.38352248072624207,
+      "learning_rate": 7.923826252464372e-05,
+      "loss": 0.2776,
+      "step": 1496
+    },
+    {
+      "epoch": 0.31615628299894405,
+      "grad_norm": 0.3260023891925812,
+      "learning_rate": 7.92107642461811e-05,
+      "loss": 0.2723,
+      "step": 1497
+    },
+    {
+      "epoch": 0.3163674762407603,
+      "grad_norm": 0.3864498734474182,
+      "learning_rate": 7.918325254800722e-05,
+      "loss": 0.2735,
+      "step": 1498
+    },
+    {
+      "epoch": 0.31657866948257657,
+      "grad_norm": 0.3672948479652405,
+      "learning_rate": 7.915572744276123e-05,
+      "loss": 0.2759,
+      "step": 1499
+    },
+    {
+      "epoch": 0.3167898627243928,
+      "grad_norm": 0.4404833912849426,
+      "learning_rate": 7.912818894308845e-05,
+      "loss": 0.2694,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3170010559662091,
+      "grad_norm": 0.33910831809043884,
+      "learning_rate": 7.91006370616403e-05,
+      "loss": 0.2725,
+      "step": 1501
+    },
+    {
+      "epoch": 0.3172122492080253,
+      "grad_norm": 0.3514660596847534,
+      "learning_rate": 7.90730718110744e-05,
+      "loss": 0.2688,
+      "step": 1502
+    },
+    {
+      "epoch": 0.3174234424498416,
+      "grad_norm": 0.3371667265892029,
+      "learning_rate": 7.90454932040545e-05,
+      "loss": 0.2703,
+      "step": 1503
+    },
+    {
+      "epoch": 0.31763463569165784,
+      "grad_norm": 0.38085615634918213,
+      "learning_rate": 7.901790125325047e-05,
+      "loss": 0.2751,
+      "step": 1504
+    },
+    {
+      "epoch": 0.31784582893347413,
+      "grad_norm": 0.3379676342010498,
+      "learning_rate": 7.899029597133835e-05,
+      "loss": 0.2677,
+      "step": 1505
+    },
+    {
+      "epoch": 0.3180570221752904,
+      "grad_norm": 0.342742919921875,
+      "learning_rate": 7.896267737100027e-05,
+      "loss": 0.2722,
+      "step": 1506
+    },
+    {
+      "epoch": 0.31826821541710665,
+      "grad_norm": 0.323000431060791,
+      "learning_rate": 7.893504546492447e-05,
+      "loss": 0.2752,
+      "step": 1507
+    },
+    {
+      "epoch": 0.31847940865892294,
+      "grad_norm": 0.3610881268978119,
+      "learning_rate": 7.890740026580532e-05,
+      "loss": 0.2726,
+      "step": 1508
+    },
+    {
+      "epoch": 0.31869060190073917,
+      "grad_norm": 0.3535216152667999,
+      "learning_rate": 7.88797417863433e-05,
+      "loss": 0.2708,
+      "step": 1509
+    },
+    {
+      "epoch": 0.31890179514255546,
+      "grad_norm": 0.34378522634506226,
+      "learning_rate": 7.885207003924498e-05,
+      "loss": 0.2738,
+      "step": 1510
+    },
+    {
+      "epoch": 0.3191129883843717,
+      "grad_norm": 0.4807360768318176,
+      "learning_rate": 7.882438503722304e-05,
+      "loss": 0.2769,
+      "step": 1511
+    },
+    {
+      "epoch": 0.319324181626188,
+      "grad_norm": 0.5306156873703003,
+      "learning_rate": 7.879668679299624e-05,
+      "loss": 0.2707,
+      "step": 1512
+    },
+    {
+      "epoch": 0.3195353748680042,
+      "grad_norm": 0.717239260673523,
+      "learning_rate": 7.876897531928943e-05,
+      "loss": 0.2754,
+      "step": 1513
+    },
+    {
+      "epoch": 0.3197465681098205,
+      "grad_norm": 0.4456329643726349,
+      "learning_rate": 7.874125062883351e-05,
+      "loss": 0.2815,
+      "step": 1514
+    },
+    {
+      "epoch": 0.3199577613516367,
+      "grad_norm": 0.39542269706726074,
+      "learning_rate": 7.871351273436551e-05,
+      "loss": 0.2828,
+      "step": 1515
+    },
+    {
+      "epoch": 0.320168954593453,
+      "grad_norm": 0.29692500829696655,
+      "learning_rate": 7.868576164862847e-05,
+      "loss": 0.273,
+      "step": 1516
+    },
+    {
+      "epoch": 0.32038014783526925,
+      "grad_norm": 0.41348621249198914,
+      "learning_rate": 7.865799738437148e-05,
+      "loss": 0.2751,
+      "step": 1517
+    },
+    {
+      "epoch": 0.32059134107708553,
+      "grad_norm": 0.4103499948978424,
+      "learning_rate": 7.863021995434978e-05,
+      "loss": 0.2719,
+      "step": 1518
+    },
+    {
+      "epoch": 0.3208025343189018,
+      "grad_norm": 0.3981059789657593,
+      "learning_rate": 7.860242937132454e-05,
+      "loss": 0.2828,
+      "step": 1519
+    },
+    {
+      "epoch": 0.32101372756071805,
+      "grad_norm": 0.4978472590446472,
+      "learning_rate": 7.857462564806306e-05,
+      "loss": 0.2744,
+      "step": 1520
+    },
+    {
+      "epoch": 0.32122492080253434,
+      "grad_norm": 0.5455237030982971,
+      "learning_rate": 7.854680879733862e-05,
+      "loss": 0.2759,
+      "step": 1521
+    },
+    {
+      "epoch": 0.3214361140443506,
+      "grad_norm": 0.5992631912231445,
+      "learning_rate": 7.851897883193057e-05,
+      "loss": 0.2766,
+      "step": 1522
+    },
+    {
+      "epoch": 0.32164730728616686,
+      "grad_norm": 0.40774473547935486,
+      "learning_rate": 7.849113576462425e-05,
+      "loss": 0.278,
+      "step": 1523
+    },
+    {
+      "epoch": 0.3218585005279831,
+      "grad_norm": 0.39149951934814453,
+      "learning_rate": 7.846327960821107e-05,
+      "loss": 0.2765,
+      "step": 1524
+    },
+    {
+      "epoch": 0.3220696937697994,
+      "grad_norm": 0.35199829936027527,
+      "learning_rate": 7.843541037548838e-05,
+      "loss": 0.2717,
+      "step": 1525
+    },
+    {
+      "epoch": 0.3222808870116156,
+      "grad_norm": 0.32971954345703125,
+      "learning_rate": 7.84075280792596e-05,
+      "loss": 0.2744,
+      "step": 1526
+    },
+    {
+      "epoch": 0.3224920802534319,
+      "grad_norm": 0.44395220279693604,
+      "learning_rate": 7.837963273233414e-05,
+      "loss": 0.2714,
+      "step": 1527
+    },
+    {
+      "epoch": 0.32270327349524813,
+      "grad_norm": 0.364523321390152,
+      "learning_rate": 7.835172434752738e-05,
+      "loss": 0.2747,
+      "step": 1528
+    },
+    {
+      "epoch": 0.3229144667370644,
+      "grad_norm": 0.48613715171813965,
+      "learning_rate": 7.832380293766071e-05,
+      "loss": 0.2767,
+      "step": 1529
+    },
+    {
+      "epoch": 0.32312565997888065,
+      "grad_norm": 0.3460061848163605,
+      "learning_rate": 7.82958685155615e-05,
+      "loss": 0.2852,
+      "step": 1530
+    },
+    {
+      "epoch": 0.32333685322069694,
+      "grad_norm": 0.541907548904419,
+      "learning_rate": 7.826792109406308e-05,
+      "loss": 0.2769,
+      "step": 1531
+    },
+    {
+      "epoch": 0.3235480464625132,
+      "grad_norm": 0.5246310234069824,
+      "learning_rate": 7.823996068600481e-05,
+      "loss": 0.2735,
+      "step": 1532
+    },
+    {
+      "epoch": 0.32375923970432946,
+      "grad_norm": 0.30062198638916016,
+      "learning_rate": 7.821198730423194e-05,
+      "loss": 0.2739,
+      "step": 1533
+    },
+    {
+      "epoch": 0.32397043294614575,
+      "grad_norm": 0.3406190574169159,
+      "learning_rate": 7.818400096159572e-05,
+      "loss": 0.2718,
+      "step": 1534
+    },
+    {
+      "epoch": 0.324181626187962,
+      "grad_norm": 0.368279367685318,
+      "learning_rate": 7.815600167095338e-05,
+      "loss": 0.2777,
+      "step": 1535
+    },
+    {
+      "epoch": 0.32439281942977827,
+      "grad_norm": 0.335226446390152,
+      "learning_rate": 7.812798944516802e-05,
+      "loss": 0.2745,
+      "step": 1536
+    },
+    {
+      "epoch": 0.3246040126715945,
+      "grad_norm": 0.3566719591617584,
+      "learning_rate": 7.809996429710877e-05,
+      "loss": 0.2743,
+      "step": 1537
+    },
+    {
+      "epoch": 0.3248152059134108,
+      "grad_norm": 0.40046578645706177,
+      "learning_rate": 7.807192623965064e-05,
+      "loss": 0.2788,
+      "step": 1538
+    },
+    {
+      "epoch": 0.325026399155227,
+      "grad_norm": 0.4136143922805786,
+      "learning_rate": 7.804387528567459e-05,
+      "loss": 0.2751,
+      "step": 1539
+    },
+    {
+      "epoch": 0.3252375923970433,
+      "grad_norm": 0.4305802583694458,
+      "learning_rate": 7.801581144806752e-05,
+      "loss": 0.2663,
+      "step": 1540
+    },
+    {
+      "epoch": 0.32544878563885954,
+      "grad_norm": 0.507694661617279,
+      "learning_rate": 7.798773473972219e-05,
+      "loss": 0.2692,
+      "step": 1541
+    },
+    {
+      "epoch": 0.3256599788806758,
+      "grad_norm": 0.46440115571022034,
+      "learning_rate": 7.795964517353735e-05,
+      "loss": 0.2767,
+      "step": 1542
+    },
+    {
+      "epoch": 0.32587117212249206,
+      "grad_norm": 0.4096996784210205,
+      "learning_rate": 7.793154276241761e-05,
+      "loss": 0.2673,
+      "step": 1543
+    },
+    {
+      "epoch": 0.32608236536430835,
+      "grad_norm": 0.34040892124176025,
+      "learning_rate": 7.790342751927351e-05,
+      "loss": 0.2806,
+      "step": 1544
+    },
+    {
+      "epoch": 0.3262935586061246,
+      "grad_norm": 0.32686254382133484,
+      "learning_rate": 7.787529945702144e-05,
+      "loss": 0.2747,
+      "step": 1545
+    },
+    {
+      "epoch": 0.32650475184794087,
+      "grad_norm": 0.3229081928730011,
+      "learning_rate": 7.784715858858374e-05,
+      "loss": 0.2725,
+      "step": 1546
+    },
+    {
+      "epoch": 0.32671594508975715,
+      "grad_norm": 0.31223878264427185,
+      "learning_rate": 7.781900492688858e-05,
+      "loss": 0.2829,
+      "step": 1547
+    },
+    {
+      "epoch": 0.3269271383315734,
+      "grad_norm": 0.37542206048965454,
+      "learning_rate": 7.779083848487001e-05,
+      "loss": 0.2779,
+      "step": 1548
+    },
+    {
+      "epoch": 0.3271383315733897,
+      "grad_norm": 0.3160516023635864,
+      "learning_rate": 7.776265927546799e-05,
+      "loss": 0.2642,
+      "step": 1549
+    },
+    {
+      "epoch": 0.3273495248152059,
+      "grad_norm": 0.379178524017334,
+      "learning_rate": 7.773446731162834e-05,
+      "loss": 0.2765,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3275607180570222,
+      "grad_norm": 0.36190932989120483,
+      "learning_rate": 7.770626260630272e-05,
+      "loss": 0.2718,
+      "step": 1551
+    },
+    {
+      "epoch": 0.3277719112988384,
+      "grad_norm": 0.3031412959098816,
+      "learning_rate": 7.767804517244861e-05,
+      "loss": 0.2783,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3279831045406547,
+      "grad_norm": 0.3426854610443115,
+      "learning_rate": 7.764981502302941e-05,
+      "loss": 0.269,
+      "step": 1553
+    },
+    {
+      "epoch": 0.32819429778247095,
+      "grad_norm": 0.4297191798686981,
+      "learning_rate": 7.762157217101434e-05,
+      "loss": 0.2728,
+      "step": 1554
+    },
+    {
+      "epoch": 0.32840549102428723,
+      "grad_norm": 0.4441051483154297,
+      "learning_rate": 7.759331662937841e-05,
+      "loss": 0.2775,
+      "step": 1555
+    },
+    {
+      "epoch": 0.32861668426610346,
+      "grad_norm": 0.37604859471321106,
+      "learning_rate": 7.75650484111025e-05,
+      "loss": 0.267,
+      "step": 1556
+    },
+    {
+      "epoch": 0.32882787750791975,
+      "grad_norm": 0.4371793270111084,
+      "learning_rate": 7.753676752917331e-05,
+      "loss": 0.2746,
+      "step": 1557
+    },
+    {
+      "epoch": 0.329039070749736,
+      "grad_norm": 0.5159371495246887,
+      "learning_rate": 7.750847399658336e-05,
+      "loss": 0.2759,
+      "step": 1558
+    },
+    {
+      "epoch": 0.32925026399155227,
+      "grad_norm": 0.32939767837524414,
+      "learning_rate": 7.748016782633098e-05,
+      "loss": 0.2766,
+      "step": 1559
+    },
+    {
+      "epoch": 0.32946145723336856,
+      "grad_norm": 0.46038082242012024,
+      "learning_rate": 7.745184903142028e-05,
+      "loss": 0.2825,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3296726504751848,
+      "grad_norm": 0.342672199010849,
+      "learning_rate": 7.742351762486122e-05,
+      "loss": 0.2778,
+      "step": 1561
+    },
+    {
+      "epoch": 0.3298838437170011,
+      "grad_norm": 0.39122793078422546,
+      "learning_rate": 7.739517361966952e-05,
+      "loss": 0.2745,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3300950369588173,
+      "grad_norm": 0.3721182942390442,
+      "learning_rate": 7.736681702886666e-05,
+      "loss": 0.2738,
+      "step": 1563
+    },
+    {
+      "epoch": 0.3303062302006336,
+      "grad_norm": 0.3386889696121216,
+      "learning_rate": 7.733844786547997e-05,
+      "loss": 0.2771,
+      "step": 1564
+    },
+    {
+      "epoch": 0.33051742344244983,
+      "grad_norm": 0.42348983883857727,
+      "learning_rate": 7.731006614254251e-05,
+      "loss": 0.2743,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3307286166842661,
+      "grad_norm": 0.37283334136009216,
+      "learning_rate": 7.728167187309313e-05,
+      "loss": 0.2724,
+      "step": 1566
+    },
+    {
+      "epoch": 0.33093980992608235,
+      "grad_norm": 0.3644734025001526,
+      "learning_rate": 7.725326507017643e-05,
+      "loss": 0.2687,
+      "step": 1567
+    },
+    {
+      "epoch": 0.33115100316789864,
+      "grad_norm": 0.47546103596687317,
+      "learning_rate": 7.722484574684276e-05,
+      "loss": 0.2815,
+      "step": 1568
+    },
+    {
+      "epoch": 0.33136219640971487,
+      "grad_norm": 0.31447702646255493,
+      "learning_rate": 7.719641391614827e-05,
+      "loss": 0.2712,
+      "step": 1569
+    },
+    {
+      "epoch": 0.33157338965153116,
+      "grad_norm": 0.43051278591156006,
+      "learning_rate": 7.716796959115478e-05,
+      "loss": 0.2755,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3317845828933474,
+      "grad_norm": 0.4824818968772888,
+      "learning_rate": 7.713951278492991e-05,
+      "loss": 0.2719,
+      "step": 1571
+    },
+    {
+      "epoch": 0.3319957761351637,
+      "grad_norm": 0.39186710119247437,
+      "learning_rate": 7.711104351054701e-05,
+      "loss": 0.2734,
+      "step": 1572
+    },
+    {
+      "epoch": 0.3322069693769799,
+      "grad_norm": 0.8596894145011902,
+      "learning_rate": 7.708256178108513e-05,
+      "loss": 0.2737,
+      "step": 1573
+    },
+    {
+      "epoch": 0.3324181626187962,
+      "grad_norm": 0.4611983597278595,
+      "learning_rate": 7.705406760962906e-05,
+      "loss": 0.2854,
+      "step": 1574
+    },
+    {
+      "epoch": 0.3326293558606125,
+      "grad_norm": 0.3612045645713806,
+      "learning_rate": 7.702556100926929e-05,
+      "loss": 0.2784,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3328405491024287,
+      "grad_norm": 0.4009132385253906,
+      "learning_rate": 7.699704199310204e-05,
+      "loss": 0.2782,
+      "step": 1576
+    },
+    {
+      "epoch": 0.333051742344245,
+      "grad_norm": 0.3800610601902008,
+      "learning_rate": 7.696851057422922e-05,
+      "loss": 0.2791,
+      "step": 1577
+    },
+    {
+      "epoch": 0.33326293558606124,
+      "grad_norm": 0.38686516880989075,
+      "learning_rate": 7.693996676575845e-05,
+      "loss": 0.2705,
+      "step": 1578
+    },
+    {
+      "epoch": 0.3334741288278775,
+      "grad_norm": 0.4652829170227051,
+      "learning_rate": 7.691141058080302e-05,
+      "loss": 0.2763,
+      "step": 1579
+    },
+    {
+      "epoch": 0.33368532206969376,
+      "grad_norm": 0.4610855281352997,
+      "learning_rate": 7.688284203248196e-05,
+      "loss": 0.2648,
+      "step": 1580
+    },
+    {
+      "epoch": 0.33389651531151004,
+      "grad_norm": 0.514232337474823,
+      "learning_rate": 7.68542611339199e-05,
+      "loss": 0.2626,
+      "step": 1581
+    },
+    {
+      "epoch": 0.3341077085533263,
+      "grad_norm": 0.40862902998924255,
+      "learning_rate": 7.682566789824722e-05,
+      "loss": 0.2702,
+      "step": 1582
+    },
+    {
+      "epoch": 0.33431890179514256,
+      "grad_norm": 0.40841519832611084,
+      "learning_rate": 7.679706233859989e-05,
+      "loss": 0.2709,
+      "step": 1583
+    },
+    {
+      "epoch": 0.3345300950369588,
+      "grad_norm": 0.3859175145626068,
+      "learning_rate": 7.676844446811961e-05,
+      "loss": 0.2682,
+      "step": 1584
+    },
+    {
+      "epoch": 0.3347412882787751,
+      "grad_norm": 0.37468379735946655,
+      "learning_rate": 7.673981429995372e-05,
+      "loss": 0.2675,
+      "step": 1585
+    },
+    {
+      "epoch": 0.3349524815205913,
+      "grad_norm": 0.30321672558784485,
+      "learning_rate": 7.671117184725518e-05,
+      "loss": 0.2799,
+      "step": 1586
+    },
+    {
+      "epoch": 0.3351636747624076,
+      "grad_norm": 0.512050986289978,
+      "learning_rate": 7.668251712318262e-05,
+      "loss": 0.2743,
+      "step": 1587
+    },
+    {
+      "epoch": 0.3353748680042239,
+      "grad_norm": 0.5246691107749939,
+      "learning_rate": 7.665385014090029e-05,
+      "loss": 0.2757,
+      "step": 1588
+    },
+    {
+      "epoch": 0.3355860612460401,
+      "grad_norm": 0.4708915650844574,
+      "learning_rate": 7.662517091357811e-05,
+      "loss": 0.2716,
+      "step": 1589
+    },
+    {
+      "epoch": 0.3357972544878564,
+      "grad_norm": 0.43306973576545715,
+      "learning_rate": 7.659647945439156e-05,
+      "loss": 0.2621,
+      "step": 1590
+    },
+    {
+      "epoch": 0.33600844772967264,
+      "grad_norm": 0.34662356972694397,
+      "learning_rate": 7.656777577652181e-05,
+      "loss": 0.2647,
+      "step": 1591
+    },
+    {
+      "epoch": 0.33621964097148893,
+      "grad_norm": 0.4028608202934265,
+      "learning_rate": 7.65390598931556e-05,
+      "loss": 0.2695,
+      "step": 1592
+    },
+    {
+      "epoch": 0.33643083421330516,
+      "grad_norm": 0.3623037338256836,
+      "learning_rate": 7.65103318174853e-05,
+      "loss": 0.2759,
+      "step": 1593
+    },
+    {
+      "epoch": 0.33664202745512145,
+      "grad_norm": 0.28270894289016724,
+      "learning_rate": 7.648159156270884e-05,
+      "loss": 0.2712,
+      "step": 1594
+    },
+    {
+      "epoch": 0.3368532206969377,
+      "grad_norm": 0.3580645024776459,
+      "learning_rate": 7.645283914202981e-05,
+      "loss": 0.2747,
+      "step": 1595
+    },
+    {
+      "epoch": 0.33706441393875397,
+      "grad_norm": 0.40753814578056335,
+      "learning_rate": 7.642407456865733e-05,
+      "loss": 0.2711,
+      "step": 1596
+    },
+    {
+      "epoch": 0.3372756071805702,
+      "grad_norm": 0.34346628189086914,
+      "learning_rate": 7.639529785580613e-05,
+      "loss": 0.2641,
+      "step": 1597
+    },
+    {
+      "epoch": 0.3374868004223865,
+      "grad_norm": 0.3840825855731964,
+      "learning_rate": 7.63665090166965e-05,
+      "loss": 0.274,
+      "step": 1598
+    },
+    {
+      "epoch": 0.3376979936642027,
+      "grad_norm": 0.31534719467163086,
+      "learning_rate": 7.633770806455436e-05,
+      "loss": 0.278,
+      "step": 1599
+    },
+    {
+      "epoch": 0.337909186906019,
+      "grad_norm": 0.4307127892971039,
+      "learning_rate": 7.630889501261109e-05,
+      "loss": 0.2729,
+      "step": 1600
+    },
+    {
+      "epoch": 0.337909186906019,
+      "eval_train_split_loss": 0.34251803159713745,
+      "eval_train_split_runtime": 6.5759,
+      "eval_train_split_samples_per_second": 1.521,
+      "eval_train_split_steps_per_second": 0.152,
+      "step": 1600
+    },
+    {
+      "epoch": 0.33812038014783524,
+      "grad_norm": 0.393484890460968,
+      "learning_rate": 7.628006987410374e-05,
+      "loss": 0.2758,
+      "step": 1601
+    },
+    {
+      "epoch": 0.33833157338965153,
+      "grad_norm": 0.4438894987106323,
+      "learning_rate": 7.625123266227483e-05,
+      "loss": 0.2762,
+      "step": 1602
+    },
+    {
+      "epoch": 0.3385427666314678,
+      "grad_norm": 0.36399173736572266,
+      "learning_rate": 7.622238339037247e-05,
+      "loss": 0.2701,
+      "step": 1603
+    },
+    {
+      "epoch": 0.33875395987328405,
+      "grad_norm": 0.390704482793808,
+      "learning_rate": 7.619352207165031e-05,
+      "loss": 0.2736,
+      "step": 1604
+    },
+    {
+      "epoch": 0.33896515311510034,
+      "grad_norm": 0.4204282760620117,
+      "learning_rate": 7.616464871936749e-05,
+      "loss": 0.2669,
+      "step": 1605
+    },
+    {
+      "epoch": 0.33917634635691657,
+      "grad_norm": 0.5263951420783997,
+      "learning_rate": 7.613576334678872e-05,
+      "loss": 0.2698,
+      "step": 1606
+    },
+    {
+      "epoch": 0.33938753959873286,
+      "grad_norm": 0.45608505606651306,
+      "learning_rate": 7.610686596718426e-05,
+      "loss": 0.2733,
+      "step": 1607
+    },
+    {
+      "epoch": 0.3395987328405491,
+      "grad_norm": 0.29659056663513184,
+      "learning_rate": 7.607795659382981e-05,
+      "loss": 0.2717,
+      "step": 1608
+    },
+    {
+      "epoch": 0.3398099260823654,
+      "grad_norm": 0.3872021734714508,
+      "learning_rate": 7.604903524000662e-05,
+      "loss": 0.2663,
+      "step": 1609
+    },
+    {
+      "epoch": 0.3400211193241816,
+      "grad_norm": 0.30935078859329224,
+      "learning_rate": 7.602010191900146e-05,
+      "loss": 0.2662,
+      "step": 1610
+    },
+    {
+      "epoch": 0.3402323125659979,
+      "grad_norm": 0.6032463312149048,
+      "learning_rate": 7.599115664410659e-05,
+      "loss": 0.2704,
+      "step": 1611
+    },
+    {
+      "epoch": 0.34044350580781413,
+      "grad_norm": 0.44057250022888184,
+      "learning_rate": 7.596219942861971e-05,
+      "loss": 0.2723,
+      "step": 1612
+    },
+    {
+      "epoch": 0.3406546990496304,
+      "grad_norm": 0.5571770668029785,
+      "learning_rate": 7.59332302858441e-05,
+      "loss": 0.2774,
+      "step": 1613
+    },
+    {
+      "epoch": 0.34086589229144665,
+      "grad_norm": 0.4296833574771881,
+      "learning_rate": 7.590424922908844e-05,
+      "loss": 0.2736,
+      "step": 1614
+    },
+    {
+      "epoch": 0.34107708553326294,
+      "grad_norm": 0.3137790560722351,
+      "learning_rate": 7.58752562716669e-05,
+      "loss": 0.2675,
+      "step": 1615
+    },
+    {
+      "epoch": 0.3412882787750792,
+      "grad_norm": 0.35367098450660706,
+      "learning_rate": 7.584625142689916e-05,
+      "loss": 0.2667,
+      "step": 1616
+    },
+    {
+      "epoch": 0.34149947201689546,
+      "grad_norm": 0.39384177327156067,
+      "learning_rate": 7.581723470811031e-05,
+      "loss": 0.2656,
+      "step": 1617
+    },
+    {
+      "epoch": 0.34171066525871174,
+      "grad_norm": 0.3373470902442932,
+      "learning_rate": 7.578820612863093e-05,
+      "loss": 0.2694,
+      "step": 1618
+    },
+    {
+      "epoch": 0.341921858500528,
+      "grad_norm": 0.35782402753829956,
+      "learning_rate": 7.575916570179701e-05,
+      "loss": 0.2727,
+      "step": 1619
+    },
+    {
+      "epoch": 0.34213305174234426,
+      "grad_norm": 0.3433518409729004,
+      "learning_rate": 7.573011344095003e-05,
+      "loss": 0.2702,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3423442449841605,
+      "grad_norm": 0.3651270270347595,
+      "learning_rate": 7.570104935943685e-05,
+      "loss": 0.2683,
+      "step": 1621
+    },
+    {
+      "epoch": 0.3425554382259768,
+      "grad_norm": 0.4034000337123871,
+      "learning_rate": 7.567197347060984e-05,
+      "loss": 0.2687,
+      "step": 1622
+    },
+    {
+      "epoch": 0.342766631467793,
+      "grad_norm": 0.3172040581703186,
+      "learning_rate": 7.56428857878267e-05,
+      "loss": 0.2665,
+      "step": 1623
+    },
+    {
+      "epoch": 0.3429778247096093,
+      "grad_norm": 0.29668810963630676,
+      "learning_rate": 7.561378632445066e-05,
+      "loss": 0.269,
+      "step": 1624
+    },
+    {
+      "epoch": 0.34318901795142553,
+      "grad_norm": 0.375493586063385,
+      "learning_rate": 7.558467509385023e-05,
+      "loss": 0.2718,
+      "step": 1625
+    },
+    {
+      "epoch": 0.3434002111932418,
+      "grad_norm": 0.4628557860851288,
+      "learning_rate": 7.555555210939944e-05,
+      "loss": 0.2722,
+      "step": 1626
+    },
+    {
+      "epoch": 0.34361140443505805,
+      "grad_norm": 0.3442576229572296,
+      "learning_rate": 7.552641738447763e-05,
+      "loss": 0.2668,
+      "step": 1627
+    },
+    {
+      "epoch": 0.34382259767687434,
+      "grad_norm": 0.35542237758636475,
+      "learning_rate": 7.549727093246963e-05,
+      "loss": 0.2718,
+      "step": 1628
+    },
+    {
+      "epoch": 0.34403379091869063,
+      "grad_norm": 0.3620069921016693,
+      "learning_rate": 7.546811276676557e-05,
+      "loss": 0.2729,
+      "step": 1629
+    },
+    {
+      "epoch": 0.34424498416050686,
+      "grad_norm": 0.3549734055995941,
+      "learning_rate": 7.543894290076103e-05,
+      "loss": 0.2733,
+      "step": 1630
+    },
+    {
+      "epoch": 0.34445617740232315,
+      "grad_norm": 0.3700680136680603,
+      "learning_rate": 7.54097613478569e-05,
+      "loss": 0.2712,
+      "step": 1631
+    },
+    {
+      "epoch": 0.3446673706441394,
+      "grad_norm": 0.3460991680622101,
+      "learning_rate": 7.538056812145948e-05,
+      "loss": 0.2691,
+      "step": 1632
+    },
+    {
+      "epoch": 0.34487856388595567,
+      "grad_norm": 0.3311843276023865,
+      "learning_rate": 7.535136323498041e-05,
+      "loss": 0.2802,
+      "step": 1633
+    },
+    {
+      "epoch": 0.3450897571277719,
+      "grad_norm": 0.37731653451919556,
+      "learning_rate": 7.532214670183678e-05,
+      "loss": 0.2808,
+      "step": 1634
+    },
+    {
+      "epoch": 0.3453009503695882,
+      "grad_norm": 0.3085135817527771,
+      "learning_rate": 7.529291853545082e-05,
+      "loss": 0.2709,
+      "step": 1635
+    },
+    {
+      "epoch": 0.3455121436114044,
+      "grad_norm": 0.34321218729019165,
+      "learning_rate": 7.526367874925033e-05,
+      "loss": 0.2692,
+      "step": 1636
+    },
+    {
+      "epoch": 0.3457233368532207,
+      "grad_norm": 0.3788973391056061,
+      "learning_rate": 7.523442735666834e-05,
+      "loss": 0.2771,
+      "step": 1637
+    },
+    {
+      "epoch": 0.34593453009503694,
+      "grad_norm": 0.328721821308136,
+      "learning_rate": 7.52051643711432e-05,
+      "loss": 0.2634,
+      "step": 1638
+    },
+    {
+      "epoch": 0.34614572333685323,
+      "grad_norm": 0.3974335789680481,
+      "learning_rate": 7.517588980611864e-05,
+      "loss": 0.2683,
+      "step": 1639
+    },
+    {
+      "epoch": 0.34635691657866946,
+      "grad_norm": 0.3098681569099426,
+      "learning_rate": 7.514660367504367e-05,
+      "loss": 0.2708,
+      "step": 1640
+    },
+    {
+      "epoch": 0.34656810982048575,
+      "grad_norm": 0.41721415519714355,
+      "learning_rate": 7.511730599137261e-05,
+      "loss": 0.2678,
+      "step": 1641
+    },
+    {
+      "epoch": 0.346779303062302,
+      "grad_norm": 0.3062261939048767,
+      "learning_rate": 7.508799676856513e-05,
+      "loss": 0.2599,
+      "step": 1642
+    },
+    {
+      "epoch": 0.34699049630411827,
+      "grad_norm": 0.30826717615127563,
+      "learning_rate": 7.505867602008615e-05,
+      "loss": 0.2683,
+      "step": 1643
+    },
+    {
+      "epoch": 0.34720168954593456,
+      "grad_norm": 0.30457305908203125,
+      "learning_rate": 7.502934375940594e-05,
+      "loss": 0.2717,
+      "step": 1644
+    },
+    {
+      "epoch": 0.3474128827877508,
+      "grad_norm": 0.3540324866771698,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.275,
+      "step": 1645
+    },
+    {
+      "epoch": 0.3476240760295671,
+      "grad_norm": 0.3178246319293976,
+      "learning_rate": 7.497064475534916e-05,
+      "loss": 0.2729,
+      "step": 1646
+    },
+    {
+      "epoch": 0.3478352692713833,
+      "grad_norm": 0.45111629366874695,
+      "learning_rate": 7.494127803893947e-05,
+      "loss": 0.2684,
+      "step": 1647
+    },
+    {
+      "epoch": 0.3480464625131996,
+      "grad_norm": 0.2734365165233612,
+      "learning_rate": 7.491189986426236e-05,
+      "loss": 0.2619,
+      "step": 1648
+    },
+    {
+      "epoch": 0.3482576557550158,
+      "grad_norm": 0.3744702637195587,
+      "learning_rate": 7.488251024481437e-05,
+      "loss": 0.2701,
+      "step": 1649
+    },
+    {
+      "epoch": 0.3484688489968321,
+      "grad_norm": 0.32984721660614014,
+      "learning_rate": 7.485310919409742e-05,
+      "loss": 0.2653,
+      "step": 1650
+    },
+    {
+      "epoch": 0.34868004223864835,
+      "grad_norm": 0.4045005142688751,
+      "learning_rate": 7.482369672561864e-05,
+      "loss": 0.2676,
+      "step": 1651
+    },
+    {
+      "epoch": 0.34889123548046463,
+      "grad_norm": 0.5495765209197998,
+      "learning_rate": 7.47942728528904e-05,
+      "loss": 0.2706,
+      "step": 1652
+    },
+    {
+      "epoch": 0.34910242872228087,
+      "grad_norm": 0.2753506898880005,
+      "learning_rate": 7.476483758943031e-05,
+      "loss": 0.2706,
+      "step": 1653
+    },
+    {
+      "epoch": 0.34931362196409715,
+      "grad_norm": 0.34794318675994873,
+      "learning_rate": 7.473539094876125e-05,
+      "loss": 0.2722,
+      "step": 1654
+    },
+    {
+      "epoch": 0.3495248152059134,
+      "grad_norm": 0.364276260137558,
+      "learning_rate": 7.470593294441124e-05,
+      "loss": 0.2683,
+      "step": 1655
+    },
+    {
+      "epoch": 0.3497360084477297,
+      "grad_norm": 0.32945653796195984,
+      "learning_rate": 7.467646358991362e-05,
+      "loss": 0.2695,
+      "step": 1656
+    },
+    {
+      "epoch": 0.34994720168954596,
+      "grad_norm": 0.39120107889175415,
+      "learning_rate": 7.464698289880688e-05,
+      "loss": 0.2705,
+      "step": 1657
+    },
+    {
+      "epoch": 0.3501583949313622,
+      "grad_norm": 0.3489273488521576,
+      "learning_rate": 7.461749088463476e-05,
+      "loss": 0.2702,
+      "step": 1658
+    },
+    {
+      "epoch": 0.3503695881731785,
+      "grad_norm": 0.573573887348175,
+      "learning_rate": 7.458798756094614e-05,
+      "loss": 0.2688,
+      "step": 1659
+    },
+    {
+      "epoch": 0.3505807814149947,
+      "grad_norm": 0.3883051574230194,
+      "learning_rate": 7.455847294129519e-05,
+      "loss": 0.2668,
+      "step": 1660
+    },
+    {
+      "epoch": 0.350791974656811,
+      "grad_norm": 0.40855589509010315,
+      "learning_rate": 7.452894703924117e-05,
+      "loss": 0.2663,
+      "step": 1661
+    },
+    {
+      "epoch": 0.35100316789862723,
+      "grad_norm": 0.399848610162735,
+      "learning_rate": 7.449940986834858e-05,
+      "loss": 0.2698,
+      "step": 1662
+    },
+    {
+      "epoch": 0.3512143611404435,
+      "grad_norm": 0.3944951593875885,
+      "learning_rate": 7.446986144218711e-05,
+      "loss": 0.2652,
+      "step": 1663
+    },
+    {
+      "epoch": 0.35142555438225975,
+      "grad_norm": 0.36734044551849365,
+      "learning_rate": 7.444030177433156e-05,
+      "loss": 0.267,
+      "step": 1664
+    },
+    {
+      "epoch": 0.35163674762407604,
+      "grad_norm": 0.40042340755462646,
+      "learning_rate": 7.441073087836193e-05,
+      "loss": 0.2712,
+      "step": 1665
+    },
+    {
+      "epoch": 0.3518479408658923,
+      "grad_norm": 0.31229841709136963,
+      "learning_rate": 7.438114876786344e-05,
+      "loss": 0.2692,
+      "step": 1666
+    },
+    {
+      "epoch": 0.35205913410770856,
+      "grad_norm": 0.32294216752052307,
+      "learning_rate": 7.435155545642633e-05,
+      "loss": 0.27,
+      "step": 1667
+    },
+    {
+      "epoch": 0.3522703273495248,
+      "grad_norm": 0.36750587821006775,
+      "learning_rate": 7.432195095764609e-05,
+      "loss": 0.2697,
+      "step": 1668
+    },
+    {
+      "epoch": 0.3524815205913411,
+      "grad_norm": 0.39099735021591187,
+      "learning_rate": 7.429233528512331e-05,
+      "loss": 0.2687,
+      "step": 1669
+    },
+    {
+      "epoch": 0.3526927138331573,
+      "grad_norm": 0.31327545642852783,
+      "learning_rate": 7.426270845246373e-05,
+      "loss": 0.2651,
+      "step": 1670
+    },
+    {
+      "epoch": 0.3529039070749736,
+      "grad_norm": 0.33532506227493286,
+      "learning_rate": 7.423307047327819e-05,
+      "loss": 0.2693,
+      "step": 1671
+    },
+    {
+      "epoch": 0.3531151003167899,
+      "grad_norm": 0.3447170853614807,
+      "learning_rate": 7.42034213611827e-05,
+      "loss": 0.27,
+      "step": 1672
+    },
+    {
+      "epoch": 0.3533262935586061,
+      "grad_norm": 0.34815552830696106,
+      "learning_rate": 7.417376112979832e-05,
+      "loss": 0.268,
+      "step": 1673
+    },
+    {
+      "epoch": 0.3535374868004224,
+      "grad_norm": 0.33157265186309814,
+      "learning_rate": 7.414408979275125e-05,
+      "loss": 0.2675,
+      "step": 1674
+    },
+    {
+      "epoch": 0.35374868004223864,
+      "grad_norm": 0.4003755450248718,
+      "learning_rate": 7.411440736367281e-05,
+      "loss": 0.2671,
+      "step": 1675
+    },
+    {
+      "epoch": 0.3539598732840549,
+      "grad_norm": 0.7965269684791565,
+      "learning_rate": 7.408471385619939e-05,
+      "loss": 0.2703,
+      "step": 1676
+    },
+    {
+      "epoch": 0.35417106652587116,
+      "grad_norm": 0.34893205761909485,
+      "learning_rate": 7.405500928397248e-05,
+      "loss": 0.2699,
+      "step": 1677
+    },
+    {
+      "epoch": 0.35438225976768745,
+      "grad_norm": 0.36381006240844727,
+      "learning_rate": 7.402529366063868e-05,
+      "loss": 0.267,
+      "step": 1678
+    },
+    {
+      "epoch": 0.3545934530095037,
+      "grad_norm": 0.3976184129714966,
+      "learning_rate": 7.399556699984958e-05,
+      "loss": 0.2708,
+      "step": 1679
+    },
+    {
+      "epoch": 0.35480464625131997,
+      "grad_norm": 0.3170059025287628,
+      "learning_rate": 7.396582931526193e-05,
+      "loss": 0.2669,
+      "step": 1680
+    },
+    {
+      "epoch": 0.3550158394931362,
+      "grad_norm": 0.3825322389602661,
+      "learning_rate": 7.393608062053753e-05,
+      "loss": 0.2759,
+      "step": 1681
+    },
+    {
+      "epoch": 0.3552270327349525,
+      "grad_norm": 0.3521248698234558,
+      "learning_rate": 7.390632092934319e-05,
+      "loss": 0.271,
+      "step": 1682
+    },
+    {
+      "epoch": 0.3554382259767687,
+      "grad_norm": 0.3249512016773224,
+      "learning_rate": 7.387655025535083e-05,
+      "loss": 0.2693,
+      "step": 1683
+    },
+    {
+      "epoch": 0.355649419218585,
+      "grad_norm": 0.4294469952583313,
+      "learning_rate": 7.384676861223738e-05,
+      "loss": 0.2751,
+      "step": 1684
+    },
+    {
+      "epoch": 0.3558606124604013,
+      "grad_norm": 0.31159165501594543,
+      "learning_rate": 7.38169760136848e-05,
+      "loss": 0.2689,
+      "step": 1685
+    },
+    {
+      "epoch": 0.3560718057022175,
+      "grad_norm": 0.42009198665618896,
+      "learning_rate": 7.378717247338015e-05,
+      "loss": 0.27,
+      "step": 1686
+    },
+    {
+      "epoch": 0.3562829989440338,
+      "grad_norm": 0.3476303219795227,
+      "learning_rate": 7.375735800501542e-05,
+      "loss": 0.2685,
+      "step": 1687
+    },
+    {
+      "epoch": 0.35649419218585005,
+      "grad_norm": 0.416236937046051,
+      "learning_rate": 7.372753262228766e-05,
+      "loss": 0.2709,
+      "step": 1688
+    },
+    {
+      "epoch": 0.35670538542766633,
+      "grad_norm": 0.30849647521972656,
+      "learning_rate": 7.3697696338899e-05,
+      "loss": 0.2721,
+      "step": 1689
+    },
+    {
+      "epoch": 0.35691657866948256,
+      "grad_norm": 0.398502379655838,
+      "learning_rate": 7.36678491685565e-05,
+      "loss": 0.268,
+      "step": 1690
+    },
+    {
+      "epoch": 0.35712777191129885,
+      "grad_norm": 0.3252418339252472,
+      "learning_rate": 7.36379911249722e-05,
+      "loss": 0.2655,
+      "step": 1691
+    },
+    {
+      "epoch": 0.3573389651531151,
+      "grad_norm": 0.3882594108581543,
+      "learning_rate": 7.360812222186324e-05,
+      "loss": 0.2721,
+      "step": 1692
+    },
+    {
+      "epoch": 0.35755015839493137,
+      "grad_norm": 0.354158490896225,
+      "learning_rate": 7.357824247295164e-05,
+      "loss": 0.2687,
+      "step": 1693
+    },
+    {
+      "epoch": 0.3577613516367476,
+      "grad_norm": 0.34175145626068115,
+      "learning_rate": 7.354835189196447e-05,
+      "loss": 0.2672,
+      "step": 1694
+    },
+    {
+      "epoch": 0.3579725448785639,
+      "grad_norm": 0.4086613059043884,
+      "learning_rate": 7.351845049263374e-05,
+      "loss": 0.272,
+      "step": 1695
+    },
+    {
+      "epoch": 0.3581837381203801,
+      "grad_norm": 0.39072686433792114,
+      "learning_rate": 7.348853828869648e-05,
+      "loss": 0.2647,
+      "step": 1696
+    },
+    {
+      "epoch": 0.3583949313621964,
+      "grad_norm": 0.34353819489479065,
+      "learning_rate": 7.345861529389461e-05,
+      "loss": 0.2689,
+      "step": 1697
+    },
+    {
+      "epoch": 0.3586061246040127,
+      "grad_norm": 0.3195265233516693,
+      "learning_rate": 7.342868152197506e-05,
+      "loss": 0.2622,
+      "step": 1698
+    },
+    {
+      "epoch": 0.35881731784582893,
+      "grad_norm": 0.3065754175186157,
+      "learning_rate": 7.339873698668971e-05,
+      "loss": 0.2529,
+      "step": 1699
+    },
+    {
+      "epoch": 0.3590285110876452,
+      "grad_norm": 0.4163798391819,
+      "learning_rate": 7.336878170179537e-05,
+      "loss": 0.2625,
+      "step": 1700
+    },
+    {
+      "epoch": 0.35923970432946145,
+      "grad_norm": 0.3789835274219513,
+      "learning_rate": 7.333881568105378e-05,
+      "loss": 0.2617,
+      "step": 1701
+    },
+    {
+      "epoch": 0.35945089757127774,
+      "grad_norm": 0.3963404595851898,
+      "learning_rate": 7.330883893823164e-05,
+      "loss": 0.2636,
+      "step": 1702
+    },
+    {
+      "epoch": 0.35966209081309397,
+      "grad_norm": 0.47819721698760986,
+      "learning_rate": 7.327885148710053e-05,
+      "loss": 0.2662,
+      "step": 1703
+    },
+    {
+      "epoch": 0.35987328405491026,
+      "grad_norm": 0.37052810192108154,
+      "learning_rate": 7.324885334143702e-05,
+      "loss": 0.27,
+      "step": 1704
+    },
+    {
+      "epoch": 0.3600844772967265,
+      "grad_norm": 0.3508071303367615,
+      "learning_rate": 7.321884451502252e-05,
+      "loss": 0.2694,
+      "step": 1705
+    },
+    {
+      "epoch": 0.3602956705385428,
+      "grad_norm": 0.33410829305648804,
+      "learning_rate": 7.318882502164337e-05,
+      "loss": 0.2648,
+      "step": 1706
+    },
+    {
+      "epoch": 0.360506863780359,
+      "grad_norm": 0.5264217257499695,
+      "learning_rate": 7.315879487509086e-05,
+      "loss": 0.2691,
+      "step": 1707
+    },
+    {
+      "epoch": 0.3607180570221753,
+      "grad_norm": 0.3285062611103058,
+      "learning_rate": 7.31287540891611e-05,
+      "loss": 0.2681,
+      "step": 1708
+    },
+    {
+      "epoch": 0.36092925026399153,
+      "grad_norm": 0.5437626838684082,
+      "learning_rate": 7.309870267765515e-05,
+      "loss": 0.2735,
+      "step": 1709
+    },
+    {
+      "epoch": 0.3611404435058078,
+      "grad_norm": 0.49418190121650696,
+      "learning_rate": 7.30686406543789e-05,
+      "loss": 0.2772,
+      "step": 1710
+    },
+    {
+      "epoch": 0.36135163674762405,
+      "grad_norm": 0.5286495685577393,
+      "learning_rate": 7.303856803314313e-05,
+      "loss": 0.2767,
+      "step": 1711
+    },
+    {
+      "epoch": 0.36156282998944034,
+      "grad_norm": 0.299612820148468,
+      "learning_rate": 7.300848482776352e-05,
+      "loss": 0.2651,
+      "step": 1712
+    },
+    {
+      "epoch": 0.3617740232312566,
+      "grad_norm": 0.4181865453720093,
+      "learning_rate": 7.297839105206058e-05,
+      "loss": 0.2684,
+      "step": 1713
+    },
+    {
+      "epoch": 0.36198521647307286,
+      "grad_norm": 0.44695696234703064,
+      "learning_rate": 7.29482867198597e-05,
+      "loss": 0.2628,
+      "step": 1714
+    },
+    {
+      "epoch": 0.36219640971488914,
+      "grad_norm": 0.43489527702331543,
+      "learning_rate": 7.291817184499107e-05,
+      "loss": 0.2671,
+      "step": 1715
+    },
+    {
+      "epoch": 0.3624076029567054,
+      "grad_norm": 0.29662656784057617,
+      "learning_rate": 7.288804644128978e-05,
+      "loss": 0.2676,
+      "step": 1716
+    },
+    {
+      "epoch": 0.36261879619852166,
+      "grad_norm": 0.3823290765285492,
+      "learning_rate": 7.285791052259574e-05,
+      "loss": 0.2634,
+      "step": 1717
+    },
+    {
+      "epoch": 0.3628299894403379,
+      "grad_norm": 0.2718668580055237,
+      "learning_rate": 7.282776410275368e-05,
+      "loss": 0.258,
+      "step": 1718
+    },
+    {
+      "epoch": 0.3630411826821542,
+      "grad_norm": 0.28235042095184326,
+      "learning_rate": 7.279760719561315e-05,
+      "loss": 0.2613,
+      "step": 1719
+    },
+    {
+      "epoch": 0.3632523759239704,
+      "grad_norm": 0.47263166308403015,
+      "learning_rate": 7.276743981502856e-05,
+      "loss": 0.2611,
+      "step": 1720
+    },
+    {
+      "epoch": 0.3634635691657867,
+      "grad_norm": 0.3715069591999054,
+      "learning_rate": 7.273726197485906e-05,
+      "loss": 0.2645,
+      "step": 1721
+    },
+    {
+      "epoch": 0.36367476240760294,
+      "grad_norm": 0.505162239074707,
+      "learning_rate": 7.27070736889687e-05,
+      "loss": 0.2749,
+      "step": 1722
+    },
+    {
+      "epoch": 0.3638859556494192,
+      "grad_norm": 0.36894533038139343,
+      "learning_rate": 7.267687497122619e-05,
+      "loss": 0.2812,
+      "step": 1723
+    },
+    {
+      "epoch": 0.36409714889123546,
+      "grad_norm": 0.41288602352142334,
+      "learning_rate": 7.26466658355052e-05,
+      "loss": 0.2804,
+      "step": 1724
+    },
+    {
+      "epoch": 0.36430834213305174,
+      "grad_norm": 0.3856338560581207,
+      "learning_rate": 7.261644629568407e-05,
+      "loss": 0.2696,
+      "step": 1725
+    },
+    {
+      "epoch": 0.36451953537486803,
+      "grad_norm": 0.381818562746048,
+      "learning_rate": 7.258621636564593e-05,
+      "loss": 0.2747,
+      "step": 1726
+    },
+    {
+      "epoch": 0.36473072861668426,
+      "grad_norm": 0.3633262515068054,
+      "learning_rate": 7.255597605927876e-05,
+      "loss": 0.2676,
+      "step": 1727
+    },
+    {
+      "epoch": 0.36494192185850055,
+      "grad_norm": 0.3567476272583008,
+      "learning_rate": 7.252572539047521e-05,
+      "loss": 0.2663,
+      "step": 1728
+    },
+    {
+      "epoch": 0.3651531151003168,
+      "grad_norm": 0.37713220715522766,
+      "learning_rate": 7.249546437313272e-05,
+      "loss": 0.2801,
+      "step": 1729
+    },
+    {
+      "epoch": 0.36536430834213307,
+      "grad_norm": 0.3345044255256653,
+      "learning_rate": 7.246519302115355e-05,
+      "loss": 0.2716,
+      "step": 1730
+    },
+    {
+      "epoch": 0.3655755015839493,
+      "grad_norm": 0.32468757033348083,
+      "learning_rate": 7.243491134844462e-05,
+      "loss": 0.2678,
+      "step": 1731
+    },
+    {
+      "epoch": 0.3657866948257656,
+      "grad_norm": 0.35353878140449524,
+      "learning_rate": 7.240461936891762e-05,
+      "loss": 0.2716,
+      "step": 1732
+    },
+    {
+      "epoch": 0.3659978880675818,
+      "grad_norm": 0.37433922290802,
+      "learning_rate": 7.237431709648902e-05,
+      "loss": 0.2646,
+      "step": 1733
+    },
+    {
+      "epoch": 0.3662090813093981,
+      "grad_norm": 0.7967010140419006,
+      "learning_rate": 7.234400454507994e-05,
+      "loss": 0.2625,
+      "step": 1734
+    },
+    {
+      "epoch": 0.36642027455121434,
+      "grad_norm": 0.36097732186317444,
+      "learning_rate": 7.23136817286163e-05,
+      "loss": 0.2687,
+      "step": 1735
+    },
+    {
+      "epoch": 0.36663146779303063,
+      "grad_norm": 0.41895434260368347,
+      "learning_rate": 7.228334866102866e-05,
+      "loss": 0.2622,
+      "step": 1736
+    },
+    {
+      "epoch": 0.36684266103484686,
+      "grad_norm": 0.37427976727485657,
+      "learning_rate": 7.225300535625236e-05,
+      "loss": 0.2694,
+      "step": 1737
+    },
+    {
+      "epoch": 0.36705385427666315,
+      "grad_norm": 0.31183329224586487,
+      "learning_rate": 7.22226518282274e-05,
+      "loss": 0.267,
+      "step": 1738
+    },
+    {
+      "epoch": 0.3672650475184794,
+      "grad_norm": 0.37305107712745667,
+      "learning_rate": 7.219228809089848e-05,
+      "loss": 0.2774,
+      "step": 1739
+    },
+    {
+      "epoch": 0.36747624076029567,
+      "grad_norm": 0.3399893343448639,
+      "learning_rate": 7.216191415821503e-05,
+      "loss": 0.2666,
+      "step": 1740
+    },
+    {
+      "epoch": 0.36768743400211196,
+      "grad_norm": 0.3763524889945984,
+      "learning_rate": 7.213153004413108e-05,
+      "loss": 0.2557,
+      "step": 1741
+    },
+    {
+      "epoch": 0.3678986272439282,
+      "grad_norm": 0.39877429604530334,
+      "learning_rate": 7.210113576260544e-05,
+      "loss": 0.2689,
+      "step": 1742
+    },
+    {
+      "epoch": 0.3681098204857445,
+      "grad_norm": 0.39208346605300903,
+      "learning_rate": 7.207073132760151e-05,
+      "loss": 0.2624,
+      "step": 1743
+    },
+    {
+      "epoch": 0.3683210137275607,
+      "grad_norm": 0.33948197960853577,
+      "learning_rate": 7.204031675308738e-05,
+      "loss": 0.2669,
+      "step": 1744
+    },
+    {
+      "epoch": 0.368532206969377,
+      "grad_norm": 0.5894697904586792,
+      "learning_rate": 7.200989205303583e-05,
+      "loss": 0.2822,
+      "step": 1745
+    },
+    {
+      "epoch": 0.36874340021119323,
+      "grad_norm": 0.4160727262496948,
+      "learning_rate": 7.197945724142426e-05,
+      "loss": 0.2716,
+      "step": 1746
+    },
+    {
+      "epoch": 0.3689545934530095,
+      "grad_norm": 0.3931577205657959,
+      "learning_rate": 7.19490123322347e-05,
+      "loss": 0.2702,
+      "step": 1747
+    },
+    {
+      "epoch": 0.36916578669482575,
+      "grad_norm": 0.39650002121925354,
+      "learning_rate": 7.191855733945387e-05,
+      "loss": 0.2656,
+      "step": 1748
+    },
+    {
+      "epoch": 0.36937697993664204,
+      "grad_norm": 0.38313668966293335,
+      "learning_rate": 7.188809227707307e-05,
+      "loss": 0.2694,
+      "step": 1749
+    },
+    {
+      "epoch": 0.36958817317845827,
+      "grad_norm": 0.3257564902305603,
+      "learning_rate": 7.185761715908825e-05,
+      "loss": 0.2716,
+      "step": 1750
+    },
+    {
+      "epoch": 0.36979936642027456,
+      "grad_norm": 0.38745754957199097,
+      "learning_rate": 7.182713199949998e-05,
+      "loss": 0.2791,
+      "step": 1751
+    },
+    {
+      "epoch": 0.3700105596620908,
+      "grad_norm": 0.3291710317134857,
+      "learning_rate": 7.179663681231345e-05,
+      "loss": 0.2744,
+      "step": 1752
+    },
+    {
+      "epoch": 0.3702217529039071,
+      "grad_norm": 0.4040099084377289,
+      "learning_rate": 7.176613161153843e-05,
+      "loss": 0.2643,
+      "step": 1753
+    },
+    {
+      "epoch": 0.37043294614572336,
+      "grad_norm": 0.35342592000961304,
+      "learning_rate": 7.173561641118933e-05,
+      "loss": 0.266,
+      "step": 1754
+    },
+    {
+      "epoch": 0.3706441393875396,
+      "grad_norm": 0.37152549624443054,
+      "learning_rate": 7.170509122528512e-05,
+      "loss": 0.2646,
+      "step": 1755
+    },
+    {
+      "epoch": 0.3708553326293559,
+      "grad_norm": 0.3697604537010193,
+      "learning_rate": 7.167455606784935e-05,
+      "loss": 0.2637,
+      "step": 1756
+    },
+    {
+      "epoch": 0.3710665258711721,
+      "grad_norm": 0.31679266691207886,
+      "learning_rate": 7.16440109529102e-05,
+      "loss": 0.2593,
+      "step": 1757
+    },
+    {
+      "epoch": 0.3712777191129884,
+      "grad_norm": 0.33945295214653015,
+      "learning_rate": 7.161345589450038e-05,
+      "loss": 0.2689,
+      "step": 1758
+    },
+    {
+      "epoch": 0.37148891235480463,
+      "grad_norm": 0.2861180901527405,
+      "learning_rate": 7.15828909066572e-05,
+      "loss": 0.2628,
+      "step": 1759
+    },
+    {
+      "epoch": 0.3717001055966209,
+      "grad_norm": 0.361687034368515,
+      "learning_rate": 7.155231600342251e-05,
+      "loss": 0.2675,
+      "step": 1760
+    },
+    {
+      "epoch": 0.37191129883843715,
+      "grad_norm": 0.3180461525917053,
+      "learning_rate": 7.15217311988427e-05,
+      "loss": 0.2756,
+      "step": 1761
+    },
+    {
+      "epoch": 0.37212249208025344,
+      "grad_norm": 0.2994893789291382,
+      "learning_rate": 7.149113650696873e-05,
+      "loss": 0.2654,
+      "step": 1762
+    },
+    {
+      "epoch": 0.3723336853220697,
+      "grad_norm": 0.3863086402416229,
+      "learning_rate": 7.146053194185613e-05,
+      "loss": 0.2667,
+      "step": 1763
+    },
+    {
+      "epoch": 0.37254487856388596,
+      "grad_norm": 0.31023648381233215,
+      "learning_rate": 7.142991751756492e-05,
+      "loss": 0.2592,
+      "step": 1764
+    },
+    {
+      "epoch": 0.3727560718057022,
+      "grad_norm": 0.39709532260894775,
+      "learning_rate": 7.139929324815965e-05,
+      "loss": 0.2683,
+      "step": 1765
+    },
+    {
+      "epoch": 0.3729672650475185,
+      "grad_norm": 0.348423033952713,
+      "learning_rate": 7.136865914770944e-05,
+      "loss": 0.2653,
+      "step": 1766
+    },
+    {
+      "epoch": 0.37317845828933477,
+      "grad_norm": 0.2806280255317688,
+      "learning_rate": 7.133801523028788e-05,
+      "loss": 0.2614,
+      "step": 1767
+    },
+    {
+      "epoch": 0.373389651531151,
+      "grad_norm": 0.34935855865478516,
+      "learning_rate": 7.130736150997308e-05,
+      "loss": 0.2636,
+      "step": 1768
+    },
+    {
+      "epoch": 0.3736008447729673,
+      "grad_norm": 0.35370802879333496,
+      "learning_rate": 7.127669800084767e-05,
+      "loss": 0.2625,
+      "step": 1769
+    },
+    {
+      "epoch": 0.3738120380147835,
+      "grad_norm": 0.3103545308113098,
+      "learning_rate": 7.124602471699877e-05,
+      "loss": 0.2581,
+      "step": 1770
+    },
+    {
+      "epoch": 0.3740232312565998,
+      "grad_norm": 0.35079506039619446,
+      "learning_rate": 7.121534167251797e-05,
+      "loss": 0.2697,
+      "step": 1771
+    },
+    {
+      "epoch": 0.37423442449841604,
+      "grad_norm": 0.3645190894603729,
+      "learning_rate": 7.118464888150139e-05,
+      "loss": 0.2756,
+      "step": 1772
+    },
+    {
+      "epoch": 0.37444561774023233,
+      "grad_norm": 0.46524515748023987,
+      "learning_rate": 7.115394635804956e-05,
+      "loss": 0.2638,
+      "step": 1773
+    },
+    {
+      "epoch": 0.37465681098204856,
+      "grad_norm": 0.3275543451309204,
+      "learning_rate": 7.112323411626755e-05,
+      "loss": 0.2682,
+      "step": 1774
+    },
+    {
+      "epoch": 0.37486800422386485,
+      "grad_norm": 0.3308124840259552,
+      "learning_rate": 7.109251217026487e-05,
+      "loss": 0.2657,
+      "step": 1775
+    },
+    {
+      "epoch": 0.3750791974656811,
+      "grad_norm": 0.3791813850402832,
+      "learning_rate": 7.106178053415545e-05,
+      "loss": 0.2567,
+      "step": 1776
+    },
+    {
+      "epoch": 0.37529039070749737,
+      "grad_norm": 0.28272420167922974,
+      "learning_rate": 7.103103922205772e-05,
+      "loss": 0.2631,
+      "step": 1777
+    },
+    {
+      "epoch": 0.3755015839493136,
+      "grad_norm": 0.3779449760913849,
+      "learning_rate": 7.100028824809456e-05,
+      "loss": 0.2661,
+      "step": 1778
+    },
+    {
+      "epoch": 0.3757127771911299,
+      "grad_norm": 0.30117306113243103,
+      "learning_rate": 7.096952762639322e-05,
+      "loss": 0.2654,
+      "step": 1779
+    },
+    {
+      "epoch": 0.3759239704329461,
+      "grad_norm": 0.32284849882125854,
+      "learning_rate": 7.093875737108549e-05,
+      "loss": 0.2612,
+      "step": 1780
+    },
+    {
+      "epoch": 0.3761351636747624,
+      "grad_norm": 0.31875574588775635,
+      "learning_rate": 7.09079774963075e-05,
+      "loss": 0.2608,
+      "step": 1781
+    },
+    {
+      "epoch": 0.3763463569165787,
+      "grad_norm": 0.44315415620803833,
+      "learning_rate": 7.087718801619981e-05,
+      "loss": 0.2673,
+      "step": 1782
+    },
+    {
+      "epoch": 0.3765575501583949,
+      "grad_norm": 0.28723564743995667,
+      "learning_rate": 7.084638894490744e-05,
+      "loss": 0.2632,
+      "step": 1783
+    },
+    {
+      "epoch": 0.3767687434002112,
+      "grad_norm": 0.30582210421562195,
+      "learning_rate": 7.081558029657976e-05,
+      "loss": 0.2672,
+      "step": 1784
+    },
+    {
+      "epoch": 0.37697993664202745,
+      "grad_norm": 0.37577682733535767,
+      "learning_rate": 7.078476208537058e-05,
+      "loss": 0.2695,
+      "step": 1785
+    },
+    {
+      "epoch": 0.37719112988384373,
+      "grad_norm": 0.3432087004184723,
+      "learning_rate": 7.075393432543808e-05,
+      "loss": 0.2702,
+      "step": 1786
+    },
+    {
+      "epoch": 0.37740232312565997,
+      "grad_norm": 0.3155047297477722,
+      "learning_rate": 7.072309703094482e-05,
+      "loss": 0.2652,
+      "step": 1787
+    },
+    {
+      "epoch": 0.37761351636747625,
+      "grad_norm": 0.37758874893188477,
+      "learning_rate": 7.06922502160578e-05,
+      "loss": 0.2593,
+      "step": 1788
+    },
+    {
+      "epoch": 0.3778247096092925,
+      "grad_norm": 0.59708172082901,
+      "learning_rate": 7.066139389494831e-05,
+      "loss": 0.2663,
+      "step": 1789
+    },
+    {
+      "epoch": 0.3780359028511088,
+      "grad_norm": 0.40301111340522766,
+      "learning_rate": 7.063052808179205e-05,
+      "loss": 0.2677,
+      "step": 1790
+    },
+    {
+      "epoch": 0.378247096092925,
+      "grad_norm": 0.35209420323371887,
+      "learning_rate": 7.059965279076909e-05,
+      "loss": 0.2727,
+      "step": 1791
+    },
+    {
+      "epoch": 0.3784582893347413,
+      "grad_norm": 0.29887133836746216,
+      "learning_rate": 7.056876803606383e-05,
+      "loss": 0.262,
+      "step": 1792
+    },
+    {
+      "epoch": 0.3786694825765575,
+      "grad_norm": 0.3138899505138397,
+      "learning_rate": 7.053787383186505e-05,
+      "loss": 0.2624,
+      "step": 1793
+    },
+    {
+      "epoch": 0.3788806758183738,
+      "grad_norm": 0.44647619128227234,
+      "learning_rate": 7.050697019236581e-05,
+      "loss": 0.2643,
+      "step": 1794
+    },
+    {
+      "epoch": 0.3790918690601901,
+      "grad_norm": 0.32584792375564575,
+      "learning_rate": 7.04760571317636e-05,
+      "loss": 0.2589,
+      "step": 1795
+    },
+    {
+      "epoch": 0.37930306230200633,
+      "grad_norm": 0.33618640899658203,
+      "learning_rate": 7.044513466426012e-05,
+      "loss": 0.2687,
+      "step": 1796
+    },
+    {
+      "epoch": 0.3795142555438226,
+      "grad_norm": 0.31376752257347107,
+      "learning_rate": 7.04142028040615e-05,
+      "loss": 0.26,
+      "step": 1797
+    },
+    {
+      "epoch": 0.37972544878563885,
+      "grad_norm": 0.2943531572818756,
+      "learning_rate": 7.038326156537814e-05,
+      "loss": 0.2655,
+      "step": 1798
+    },
+    {
+      "epoch": 0.37993664202745514,
+      "grad_norm": 0.34809890389442444,
+      "learning_rate": 7.035231096242472e-05,
+      "loss": 0.2607,
+      "step": 1799
+    },
+    {
+      "epoch": 0.3801478352692714,
+      "grad_norm": 0.4191089868545532,
+      "learning_rate": 7.032135100942027e-05,
+      "loss": 0.2641,
+      "step": 1800
+    },
+    {
+      "epoch": 0.3801478352692714,
+      "eval_train_split_loss": 0.34744828939437866,
+      "eval_train_split_runtime": 6.5387,
+      "eval_train_split_samples_per_second": 1.529,
+      "eval_train_split_steps_per_second": 0.153,
+      "step": 1800
+    },
+    {
+      "epoch": 0.38035902851108766,
+      "grad_norm": 0.3785688877105713,
+      "learning_rate": 7.029038172058809e-05,
+      "loss": 0.2672,
+      "step": 1801
+    },
+    {
+      "epoch": 0.3805702217529039,
+      "grad_norm": 0.3919917345046997,
+      "learning_rate": 7.025940311015576e-05,
+      "loss": 0.2686,
+      "step": 1802
+    },
+    {
+      "epoch": 0.3807814149947202,
+      "grad_norm": 0.35662326216697693,
+      "learning_rate": 7.022841519235519e-05,
+      "loss": 0.2612,
+      "step": 1803
+    },
+    {
+      "epoch": 0.3809926082365364,
+      "grad_norm": 0.3849683105945587,
+      "learning_rate": 7.019741798142251e-05,
+      "loss": 0.2573,
+      "step": 1804
+    },
+    {
+      "epoch": 0.3812038014783527,
+      "grad_norm": 0.3740971088409424,
+      "learning_rate": 7.016641149159815e-05,
+      "loss": 0.264,
+      "step": 1805
+    },
+    {
+      "epoch": 0.38141499472016893,
+      "grad_norm": 0.33464938402175903,
+      "learning_rate": 7.013539573712679e-05,
+      "loss": 0.2599,
+      "step": 1806
+    },
+    {
+      "epoch": 0.3816261879619852,
+      "grad_norm": 0.2869853079319,
+      "learning_rate": 7.010437073225737e-05,
+      "loss": 0.2574,
+      "step": 1807
+    },
+    {
+      "epoch": 0.38183738120380145,
+      "grad_norm": 0.29229414463043213,
+      "learning_rate": 7.007333649124308e-05,
+      "loss": 0.2607,
+      "step": 1808
+    },
+    {
+      "epoch": 0.38204857444561774,
+      "grad_norm": 0.36321312189102173,
+      "learning_rate": 7.004229302834138e-05,
+      "loss": 0.2667,
+      "step": 1809
+    },
+    {
+      "epoch": 0.382259767687434,
+      "grad_norm": 0.2793180048465729,
+      "learning_rate": 7.00112403578139e-05,
+      "loss": 0.2674,
+      "step": 1810
+    },
+    {
+      "epoch": 0.38247096092925026,
+      "grad_norm": 0.29937034845352173,
+      "learning_rate": 6.998017849392657e-05,
+      "loss": 0.2613,
+      "step": 1811
+    },
+    {
+      "epoch": 0.38268215417106655,
+      "grad_norm": 0.613596498966217,
+      "learning_rate": 6.994910745094951e-05,
+      "loss": 0.2601,
+      "step": 1812
+    },
+    {
+      "epoch": 0.3828933474128828,
+      "grad_norm": 0.3137381374835968,
+      "learning_rate": 6.991802724315706e-05,
+      "loss": 0.2645,
+      "step": 1813
+    },
+    {
+      "epoch": 0.38310454065469907,
+      "grad_norm": 0.39771923422813416,
+      "learning_rate": 6.988693788482774e-05,
+      "loss": 0.2669,
+      "step": 1814
+    },
+    {
+      "epoch": 0.3833157338965153,
+      "grad_norm": 0.3305271565914154,
+      "learning_rate": 6.985583939024437e-05,
+      "loss": 0.2611,
+      "step": 1815
+    },
+    {
+      "epoch": 0.3835269271383316,
+      "grad_norm": 0.4109390079975128,
+      "learning_rate": 6.982473177369386e-05,
+      "loss": 0.2619,
+      "step": 1816
+    },
+    {
+      "epoch": 0.3837381203801478,
+      "grad_norm": 0.5110982656478882,
+      "learning_rate": 6.979361504946736e-05,
+      "loss": 0.2603,
+      "step": 1817
+    },
+    {
+      "epoch": 0.3839493136219641,
+      "grad_norm": 0.46137967705726624,
+      "learning_rate": 6.97624892318602e-05,
+      "loss": 0.264,
+      "step": 1818
+    },
+    {
+      "epoch": 0.38416050686378034,
+      "grad_norm": 0.359939306974411,
+      "learning_rate": 6.973135433517188e-05,
+      "loss": 0.2601,
+      "step": 1819
+    },
+    {
+      "epoch": 0.3843717001055966,
+      "grad_norm": 0.3532114028930664,
+      "learning_rate": 6.970021037370609e-05,
+      "loss": 0.2659,
+      "step": 1820
+    },
+    {
+      "epoch": 0.38458289334741286,
+      "grad_norm": 0.4738299250602722,
+      "learning_rate": 6.966905736177065e-05,
+      "loss": 0.2684,
+      "step": 1821
+    },
+    {
+      "epoch": 0.38479408658922915,
+      "grad_norm": 0.34422430396080017,
+      "learning_rate": 6.963789531367759e-05,
+      "loss": 0.2601,
+      "step": 1822
+    },
+    {
+      "epoch": 0.38500527983104543,
+      "grad_norm": 0.507690966129303,
+      "learning_rate": 6.960672424374303e-05,
+      "loss": 0.2619,
+      "step": 1823
+    },
+    {
+      "epoch": 0.38521647307286166,
+      "grad_norm": 0.36883941292762756,
+      "learning_rate": 6.957554416628729e-05,
+      "loss": 0.2627,
+      "step": 1824
+    },
+    {
+      "epoch": 0.38542766631467795,
+      "grad_norm": 0.41659849882125854,
+      "learning_rate": 6.954435509563478e-05,
+      "loss": 0.2614,
+      "step": 1825
+    },
+    {
+      "epoch": 0.3856388595564942,
+      "grad_norm": 0.40013259649276733,
+      "learning_rate": 6.951315704611408e-05,
+      "loss": 0.2703,
+      "step": 1826
+    },
+    {
+      "epoch": 0.38585005279831047,
+      "grad_norm": 0.44765275716781616,
+      "learning_rate": 6.948195003205786e-05,
+      "loss": 0.2687,
+      "step": 1827
+    },
+    {
+      "epoch": 0.3860612460401267,
+      "grad_norm": 0.35489124059677124,
+      "learning_rate": 6.945073406780296e-05,
+      "loss": 0.2644,
+      "step": 1828
+    },
+    {
+      "epoch": 0.386272439281943,
+      "grad_norm": 0.41339606046676636,
+      "learning_rate": 6.941950916769028e-05,
+      "loss": 0.2611,
+      "step": 1829
+    },
+    {
+      "epoch": 0.3864836325237592,
+      "grad_norm": 0.3601471185684204,
+      "learning_rate": 6.938827534606483e-05,
+      "loss": 0.2629,
+      "step": 1830
+    },
+    {
+      "epoch": 0.3866948257655755,
+      "grad_norm": 0.3806331157684326,
+      "learning_rate": 6.935703261727578e-05,
+      "loss": 0.2604,
+      "step": 1831
+    },
+    {
+      "epoch": 0.38690601900739174,
+      "grad_norm": 0.337954044342041,
+      "learning_rate": 6.932578099567628e-05,
+      "loss": 0.2567,
+      "step": 1832
+    },
+    {
+      "epoch": 0.38711721224920803,
+      "grad_norm": 0.41478219628334045,
+      "learning_rate": 6.929452049562371e-05,
+      "loss": 0.2637,
+      "step": 1833
+    },
+    {
+      "epoch": 0.38732840549102426,
+      "grad_norm": 0.3226884603500366,
+      "learning_rate": 6.926325113147937e-05,
+      "loss": 0.2606,
+      "step": 1834
+    },
+    {
+      "epoch": 0.38753959873284055,
+      "grad_norm": 0.4262640178203583,
+      "learning_rate": 6.923197291760876e-05,
+      "loss": 0.2693,
+      "step": 1835
+    },
+    {
+      "epoch": 0.3877507919746568,
+      "grad_norm": 0.36255720257759094,
+      "learning_rate": 6.92006858683814e-05,
+      "loss": 0.2631,
+      "step": 1836
+    },
+    {
+      "epoch": 0.38796198521647307,
+      "grad_norm": 0.4102550446987152,
+      "learning_rate": 6.916938999817084e-05,
+      "loss": 0.265,
+      "step": 1837
+    },
+    {
+      "epoch": 0.38817317845828936,
+      "grad_norm": 0.3359261751174927,
+      "learning_rate": 6.913808532135473e-05,
+      "loss": 0.2637,
+      "step": 1838
+    },
+    {
+      "epoch": 0.3883843717001056,
+      "grad_norm": 0.32953017950057983,
+      "learning_rate": 6.910677185231473e-05,
+      "loss": 0.2596,
+      "step": 1839
+    },
+    {
+      "epoch": 0.3885955649419219,
+      "grad_norm": 0.36355462670326233,
+      "learning_rate": 6.907544960543658e-05,
+      "loss": 0.2561,
+      "step": 1840
+    },
+    {
+      "epoch": 0.3888067581837381,
+      "grad_norm": 0.3528628647327423,
+      "learning_rate": 6.904411859511001e-05,
+      "loss": 0.2596,
+      "step": 1841
+    },
+    {
+      "epoch": 0.3890179514255544,
+      "grad_norm": 0.3596561849117279,
+      "learning_rate": 6.901277883572879e-05,
+      "loss": 0.2604,
+      "step": 1842
+    },
+    {
+      "epoch": 0.38922914466737063,
+      "grad_norm": 0.30548587441444397,
+      "learning_rate": 6.898143034169072e-05,
+      "loss": 0.2639,
+      "step": 1843
+    },
+    {
+      "epoch": 0.3894403379091869,
+      "grad_norm": 0.3815968930721283,
+      "learning_rate": 6.895007312739758e-05,
+      "loss": 0.2649,
+      "step": 1844
+    },
+    {
+      "epoch": 0.38965153115100315,
+      "grad_norm": 0.2936277389526367,
+      "learning_rate": 6.891870720725522e-05,
+      "loss": 0.2638,
+      "step": 1845
+    },
+    {
+      "epoch": 0.38986272439281944,
+      "grad_norm": 0.35309407114982605,
+      "learning_rate": 6.888733259567342e-05,
+      "loss": 0.263,
+      "step": 1846
+    },
+    {
+      "epoch": 0.39007391763463567,
+      "grad_norm": 0.3298608362674713,
+      "learning_rate": 6.885594930706601e-05,
+      "loss": 0.2648,
+      "step": 1847
+    },
+    {
+      "epoch": 0.39028511087645196,
+      "grad_norm": 0.34277698397636414,
+      "learning_rate": 6.882455735585077e-05,
+      "loss": 0.2596,
+      "step": 1848
+    },
+    {
+      "epoch": 0.3904963041182682,
+      "grad_norm": 0.33773091435432434,
+      "learning_rate": 6.879315675644946e-05,
+      "loss": 0.2575,
+      "step": 1849
+    },
+    {
+      "epoch": 0.3907074973600845,
+      "grad_norm": 0.4637722671031952,
+      "learning_rate": 6.876174752328785e-05,
+      "loss": 0.2692,
+      "step": 1850
+    },
+    {
+      "epoch": 0.39091869060190076,
+      "grad_norm": 0.4289977550506592,
+      "learning_rate": 6.873032967079561e-05,
+      "loss": 0.2651,
+      "step": 1851
+    },
+    {
+      "epoch": 0.391129883843717,
+      "grad_norm": 0.5879898071289062,
+      "learning_rate": 6.869890321340642e-05,
+      "loss": 0.263,
+      "step": 1852
+    },
+    {
+      "epoch": 0.3913410770855333,
+      "grad_norm": 0.31823113560676575,
+      "learning_rate": 6.866746816555794e-05,
+      "loss": 0.2626,
+      "step": 1853
+    },
+    {
+      "epoch": 0.3915522703273495,
+      "grad_norm": 0.40310221910476685,
+      "learning_rate": 6.86360245416917e-05,
+      "loss": 0.2672,
+      "step": 1854
+    },
+    {
+      "epoch": 0.3917634635691658,
+      "grad_norm": 0.30710679292678833,
+      "learning_rate": 6.860457235625322e-05,
+      "loss": 0.2651,
+      "step": 1855
+    },
+    {
+      "epoch": 0.39197465681098204,
+      "grad_norm": 0.3857133388519287,
+      "learning_rate": 6.857311162369193e-05,
+      "loss": 0.2619,
+      "step": 1856
+    },
+    {
+      "epoch": 0.3921858500527983,
+      "grad_norm": 0.3757339417934418,
+      "learning_rate": 6.854164235846123e-05,
+      "loss": 0.2578,
+      "step": 1857
+    },
+    {
+      "epoch": 0.39239704329461456,
+      "grad_norm": 0.3872772455215454,
+      "learning_rate": 6.851016457501836e-05,
+      "loss": 0.2572,
+      "step": 1858
+    },
+    {
+      "epoch": 0.39260823653643084,
+      "grad_norm": 0.3773196041584015,
+      "learning_rate": 6.847867828782458e-05,
+      "loss": 0.2596,
+      "step": 1859
+    },
+    {
+      "epoch": 0.3928194297782471,
+      "grad_norm": 0.32978305220603943,
+      "learning_rate": 6.844718351134496e-05,
+      "loss": 0.2627,
+      "step": 1860
+    },
+    {
+      "epoch": 0.39303062302006336,
+      "grad_norm": 0.4243892729282379,
+      "learning_rate": 6.841568026004852e-05,
+      "loss": 0.2569,
+      "step": 1861
+    },
+    {
+      "epoch": 0.3932418162618796,
+      "grad_norm": 0.35004329681396484,
+      "learning_rate": 6.838416854840816e-05,
+      "loss": 0.2617,
+      "step": 1862
+    },
+    {
+      "epoch": 0.3934530095036959,
+      "grad_norm": 0.31010955572128296,
+      "learning_rate": 6.835264839090068e-05,
+      "loss": 0.266,
+      "step": 1863
+    },
+    {
+      "epoch": 0.39366420274551217,
+      "grad_norm": 0.3802405595779419,
+      "learning_rate": 6.832111980200672e-05,
+      "loss": 0.2639,
+      "step": 1864
+    },
+    {
+      "epoch": 0.3938753959873284,
+      "grad_norm": 0.32143861055374146,
+      "learning_rate": 6.828958279621085e-05,
+      "loss": 0.2571,
+      "step": 1865
+    },
+    {
+      "epoch": 0.3940865892291447,
+      "grad_norm": 0.3755107820034027,
+      "learning_rate": 6.825803738800148e-05,
+      "loss": 0.2596,
+      "step": 1866
+    },
+    {
+      "epoch": 0.3942977824709609,
+      "grad_norm": 0.3363843560218811,
+      "learning_rate": 6.822648359187085e-05,
+      "loss": 0.2589,
+      "step": 1867
+    },
+    {
+      "epoch": 0.3945089757127772,
+      "grad_norm": 0.4123755991458893,
+      "learning_rate": 6.819492142231513e-05,
+      "loss": 0.2558,
+      "step": 1868
+    },
+    {
+      "epoch": 0.39472016895459344,
+      "grad_norm": 0.3600991666316986,
+      "learning_rate": 6.816335089383426e-05,
+      "loss": 0.2678,
+      "step": 1869
+    },
+    {
+      "epoch": 0.39493136219640973,
+      "grad_norm": 0.3160013258457184,
+      "learning_rate": 6.813177202093202e-05,
+      "loss": 0.261,
+      "step": 1870
+    },
+    {
+      "epoch": 0.39514255543822596,
+      "grad_norm": 0.30131232738494873,
+      "learning_rate": 6.810018481811614e-05,
+      "loss": 0.2607,
+      "step": 1871
+    },
+    {
+      "epoch": 0.39535374868004225,
+      "grad_norm": 0.34662073850631714,
+      "learning_rate": 6.8068589299898e-05,
+      "loss": 0.2625,
+      "step": 1872
+    },
+    {
+      "epoch": 0.3955649419218585,
+      "grad_norm": 0.28382354974746704,
+      "learning_rate": 6.803698548079294e-05,
+      "loss": 0.2619,
+      "step": 1873
+    },
+    {
+      "epoch": 0.39577613516367477,
+      "grad_norm": 0.3658379912376404,
+      "learning_rate": 6.800537337532005e-05,
+      "loss": 0.2665,
+      "step": 1874
+    },
+    {
+      "epoch": 0.395987328405491,
+      "grad_norm": 0.29414689540863037,
+      "learning_rate": 6.797375299800224e-05,
+      "loss": 0.2631,
+      "step": 1875
+    },
+    {
+      "epoch": 0.3961985216473073,
+      "grad_norm": 0.30097514390945435,
+      "learning_rate": 6.794212436336624e-05,
+      "loss": 0.2637,
+      "step": 1876
+    },
+    {
+      "epoch": 0.3964097148891235,
+      "grad_norm": 0.3043946921825409,
+      "learning_rate": 6.791048748594251e-05,
+      "loss": 0.2598,
+      "step": 1877
+    },
+    {
+      "epoch": 0.3966209081309398,
+      "grad_norm": 0.3490395247936249,
+      "learning_rate": 6.787884238026541e-05,
+      "loss": 0.2606,
+      "step": 1878
+    },
+    {
+      "epoch": 0.3968321013727561,
+      "grad_norm": 0.33023717999458313,
+      "learning_rate": 6.784718906087296e-05,
+      "loss": 0.2533,
+      "step": 1879
+    },
+    {
+      "epoch": 0.39704329461457233,
+      "grad_norm": 0.392117440700531,
+      "learning_rate": 6.7815527542307e-05,
+      "loss": 0.2588,
+      "step": 1880
+    },
+    {
+      "epoch": 0.3972544878563886,
+      "grad_norm": 0.333850234746933,
+      "learning_rate": 6.778385783911316e-05,
+      "loss": 0.2614,
+      "step": 1881
+    },
+    {
+      "epoch": 0.39746568109820485,
+      "grad_norm": 0.3338703215122223,
+      "learning_rate": 6.775217996584083e-05,
+      "loss": 0.2568,
+      "step": 1882
+    },
+    {
+      "epoch": 0.39767687434002114,
+      "grad_norm": 0.4263315200805664,
+      "learning_rate": 6.77204939370431e-05,
+      "loss": 0.2622,
+      "step": 1883
+    },
+    {
+      "epoch": 0.39788806758183737,
+      "grad_norm": 0.2845138907432556,
+      "learning_rate": 6.768879976727687e-05,
+      "loss": 0.2548,
+      "step": 1884
+    },
+    {
+      "epoch": 0.39809926082365366,
+      "grad_norm": 0.4653513729572296,
+      "learning_rate": 6.765709747110274e-05,
+      "loss": 0.2663,
+      "step": 1885
+    },
+    {
+      "epoch": 0.3983104540654699,
+      "grad_norm": 0.3557477295398712,
+      "learning_rate": 6.762538706308504e-05,
+      "loss": 0.261,
+      "step": 1886
+    },
+    {
+      "epoch": 0.3985216473072862,
+      "grad_norm": 0.42685747146606445,
+      "learning_rate": 6.759366855779187e-05,
+      "loss": 0.2656,
+      "step": 1887
+    },
+    {
+      "epoch": 0.3987328405491024,
+      "grad_norm": 0.2981302738189697,
+      "learning_rate": 6.756194196979501e-05,
+      "loss": 0.258,
+      "step": 1888
+    },
+    {
+      "epoch": 0.3989440337909187,
+      "grad_norm": 0.3768289089202881,
+      "learning_rate": 6.753020731366997e-05,
+      "loss": 0.2624,
+      "step": 1889
+    },
+    {
+      "epoch": 0.3991552270327349,
+      "grad_norm": 0.3319506347179413,
+      "learning_rate": 6.749846460399594e-05,
+      "loss": 0.2567,
+      "step": 1890
+    },
+    {
+      "epoch": 0.3993664202745512,
+      "grad_norm": 0.33027949929237366,
+      "learning_rate": 6.746671385535586e-05,
+      "loss": 0.2607,
+      "step": 1891
+    },
+    {
+      "epoch": 0.3995776135163675,
+      "grad_norm": 0.3836093842983246,
+      "learning_rate": 6.743495508233634e-05,
+      "loss": 0.2586,
+      "step": 1892
+    },
+    {
+      "epoch": 0.39978880675818373,
+      "grad_norm": 0.2776843309402466,
+      "learning_rate": 6.740318829952761e-05,
+      "loss": 0.2538,
+      "step": 1893
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3244691789150238,
+      "learning_rate": 6.737141352152373e-05,
+      "loss": 0.2555,
+      "step": 1894
+    },
+    {
+      "epoch": 0.40021119324181625,
+      "grad_norm": 0.27027779817581177,
+      "learning_rate": 6.73396307629223e-05,
+      "loss": 0.255,
+      "step": 1895
+    },
+    {
+      "epoch": 0.40042238648363254,
+      "grad_norm": 0.28744199872016907,
+      "learning_rate": 6.730784003832461e-05,
+      "loss": 0.2508,
+      "step": 1896
+    },
+    {
+      "epoch": 0.4006335797254488,
+      "grad_norm": 0.26639190316200256,
+      "learning_rate": 6.727604136233565e-05,
+      "loss": 0.2598,
+      "step": 1897
+    },
+    {
+      "epoch": 0.40084477296726506,
+      "grad_norm": 0.3411286771297455,
+      "learning_rate": 6.724423474956406e-05,
+      "loss": 0.2713,
+      "step": 1898
+    },
+    {
+      "epoch": 0.4010559662090813,
+      "grad_norm": 0.3124578297138214,
+      "learning_rate": 6.72124202146221e-05,
+      "loss": 0.2632,
+      "step": 1899
+    },
+    {
+      "epoch": 0.4012671594508976,
+      "grad_norm": 0.4226846396923065,
+      "learning_rate": 6.718059777212567e-05,
+      "loss": 0.2663,
+      "step": 1900
+    },
+    {
+      "epoch": 0.4014783526927138,
+      "grad_norm": 0.3878417909145355,
+      "learning_rate": 6.714876743669429e-05,
+      "loss": 0.2639,
+      "step": 1901
+    },
+    {
+      "epoch": 0.4016895459345301,
+      "grad_norm": 0.3209696412086487,
+      "learning_rate": 6.711692922295117e-05,
+      "loss": 0.2618,
+      "step": 1902
+    },
+    {
+      "epoch": 0.40190073917634633,
+      "grad_norm": 0.41655659675598145,
+      "learning_rate": 6.70850831455231e-05,
+      "loss": 0.261,
+      "step": 1903
+    },
+    {
+      "epoch": 0.4021119324181626,
+      "grad_norm": 0.34424808621406555,
+      "learning_rate": 6.705322921904043e-05,
+      "loss": 0.2618,
+      "step": 1904
+    },
+    {
+      "epoch": 0.40232312565997885,
+      "grad_norm": 0.3627392649650574,
+      "learning_rate": 6.702136745813721e-05,
+      "loss": 0.2647,
+      "step": 1905
+    },
+    {
+      "epoch": 0.40253431890179514,
+      "grad_norm": 0.4215112328529358,
+      "learning_rate": 6.698949787745105e-05,
+      "loss": 0.2614,
+      "step": 1906
+    },
+    {
+      "epoch": 0.40274551214361143,
+      "grad_norm": 0.3823428750038147,
+      "learning_rate": 6.69576204916231e-05,
+      "loss": 0.2559,
+      "step": 1907
+    },
+    {
+      "epoch": 0.40295670538542766,
+      "grad_norm": 0.3599855601787567,
+      "learning_rate": 6.692573531529818e-05,
+      "loss": 0.2612,
+      "step": 1908
+    },
+    {
+      "epoch": 0.40316789862724395,
+      "grad_norm": 2.212090253829956,
+      "learning_rate": 6.689384236312465e-05,
+      "loss": 0.2664,
+      "step": 1909
+    },
+    {
+      "epoch": 0.4033790918690602,
+      "grad_norm": 0.49254605174064636,
+      "learning_rate": 6.686194164975445e-05,
+      "loss": 0.2675,
+      "step": 1910
+    },
+    {
+      "epoch": 0.40359028511087647,
+      "grad_norm": 0.3318254053592682,
+      "learning_rate": 6.683003318984304e-05,
+      "loss": 0.2637,
+      "step": 1911
+    },
+    {
+      "epoch": 0.4038014783526927,
+      "grad_norm": 0.4154026508331299,
+      "learning_rate": 6.679811699804953e-05,
+      "loss": 0.2566,
+      "step": 1912
+    },
+    {
+      "epoch": 0.404012671594509,
+      "grad_norm": 0.38044416904449463,
+      "learning_rate": 6.676619308903652e-05,
+      "loss": 0.2568,
+      "step": 1913
+    },
+    {
+      "epoch": 0.4042238648363252,
+      "grad_norm": 0.3209648132324219,
+      "learning_rate": 6.673426147747014e-05,
+      "loss": 0.2596,
+      "step": 1914
+    },
+    {
+      "epoch": 0.4044350580781415,
+      "grad_norm": 0.30597540736198425,
+      "learning_rate": 6.67023221780201e-05,
+      "loss": 0.2554,
+      "step": 1915
+    },
+    {
+      "epoch": 0.40464625131995774,
+      "grad_norm": 0.31292611360549927,
+      "learning_rate": 6.667037520535963e-05,
+      "loss": 0.2582,
+      "step": 1916
+    },
+    {
+      "epoch": 0.404857444561774,
+      "grad_norm": 0.3791355788707733,
+      "learning_rate": 6.66384205741655e-05,
+      "loss": 0.2658,
+      "step": 1917
+    },
+    {
+      "epoch": 0.40506863780359026,
+      "grad_norm": 0.3750941753387451,
+      "learning_rate": 6.660645829911794e-05,
+      "loss": 0.2573,
+      "step": 1918
+    },
+    {
+      "epoch": 0.40527983104540655,
+      "grad_norm": 0.8172658085823059,
+      "learning_rate": 6.657448839490076e-05,
+      "loss": 0.2607,
+      "step": 1919
+    },
+    {
+      "epoch": 0.40549102428722283,
+      "grad_norm": 0.44030237197875977,
+      "learning_rate": 6.654251087620125e-05,
+      "loss": 0.2629,
+      "step": 1920
+    },
+    {
+      "epoch": 0.40570221752903907,
+      "grad_norm": 0.42775893211364746,
+      "learning_rate": 6.651052575771018e-05,
+      "loss": 0.2648,
+      "step": 1921
+    },
+    {
+      "epoch": 0.40591341077085535,
+      "grad_norm": 0.44088008999824524,
+      "learning_rate": 6.647853305412182e-05,
+      "loss": 0.2616,
+      "step": 1922
+    },
+    {
+      "epoch": 0.4061246040126716,
+      "grad_norm": 0.3638485372066498,
+      "learning_rate": 6.644653278013396e-05,
+      "loss": 0.2625,
+      "step": 1923
+    },
+    {
+      "epoch": 0.4063357972544879,
+      "grad_norm": 0.4363839328289032,
+      "learning_rate": 6.641452495044782e-05,
+      "loss": 0.2637,
+      "step": 1924
+    },
+    {
+      "epoch": 0.4065469904963041,
+      "grad_norm": 0.4155949056148529,
+      "learning_rate": 6.638250957976813e-05,
+      "loss": 0.2635,
+      "step": 1925
+    },
+    {
+      "epoch": 0.4067581837381204,
+      "grad_norm": 0.36409464478492737,
+      "learning_rate": 6.635048668280301e-05,
+      "loss": 0.2581,
+      "step": 1926
+    },
+    {
+      "epoch": 0.4069693769799366,
+      "grad_norm": 0.330169141292572,
+      "learning_rate": 6.631845627426418e-05,
+      "loss": 0.2605,
+      "step": 1927
+    },
+    {
+      "epoch": 0.4071805702217529,
+      "grad_norm": 0.32610511779785156,
+      "learning_rate": 6.628641836886663e-05,
+      "loss": 0.2586,
+      "step": 1928
+    },
+    {
+      "epoch": 0.40739176346356915,
+      "grad_norm": 0.3227826654911041,
+      "learning_rate": 6.625437298132896e-05,
+      "loss": 0.264,
+      "step": 1929
+    },
+    {
+      "epoch": 0.40760295670538543,
+      "grad_norm": 0.2977607548236847,
+      "learning_rate": 6.62223201263731e-05,
+      "loss": 0.2586,
+      "step": 1930
+    },
+    {
+      "epoch": 0.40781414994720167,
+      "grad_norm": 0.35836920142173767,
+      "learning_rate": 6.619025981872447e-05,
+      "loss": 0.2533,
+      "step": 1931
+    },
+    {
+      "epoch": 0.40802534318901795,
+      "grad_norm": 0.37403273582458496,
+      "learning_rate": 6.615819207311185e-05,
+      "loss": 0.2574,
+      "step": 1932
+    },
+    {
+      "epoch": 0.40823653643083424,
+      "grad_norm": 0.4140656590461731,
+      "learning_rate": 6.612611690426751e-05,
+      "loss": 0.2574,
+      "step": 1933
+    },
+    {
+      "epoch": 0.4084477296726505,
+      "grad_norm": 0.3286362588405609,
+      "learning_rate": 6.609403432692709e-05,
+      "loss": 0.2544,
+      "step": 1934
+    },
+    {
+      "epoch": 0.40865892291446676,
+      "grad_norm": 0.3253157138824463,
+      "learning_rate": 6.606194435582964e-05,
+      "loss": 0.2579,
+      "step": 1935
+    },
+    {
+      "epoch": 0.408870116156283,
+      "grad_norm": 0.29681751132011414,
+      "learning_rate": 6.602984700571759e-05,
+      "loss": 0.254,
+      "step": 1936
+    },
+    {
+      "epoch": 0.4090813093980993,
+      "grad_norm": 0.30146291851997375,
+      "learning_rate": 6.59977422913368e-05,
+      "loss": 0.2552,
+      "step": 1937
+    },
+    {
+      "epoch": 0.4092925026399155,
+      "grad_norm": 0.3405838906764984,
+      "learning_rate": 6.596563022743646e-05,
+      "loss": 0.2582,
+      "step": 1938
+    },
+    {
+      "epoch": 0.4095036958817318,
+      "grad_norm": 0.3043459355831146,
+      "learning_rate": 6.593351082876921e-05,
+      "loss": 0.2573,
+      "step": 1939
+    },
+    {
+      "epoch": 0.40971488912354803,
+      "grad_norm": 0.28687623143196106,
+      "learning_rate": 6.590138411009099e-05,
+      "loss": 0.2499,
+      "step": 1940
+    },
+    {
+      "epoch": 0.4099260823653643,
+      "grad_norm": 0.4864982068538666,
+      "learning_rate": 6.586925008616113e-05,
+      "loss": 0.2586,
+      "step": 1941
+    },
+    {
+      "epoch": 0.41013727560718055,
+      "grad_norm": 0.32113856077194214,
+      "learning_rate": 6.583710877174233e-05,
+      "loss": 0.2606,
+      "step": 1942
+    },
+    {
+      "epoch": 0.41034846884899684,
+      "grad_norm": 0.40114903450012207,
+      "learning_rate": 6.580496018160059e-05,
+      "loss": 0.2678,
+      "step": 1943
+    },
+    {
+      "epoch": 0.41055966209081307,
+      "grad_norm": 0.2755469083786011,
+      "learning_rate": 6.577280433050535e-05,
+      "loss": 0.2621,
+      "step": 1944
+    },
+    {
+      "epoch": 0.41077085533262936,
+      "grad_norm": 0.32695573568344116,
+      "learning_rate": 6.574064123322925e-05,
+      "loss": 0.2624,
+      "step": 1945
+    },
+    {
+      "epoch": 0.4109820485744456,
+      "grad_norm": 0.30938369035720825,
+      "learning_rate": 6.570847090454838e-05,
+      "loss": 0.2618,
+      "step": 1946
+    },
+    {
+      "epoch": 0.4111932418162619,
+      "grad_norm": 0.3177734315395355,
+      "learning_rate": 6.567629335924208e-05,
+      "loss": 0.2555,
+      "step": 1947
+    },
+    {
+      "epoch": 0.41140443505807817,
+      "grad_norm": 0.3019404709339142,
+      "learning_rate": 6.564410861209306e-05,
+      "loss": 0.2548,
+      "step": 1948
+    },
+    {
+      "epoch": 0.4116156282998944,
+      "grad_norm": 0.3436680734157562,
+      "learning_rate": 6.561191667788726e-05,
+      "loss": 0.2556,
+      "step": 1949
+    },
+    {
+      "epoch": 0.4118268215417107,
+      "grad_norm": 0.3854840397834778,
+      "learning_rate": 6.557971757141401e-05,
+      "loss": 0.2605,
+      "step": 1950
+    },
+    {
+      "epoch": 0.4120380147835269,
+      "grad_norm": 0.36365339159965515,
+      "learning_rate": 6.55475113074659e-05,
+      "loss": 0.2599,
+      "step": 1951
+    },
+    {
+      "epoch": 0.4122492080253432,
+      "grad_norm": 0.4324885904788971,
+      "learning_rate": 6.551529790083876e-05,
+      "loss": 0.2561,
+      "step": 1952
+    },
+    {
+      "epoch": 0.41246040126715944,
+      "grad_norm": 0.3663138449192047,
+      "learning_rate": 6.548307736633178e-05,
+      "loss": 0.2584,
+      "step": 1953
+    },
+    {
+      "epoch": 0.4126715945089757,
+      "grad_norm": 0.3217354416847229,
+      "learning_rate": 6.545084971874738e-05,
+      "loss": 0.2565,
+      "step": 1954
+    },
+    {
+      "epoch": 0.41288278775079196,
+      "grad_norm": 0.5375369787216187,
+      "learning_rate": 6.541861497289126e-05,
+      "loss": 0.2551,
+      "step": 1955
+    },
+    {
+      "epoch": 0.41309398099260825,
+      "grad_norm": 0.5009047389030457,
+      "learning_rate": 6.538637314357237e-05,
+      "loss": 0.2623,
+      "step": 1956
+    },
+    {
+      "epoch": 0.4133051742344245,
+      "grad_norm": 0.4821934103965759,
+      "learning_rate": 6.535412424560293e-05,
+      "loss": 0.2684,
+      "step": 1957
+    },
+    {
+      "epoch": 0.41351636747624076,
+      "grad_norm": 0.31959158182144165,
+      "learning_rate": 6.532186829379839e-05,
+      "loss": 0.2622,
+      "step": 1958
+    },
+    {
+      "epoch": 0.413727560718057,
+      "grad_norm": 0.3432377576828003,
+      "learning_rate": 6.528960530297745e-05,
+      "loss": 0.26,
+      "step": 1959
+    },
+    {
+      "epoch": 0.4139387539598733,
+      "grad_norm": 0.37346595525741577,
+      "learning_rate": 6.525733528796207e-05,
+      "loss": 0.2569,
+      "step": 1960
+    },
+    {
+      "epoch": 0.41414994720168957,
+      "grad_norm": 0.3408629596233368,
+      "learning_rate": 6.522505826357737e-05,
+      "loss": 0.2554,
+      "step": 1961
+    },
+    {
+      "epoch": 0.4143611404435058,
+      "grad_norm": 0.2720877230167389,
+      "learning_rate": 6.519277424465177e-05,
+      "loss": 0.2521,
+      "step": 1962
+    },
+    {
+      "epoch": 0.4145723336853221,
+      "grad_norm": 0.3213403522968292,
+      "learning_rate": 6.516048324601684e-05,
+      "loss": 0.254,
+      "step": 1963
+    },
+    {
+      "epoch": 0.4147835269271383,
+      "grad_norm": 0.30583125352859497,
+      "learning_rate": 6.51281852825074e-05,
+      "loss": 0.2553,
+      "step": 1964
+    },
+    {
+      "epoch": 0.4149947201689546,
+      "grad_norm": 0.35771241784095764,
+      "learning_rate": 6.509588036896143e-05,
+      "loss": 0.2582,
+      "step": 1965
+    },
+    {
+      "epoch": 0.41520591341077084,
+      "grad_norm": 0.361017107963562,
+      "learning_rate": 6.506356852022013e-05,
+      "loss": 0.2617,
+      "step": 1966
+    },
+    {
+      "epoch": 0.41541710665258713,
+      "grad_norm": 0.28236308693885803,
+      "learning_rate": 6.503124975112788e-05,
+      "loss": 0.2578,
+      "step": 1967
+    },
+    {
+      "epoch": 0.41562829989440336,
+      "grad_norm": 0.31907933950424194,
+      "learning_rate": 6.499892407653227e-05,
+      "loss": 0.2667,
+      "step": 1968
+    },
+    {
+      "epoch": 0.41583949313621965,
+      "grad_norm": 0.47929680347442627,
+      "learning_rate": 6.496659151128397e-05,
+      "loss": 0.2572,
+      "step": 1969
+    },
+    {
+      "epoch": 0.4160506863780359,
+      "grad_norm": 0.2947295606136322,
+      "learning_rate": 6.493425207023693e-05,
+      "loss": 0.2519,
+      "step": 1970
+    },
+    {
+      "epoch": 0.41626187961985217,
+      "grad_norm": 0.34185075759887695,
+      "learning_rate": 6.490190576824819e-05,
+      "loss": 0.2581,
+      "step": 1971
+    },
+    {
+      "epoch": 0.4164730728616684,
+      "grad_norm": 0.2819492220878601,
+      "learning_rate": 6.486955262017795e-05,
+      "loss": 0.2589,
+      "step": 1972
+    },
+    {
+      "epoch": 0.4166842661034847,
+      "grad_norm": 0.2985383868217468,
+      "learning_rate": 6.483719264088956e-05,
+      "loss": 0.2547,
+      "step": 1973
+    },
+    {
+      "epoch": 0.4168954593453009,
+      "grad_norm": 0.32047462463378906,
+      "learning_rate": 6.480482584524951e-05,
+      "loss": 0.2551,
+      "step": 1974
+    },
+    {
+      "epoch": 0.4171066525871172,
+      "grad_norm": 0.3490952253341675,
+      "learning_rate": 6.477245224812745e-05,
+      "loss": 0.258,
+      "step": 1975
+    },
+    {
+      "epoch": 0.4173178458289335,
+      "grad_norm": 0.39916518330574036,
+      "learning_rate": 6.474007186439613e-05,
+      "loss": 0.2554,
+      "step": 1976
+    },
+    {
+      "epoch": 0.41752903907074973,
+      "grad_norm": 0.36380985379219055,
+      "learning_rate": 6.470768470893136e-05,
+      "loss": 0.2625,
+      "step": 1977
+    },
+    {
+      "epoch": 0.417740232312566,
+      "grad_norm": 0.28608936071395874,
+      "learning_rate": 6.467529079661216e-05,
+      "loss": 0.2553,
+      "step": 1978
+    },
+    {
+      "epoch": 0.41795142555438225,
+      "grad_norm": 0.35926151275634766,
+      "learning_rate": 6.464289014232062e-05,
+      "loss": 0.2545,
+      "step": 1979
+    },
+    {
+      "epoch": 0.41816261879619854,
+      "grad_norm": 0.2857661843299866,
+      "learning_rate": 6.461048276094189e-05,
+      "loss": 0.255,
+      "step": 1980
+    },
+    {
+      "epoch": 0.41837381203801477,
+      "grad_norm": 0.3653435707092285,
+      "learning_rate": 6.457806866736424e-05,
+      "loss": 0.254,
+      "step": 1981
+    },
+    {
+      "epoch": 0.41858500527983106,
+      "grad_norm": 0.30634230375289917,
+      "learning_rate": 6.454564787647904e-05,
+      "loss": 0.2561,
+      "step": 1982
+    },
+    {
+      "epoch": 0.4187961985216473,
+      "grad_norm": 0.30079784989356995,
+      "learning_rate": 6.451322040318073e-05,
+      "loss": 0.2593,
+      "step": 1983
+    },
+    {
+      "epoch": 0.4190073917634636,
+      "grad_norm": 0.355526864528656,
+      "learning_rate": 6.448078626236676e-05,
+      "loss": 0.2509,
+      "step": 1984
+    },
+    {
+      "epoch": 0.4192185850052798,
+      "grad_norm": 0.3241296112537384,
+      "learning_rate": 6.444834546893774e-05,
+      "loss": 0.2495,
+      "step": 1985
+    },
+    {
+      "epoch": 0.4194297782470961,
+      "grad_norm": 0.31503885984420776,
+      "learning_rate": 6.441589803779727e-05,
+      "loss": 0.2592,
+      "step": 1986
+    },
+    {
+      "epoch": 0.41964097148891233,
+      "grad_norm": 0.3563302159309387,
+      "learning_rate": 6.438344398385199e-05,
+      "loss": 0.2554,
+      "step": 1987
+    },
+    {
+      "epoch": 0.4198521647307286,
+      "grad_norm": 0.310743123292923,
+      "learning_rate": 6.435098332201166e-05,
+      "loss": 0.2573,
+      "step": 1988
+    },
+    {
+      "epoch": 0.4200633579725449,
+      "grad_norm": 0.3609410226345062,
+      "learning_rate": 6.4318516067189e-05,
+      "loss": 0.262,
+      "step": 1989
+    },
+    {
+      "epoch": 0.42027455121436114,
+      "grad_norm": 0.34922000765800476,
+      "learning_rate": 6.42860422342998e-05,
+      "loss": 0.2612,
+      "step": 1990
+    },
+    {
+      "epoch": 0.4204857444561774,
+      "grad_norm": 0.3274752199649811,
+      "learning_rate": 6.425356183826283e-05,
+      "loss": 0.2626,
+      "step": 1991
+    },
+    {
+      "epoch": 0.42069693769799366,
+      "grad_norm": 0.47317004203796387,
+      "learning_rate": 6.422107489399993e-05,
+      "loss": 0.2616,
+      "step": 1992
+    },
+    {
+      "epoch": 0.42090813093980994,
+      "grad_norm": 0.4474397897720337,
+      "learning_rate": 6.41885814164359e-05,
+      "loss": 0.2601,
+      "step": 1993
+    },
+    {
+      "epoch": 0.4211193241816262,
+      "grad_norm": 0.3935490846633911,
+      "learning_rate": 6.415608142049859e-05,
+      "loss": 0.2578,
+      "step": 1994
+    },
+    {
+      "epoch": 0.42133051742344246,
+      "grad_norm": 0.42940765619277954,
+      "learning_rate": 6.412357492111876e-05,
+      "loss": 0.2526,
+      "step": 1995
+    },
+    {
+      "epoch": 0.4215417106652587,
+      "grad_norm": 0.32802504301071167,
+      "learning_rate": 6.409106193323027e-05,
+      "loss": 0.2539,
+      "step": 1996
+    },
+    {
+      "epoch": 0.421752903907075,
+      "grad_norm": 0.3647957742214203,
+      "learning_rate": 6.40585424717699e-05,
+      "loss": 0.2556,
+      "step": 1997
+    },
+    {
+      "epoch": 0.4219640971488912,
+      "grad_norm": 0.28577372431755066,
+      "learning_rate": 6.402601655167739e-05,
+      "loss": 0.2521,
+      "step": 1998
+    },
+    {
+      "epoch": 0.4221752903907075,
+      "grad_norm": 0.35097217559814453,
+      "learning_rate": 6.399348418789545e-05,
+      "loss": 0.2514,
+      "step": 1999
+    },
+    {
+      "epoch": 0.42238648363252373,
+      "grad_norm": 0.34535184502601624,
+      "learning_rate": 6.396094539536981e-05,
+      "loss": 0.2573,
+      "step": 2000
+    },
+    {
+      "epoch": 0.42238648363252373,
+      "eval_train_split_loss": 0.3918624520301819,
+      "eval_train_split_runtime": 6.4803,
+      "eval_train_split_samples_per_second": 1.543,
+      "eval_train_split_steps_per_second": 0.154,
+      "step": 2000
+    },
+    {
+      "epoch": 0.42259767687434,
+      "grad_norm": 0.3618689477443695,
+      "learning_rate": 6.392840018904908e-05,
+      "loss": 0.2614,
+      "step": 2001
+    },
+    {
+      "epoch": 0.4228088701161563,
+      "grad_norm": 0.3648746907711029,
+      "learning_rate": 6.389584858388484e-05,
+      "loss": 0.2575,
+      "step": 2002
+    },
+    {
+      "epoch": 0.42302006335797254,
+      "grad_norm": 0.3827821612358093,
+      "learning_rate": 6.386329059483165e-05,
+      "loss": 0.2592,
+      "step": 2003
+    },
+    {
+      "epoch": 0.42323125659978883,
+      "grad_norm": 0.30110910534858704,
+      "learning_rate": 6.383072623684696e-05,
+      "loss": 0.2573,
+      "step": 2004
+    },
+    {
+      "epoch": 0.42344244984160506,
+      "grad_norm": 0.3633120656013489,
+      "learning_rate": 6.379815552489112e-05,
+      "loss": 0.2588,
+      "step": 2005
+    },
+    {
+      "epoch": 0.42365364308342135,
+      "grad_norm": 0.32551953196525574,
+      "learning_rate": 6.376557847392747e-05,
+      "loss": 0.2574,
+      "step": 2006
+    },
+    {
+      "epoch": 0.4238648363252376,
+      "grad_norm": 0.3477630615234375,
+      "learning_rate": 6.373299509892223e-05,
+      "loss": 0.2526,
+      "step": 2007
+    },
+    {
+      "epoch": 0.42407602956705387,
+      "grad_norm": 0.31732550263404846,
+      "learning_rate": 6.37004054148445e-05,
+      "loss": 0.2508,
+      "step": 2008
+    },
+    {
+      "epoch": 0.4242872228088701,
+      "grad_norm": 0.3623895049095154,
+      "learning_rate": 6.36678094366663e-05,
+      "loss": 0.2522,
+      "step": 2009
+    },
+    {
+      "epoch": 0.4244984160506864,
+      "grad_norm": 0.3724652826786041,
+      "learning_rate": 6.363520717936256e-05,
+      "loss": 0.2587,
+      "step": 2010
+    },
+    {
+      "epoch": 0.4247096092925026,
+      "grad_norm": 0.3066050708293915,
+      "learning_rate": 6.360259865791106e-05,
+      "loss": 0.2588,
+      "step": 2011
+    },
+    {
+      "epoch": 0.4249208025343189,
+      "grad_norm": 0.3705737292766571,
+      "learning_rate": 6.356998388729247e-05,
+      "loss": 0.256,
+      "step": 2012
+    },
+    {
+      "epoch": 0.42513199577613514,
+      "grad_norm": 0.3478221297264099,
+      "learning_rate": 6.353736288249037e-05,
+      "loss": 0.2606,
+      "step": 2013
+    },
+    {
+      "epoch": 0.42534318901795143,
+      "grad_norm": 0.407416433095932,
+      "learning_rate": 6.350473565849112e-05,
+      "loss": 0.2642,
+      "step": 2014
+    },
+    {
+      "epoch": 0.42555438225976766,
+      "grad_norm": 0.37096425890922546,
+      "learning_rate": 6.347210223028403e-05,
+      "loss": 0.2563,
+      "step": 2015
+    },
+    {
+      "epoch": 0.42576557550158395,
+      "grad_norm": 0.4088791310787201,
+      "learning_rate": 6.343946261286117e-05,
+      "loss": 0.2543,
+      "step": 2016
+    },
+    {
+      "epoch": 0.42597676874340024,
+      "grad_norm": 0.3512021601200104,
+      "learning_rate": 6.340681682121755e-05,
+      "loss": 0.2572,
+      "step": 2017
+    },
+    {
+      "epoch": 0.42618796198521647,
+      "grad_norm": 0.331222265958786,
+      "learning_rate": 6.337416487035095e-05,
+      "loss": 0.2564,
+      "step": 2018
+    },
+    {
+      "epoch": 0.42639915522703276,
+      "grad_norm": 0.30013251304626465,
+      "learning_rate": 6.334150677526196e-05,
+      "loss": 0.2623,
+      "step": 2019
+    },
+    {
+      "epoch": 0.426610348468849,
+      "grad_norm": 0.30134907364845276,
+      "learning_rate": 6.330884255095408e-05,
+      "loss": 0.2583,
+      "step": 2020
+    },
+    {
+      "epoch": 0.4268215417106653,
+      "grad_norm": 0.29840895533561707,
+      "learning_rate": 6.327617221243356e-05,
+      "loss": 0.2556,
+      "step": 2021
+    },
+    {
+      "epoch": 0.4270327349524815,
+      "grad_norm": 0.3379371464252472,
+      "learning_rate": 6.324349577470944e-05,
+      "loss": 0.2529,
+      "step": 2022
+    },
+    {
+      "epoch": 0.4272439281942978,
+      "grad_norm": 0.319657564163208,
+      "learning_rate": 6.321081325279365e-05,
+      "loss": 0.2452,
+      "step": 2023
+    },
+    {
+      "epoch": 0.427455121436114,
+      "grad_norm": 0.3542124629020691,
+      "learning_rate": 6.317812466170083e-05,
+      "loss": 0.2482,
+      "step": 2024
+    },
+    {
+      "epoch": 0.4276663146779303,
+      "grad_norm": 0.33450204133987427,
+      "learning_rate": 6.314543001644844e-05,
+      "loss": 0.2535,
+      "step": 2025
+    },
+    {
+      "epoch": 0.42787750791974655,
+      "grad_norm": 0.31855499744415283,
+      "learning_rate": 6.311272933205672e-05,
+      "loss": 0.258,
+      "step": 2026
+    },
+    {
+      "epoch": 0.42808870116156283,
+      "grad_norm": 0.3302219808101654,
+      "learning_rate": 6.308002262354869e-05,
+      "loss": 0.2629,
+      "step": 2027
+    },
+    {
+      "epoch": 0.42829989440337907,
+      "grad_norm": 0.28444012999534607,
+      "learning_rate": 6.304730990595011e-05,
+      "loss": 0.2525,
+      "step": 2028
+    },
+    {
+      "epoch": 0.42851108764519535,
+      "grad_norm": 0.3347005546092987,
+      "learning_rate": 6.301459119428956e-05,
+      "loss": 0.2581,
+      "step": 2029
+    },
+    {
+      "epoch": 0.42872228088701164,
+      "grad_norm": 0.2904024124145508,
+      "learning_rate": 6.298186650359832e-05,
+      "loss": 0.2555,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4289334741288279,
+      "grad_norm": 0.33425018191337585,
+      "learning_rate": 6.294913584891043e-05,
+      "loss": 0.2499,
+      "step": 2031
+    },
+    {
+      "epoch": 0.42914466737064416,
+      "grad_norm": 0.27351266145706177,
+      "learning_rate": 6.291639924526268e-05,
+      "loss": 0.255,
+      "step": 2032
+    },
+    {
+      "epoch": 0.4293558606124604,
+      "grad_norm": 0.27451226115226746,
+      "learning_rate": 6.288365670769459e-05,
+      "loss": 0.2567,
+      "step": 2033
+    },
+    {
+      "epoch": 0.4295670538542767,
+      "grad_norm": 0.3020147383213043,
+      "learning_rate": 6.285090825124835e-05,
+      "loss": 0.2593,
+      "step": 2034
+    },
+    {
+      "epoch": 0.4297782470960929,
+      "grad_norm": 0.28530821204185486,
+      "learning_rate": 6.281815389096903e-05,
+      "loss": 0.2527,
+      "step": 2035
+    },
+    {
+      "epoch": 0.4299894403379092,
+      "grad_norm": 0.25803953409194946,
+      "learning_rate": 6.27853936419042e-05,
+      "loss": 0.2577,
+      "step": 2036
+    },
+    {
+      "epoch": 0.43020063357972543,
+      "grad_norm": 0.2974567413330078,
+      "learning_rate": 6.275262751910428e-05,
+      "loss": 0.2595,
+      "step": 2037
+    },
+    {
+      "epoch": 0.4304118268215417,
+      "grad_norm": 0.27856704592704773,
+      "learning_rate": 6.271985553762238e-05,
+      "loss": 0.253,
+      "step": 2038
+    },
+    {
+      "epoch": 0.43062302006335795,
+      "grad_norm": 0.27032896876335144,
+      "learning_rate": 6.26870777125142e-05,
+      "loss": 0.2558,
+      "step": 2039
+    },
+    {
+      "epoch": 0.43083421330517424,
+      "grad_norm": 0.32427453994750977,
+      "learning_rate": 6.265429405883825e-05,
+      "loss": 0.259,
+      "step": 2040
+    },
+    {
+      "epoch": 0.4310454065469905,
+      "grad_norm": 0.2770988941192627,
+      "learning_rate": 6.262150459165563e-05,
+      "loss": 0.2554,
+      "step": 2041
+    },
+    {
+      "epoch": 0.43125659978880676,
+      "grad_norm": 0.2837235927581787,
+      "learning_rate": 6.258870932603018e-05,
+      "loss": 0.2539,
+      "step": 2042
+    },
+    {
+      "epoch": 0.431467793030623,
+      "grad_norm": 0.2972002327442169,
+      "learning_rate": 6.255590827702832e-05,
+      "loss": 0.2571,
+      "step": 2043
+    },
+    {
+      "epoch": 0.4316789862724393,
+      "grad_norm": 0.28603243827819824,
+      "learning_rate": 6.25231014597192e-05,
+      "loss": 0.2544,
+      "step": 2044
+    },
+    {
+      "epoch": 0.43189017951425557,
+      "grad_norm": 0.2790755033493042,
+      "learning_rate": 6.249028888917459e-05,
+      "loss": 0.2609,
+      "step": 2045
+    },
+    {
+      "epoch": 0.4321013727560718,
+      "grad_norm": 0.24967117607593536,
+      "learning_rate": 6.245747058046891e-05,
+      "loss": 0.27,
+      "step": 2046
+    },
+    {
+      "epoch": 0.4323125659978881,
+      "grad_norm": 0.3103407025337219,
+      "learning_rate": 6.24246465486792e-05,
+      "loss": 0.2647,
+      "step": 2047
+    },
+    {
+      "epoch": 0.4325237592397043,
+      "grad_norm": 0.28606992959976196,
+      "learning_rate": 6.239181680888516e-05,
+      "loss": 0.2586,
+      "step": 2048
+    },
+    {
+      "epoch": 0.4327349524815206,
+      "grad_norm": 0.3159793019294739,
+      "learning_rate": 6.235898137616907e-05,
+      "loss": 0.2582,
+      "step": 2049
+    },
+    {
+      "epoch": 0.43294614572333684,
+      "grad_norm": 0.28344643115997314,
+      "learning_rate": 6.232614026561587e-05,
+      "loss": 0.2627,
+      "step": 2050
+    },
+    {
+      "epoch": 0.4331573389651531,
+      "grad_norm": 0.32384514808654785,
+      "learning_rate": 6.229329349231306e-05,
+      "loss": 0.2524,
+      "step": 2051
+    },
+    {
+      "epoch": 0.43336853220696936,
+      "grad_norm": 0.3373807668685913,
+      "learning_rate": 6.22604410713508e-05,
+      "loss": 0.2582,
+      "step": 2052
+    },
+    {
+      "epoch": 0.43357972544878565,
+      "grad_norm": 0.3135640621185303,
+      "learning_rate": 6.222758301782182e-05,
+      "loss": 0.2553,
+      "step": 2053
+    },
+    {
+      "epoch": 0.4337909186906019,
+      "grad_norm": 0.28483468294143677,
+      "learning_rate": 6.21947193468214e-05,
+      "loss": 0.2567,
+      "step": 2054
+    },
+    {
+      "epoch": 0.43400211193241817,
+      "grad_norm": 0.33018743991851807,
+      "learning_rate": 6.216185007344744e-05,
+      "loss": 0.2576,
+      "step": 2055
+    },
+    {
+      "epoch": 0.4342133051742344,
+      "grad_norm": 0.29535144567489624,
+      "learning_rate": 6.212897521280045e-05,
+      "loss": 0.2536,
+      "step": 2056
+    },
+    {
+      "epoch": 0.4344244984160507,
+      "grad_norm": 0.33571016788482666,
+      "learning_rate": 6.209609477998338e-05,
+      "loss": 0.2548,
+      "step": 2057
+    },
+    {
+      "epoch": 0.434635691657867,
+      "grad_norm": 0.31034013628959656,
+      "learning_rate": 6.20632087901019e-05,
+      "loss": 0.256,
+      "step": 2058
+    },
+    {
+      "epoch": 0.4348468848996832,
+      "grad_norm": 0.28560197353363037,
+      "learning_rate": 6.203031725826412e-05,
+      "loss": 0.2521,
+      "step": 2059
+    },
+    {
+      "epoch": 0.4350580781414995,
+      "grad_norm": 0.29830431938171387,
+      "learning_rate": 6.199742019958073e-05,
+      "loss": 0.2484,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4352692713833157,
+      "grad_norm": 0.39499348402023315,
+      "learning_rate": 6.196451762916497e-05,
+      "loss": 0.2578,
+      "step": 2061
+    },
+    {
+      "epoch": 0.435480464625132,
+      "grad_norm": 0.28458210825920105,
+      "learning_rate": 6.193160956213262e-05,
+      "loss": 0.2554,
+      "step": 2062
+    },
+    {
+      "epoch": 0.43569165786694825,
+      "grad_norm": 0.30688294768333435,
+      "learning_rate": 6.189869601360195e-05,
+      "loss": 0.2582,
+      "step": 2063
+    },
+    {
+      "epoch": 0.43590285110876453,
+      "grad_norm": 0.39746373891830444,
+      "learning_rate": 6.186577699869373e-05,
+      "loss": 0.2567,
+      "step": 2064
+    },
+    {
+      "epoch": 0.43611404435058077,
+      "grad_norm": 0.33651888370513916,
+      "learning_rate": 6.183285253253135e-05,
+      "loss": 0.2576,
+      "step": 2065
+    },
+    {
+      "epoch": 0.43632523759239705,
+      "grad_norm": 0.4524444341659546,
+      "learning_rate": 6.17999226302406e-05,
+      "loss": 0.2573,
+      "step": 2066
+    },
+    {
+      "epoch": 0.4365364308342133,
+      "grad_norm": 0.29519718885421753,
+      "learning_rate": 6.176698730694977e-05,
+      "loss": 0.2525,
+      "step": 2067
+    },
+    {
+      "epoch": 0.4367476240760296,
+      "grad_norm": 0.3804580867290497,
+      "learning_rate": 6.173404657778972e-05,
+      "loss": 0.2547,
+      "step": 2068
+    },
+    {
+      "epoch": 0.4369588173178458,
+      "grad_norm": 0.3190968930721283,
+      "learning_rate": 6.17011004578937e-05,
+      "loss": 0.2533,
+      "step": 2069
+    },
+    {
+      "epoch": 0.4371700105596621,
+      "grad_norm": 0.35526371002197266,
+      "learning_rate": 6.166814896239752e-05,
+      "loss": 0.2554,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4373812038014784,
+      "grad_norm": 0.30569711327552795,
+      "learning_rate": 6.163519210643939e-05,
+      "loss": 0.251,
+      "step": 2071
+    },
+    {
+      "epoch": 0.4375923970432946,
+      "grad_norm": 0.4011141359806061,
+      "learning_rate": 6.160222990516001e-05,
+      "loss": 0.2542,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4378035902851109,
+      "grad_norm": 0.24943578243255615,
+      "learning_rate": 6.156926237370259e-05,
+      "loss": 0.2482,
+      "step": 2073
+    },
+    {
+      "epoch": 0.43801478352692713,
+      "grad_norm": 0.3444424271583557,
+      "learning_rate": 6.153628952721267e-05,
+      "loss": 0.2536,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4382259767687434,
+      "grad_norm": 0.33241593837738037,
+      "learning_rate": 6.150331138083832e-05,
+      "loss": 0.2548,
+      "step": 2075
+    },
+    {
+      "epoch": 0.43843717001055965,
+      "grad_norm": 0.343552827835083,
+      "learning_rate": 6.147032794973004e-05,
+      "loss": 0.2567,
+      "step": 2076
+    },
+    {
+      "epoch": 0.43864836325237594,
+      "grad_norm": 0.278913289308548,
+      "learning_rate": 6.143733924904075e-05,
+      "loss": 0.2565,
+      "step": 2077
+    },
+    {
+      "epoch": 0.43885955649419217,
+      "grad_norm": 0.3742517828941345,
+      "learning_rate": 6.140434529392575e-05,
+      "loss": 0.252,
+      "step": 2078
+    },
+    {
+      "epoch": 0.43907074973600846,
+      "grad_norm": 0.3341681957244873,
+      "learning_rate": 6.13713460995428e-05,
+      "loss": 0.2529,
+      "step": 2079
+    },
+    {
+      "epoch": 0.4392819429778247,
+      "grad_norm": 0.33502617478370667,
+      "learning_rate": 6.133834168105206e-05,
+      "loss": 0.2536,
+      "step": 2080
+    },
+    {
+      "epoch": 0.439493136219641,
+      "grad_norm": 0.3368094563484192,
+      "learning_rate": 6.130533205361606e-05,
+      "loss": 0.2559,
+      "step": 2081
+    },
+    {
+      "epoch": 0.4397043294614572,
+      "grad_norm": 0.3631986379623413,
+      "learning_rate": 6.127231723239978e-05,
+      "loss": 0.2555,
+      "step": 2082
+    },
+    {
+      "epoch": 0.4399155227032735,
+      "grad_norm": 0.29809483885765076,
+      "learning_rate": 6.123929723257053e-05,
+      "loss": 0.2497,
+      "step": 2083
+    },
+    {
+      "epoch": 0.44012671594508973,
+      "grad_norm": 0.32889872789382935,
+      "learning_rate": 6.120627206929804e-05,
+      "loss": 0.2529,
+      "step": 2084
+    },
+    {
+      "epoch": 0.440337909186906,
+      "grad_norm": 0.31574496626853943,
+      "learning_rate": 6.117324175775435e-05,
+      "loss": 0.2483,
+      "step": 2085
+    },
+    {
+      "epoch": 0.4405491024287223,
+      "grad_norm": 0.4180498421192169,
+      "learning_rate": 6.114020631311396e-05,
+      "loss": 0.2518,
+      "step": 2086
+    },
+    {
+      "epoch": 0.44076029567053854,
+      "grad_norm": 0.3480062782764435,
+      "learning_rate": 6.110716575055365e-05,
+      "loss": 0.2573,
+      "step": 2087
+    },
+    {
+      "epoch": 0.4409714889123548,
+      "grad_norm": 0.37658679485321045,
+      "learning_rate": 6.107412008525258e-05,
+      "loss": 0.2618,
+      "step": 2088
+    },
+    {
+      "epoch": 0.44118268215417106,
+      "grad_norm": 0.31667643785476685,
+      "learning_rate": 6.104106933239226e-05,
+      "loss": 0.2515,
+      "step": 2089
+    },
+    {
+      "epoch": 0.44139387539598735,
+      "grad_norm": 0.3231106102466583,
+      "learning_rate": 6.100801350715652e-05,
+      "loss": 0.2503,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4416050686378036,
+      "grad_norm": 0.32863879203796387,
+      "learning_rate": 6.097495262473156e-05,
+      "loss": 0.2519,
+      "step": 2091
+    },
+    {
+      "epoch": 0.44181626187961986,
+      "grad_norm": 0.6909302473068237,
+      "learning_rate": 6.094188670030583e-05,
+      "loss": 0.2537,
+      "step": 2092
+    },
+    {
+      "epoch": 0.4420274551214361,
+      "grad_norm": 0.3820771872997284,
+      "learning_rate": 6.090881574907016e-05,
+      "loss": 0.2529,
+      "step": 2093
+    },
+    {
+      "epoch": 0.4422386483632524,
+      "grad_norm": 0.47600027918815613,
+      "learning_rate": 6.087573978621768e-05,
+      "loss": 0.258,
+      "step": 2094
+    },
+    {
+      "epoch": 0.4424498416050686,
+      "grad_norm": 0.3532494604587555,
+      "learning_rate": 6.0842658826943777e-05,
+      "loss": 0.2481,
+      "step": 2095
+    },
+    {
+      "epoch": 0.4426610348468849,
+      "grad_norm": 0.40473610162734985,
+      "learning_rate": 6.0809572886446195e-05,
+      "loss": 0.257,
+      "step": 2096
+    },
+    {
+      "epoch": 0.44287222808870114,
+      "grad_norm": 0.31367939710617065,
+      "learning_rate": 6.077648197992495e-05,
+      "loss": 0.2588,
+      "step": 2097
+    },
+    {
+      "epoch": 0.4430834213305174,
+      "grad_norm": 0.4002780020236969,
+      "learning_rate": 6.0743386122582303e-05,
+      "loss": 0.2565,
+      "step": 2098
+    },
+    {
+      "epoch": 0.4432946145723337,
+      "grad_norm": 0.33418652415275574,
+      "learning_rate": 6.071028532962282e-05,
+      "loss": 0.2494,
+      "step": 2099
+    },
+    {
+      "epoch": 0.44350580781414994,
+      "grad_norm": 0.3279760777950287,
+      "learning_rate": 6.0677179616253344e-05,
+      "loss": 0.253,
+      "step": 2100
+    },
+    {
+      "epoch": 0.44371700105596623,
+      "grad_norm": 0.30447155237197876,
+      "learning_rate": 6.0644068997682944e-05,
+      "loss": 0.2468,
+      "step": 2101
+    },
+    {
+      "epoch": 0.44392819429778246,
+      "grad_norm": 0.305851012468338,
+      "learning_rate": 6.061095348912298e-05,
+      "loss": 0.2531,
+      "step": 2102
+    },
+    {
+      "epoch": 0.44413938753959875,
+      "grad_norm": 0.2829887866973877,
+      "learning_rate": 6.057783310578703e-05,
+      "loss": 0.2542,
+      "step": 2103
+    },
+    {
+      "epoch": 0.444350580781415,
+      "grad_norm": 0.41406047344207764,
+      "learning_rate": 6.054470786289094e-05,
+      "loss": 0.2544,
+      "step": 2104
+    },
+    {
+      "epoch": 0.44456177402323127,
+      "grad_norm": 0.39595961570739746,
+      "learning_rate": 6.051157777565274e-05,
+      "loss": 0.2585,
+      "step": 2105
+    },
+    {
+      "epoch": 0.4447729672650475,
+      "grad_norm": 0.3318381905555725,
+      "learning_rate": 6.047844285929274e-05,
+      "loss": 0.2581,
+      "step": 2106
+    },
+    {
+      "epoch": 0.4449841605068638,
+      "grad_norm": 0.37825435400009155,
+      "learning_rate": 6.0445303129033425e-05,
+      "loss": 0.2608,
+      "step": 2107
+    },
+    {
+      "epoch": 0.44519535374868,
+      "grad_norm": 0.35030773282051086,
+      "learning_rate": 6.0412158600099556e-05,
+      "loss": 0.2578,
+      "step": 2108
+    },
+    {
+      "epoch": 0.4454065469904963,
+      "grad_norm": 0.3581310212612152,
+      "learning_rate": 6.0379009287718014e-05,
+      "loss": 0.2495,
+      "step": 2109
+    },
+    {
+      "epoch": 0.44561774023231254,
+      "grad_norm": 0.3708845376968384,
+      "learning_rate": 6.034585520711792e-05,
+      "loss": 0.2507,
+      "step": 2110
+    },
+    {
+      "epoch": 0.44582893347412883,
+      "grad_norm": 0.3534052073955536,
+      "learning_rate": 6.0312696373530585e-05,
+      "loss": 0.2519,
+      "step": 2111
+    },
+    {
+      "epoch": 0.44604012671594506,
+      "grad_norm": 0.3367851674556732,
+      "learning_rate": 6.0279532802189544e-05,
+      "loss": 0.2556,
+      "step": 2112
+    },
+    {
+      "epoch": 0.44625131995776135,
+      "grad_norm": 0.31018221378326416,
+      "learning_rate": 6.0246364508330385e-05,
+      "loss": 0.2536,
+      "step": 2113
+    },
+    {
+      "epoch": 0.44646251319957764,
+      "grad_norm": 0.3064590096473694,
+      "learning_rate": 6.021319150719103e-05,
+      "loss": 0.2505,
+      "step": 2114
+    },
+    {
+      "epoch": 0.44667370644139387,
+      "grad_norm": 0.3618181645870209,
+      "learning_rate": 6.018001381401143e-05,
+      "loss": 0.2523,
+      "step": 2115
+    },
+    {
+      "epoch": 0.44688489968321016,
+      "grad_norm": 0.2696429193019867,
+      "learning_rate": 6.014683144403375e-05,
+      "loss": 0.2501,
+      "step": 2116
+    },
+    {
+      "epoch": 0.4470960929250264,
+      "grad_norm": 0.28446897864341736,
+      "learning_rate": 6.01136444125023e-05,
+      "loss": 0.252,
+      "step": 2117
+    },
+    {
+      "epoch": 0.4473072861668427,
+      "grad_norm": 0.2696904242038727,
+      "learning_rate": 6.008045273466354e-05,
+      "loss": 0.2541,
+      "step": 2118
+    },
+    {
+      "epoch": 0.4475184794086589,
+      "grad_norm": 0.2875731885433197,
+      "learning_rate": 6.004725642576604e-05,
+      "loss": 0.2556,
+      "step": 2119
+    },
+    {
+      "epoch": 0.4477296726504752,
+      "grad_norm": 0.48679593205451965,
+      "learning_rate": 6.001405550106052e-05,
+      "loss": 0.2563,
+      "step": 2120
+    },
+    {
+      "epoch": 0.44794086589229143,
+      "grad_norm": 0.30276724696159363,
+      "learning_rate": 5.998084997579978e-05,
+      "loss": 0.2522,
+      "step": 2121
+    },
+    {
+      "epoch": 0.4481520591341077,
+      "grad_norm": 0.40052706003189087,
+      "learning_rate": 5.9947639865238804e-05,
+      "loss": 0.2577,
+      "step": 2122
+    },
+    {
+      "epoch": 0.44836325237592395,
+      "grad_norm": 0.28717052936553955,
+      "learning_rate": 5.9914425184634614e-05,
+      "loss": 0.2561,
+      "step": 2123
+    },
+    {
+      "epoch": 0.44857444561774024,
+      "grad_norm": 0.43145620822906494,
+      "learning_rate": 5.9881205949246363e-05,
+      "loss": 0.2529,
+      "step": 2124
+    },
+    {
+      "epoch": 0.44878563885955647,
+      "grad_norm": 0.31818121671676636,
+      "learning_rate": 5.9847982174335316e-05,
+      "loss": 0.2638,
+      "step": 2125
+    },
+    {
+      "epoch": 0.44899683210137276,
+      "grad_norm": 0.3760785758495331,
+      "learning_rate": 5.981475387516476e-05,
+      "loss": 0.265,
+      "step": 2126
+    },
+    {
+      "epoch": 0.44920802534318904,
+      "grad_norm": 0.3204856514930725,
+      "learning_rate": 5.9781521067000144e-05,
+      "loss": 0.2549,
+      "step": 2127
+    },
+    {
+      "epoch": 0.4494192185850053,
+      "grad_norm": 0.34303295612335205,
+      "learning_rate": 5.9748283765108915e-05,
+      "loss": 0.2527,
+      "step": 2128
+    },
+    {
+      "epoch": 0.44963041182682156,
+      "grad_norm": 0.4593763053417206,
+      "learning_rate": 5.971504198476063e-05,
+      "loss": 0.2507,
+      "step": 2129
+    },
+    {
+      "epoch": 0.4498416050686378,
+      "grad_norm": 0.36021262407302856,
+      "learning_rate": 5.96817957412269e-05,
+      "loss": 0.2542,
+      "step": 2130
+    },
+    {
+      "epoch": 0.4500527983104541,
+      "grad_norm": 0.3517497479915619,
+      "learning_rate": 5.9648545049781325e-05,
+      "loss": 0.2546,
+      "step": 2131
+    },
+    {
+      "epoch": 0.4502639915522703,
+      "grad_norm": 0.36130425333976746,
+      "learning_rate": 5.961528992569967e-05,
+      "loss": 0.257,
+      "step": 2132
+    },
+    {
+      "epoch": 0.4504751847940866,
+      "grad_norm": 0.38368016481399536,
+      "learning_rate": 5.9582030384259615e-05,
+      "loss": 0.2549,
+      "step": 2133
+    },
+    {
+      "epoch": 0.45068637803590283,
+      "grad_norm": 0.2992406189441681,
+      "learning_rate": 5.954876644074091e-05,
+      "loss": 0.2555,
+      "step": 2134
+    },
+    {
+      "epoch": 0.4508975712777191,
+      "grad_norm": 0.3389708399772644,
+      "learning_rate": 5.9515498110425386e-05,
+      "loss": 0.2524,
+      "step": 2135
+    },
+    {
+      "epoch": 0.45110876451953535,
+      "grad_norm": 0.3367575407028198,
+      "learning_rate": 5.9482225408596794e-05,
+      "loss": 0.2497,
+      "step": 2136
+    },
+    {
+      "epoch": 0.45131995776135164,
+      "grad_norm": 0.31120380759239197,
+      "learning_rate": 5.9448948350540964e-05,
+      "loss": 0.2567,
+      "step": 2137
+    },
+    {
+      "epoch": 0.4515311510031679,
+      "grad_norm": 0.3055136501789093,
+      "learning_rate": 5.94156669515457e-05,
+      "loss": 0.2518,
+      "step": 2138
+    },
+    {
+      "epoch": 0.45174234424498416,
+      "grad_norm": 0.45693567395210266,
+      "learning_rate": 5.9382381226900805e-05,
+      "loss": 0.25,
+      "step": 2139
+    },
+    {
+      "epoch": 0.45195353748680045,
+      "grad_norm": 0.3099658191204071,
+      "learning_rate": 5.934909119189805e-05,
+      "loss": 0.2569,
+      "step": 2140
+    },
+    {
+      "epoch": 0.4521647307286167,
+      "grad_norm": 0.45669350028038025,
+      "learning_rate": 5.931579686183122e-05,
+      "loss": 0.2668,
+      "step": 2141
+    },
+    {
+      "epoch": 0.45237592397043297,
+      "grad_norm": 0.33291175961494446,
+      "learning_rate": 5.9282498251996044e-05,
+      "loss": 0.2509,
+      "step": 2142
+    },
+    {
+      "epoch": 0.4525871172122492,
+      "grad_norm": 0.42765241861343384,
+      "learning_rate": 5.9249195377690244e-05,
+      "loss": 0.2603,
+      "step": 2143
+    },
+    {
+      "epoch": 0.4527983104540655,
+      "grad_norm": 0.3447345793247223,
+      "learning_rate": 5.921588825421347e-05,
+      "loss": 0.2534,
+      "step": 2144
+    },
+    {
+      "epoch": 0.4530095036958817,
+      "grad_norm": 0.37823477387428284,
+      "learning_rate": 5.918257689686736e-05,
+      "loss": 0.2706,
+      "step": 2145
+    },
+    {
+      "epoch": 0.453220696937698,
+      "grad_norm": 0.3178291618824005,
+      "learning_rate": 5.914926132095546e-05,
+      "loss": 0.2615,
+      "step": 2146
+    },
+    {
+      "epoch": 0.45343189017951424,
+      "grad_norm": 0.3365192115306854,
+      "learning_rate": 5.911594154178328e-05,
+      "loss": 0.2546,
+      "step": 2147
+    },
+    {
+      "epoch": 0.45364308342133053,
+      "grad_norm": 0.4294624328613281,
+      "learning_rate": 5.908261757465825e-05,
+      "loss": 0.2553,
+      "step": 2148
+    },
+    {
+      "epoch": 0.45385427666314676,
+      "grad_norm": 0.3494749069213867,
+      "learning_rate": 5.9049289434889735e-05,
+      "loss": 0.2486,
+      "step": 2149
+    },
+    {
+      "epoch": 0.45406546990496305,
+      "grad_norm": 0.34204429388046265,
+      "learning_rate": 5.9015957137789e-05,
+      "loss": 0.2515,
+      "step": 2150
+    },
+    {
+      "epoch": 0.4542766631467793,
+      "grad_norm": 0.5559676289558411,
+      "learning_rate": 5.898262069866922e-05,
+      "loss": 0.2519,
+      "step": 2151
+    },
+    {
+      "epoch": 0.45448785638859557,
+      "grad_norm": 0.37397411465644836,
+      "learning_rate": 5.894928013284551e-05,
+      "loss": 0.2562,
+      "step": 2152
+    },
+    {
+      "epoch": 0.4546990496304118,
+      "grad_norm": 0.33559131622314453,
+      "learning_rate": 5.891593545563485e-05,
+      "loss": 0.256,
+      "step": 2153
+    },
+    {
+      "epoch": 0.4549102428722281,
+      "grad_norm": 0.3471549451351166,
+      "learning_rate": 5.888258668235607e-05,
+      "loss": 0.2522,
+      "step": 2154
+    },
+    {
+      "epoch": 0.4551214361140444,
+      "grad_norm": 0.3610290288925171,
+      "learning_rate": 5.884923382832996e-05,
+      "loss": 0.2511,
+      "step": 2155
+    },
+    {
+      "epoch": 0.4553326293558606,
+      "grad_norm": 0.3021371364593506,
+      "learning_rate": 5.881587690887914e-05,
+      "loss": 0.2479,
+      "step": 2156
+    },
+    {
+      "epoch": 0.4555438225976769,
+      "grad_norm": 0.35525020956993103,
+      "learning_rate": 5.8782515939328106e-05,
+      "loss": 0.2586,
+      "step": 2157
+    },
+    {
+      "epoch": 0.4557550158394931,
+      "grad_norm": 0.284553587436676,
+      "learning_rate": 5.874915093500321e-05,
+      "loss": 0.2529,
+      "step": 2158
+    },
+    {
+      "epoch": 0.4559662090813094,
+      "grad_norm": 0.3198205828666687,
+      "learning_rate": 5.871578191123266e-05,
+      "loss": 0.249,
+      "step": 2159
+    },
+    {
+      "epoch": 0.45617740232312565,
+      "grad_norm": 0.4199419915676117,
+      "learning_rate": 5.868240888334653e-05,
+      "loss": 0.2478,
+      "step": 2160
+    },
+    {
+      "epoch": 0.45638859556494193,
+      "grad_norm": 0.33335763216018677,
+      "learning_rate": 5.864903186667668e-05,
+      "loss": 0.2517,
+      "step": 2161
+    },
+    {
+      "epoch": 0.45659978880675817,
+      "grad_norm": 0.30836915969848633,
+      "learning_rate": 5.861565087655688e-05,
+      "loss": 0.2522,
+      "step": 2162
+    },
+    {
+      "epoch": 0.45681098204857445,
+      "grad_norm": 0.3462926149368286,
+      "learning_rate": 5.858226592832265e-05,
+      "loss": 0.2475,
+      "step": 2163
+    },
+    {
+      "epoch": 0.4570221752903907,
+      "grad_norm": 0.2998363971710205,
+      "learning_rate": 5.85488770373114e-05,
+      "loss": 0.248,
+      "step": 2164
+    },
+    {
+      "epoch": 0.457233368532207,
+      "grad_norm": 0.3168848156929016,
+      "learning_rate": 5.8515484218862284e-05,
+      "loss": 0.2453,
+      "step": 2165
+    },
+    {
+      "epoch": 0.4574445617740232,
+      "grad_norm": 0.30999934673309326,
+      "learning_rate": 5.8482087488316274e-05,
+      "loss": 0.2501,
+      "step": 2166
+    },
+    {
+      "epoch": 0.4576557550158395,
+      "grad_norm": 0.36397939920425415,
+      "learning_rate": 5.844868686101621e-05,
+      "loss": 0.2533,
+      "step": 2167
+    },
+    {
+      "epoch": 0.4578669482576558,
+      "grad_norm": 0.3662760853767395,
+      "learning_rate": 5.8415282352306624e-05,
+      "loss": 0.256,
+      "step": 2168
+    },
+    {
+      "epoch": 0.458078141499472,
+      "grad_norm": 0.3037903308868408,
+      "learning_rate": 5.838187397753386e-05,
+      "loss": 0.2577,
+      "step": 2169
+    },
+    {
+      "epoch": 0.4582893347412883,
+      "grad_norm": 0.32793018221855164,
+      "learning_rate": 5.8348461752046116e-05,
+      "loss": 0.2663,
+      "step": 2170
+    },
+    {
+      "epoch": 0.45850052798310453,
+      "grad_norm": 0.33678099513053894,
+      "learning_rate": 5.831504569119325e-05,
+      "loss": 0.2517,
+      "step": 2171
+    },
+    {
+      "epoch": 0.4587117212249208,
+      "grad_norm": 0.36083343625068665,
+      "learning_rate": 5.828162581032691e-05,
+      "loss": 0.2497,
+      "step": 2172
+    },
+    {
+      "epoch": 0.45892291446673705,
+      "grad_norm": 0.3053895831108093,
+      "learning_rate": 5.824820212480057e-05,
+      "loss": 0.2484,
+      "step": 2173
+    },
+    {
+      "epoch": 0.45913410770855334,
+      "grad_norm": 0.3070761263370514,
+      "learning_rate": 5.821477464996936e-05,
+      "loss": 0.2422,
+      "step": 2174
+    },
+    {
+      "epoch": 0.4593453009503696,
+      "grad_norm": 0.32045868039131165,
+      "learning_rate": 5.8181343401190205e-05,
+      "loss": 0.2526,
+      "step": 2175
+    },
+    {
+      "epoch": 0.45955649419218586,
+      "grad_norm": 0.5066584944725037,
+      "learning_rate": 5.8147908393821736e-05,
+      "loss": 0.2546,
+      "step": 2176
+    },
+    {
+      "epoch": 0.4597676874340021,
+      "grad_norm": 0.32927635312080383,
+      "learning_rate": 5.8114469643224334e-05,
+      "loss": 0.2577,
+      "step": 2177
+    },
+    {
+      "epoch": 0.4599788806758184,
+      "grad_norm": 0.3166161775588989,
+      "learning_rate": 5.8081027164760094e-05,
+      "loss": 0.2541,
+      "step": 2178
+    },
+    {
+      "epoch": 0.4601900739176346,
+      "grad_norm": 0.3751268982887268,
+      "learning_rate": 5.8047580973792805e-05,
+      "loss": 0.2526,
+      "step": 2179
+    },
+    {
+      "epoch": 0.4604012671594509,
+      "grad_norm": 0.29679134488105774,
+      "learning_rate": 5.801413108568797e-05,
+      "loss": 0.2521,
+      "step": 2180
+    },
+    {
+      "epoch": 0.46061246040126713,
+      "grad_norm": 0.34050193428993225,
+      "learning_rate": 5.798067751581282e-05,
+      "loss": 0.2441,
+      "step": 2181
+    },
+    {
+      "epoch": 0.4608236536430834,
+      "grad_norm": 0.30644261837005615,
+      "learning_rate": 5.794722027953623e-05,
+      "loss": 0.2468,
+      "step": 2182
+    },
+    {
+      "epoch": 0.4610348468848997,
+      "grad_norm": 0.331982284784317,
+      "learning_rate": 5.791375939222879e-05,
+      "loss": 0.2427,
+      "step": 2183
+    },
+    {
+      "epoch": 0.46124604012671594,
+      "grad_norm": 0.391683429479599,
+      "learning_rate": 5.788029486926275e-05,
+      "loss": 0.2517,
+      "step": 2184
+    },
+    {
+      "epoch": 0.4614572333685322,
+      "grad_norm": 0.385872483253479,
+      "learning_rate": 5.784682672601207e-05,
+      "loss": 0.2543,
+      "step": 2185
+    },
+    {
+      "epoch": 0.46166842661034846,
+      "grad_norm": 0.39227354526519775,
+      "learning_rate": 5.78133549778523e-05,
+      "loss": 0.2627,
+      "step": 2186
+    },
+    {
+      "epoch": 0.46187961985216475,
+      "grad_norm": 0.33782193064689636,
+      "learning_rate": 5.7779879640160716e-05,
+      "loss": 0.2509,
+      "step": 2187
+    },
+    {
+      "epoch": 0.462090813093981,
+      "grad_norm": 0.3989555537700653,
+      "learning_rate": 5.774640072831622e-05,
+      "loss": 0.2446,
+      "step": 2188
+    },
+    {
+      "epoch": 0.46230200633579727,
+      "grad_norm": 0.32866227626800537,
+      "learning_rate": 5.771291825769932e-05,
+      "loss": 0.2471,
+      "step": 2189
+    },
+    {
+      "epoch": 0.4625131995776135,
+      "grad_norm": 0.3718309998512268,
+      "learning_rate": 5.7679432243692236e-05,
+      "loss": 0.2511,
+      "step": 2190
+    },
+    {
+      "epoch": 0.4627243928194298,
+      "grad_norm": 0.32004842162132263,
+      "learning_rate": 5.764594270167874e-05,
+      "loss": 0.2465,
+      "step": 2191
+    },
+    {
+      "epoch": 0.462935586061246,
+      "grad_norm": 0.37257716059684753,
+      "learning_rate": 5.7612449647044265e-05,
+      "loss": 0.2532,
+      "step": 2192
+    },
+    {
+      "epoch": 0.4631467793030623,
+      "grad_norm": 0.31757742166519165,
+      "learning_rate": 5.7578953095175844e-05,
+      "loss": 0.2538,
+      "step": 2193
+    },
+    {
+      "epoch": 0.46335797254487854,
+      "grad_norm": 0.34105828404426575,
+      "learning_rate": 5.754545306146214e-05,
+      "loss": 0.2502,
+      "step": 2194
+    },
+    {
+      "epoch": 0.4635691657866948,
+      "grad_norm": 0.2900277376174927,
+      "learning_rate": 5.7511949561293363e-05,
+      "loss": 0.2482,
+      "step": 2195
+    },
+    {
+      "epoch": 0.4637803590285111,
+      "grad_norm": 0.3169609010219574,
+      "learning_rate": 5.7478442610061375e-05,
+      "loss": 0.2515,
+      "step": 2196
+    },
+    {
+      "epoch": 0.46399155227032735,
+      "grad_norm": 0.35615748167037964,
+      "learning_rate": 5.7444932223159595e-05,
+      "loss": 0.2496,
+      "step": 2197
+    },
+    {
+      "epoch": 0.46420274551214363,
+      "grad_norm": 0.31895196437835693,
+      "learning_rate": 5.7411418415982996e-05,
+      "loss": 0.2525,
+      "step": 2198
+    },
+    {
+      "epoch": 0.46441393875395987,
+      "grad_norm": 0.3096584379673004,
+      "learning_rate": 5.737790120392821e-05,
+      "loss": 0.2605,
+      "step": 2199
+    },
+    {
+      "epoch": 0.46462513199577615,
+      "grad_norm": 0.34887707233428955,
+      "learning_rate": 5.734438060239331e-05,
+      "loss": 0.2546,
+      "step": 2200
+    },
+    {
+      "epoch": 0.46462513199577615,
+      "eval_train_split_loss": 0.35434871912002563,
+      "eval_train_split_runtime": 8.0071,
+      "eval_train_split_samples_per_second": 1.249,
+      "eval_train_split_steps_per_second": 0.125,
+      "step": 2200
+    },
+    {
+      "epoch": 0.4648363252375924,
+      "grad_norm": 0.2847842276096344,
+      "learning_rate": 5.7310856626778e-05,
+      "loss": 0.2542,
+      "step": 2201
+    },
+    {
+      "epoch": 0.4650475184794087,
+      "grad_norm": 0.3099900782108307,
+      "learning_rate": 5.727732929248356e-05,
+      "loss": 0.2512,
+      "step": 2202
+    },
+    {
+      "epoch": 0.4652587117212249,
+      "grad_norm": 0.37871041893959045,
+      "learning_rate": 5.724379861491275e-05,
+      "loss": 0.2538,
+      "step": 2203
+    },
+    {
+      "epoch": 0.4654699049630412,
+      "grad_norm": 0.3118206262588501,
+      "learning_rate": 5.721026460946988e-05,
+      "loss": 0.2531,
+      "step": 2204
+    },
+    {
+      "epoch": 0.4656810982048574,
+      "grad_norm": 0.28658244013786316,
+      "learning_rate": 5.7176727291560814e-05,
+      "loss": 0.2494,
+      "step": 2205
+    },
+    {
+      "epoch": 0.4658922914466737,
+      "grad_norm": 0.3070621192455292,
+      "learning_rate": 5.7143186676592944e-05,
+      "loss": 0.249,
+      "step": 2206
+    },
+    {
+      "epoch": 0.46610348468848994,
+      "grad_norm": 0.2591784596443176,
+      "learning_rate": 5.7109642779975105e-05,
+      "loss": 0.2478,
+      "step": 2207
+    },
+    {
+      "epoch": 0.46631467793030623,
+      "grad_norm": 1.7409822940826416,
+      "learning_rate": 5.707609561711774e-05,
+      "loss": 0.2482,
+      "step": 2208
+    },
+    {
+      "epoch": 0.4665258711721225,
+      "grad_norm": 0.773737370967865,
+      "learning_rate": 5.7042545203432755e-05,
+      "loss": 0.2573,
+      "step": 2209
+    },
+    {
+      "epoch": 0.46673706441393875,
+      "grad_norm": 0.35473713278770447,
+      "learning_rate": 5.7008991554333466e-05,
+      "loss": 0.2538,
+      "step": 2210
+    },
+    {
+      "epoch": 0.46694825765575504,
+      "grad_norm": 0.33380743861198425,
+      "learning_rate": 5.697543468523482e-05,
+      "loss": 0.256,
+      "step": 2211
+    },
+    {
+      "epoch": 0.46715945089757127,
+      "grad_norm": 0.30332252383232117,
+      "learning_rate": 5.694187461155314e-05,
+      "loss": 0.2506,
+      "step": 2212
+    },
+    {
+      "epoch": 0.46737064413938756,
+      "grad_norm": 0.27759674191474915,
+      "learning_rate": 5.6908311348706255e-05,
+      "loss": 0.2473,
+      "step": 2213
+    },
+    {
+      "epoch": 0.4675818373812038,
+      "grad_norm": 0.34392431378364563,
+      "learning_rate": 5.687474491211346e-05,
+      "loss": 0.2509,
+      "step": 2214
+    },
+    {
+      "epoch": 0.4677930306230201,
+      "grad_norm": 0.30198416113853455,
+      "learning_rate": 5.6841175317195515e-05,
+      "loss": 0.2526,
+      "step": 2215
+    },
+    {
+      "epoch": 0.4680042238648363,
+      "grad_norm": 0.27786314487457275,
+      "learning_rate": 5.680760257937461e-05,
+      "loss": 0.2495,
+      "step": 2216
+    },
+    {
+      "epoch": 0.4682154171066526,
+      "grad_norm": 0.331228643655777,
+      "learning_rate": 5.677402671407438e-05,
+      "loss": 0.2524,
+      "step": 2217
+    },
+    {
+      "epoch": 0.46842661034846883,
+      "grad_norm": 0.30430442094802856,
+      "learning_rate": 5.6740447736719915e-05,
+      "loss": 0.2557,
+      "step": 2218
+    },
+    {
+      "epoch": 0.4686378035902851,
+      "grad_norm": 0.34494727849960327,
+      "learning_rate": 5.670686566273773e-05,
+      "loss": 0.2506,
+      "step": 2219
+    },
+    {
+      "epoch": 0.46884899683210135,
+      "grad_norm": 0.3453083634376526,
+      "learning_rate": 5.667328050755576e-05,
+      "loss": 0.2488,
+      "step": 2220
+    },
+    {
+      "epoch": 0.46906019007391764,
+      "grad_norm": 0.3131173253059387,
+      "learning_rate": 5.663969228660335e-05,
+      "loss": 0.2499,
+      "step": 2221
+    },
+    {
+      "epoch": 0.46927138331573387,
+      "grad_norm": 0.37547609210014343,
+      "learning_rate": 5.6606101015311255e-05,
+      "loss": 0.2497,
+      "step": 2222
+    },
+    {
+      "epoch": 0.46948257655755016,
+      "grad_norm": 0.31078779697418213,
+      "learning_rate": 5.657250670911165e-05,
+      "loss": 0.252,
+      "step": 2223
+    },
+    {
+      "epoch": 0.46969376979936645,
+      "grad_norm": 0.383341521024704,
+      "learning_rate": 5.653890938343805e-05,
+      "loss": 0.2536,
+      "step": 2224
+    },
+    {
+      "epoch": 0.4699049630411827,
+      "grad_norm": 0.35261204838752747,
+      "learning_rate": 5.6505309053725444e-05,
+      "loss": 0.2551,
+      "step": 2225
+    },
+    {
+      "epoch": 0.47011615628299896,
+      "grad_norm": 0.3596508502960205,
+      "learning_rate": 5.647170573541014e-05,
+      "loss": 0.2497,
+      "step": 2226
+    },
+    {
+      "epoch": 0.4703273495248152,
+      "grad_norm": 0.3035337030887604,
+      "learning_rate": 5.643809944392979e-05,
+      "loss": 0.2498,
+      "step": 2227
+    },
+    {
+      "epoch": 0.4705385427666315,
+      "grad_norm": 0.5890151262283325,
+      "learning_rate": 5.64044901947235e-05,
+      "loss": 0.2623,
+      "step": 2228
+    },
+    {
+      "epoch": 0.4707497360084477,
+      "grad_norm": 0.3171532154083252,
+      "learning_rate": 5.6370878003231674e-05,
+      "loss": 0.251,
+      "step": 2229
+    },
+    {
+      "epoch": 0.470960929250264,
+      "grad_norm": 0.3569689393043518,
+      "learning_rate": 5.633726288489609e-05,
+      "loss": 0.2525,
+      "step": 2230
+    },
+    {
+      "epoch": 0.47117212249208024,
+      "grad_norm": 0.3469400107860565,
+      "learning_rate": 5.630364485515984e-05,
+      "loss": 0.2521,
+      "step": 2231
+    },
+    {
+      "epoch": 0.4713833157338965,
+      "grad_norm": 0.39287418127059937,
+      "learning_rate": 5.62700239294674e-05,
+      "loss": 0.2511,
+      "step": 2232
+    },
+    {
+      "epoch": 0.47159450897571276,
+      "grad_norm": 0.3524969518184662,
+      "learning_rate": 5.623640012326454e-05,
+      "loss": 0.2512,
+      "step": 2233
+    },
+    {
+      "epoch": 0.47180570221752904,
+      "grad_norm": 0.3939574658870697,
+      "learning_rate": 5.620277345199837e-05,
+      "loss": 0.2529,
+      "step": 2234
+    },
+    {
+      "epoch": 0.4720168954593453,
+      "grad_norm": 0.3080081641674042,
+      "learning_rate": 5.616914393111732e-05,
+      "loss": 0.247,
+      "step": 2235
+    },
+    {
+      "epoch": 0.47222808870116156,
+      "grad_norm": 0.39265573024749756,
+      "learning_rate": 5.613551157607109e-05,
+      "loss": 0.2528,
+      "step": 2236
+    },
+    {
+      "epoch": 0.47243928194297785,
+      "grad_norm": 0.3170768916606903,
+      "learning_rate": 5.610187640231076e-05,
+      "loss": 0.2539,
+      "step": 2237
+    },
+    {
+      "epoch": 0.4726504751847941,
+      "grad_norm": 0.39708417654037476,
+      "learning_rate": 5.606823842528862e-05,
+      "loss": 0.2526,
+      "step": 2238
+    },
+    {
+      "epoch": 0.47286166842661037,
+      "grad_norm": 0.31597837805747986,
+      "learning_rate": 5.60345976604583e-05,
+      "loss": 0.2539,
+      "step": 2239
+    },
+    {
+      "epoch": 0.4730728616684266,
+      "grad_norm": 0.32596608996391296,
+      "learning_rate": 5.6000954123274704e-05,
+      "loss": 0.2476,
+      "step": 2240
+    },
+    {
+      "epoch": 0.4732840549102429,
+      "grad_norm": 0.2788051664829254,
+      "learning_rate": 5.596730782919401e-05,
+      "loss": 0.2448,
+      "step": 2241
+    },
+    {
+      "epoch": 0.4734952481520591,
+      "grad_norm": 0.35000288486480713,
+      "learning_rate": 5.5933658793673607e-05,
+      "loss": 0.251,
+      "step": 2242
+    },
+    {
+      "epoch": 0.4737064413938754,
+      "grad_norm": 0.3339793086051941,
+      "learning_rate": 5.590000703217224e-05,
+      "loss": 0.2497,
+      "step": 2243
+    },
+    {
+      "epoch": 0.47391763463569164,
+      "grad_norm": 0.3502020239830017,
+      "learning_rate": 5.586635256014986e-05,
+      "loss": 0.2546,
+      "step": 2244
+    },
+    {
+      "epoch": 0.47412882787750793,
+      "grad_norm": 0.2961342930793762,
+      "learning_rate": 5.583269539306762e-05,
+      "loss": 0.2496,
+      "step": 2245
+    },
+    {
+      "epoch": 0.47434002111932416,
+      "grad_norm": 0.3587300479412079,
+      "learning_rate": 5.579903554638799e-05,
+      "loss": 0.2534,
+      "step": 2246
+    },
+    {
+      "epoch": 0.47455121436114045,
+      "grad_norm": 0.3355439305305481,
+      "learning_rate": 5.576537303557463e-05,
+      "loss": 0.2484,
+      "step": 2247
+    },
+    {
+      "epoch": 0.4747624076029567,
+      "grad_norm": 0.3442060351371765,
+      "learning_rate": 5.57317078760924e-05,
+      "loss": 0.2464,
+      "step": 2248
+    },
+    {
+      "epoch": 0.47497360084477297,
+      "grad_norm": 0.29550787806510925,
+      "learning_rate": 5.569804008340743e-05,
+      "loss": 0.2497,
+      "step": 2249
+    },
+    {
+      "epoch": 0.4751847940865892,
+      "grad_norm": 0.322881817817688,
+      "learning_rate": 5.566436967298703e-05,
+      "loss": 0.25,
+      "step": 2250
+    },
+    {
+      "epoch": 0.4753959873284055,
+      "grad_norm": 0.3460955321788788,
+      "learning_rate": 5.5630696660299694e-05,
+      "loss": 0.253,
+      "step": 2251
+    },
+    {
+      "epoch": 0.4756071805702218,
+      "grad_norm": 0.43800613284111023,
+      "learning_rate": 5.559702106081517e-05,
+      "loss": 0.2475,
+      "step": 2252
+    },
+    {
+      "epoch": 0.475818373812038,
+      "grad_norm": 0.26661649346351624,
+      "learning_rate": 5.556334289000433e-05,
+      "loss": 0.245,
+      "step": 2253
+    },
+    {
+      "epoch": 0.4760295670538543,
+      "grad_norm": 0.3388286828994751,
+      "learning_rate": 5.5529662163339266e-05,
+      "loss": 0.2431,
+      "step": 2254
+    },
+    {
+      "epoch": 0.47624076029567053,
+      "grad_norm": 0.30181705951690674,
+      "learning_rate": 5.5495978896293244e-05,
+      "loss": 0.2483,
+      "step": 2255
+    },
+    {
+      "epoch": 0.4764519535374868,
+      "grad_norm": 0.2626550495624542,
+      "learning_rate": 5.546229310434067e-05,
+      "loss": 0.2525,
+      "step": 2256
+    },
+    {
+      "epoch": 0.47666314677930305,
+      "grad_norm": 0.34363436698913574,
+      "learning_rate": 5.542860480295712e-05,
+      "loss": 0.2536,
+      "step": 2257
+    },
+    {
+      "epoch": 0.47687434002111934,
+      "grad_norm": 0.28438371419906616,
+      "learning_rate": 5.539491400761937e-05,
+      "loss": 0.2535,
+      "step": 2258
+    },
+    {
+      "epoch": 0.47708553326293557,
+      "grad_norm": 0.3227217197418213,
+      "learning_rate": 5.536122073380525e-05,
+      "loss": 0.2499,
+      "step": 2259
+    },
+    {
+      "epoch": 0.47729672650475186,
+      "grad_norm": 0.3044903576374054,
+      "learning_rate": 5.532752499699381e-05,
+      "loss": 0.25,
+      "step": 2260
+    },
+    {
+      "epoch": 0.4775079197465681,
+      "grad_norm": 0.2860196828842163,
+      "learning_rate": 5.529382681266521e-05,
+      "loss": 0.2498,
+      "step": 2261
+    },
+    {
+      "epoch": 0.4777191129883844,
+      "grad_norm": 0.31576815247535706,
+      "learning_rate": 5.52601261963007e-05,
+      "loss": 0.2446,
+      "step": 2262
+    },
+    {
+      "epoch": 0.4779303062302006,
+      "grad_norm": 0.3119003474712372,
+      "learning_rate": 5.522642316338268e-05,
+      "loss": 0.2449,
+      "step": 2263
+    },
+    {
+      "epoch": 0.4781414994720169,
+      "grad_norm": 0.34888410568237305,
+      "learning_rate": 5.519271772939466e-05,
+      "loss": 0.2491,
+      "step": 2264
+    },
+    {
+      "epoch": 0.4783526927138332,
+      "grad_norm": 0.3391266167163849,
+      "learning_rate": 5.5159009909821246e-05,
+      "loss": 0.2501,
+      "step": 2265
+    },
+    {
+      "epoch": 0.4785638859556494,
+      "grad_norm": 0.33687084913253784,
+      "learning_rate": 5.5125299720148146e-05,
+      "loss": 0.2544,
+      "step": 2266
+    },
+    {
+      "epoch": 0.4787750791974657,
+      "grad_norm": 0.2791455090045929,
+      "learning_rate": 5.509158717586215e-05,
+      "loss": 0.2487,
+      "step": 2267
+    },
+    {
+      "epoch": 0.47898627243928193,
+      "grad_norm": 0.2468070685863495,
+      "learning_rate": 5.505787229245112e-05,
+      "loss": 0.2464,
+      "step": 2268
+    },
+    {
+      "epoch": 0.4791974656810982,
+      "grad_norm": 0.30782583355903625,
+      "learning_rate": 5.5024155085404005e-05,
+      "loss": 0.2473,
+      "step": 2269
+    },
+    {
+      "epoch": 0.47940865892291445,
+      "grad_norm": 0.2594704031944275,
+      "learning_rate": 5.499043557021083e-05,
+      "loss": 0.2505,
+      "step": 2270
+    },
+    {
+      "epoch": 0.47961985216473074,
+      "grad_norm": 0.2937149703502655,
+      "learning_rate": 5.4956713762362664e-05,
+      "loss": 0.2505,
+      "step": 2271
+    },
+    {
+      "epoch": 0.479831045406547,
+      "grad_norm": 0.274291455745697,
+      "learning_rate": 5.4922989677351634e-05,
+      "loss": 0.2494,
+      "step": 2272
+    },
+    {
+      "epoch": 0.48004223864836326,
+      "grad_norm": 0.3258136808872223,
+      "learning_rate": 5.4889263330670924e-05,
+      "loss": 0.248,
+      "step": 2273
+    },
+    {
+      "epoch": 0.4802534318901795,
+      "grad_norm": 0.32571765780448914,
+      "learning_rate": 5.485553473781474e-05,
+      "loss": 0.2524,
+      "step": 2274
+    },
+    {
+      "epoch": 0.4804646251319958,
+      "grad_norm": 0.36827027797698975,
+      "learning_rate": 5.4821803914278336e-05,
+      "loss": 0.254,
+      "step": 2275
+    },
+    {
+      "epoch": 0.480675818373812,
+      "grad_norm": 0.38788777589797974,
+      "learning_rate": 5.478807087555797e-05,
+      "loss": 0.2498,
+      "step": 2276
+    },
+    {
+      "epoch": 0.4808870116156283,
+      "grad_norm": 0.3337920308113098,
+      "learning_rate": 5.475433563715093e-05,
+      "loss": 0.248,
+      "step": 2277
+    },
+    {
+      "epoch": 0.4810982048574446,
+      "grad_norm": 0.37426120042800903,
+      "learning_rate": 5.472059821455554e-05,
+      "loss": 0.2523,
+      "step": 2278
+    },
+    {
+      "epoch": 0.4813093980992608,
+      "grad_norm": 0.30647310614585876,
+      "learning_rate": 5.4686858623271086e-05,
+      "loss": 0.2479,
+      "step": 2279
+    },
+    {
+      "epoch": 0.4815205913410771,
+      "grad_norm": 0.362946093082428,
+      "learning_rate": 5.465311687879785e-05,
+      "loss": 0.2464,
+      "step": 2280
+    },
+    {
+      "epoch": 0.48173178458289334,
+      "grad_norm": 0.3455063998699188,
+      "learning_rate": 5.461937299663715e-05,
+      "loss": 0.2491,
+      "step": 2281
+    },
+    {
+      "epoch": 0.48194297782470963,
+      "grad_norm": 0.3110775947570801,
+      "learning_rate": 5.458562699229125e-05,
+      "loss": 0.2496,
+      "step": 2282
+    },
+    {
+      "epoch": 0.48215417106652586,
+      "grad_norm": 0.270971417427063,
+      "learning_rate": 5.455187888126336e-05,
+      "loss": 0.2508,
+      "step": 2283
+    },
+    {
+      "epoch": 0.48236536430834215,
+      "grad_norm": 0.3419160842895508,
+      "learning_rate": 5.451812867905775e-05,
+      "loss": 0.2434,
+      "step": 2284
+    },
+    {
+      "epoch": 0.4825765575501584,
+      "grad_norm": 0.31939828395843506,
+      "learning_rate": 5.4484376401179534e-05,
+      "loss": 0.2429,
+      "step": 2285
+    },
+    {
+      "epoch": 0.48278775079197467,
+      "grad_norm": 0.3361072242259979,
+      "learning_rate": 5.445062206313489e-05,
+      "loss": 0.2534,
+      "step": 2286
+    },
+    {
+      "epoch": 0.4829989440337909,
+      "grad_norm": 0.34176942706108093,
+      "learning_rate": 5.4416865680430865e-05,
+      "loss": 0.2507,
+      "step": 2287
+    },
+    {
+      "epoch": 0.4832101372756072,
+      "grad_norm": 0.39322641491889954,
+      "learning_rate": 5.4383107268575476e-05,
+      "loss": 0.2486,
+      "step": 2288
+    },
+    {
+      "epoch": 0.4834213305174234,
+      "grad_norm": 0.47020938992500305,
+      "learning_rate": 5.434934684307767e-05,
+      "loss": 0.2499,
+      "step": 2289
+    },
+    {
+      "epoch": 0.4836325237592397,
+      "grad_norm": 0.3252333700656891,
+      "learning_rate": 5.431558441944731e-05,
+      "loss": 0.2522,
+      "step": 2290
+    },
+    {
+      "epoch": 0.48384371700105594,
+      "grad_norm": 0.549699604511261,
+      "learning_rate": 5.4281820013195195e-05,
+      "loss": 0.2539,
+      "step": 2291
+    },
+    {
+      "epoch": 0.4840549102428722,
+      "grad_norm": 0.4672764241695404,
+      "learning_rate": 5.4248053639833006e-05,
+      "loss": 0.2517,
+      "step": 2292
+    },
+    {
+      "epoch": 0.4842661034846885,
+      "grad_norm": 0.3819134831428528,
+      "learning_rate": 5.421428531487339e-05,
+      "loss": 0.2488,
+      "step": 2293
+    },
+    {
+      "epoch": 0.48447729672650475,
+      "grad_norm": 0.37068068981170654,
+      "learning_rate": 5.418051505382979e-05,
+      "loss": 0.2518,
+      "step": 2294
+    },
+    {
+      "epoch": 0.48468848996832103,
+      "grad_norm": 0.3229024112224579,
+      "learning_rate": 5.414674287221663e-05,
+      "loss": 0.2506,
+      "step": 2295
+    },
+    {
+      "epoch": 0.48489968321013727,
+      "grad_norm": 0.2873823046684265,
+      "learning_rate": 5.411296878554918e-05,
+      "loss": 0.24,
+      "step": 2296
+    },
+    {
+      "epoch": 0.48511087645195355,
+      "grad_norm": 0.31863290071487427,
+      "learning_rate": 5.407919280934357e-05,
+      "loss": 0.2399,
+      "step": 2297
+    },
+    {
+      "epoch": 0.4853220696937698,
+      "grad_norm": 0.28696224093437195,
+      "learning_rate": 5.40454149591168e-05,
+      "loss": 0.2463,
+      "step": 2298
+    },
+    {
+      "epoch": 0.4855332629355861,
+      "grad_norm": 0.26987582445144653,
+      "learning_rate": 5.4011635250386794e-05,
+      "loss": 0.2437,
+      "step": 2299
+    },
+    {
+      "epoch": 0.4857444561774023,
+      "grad_norm": 0.2787511348724365,
+      "learning_rate": 5.397785369867226e-05,
+      "loss": 0.2443,
+      "step": 2300
+    },
+    {
+      "epoch": 0.4859556494192186,
+      "grad_norm": 0.27399760484695435,
+      "learning_rate": 5.3944070319492746e-05,
+      "loss": 0.2451,
+      "step": 2301
+    },
+    {
+      "epoch": 0.4861668426610348,
+      "grad_norm": 0.34276965260505676,
+      "learning_rate": 5.39102851283687e-05,
+      "loss": 0.24,
+      "step": 2302
+    },
+    {
+      "epoch": 0.4863780359028511,
+      "grad_norm": 0.3174133598804474,
+      "learning_rate": 5.387649814082136e-05,
+      "loss": 0.2426,
+      "step": 2303
+    },
+    {
+      "epoch": 0.48658922914466735,
+      "grad_norm": 0.3834768533706665,
+      "learning_rate": 5.384270937237279e-05,
+      "loss": 0.2439,
+      "step": 2304
+    },
+    {
+      "epoch": 0.48680042238648363,
+      "grad_norm": 0.29933610558509827,
+      "learning_rate": 5.380891883854591e-05,
+      "loss": 0.2466,
+      "step": 2305
+    },
+    {
+      "epoch": 0.4870116156282999,
+      "grad_norm": 0.30304303765296936,
+      "learning_rate": 5.3775126554864386e-05,
+      "loss": 0.251,
+      "step": 2306
+    },
+    {
+      "epoch": 0.48722280887011615,
+      "grad_norm": 0.3319176435470581,
+      "learning_rate": 5.374133253685274e-05,
+      "loss": 0.2533,
+      "step": 2307
+    },
+    {
+      "epoch": 0.48743400211193244,
+      "grad_norm": 0.2860405147075653,
+      "learning_rate": 5.370753680003627e-05,
+      "loss": 0.247,
+      "step": 2308
+    },
+    {
+      "epoch": 0.4876451953537487,
+      "grad_norm": 0.38418474793434143,
+      "learning_rate": 5.367373935994109e-05,
+      "loss": 0.2493,
+      "step": 2309
+    },
+    {
+      "epoch": 0.48785638859556496,
+      "grad_norm": 0.3299623429775238,
+      "learning_rate": 5.3639940232094045e-05,
+      "loss": 0.2508,
+      "step": 2310
+    },
+    {
+      "epoch": 0.4880675818373812,
+      "grad_norm": 0.31489482522010803,
+      "learning_rate": 5.360613943202281e-05,
+      "loss": 0.2465,
+      "step": 2311
+    },
+    {
+      "epoch": 0.4882787750791975,
+      "grad_norm": 0.38932004570961,
+      "learning_rate": 5.3572336975255786e-05,
+      "loss": 0.2481,
+      "step": 2312
+    },
+    {
+      "epoch": 0.4884899683210137,
+      "grad_norm": 0.32240012288093567,
+      "learning_rate": 5.3538532877322144e-05,
+      "loss": 0.2479,
+      "step": 2313
+    },
+    {
+      "epoch": 0.48870116156283,
+      "grad_norm": 0.3596841096878052,
+      "learning_rate": 5.3504727153751865e-05,
+      "loss": 0.2452,
+      "step": 2314
+    },
+    {
+      "epoch": 0.48891235480464623,
+      "grad_norm": 0.355118989944458,
+      "learning_rate": 5.347091982007557e-05,
+      "loss": 0.2484,
+      "step": 2315
+    },
+    {
+      "epoch": 0.4891235480464625,
+      "grad_norm": 0.3053291440010071,
+      "learning_rate": 5.3437110891824725e-05,
+      "loss": 0.243,
+      "step": 2316
+    },
+    {
+      "epoch": 0.48933474128827875,
+      "grad_norm": 0.3202095031738281,
+      "learning_rate": 5.340330038453146e-05,
+      "loss": 0.2451,
+      "step": 2317
+    },
+    {
+      "epoch": 0.48954593453009504,
+      "grad_norm": 0.3074049651622772,
+      "learning_rate": 5.3369488313728656e-05,
+      "loss": 0.2474,
+      "step": 2318
+    },
+    {
+      "epoch": 0.48975712777191127,
+      "grad_norm": 0.3481866419315338,
+      "learning_rate": 5.333567469494991e-05,
+      "loss": 0.2458,
+      "step": 2319
+    },
+    {
+      "epoch": 0.48996832101372756,
+      "grad_norm": 0.3430945575237274,
+      "learning_rate": 5.330185954372955e-05,
+      "loss": 0.2433,
+      "step": 2320
+    },
+    {
+      "epoch": 0.49017951425554385,
+      "grad_norm": 0.31485798954963684,
+      "learning_rate": 5.326804287560254e-05,
+      "loss": 0.249,
+      "step": 2321
+    },
+    {
+      "epoch": 0.4903907074973601,
+      "grad_norm": 0.30816638469696045,
+      "learning_rate": 5.323422470610463e-05,
+      "loss": 0.2538,
+      "step": 2322
+    },
+    {
+      "epoch": 0.49060190073917637,
+      "grad_norm": 0.3114059567451477,
+      "learning_rate": 5.320040505077222e-05,
+      "loss": 0.2487,
+      "step": 2323
+    },
+    {
+      "epoch": 0.4908130939809926,
+      "grad_norm": 0.320646196603775,
+      "learning_rate": 5.3166583925142355e-05,
+      "loss": 0.2434,
+      "step": 2324
+    },
+    {
+      "epoch": 0.4910242872228089,
+      "grad_norm": 0.5643219947814941,
+      "learning_rate": 5.3132761344752825e-05,
+      "loss": 0.2399,
+      "step": 2325
+    },
+    {
+      "epoch": 0.4912354804646251,
+      "grad_norm": 0.32922273874282837,
+      "learning_rate": 5.309893732514205e-05,
+      "loss": 0.2448,
+      "step": 2326
+    },
+    {
+      "epoch": 0.4914466737064414,
+      "grad_norm": 0.34154757857322693,
+      "learning_rate": 5.3065111881849096e-05,
+      "loss": 0.2473,
+      "step": 2327
+    },
+    {
+      "epoch": 0.49165786694825764,
+      "grad_norm": 0.34457600116729736,
+      "learning_rate": 5.303128503041372e-05,
+      "loss": 0.2446,
+      "step": 2328
+    },
+    {
+      "epoch": 0.4918690601900739,
+      "grad_norm": 0.31607744097709656,
+      "learning_rate": 5.299745678637631e-05,
+      "loss": 0.2481,
+      "step": 2329
+    },
+    {
+      "epoch": 0.49208025343189016,
+      "grad_norm": 0.33459317684173584,
+      "learning_rate": 5.296362716527788e-05,
+      "loss": 0.2515,
+      "step": 2330
+    },
+    {
+      "epoch": 0.49229144667370645,
+      "grad_norm": 0.434524267911911,
+      "learning_rate": 5.292979618266012e-05,
+      "loss": 0.2454,
+      "step": 2331
+    },
+    {
+      "epoch": 0.4925026399155227,
+      "grad_norm": 0.474443644285202,
+      "learning_rate": 5.289596385406527e-05,
+      "loss": 0.2421,
+      "step": 2332
+    },
+    {
+      "epoch": 0.49271383315733897,
+      "grad_norm": 0.47752588987350464,
+      "learning_rate": 5.286213019503625e-05,
+      "loss": 0.2478,
+      "step": 2333
+    },
+    {
+      "epoch": 0.49292502639915525,
+      "grad_norm": 0.4916783571243286,
+      "learning_rate": 5.2828295221116606e-05,
+      "loss": 0.2489,
+      "step": 2334
+    },
+    {
+      "epoch": 0.4931362196409715,
+      "grad_norm": 0.5210815668106079,
+      "learning_rate": 5.279445894785041e-05,
+      "loss": 0.2523,
+      "step": 2335
+    },
+    {
+      "epoch": 0.4933474128827878,
+      "grad_norm": 0.6040790677070618,
+      "learning_rate": 5.276062139078239e-05,
+      "loss": 0.2404,
+      "step": 2336
+    },
+    {
+      "epoch": 0.493558606124604,
+      "grad_norm": 0.5170353651046753,
+      "learning_rate": 5.272678256545788e-05,
+      "loss": 0.243,
+      "step": 2337
+    },
+    {
+      "epoch": 0.4937697993664203,
+      "grad_norm": 0.47273164987564087,
+      "learning_rate": 5.2692942487422744e-05,
+      "loss": 0.2432,
+      "step": 2338
+    },
+    {
+      "epoch": 0.4939809926082365,
+      "grad_norm": 0.38265612721443176,
+      "learning_rate": 5.2659101172223436e-05,
+      "loss": 0.2411,
+      "step": 2339
+    },
+    {
+      "epoch": 0.4941921858500528,
+      "grad_norm": 0.3589935898780823,
+      "learning_rate": 5.2625258635407e-05,
+      "loss": 0.2494,
+      "step": 2340
+    },
+    {
+      "epoch": 0.49440337909186904,
+      "grad_norm": 0.38947954773902893,
+      "learning_rate": 5.259141489252104e-05,
+      "loss": 0.2545,
+      "step": 2341
+    },
+    {
+      "epoch": 0.49461457233368533,
+      "grad_norm": 0.4424537122249603,
+      "learning_rate": 5.25575699591137e-05,
+      "loss": 0.2508,
+      "step": 2342
+    },
+    {
+      "epoch": 0.49482576557550156,
+      "grad_norm": 0.41179656982421875,
+      "learning_rate": 5.2523723850733644e-05,
+      "loss": 0.2505,
+      "step": 2343
+    },
+    {
+      "epoch": 0.49503695881731785,
+      "grad_norm": 0.4141855537891388,
+      "learning_rate": 5.248987658293014e-05,
+      "loss": 0.2493,
+      "step": 2344
+    },
+    {
+      "epoch": 0.4952481520591341,
+      "grad_norm": 0.33815762400627136,
+      "learning_rate": 5.2456028171252935e-05,
+      "loss": 0.2457,
+      "step": 2345
+    },
+    {
+      "epoch": 0.49545934530095037,
+      "grad_norm": 0.4981009066104889,
+      "learning_rate": 5.242217863125234e-05,
+      "loss": 0.2466,
+      "step": 2346
+    },
+    {
+      "epoch": 0.49567053854276666,
+      "grad_norm": 0.3640243709087372,
+      "learning_rate": 5.2388327978479133e-05,
+      "loss": 0.2469,
+      "step": 2347
+    },
+    {
+      "epoch": 0.4958817317845829,
+      "grad_norm": 0.40198618173599243,
+      "learning_rate": 5.2354476228484674e-05,
+      "loss": 0.2463,
+      "step": 2348
+    },
+    {
+      "epoch": 0.4960929250263992,
+      "grad_norm": 0.5164361596107483,
+      "learning_rate": 5.2320623396820765e-05,
+      "loss": 0.2552,
+      "step": 2349
+    },
+    {
+      "epoch": 0.4963041182682154,
+      "grad_norm": 0.7245284914970398,
+      "learning_rate": 5.228676949903973e-05,
+      "loss": 0.2527,
+      "step": 2350
+    },
+    {
+      "epoch": 0.4965153115100317,
+      "grad_norm": 0.3301011323928833,
+      "learning_rate": 5.225291455069438e-05,
+      "loss": 0.2405,
+      "step": 2351
+    },
+    {
+      "epoch": 0.49672650475184793,
+      "grad_norm": 0.41319355368614197,
+      "learning_rate": 5.221905856733803e-05,
+      "loss": 0.2445,
+      "step": 2352
+    },
+    {
+      "epoch": 0.4969376979936642,
+      "grad_norm": 0.36408841609954834,
+      "learning_rate": 5.218520156452442e-05,
+      "loss": 0.2542,
+      "step": 2353
+    },
+    {
+      "epoch": 0.49714889123548045,
+      "grad_norm": 0.34712719917297363,
+      "learning_rate": 5.215134355780781e-05,
+      "loss": 0.2484,
+      "step": 2354
+    },
+    {
+      "epoch": 0.49736008447729674,
+      "grad_norm": 0.33553117513656616,
+      "learning_rate": 5.211748456274291e-05,
+      "loss": 0.2419,
+      "step": 2355
+    },
+    {
+      "epoch": 0.49757127771911297,
+      "grad_norm": 0.3114912211894989,
+      "learning_rate": 5.208362459488483e-05,
+      "loss": 0.2446,
+      "step": 2356
+    },
+    {
+      "epoch": 0.49778247096092926,
+      "grad_norm": 0.4395773112773895,
+      "learning_rate": 5.204976366978922e-05,
+      "loss": 0.2421,
+      "step": 2357
+    },
+    {
+      "epoch": 0.4979936642027455,
+      "grad_norm": 0.30478546023368835,
+      "learning_rate": 5.201590180301209e-05,
+      "loss": 0.2529,
+      "step": 2358
+    },
+    {
+      "epoch": 0.4982048574445618,
+      "grad_norm": 0.48605501651763916,
+      "learning_rate": 5.1982039010109926e-05,
+      "loss": 0.2612,
+      "step": 2359
+    },
+    {
+      "epoch": 0.498416050686378,
+      "grad_norm": 0.3588312864303589,
+      "learning_rate": 5.1948175306639625e-05,
+      "loss": 0.2525,
+      "step": 2360
+    },
+    {
+      "epoch": 0.4986272439281943,
+      "grad_norm": 0.4551800489425659,
+      "learning_rate": 5.19143107081585e-05,
+      "loss": 0.2501,
+      "step": 2361
+    },
+    {
+      "epoch": 0.4988384371700106,
+      "grad_norm": 0.43389055132865906,
+      "learning_rate": 5.188044523022428e-05,
+      "loss": 0.2545,
+      "step": 2362
+    },
+    {
+      "epoch": 0.4990496304118268,
+      "grad_norm": 0.39127007126808167,
+      "learning_rate": 5.184657888839512e-05,
+      "loss": 0.252,
+      "step": 2363
+    },
+    {
+      "epoch": 0.4992608236536431,
+      "grad_norm": 0.4039541780948639,
+      "learning_rate": 5.181271169822951e-05,
+      "loss": 0.2572,
+      "step": 2364
+    },
+    {
+      "epoch": 0.49947201689545934,
+      "grad_norm": 2.5759263038635254,
+      "learning_rate": 5.177884367528637e-05,
+      "loss": 0.2488,
+      "step": 2365
+    },
+    {
+      "epoch": 0.4996832101372756,
+      "grad_norm": 0.4603480100631714,
+      "learning_rate": 5.174497483512506e-05,
+      "loss": 0.2521,
+      "step": 2366
+    },
+    {
+      "epoch": 0.49989440337909186,
+      "grad_norm": 0.3227546215057373,
+      "learning_rate": 5.17111051933052e-05,
+      "loss": 0.2483,
+      "step": 2367
+    },
+    {
+      "epoch": 0.5001055966209081,
+      "grad_norm": 0.4672887623310089,
+      "learning_rate": 5.1677234765386825e-05,
+      "loss": 0.2534,
+      "step": 2368
+    },
+    {
+      "epoch": 0.5003167898627244,
+      "grad_norm": 0.3087901771068573,
+      "learning_rate": 5.16433635669304e-05,
+      "loss": 0.2544,
+      "step": 2369
+    },
+    {
+      "epoch": 0.5005279831045406,
+      "grad_norm": 0.4238470792770386,
+      "learning_rate": 5.160949161349665e-05,
+      "loss": 0.2555,
+      "step": 2370
+    },
+    {
+      "epoch": 0.5007391763463569,
+      "grad_norm": 0.3120185136795044,
+      "learning_rate": 5.1575618920646674e-05,
+      "loss": 0.25,
+      "step": 2371
+    },
+    {
+      "epoch": 0.5009503695881732,
+      "grad_norm": 0.4243412911891937,
+      "learning_rate": 5.154174550394193e-05,
+      "loss": 0.2435,
+      "step": 2372
+    },
+    {
+      "epoch": 0.5011615628299895,
+      "grad_norm": 0.3605717420578003,
+      "learning_rate": 5.150787137894422e-05,
+      "loss": 0.2408,
+      "step": 2373
+    },
+    {
+      "epoch": 0.5013727560718056,
+      "grad_norm": 0.3556000590324402,
+      "learning_rate": 5.1473996561215595e-05,
+      "loss": 0.243,
+      "step": 2374
+    },
+    {
+      "epoch": 0.5015839493136219,
+      "grad_norm": 0.3789392113685608,
+      "learning_rate": 5.1440121066318526e-05,
+      "loss": 0.2453,
+      "step": 2375
+    },
+    {
+      "epoch": 0.5017951425554382,
+      "grad_norm": 0.29179126024246216,
+      "learning_rate": 5.140624490981575e-05,
+      "loss": 0.2474,
+      "step": 2376
+    },
+    {
+      "epoch": 0.5020063357972545,
+      "grad_norm": 0.3459119200706482,
+      "learning_rate": 5.1372368107270254e-05,
+      "loss": 0.2412,
+      "step": 2377
+    },
+    {
+      "epoch": 0.5022175290390708,
+      "grad_norm": 0.30347907543182373,
+      "learning_rate": 5.133849067424542e-05,
+      "loss": 0.2429,
+      "step": 2378
+    },
+    {
+      "epoch": 0.502428722280887,
+      "grad_norm": 0.303317129611969,
+      "learning_rate": 5.130461262630486e-05,
+      "loss": 0.2406,
+      "step": 2379
+    },
+    {
+      "epoch": 0.5026399155227033,
+      "grad_norm": 0.45248401165008545,
+      "learning_rate": 5.127073397901248e-05,
+      "loss": 0.2434,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5028511087645195,
+      "grad_norm": 0.3647536337375641,
+      "learning_rate": 5.123685474793245e-05,
+      "loss": 0.2583,
+      "step": 2381
+    },
+    {
+      "epoch": 0.5030623020063358,
+      "grad_norm": 0.31268417835235596,
+      "learning_rate": 5.1202974948629236e-05,
+      "loss": 0.2468,
+      "step": 2382
+    },
+    {
+      "epoch": 0.503273495248152,
+      "grad_norm": 0.33243227005004883,
+      "learning_rate": 5.1169094596667546e-05,
+      "loss": 0.2452,
+      "step": 2383
+    },
+    {
+      "epoch": 0.5034846884899683,
+      "grad_norm": 0.6423991918563843,
+      "learning_rate": 5.113521370761235e-05,
+      "loss": 0.25,
+      "step": 2384
+    },
+    {
+      "epoch": 0.5036958817317846,
+      "grad_norm": 0.35589542984962463,
+      "learning_rate": 5.110133229702886e-05,
+      "loss": 0.2419,
+      "step": 2385
+    },
+    {
+      "epoch": 0.5039070749736009,
+      "grad_norm": 0.3177073299884796,
+      "learning_rate": 5.1067450380482506e-05,
+      "loss": 0.2508,
+      "step": 2386
+    },
+    {
+      "epoch": 0.504118268215417,
+      "grad_norm": 0.3196389079093933,
+      "learning_rate": 5.103356797353901e-05,
+      "loss": 0.2466,
+      "step": 2387
+    },
+    {
+      "epoch": 0.5043294614572333,
+      "grad_norm": 0.3353049159049988,
+      "learning_rate": 5.099968509176425e-05,
+      "loss": 0.2434,
+      "step": 2388
+    },
+    {
+      "epoch": 0.5045406546990496,
+      "grad_norm": 0.34590771794319153,
+      "learning_rate": 5.096580175072436e-05,
+      "loss": 0.2508,
+      "step": 2389
+    },
+    {
+      "epoch": 0.5047518479408659,
+      "grad_norm": 0.39763325452804565,
+      "learning_rate": 5.093191796598571e-05,
+      "loss": 0.2456,
+      "step": 2390
+    },
+    {
+      "epoch": 0.5049630411826822,
+      "grad_norm": 0.5896880626678467,
+      "learning_rate": 5.089803375311479e-05,
+      "loss": 0.2378,
+      "step": 2391
+    },
+    {
+      "epoch": 0.5051742344244984,
+      "grad_norm": 0.4627149999141693,
+      "learning_rate": 5.086414912767837e-05,
+      "loss": 0.2471,
+      "step": 2392
+    },
+    {
+      "epoch": 0.5053854276663147,
+      "grad_norm": 0.7557931542396545,
+      "learning_rate": 5.0830264105243385e-05,
+      "loss": 0.2476,
+      "step": 2393
+    },
+    {
+      "epoch": 0.505596620908131,
+      "grad_norm": 0.5070100426673889,
+      "learning_rate": 5.079637870137693e-05,
+      "loss": 0.2444,
+      "step": 2394
+    },
+    {
+      "epoch": 0.5058078141499472,
+      "grad_norm": 0.5653350949287415,
+      "learning_rate": 5.07624929316463e-05,
+      "loss": 0.2523,
+      "step": 2395
+    },
+    {
+      "epoch": 0.5060190073917634,
+      "grad_norm": 0.3910912871360779,
+      "learning_rate": 5.072860681161895e-05,
+      "loss": 0.2488,
+      "step": 2396
+    },
+    {
+      "epoch": 0.5062302006335797,
+      "grad_norm": 0.3381254971027374,
+      "learning_rate": 5.069472035686249e-05,
+      "loss": 0.2429,
+      "step": 2397
+    },
+    {
+      "epoch": 0.506441393875396,
+      "grad_norm": 0.4876307547092438,
+      "learning_rate": 5.06608335829447e-05,
+      "loss": 0.2497,
+      "step": 2398
+    },
+    {
+      "epoch": 0.5066525871172123,
+      "grad_norm": 0.3360268771648407,
+      "learning_rate": 5.0626946505433485e-05,
+      "loss": 0.2436,
+      "step": 2399
+    },
+    {
+      "epoch": 0.5068637803590285,
+      "grad_norm": 0.343149870634079,
+      "learning_rate": 5.059305913989689e-05,
+      "loss": 0.2464,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5068637803590285,
+      "eval_train_split_loss": 0.3634510636329651,
+      "eval_train_split_runtime": 6.4148,
+      "eval_train_split_samples_per_second": 1.559,
+      "eval_train_split_steps_per_second": 0.156,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5070749736008447,
+      "grad_norm": 0.45569247007369995,
+      "learning_rate": 5.0559171501903135e-05,
+      "loss": 0.2476,
+      "step": 2401
+    },
+    {
+      "epoch": 0.507286166842661,
+      "grad_norm": 0.43286779522895813,
+      "learning_rate": 5.05252836070205e-05,
+      "loss": 0.2417,
+      "step": 2402
+    },
+    {
+      "epoch": 0.5074973600844773,
+      "grad_norm": 0.3838598430156708,
+      "learning_rate": 5.049139547081744e-05,
+      "loss": 0.2529,
+      "step": 2403
+    },
+    {
+      "epoch": 0.5077085533262936,
+      "grad_norm": 0.4620312750339508,
+      "learning_rate": 5.045750710886248e-05,
+      "loss": 0.252,
+      "step": 2404
+    },
+    {
+      "epoch": 0.5079197465681098,
+      "grad_norm": 0.3307584226131439,
+      "learning_rate": 5.042361853672428e-05,
+      "loss": 0.2421,
+      "step": 2405
+    },
+    {
+      "epoch": 0.5081309398099261,
+      "grad_norm": 0.42339271306991577,
+      "learning_rate": 5.038972976997156e-05,
+      "loss": 0.2356,
+      "step": 2406
+    },
+    {
+      "epoch": 0.5083421330517424,
+      "grad_norm": 0.4009271562099457,
+      "learning_rate": 5.0355840824173174e-05,
+      "loss": 0.2352,
+      "step": 2407
+    },
+    {
+      "epoch": 0.5085533262935586,
+      "grad_norm": 0.3505074083805084,
+      "learning_rate": 5.032195171489804e-05,
+      "loss": 0.2468,
+      "step": 2408
+    },
+    {
+      "epoch": 0.5087645195353748,
+      "grad_norm": 0.43215763568878174,
+      "learning_rate": 5.028806245771511e-05,
+      "loss": 0.2448,
+      "step": 2409
+    },
+    {
+      "epoch": 0.5089757127771911,
+      "grad_norm": 0.3861945569515228,
+      "learning_rate": 5.0254173068193474e-05,
+      "loss": 0.2526,
+      "step": 2410
+    },
+    {
+      "epoch": 0.5091869060190074,
+      "grad_norm": 0.3654235899448395,
+      "learning_rate": 5.022028356190226e-05,
+      "loss": 0.2457,
+      "step": 2411
+    },
+    {
+      "epoch": 0.5093980992608237,
+      "grad_norm": 0.2963882386684418,
+      "learning_rate": 5.0186393954410594e-05,
+      "loss": 0.2438,
+      "step": 2412
+    },
+    {
+      "epoch": 0.5096092925026399,
+      "grad_norm": 0.30946096777915955,
+      "learning_rate": 5.0152504261287725e-05,
+      "loss": 0.2404,
+      "step": 2413
+    },
+    {
+      "epoch": 0.5098204857444562,
+      "grad_norm": 0.3201937973499298,
+      "learning_rate": 5.011861449810291e-05,
+      "loss": 0.2441,
+      "step": 2414
+    },
+    {
+      "epoch": 0.5100316789862724,
+      "grad_norm": 0.36060380935668945,
+      "learning_rate": 5.008472468042543e-05,
+      "loss": 0.2422,
+      "step": 2415
+    },
+    {
+      "epoch": 0.5102428722280887,
+      "grad_norm": 0.30341824889183044,
+      "learning_rate": 5.005083482382461e-05,
+      "loss": 0.2435,
+      "step": 2416
+    },
+    {
+      "epoch": 0.510454065469905,
+      "grad_norm": 0.36605948209762573,
+      "learning_rate": 5.001694494386976e-05,
+      "loss": 0.2484,
+      "step": 2417
+    },
+    {
+      "epoch": 0.5106652587117212,
+      "grad_norm": 0.3294595181941986,
+      "learning_rate": 4.998305505613025e-05,
+      "loss": 0.2488,
+      "step": 2418
+    },
+    {
+      "epoch": 0.5108764519535375,
+      "grad_norm": 0.355433851480484,
+      "learning_rate": 4.9949165176175396e-05,
+      "loss": 0.2486,
+      "step": 2419
+    },
+    {
+      "epoch": 0.5110876451953538,
+      "grad_norm": 0.35228458046913147,
+      "learning_rate": 4.991527531957458e-05,
+      "loss": 0.2469,
+      "step": 2420
+    },
+    {
+      "epoch": 0.5112988384371701,
+      "grad_norm": 0.3305721879005432,
+      "learning_rate": 4.98813855018971e-05,
+      "loss": 0.2471,
+      "step": 2421
+    },
+    {
+      "epoch": 0.5115100316789862,
+      "grad_norm": 0.36095401644706726,
+      "learning_rate": 4.984749573871227e-05,
+      "loss": 0.2507,
+      "step": 2422
+    },
+    {
+      "epoch": 0.5117212249208025,
+      "grad_norm": 0.29242008924484253,
+      "learning_rate": 4.981360604558942e-05,
+      "loss": 0.2494,
+      "step": 2423
+    },
+    {
+      "epoch": 0.5119324181626188,
+      "grad_norm": 0.3420129716396332,
+      "learning_rate": 4.9779716438097764e-05,
+      "loss": 0.2472,
+      "step": 2424
+    },
+    {
+      "epoch": 0.5121436114044351,
+      "grad_norm": 0.27701663970947266,
+      "learning_rate": 4.9745826931806524e-05,
+      "loss": 0.2405,
+      "step": 2425
+    },
+    {
+      "epoch": 0.5123548046462513,
+      "grad_norm": 0.3403935134410858,
+      "learning_rate": 4.97119375422849e-05,
+      "loss": 0.2517,
+      "step": 2426
+    },
+    {
+      "epoch": 0.5125659978880676,
+      "grad_norm": 0.39167460799217224,
+      "learning_rate": 4.9678048285101985e-05,
+      "loss": 0.2416,
+      "step": 2427
+    },
+    {
+      "epoch": 0.5127771911298838,
+      "grad_norm": 0.31067195534706116,
+      "learning_rate": 4.9644159175826824e-05,
+      "loss": 0.2425,
+      "step": 2428
+    },
+    {
+      "epoch": 0.5129883843717001,
+      "grad_norm": 0.34942004084587097,
+      "learning_rate": 4.961027023002845e-05,
+      "loss": 0.2457,
+      "step": 2429
+    },
+    {
+      "epoch": 0.5131995776135164,
+      "grad_norm": 0.33815526962280273,
+      "learning_rate": 4.957638146327574e-05,
+      "loss": 0.2514,
+      "step": 2430
+    },
+    {
+      "epoch": 0.5134107708553326,
+      "grad_norm": 0.31134435534477234,
+      "learning_rate": 4.9542492891137526e-05,
+      "loss": 0.2505,
+      "step": 2431
+    },
+    {
+      "epoch": 0.5136219640971489,
+      "grad_norm": 0.30949246883392334,
+      "learning_rate": 4.950860452918257e-05,
+      "loss": 0.2461,
+      "step": 2432
+    },
+    {
+      "epoch": 0.5138331573389652,
+      "grad_norm": 0.2828996479511261,
+      "learning_rate": 4.94747163929795e-05,
+      "loss": 0.2454,
+      "step": 2433
+    },
+    {
+      "epoch": 0.5140443505807815,
+      "grad_norm": 0.3116971254348755,
+      "learning_rate": 4.9440828498096884e-05,
+      "loss": 0.2477,
+      "step": 2434
+    },
+    {
+      "epoch": 0.5142555438225976,
+      "grad_norm": 0.332029789686203,
+      "learning_rate": 4.940694086010312e-05,
+      "loss": 0.2436,
+      "step": 2435
+    },
+    {
+      "epoch": 0.5144667370644139,
+      "grad_norm": 0.3987499475479126,
+      "learning_rate": 4.937305349456652e-05,
+      "loss": 0.2483,
+      "step": 2436
+    },
+    {
+      "epoch": 0.5146779303062302,
+      "grad_norm": 0.3086245656013489,
+      "learning_rate": 4.933916641705532e-05,
+      "loss": 0.2446,
+      "step": 2437
+    },
+    {
+      "epoch": 0.5148891235480465,
+      "grad_norm": 0.3546656668186188,
+      "learning_rate": 4.930527964313752e-05,
+      "loss": 0.242,
+      "step": 2438
+    },
+    {
+      "epoch": 0.5151003167898627,
+      "grad_norm": 0.2891584038734436,
+      "learning_rate": 4.927139318838105e-05,
+      "loss": 0.2425,
+      "step": 2439
+    },
+    {
+      "epoch": 0.515311510031679,
+      "grad_norm": 0.26856672763824463,
+      "learning_rate": 4.923750706835371e-05,
+      "loss": 0.2428,
+      "step": 2440
+    },
+    {
+      "epoch": 0.5155227032734953,
+      "grad_norm": 0.3686404526233673,
+      "learning_rate": 4.920362129862308e-05,
+      "loss": 0.2417,
+      "step": 2441
+    },
+    {
+      "epoch": 0.5157338965153115,
+      "grad_norm": 0.3208087086677551,
+      "learning_rate": 4.916973589475662e-05,
+      "loss": 0.2495,
+      "step": 2442
+    },
+    {
+      "epoch": 0.5159450897571277,
+      "grad_norm": 0.29874497652053833,
+      "learning_rate": 4.9135850872321646e-05,
+      "loss": 0.2495,
+      "step": 2443
+    },
+    {
+      "epoch": 0.516156282998944,
+      "grad_norm": 0.3781653344631195,
+      "learning_rate": 4.910196624688523e-05,
+      "loss": 0.2433,
+      "step": 2444
+    },
+    {
+      "epoch": 0.5163674762407603,
+      "grad_norm": 0.2824157476425171,
+      "learning_rate": 4.9068082034014305e-05,
+      "loss": 0.2429,
+      "step": 2445
+    },
+    {
+      "epoch": 0.5165786694825766,
+      "grad_norm": 0.27436044812202454,
+      "learning_rate": 4.903419824927565e-05,
+      "loss": 0.2437,
+      "step": 2446
+    },
+    {
+      "epoch": 0.5167898627243929,
+      "grad_norm": 0.2979915738105774,
+      "learning_rate": 4.900031490823576e-05,
+      "loss": 0.2404,
+      "step": 2447
+    },
+    {
+      "epoch": 0.517001055966209,
+      "grad_norm": 0.29230400919914246,
+      "learning_rate": 4.8966432026461e-05,
+      "loss": 0.2401,
+      "step": 2448
+    },
+    {
+      "epoch": 0.5172122492080253,
+      "grad_norm": 0.33911338448524475,
+      "learning_rate": 4.893254961951751e-05,
+      "loss": 0.2399,
+      "step": 2449
+    },
+    {
+      "epoch": 0.5174234424498416,
+      "grad_norm": 0.2781420350074768,
+      "learning_rate": 4.889866770297116e-05,
+      "loss": 0.2457,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5176346356916579,
+      "grad_norm": 0.37592723965644836,
+      "learning_rate": 4.886478629238766e-05,
+      "loss": 0.2418,
+      "step": 2451
+    },
+    {
+      "epoch": 0.5178458289334741,
+      "grad_norm": 0.29291772842407227,
+      "learning_rate": 4.883090540333247e-05,
+      "loss": 0.2413,
+      "step": 2452
+    },
+    {
+      "epoch": 0.5180570221752904,
+      "grad_norm": 0.32546523213386536,
+      "learning_rate": 4.8797025051370775e-05,
+      "loss": 0.2419,
+      "step": 2453
+    },
+    {
+      "epoch": 0.5182682154171067,
+      "grad_norm": 0.30910831689834595,
+      "learning_rate": 4.876314525206756e-05,
+      "loss": 0.2429,
+      "step": 2454
+    },
+    {
+      "epoch": 0.518479408658923,
+      "grad_norm": 0.40409842133522034,
+      "learning_rate": 4.8729266020987553e-05,
+      "loss": 0.2471,
+      "step": 2455
+    },
+    {
+      "epoch": 0.5186906019007391,
+      "grad_norm": 0.3105809986591339,
+      "learning_rate": 4.869538737369515e-05,
+      "loss": 0.2459,
+      "step": 2456
+    },
+    {
+      "epoch": 0.5189017951425554,
+      "grad_norm": 0.31150949001312256,
+      "learning_rate": 4.8661509325754593e-05,
+      "loss": 0.2374,
+      "step": 2457
+    },
+    {
+      "epoch": 0.5191129883843717,
+      "grad_norm": 0.2947373390197754,
+      "learning_rate": 4.862763189272976e-05,
+      "loss": 0.2438,
+      "step": 2458
+    },
+    {
+      "epoch": 0.519324181626188,
+      "grad_norm": 0.32368719577789307,
+      "learning_rate": 4.859375509018426e-05,
+      "loss": 0.2458,
+      "step": 2459
+    },
+    {
+      "epoch": 0.5195353748680043,
+      "grad_norm": 0.25585466623306274,
+      "learning_rate": 4.855987893368148e-05,
+      "loss": 0.2408,
+      "step": 2460
+    },
+    {
+      "epoch": 0.5197465681098205,
+      "grad_norm": 0.43974122405052185,
+      "learning_rate": 4.85260034387844e-05,
+      "loss": 0.2441,
+      "step": 2461
+    },
+    {
+      "epoch": 0.5199577613516367,
+      "grad_norm": 0.3184865117073059,
+      "learning_rate": 4.849212862105579e-05,
+      "loss": 0.2452,
+      "step": 2462
+    },
+    {
+      "epoch": 0.520168954593453,
+      "grad_norm": 0.2931613326072693,
+      "learning_rate": 4.8458254496058073e-05,
+      "loss": 0.243,
+      "step": 2463
+    },
+    {
+      "epoch": 0.5203801478352693,
+      "grad_norm": 0.286590576171875,
+      "learning_rate": 4.842438107935333e-05,
+      "loss": 0.2456,
+      "step": 2464
+    },
+    {
+      "epoch": 0.5205913410770855,
+      "grad_norm": 0.3682754635810852,
+      "learning_rate": 4.8390508386503355e-05,
+      "loss": 0.2437,
+      "step": 2465
+    },
+    {
+      "epoch": 0.5208025343189018,
+      "grad_norm": 0.3016096353530884,
+      "learning_rate": 4.835663643306961e-05,
+      "loss": 0.2444,
+      "step": 2466
+    },
+    {
+      "epoch": 0.5210137275607181,
+      "grad_norm": 0.31683218479156494,
+      "learning_rate": 4.832276523461317e-05,
+      "loss": 0.2466,
+      "step": 2467
+    },
+    {
+      "epoch": 0.5212249208025344,
+      "grad_norm": 0.3059089481830597,
+      "learning_rate": 4.828889480669482e-05,
+      "loss": 0.2426,
+      "step": 2468
+    },
+    {
+      "epoch": 0.5214361140443505,
+      "grad_norm": 0.4980098605155945,
+      "learning_rate": 4.825502516487497e-05,
+      "loss": 0.2437,
+      "step": 2469
+    },
+    {
+      "epoch": 0.5216473072861668,
+      "grad_norm": 0.4343259632587433,
+      "learning_rate": 4.822115632471362e-05,
+      "loss": 0.2432,
+      "step": 2470
+    },
+    {
+      "epoch": 0.5218585005279831,
+      "grad_norm": 0.7923051714897156,
+      "learning_rate": 4.81872883017705e-05,
+      "loss": 0.2464,
+      "step": 2471
+    },
+    {
+      "epoch": 0.5220696937697994,
+      "grad_norm": 0.461615651845932,
+      "learning_rate": 4.8153421111604915e-05,
+      "loss": 0.2478,
+      "step": 2472
+    },
+    {
+      "epoch": 0.5222808870116157,
+      "grad_norm": 0.323148250579834,
+      "learning_rate": 4.811955476977572e-05,
+      "loss": 0.244,
+      "step": 2473
+    },
+    {
+      "epoch": 0.5224920802534319,
+      "grad_norm": 0.35718056559562683,
+      "learning_rate": 4.8085689291841505e-05,
+      "loss": 0.2435,
+      "step": 2474
+    },
+    {
+      "epoch": 0.5227032734952481,
+      "grad_norm": 0.36730629205703735,
+      "learning_rate": 4.80518246933604e-05,
+      "loss": 0.2453,
+      "step": 2475
+    },
+    {
+      "epoch": 0.5229144667370644,
+      "grad_norm": 0.33534738421440125,
+      "learning_rate": 4.8017960989890086e-05,
+      "loss": 0.245,
+      "step": 2476
+    },
+    {
+      "epoch": 0.5231256599788807,
+      "grad_norm": 0.3866141438484192,
+      "learning_rate": 4.798409819698792e-05,
+      "loss": 0.2424,
+      "step": 2477
+    },
+    {
+      "epoch": 0.5233368532206969,
+      "grad_norm": 0.35051649808883667,
+      "learning_rate": 4.7950236330210786e-05,
+      "loss": 0.2505,
+      "step": 2478
+    },
+    {
+      "epoch": 0.5235480464625132,
+      "grad_norm": 0.39917439222335815,
+      "learning_rate": 4.791637540511518e-05,
+      "loss": 0.2513,
+      "step": 2479
+    },
+    {
+      "epoch": 0.5237592397043295,
+      "grad_norm": 0.33124399185180664,
+      "learning_rate": 4.788251543725711e-05,
+      "loss": 0.2537,
+      "step": 2480
+    },
+    {
+      "epoch": 0.5239704329461458,
+      "grad_norm": 0.3661513328552246,
+      "learning_rate": 4.784865644219218e-05,
+      "loss": 0.2402,
+      "step": 2481
+    },
+    {
+      "epoch": 0.5241816261879619,
+      "grad_norm": 0.33978548645973206,
+      "learning_rate": 4.781479843547559e-05,
+      "loss": 0.2487,
+      "step": 2482
+    },
+    {
+      "epoch": 0.5243928194297782,
+      "grad_norm": 0.3663939833641052,
+      "learning_rate": 4.778094143266199e-05,
+      "loss": 0.2458,
+      "step": 2483
+    },
+    {
+      "epoch": 0.5246040126715945,
+      "grad_norm": 0.3182114064693451,
+      "learning_rate": 4.7747085449305626e-05,
+      "loss": 0.2473,
+      "step": 2484
+    },
+    {
+      "epoch": 0.5248152059134108,
+      "grad_norm": 0.29554444551467896,
+      "learning_rate": 4.771323050096028e-05,
+      "loss": 0.2413,
+      "step": 2485
+    },
+    {
+      "epoch": 0.5250263991552271,
+      "grad_norm": 0.3095877468585968,
+      "learning_rate": 4.767937660317926e-05,
+      "loss": 0.2369,
+      "step": 2486
+    },
+    {
+      "epoch": 0.5252375923970433,
+      "grad_norm": 0.3705909252166748,
+      "learning_rate": 4.764552377151534e-05,
+      "loss": 0.2459,
+      "step": 2487
+    },
+    {
+      "epoch": 0.5254487856388596,
+      "grad_norm": 0.3803170621395111,
+      "learning_rate": 4.761167202152087e-05,
+      "loss": 0.2366,
+      "step": 2488
+    },
+    {
+      "epoch": 0.5256599788806758,
+      "grad_norm": 0.40566372871398926,
+      "learning_rate": 4.757782136874769e-05,
+      "loss": 0.2401,
+      "step": 2489
+    },
+    {
+      "epoch": 0.5258711721224921,
+      "grad_norm": 0.3405367136001587,
+      "learning_rate": 4.7543971828747076e-05,
+      "loss": 0.2409,
+      "step": 2490
+    },
+    {
+      "epoch": 0.5260823653643083,
+      "grad_norm": 0.3618094325065613,
+      "learning_rate": 4.7510123417069876e-05,
+      "loss": 0.2406,
+      "step": 2491
+    },
+    {
+      "epoch": 0.5262935586061246,
+      "grad_norm": 0.2900538444519043,
+      "learning_rate": 4.747627614926636e-05,
+      "loss": 0.2468,
+      "step": 2492
+    },
+    {
+      "epoch": 0.5265047518479409,
+      "grad_norm": 0.30922356247901917,
+      "learning_rate": 4.744243004088632e-05,
+      "loss": 0.2432,
+      "step": 2493
+    },
+    {
+      "epoch": 0.5267159450897572,
+      "grad_norm": 0.35436689853668213,
+      "learning_rate": 4.740858510747897e-05,
+      "loss": 0.2498,
+      "step": 2494
+    },
+    {
+      "epoch": 0.5269271383315733,
+      "grad_norm": 0.2993603050708771,
+      "learning_rate": 4.7374741364593e-05,
+      "loss": 0.2449,
+      "step": 2495
+    },
+    {
+      "epoch": 0.5271383315733896,
+      "grad_norm": 0.3374149799346924,
+      "learning_rate": 4.7340898827776575e-05,
+      "loss": 0.2447,
+      "step": 2496
+    },
+    {
+      "epoch": 0.5273495248152059,
+      "grad_norm": 0.35211801528930664,
+      "learning_rate": 4.7307057512577274e-05,
+      "loss": 0.2433,
+      "step": 2497
+    },
+    {
+      "epoch": 0.5275607180570222,
+      "grad_norm": 0.2948056161403656,
+      "learning_rate": 4.7273217434542125e-05,
+      "loss": 0.2435,
+      "step": 2498
+    },
+    {
+      "epoch": 0.5277719112988385,
+      "grad_norm": 0.3133615553379059,
+      "learning_rate": 4.723937860921762e-05,
+      "loss": 0.2479,
+      "step": 2499
+    },
+    {
+      "epoch": 0.5279831045406547,
+      "grad_norm": 0.35012996196746826,
+      "learning_rate": 4.720554105214961e-05,
+      "loss": 0.2417,
+      "step": 2500
+    },
+    {
+      "epoch": 0.528194297782471,
+      "grad_norm": 0.3287217915058136,
+      "learning_rate": 4.7171704778883406e-05,
+      "loss": 0.2458,
+      "step": 2501
+    },
+    {
+      "epoch": 0.5284054910242872,
+      "grad_norm": 0.46845659613609314,
+      "learning_rate": 4.713786980496376e-05,
+      "loss": 0.2451,
+      "step": 2502
+    },
+    {
+      "epoch": 0.5286166842661035,
+      "grad_norm": 0.34826886653900146,
+      "learning_rate": 4.710403614593475e-05,
+      "loss": 0.2426,
+      "step": 2503
+    },
+    {
+      "epoch": 0.5288278775079197,
+      "grad_norm": 0.31869006156921387,
+      "learning_rate": 4.70702038173399e-05,
+      "loss": 0.2469,
+      "step": 2504
+    },
+    {
+      "epoch": 0.529039070749736,
+      "grad_norm": 0.2780133783817291,
+      "learning_rate": 4.703637283472213e-05,
+      "loss": 0.2443,
+      "step": 2505
+    },
+    {
+      "epoch": 0.5292502639915523,
+      "grad_norm": 0.41863563656806946,
+      "learning_rate": 4.7002543213623695e-05,
+      "loss": 0.2453,
+      "step": 2506
+    },
+    {
+      "epoch": 0.5294614572333686,
+      "grad_norm": 0.3727928698062897,
+      "learning_rate": 4.696871496958628e-05,
+      "loss": 0.2452,
+      "step": 2507
+    },
+    {
+      "epoch": 0.5296726504751847,
+      "grad_norm": 0.6247018575668335,
+      "learning_rate": 4.693488811815092e-05,
+      "loss": 0.249,
+      "step": 2508
+    },
+    {
+      "epoch": 0.529883843717001,
+      "grad_norm": 0.34229594469070435,
+      "learning_rate": 4.6901062674857956e-05,
+      "loss": 0.244,
+      "step": 2509
+    },
+    {
+      "epoch": 0.5300950369588173,
+      "grad_norm": 0.3541579246520996,
+      "learning_rate": 4.686723865524718e-05,
+      "loss": 0.2463,
+      "step": 2510
+    },
+    {
+      "epoch": 0.5303062302006336,
+      "grad_norm": 0.434471994638443,
+      "learning_rate": 4.683341607485765e-05,
+      "loss": 0.2436,
+      "step": 2511
+    },
+    {
+      "epoch": 0.5305174234424498,
+      "grad_norm": 0.322255402803421,
+      "learning_rate": 4.6799594949227786e-05,
+      "loss": 0.2453,
+      "step": 2512
+    },
+    {
+      "epoch": 0.5307286166842661,
+      "grad_norm": 0.27160724997520447,
+      "learning_rate": 4.676577529389538e-05,
+      "loss": 0.2378,
+      "step": 2513
+    },
+    {
+      "epoch": 0.5309398099260824,
+      "grad_norm": 0.3310839831829071,
+      "learning_rate": 4.673195712439747e-05,
+      "loss": 0.2423,
+      "step": 2514
+    },
+    {
+      "epoch": 0.5311510031678986,
+      "grad_norm": 0.32830238342285156,
+      "learning_rate": 4.669814045627046e-05,
+      "loss": 0.2434,
+      "step": 2515
+    },
+    {
+      "epoch": 0.5313621964097149,
+      "grad_norm": 0.3707089126110077,
+      "learning_rate": 4.66643253050501e-05,
+      "loss": 0.2442,
+      "step": 2516
+    },
+    {
+      "epoch": 0.5315733896515311,
+      "grad_norm": 0.41235464811325073,
+      "learning_rate": 4.6630511686271356e-05,
+      "loss": 0.246,
+      "step": 2517
+    },
+    {
+      "epoch": 0.5317845828933474,
+      "grad_norm": 0.41174259781837463,
+      "learning_rate": 4.6596699615468544e-05,
+      "loss": 0.2486,
+      "step": 2518
+    },
+    {
+      "epoch": 0.5319957761351637,
+      "grad_norm": 0.3674201965332031,
+      "learning_rate": 4.656288910817529e-05,
+      "loss": 0.2468,
+      "step": 2519
+    },
+    {
+      "epoch": 0.53220696937698,
+      "grad_norm": 0.3828747868537903,
+      "learning_rate": 4.652908017992443e-05,
+      "loss": 0.243,
+      "step": 2520
+    },
+    {
+      "epoch": 0.5324181626187962,
+      "grad_norm": 0.342459112405777,
+      "learning_rate": 4.6495272846248146e-05,
+      "loss": 0.243,
+      "step": 2521
+    },
+    {
+      "epoch": 0.5326293558606124,
+      "grad_norm": 0.33182433247566223,
+      "learning_rate": 4.646146712267787e-05,
+      "loss": 0.2465,
+      "step": 2522
+    },
+    {
+      "epoch": 0.5328405491024287,
+      "grad_norm": 0.3597458004951477,
+      "learning_rate": 4.6427663024744226e-05,
+      "loss": 0.2446,
+      "step": 2523
+    },
+    {
+      "epoch": 0.533051742344245,
+      "grad_norm": 0.37902188301086426,
+      "learning_rate": 4.6393860567977205e-05,
+      "loss": 0.2446,
+      "step": 2524
+    },
+    {
+      "epoch": 0.5332629355860612,
+      "grad_norm": 0.3423847258090973,
+      "learning_rate": 4.636005976790597e-05,
+      "loss": 0.2385,
+      "step": 2525
+    },
+    {
+      "epoch": 0.5334741288278775,
+      "grad_norm": 0.32062220573425293,
+      "learning_rate": 4.632626064005892e-05,
+      "loss": 0.2384,
+      "step": 2526
+    },
+    {
+      "epoch": 0.5336853220696938,
+      "grad_norm": 0.38496989011764526,
+      "learning_rate": 4.629246319996373e-05,
+      "loss": 0.2527,
+      "step": 2527
+    },
+    {
+      "epoch": 0.5338965153115101,
+      "grad_norm": 0.3216980993747711,
+      "learning_rate": 4.625866746314728e-05,
+      "loss": 0.2444,
+      "step": 2528
+    },
+    {
+      "epoch": 0.5341077085533263,
+      "grad_norm": 0.3538614809513092,
+      "learning_rate": 4.622487344513563e-05,
+      "loss": 0.2418,
+      "step": 2529
+    },
+    {
+      "epoch": 0.5343189017951425,
+      "grad_norm": 0.4617953896522522,
+      "learning_rate": 4.619108116145411e-05,
+      "loss": 0.2458,
+      "step": 2530
+    },
+    {
+      "epoch": 0.5345300950369588,
+      "grad_norm": 0.36078372597694397,
+      "learning_rate": 4.615729062762723e-05,
+      "loss": 0.2409,
+      "step": 2531
+    },
+    {
+      "epoch": 0.5347412882787751,
+      "grad_norm": 0.2768119275569916,
+      "learning_rate": 4.612350185917865e-05,
+      "loss": 0.2446,
+      "step": 2532
+    },
+    {
+      "epoch": 0.5349524815205914,
+      "grad_norm": 0.33572012186050415,
+      "learning_rate": 4.608971487163131e-05,
+      "loss": 0.241,
+      "step": 2533
+    },
+    {
+      "epoch": 0.5351636747624076,
+      "grad_norm": 0.4141598045825958,
+      "learning_rate": 4.605592968050725e-05,
+      "loss": 0.2331,
+      "step": 2534
+    },
+    {
+      "epoch": 0.5353748680042238,
+      "grad_norm": 0.32303887605667114,
+      "learning_rate": 4.602214630132775e-05,
+      "loss": 0.2393,
+      "step": 2535
+    },
+    {
+      "epoch": 0.5355860612460401,
+      "grad_norm": 0.3237294554710388,
+      "learning_rate": 4.598836474961322e-05,
+      "loss": 0.2404,
+      "step": 2536
+    },
+    {
+      "epoch": 0.5357972544878564,
+      "grad_norm": 0.3987627625465393,
+      "learning_rate": 4.595458504088319e-05,
+      "loss": 0.2384,
+      "step": 2537
+    },
+    {
+      "epoch": 0.5360084477296726,
+      "grad_norm": 0.3130502700805664,
+      "learning_rate": 4.5920807190656446e-05,
+      "loss": 0.2375,
+      "step": 2538
+    },
+    {
+      "epoch": 0.5362196409714889,
+      "grad_norm": 0.3832043707370758,
+      "learning_rate": 4.588703121445084e-05,
+      "loss": 0.2412,
+      "step": 2539
+    },
+    {
+      "epoch": 0.5364308342133052,
+      "grad_norm": 0.3420877158641815,
+      "learning_rate": 4.585325712778337e-05,
+      "loss": 0.2435,
+      "step": 2540
+    },
+    {
+      "epoch": 0.5366420274551215,
+      "grad_norm": 0.7228314280509949,
+      "learning_rate": 4.581948494617021e-05,
+      "loss": 0.2327,
+      "step": 2541
+    },
+    {
+      "epoch": 0.5368532206969377,
+      "grad_norm": 0.38171806931495667,
+      "learning_rate": 4.578571468512663e-05,
+      "loss": 0.2403,
+      "step": 2542
+    },
+    {
+      "epoch": 0.5370644139387539,
+      "grad_norm": 0.33425435423851013,
+      "learning_rate": 4.5751946360166985e-05,
+      "loss": 0.239,
+      "step": 2543
+    },
+    {
+      "epoch": 0.5372756071805702,
+      "grad_norm": 0.31825241446495056,
+      "learning_rate": 4.5718179986804816e-05,
+      "loss": 0.2373,
+      "step": 2544
+    },
+    {
+      "epoch": 0.5374868004223865,
+      "grad_norm": 0.3877134621143341,
+      "learning_rate": 4.568441558055271e-05,
+      "loss": 0.2405,
+      "step": 2545
+    },
+    {
+      "epoch": 0.5376979936642028,
+      "grad_norm": 0.39703503251075745,
+      "learning_rate": 4.5650653156922345e-05,
+      "loss": 0.2437,
+      "step": 2546
+    },
+    {
+      "epoch": 0.537909186906019,
+      "grad_norm": 0.5174095630645752,
+      "learning_rate": 4.5616892731424536e-05,
+      "loss": 0.2446,
+      "step": 2547
+    },
+    {
+      "epoch": 0.5381203801478353,
+      "grad_norm": 0.35321295261383057,
+      "learning_rate": 4.558313431956914e-05,
+      "loss": 0.2509,
+      "step": 2548
+    },
+    {
+      "epoch": 0.5383315733896515,
+      "grad_norm": 0.4709426760673523,
+      "learning_rate": 4.554937793686512e-05,
+      "loss": 0.2469,
+      "step": 2549
+    },
+    {
+      "epoch": 0.5385427666314678,
+      "grad_norm": 0.3242552876472473,
+      "learning_rate": 4.551562359882047e-05,
+      "loss": 0.2456,
+      "step": 2550
+    },
+    {
+      "epoch": 0.538753959873284,
+      "grad_norm": 0.34126734733581543,
+      "learning_rate": 4.548187132094226e-05,
+      "loss": 0.2418,
+      "step": 2551
+    },
+    {
+      "epoch": 0.5389651531151003,
+      "grad_norm": 0.4411517083644867,
+      "learning_rate": 4.5448121118736645e-05,
+      "loss": 0.2398,
+      "step": 2552
+    },
+    {
+      "epoch": 0.5391763463569166,
+      "grad_norm": 0.2990863621234894,
+      "learning_rate": 4.541437300770877e-05,
+      "loss": 0.2398,
+      "step": 2553
+    },
+    {
+      "epoch": 0.5393875395987329,
+      "grad_norm": 0.4128507971763611,
+      "learning_rate": 4.5380627003362854e-05,
+      "loss": 0.2452,
+      "step": 2554
+    },
+    {
+      "epoch": 0.5395987328405492,
+      "grad_norm": 0.3103533983230591,
+      "learning_rate": 4.534688312120215e-05,
+      "loss": 0.2435,
+      "step": 2555
+    },
+    {
+      "epoch": 0.5398099260823653,
+      "grad_norm": 0.3320658206939697,
+      "learning_rate": 4.531314137672893e-05,
+      "loss": 0.2378,
+      "step": 2556
+    },
+    {
+      "epoch": 0.5400211193241816,
+      "grad_norm": 0.2993775010108948,
+      "learning_rate": 4.527940178544446e-05,
+      "loss": 0.2396,
+      "step": 2557
+    },
+    {
+      "epoch": 0.5402323125659979,
+      "grad_norm": 0.3427269160747528,
+      "learning_rate": 4.5245664362849075e-05,
+      "loss": 0.2408,
+      "step": 2558
+    },
+    {
+      "epoch": 0.5404435058078142,
+      "grad_norm": 0.9816110134124756,
+      "learning_rate": 4.521192912444205e-05,
+      "loss": 0.2399,
+      "step": 2559
+    },
+    {
+      "epoch": 0.5406546990496304,
+      "grad_norm": 0.3890838623046875,
+      "learning_rate": 4.5178196085721675e-05,
+      "loss": 0.2423,
+      "step": 2560
+    },
+    {
+      "epoch": 0.5408658922914467,
+      "grad_norm": 0.4922562837600708,
+      "learning_rate": 4.5144465262185265e-05,
+      "loss": 0.2446,
+      "step": 2561
+    },
+    {
+      "epoch": 0.541077085533263,
+      "grad_norm": 0.37733617424964905,
+      "learning_rate": 4.5110736669329074e-05,
+      "loss": 0.2432,
+      "step": 2562
+    },
+    {
+      "epoch": 0.5412882787750792,
+      "grad_norm": 0.36305588483810425,
+      "learning_rate": 4.507701032264837e-05,
+      "loss": 0.235,
+      "step": 2563
+    },
+    {
+      "epoch": 0.5414994720168954,
+      "grad_norm": 0.3201233744621277,
+      "learning_rate": 4.504328623763735e-05,
+      "loss": 0.2378,
+      "step": 2564
+    },
+    {
+      "epoch": 0.5417106652587117,
+      "grad_norm": 0.432666152715683,
+      "learning_rate": 4.500956442978918e-05,
+      "loss": 0.2411,
+      "step": 2565
+    },
+    {
+      "epoch": 0.541921858500528,
+      "grad_norm": 0.30984553694725037,
+      "learning_rate": 4.497584491459601e-05,
+      "loss": 0.2458,
+      "step": 2566
+    },
+    {
+      "epoch": 0.5421330517423443,
+      "grad_norm": 0.28294825553894043,
+      "learning_rate": 4.49421277075489e-05,
+      "loss": 0.2409,
+      "step": 2567
+    },
+    {
+      "epoch": 0.5423442449841606,
+      "grad_norm": 0.3325209617614746,
+      "learning_rate": 4.4908412824137855e-05,
+      "loss": 0.2398,
+      "step": 2568
+    },
+    {
+      "epoch": 0.5425554382259767,
+      "grad_norm": 0.3291946351528168,
+      "learning_rate": 4.487470027985186e-05,
+      "loss": 0.2432,
+      "step": 2569
+    },
+    {
+      "epoch": 0.542766631467793,
+      "grad_norm": 0.3128783404827118,
+      "learning_rate": 4.484099009017876e-05,
+      "loss": 0.2412,
+      "step": 2570
+    },
+    {
+      "epoch": 0.5429778247096093,
+      "grad_norm": 0.38186731934547424,
+      "learning_rate": 4.480728227060534e-05,
+      "loss": 0.2434,
+      "step": 2571
+    },
+    {
+      "epoch": 0.5431890179514256,
+      "grad_norm": 0.3452456593513489,
+      "learning_rate": 4.477357683661734e-05,
+      "loss": 0.2387,
+      "step": 2572
+    },
+    {
+      "epoch": 0.5434002111932418,
+      "grad_norm": 0.3313617706298828,
+      "learning_rate": 4.4739873803699326e-05,
+      "loss": 0.2408,
+      "step": 2573
+    },
+    {
+      "epoch": 0.5436114044350581,
+      "grad_norm": 0.3829326629638672,
+      "learning_rate": 4.4706173187334796e-05,
+      "loss": 0.2439,
+      "step": 2574
+    },
+    {
+      "epoch": 0.5438225976768744,
+      "grad_norm": 0.33880895376205444,
+      "learning_rate": 4.46724750030062e-05,
+      "loss": 0.2396,
+      "step": 2575
+    },
+    {
+      "epoch": 0.5440337909186906,
+      "grad_norm": 0.34759393334388733,
+      "learning_rate": 4.463877926619476e-05,
+      "loss": 0.2417,
+      "step": 2576
+    },
+    {
+      "epoch": 0.5442449841605068,
+      "grad_norm": 0.30867862701416016,
+      "learning_rate": 4.4605085992380647e-05,
+      "loss": 0.2502,
+      "step": 2577
+    },
+    {
+      "epoch": 0.5444561774023231,
+      "grad_norm": 0.3382805287837982,
+      "learning_rate": 4.4571395197042894e-05,
+      "loss": 0.2571,
+      "step": 2578
+    },
+    {
+      "epoch": 0.5446673706441394,
+      "grad_norm": 0.3039363622665405,
+      "learning_rate": 4.453770689565934e-05,
+      "loss": 0.2371,
+      "step": 2579
+    },
+    {
+      "epoch": 0.5448785638859557,
+      "grad_norm": 0.33264318108558655,
+      "learning_rate": 4.450402110370677e-05,
+      "loss": 0.2437,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5450897571277719,
+      "grad_norm": 0.34276655316352844,
+      "learning_rate": 4.4470337836660746e-05,
+      "loss": 0.2398,
+      "step": 2581
+    },
+    {
+      "epoch": 0.5453009503695881,
+      "grad_norm": 0.36068665981292725,
+      "learning_rate": 4.4436657109995675e-05,
+      "loss": 0.243,
+      "step": 2582
+    },
+    {
+      "epoch": 0.5455121436114044,
+      "grad_norm": 0.4168225824832916,
+      "learning_rate": 4.440297893918484e-05,
+      "loss": 0.2472,
+      "step": 2583
+    },
+    {
+      "epoch": 0.5457233368532207,
+      "grad_norm": 0.34001675248146057,
+      "learning_rate": 4.4369303339700324e-05,
+      "loss": 0.2381,
+      "step": 2584
+    },
+    {
+      "epoch": 0.545934530095037,
+      "grad_norm": 0.3411628305912018,
+      "learning_rate": 4.433563032701298e-05,
+      "loss": 0.2346,
+      "step": 2585
+    },
+    {
+      "epoch": 0.5461457233368532,
+      "grad_norm": 0.2747872769832611,
+      "learning_rate": 4.4301959916592585e-05,
+      "loss": 0.2396,
+      "step": 2586
+    },
+    {
+      "epoch": 0.5463569165786695,
+      "grad_norm": 0.3062359690666199,
+      "learning_rate": 4.426829212390762e-05,
+      "loss": 0.2381,
+      "step": 2587
+    },
+    {
+      "epoch": 0.5465681098204858,
+      "grad_norm": 0.31404098868370056,
+      "learning_rate": 4.423462696442538e-05,
+      "loss": 0.248,
+      "step": 2588
+    },
+    {
+      "epoch": 0.546779303062302,
+      "grad_norm": 0.39777475595474243,
+      "learning_rate": 4.420096445361202e-05,
+      "loss": 0.2447,
+      "step": 2589
+    },
+    {
+      "epoch": 0.5469904963041182,
+      "grad_norm": 0.29142963886260986,
+      "learning_rate": 4.416730460693239e-05,
+      "loss": 0.2396,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5472016895459345,
+      "grad_norm": 0.3645882308483124,
+      "learning_rate": 4.413364743985015e-05,
+      "loss": 0.244,
+      "step": 2591
+    },
+    {
+      "epoch": 0.5474128827877508,
+      "grad_norm": 0.3377965986728668,
+      "learning_rate": 4.4099992967827766e-05,
+      "loss": 0.2364,
+      "step": 2592
+    },
+    {
+      "epoch": 0.5476240760295671,
+      "grad_norm": 0.35237956047058105,
+      "learning_rate": 4.406634120632639e-05,
+      "loss": 0.2392,
+      "step": 2593
+    },
+    {
+      "epoch": 0.5478352692713833,
+      "grad_norm": 0.2971462309360504,
+      "learning_rate": 4.403269217080601e-05,
+      "loss": 0.2405,
+      "step": 2594
+    },
+    {
+      "epoch": 0.5480464625131996,
+      "grad_norm": 0.35137203335762024,
+      "learning_rate": 4.3999045876725314e-05,
+      "loss": 0.2376,
+      "step": 2595
+    },
+    {
+      "epoch": 0.5482576557550158,
+      "grad_norm": 0.335537850856781,
+      "learning_rate": 4.3965402339541705e-05,
+      "loss": 0.2379,
+      "step": 2596
+    },
+    {
+      "epoch": 0.5484688489968321,
+      "grad_norm": 0.3896923065185547,
+      "learning_rate": 4.3931761574711386e-05,
+      "loss": 0.2456,
+      "step": 2597
+    },
+    {
+      "epoch": 0.5486800422386484,
+      "grad_norm": 0.2995104491710663,
+      "learning_rate": 4.3898123597689265e-05,
+      "loss": 0.2436,
+      "step": 2598
+    },
+    {
+      "epoch": 0.5488912354804646,
+      "grad_norm": 0.3067997395992279,
+      "learning_rate": 4.386448842392891e-05,
+      "loss": 0.2431,
+      "step": 2599
+    },
+    {
+      "epoch": 0.5491024287222809,
+      "grad_norm": 0.35092616081237793,
+      "learning_rate": 4.38308560688827e-05,
+      "loss": 0.2388,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5491024287222809,
+      "eval_train_split_loss": 0.31974321603775024,
+      "eval_train_split_runtime": 10.1393,
+      "eval_train_split_samples_per_second": 0.986,
+      "eval_train_split_steps_per_second": 0.099,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5493136219640972,
+      "grad_norm": 0.3173377513885498,
+      "learning_rate": 4.3797226548001646e-05,
+      "loss": 0.2521,
+      "step": 2601
+    },
+    {
+      "epoch": 0.5495248152059135,
+      "grad_norm": 0.30894333124160767,
+      "learning_rate": 4.3763599876735465e-05,
+      "loss": 0.2397,
+      "step": 2602
+    },
+    {
+      "epoch": 0.5497360084477296,
+      "grad_norm": 0.2927210330963135,
+      "learning_rate": 4.372997607053261e-05,
+      "loss": 0.2362,
+      "step": 2603
+    },
+    {
+      "epoch": 0.5499472016895459,
+      "grad_norm": 0.28799283504486084,
+      "learning_rate": 4.369635514484018e-05,
+      "loss": 0.2406,
+      "step": 2604
+    },
+    {
+      "epoch": 0.5501583949313622,
+      "grad_norm": 0.28456762433052063,
+      "learning_rate": 4.366273711510392e-05,
+      "loss": 0.2382,
+      "step": 2605
+    },
+    {
+      "epoch": 0.5503695881731785,
+      "grad_norm": 0.2770291268825531,
+      "learning_rate": 4.362912199676833e-05,
+      "loss": 0.2332,
+      "step": 2606
+    },
+    {
+      "epoch": 0.5505807814149947,
+      "grad_norm": 0.351909339427948,
+      "learning_rate": 4.35955098052765e-05,
+      "loss": 0.2413,
+      "step": 2607
+    },
+    {
+      "epoch": 0.550791974656811,
+      "grad_norm": 0.41633135080337524,
+      "learning_rate": 4.3561900556070225e-05,
+      "loss": 0.2414,
+      "step": 2608
+    },
+    {
+      "epoch": 0.5510031678986272,
+      "grad_norm": 0.31283244490623474,
+      "learning_rate": 4.352829426458989e-05,
+      "loss": 0.2463,
+      "step": 2609
+    },
+    {
+      "epoch": 0.5512143611404435,
+      "grad_norm": 0.5374879240989685,
+      "learning_rate": 4.349469094627456e-05,
+      "loss": 0.234,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5514255543822598,
+      "grad_norm": 0.3588460385799408,
+      "learning_rate": 4.346109061656195e-05,
+      "loss": 0.2399,
+      "step": 2611
+    },
+    {
+      "epoch": 0.551636747624076,
+      "grad_norm": 0.3367220461368561,
+      "learning_rate": 4.342749329088837e-05,
+      "loss": 0.2447,
+      "step": 2612
+    },
+    {
+      "epoch": 0.5518479408658923,
+      "grad_norm": 0.3001561164855957,
+      "learning_rate": 4.339389898468875e-05,
+      "loss": 0.2416,
+      "step": 2613
+    },
+    {
+      "epoch": 0.5520591341077086,
+      "grad_norm": 0.3334922790527344,
+      "learning_rate": 4.336030771339665e-05,
+      "loss": 0.2361,
+      "step": 2614
+    },
+    {
+      "epoch": 0.5522703273495249,
+      "grad_norm": 0.3377707600593567,
+      "learning_rate": 4.332671949244426e-05,
+      "loss": 0.2368,
+      "step": 2615
+    },
+    {
+      "epoch": 0.552481520591341,
+      "grad_norm": 0.28419122099876404,
+      "learning_rate": 4.3293134337262275e-05,
+      "loss": 0.2402,
+      "step": 2616
+    },
+    {
+      "epoch": 0.5526927138331573,
+      "grad_norm": 0.2962735891342163,
+      "learning_rate": 4.32595522632801e-05,
+      "loss": 0.2362,
+      "step": 2617
+    },
+    {
+      "epoch": 0.5529039070749736,
+      "grad_norm": 0.49866124987602234,
+      "learning_rate": 4.322597328592565e-05,
+      "loss": 0.2378,
+      "step": 2618
+    },
+    {
+      "epoch": 0.5531151003167899,
+      "grad_norm": 0.4965287446975708,
+      "learning_rate": 4.319239742062541e-05,
+      "loss": 0.2393,
+      "step": 2619
+    },
+    {
+      "epoch": 0.5533262935586061,
+      "grad_norm": 0.39480122923851013,
+      "learning_rate": 4.31588246828045e-05,
+      "loss": 0.2476,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5535374868004224,
+      "grad_norm": 0.3378801941871643,
+      "learning_rate": 4.3125255087886535e-05,
+      "loss": 0.2439,
+      "step": 2621
+    },
+    {
+      "epoch": 0.5537486800422387,
+      "grad_norm": 0.26894721388816833,
+      "learning_rate": 4.309168865129376e-05,
+      "loss": 0.2397,
+      "step": 2622
+    },
+    {
+      "epoch": 0.5539598732840549,
+      "grad_norm": 0.3291214406490326,
+      "learning_rate": 4.305812538844687e-05,
+      "loss": 0.2439,
+      "step": 2623
+    },
+    {
+      "epoch": 0.5541710665258712,
+      "grad_norm": 0.2929019629955292,
+      "learning_rate": 4.302456531476518e-05,
+      "loss": 0.2383,
+      "step": 2624
+    },
+    {
+      "epoch": 0.5543822597676874,
+      "grad_norm": 0.27708759903907776,
+      "learning_rate": 4.2991008445666545e-05,
+      "loss": 0.2366,
+      "step": 2625
+    },
+    {
+      "epoch": 0.5545934530095037,
+      "grad_norm": 0.2934189736843109,
+      "learning_rate": 4.295745479656728e-05,
+      "loss": 0.2327,
+      "step": 2626
+    },
+    {
+      "epoch": 0.55480464625132,
+      "grad_norm": 0.33838939666748047,
+      "learning_rate": 4.292390438288225e-05,
+      "loss": 0.2363,
+      "step": 2627
+    },
+    {
+      "epoch": 0.5550158394931363,
+      "grad_norm": 0.3508434295654297,
+      "learning_rate": 4.28903572200249e-05,
+      "loss": 0.2387,
+      "step": 2628
+    },
+    {
+      "epoch": 0.5552270327349524,
+      "grad_norm": 0.4824860990047455,
+      "learning_rate": 4.285681332340708e-05,
+      "loss": 0.2392,
+      "step": 2629
+    },
+    {
+      "epoch": 0.5554382259767687,
+      "grad_norm": 0.3451348543167114,
+      "learning_rate": 4.282327270843919e-05,
+      "loss": 0.2435,
+      "step": 2630
+    },
+    {
+      "epoch": 0.555649419218585,
+      "grad_norm": 0.30672183632850647,
+      "learning_rate": 4.278973539053014e-05,
+      "loss": 0.2378,
+      "step": 2631
+    },
+    {
+      "epoch": 0.5558606124604013,
+      "grad_norm": 0.3122811019420624,
+      "learning_rate": 4.275620138508727e-05,
+      "loss": 0.2358,
+      "step": 2632
+    },
+    {
+      "epoch": 0.5560718057022175,
+      "grad_norm": 0.45386746525764465,
+      "learning_rate": 4.2722670707516444e-05,
+      "loss": 0.2435,
+      "step": 2633
+    },
+    {
+      "epoch": 0.5562829989440338,
+      "grad_norm": 0.33855554461479187,
+      "learning_rate": 4.2689143373222004e-05,
+      "loss": 0.2428,
+      "step": 2634
+    },
+    {
+      "epoch": 0.5564941921858501,
+      "grad_norm": 0.30780982971191406,
+      "learning_rate": 4.26556193976067e-05,
+      "loss": 0.233,
+      "step": 2635
+    },
+    {
+      "epoch": 0.5567053854276663,
+      "grad_norm": 0.27518513798713684,
+      "learning_rate": 4.262209879607181e-05,
+      "loss": 0.2355,
+      "step": 2636
+    },
+    {
+      "epoch": 0.5569165786694826,
+      "grad_norm": 0.5082710385322571,
+      "learning_rate": 4.258858158401701e-05,
+      "loss": 0.2387,
+      "step": 2637
+    },
+    {
+      "epoch": 0.5571277719112988,
+      "grad_norm": 0.36925703287124634,
+      "learning_rate": 4.255506777684041e-05,
+      "loss": 0.2418,
+      "step": 2638
+    },
+    {
+      "epoch": 0.5573389651531151,
+      "grad_norm": 0.3392466902732849,
+      "learning_rate": 4.252155738993863e-05,
+      "loss": 0.2402,
+      "step": 2639
+    },
+    {
+      "epoch": 0.5575501583949314,
+      "grad_norm": 0.331324964761734,
+      "learning_rate": 4.248805043870665e-05,
+      "loss": 0.2389,
+      "step": 2640
+    },
+    {
+      "epoch": 0.5577613516367477,
+      "grad_norm": 0.3251309096813202,
+      "learning_rate": 4.2454546938537866e-05,
+      "loss": 0.2417,
+      "step": 2641
+    },
+    {
+      "epoch": 0.5579725448785638,
+      "grad_norm": 0.3235847055912018,
+      "learning_rate": 4.242104690482416e-05,
+      "loss": 0.2419,
+      "step": 2642
+    },
+    {
+      "epoch": 0.5581837381203801,
+      "grad_norm": 0.2962948977947235,
+      "learning_rate": 4.238755035295574e-05,
+      "loss": 0.2334,
+      "step": 2643
+    },
+    {
+      "epoch": 0.5583949313621964,
+      "grad_norm": 0.34954577684402466,
+      "learning_rate": 4.235405729832126e-05,
+      "loss": 0.239,
+      "step": 2644
+    },
+    {
+      "epoch": 0.5586061246040127,
+      "grad_norm": 0.6105573177337646,
+      "learning_rate": 4.2320567756307776e-05,
+      "loss": 0.2415,
+      "step": 2645
+    },
+    {
+      "epoch": 0.5588173178458289,
+      "grad_norm": 0.3513045608997345,
+      "learning_rate": 4.228708174230068e-05,
+      "loss": 0.2436,
+      "step": 2646
+    },
+    {
+      "epoch": 0.5590285110876452,
+      "grad_norm": 0.3588605225086212,
+      "learning_rate": 4.2253599271683794e-05,
+      "loss": 0.2454,
+      "step": 2647
+    },
+    {
+      "epoch": 0.5592397043294615,
+      "grad_norm": 0.2586362659931183,
+      "learning_rate": 4.22201203598393e-05,
+      "loss": 0.239,
+      "step": 2648
+    },
+    {
+      "epoch": 0.5594508975712777,
+      "grad_norm": 0.3491639792919159,
+      "learning_rate": 4.2186645022147706e-05,
+      "loss": 0.2415,
+      "step": 2649
+    },
+    {
+      "epoch": 0.5596620908130939,
+      "grad_norm": 0.26931047439575195,
+      "learning_rate": 4.2153173273987946e-05,
+      "loss": 0.2374,
+      "step": 2650
+    },
+    {
+      "epoch": 0.5598732840549102,
+      "grad_norm": 0.4281803369522095,
+      "learning_rate": 4.211970513073726e-05,
+      "loss": 0.2399,
+      "step": 2651
+    },
+    {
+      "epoch": 0.5600844772967265,
+      "grad_norm": 0.27818337082862854,
+      "learning_rate": 4.208624060777122e-05,
+      "loss": 0.2368,
+      "step": 2652
+    },
+    {
+      "epoch": 0.5602956705385428,
+      "grad_norm": 0.36896800994873047,
+      "learning_rate": 4.205277972046378e-05,
+      "loss": 0.2412,
+      "step": 2653
+    },
+    {
+      "epoch": 0.5605068637803591,
+      "grad_norm": 0.4811095893383026,
+      "learning_rate": 4.20193224841872e-05,
+      "loss": 0.2436,
+      "step": 2654
+    },
+    {
+      "epoch": 0.5607180570221753,
+      "grad_norm": 0.3735496997833252,
+      "learning_rate": 4.1985868914312035e-05,
+      "loss": 0.2381,
+      "step": 2655
+    },
+    {
+      "epoch": 0.5609292502639915,
+      "grad_norm": 0.3142243027687073,
+      "learning_rate": 4.1952419026207207e-05,
+      "loss": 0.2396,
+      "step": 2656
+    },
+    {
+      "epoch": 0.5611404435058078,
+      "grad_norm": 0.3057881295681,
+      "learning_rate": 4.191897283523993e-05,
+      "loss": 0.2362,
+      "step": 2657
+    },
+    {
+      "epoch": 0.5613516367476241,
+      "grad_norm": 0.32123610377311707,
+      "learning_rate": 4.188553035677567e-05,
+      "loss": 0.2418,
+      "step": 2658
+    },
+    {
+      "epoch": 0.5615628299894403,
+      "grad_norm": 0.4047769010066986,
+      "learning_rate": 4.1852091606178276e-05,
+      "loss": 0.2408,
+      "step": 2659
+    },
+    {
+      "epoch": 0.5617740232312566,
+      "grad_norm": 0.29227349162101746,
+      "learning_rate": 4.181865659880982e-05,
+      "loss": 0.2393,
+      "step": 2660
+    },
+    {
+      "epoch": 0.5619852164730729,
+      "grad_norm": 0.32578834891319275,
+      "learning_rate": 4.1785225350030656e-05,
+      "loss": 0.237,
+      "step": 2661
+    },
+    {
+      "epoch": 0.5621964097148892,
+      "grad_norm": 0.27291789650917053,
+      "learning_rate": 4.175179787519944e-05,
+      "loss": 0.239,
+      "step": 2662
+    },
+    {
+      "epoch": 0.5624076029567053,
+      "grad_norm": 0.3171214461326599,
+      "learning_rate": 4.1718374189673085e-05,
+      "loss": 0.2423,
+      "step": 2663
+    },
+    {
+      "epoch": 0.5626187961985216,
+      "grad_norm": 0.28598618507385254,
+      "learning_rate": 4.1684954308806765e-05,
+      "loss": 0.2386,
+      "step": 2664
+    },
+    {
+      "epoch": 0.5628299894403379,
+      "grad_norm": 0.3515963554382324,
+      "learning_rate": 4.16515382479539e-05,
+      "loss": 0.2408,
+      "step": 2665
+    },
+    {
+      "epoch": 0.5630411826821542,
+      "grad_norm": 0.35575324296951294,
+      "learning_rate": 4.161812602246613e-05,
+      "loss": 0.2423,
+      "step": 2666
+    },
+    {
+      "epoch": 0.5632523759239705,
+      "grad_norm": 1.021469235420227,
+      "learning_rate": 4.1584717647693394e-05,
+      "loss": 0.2416,
+      "step": 2667
+    },
+    {
+      "epoch": 0.5634635691657867,
+      "grad_norm": 0.29505306482315063,
+      "learning_rate": 4.1551313138983814e-05,
+      "loss": 0.2403,
+      "step": 2668
+    },
+    {
+      "epoch": 0.563674762407603,
+      "grad_norm": 0.2982150614261627,
+      "learning_rate": 4.1517912511683724e-05,
+      "loss": 0.2396,
+      "step": 2669
+    },
+    {
+      "epoch": 0.5638859556494192,
+      "grad_norm": 0.3359386920928955,
+      "learning_rate": 4.1484515781137735e-05,
+      "loss": 0.2408,
+      "step": 2670
+    },
+    {
+      "epoch": 0.5640971488912355,
+      "grad_norm": 0.28253883123397827,
+      "learning_rate": 4.145112296268862e-05,
+      "loss": 0.2364,
+      "step": 2671
+    },
+    {
+      "epoch": 0.5643083421330517,
+      "grad_norm": 0.32391294836997986,
+      "learning_rate": 4.141773407167735e-05,
+      "loss": 0.2408,
+      "step": 2672
+    },
+    {
+      "epoch": 0.564519535374868,
+      "grad_norm": 0.2761439383029938,
+      "learning_rate": 4.138434912344313e-05,
+      "loss": 0.24,
+      "step": 2673
+    },
+    {
+      "epoch": 0.5647307286166843,
+      "grad_norm": 0.30525922775268555,
+      "learning_rate": 4.135096813332333e-05,
+      "loss": 0.2393,
+      "step": 2674
+    },
+    {
+      "epoch": 0.5649419218585006,
+      "grad_norm": 0.3637843430042267,
+      "learning_rate": 4.131759111665349e-05,
+      "loss": 0.2436,
+      "step": 2675
+    },
+    {
+      "epoch": 0.5651531151003167,
+      "grad_norm": 0.28709015250205994,
+      "learning_rate": 4.128421808876735e-05,
+      "loss": 0.2482,
+      "step": 2676
+    },
+    {
+      "epoch": 0.565364308342133,
+      "grad_norm": 0.32162970304489136,
+      "learning_rate": 4.125084906499682e-05,
+      "loss": 0.2372,
+      "step": 2677
+    },
+    {
+      "epoch": 0.5655755015839493,
+      "grad_norm": 0.30411937832832336,
+      "learning_rate": 4.1217484060671905e-05,
+      "loss": 0.2368,
+      "step": 2678
+    },
+    {
+      "epoch": 0.5657866948257656,
+      "grad_norm": 0.3315877318382263,
+      "learning_rate": 4.118412309112087e-05,
+      "loss": 0.232,
+      "step": 2679
+    },
+    {
+      "epoch": 0.5659978880675819,
+      "grad_norm": 0.3685995638370514,
+      "learning_rate": 4.1150766171670044e-05,
+      "loss": 0.2354,
+      "step": 2680
+    },
+    {
+      "epoch": 0.5662090813093981,
+      "grad_norm": 0.2819352149963379,
+      "learning_rate": 4.1117413317643944e-05,
+      "loss": 0.2361,
+      "step": 2681
+    },
+    {
+      "epoch": 0.5664202745512144,
+      "grad_norm": 0.2675110101699829,
+      "learning_rate": 4.1084064544365175e-05,
+      "loss": 0.2347,
+      "step": 2682
+    },
+    {
+      "epoch": 0.5666314677930306,
+      "grad_norm": 0.28310370445251465,
+      "learning_rate": 4.105071986715449e-05,
+      "loss": 0.2424,
+      "step": 2683
+    },
+    {
+      "epoch": 0.5668426610348469,
+      "grad_norm": 0.3276199400424957,
+      "learning_rate": 4.1017379301330786e-05,
+      "loss": 0.2406,
+      "step": 2684
+    },
+    {
+      "epoch": 0.5670538542766631,
+      "grad_norm": 0.3665010929107666,
+      "learning_rate": 4.098404286221101e-05,
+      "loss": 0.2397,
+      "step": 2685
+    },
+    {
+      "epoch": 0.5672650475184794,
+      "grad_norm": 0.3483017683029175,
+      "learning_rate": 4.095071056511027e-05,
+      "loss": 0.2344,
+      "step": 2686
+    },
+    {
+      "epoch": 0.5674762407602957,
+      "grad_norm": 0.3165411353111267,
+      "learning_rate": 4.091738242534177e-05,
+      "loss": 0.2377,
+      "step": 2687
+    },
+    {
+      "epoch": 0.567687434002112,
+      "grad_norm": 0.3343142569065094,
+      "learning_rate": 4.088405845821674e-05,
+      "loss": 0.2418,
+      "step": 2688
+    },
+    {
+      "epoch": 0.5678986272439281,
+      "grad_norm": 0.32424667477607727,
+      "learning_rate": 4.085073867904455e-05,
+      "loss": 0.2395,
+      "step": 2689
+    },
+    {
+      "epoch": 0.5681098204857444,
+      "grad_norm": 0.31760403513908386,
+      "learning_rate": 4.0817423103132654e-05,
+      "loss": 0.2395,
+      "step": 2690
+    },
+    {
+      "epoch": 0.5683210137275607,
+      "grad_norm": 0.2995377480983734,
+      "learning_rate": 4.078411174578655e-05,
+      "loss": 0.2398,
+      "step": 2691
+    },
+    {
+      "epoch": 0.568532206969377,
+      "grad_norm": 0.3205535411834717,
+      "learning_rate": 4.075080462230977e-05,
+      "loss": 0.2367,
+      "step": 2692
+    },
+    {
+      "epoch": 0.5687434002111933,
+      "grad_norm": 0.27503702044487,
+      "learning_rate": 4.071750174800396e-05,
+      "loss": 0.2346,
+      "step": 2693
+    },
+    {
+      "epoch": 0.5689545934530095,
+      "grad_norm": 0.2985975444316864,
+      "learning_rate": 4.068420313816878e-05,
+      "loss": 0.2412,
+      "step": 2694
+    },
+    {
+      "epoch": 0.5691657866948258,
+      "grad_norm": 0.4202212989330292,
+      "learning_rate": 4.065090880810196e-05,
+      "loss": 0.24,
+      "step": 2695
+    },
+    {
+      "epoch": 0.569376979936642,
+      "grad_norm": 0.3040681481361389,
+      "learning_rate": 4.061761877309921e-05,
+      "loss": 0.2365,
+      "step": 2696
+    },
+    {
+      "epoch": 0.5695881731784583,
+      "grad_norm": 0.34884291887283325,
+      "learning_rate": 4.058433304845429e-05,
+      "loss": 0.2471,
+      "step": 2697
+    },
+    {
+      "epoch": 0.5697993664202745,
+      "grad_norm": 0.29586631059646606,
+      "learning_rate": 4.055105164945904e-05,
+      "loss": 0.2451,
+      "step": 2698
+    },
+    {
+      "epoch": 0.5700105596620908,
+      "grad_norm": 0.3705187737941742,
+      "learning_rate": 4.051777459140321e-05,
+      "loss": 0.2397,
+      "step": 2699
+    },
+    {
+      "epoch": 0.5702217529039071,
+      "grad_norm": 0.4128904342651367,
+      "learning_rate": 4.048450188957462e-05,
+      "loss": 0.2359,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5704329461457234,
+      "grad_norm": 0.6036512851715088,
+      "learning_rate": 4.04512335592591e-05,
+      "loss": 0.2411,
+      "step": 2701
+    },
+    {
+      "epoch": 0.5706441393875396,
+      "grad_norm": 0.3121773898601532,
+      "learning_rate": 4.0417969615740416e-05,
+      "loss": 0.2356,
+      "step": 2702
+    },
+    {
+      "epoch": 0.5708553326293558,
+      "grad_norm": 0.2861095368862152,
+      "learning_rate": 4.038471007430035e-05,
+      "loss": 0.2414,
+      "step": 2703
+    },
+    {
+      "epoch": 0.5710665258711721,
+      "grad_norm": 0.36872532963752747,
+      "learning_rate": 4.035145495021868e-05,
+      "loss": 0.239,
+      "step": 2704
+    },
+    {
+      "epoch": 0.5712777191129884,
+      "grad_norm": 0.3073825240135193,
+      "learning_rate": 4.0318204258773126e-05,
+      "loss": 0.2452,
+      "step": 2705
+    },
+    {
+      "epoch": 0.5714889123548047,
+      "grad_norm": 0.2971263527870178,
+      "learning_rate": 4.028495801523937e-05,
+      "loss": 0.2481,
+      "step": 2706
+    },
+    {
+      "epoch": 0.5717001055966209,
+      "grad_norm": 0.29337817430496216,
+      "learning_rate": 4.02517162348911e-05,
+      "loss": 0.2417,
+      "step": 2707
+    },
+    {
+      "epoch": 0.5719112988384372,
+      "grad_norm": 0.3018752336502075,
+      "learning_rate": 4.021847893299987e-05,
+      "loss": 0.2349,
+      "step": 2708
+    },
+    {
+      "epoch": 0.5721224920802535,
+      "grad_norm": 0.3428551256656647,
+      "learning_rate": 4.018524612483524e-05,
+      "loss": 0.2395,
+      "step": 2709
+    },
+    {
+      "epoch": 0.5723336853220697,
+      "grad_norm": 0.4351382255554199,
+      "learning_rate": 4.015201782566471e-05,
+      "loss": 0.237,
+      "step": 2710
+    },
+    {
+      "epoch": 0.5725448785638859,
+      "grad_norm": 0.2916623055934906,
+      "learning_rate": 4.011879405075364e-05,
+      "loss": 0.2407,
+      "step": 2711
+    },
+    {
+      "epoch": 0.5727560718057022,
+      "grad_norm": 0.40970560908317566,
+      "learning_rate": 4.0085574815365404e-05,
+      "loss": 0.2369,
+      "step": 2712
+    },
+    {
+      "epoch": 0.5729672650475185,
+      "grad_norm": 0.4280383884906769,
+      "learning_rate": 4.005236013476122e-05,
+      "loss": 0.234,
+      "step": 2713
+    },
+    {
+      "epoch": 0.5731784582893348,
+      "grad_norm": 0.3927813470363617,
+      "learning_rate": 4.001915002420023e-05,
+      "loss": 0.2351,
+      "step": 2714
+    },
+    {
+      "epoch": 0.573389651531151,
+      "grad_norm": 0.3551713526248932,
+      "learning_rate": 3.99859444989395e-05,
+      "loss": 0.2391,
+      "step": 2715
+    },
+    {
+      "epoch": 0.5736008447729672,
+      "grad_norm": 0.33737626671791077,
+      "learning_rate": 3.995274357423398e-05,
+      "loss": 0.2491,
+      "step": 2716
+    },
+    {
+      "epoch": 0.5738120380147835,
+      "grad_norm": 0.46935224533081055,
+      "learning_rate": 3.991954726533646e-05,
+      "loss": 0.248,
+      "step": 2717
+    },
+    {
+      "epoch": 0.5740232312565998,
+      "grad_norm": 0.41138219833374023,
+      "learning_rate": 3.9886355587497704e-05,
+      "loss": 0.2457,
+      "step": 2718
+    },
+    {
+      "epoch": 0.574234424498416,
+      "grad_norm": 0.28828004002571106,
+      "learning_rate": 3.985316855596626e-05,
+      "loss": 0.2408,
+      "step": 2719
+    },
+    {
+      "epoch": 0.5744456177402323,
+      "grad_norm": 0.34511035680770874,
+      "learning_rate": 3.981998618598858e-05,
+      "loss": 0.2415,
+      "step": 2720
+    },
+    {
+      "epoch": 0.5746568109820486,
+      "grad_norm": 0.4258115589618683,
+      "learning_rate": 3.978680849280899e-05,
+      "loss": 0.2355,
+      "step": 2721
+    },
+    {
+      "epoch": 0.5748680042238649,
+      "grad_norm": 0.5778120756149292,
+      "learning_rate": 3.975363549166961e-05,
+      "loss": 0.2394,
+      "step": 2722
+    },
+    {
+      "epoch": 0.5750791974656811,
+      "grad_norm": 0.3180156648159027,
+      "learning_rate": 3.9720467197810474e-05,
+      "loss": 0.241,
+      "step": 2723
+    },
+    {
+      "epoch": 0.5752903907074973,
+      "grad_norm": 0.38081789016723633,
+      "learning_rate": 3.968730362646942e-05,
+      "loss": 0.237,
+      "step": 2724
+    },
+    {
+      "epoch": 0.5755015839493136,
+      "grad_norm": 0.3335338532924652,
+      "learning_rate": 3.965414479288209e-05,
+      "loss": 0.2395,
+      "step": 2725
+    },
+    {
+      "epoch": 0.5757127771911299,
+      "grad_norm": 0.39817047119140625,
+      "learning_rate": 3.9620990712282e-05,
+      "loss": 0.2455,
+      "step": 2726
+    },
+    {
+      "epoch": 0.5759239704329462,
+      "grad_norm": 0.3528037965297699,
+      "learning_rate": 3.958784139990046e-05,
+      "loss": 0.2362,
+      "step": 2727
+    },
+    {
+      "epoch": 0.5761351636747624,
+      "grad_norm": 0.4104713797569275,
+      "learning_rate": 3.9554696870966566e-05,
+      "loss": 0.2382,
+      "step": 2728
+    },
+    {
+      "epoch": 0.5763463569165787,
+      "grad_norm": 0.333295613527298,
+      "learning_rate": 3.9521557140707264e-05,
+      "loss": 0.2405,
+      "step": 2729
+    },
+    {
+      "epoch": 0.5765575501583949,
+      "grad_norm": 0.3023427426815033,
+      "learning_rate": 3.948842222434728e-05,
+      "loss": 0.2367,
+      "step": 2730
+    },
+    {
+      "epoch": 0.5767687434002112,
+      "grad_norm": 0.29385069012641907,
+      "learning_rate": 3.945529213710908e-05,
+      "loss": 0.2321,
+      "step": 2731
+    },
+    {
+      "epoch": 0.5769799366420274,
+      "grad_norm": 0.3327852189540863,
+      "learning_rate": 3.942216689421298e-05,
+      "loss": 0.233,
+      "step": 2732
+    },
+    {
+      "epoch": 0.5771911298838437,
+      "grad_norm": 0.32334765791893005,
+      "learning_rate": 3.9389046510877044e-05,
+      "loss": 0.2408,
+      "step": 2733
+    },
+    {
+      "epoch": 0.57740232312566,
+      "grad_norm": 0.2967278063297272,
+      "learning_rate": 3.935593100231707e-05,
+      "loss": 0.2405,
+      "step": 2734
+    },
+    {
+      "epoch": 0.5776135163674763,
+      "grad_norm": 0.29247260093688965,
+      "learning_rate": 3.9322820383746675e-05,
+      "loss": 0.2391,
+      "step": 2735
+    },
+    {
+      "epoch": 0.5778247096092926,
+      "grad_norm": 0.3558351695537567,
+      "learning_rate": 3.928971467037718e-05,
+      "loss": 0.2385,
+      "step": 2736
+    },
+    {
+      "epoch": 0.5780359028511087,
+      "grad_norm": 1.134796380996704,
+      "learning_rate": 3.9256613877417715e-05,
+      "loss": 0.2392,
+      "step": 2737
+    },
+    {
+      "epoch": 0.578247096092925,
+      "grad_norm": 0.3262888789176941,
+      "learning_rate": 3.9223518020075064e-05,
+      "loss": 0.2375,
+      "step": 2738
+    },
+    {
+      "epoch": 0.5784582893347413,
+      "grad_norm": 0.3076988756656647,
+      "learning_rate": 3.91904271135538e-05,
+      "loss": 0.241,
+      "step": 2739
+    },
+    {
+      "epoch": 0.5786694825765576,
+      "grad_norm": 0.30255013704299927,
+      "learning_rate": 3.915734117305624e-05,
+      "loss": 0.2412,
+      "step": 2740
+    },
+    {
+      "epoch": 0.5788806758183738,
+      "grad_norm": 0.3603900671005249,
+      "learning_rate": 3.912426021378235e-05,
+      "loss": 0.2425,
+      "step": 2741
+    },
+    {
+      "epoch": 0.5790918690601901,
+      "grad_norm": 0.3081553280353546,
+      "learning_rate": 3.9091184250929844e-05,
+      "loss": 0.2326,
+      "step": 2742
+    },
+    {
+      "epoch": 0.5793030623020063,
+      "grad_norm": 0.2931198477745056,
+      "learning_rate": 3.9058113299694175e-05,
+      "loss": 0.2443,
+      "step": 2743
+    },
+    {
+      "epoch": 0.5795142555438226,
+      "grad_norm": 0.2722361385822296,
+      "learning_rate": 3.9025047375268456e-05,
+      "loss": 0.2441,
+      "step": 2744
+    },
+    {
+      "epoch": 0.5797254487856388,
+      "grad_norm": 0.30264535546302795,
+      "learning_rate": 3.8991986492843475e-05,
+      "loss": 0.2395,
+      "step": 2745
+    },
+    {
+      "epoch": 0.5799366420274551,
+      "grad_norm": 0.30972152948379517,
+      "learning_rate": 3.895893066760774e-05,
+      "loss": 0.2343,
+      "step": 2746
+    },
+    {
+      "epoch": 0.5801478352692714,
+      "grad_norm": 0.3400992751121521,
+      "learning_rate": 3.892587991474743e-05,
+      "loss": 0.2336,
+      "step": 2747
+    },
+    {
+      "epoch": 0.5803590285110877,
+      "grad_norm": 0.38928690552711487,
+      "learning_rate": 3.889283424944636e-05,
+      "loss": 0.24,
+      "step": 2748
+    },
+    {
+      "epoch": 0.580570221752904,
+      "grad_norm": 0.3850376307964325,
+      "learning_rate": 3.8859793686886045e-05,
+      "loss": 0.2365,
+      "step": 2749
+    },
+    {
+      "epoch": 0.5807814149947201,
+      "grad_norm": 0.33159908652305603,
+      "learning_rate": 3.882675824224565e-05,
+      "loss": 0.2403,
+      "step": 2750
+    },
+    {
+      "epoch": 0.5809926082365364,
+      "grad_norm": 0.3491412103176117,
+      "learning_rate": 3.879372793070198e-05,
+      "loss": 0.2389,
+      "step": 2751
+    },
+    {
+      "epoch": 0.5812038014783527,
+      "grad_norm": 0.31051376461982727,
+      "learning_rate": 3.876070276742948e-05,
+      "loss": 0.2378,
+      "step": 2752
+    },
+    {
+      "epoch": 0.581414994720169,
+      "grad_norm": 0.34489813446998596,
+      "learning_rate": 3.872768276760022e-05,
+      "loss": 0.2386,
+      "step": 2753
+    },
+    {
+      "epoch": 0.5816261879619852,
+      "grad_norm": 0.2970786392688751,
+      "learning_rate": 3.8694667946383946e-05,
+      "loss": 0.2387,
+      "step": 2754
+    },
+    {
+      "epoch": 0.5818373812038015,
+      "grad_norm": 0.44803062081336975,
+      "learning_rate": 3.866165831894796e-05,
+      "loss": 0.2337,
+      "step": 2755
+    },
+    {
+      "epoch": 0.5820485744456178,
+      "grad_norm": 0.38751617074012756,
+      "learning_rate": 3.86286539004572e-05,
+      "loss": 0.2381,
+      "step": 2756
+    },
+    {
+      "epoch": 0.582259767687434,
+      "grad_norm": 0.32406339049339294,
+      "learning_rate": 3.8595654706074266e-05,
+      "loss": 0.2415,
+      "step": 2757
+    },
+    {
+      "epoch": 0.5824709609292502,
+      "grad_norm": 0.304618239402771,
+      "learning_rate": 3.8562660750959265e-05,
+      "loss": 0.2388,
+      "step": 2758
+    },
+    {
+      "epoch": 0.5826821541710665,
+      "grad_norm": 0.3215314447879791,
+      "learning_rate": 3.852967205026995e-05,
+      "loss": 0.2359,
+      "step": 2759
+    },
+    {
+      "epoch": 0.5828933474128828,
+      "grad_norm": 0.3494666814804077,
+      "learning_rate": 3.8496688619161685e-05,
+      "loss": 0.2391,
+      "step": 2760
+    },
+    {
+      "epoch": 0.5831045406546991,
+      "grad_norm": 0.2889004051685333,
+      "learning_rate": 3.846371047278735e-05,
+      "loss": 0.2357,
+      "step": 2761
+    },
+    {
+      "epoch": 0.5833157338965154,
+      "grad_norm": 0.38874292373657227,
+      "learning_rate": 3.843073762629743e-05,
+      "loss": 0.2387,
+      "step": 2762
+    },
+    {
+      "epoch": 0.5835269271383315,
+      "grad_norm": 0.3388304114341736,
+      "learning_rate": 3.8397770094839994e-05,
+      "loss": 0.243,
+      "step": 2763
+    },
+    {
+      "epoch": 0.5837381203801478,
+      "grad_norm": 0.2755889594554901,
+      "learning_rate": 3.836480789356063e-05,
+      "loss": 0.2393,
+      "step": 2764
+    },
+    {
+      "epoch": 0.5839493136219641,
+      "grad_norm": 0.32202860713005066,
+      "learning_rate": 3.8331851037602494e-05,
+      "loss": 0.2418,
+      "step": 2765
+    },
+    {
+      "epoch": 0.5841605068637804,
+      "grad_norm": 0.31637251377105713,
+      "learning_rate": 3.8298899542106315e-05,
+      "loss": 0.2347,
+      "step": 2766
+    },
+    {
+      "epoch": 0.5843717001055966,
+      "grad_norm": 0.2785618305206299,
+      "learning_rate": 3.826595342221029e-05,
+      "loss": 0.2367,
+      "step": 2767
+    },
+    {
+      "epoch": 0.5845828933474129,
+      "grad_norm": 0.28589925169944763,
+      "learning_rate": 3.823301269305024e-05,
+      "loss": 0.2386,
+      "step": 2768
+    },
+    {
+      "epoch": 0.5847940865892292,
+      "grad_norm": 0.3365243077278137,
+      "learning_rate": 3.8200077369759423e-05,
+      "loss": 0.2397,
+      "step": 2769
+    },
+    {
+      "epoch": 0.5850052798310454,
+      "grad_norm": 0.3319475054740906,
+      "learning_rate": 3.816714746746865e-05,
+      "loss": 0.2377,
+      "step": 2770
+    },
+    {
+      "epoch": 0.5852164730728616,
+      "grad_norm": 0.27581727504730225,
+      "learning_rate": 3.813422300130627e-05,
+      "loss": 0.2398,
+      "step": 2771
+    },
+    {
+      "epoch": 0.5854276663146779,
+      "grad_norm": 0.3807653486728668,
+      "learning_rate": 3.810130398639807e-05,
+      "loss": 0.2359,
+      "step": 2772
+    },
+    {
+      "epoch": 0.5856388595564942,
+      "grad_norm": 0.2617390751838684,
+      "learning_rate": 3.806839043786738e-05,
+      "loss": 0.2366,
+      "step": 2773
+    },
+    {
+      "epoch": 0.5858500527983105,
+      "grad_norm": 0.2992992103099823,
+      "learning_rate": 3.803548237083503e-05,
+      "loss": 0.2367,
+      "step": 2774
+    },
+    {
+      "epoch": 0.5860612460401268,
+      "grad_norm": 0.35363104939460754,
+      "learning_rate": 3.800257980041928e-05,
+      "loss": 0.2347,
+      "step": 2775
+    },
+    {
+      "epoch": 0.586272439281943,
+      "grad_norm": 0.3222738802433014,
+      "learning_rate": 3.796968274173588e-05,
+      "loss": 0.2366,
+      "step": 2776
+    },
+    {
+      "epoch": 0.5864836325237592,
+      "grad_norm": 0.31016862392425537,
+      "learning_rate": 3.7936791209898106e-05,
+      "loss": 0.2333,
+      "step": 2777
+    },
+    {
+      "epoch": 0.5866948257655755,
+      "grad_norm": 0.278100848197937,
+      "learning_rate": 3.790390522001662e-05,
+      "loss": 0.2323,
+      "step": 2778
+    },
+    {
+      "epoch": 0.5869060190073918,
+      "grad_norm": 0.28183016180992126,
+      "learning_rate": 3.7871024787199566e-05,
+      "loss": 0.2376,
+      "step": 2779
+    },
+    {
+      "epoch": 0.587117212249208,
+      "grad_norm": 0.3195381462574005,
+      "learning_rate": 3.783814992655256e-05,
+      "loss": 0.2387,
+      "step": 2780
+    },
+    {
+      "epoch": 0.5873284054910243,
+      "grad_norm": 0.30770155787467957,
+      "learning_rate": 3.78052806531786e-05,
+      "loss": 0.2357,
+      "step": 2781
+    },
+    {
+      "epoch": 0.5875395987328406,
+      "grad_norm": 0.3132328391075134,
+      "learning_rate": 3.777241698217819e-05,
+      "loss": 0.2377,
+      "step": 2782
+    },
+    {
+      "epoch": 0.5877507919746568,
+      "grad_norm": 0.3161110579967499,
+      "learning_rate": 3.773955892864921e-05,
+      "loss": 0.2399,
+      "step": 2783
+    },
+    {
+      "epoch": 0.587961985216473,
+      "grad_norm": 0.2893757224082947,
+      "learning_rate": 3.770670650768694e-05,
+      "loss": 0.235,
+      "step": 2784
+    },
+    {
+      "epoch": 0.5881731784582893,
+      "grad_norm": 0.30925253033638,
+      "learning_rate": 3.767385973438415e-05,
+      "loss": 0.2355,
+      "step": 2785
+    },
+    {
+      "epoch": 0.5883843717001056,
+      "grad_norm": 0.3148619532585144,
+      "learning_rate": 3.764101862383096e-05,
+      "loss": 0.2456,
+      "step": 2786
+    },
+    {
+      "epoch": 0.5885955649419219,
+      "grad_norm": 0.29597848653793335,
+      "learning_rate": 3.760818319111486e-05,
+      "loss": 0.2353,
+      "step": 2787
+    },
+    {
+      "epoch": 0.5888067581837381,
+      "grad_norm": 0.3165985643863678,
+      "learning_rate": 3.757535345132081e-05,
+      "loss": 0.238,
+      "step": 2788
+    },
+    {
+      "epoch": 0.5890179514255544,
+      "grad_norm": 0.35007795691490173,
+      "learning_rate": 3.754252941953112e-05,
+      "loss": 0.2357,
+      "step": 2789
+    },
+    {
+      "epoch": 0.5892291446673706,
+      "grad_norm": 0.3170489966869354,
+      "learning_rate": 3.750971111082542e-05,
+      "loss": 0.2405,
+      "step": 2790
+    },
+    {
+      "epoch": 0.5894403379091869,
+      "grad_norm": 0.28090429306030273,
+      "learning_rate": 3.747689854028081e-05,
+      "loss": 0.2472,
+      "step": 2791
+    },
+    {
+      "epoch": 0.5896515311510032,
+      "grad_norm": 0.3219681680202484,
+      "learning_rate": 3.7444091722971696e-05,
+      "loss": 0.24,
+      "step": 2792
+    },
+    {
+      "epoch": 0.5898627243928194,
+      "grad_norm": 0.25537997484207153,
+      "learning_rate": 3.741129067396982e-05,
+      "loss": 0.2364,
+      "step": 2793
+    },
+    {
+      "epoch": 0.5900739176346357,
+      "grad_norm": 0.3010719120502472,
+      "learning_rate": 3.737849540834437e-05,
+      "loss": 0.239,
+      "step": 2794
+    },
+    {
+      "epoch": 0.590285110876452,
+      "grad_norm": 0.3105645775794983,
+      "learning_rate": 3.734570594116176e-05,
+      "loss": 0.2357,
+      "step": 2795
+    },
+    {
+      "epoch": 0.5904963041182683,
+      "grad_norm": 0.26453015208244324,
+      "learning_rate": 3.73129222874858e-05,
+      "loss": 0.2384,
+      "step": 2796
+    },
+    {
+      "epoch": 0.5907074973600844,
+      "grad_norm": 0.27536654472351074,
+      "learning_rate": 3.728014446237765e-05,
+      "loss": 0.2429,
+      "step": 2797
+    },
+    {
+      "epoch": 0.5909186906019007,
+      "grad_norm": 0.2647611200809479,
+      "learning_rate": 3.7247372480895714e-05,
+      "loss": 0.2496,
+      "step": 2798
+    },
+    {
+      "epoch": 0.591129883843717,
+      "grad_norm": 0.27618610858917236,
+      "learning_rate": 3.721460635809581e-05,
+      "loss": 0.2398,
+      "step": 2799
+    },
+    {
+      "epoch": 0.5913410770855333,
+      "grad_norm": 0.30390408635139465,
+      "learning_rate": 3.7181846109031005e-05,
+      "loss": 0.2346,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5913410770855333,
+      "eval_train_split_loss": 0.3575657904148102,
+      "eval_train_split_runtime": 11.431,
+      "eval_train_split_samples_per_second": 0.875,
+      "eval_train_split_steps_per_second": 0.087,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5915522703273495,
+      "grad_norm": 0.6739718914031982,
+      "learning_rate": 3.714909174875164e-05,
+      "loss": 0.2353,
+      "step": 2801
+    },
+    {
+      "epoch": 0.5917634635691658,
+      "grad_norm": 0.3523328900337219,
+      "learning_rate": 3.711634329230543e-05,
+      "loss": 0.2352,
+      "step": 2802
+    },
+    {
+      "epoch": 0.591974656810982,
+      "grad_norm": 0.36691582202911377,
+      "learning_rate": 3.708360075473735e-05,
+      "loss": 0.2342,
+      "step": 2803
+    },
+    {
+      "epoch": 0.5921858500527983,
+      "grad_norm": 0.3202178478240967,
+      "learning_rate": 3.705086415108958e-05,
+      "loss": 0.2391,
+      "step": 2804
+    },
+    {
+      "epoch": 0.5923970432946146,
+      "grad_norm": 0.37684398889541626,
+      "learning_rate": 3.701813349640169e-05,
+      "loss": 0.2398,
+      "step": 2805
+    },
+    {
+      "epoch": 0.5926082365364308,
+      "grad_norm": 0.31868696212768555,
+      "learning_rate": 3.698540880571046e-05,
+      "loss": 0.2354,
+      "step": 2806
+    },
+    {
+      "epoch": 0.5928194297782471,
+      "grad_norm": 0.37812554836273193,
+      "learning_rate": 3.69526900940499e-05,
+      "loss": 0.2446,
+      "step": 2807
+    },
+    {
+      "epoch": 0.5930306230200634,
+      "grad_norm": 0.2850356996059418,
+      "learning_rate": 3.6919977376451335e-05,
+      "loss": 0.2442,
+      "step": 2808
+    },
+    {
+      "epoch": 0.5932418162618797,
+      "grad_norm": 0.3159286379814148,
+      "learning_rate": 3.688727066794329e-05,
+      "loss": 0.2412,
+      "step": 2809
+    },
+    {
+      "epoch": 0.5934530095036958,
+      "grad_norm": 0.46548140048980713,
+      "learning_rate": 3.685456998355158e-05,
+      "loss": 0.2427,
+      "step": 2810
+    },
+    {
+      "epoch": 0.5936642027455121,
+      "grad_norm": 0.28511151671409607,
+      "learning_rate": 3.6821875338299185e-05,
+      "loss": 0.2411,
+      "step": 2811
+    },
+    {
+      "epoch": 0.5938753959873284,
+      "grad_norm": 0.33745625615119934,
+      "learning_rate": 3.6789186747206345e-05,
+      "loss": 0.2357,
+      "step": 2812
+    },
+    {
+      "epoch": 0.5940865892291447,
+      "grad_norm": 0.33730438351631165,
+      "learning_rate": 3.6756504225290554e-05,
+      "loss": 0.2362,
+      "step": 2813
+    },
+    {
+      "epoch": 0.5942977824709609,
+      "grad_norm": 0.3079628646373749,
+      "learning_rate": 3.672382778756645e-05,
+      "loss": 0.2315,
+      "step": 2814
+    },
+    {
+      "epoch": 0.5945089757127772,
+      "grad_norm": 0.32386451959609985,
+      "learning_rate": 3.669115744904591e-05,
+      "loss": 0.2324,
+      "step": 2815
+    },
+    {
+      "epoch": 0.5947201689545935,
+      "grad_norm": 0.29547902941703796,
+      "learning_rate": 3.665849322473804e-05,
+      "loss": 0.235,
+      "step": 2816
+    },
+    {
+      "epoch": 0.5949313621964097,
+      "grad_norm": 0.41414907574653625,
+      "learning_rate": 3.662583512964907e-05,
+      "loss": 0.2311,
+      "step": 2817
+    },
+    {
+      "epoch": 0.595142555438226,
+      "grad_norm": 0.31273752450942993,
+      "learning_rate": 3.659318317878245e-05,
+      "loss": 0.234,
+      "step": 2818
+    },
+    {
+      "epoch": 0.5953537486800422,
+      "grad_norm": 0.28896960616111755,
+      "learning_rate": 3.656053738713884e-05,
+      "loss": 0.2383,
+      "step": 2819
+    },
+    {
+      "epoch": 0.5955649419218585,
+      "grad_norm": 0.3632212281227112,
+      "learning_rate": 3.6527897769715995e-05,
+      "loss": 0.2319,
+      "step": 2820
+    },
+    {
+      "epoch": 0.5957761351636748,
+      "grad_norm": 0.3228650987148285,
+      "learning_rate": 3.649526434150888e-05,
+      "loss": 0.2328,
+      "step": 2821
+    },
+    {
+      "epoch": 0.5959873284054911,
+      "grad_norm": 0.36897674202919006,
+      "learning_rate": 3.6462637117509647e-05,
+      "loss": 0.2365,
+      "step": 2822
+    },
+    {
+      "epoch": 0.5961985216473072,
+      "grad_norm": 0.29380524158477783,
+      "learning_rate": 3.6430016112707524e-05,
+      "loss": 0.2372,
+      "step": 2823
+    },
+    {
+      "epoch": 0.5964097148891235,
+      "grad_norm": 0.2964640259742737,
+      "learning_rate": 3.6397401342088954e-05,
+      "loss": 0.2368,
+      "step": 2824
+    },
+    {
+      "epoch": 0.5966209081309398,
+      "grad_norm": 0.3109495937824249,
+      "learning_rate": 3.636479282063745e-05,
+      "loss": 0.237,
+      "step": 2825
+    },
+    {
+      "epoch": 0.5968321013727561,
+      "grad_norm": 0.36109986901283264,
+      "learning_rate": 3.63321905633337e-05,
+      "loss": 0.2432,
+      "step": 2826
+    },
+    {
+      "epoch": 0.5970432946145723,
+      "grad_norm": 0.3430037498474121,
+      "learning_rate": 3.629959458515551e-05,
+      "loss": 0.2357,
+      "step": 2827
+    },
+    {
+      "epoch": 0.5972544878563886,
+      "grad_norm": 0.3241243362426758,
+      "learning_rate": 3.626700490107778e-05,
+      "loss": 0.2352,
+      "step": 2828
+    },
+    {
+      "epoch": 0.5974656810982049,
+      "grad_norm": 0.331821084022522,
+      "learning_rate": 3.6234421526072516e-05,
+      "loss": 0.2403,
+      "step": 2829
+    },
+    {
+      "epoch": 0.5976768743400211,
+      "grad_norm": 0.3001629710197449,
+      "learning_rate": 3.620184447510888e-05,
+      "loss": 0.2351,
+      "step": 2830
+    },
+    {
+      "epoch": 0.5978880675818374,
+      "grad_norm": 0.37214118242263794,
+      "learning_rate": 3.6169273763153056e-05,
+      "loss": 0.2332,
+      "step": 2831
+    },
+    {
+      "epoch": 0.5980992608236536,
+      "grad_norm": 0.4331023097038269,
+      "learning_rate": 3.6136709405168344e-05,
+      "loss": 0.2361,
+      "step": 2832
+    },
+    {
+      "epoch": 0.5983104540654699,
+      "grad_norm": 0.3331824541091919,
+      "learning_rate": 3.610415141611516e-05,
+      "loss": 0.2393,
+      "step": 2833
+    },
+    {
+      "epoch": 0.5985216473072862,
+      "grad_norm": 0.27378085255622864,
+      "learning_rate": 3.607159981095094e-05,
+      "loss": 0.2293,
+      "step": 2834
+    },
+    {
+      "epoch": 0.5987328405491025,
+      "grad_norm": 0.32710573077201843,
+      "learning_rate": 3.60390546046302e-05,
+      "loss": 0.2372,
+      "step": 2835
+    },
+    {
+      "epoch": 0.5989440337909187,
+      "grad_norm": 0.3682336211204529,
+      "learning_rate": 3.600651581210457e-05,
+      "loss": 0.236,
+      "step": 2836
+    },
+    {
+      "epoch": 0.5991552270327349,
+      "grad_norm": 0.3328309953212738,
+      "learning_rate": 3.597398344832263e-05,
+      "loss": 0.2358,
+      "step": 2837
+    },
+    {
+      "epoch": 0.5993664202745512,
+      "grad_norm": 0.27938058972358704,
+      "learning_rate": 3.594145752823012e-05,
+      "loss": 0.2332,
+      "step": 2838
+    },
+    {
+      "epoch": 0.5995776135163675,
+      "grad_norm": 0.317613810300827,
+      "learning_rate": 3.590893806676974e-05,
+      "loss": 0.2357,
+      "step": 2839
+    },
+    {
+      "epoch": 0.5997888067581837,
+      "grad_norm": 0.5353885889053345,
+      "learning_rate": 3.587642507888124e-05,
+      "loss": 0.232,
+      "step": 2840
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.35078856348991394,
+      "learning_rate": 3.584391857950143e-05,
+      "loss": 0.2339,
+      "step": 2841
+    },
+    {
+      "epoch": 0.6002111932418163,
+      "grad_norm": 0.3239556849002838,
+      "learning_rate": 3.581141858356412e-05,
+      "loss": 0.2388,
+      "step": 2842
+    },
+    {
+      "epoch": 0.6004223864836326,
+      "grad_norm": 0.36350443959236145,
+      "learning_rate": 3.577892510600008e-05,
+      "loss": 0.2343,
+      "step": 2843
+    },
+    {
+      "epoch": 0.6006335797254488,
+      "grad_norm": 0.3176441788673401,
+      "learning_rate": 3.574643816173717e-05,
+      "loss": 0.2334,
+      "step": 2844
+    },
+    {
+      "epoch": 0.600844772967265,
+      "grad_norm": 0.33995452523231506,
+      "learning_rate": 3.571395776570023e-05,
+      "loss": 0.2371,
+      "step": 2845
+    },
+    {
+      "epoch": 0.6010559662090813,
+      "grad_norm": 0.5550170540809631,
+      "learning_rate": 3.568148393281099e-05,
+      "loss": 0.2424,
+      "step": 2846
+    },
+    {
+      "epoch": 0.6012671594508976,
+      "grad_norm": 0.46955397725105286,
+      "learning_rate": 3.564901667798835e-05,
+      "loss": 0.2371,
+      "step": 2847
+    },
+    {
+      "epoch": 0.6014783526927139,
+      "grad_norm": 0.35430672764778137,
+      "learning_rate": 3.561655601614802e-05,
+      "loss": 0.2401,
+      "step": 2848
+    },
+    {
+      "epoch": 0.6016895459345301,
+      "grad_norm": 0.3873371481895447,
+      "learning_rate": 3.558410196220274e-05,
+      "loss": 0.2345,
+      "step": 2849
+    },
+    {
+      "epoch": 0.6019007391763463,
+      "grad_norm": 0.2998705506324768,
+      "learning_rate": 3.555165453106228e-05,
+      "loss": 0.2378,
+      "step": 2850
+    },
+    {
+      "epoch": 0.6021119324181626,
+      "grad_norm": 0.3094276785850525,
+      "learning_rate": 3.551921373763325e-05,
+      "loss": 0.234,
+      "step": 2851
+    },
+    {
+      "epoch": 0.6023231256599789,
+      "grad_norm": 0.300932377576828,
+      "learning_rate": 3.548677959681929e-05,
+      "loss": 0.238,
+      "step": 2852
+    },
+    {
+      "epoch": 0.6025343189017951,
+      "grad_norm": 0.2968694269657135,
+      "learning_rate": 3.545435212352097e-05,
+      "loss": 0.2352,
+      "step": 2853
+    },
+    {
+      "epoch": 0.6027455121436114,
+      "grad_norm": 0.2750983238220215,
+      "learning_rate": 3.542193133263576e-05,
+      "loss": 0.2327,
+      "step": 2854
+    },
+    {
+      "epoch": 0.6029567053854277,
+      "grad_norm": 0.3573574423789978,
+      "learning_rate": 3.5389517239058126e-05,
+      "loss": 0.2338,
+      "step": 2855
+    },
+    {
+      "epoch": 0.603167898627244,
+      "grad_norm": 0.2850675880908966,
+      "learning_rate": 3.5357109857679405e-05,
+      "loss": 0.2313,
+      "step": 2856
+    },
+    {
+      "epoch": 0.6033790918690601,
+      "grad_norm": 0.2699264585971832,
+      "learning_rate": 3.532470920338784e-05,
+      "loss": 0.2357,
+      "step": 2857
+    },
+    {
+      "epoch": 0.6035902851108764,
+      "grad_norm": 0.2879468500614166,
+      "learning_rate": 3.5292315291068644e-05,
+      "loss": 0.2359,
+      "step": 2858
+    },
+    {
+      "epoch": 0.6038014783526927,
+      "grad_norm": 0.6476089954376221,
+      "learning_rate": 3.5259928135603904e-05,
+      "loss": 0.2323,
+      "step": 2859
+    },
+    {
+      "epoch": 0.604012671594509,
+      "grad_norm": 0.4889545440673828,
+      "learning_rate": 3.5227547751872546e-05,
+      "loss": 0.2336,
+      "step": 2860
+    },
+    {
+      "epoch": 0.6042238648363253,
+      "grad_norm": 0.4603084921836853,
+      "learning_rate": 3.519517415475049e-05,
+      "loss": 0.2382,
+      "step": 2861
+    },
+    {
+      "epoch": 0.6044350580781415,
+      "grad_norm": 0.3143494725227356,
+      "learning_rate": 3.516280735911046e-05,
+      "loss": 0.2383,
+      "step": 2862
+    },
+    {
+      "epoch": 0.6046462513199578,
+      "grad_norm": 0.4366181492805481,
+      "learning_rate": 3.5130447379822076e-05,
+      "loss": 0.2336,
+      "step": 2863
+    },
+    {
+      "epoch": 0.604857444561774,
+      "grad_norm": 0.33190834522247314,
+      "learning_rate": 3.5098094231751825e-05,
+      "loss": 0.2321,
+      "step": 2864
+    },
+    {
+      "epoch": 0.6050686378035903,
+      "grad_norm": 0.2788475751876831,
+      "learning_rate": 3.5065747929763096e-05,
+      "loss": 0.2327,
+      "step": 2865
+    },
+    {
+      "epoch": 0.6052798310454065,
+      "grad_norm": 0.3899335265159607,
+      "learning_rate": 3.503340848871603e-05,
+      "loss": 0.2341,
+      "step": 2866
+    },
+    {
+      "epoch": 0.6054910242872228,
+      "grad_norm": 0.31378549337387085,
+      "learning_rate": 3.5001075923467755e-05,
+      "loss": 0.2359,
+      "step": 2867
+    },
+    {
+      "epoch": 0.6057022175290391,
+      "grad_norm": 0.3260684311389923,
+      "learning_rate": 3.49687502488721e-05,
+      "loss": 0.2328,
+      "step": 2868
+    },
+    {
+      "epoch": 0.6059134107708554,
+      "grad_norm": 0.3149801194667816,
+      "learning_rate": 3.493643147977988e-05,
+      "loss": 0.2347,
+      "step": 2869
+    },
+    {
+      "epoch": 0.6061246040126715,
+      "grad_norm": 0.3331005871295929,
+      "learning_rate": 3.4904119631038586e-05,
+      "loss": 0.2369,
+      "step": 2870
+    },
+    {
+      "epoch": 0.6063357972544878,
+      "grad_norm": 0.2900441586971283,
+      "learning_rate": 3.4871814717492606e-05,
+      "loss": 0.2327,
+      "step": 2871
+    },
+    {
+      "epoch": 0.6065469904963041,
+      "grad_norm": 0.32249966263771057,
+      "learning_rate": 3.483951675398316e-05,
+      "loss": 0.2346,
+      "step": 2872
+    },
+    {
+      "epoch": 0.6067581837381204,
+      "grad_norm": 0.3198586106300354,
+      "learning_rate": 3.480722575534825e-05,
+      "loss": 0.2377,
+      "step": 2873
+    },
+    {
+      "epoch": 0.6069693769799367,
+      "grad_norm": 0.3268449604511261,
+      "learning_rate": 3.4774941736422625e-05,
+      "loss": 0.2354,
+      "step": 2874
+    },
+    {
+      "epoch": 0.6071805702217529,
+      "grad_norm": 0.3610830008983612,
+      "learning_rate": 3.474266471203794e-05,
+      "loss": 0.2284,
+      "step": 2875
+    },
+    {
+      "epoch": 0.6073917634635692,
+      "grad_norm": 0.2985588312149048,
+      "learning_rate": 3.471039469702256e-05,
+      "loss": 0.233,
+      "step": 2876
+    },
+    {
+      "epoch": 0.6076029567053854,
+      "grad_norm": 0.3288106620311737,
+      "learning_rate": 3.4678131706201625e-05,
+      "loss": 0.2357,
+      "step": 2877
+    },
+    {
+      "epoch": 0.6078141499472017,
+      "grad_norm": 0.35579240322113037,
+      "learning_rate": 3.464587575439709e-05,
+      "loss": 0.2359,
+      "step": 2878
+    },
+    {
+      "epoch": 0.6080253431890179,
+      "grad_norm": 0.45609766244888306,
+      "learning_rate": 3.461362685642766e-05,
+      "loss": 0.2334,
+      "step": 2879
+    },
+    {
+      "epoch": 0.6082365364308342,
+      "grad_norm": 0.526413083076477,
+      "learning_rate": 3.458138502710876e-05,
+      "loss": 0.2334,
+      "step": 2880
+    },
+    {
+      "epoch": 0.6084477296726505,
+      "grad_norm": 0.34149619936943054,
+      "learning_rate": 3.4549150281252636e-05,
+      "loss": 0.2341,
+      "step": 2881
+    },
+    {
+      "epoch": 0.6086589229144668,
+      "grad_norm": 0.4005584418773651,
+      "learning_rate": 3.451692263366822e-05,
+      "loss": 0.2321,
+      "step": 2882
+    },
+    {
+      "epoch": 0.608870116156283,
+      "grad_norm": 0.30994361639022827,
+      "learning_rate": 3.448470209916125e-05,
+      "loss": 0.2302,
+      "step": 2883
+    },
+    {
+      "epoch": 0.6090813093980992,
+      "grad_norm": 0.34466642141342163,
+      "learning_rate": 3.445248869253413e-05,
+      "loss": 0.2369,
+      "step": 2884
+    },
+    {
+      "epoch": 0.6092925026399155,
+      "grad_norm": 0.29581400752067566,
+      "learning_rate": 3.442028242858599e-05,
+      "loss": 0.2432,
+      "step": 2885
+    },
+    {
+      "epoch": 0.6095036958817318,
+      "grad_norm": 0.3178180456161499,
+      "learning_rate": 3.438808332211275e-05,
+      "loss": 0.2363,
+      "step": 2886
+    },
+    {
+      "epoch": 0.6097148891235481,
+      "grad_norm": 0.29361653327941895,
+      "learning_rate": 3.435589138790697e-05,
+      "loss": 0.2375,
+      "step": 2887
+    },
+    {
+      "epoch": 0.6099260823653643,
+      "grad_norm": 0.37641000747680664,
+      "learning_rate": 3.432370664075792e-05,
+      "loss": 0.2384,
+      "step": 2888
+    },
+    {
+      "epoch": 0.6101372756071806,
+      "grad_norm": 0.4002230167388916,
+      "learning_rate": 3.4291529095451636e-05,
+      "loss": 0.2341,
+      "step": 2889
+    },
+    {
+      "epoch": 0.6103484688489969,
+      "grad_norm": 0.317348837852478,
+      "learning_rate": 3.4259358766770766e-05,
+      "loss": 0.2281,
+      "step": 2890
+    },
+    {
+      "epoch": 0.6105596620908131,
+      "grad_norm": 0.26942178606987,
+      "learning_rate": 3.422719566949467e-05,
+      "loss": 0.2352,
+      "step": 2891
+    },
+    {
+      "epoch": 0.6107708553326293,
+      "grad_norm": 0.29976487159729004,
+      "learning_rate": 3.419503981839941e-05,
+      "loss": 0.2354,
+      "step": 2892
+    },
+    {
+      "epoch": 0.6109820485744456,
+      "grad_norm": 0.3836342990398407,
+      "learning_rate": 3.4162891228257686e-05,
+      "loss": 0.2342,
+      "step": 2893
+    },
+    {
+      "epoch": 0.6111932418162619,
+      "grad_norm": 0.3051086664199829,
+      "learning_rate": 3.413074991383887e-05,
+      "loss": 0.2329,
+      "step": 2894
+    },
+    {
+      "epoch": 0.6114044350580782,
+      "grad_norm": 0.3116711974143982,
+      "learning_rate": 3.409861588990902e-05,
+      "loss": 0.2303,
+      "step": 2895
+    },
+    {
+      "epoch": 0.6116156282998944,
+      "grad_norm": 0.33673545718193054,
+      "learning_rate": 3.4066489171230786e-05,
+      "loss": 0.2299,
+      "step": 2896
+    },
+    {
+      "epoch": 0.6118268215417106,
+      "grad_norm": 0.3039020299911499,
+      "learning_rate": 3.403436977256355e-05,
+      "loss": 0.2323,
+      "step": 2897
+    },
+    {
+      "epoch": 0.6120380147835269,
+      "grad_norm": 0.5289638042449951,
+      "learning_rate": 3.4002257708663224e-05,
+      "loss": 0.2359,
+      "step": 2898
+    },
+    {
+      "epoch": 0.6122492080253432,
+      "grad_norm": 0.2867031395435333,
+      "learning_rate": 3.397015299428242e-05,
+      "loss": 0.235,
+      "step": 2899
+    },
+    {
+      "epoch": 0.6124604012671595,
+      "grad_norm": 0.29067304730415344,
+      "learning_rate": 3.393805564417038e-05,
+      "loss": 0.2352,
+      "step": 2900
+    },
+    {
+      "epoch": 0.6126715945089757,
+      "grad_norm": 0.30219727754592896,
+      "learning_rate": 3.390596567307293e-05,
+      "loss": 0.2355,
+      "step": 2901
+    },
+    {
+      "epoch": 0.612882787750792,
+      "grad_norm": 0.276287317276001,
+      "learning_rate": 3.387388309573249e-05,
+      "loss": 0.2367,
+      "step": 2902
+    },
+    {
+      "epoch": 0.6130939809926083,
+      "grad_norm": 0.5059013962745667,
+      "learning_rate": 3.3841807926888155e-05,
+      "loss": 0.2329,
+      "step": 2903
+    },
+    {
+      "epoch": 0.6133051742344245,
+      "grad_norm": 0.2947138249874115,
+      "learning_rate": 3.380974018127555e-05,
+      "loss": 0.2369,
+      "step": 2904
+    },
+    {
+      "epoch": 0.6135163674762407,
+      "grad_norm": 0.3310089111328125,
+      "learning_rate": 3.37776798736269e-05,
+      "loss": 0.2324,
+      "step": 2905
+    },
+    {
+      "epoch": 0.613727560718057,
+      "grad_norm": 0.2948853671550751,
+      "learning_rate": 3.374562701867106e-05,
+      "loss": 0.2328,
+      "step": 2906
+    },
+    {
+      "epoch": 0.6139387539598733,
+      "grad_norm": 0.3049337863922119,
+      "learning_rate": 3.371358163113338e-05,
+      "loss": 0.235,
+      "step": 2907
+    },
+    {
+      "epoch": 0.6141499472016896,
+      "grad_norm": 0.27537792921066284,
+      "learning_rate": 3.368154372573584e-05,
+      "loss": 0.2346,
+      "step": 2908
+    },
+    {
+      "epoch": 0.6143611404435058,
+      "grad_norm": 0.3279399871826172,
+      "learning_rate": 3.3649513317197e-05,
+      "loss": 0.2311,
+      "step": 2909
+    },
+    {
+      "epoch": 0.614572333685322,
+      "grad_norm": 0.29331815242767334,
+      "learning_rate": 3.361749042023189e-05,
+      "loss": 0.2338,
+      "step": 2910
+    },
+    {
+      "epoch": 0.6147835269271383,
+      "grad_norm": 0.34321916103363037,
+      "learning_rate": 3.3585475049552183e-05,
+      "loss": 0.2356,
+      "step": 2911
+    },
+    {
+      "epoch": 0.6149947201689546,
+      "grad_norm": 0.30281415581703186,
+      "learning_rate": 3.3553467219866056e-05,
+      "loss": 0.233,
+      "step": 2912
+    },
+    {
+      "epoch": 0.6152059134107709,
+      "grad_norm": 0.33479446172714233,
+      "learning_rate": 3.352146694587819e-05,
+      "loss": 0.2327,
+      "step": 2913
+    },
+    {
+      "epoch": 0.6154171066525871,
+      "grad_norm": 0.3250405788421631,
+      "learning_rate": 3.348947424228984e-05,
+      "loss": 0.2324,
+      "step": 2914
+    },
+    {
+      "epoch": 0.6156282998944034,
+      "grad_norm": 0.29244932532310486,
+      "learning_rate": 3.345748912379878e-05,
+      "loss": 0.2331,
+      "step": 2915
+    },
+    {
+      "epoch": 0.6158394931362197,
+      "grad_norm": 0.32867908477783203,
+      "learning_rate": 3.342551160509925e-05,
+      "loss": 0.2341,
+      "step": 2916
+    },
+    {
+      "epoch": 0.616050686378036,
+      "grad_norm": 0.27620774507522583,
+      "learning_rate": 3.339354170088207e-05,
+      "loss": 0.2307,
+      "step": 2917
+    },
+    {
+      "epoch": 0.6162618796198521,
+      "grad_norm": 0.30944177508354187,
+      "learning_rate": 3.336157942583453e-05,
+      "loss": 0.2333,
+      "step": 2918
+    },
+    {
+      "epoch": 0.6164730728616684,
+      "grad_norm": 0.2853928506374359,
+      "learning_rate": 3.332962479464037e-05,
+      "loss": 0.2347,
+      "step": 2919
+    },
+    {
+      "epoch": 0.6166842661034847,
+      "grad_norm": 0.27635249495506287,
+      "learning_rate": 3.329767782197991e-05,
+      "loss": 0.2316,
+      "step": 2920
+    },
+    {
+      "epoch": 0.616895459345301,
+      "grad_norm": 0.3172418177127838,
+      "learning_rate": 3.326573852252989e-05,
+      "loss": 0.2381,
+      "step": 2921
+    },
+    {
+      "epoch": 0.6171066525871172,
+      "grad_norm": 0.2818737328052521,
+      "learning_rate": 3.323380691096348e-05,
+      "loss": 0.2322,
+      "step": 2922
+    },
+    {
+      "epoch": 0.6173178458289335,
+      "grad_norm": 0.2827300727367401,
+      "learning_rate": 3.320188300195047e-05,
+      "loss": 0.2349,
+      "step": 2923
+    },
+    {
+      "epoch": 0.6175290390707497,
+      "grad_norm": 0.27501359581947327,
+      "learning_rate": 3.316996681015695e-05,
+      "loss": 0.2361,
+      "step": 2924
+    },
+    {
+      "epoch": 0.617740232312566,
+      "grad_norm": 0.37436890602111816,
+      "learning_rate": 3.313805835024556e-05,
+      "loss": 0.2325,
+      "step": 2925
+    },
+    {
+      "epoch": 0.6179514255543822,
+      "grad_norm": 0.4037608802318573,
+      "learning_rate": 3.310615763687536e-05,
+      "loss": 0.2388,
+      "step": 2926
+    },
+    {
+      "epoch": 0.6181626187961985,
+      "grad_norm": 0.28328436613082886,
+      "learning_rate": 3.307426468470182e-05,
+      "loss": 0.2386,
+      "step": 2927
+    },
+    {
+      "epoch": 0.6183738120380148,
+      "grad_norm": 0.28247496485710144,
+      "learning_rate": 3.3042379508376906e-05,
+      "loss": 0.2355,
+      "step": 2928
+    },
+    {
+      "epoch": 0.6185850052798311,
+      "grad_norm": 0.3310779631137848,
+      "learning_rate": 3.301050212254898e-05,
+      "loss": 0.2374,
+      "step": 2929
+    },
+    {
+      "epoch": 0.6187961985216474,
+      "grad_norm": 0.2984446883201599,
+      "learning_rate": 3.297863254186279e-05,
+      "loss": 0.233,
+      "step": 2930
+    },
+    {
+      "epoch": 0.6190073917634635,
+      "grad_norm": 0.29901978373527527,
+      "learning_rate": 3.2946770780959565e-05,
+      "loss": 0.228,
+      "step": 2931
+    },
+    {
+      "epoch": 0.6192185850052798,
+      "grad_norm": 0.2637610137462616,
+      "learning_rate": 3.2914916854476926e-05,
+      "loss": 0.2363,
+      "step": 2932
+    },
+    {
+      "epoch": 0.6194297782470961,
+      "grad_norm": 0.42979711294174194,
+      "learning_rate": 3.288307077704883e-05,
+      "loss": 0.2347,
+      "step": 2933
+    },
+    {
+      "epoch": 0.6196409714889124,
+      "grad_norm": 0.2954546809196472,
+      "learning_rate": 3.285123256330571e-05,
+      "loss": 0.2355,
+      "step": 2934
+    },
+    {
+      "epoch": 0.6198521647307286,
+      "grad_norm": 0.5757325887680054,
+      "learning_rate": 3.2819402227874365e-05,
+      "loss": 0.2372,
+      "step": 2935
+    },
+    {
+      "epoch": 0.6200633579725449,
+      "grad_norm": 0.3732314705848694,
+      "learning_rate": 3.278757978537792e-05,
+      "loss": 0.2354,
+      "step": 2936
+    },
+    {
+      "epoch": 0.6202745512143611,
+      "grad_norm": 0.30068323016166687,
+      "learning_rate": 3.275576525043594e-05,
+      "loss": 0.2344,
+      "step": 2937
+    },
+    {
+      "epoch": 0.6204857444561774,
+      "grad_norm": 0.29688704013824463,
+      "learning_rate": 3.2723958637664345e-05,
+      "loss": 0.2281,
+      "step": 2938
+    },
+    {
+      "epoch": 0.6206969376979936,
+      "grad_norm": 0.29557088017463684,
+      "learning_rate": 3.269215996167541e-05,
+      "loss": 0.2312,
+      "step": 2939
+    },
+    {
+      "epoch": 0.6209081309398099,
+      "grad_norm": 0.355067640542984,
+      "learning_rate": 3.2660369237077724e-05,
+      "loss": 0.2304,
+      "step": 2940
+    },
+    {
+      "epoch": 0.6211193241816262,
+      "grad_norm": 0.2888340950012207,
+      "learning_rate": 3.262858647847627e-05,
+      "loss": 0.2331,
+      "step": 2941
+    },
+    {
+      "epoch": 0.6213305174234425,
+      "grad_norm": 0.25456440448760986,
+      "learning_rate": 3.2596811700472384e-05,
+      "loss": 0.2306,
+      "step": 2942
+    },
+    {
+      "epoch": 0.6215417106652588,
+      "grad_norm": 0.2763586640357971,
+      "learning_rate": 3.256504491766369e-05,
+      "loss": 0.2385,
+      "step": 2943
+    },
+    {
+      "epoch": 0.6217529039070749,
+      "grad_norm": 0.29783007502555847,
+      "learning_rate": 3.2533286144644134e-05,
+      "loss": 0.2332,
+      "step": 2944
+    },
+    {
+      "epoch": 0.6219640971488912,
+      "grad_norm": 0.3333622217178345,
+      "learning_rate": 3.250153539600407e-05,
+      "loss": 0.2373,
+      "step": 2945
+    },
+    {
+      "epoch": 0.6221752903907075,
+      "grad_norm": 0.2832644283771515,
+      "learning_rate": 3.246979268633005e-05,
+      "loss": 0.242,
+      "step": 2946
+    },
+    {
+      "epoch": 0.6223864836325238,
+      "grad_norm": 0.39853474497795105,
+      "learning_rate": 3.2438058030205e-05,
+      "loss": 0.2337,
+      "step": 2947
+    },
+    {
+      "epoch": 0.62259767687434,
+      "grad_norm": 0.3442070484161377,
+      "learning_rate": 3.2406331442208146e-05,
+      "loss": 0.2351,
+      "step": 2948
+    },
+    {
+      "epoch": 0.6228088701161563,
+      "grad_norm": 0.3147808313369751,
+      "learning_rate": 3.2374612936914974e-05,
+      "loss": 0.2253,
+      "step": 2949
+    },
+    {
+      "epoch": 0.6230200633579726,
+      "grad_norm": 0.32762184739112854,
+      "learning_rate": 3.2342902528897276e-05,
+      "loss": 0.2287,
+      "step": 2950
+    },
+    {
+      "epoch": 0.6232312565997888,
+      "grad_norm": 0.38223567605018616,
+      "learning_rate": 3.231120023272314e-05,
+      "loss": 0.2315,
+      "step": 2951
+    },
+    {
+      "epoch": 0.623442449841605,
+      "grad_norm": 0.32931822538375854,
+      "learning_rate": 3.22795060629569e-05,
+      "loss": 0.2326,
+      "step": 2952
+    },
+    {
+      "epoch": 0.6236536430834213,
+      "grad_norm": 0.3535999059677124,
+      "learning_rate": 3.224782003415918e-05,
+      "loss": 0.2299,
+      "step": 2953
+    },
+    {
+      "epoch": 0.6238648363252376,
+      "grad_norm": 0.39683961868286133,
+      "learning_rate": 3.221614216088684e-05,
+      "loss": 0.2351,
+      "step": 2954
+    },
+    {
+      "epoch": 0.6240760295670539,
+      "grad_norm": 0.28678348660469055,
+      "learning_rate": 3.2184472457693006e-05,
+      "loss": 0.2322,
+      "step": 2955
+    },
+    {
+      "epoch": 0.6242872228088702,
+      "grad_norm": 0.5411520004272461,
+      "learning_rate": 3.215281093912707e-05,
+      "loss": 0.2333,
+      "step": 2956
+    },
+    {
+      "epoch": 0.6244984160506863,
+      "grad_norm": 0.3360096514225006,
+      "learning_rate": 3.212115761973461e-05,
+      "loss": 0.2308,
+      "step": 2957
+    },
+    {
+      "epoch": 0.6247096092925026,
+      "grad_norm": 0.36475569009780884,
+      "learning_rate": 3.2089512514057477e-05,
+      "loss": 0.2338,
+      "step": 2958
+    },
+    {
+      "epoch": 0.6249208025343189,
+      "grad_norm": 0.3333730697631836,
+      "learning_rate": 3.205787563663378e-05,
+      "loss": 0.2396,
+      "step": 2959
+    },
+    {
+      "epoch": 0.6251319957761352,
+      "grad_norm": 0.7416667938232422,
+      "learning_rate": 3.202624700199777e-05,
+      "loss": 0.2339,
+      "step": 2960
+    },
+    {
+      "epoch": 0.6253431890179514,
+      "grad_norm": 0.3209763169288635,
+      "learning_rate": 3.199462662467996e-05,
+      "loss": 0.2265,
+      "step": 2961
+    },
+    {
+      "epoch": 0.6255543822597677,
+      "grad_norm": 0.3189677298069,
+      "learning_rate": 3.196301451920708e-05,
+      "loss": 0.2332,
+      "step": 2962
+    },
+    {
+      "epoch": 0.625765575501584,
+      "grad_norm": 0.48028862476348877,
+      "learning_rate": 3.1931410700102016e-05,
+      "loss": 0.2361,
+      "step": 2963
+    },
+    {
+      "epoch": 0.6259767687434002,
+      "grad_norm": 0.3552095592021942,
+      "learning_rate": 3.189981518188388e-05,
+      "loss": 0.2326,
+      "step": 2964
+    },
+    {
+      "epoch": 0.6261879619852164,
+      "grad_norm": 0.3513357937335968,
+      "learning_rate": 3.1868227979067984e-05,
+      "loss": 0.2335,
+      "step": 2965
+    },
+    {
+      "epoch": 0.6263991552270327,
+      "grad_norm": 0.34849846363067627,
+      "learning_rate": 3.1836649106165775e-05,
+      "loss": 0.2271,
+      "step": 2966
+    },
+    {
+      "epoch": 0.626610348468849,
+      "grad_norm": 0.26313716173171997,
+      "learning_rate": 3.1805078577684884e-05,
+      "loss": 0.2277,
+      "step": 2967
+    },
+    {
+      "epoch": 0.6268215417106653,
+      "grad_norm": 0.3002517819404602,
+      "learning_rate": 3.177351640812916e-05,
+      "loss": 0.2349,
+      "step": 2968
+    },
+    {
+      "epoch": 0.6270327349524816,
+      "grad_norm": 0.3471336364746094,
+      "learning_rate": 3.1741962611998534e-05,
+      "loss": 0.2328,
+      "step": 2969
+    },
+    {
+      "epoch": 0.6272439281942978,
+      "grad_norm": 0.30390119552612305,
+      "learning_rate": 3.1710417203789155e-05,
+      "loss": 0.2321,
+      "step": 2970
+    },
+    {
+      "epoch": 0.627455121436114,
+      "grad_norm": 0.28932949900627136,
+      "learning_rate": 3.16788801979933e-05,
+      "loss": 0.2345,
+      "step": 2971
+    },
+    {
+      "epoch": 0.6276663146779303,
+      "grad_norm": 0.3197578489780426,
+      "learning_rate": 3.164735160909934e-05,
+      "loss": 0.232,
+      "step": 2972
+    },
+    {
+      "epoch": 0.6278775079197466,
+      "grad_norm": 0.30003684759140015,
+      "learning_rate": 3.161583145159185e-05,
+      "loss": 0.2286,
+      "step": 2973
+    },
+    {
+      "epoch": 0.6280887011615628,
+      "grad_norm": 0.2833123803138733,
+      "learning_rate": 3.15843197399515e-05,
+      "loss": 0.2337,
+      "step": 2974
+    },
+    {
+      "epoch": 0.6282998944033791,
+      "grad_norm": 0.36067500710487366,
+      "learning_rate": 3.155281648865504e-05,
+      "loss": 0.2271,
+      "step": 2975
+    },
+    {
+      "epoch": 0.6285110876451954,
+      "grad_norm": 0.32001587748527527,
+      "learning_rate": 3.152132171217543e-05,
+      "loss": 0.2311,
+      "step": 2976
+    },
+    {
+      "epoch": 0.6287222808870117,
+      "grad_norm": 0.3431919515132904,
+      "learning_rate": 3.148983542498164e-05,
+      "loss": 0.2339,
+      "step": 2977
+    },
+    {
+      "epoch": 0.6289334741288278,
+      "grad_norm": 0.286998987197876,
+      "learning_rate": 3.145835764153878e-05,
+      "loss": 0.2332,
+      "step": 2978
+    },
+    {
+      "epoch": 0.6291446673706441,
+      "grad_norm": 0.3083532452583313,
+      "learning_rate": 3.1426888376308084e-05,
+      "loss": 0.2349,
+      "step": 2979
+    },
+    {
+      "epoch": 0.6293558606124604,
+      "grad_norm": 0.36541271209716797,
+      "learning_rate": 3.1395427643746796e-05,
+      "loss": 0.2371,
+      "step": 2980
+    },
+    {
+      "epoch": 0.6295670538542767,
+      "grad_norm": 0.29836100339889526,
+      "learning_rate": 3.136397545830831e-05,
+      "loss": 0.2336,
+      "step": 2981
+    },
+    {
+      "epoch": 0.629778247096093,
+      "grad_norm": 0.2602342367172241,
+      "learning_rate": 3.133253183444208e-05,
+      "loss": 0.2377,
+      "step": 2982
+    },
+    {
+      "epoch": 0.6299894403379092,
+      "grad_norm": 0.3169727027416229,
+      "learning_rate": 3.1301096786593575e-05,
+      "loss": 0.2294,
+      "step": 2983
+    },
+    {
+      "epoch": 0.6302006335797254,
+      "grad_norm": 0.3921878933906555,
+      "learning_rate": 3.12696703292044e-05,
+      "loss": 0.2242,
+      "step": 2984
+    },
+    {
+      "epoch": 0.6304118268215417,
+      "grad_norm": 0.36334311962127686,
+      "learning_rate": 3.123825247671217e-05,
+      "loss": 0.2335,
+      "step": 2985
+    },
+    {
+      "epoch": 0.630623020063358,
+      "grad_norm": 0.2874634563922882,
+      "learning_rate": 3.1206843243550546e-05,
+      "loss": 0.2301,
+      "step": 2986
+    },
+    {
+      "epoch": 0.6308342133051742,
+      "grad_norm": 0.3294402062892914,
+      "learning_rate": 3.117544264414923e-05,
+      "loss": 0.2305,
+      "step": 2987
+    },
+    {
+      "epoch": 0.6310454065469905,
+      "grad_norm": 0.32110047340393066,
+      "learning_rate": 3.1144050692934e-05,
+      "loss": 0.23,
+      "step": 2988
+    },
+    {
+      "epoch": 0.6312565997888068,
+      "grad_norm": 0.29757383465766907,
+      "learning_rate": 3.111266740432658e-05,
+      "loss": 0.2295,
+      "step": 2989
+    },
+    {
+      "epoch": 0.6314677930306231,
+      "grad_norm": 0.308436781167984,
+      "learning_rate": 3.108129279274479e-05,
+      "loss": 0.2352,
+      "step": 2990
+    },
+    {
+      "epoch": 0.6316789862724392,
+      "grad_norm": 0.32039111852645874,
+      "learning_rate": 3.104992687260244e-05,
+      "loss": 0.2338,
+      "step": 2991
+    },
+    {
+      "epoch": 0.6318901795142555,
+      "grad_norm": 0.3341165781021118,
+      "learning_rate": 3.101856965830931e-05,
+      "loss": 0.2318,
+      "step": 2992
+    },
+    {
+      "epoch": 0.6321013727560718,
+      "grad_norm": 0.29674389958381653,
+      "learning_rate": 3.098722116427123e-05,
+      "loss": 0.2278,
+      "step": 2993
+    },
+    {
+      "epoch": 0.6323125659978881,
+      "grad_norm": 0.28486210107803345,
+      "learning_rate": 3.095588140489002e-05,
+      "loss": 0.2318,
+      "step": 2994
+    },
+    {
+      "epoch": 0.6325237592397043,
+      "grad_norm": 0.3269345760345459,
+      "learning_rate": 3.092455039456343e-05,
+      "loss": 0.2294,
+      "step": 2995
+    },
+    {
+      "epoch": 0.6327349524815206,
+      "grad_norm": 0.2724529206752777,
+      "learning_rate": 3.089322814768527e-05,
+      "loss": 0.2281,
+      "step": 2996
+    },
+    {
+      "epoch": 0.6329461457233369,
+      "grad_norm": 0.5205068588256836,
+      "learning_rate": 3.086191467864528e-05,
+      "loss": 0.2274,
+      "step": 2997
+    },
+    {
+      "epoch": 0.6331573389651531,
+      "grad_norm": 0.3308090567588806,
+      "learning_rate": 3.0830610001829174e-05,
+      "loss": 0.2321,
+      "step": 2998
+    },
+    {
+      "epoch": 0.6333685322069694,
+      "grad_norm": 0.3041190505027771,
+      "learning_rate": 3.079931413161862e-05,
+      "loss": 0.2363,
+      "step": 2999
+    },
+    {
+      "epoch": 0.6335797254487856,
+      "grad_norm": 0.28160884976387024,
+      "learning_rate": 3.076802708239124e-05,
+      "loss": 0.2247,
+      "step": 3000
+    },
+    {
+      "epoch": 0.6335797254487856,
+      "eval_train_split_loss": 0.3461324870586395,
+      "eval_train_split_runtime": 10.2854,
+      "eval_train_split_samples_per_second": 0.972,
+      "eval_train_split_steps_per_second": 0.097,
+      "step": 3000
+    },
+    {
+      "epoch": 0.6337909186906019,
+      "grad_norm": 0.306832492351532,
+      "learning_rate": 3.073674886852064e-05,
+      "loss": 0.2339,
+      "step": 3001
+    },
+    {
+      "epoch": 0.6340021119324182,
+      "grad_norm": 0.3131411075592041,
+      "learning_rate": 3.070547950437632e-05,
+      "loss": 0.2308,
+      "step": 3002
+    },
+    {
+      "epoch": 0.6342133051742345,
+      "grad_norm": 0.331999272108078,
+      "learning_rate": 3.067421900432372e-05,
+      "loss": 0.2296,
+      "step": 3003
+    },
+    {
+      "epoch": 0.6344244984160506,
+      "grad_norm": 0.365161657333374,
+      "learning_rate": 3.064296738272424e-05,
+      "loss": 0.2284,
+      "step": 3004
+    },
+    {
+      "epoch": 0.6346356916578669,
+      "grad_norm": 0.306587278842926,
+      "learning_rate": 3.061172465393518e-05,
+      "loss": 0.2295,
+      "step": 3005
+    },
+    {
+      "epoch": 0.6348468848996832,
+      "grad_norm": 0.3394300639629364,
+      "learning_rate": 3.058049083230974e-05,
+      "loss": 0.2323,
+      "step": 3006
+    },
+    {
+      "epoch": 0.6350580781414995,
+      "grad_norm": 0.32192662358283997,
+      "learning_rate": 3.0549265932197055e-05,
+      "loss": 0.2339,
+      "step": 3007
+    },
+    {
+      "epoch": 0.6352692713833157,
+      "grad_norm": 0.28615298867225647,
+      "learning_rate": 3.051804996794215e-05,
+      "loss": 0.2299,
+      "step": 3008
+    },
+    {
+      "epoch": 0.635480464625132,
+      "grad_norm": 0.32055121660232544,
+      "learning_rate": 3.0486842953885936e-05,
+      "loss": 0.2318,
+      "step": 3009
+    },
+    {
+      "epoch": 0.6356916578669483,
+      "grad_norm": 0.3251102864742279,
+      "learning_rate": 3.0455644904365234e-05,
+      "loss": 0.2298,
+      "step": 3010
+    },
+    {
+      "epoch": 0.6359028511087645,
+      "grad_norm": 0.3095821738243103,
+      "learning_rate": 3.0424455833712717e-05,
+      "loss": 0.2353,
+      "step": 3011
+    },
+    {
+      "epoch": 0.6361140443505808,
+      "grad_norm": 0.35642382502555847,
+      "learning_rate": 3.039327575625698e-05,
+      "loss": 0.231,
+      "step": 3012
+    },
+    {
+      "epoch": 0.636325237592397,
+      "grad_norm": 0.33006930351257324,
+      "learning_rate": 3.0362104686322424e-05,
+      "loss": 0.237,
+      "step": 3013
+    },
+    {
+      "epoch": 0.6365364308342133,
+      "grad_norm": 0.2921389043331146,
+      "learning_rate": 3.0330942638229344e-05,
+      "loss": 0.2303,
+      "step": 3014
+    },
+    {
+      "epoch": 0.6367476240760296,
+      "grad_norm": 0.3226618766784668,
+      "learning_rate": 3.029978962629393e-05,
+      "loss": 0.2342,
+      "step": 3015
+    },
+    {
+      "epoch": 0.6369588173178459,
+      "grad_norm": 0.3877405524253845,
+      "learning_rate": 3.0268645664828134e-05,
+      "loss": 0.2371,
+      "step": 3016
+    },
+    {
+      "epoch": 0.637170010559662,
+      "grad_norm": 0.3516109883785248,
+      "learning_rate": 3.023751076813981e-05,
+      "loss": 0.2342,
+      "step": 3017
+    },
+    {
+      "epoch": 0.6373812038014783,
+      "grad_norm": 0.27809739112854004,
+      "learning_rate": 3.020638495053266e-05,
+      "loss": 0.2386,
+      "step": 3018
+    },
+    {
+      "epoch": 0.6375923970432946,
+      "grad_norm": 0.27742424607276917,
+      "learning_rate": 3.0175268226306158e-05,
+      "loss": 0.2316,
+      "step": 3019
+    },
+    {
+      "epoch": 0.6378035902851109,
+      "grad_norm": 0.33293482661247253,
+      "learning_rate": 3.0144160609755633e-05,
+      "loss": 0.2295,
+      "step": 3020
+    },
+    {
+      "epoch": 0.6380147835269271,
+      "grad_norm": 0.3061244487762451,
+      "learning_rate": 3.0113062115172263e-05,
+      "loss": 0.2356,
+      "step": 3021
+    },
+    {
+      "epoch": 0.6382259767687434,
+      "grad_norm": 0.32253527641296387,
+      "learning_rate": 3.0081972756842967e-05,
+      "loss": 0.2335,
+      "step": 3022
+    },
+    {
+      "epoch": 0.6384371700105597,
+      "grad_norm": 0.3369770348072052,
+      "learning_rate": 3.0050892549050502e-05,
+      "loss": 0.2341,
+      "step": 3023
+    },
+    {
+      "epoch": 0.638648363252376,
+      "grad_norm": 0.31085020303726196,
+      "learning_rate": 3.001982150607345e-05,
+      "loss": 0.2323,
+      "step": 3024
+    },
+    {
+      "epoch": 0.6388595564941922,
+      "grad_norm": 0.34925466775894165,
+      "learning_rate": 2.9988759642186097e-05,
+      "loss": 0.235,
+      "step": 3025
+    },
+    {
+      "epoch": 0.6390707497360084,
+      "grad_norm": 0.3109789490699768,
+      "learning_rate": 2.995770697165864e-05,
+      "loss": 0.233,
+      "step": 3026
+    },
+    {
+      "epoch": 0.6392819429778247,
+      "grad_norm": 0.3246631324291229,
+      "learning_rate": 2.9926663508756924e-05,
+      "loss": 0.2265,
+      "step": 3027
+    },
+    {
+      "epoch": 0.639493136219641,
+      "grad_norm": 0.5469430088996887,
+      "learning_rate": 2.989562926774263e-05,
+      "loss": 0.2264,
+      "step": 3028
+    },
+    {
+      "epoch": 0.6397043294614573,
+      "grad_norm": 0.4118088483810425,
+      "learning_rate": 2.986460426287323e-05,
+      "loss": 0.2322,
+      "step": 3029
+    },
+    {
+      "epoch": 0.6399155227032735,
+      "grad_norm": 0.34733065962791443,
+      "learning_rate": 2.9833588508401866e-05,
+      "loss": 0.228,
+      "step": 3030
+    },
+    {
+      "epoch": 0.6401267159450897,
+      "grad_norm": 0.6411674618721008,
+      "learning_rate": 2.9802582018577495e-05,
+      "loss": 0.2326,
+      "step": 3031
+    },
+    {
+      "epoch": 0.640337909186906,
+      "grad_norm": 0.34577682614326477,
+      "learning_rate": 2.9771584807644825e-05,
+      "loss": 0.2407,
+      "step": 3032
+    },
+    {
+      "epoch": 0.6405491024287223,
+      "grad_norm": 0.34006422758102417,
+      "learning_rate": 2.9740596889844246e-05,
+      "loss": 0.2334,
+      "step": 3033
+    },
+    {
+      "epoch": 0.6407602956705385,
+      "grad_norm": 0.3779010474681854,
+      "learning_rate": 2.970961827941192e-05,
+      "loss": 0.2342,
+      "step": 3034
+    },
+    {
+      "epoch": 0.6409714889123548,
+      "grad_norm": 0.33080193400382996,
+      "learning_rate": 2.9678648990579748e-05,
+      "loss": 0.2335,
+      "step": 3035
+    },
+    {
+      "epoch": 0.6411826821541711,
+      "grad_norm": 0.3537871837615967,
+      "learning_rate": 2.96476890375753e-05,
+      "loss": 0.2288,
+      "step": 3036
+    },
+    {
+      "epoch": 0.6413938753959874,
+      "grad_norm": 0.3486878275871277,
+      "learning_rate": 2.961673843462187e-05,
+      "loss": 0.2282,
+      "step": 3037
+    },
+    {
+      "epoch": 0.6416050686378036,
+      "grad_norm": 0.2965608239173889,
+      "learning_rate": 2.9585797195938503e-05,
+      "loss": 0.2277,
+      "step": 3038
+    },
+    {
+      "epoch": 0.6418162618796198,
+      "grad_norm": 0.3657856285572052,
+      "learning_rate": 2.9554865335739877e-05,
+      "loss": 0.2303,
+      "step": 3039
+    },
+    {
+      "epoch": 0.6420274551214361,
+      "grad_norm": 0.37707945704460144,
+      "learning_rate": 2.9523942868236417e-05,
+      "loss": 0.229,
+      "step": 3040
+    },
+    {
+      "epoch": 0.6422386483632524,
+      "grad_norm": 0.3631666302680969,
+      "learning_rate": 2.94930298076342e-05,
+      "loss": 0.2363,
+      "step": 3041
+    },
+    {
+      "epoch": 0.6424498416050687,
+      "grad_norm": 0.37196728587150574,
+      "learning_rate": 2.9462126168134964e-05,
+      "loss": 0.2327,
+      "step": 3042
+    },
+    {
+      "epoch": 0.6426610348468849,
+      "grad_norm": 1.1121355295181274,
+      "learning_rate": 2.9431231963936174e-05,
+      "loss": 0.2282,
+      "step": 3043
+    },
+    {
+      "epoch": 0.6428722280887011,
+      "grad_norm": 0.39870283007621765,
+      "learning_rate": 2.9400347209230927e-05,
+      "loss": 0.231,
+      "step": 3044
+    },
+    {
+      "epoch": 0.6430834213305174,
+      "grad_norm": 0.33748286962509155,
+      "learning_rate": 2.9369471918207958e-05,
+      "loss": 0.2252,
+      "step": 3045
+    },
+    {
+      "epoch": 0.6432946145723337,
+      "grad_norm": 0.3920283019542694,
+      "learning_rate": 2.9338606105051703e-05,
+      "loss": 0.2268,
+      "step": 3046
+    },
+    {
+      "epoch": 0.6435058078141499,
+      "grad_norm": 0.3739641010761261,
+      "learning_rate": 2.9307749783942228e-05,
+      "loss": 0.2357,
+      "step": 3047
+    },
+    {
+      "epoch": 0.6437170010559662,
+      "grad_norm": 0.33909180760383606,
+      "learning_rate": 2.9276902969055185e-05,
+      "loss": 0.2261,
+      "step": 3048
+    },
+    {
+      "epoch": 0.6439281942977825,
+      "grad_norm": 0.30689123272895813,
+      "learning_rate": 2.924606567456194e-05,
+      "loss": 0.2275,
+      "step": 3049
+    },
+    {
+      "epoch": 0.6441393875395988,
+      "grad_norm": 0.3837328255176544,
+      "learning_rate": 2.9215237914629445e-05,
+      "loss": 0.2304,
+      "step": 3050
+    },
+    {
+      "epoch": 0.644350580781415,
+      "grad_norm": 0.29920318722724915,
+      "learning_rate": 2.918441970342024e-05,
+      "loss": 0.2312,
+      "step": 3051
+    },
+    {
+      "epoch": 0.6445617740232312,
+      "grad_norm": 0.3119010329246521,
+      "learning_rate": 2.9153611055092583e-05,
+      "loss": 0.2301,
+      "step": 3052
+    },
+    {
+      "epoch": 0.6447729672650475,
+      "grad_norm": 0.32063332200050354,
+      "learning_rate": 2.9122811983800184e-05,
+      "loss": 0.2369,
+      "step": 3053
+    },
+    {
+      "epoch": 0.6449841605068638,
+      "grad_norm": 0.27865931391716003,
+      "learning_rate": 2.9092022503692505e-05,
+      "loss": 0.237,
+      "step": 3054
+    },
+    {
+      "epoch": 0.6451953537486801,
+      "grad_norm": 0.28188279271125793,
+      "learning_rate": 2.906124262891451e-05,
+      "loss": 0.2333,
+      "step": 3055
+    },
+    {
+      "epoch": 0.6454065469904963,
+      "grad_norm": 0.32067686319351196,
+      "learning_rate": 2.9030472373606778e-05,
+      "loss": 0.2366,
+      "step": 3056
+    },
+    {
+      "epoch": 0.6456177402323126,
+      "grad_norm": 0.2906336486339569,
+      "learning_rate": 2.8999711751905455e-05,
+      "loss": 0.2321,
+      "step": 3057
+    },
+    {
+      "epoch": 0.6458289334741288,
+      "grad_norm": 0.3684746325016022,
+      "learning_rate": 2.8968960777942283e-05,
+      "loss": 0.2334,
+      "step": 3058
+    },
+    {
+      "epoch": 0.6460401267159451,
+      "grad_norm": 0.4433421790599823,
+      "learning_rate": 2.8938219465844562e-05,
+      "loss": 0.233,
+      "step": 3059
+    },
+    {
+      "epoch": 0.6462513199577613,
+      "grad_norm": 0.3329778015613556,
+      "learning_rate": 2.8907487829735147e-05,
+      "loss": 0.2354,
+      "step": 3060
+    },
+    {
+      "epoch": 0.6464625131995776,
+      "grad_norm": 0.31395289301872253,
+      "learning_rate": 2.8876765883732447e-05,
+      "loss": 0.2385,
+      "step": 3061
+    },
+    {
+      "epoch": 0.6466737064413939,
+      "grad_norm": 0.31003081798553467,
+      "learning_rate": 2.8846053641950444e-05,
+      "loss": 0.2323,
+      "step": 3062
+    },
+    {
+      "epoch": 0.6468848996832102,
+      "grad_norm": 0.3135981261730194,
+      "learning_rate": 2.8815351118498623e-05,
+      "loss": 0.2275,
+      "step": 3063
+    },
+    {
+      "epoch": 0.6470960929250263,
+      "grad_norm": 0.4058052599430084,
+      "learning_rate": 2.8784658327482027e-05,
+      "loss": 0.2317,
+      "step": 3064
+    },
+    {
+      "epoch": 0.6473072861668426,
+      "grad_norm": 0.32327064871788025,
+      "learning_rate": 2.8753975283001234e-05,
+      "loss": 0.2231,
+      "step": 3065
+    },
+    {
+      "epoch": 0.6475184794086589,
+      "grad_norm": 0.33835169672966003,
+      "learning_rate": 2.872330199915233e-05,
+      "loss": 0.2298,
+      "step": 3066
+    },
+    {
+      "epoch": 0.6477296726504752,
+      "grad_norm": 0.29836052656173706,
+      "learning_rate": 2.8692638490026925e-05,
+      "loss": 0.2363,
+      "step": 3067
+    },
+    {
+      "epoch": 0.6479408658922915,
+      "grad_norm": 0.36531150341033936,
+      "learning_rate": 2.8661984769712135e-05,
+      "loss": 0.236,
+      "step": 3068
+    },
+    {
+      "epoch": 0.6481520591341077,
+      "grad_norm": 0.37575194239616394,
+      "learning_rate": 2.8631340852290573e-05,
+      "loss": 0.2386,
+      "step": 3069
+    },
+    {
+      "epoch": 0.648363252375924,
+      "grad_norm": 0.4474295377731323,
+      "learning_rate": 2.860070675184036e-05,
+      "loss": 0.2321,
+      "step": 3070
+    },
+    {
+      "epoch": 0.6485744456177402,
+      "grad_norm": 0.33668723702430725,
+      "learning_rate": 2.85700824824351e-05,
+      "loss": 0.2246,
+      "step": 3071
+    },
+    {
+      "epoch": 0.6487856388595565,
+      "grad_norm": 0.29570871591567993,
+      "learning_rate": 2.8539468058143888e-05,
+      "loss": 0.2282,
+      "step": 3072
+    },
+    {
+      "epoch": 0.6489968321013727,
+      "grad_norm": 0.3341172933578491,
+      "learning_rate": 2.8508863493031278e-05,
+      "loss": 0.2318,
+      "step": 3073
+    },
+    {
+      "epoch": 0.649208025343189,
+      "grad_norm": 0.348532497882843,
+      "learning_rate": 2.8478268801157325e-05,
+      "loss": 0.235,
+      "step": 3074
+    },
+    {
+      "epoch": 0.6494192185850053,
+      "grad_norm": 0.30277878046035767,
+      "learning_rate": 2.844768399657751e-05,
+      "loss": 0.233,
+      "step": 3075
+    },
+    {
+      "epoch": 0.6496304118268216,
+      "grad_norm": 0.3137795925140381,
+      "learning_rate": 2.8417109093342815e-05,
+      "loss": 0.2265,
+      "step": 3076
+    },
+    {
+      "epoch": 0.6498416050686378,
+      "grad_norm": 0.40547239780426025,
+      "learning_rate": 2.8386544105499624e-05,
+      "loss": 0.2314,
+      "step": 3077
+    },
+    {
+      "epoch": 0.650052798310454,
+      "grad_norm": 0.37535548210144043,
+      "learning_rate": 2.8355989047089815e-05,
+      "loss": 0.24,
+      "step": 3078
+    },
+    {
+      "epoch": 0.6502639915522703,
+      "grad_norm": 0.33290621638298035,
+      "learning_rate": 2.8325443932150647e-05,
+      "loss": 0.2281,
+      "step": 3079
+    },
+    {
+      "epoch": 0.6504751847940866,
+      "grad_norm": 0.3013865649700165,
+      "learning_rate": 2.829490877471491e-05,
+      "loss": 0.2285,
+      "step": 3080
+    },
+    {
+      "epoch": 0.6506863780359029,
+      "grad_norm": 0.3188580870628357,
+      "learning_rate": 2.8264383588810695e-05,
+      "loss": 0.2281,
+      "step": 3081
+    },
+    {
+      "epoch": 0.6508975712777191,
+      "grad_norm": 0.32798150181770325,
+      "learning_rate": 2.823386838846156e-05,
+      "loss": 0.2315,
+      "step": 3082
+    },
+    {
+      "epoch": 0.6511087645195354,
+      "grad_norm": 0.3247312605381012,
+      "learning_rate": 2.820336318768657e-05,
+      "loss": 0.2349,
+      "step": 3083
+    },
+    {
+      "epoch": 0.6513199577613517,
+      "grad_norm": 0.2953028082847595,
+      "learning_rate": 2.8172868000500012e-05,
+      "loss": 0.2305,
+      "step": 3084
+    },
+    {
+      "epoch": 0.6515311510031679,
+      "grad_norm": 0.5457118153572083,
+      "learning_rate": 2.8142382840911746e-05,
+      "loss": 0.2352,
+      "step": 3085
+    },
+    {
+      "epoch": 0.6517423442449841,
+      "grad_norm": 0.33869674801826477,
+      "learning_rate": 2.8111907722926944e-05,
+      "loss": 0.2316,
+      "step": 3086
+    },
+    {
+      "epoch": 0.6519535374868004,
+      "grad_norm": 0.3335368037223816,
+      "learning_rate": 2.8081442660546125e-05,
+      "loss": 0.2316,
+      "step": 3087
+    },
+    {
+      "epoch": 0.6521647307286167,
+      "grad_norm": 0.34322983026504517,
+      "learning_rate": 2.805098766776529e-05,
+      "loss": 0.2365,
+      "step": 3088
+    },
+    {
+      "epoch": 0.652375923970433,
+      "grad_norm": 0.27676063776016235,
+      "learning_rate": 2.8020542758575757e-05,
+      "loss": 0.2367,
+      "step": 3089
+    },
+    {
+      "epoch": 0.6525871172122492,
+      "grad_norm": 0.3347737193107605,
+      "learning_rate": 2.799010794696416e-05,
+      "loss": 0.237,
+      "step": 3090
+    },
+    {
+      "epoch": 0.6527983104540654,
+      "grad_norm": 0.39903509616851807,
+      "learning_rate": 2.7959683246912614e-05,
+      "loss": 0.2327,
+      "step": 3091
+    },
+    {
+      "epoch": 0.6530095036958817,
+      "grad_norm": 0.3151611089706421,
+      "learning_rate": 2.7929268672398524e-05,
+      "loss": 0.2291,
+      "step": 3092
+    },
+    {
+      "epoch": 0.653220696937698,
+      "grad_norm": 0.2840743958950043,
+      "learning_rate": 2.7898864237394573e-05,
+      "loss": 0.2285,
+      "step": 3093
+    },
+    {
+      "epoch": 0.6534318901795143,
+      "grad_norm": 0.39600080251693726,
+      "learning_rate": 2.7868469955868927e-05,
+      "loss": 0.2308,
+      "step": 3094
+    },
+    {
+      "epoch": 0.6536430834213305,
+      "grad_norm": 0.33216971158981323,
+      "learning_rate": 2.7838085841785005e-05,
+      "loss": 0.2315,
+      "step": 3095
+    },
+    {
+      "epoch": 0.6538542766631468,
+      "grad_norm": 0.34497228264808655,
+      "learning_rate": 2.780771190910152e-05,
+      "loss": 0.2339,
+      "step": 3096
+    },
+    {
+      "epoch": 0.6540654699049631,
+      "grad_norm": 0.4864636957645416,
+      "learning_rate": 2.7777348171772604e-05,
+      "loss": 0.2308,
+      "step": 3097
+    },
+    {
+      "epoch": 0.6542766631467793,
+      "grad_norm": 0.38269147276878357,
+      "learning_rate": 2.774699464374765e-05,
+      "loss": 0.2301,
+      "step": 3098
+    },
+    {
+      "epoch": 0.6544878563885955,
+      "grad_norm": 0.32355353236198425,
+      "learning_rate": 2.7716651338971345e-05,
+      "loss": 0.227,
+      "step": 3099
+    },
+    {
+      "epoch": 0.6546990496304118,
+      "grad_norm": 0.3168317377567291,
+      "learning_rate": 2.7686318271383714e-05,
+      "loss": 0.2291,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6549102428722281,
+      "grad_norm": 0.3341856002807617,
+      "learning_rate": 2.7655995454920065e-05,
+      "loss": 0.232,
+      "step": 3101
+    },
+    {
+      "epoch": 0.6551214361140444,
+      "grad_norm": 0.32725241780281067,
+      "learning_rate": 2.7625682903510996e-05,
+      "loss": 0.2301,
+      "step": 3102
+    },
+    {
+      "epoch": 0.6553326293558606,
+      "grad_norm": 0.29094448685646057,
+      "learning_rate": 2.7595380631082384e-05,
+      "loss": 0.236,
+      "step": 3103
+    },
+    {
+      "epoch": 0.6555438225976769,
+      "grad_norm": 0.32380586862564087,
+      "learning_rate": 2.7565088651555397e-05,
+      "loss": 0.236,
+      "step": 3104
+    },
+    {
+      "epoch": 0.6557550158394931,
+      "grad_norm": 0.3369513750076294,
+      "learning_rate": 2.7534806978846465e-05,
+      "loss": 0.2327,
+      "step": 3105
+    },
+    {
+      "epoch": 0.6559662090813094,
+      "grad_norm": 0.3776058256626129,
+      "learning_rate": 2.7504535626867288e-05,
+      "loss": 0.233,
+      "step": 3106
+    },
+    {
+      "epoch": 0.6561774023231257,
+      "grad_norm": 0.3150985836982727,
+      "learning_rate": 2.747427460952482e-05,
+      "loss": 0.2331,
+      "step": 3107
+    },
+    {
+      "epoch": 0.6563885955649419,
+      "grad_norm": 0.39139804244041443,
+      "learning_rate": 2.7444023940721266e-05,
+      "loss": 0.2314,
+      "step": 3108
+    },
+    {
+      "epoch": 0.6565997888067582,
+      "grad_norm": 0.3785502314567566,
+      "learning_rate": 2.7413783634354072e-05,
+      "loss": 0.2338,
+      "step": 3109
+    },
+    {
+      "epoch": 0.6568109820485745,
+      "grad_norm": 0.6826025247573853,
+      "learning_rate": 2.7383553704315944e-05,
+      "loss": 0.2353,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6570221752903908,
+      "grad_norm": 0.359048068523407,
+      "learning_rate": 2.735333416449481e-05,
+      "loss": 0.2314,
+      "step": 3111
+    },
+    {
+      "epoch": 0.6572333685322069,
+      "grad_norm": 0.43808430433273315,
+      "learning_rate": 2.7323125028773788e-05,
+      "loss": 0.23,
+      "step": 3112
+    },
+    {
+      "epoch": 0.6574445617740232,
+      "grad_norm": 0.6064799427986145,
+      "learning_rate": 2.729292631103133e-05,
+      "loss": 0.2334,
+      "step": 3113
+    },
+    {
+      "epoch": 0.6576557550158395,
+      "grad_norm": 0.3116583526134491,
+      "learning_rate": 2.7262738025140944e-05,
+      "loss": 0.2325,
+      "step": 3114
+    },
+    {
+      "epoch": 0.6578669482576558,
+      "grad_norm": 0.35011252760887146,
+      "learning_rate": 2.7232560184971434e-05,
+      "loss": 0.2319,
+      "step": 3115
+    },
+    {
+      "epoch": 0.658078141499472,
+      "grad_norm": 0.31987500190734863,
+      "learning_rate": 2.7202392804386855e-05,
+      "loss": 0.2312,
+      "step": 3116
+    },
+    {
+      "epoch": 0.6582893347412883,
+      "grad_norm": 0.36038339138031006,
+      "learning_rate": 2.717223589724634e-05,
+      "loss": 0.233,
+      "step": 3117
+    },
+    {
+      "epoch": 0.6585005279831045,
+      "grad_norm": 0.35408055782318115,
+      "learning_rate": 2.7142089477404258e-05,
+      "loss": 0.2297,
+      "step": 3118
+    },
+    {
+      "epoch": 0.6587117212249208,
+      "grad_norm": 0.49867188930511475,
+      "learning_rate": 2.7111953558710236e-05,
+      "loss": 0.2346,
+      "step": 3119
+    },
+    {
+      "epoch": 0.6589229144667371,
+      "grad_norm": 0.39370080828666687,
+      "learning_rate": 2.7081828155008953e-05,
+      "loss": 0.234,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6591341077085533,
+      "grad_norm": 0.3373044729232788,
+      "learning_rate": 2.705171328014031e-05,
+      "loss": 0.2309,
+      "step": 3121
+    },
+    {
+      "epoch": 0.6593453009503696,
+      "grad_norm": 0.7185695767402649,
+      "learning_rate": 2.7021608947939438e-05,
+      "loss": 0.2283,
+      "step": 3122
+    },
+    {
+      "epoch": 0.6595564941921859,
+      "grad_norm": 0.39901304244995117,
+      "learning_rate": 2.699151517223649e-05,
+      "loss": 0.2328,
+      "step": 3123
+    },
+    {
+      "epoch": 0.6597676874340022,
+      "grad_norm": 0.4428529739379883,
+      "learning_rate": 2.6961431966856864e-05,
+      "loss": 0.2371,
+      "step": 3124
+    },
+    {
+      "epoch": 0.6599788806758183,
+      "grad_norm": 0.38921934366226196,
+      "learning_rate": 2.6931359345621127e-05,
+      "loss": 0.2377,
+      "step": 3125
+    },
+    {
+      "epoch": 0.6601900739176346,
+      "grad_norm": 0.31043264269828796,
+      "learning_rate": 2.6901297322344854e-05,
+      "loss": 0.2295,
+      "step": 3126
+    },
+    {
+      "epoch": 0.6604012671594509,
+      "grad_norm": 0.3671751320362091,
+      "learning_rate": 2.6871245910838893e-05,
+      "loss": 0.2303,
+      "step": 3127
+    },
+    {
+      "epoch": 0.6606124604012672,
+      "grad_norm": 0.3833148181438446,
+      "learning_rate": 2.6841205124909164e-05,
+      "loss": 0.2287,
+      "step": 3128
+    },
+    {
+      "epoch": 0.6608236536430834,
+      "grad_norm": 0.36144009232521057,
+      "learning_rate": 2.6811174978356624e-05,
+      "loss": 0.2293,
+      "step": 3129
+    },
+    {
+      "epoch": 0.6610348468848997,
+      "grad_norm": 0.320023775100708,
+      "learning_rate": 2.6781155484977493e-05,
+      "loss": 0.2287,
+      "step": 3130
+    },
+    {
+      "epoch": 0.661246040126716,
+      "grad_norm": 0.38651448488235474,
+      "learning_rate": 2.675114665856299e-05,
+      "loss": 0.2244,
+      "step": 3131
+    },
+    {
+      "epoch": 0.6614572333685322,
+      "grad_norm": 0.3392692804336548,
+      "learning_rate": 2.6721148512899475e-05,
+      "loss": 0.2283,
+      "step": 3132
+    },
+    {
+      "epoch": 0.6616684266103484,
+      "grad_norm": 0.35928863286972046,
+      "learning_rate": 2.6691161061768376e-05,
+      "loss": 0.2254,
+      "step": 3133
+    },
+    {
+      "epoch": 0.6618796198521647,
+      "grad_norm": 0.30086737871170044,
+      "learning_rate": 2.6661184318946225e-05,
+      "loss": 0.2277,
+      "step": 3134
+    },
+    {
+      "epoch": 0.662090813093981,
+      "grad_norm": 0.2801881730556488,
+      "learning_rate": 2.663121829820464e-05,
+      "loss": 0.2347,
+      "step": 3135
+    },
+    {
+      "epoch": 0.6623020063357973,
+      "grad_norm": 0.3459729850292206,
+      "learning_rate": 2.6601263013310297e-05,
+      "loss": 0.2292,
+      "step": 3136
+    },
+    {
+      "epoch": 0.6625131995776136,
+      "grad_norm": 0.36526989936828613,
+      "learning_rate": 2.657131847802494e-05,
+      "loss": 0.2325,
+      "step": 3137
+    },
+    {
+      "epoch": 0.6627243928194297,
+      "grad_norm": 0.3507133424282074,
+      "learning_rate": 2.6541384706105404e-05,
+      "loss": 0.2344,
+      "step": 3138
+    },
+    {
+      "epoch": 0.662935586061246,
+      "grad_norm": 0.37101882696151733,
+      "learning_rate": 2.6511461711303543e-05,
+      "loss": 0.2328,
+      "step": 3139
+    },
+    {
+      "epoch": 0.6631467793030623,
+      "grad_norm": 0.3297421932220459,
+      "learning_rate": 2.6481549507366265e-05,
+      "loss": 0.2262,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6633579725448786,
+      "grad_norm": 0.34943827986717224,
+      "learning_rate": 2.645164810803555e-05,
+      "loss": 0.2271,
+      "step": 3141
+    },
+    {
+      "epoch": 0.6635691657866948,
+      "grad_norm": 0.303551584482193,
+      "learning_rate": 2.6421757527048373e-05,
+      "loss": 0.2275,
+      "step": 3142
+    },
+    {
+      "epoch": 0.6637803590285111,
+      "grad_norm": 0.3352566361427307,
+      "learning_rate": 2.6391877778136782e-05,
+      "loss": 0.2265,
+      "step": 3143
+    },
+    {
+      "epoch": 0.6639915522703274,
+      "grad_norm": 0.3000815808773041,
+      "learning_rate": 2.6362008875027804e-05,
+      "loss": 0.2261,
+      "step": 3144
+    },
+    {
+      "epoch": 0.6642027455121436,
+      "grad_norm": 0.3332419693470001,
+      "learning_rate": 2.6332150831443524e-05,
+      "loss": 0.2268,
+      "step": 3145
+    },
+    {
+      "epoch": 0.6644139387539598,
+      "grad_norm": 0.3122781813144684,
+      "learning_rate": 2.6302303661101e-05,
+      "loss": 0.2259,
+      "step": 3146
+    },
+    {
+      "epoch": 0.6646251319957761,
+      "grad_norm": 0.31545373797416687,
+      "learning_rate": 2.627246737771234e-05,
+      "loss": 0.237,
+      "step": 3147
+    },
+    {
+      "epoch": 0.6648363252375924,
+      "grad_norm": 0.30331748723983765,
+      "learning_rate": 2.6242641994984606e-05,
+      "loss": 0.2313,
+      "step": 3148
+    },
+    {
+      "epoch": 0.6650475184794087,
+      "grad_norm": 0.35763904452323914,
+      "learning_rate": 2.6212827526619877e-05,
+      "loss": 0.2326,
+      "step": 3149
+    },
+    {
+      "epoch": 0.665258711721225,
+      "grad_norm": 0.3380158841609955,
+      "learning_rate": 2.6183023986315202e-05,
+      "loss": 0.2272,
+      "step": 3150
+    },
+    {
+      "epoch": 0.6654699049630411,
+      "grad_norm": 0.292534202337265,
+      "learning_rate": 2.6153231387762637e-05,
+      "loss": 0.2306,
+      "step": 3151
+    },
+    {
+      "epoch": 0.6656810982048574,
+      "grad_norm": 0.33485734462738037,
+      "learning_rate": 2.6123449744649187e-05,
+      "loss": 0.2269,
+      "step": 3152
+    },
+    {
+      "epoch": 0.6658922914466737,
+      "grad_norm": 0.3435097932815552,
+      "learning_rate": 2.609367907065682e-05,
+      "loss": 0.2309,
+      "step": 3153
+    },
+    {
+      "epoch": 0.66610348468849,
+      "grad_norm": 0.31536155939102173,
+      "learning_rate": 2.606391937946247e-05,
+      "loss": 0.2279,
+      "step": 3154
+    },
+    {
+      "epoch": 0.6663146779303062,
+      "grad_norm": 0.32368144392967224,
+      "learning_rate": 2.6034170684738064e-05,
+      "loss": 0.2277,
+      "step": 3155
+    },
+    {
+      "epoch": 0.6665258711721225,
+      "grad_norm": 0.3069643974304199,
+      "learning_rate": 2.600443300015044e-05,
+      "loss": 0.2332,
+      "step": 3156
+    },
+    {
+      "epoch": 0.6667370644139388,
+      "grad_norm": 0.3046828806400299,
+      "learning_rate": 2.597470633936133e-05,
+      "loss": 0.2334,
+      "step": 3157
+    },
+    {
+      "epoch": 0.666948257655755,
+      "grad_norm": 0.3183800280094147,
+      "learning_rate": 2.5944990716027507e-05,
+      "loss": 0.2285,
+      "step": 3158
+    },
+    {
+      "epoch": 0.6671594508975712,
+      "grad_norm": 0.35319313406944275,
+      "learning_rate": 2.591528614380062e-05,
+      "loss": 0.2288,
+      "step": 3159
+    },
+    {
+      "epoch": 0.6673706441393875,
+      "grad_norm": 0.9097281098365784,
+      "learning_rate": 2.588559263632719e-05,
+      "loss": 0.2287,
+      "step": 3160
+    },
+    {
+      "epoch": 0.6675818373812038,
+      "grad_norm": 0.343339204788208,
+      "learning_rate": 2.585591020724875e-05,
+      "loss": 0.2383,
+      "step": 3161
+    },
+    {
+      "epoch": 0.6677930306230201,
+      "grad_norm": 0.28204014897346497,
+      "learning_rate": 2.5826238870201708e-05,
+      "loss": 0.2292,
+      "step": 3162
+    },
+    {
+      "epoch": 0.6680042238648364,
+      "grad_norm": 0.3989930748939514,
+      "learning_rate": 2.5796578638817303e-05,
+      "loss": 0.2277,
+      "step": 3163
+    },
+    {
+      "epoch": 0.6682154171066526,
+      "grad_norm": 0.2759205102920532,
+      "learning_rate": 2.57669295267218e-05,
+      "loss": 0.2281,
+      "step": 3164
+    },
+    {
+      "epoch": 0.6684266103484688,
+      "grad_norm": 0.3290959894657135,
+      "learning_rate": 2.573729154753629e-05,
+      "loss": 0.2257,
+      "step": 3165
+    },
+    {
+      "epoch": 0.6686378035902851,
+      "grad_norm": 0.313585102558136,
+      "learning_rate": 2.5707664714876688e-05,
+      "loss": 0.2318,
+      "step": 3166
+    },
+    {
+      "epoch": 0.6688489968321014,
+      "grad_norm": 0.4770658612251282,
+      "learning_rate": 2.5678049042353912e-05,
+      "loss": 0.2345,
+      "step": 3167
+    },
+    {
+      "epoch": 0.6690601900739176,
+      "grad_norm": 0.36731061339378357,
+      "learning_rate": 2.5648444543573697e-05,
+      "loss": 0.2321,
+      "step": 3168
+    },
+    {
+      "epoch": 0.6692713833157339,
+      "grad_norm": 0.7849541902542114,
+      "learning_rate": 2.5618851232136574e-05,
+      "loss": 0.226,
+      "step": 3169
+    },
+    {
+      "epoch": 0.6694825765575502,
+      "grad_norm": 0.3591426908969879,
+      "learning_rate": 2.5589269121638066e-05,
+      "loss": 0.2325,
+      "step": 3170
+    },
+    {
+      "epoch": 0.6696937697993665,
+      "grad_norm": 0.37090346217155457,
+      "learning_rate": 2.5559698225668456e-05,
+      "loss": 0.2318,
+      "step": 3171
+    },
+    {
+      "epoch": 0.6699049630411826,
+      "grad_norm": 0.45854651927948,
+      "learning_rate": 2.553013855781291e-05,
+      "loss": 0.2372,
+      "step": 3172
+    },
+    {
+      "epoch": 0.6701161562829989,
+      "grad_norm": 0.3254079520702362,
+      "learning_rate": 2.5500590131651424e-05,
+      "loss": 0.23,
+      "step": 3173
+    },
+    {
+      "epoch": 0.6703273495248152,
+      "grad_norm": 0.3534514307975769,
+      "learning_rate": 2.5471052960758846e-05,
+      "loss": 0.2373,
+      "step": 3174
+    },
+    {
+      "epoch": 0.6705385427666315,
+      "grad_norm": 0.35699227452278137,
+      "learning_rate": 2.544152705870483e-05,
+      "loss": 0.2368,
+      "step": 3175
+    },
+    {
+      "epoch": 0.6707497360084478,
+      "grad_norm": 0.6863391995429993,
+      "learning_rate": 2.5412012439053863e-05,
+      "loss": 0.2352,
+      "step": 3176
+    },
+    {
+      "epoch": 0.670960929250264,
+      "grad_norm": 0.46611514687538147,
+      "learning_rate": 2.5382509115365255e-05,
+      "loss": 0.2356,
+      "step": 3177
+    },
+    {
+      "epoch": 0.6711721224920802,
+      "grad_norm": 0.30108892917633057,
+      "learning_rate": 2.535301710119312e-05,
+      "loss": 0.2283,
+      "step": 3178
+    },
+    {
+      "epoch": 0.6713833157338965,
+      "grad_norm": 0.37964868545532227,
+      "learning_rate": 2.532353641008639e-05,
+      "loss": 0.2327,
+      "step": 3179
+    },
+    {
+      "epoch": 0.6715945089757128,
+      "grad_norm": 0.3871065080165863,
+      "learning_rate": 2.5294067055588765e-05,
+      "loss": 0.2338,
+      "step": 3180
+    },
+    {
+      "epoch": 0.671805702217529,
+      "grad_norm": 0.3388059735298157,
+      "learning_rate": 2.5264609051238775e-05,
+      "loss": 0.2313,
+      "step": 3181
+    },
+    {
+      "epoch": 0.6720168954593453,
+      "grad_norm": 0.3428068161010742,
+      "learning_rate": 2.5235162410569695e-05,
+      "loss": 0.2262,
+      "step": 3182
+    },
+    {
+      "epoch": 0.6722280887011616,
+      "grad_norm": 0.3202349543571472,
+      "learning_rate": 2.5205727147109615e-05,
+      "loss": 0.2343,
+      "step": 3183
+    },
+    {
+      "epoch": 0.6724392819429779,
+      "grad_norm": 0.3188331127166748,
+      "learning_rate": 2.5176303274381375e-05,
+      "loss": 0.2343,
+      "step": 3184
+    },
+    {
+      "epoch": 0.672650475184794,
+      "grad_norm": 0.3062594532966614,
+      "learning_rate": 2.5146890805902573e-05,
+      "loss": 0.2307,
+      "step": 3185
+    },
+    {
+      "epoch": 0.6728616684266103,
+      "grad_norm": 0.3167622685432434,
+      "learning_rate": 2.5117489755185646e-05,
+      "loss": 0.2289,
+      "step": 3186
+    },
+    {
+      "epoch": 0.6730728616684266,
+      "grad_norm": 0.320989727973938,
+      "learning_rate": 2.5088100135737668e-05,
+      "loss": 0.2365,
+      "step": 3187
+    },
+    {
+      "epoch": 0.6732840549102429,
+      "grad_norm": 0.31094008684158325,
+      "learning_rate": 2.505872196106051e-05,
+      "loss": 0.2317,
+      "step": 3188
+    },
+    {
+      "epoch": 0.6734952481520592,
+      "grad_norm": 0.3370620906352997,
+      "learning_rate": 2.502935524465086e-05,
+      "loss": 0.2291,
+      "step": 3189
+    },
+    {
+      "epoch": 0.6737064413938754,
+      "grad_norm": 0.3751080334186554,
+      "learning_rate": 2.500000000000001e-05,
+      "loss": 0.2303,
+      "step": 3190
+    },
+    {
+      "epoch": 0.6739176346356917,
+      "grad_norm": 0.3223649859428406,
+      "learning_rate": 2.4970656240594055e-05,
+      "loss": 0.2294,
+      "step": 3191
+    },
+    {
+      "epoch": 0.6741288278775079,
+      "grad_norm": 0.3722957670688629,
+      "learning_rate": 2.4941323979913854e-05,
+      "loss": 0.2306,
+      "step": 3192
+    },
+    {
+      "epoch": 0.6743400211193242,
+      "grad_norm": 0.32018133997917175,
+      "learning_rate": 2.491200323143489e-05,
+      "loss": 0.2333,
+      "step": 3193
+    },
+    {
+      "epoch": 0.6745512143611404,
+      "grad_norm": 0.3401433229446411,
+      "learning_rate": 2.4882694008627388e-05,
+      "loss": 0.2308,
+      "step": 3194
+    },
+    {
+      "epoch": 0.6747624076029567,
+      "grad_norm": 0.32853972911834717,
+      "learning_rate": 2.4853396324956358e-05,
+      "loss": 0.2283,
+      "step": 3195
+    },
+    {
+      "epoch": 0.674973600844773,
+      "grad_norm": 0.3435714840888977,
+      "learning_rate": 2.4824110193881382e-05,
+      "loss": 0.2332,
+      "step": 3196
+    },
+    {
+      "epoch": 0.6751847940865893,
+      "grad_norm": 0.29699403047561646,
+      "learning_rate": 2.47948356288568e-05,
+      "loss": 0.2364,
+      "step": 3197
+    },
+    {
+      "epoch": 0.6753959873284054,
+      "grad_norm": 0.3139247000217438,
+      "learning_rate": 2.4765572643331687e-05,
+      "loss": 0.2308,
+      "step": 3198
+    },
+    {
+      "epoch": 0.6756071805702217,
+      "grad_norm": 0.3940269649028778,
+      "learning_rate": 2.4736321250749667e-05,
+      "loss": 0.2283,
+      "step": 3199
+    },
+    {
+      "epoch": 0.675818373812038,
+      "grad_norm": 0.30655792355537415,
+      "learning_rate": 2.470708146454918e-05,
+      "loss": 0.2219,
+      "step": 3200
+    },
+    {
+      "epoch": 0.675818373812038,
+      "eval_train_split_loss": 0.3168491721153259,
+      "eval_train_split_runtime": 7.4055,
+      "eval_train_split_samples_per_second": 1.35,
+      "eval_train_split_steps_per_second": 0.135,
+      "step": 3200
+    },
+    {
+      "epoch": 0.6760295670538543,
+      "grad_norm": 0.3162849247455597,
+      "learning_rate": 2.4677853298163263e-05,
+      "loss": 0.2258,
+      "step": 3201
+    },
+    {
+      "epoch": 0.6762407602956705,
+      "grad_norm": 0.3467232286930084,
+      "learning_rate": 2.4648636765019573e-05,
+      "loss": 0.2273,
+      "step": 3202
+    },
+    {
+      "epoch": 0.6764519535374868,
+      "grad_norm": 0.30403372645378113,
+      "learning_rate": 2.461943187854052e-05,
+      "loss": 0.2205,
+      "step": 3203
+    },
+    {
+      "epoch": 0.6766631467793031,
+      "grad_norm": 0.2926388084888458,
+      "learning_rate": 2.4590238652143122e-05,
+      "loss": 0.225,
+      "step": 3204
+    },
+    {
+      "epoch": 0.6768743400211193,
+      "grad_norm": 0.302139014005661,
+      "learning_rate": 2.456105709923897e-05,
+      "loss": 0.2289,
+      "step": 3205
+    },
+    {
+      "epoch": 0.6770855332629356,
+      "grad_norm": 0.31515467166900635,
+      "learning_rate": 2.4531887233234424e-05,
+      "loss": 0.2323,
+      "step": 3206
+    },
+    {
+      "epoch": 0.6772967265047518,
+      "grad_norm": 0.39052391052246094,
+      "learning_rate": 2.450272906753039e-05,
+      "loss": 0.2277,
+      "step": 3207
+    },
+    {
+      "epoch": 0.6775079197465681,
+      "grad_norm": 0.45170503854751587,
+      "learning_rate": 2.4473582615522366e-05,
+      "loss": 0.2241,
+      "step": 3208
+    },
+    {
+      "epoch": 0.6777191129883844,
+      "grad_norm": 0.3238046169281006,
+      "learning_rate": 2.444444789060058e-05,
+      "loss": 0.2285,
+      "step": 3209
+    },
+    {
+      "epoch": 0.6779303062302007,
+      "grad_norm": 0.2947711646556854,
+      "learning_rate": 2.4415324906149778e-05,
+      "loss": 0.2273,
+      "step": 3210
+    },
+    {
+      "epoch": 0.6781414994720169,
+      "grad_norm": 0.489774614572525,
+      "learning_rate": 2.4386213675549358e-05,
+      "loss": 0.2299,
+      "step": 3211
+    },
+    {
+      "epoch": 0.6783526927138331,
+      "grad_norm": 0.4315170347690582,
+      "learning_rate": 2.4357114212173292e-05,
+      "loss": 0.2285,
+      "step": 3212
+    },
+    {
+      "epoch": 0.6785638859556494,
+      "grad_norm": 0.3119152784347534,
+      "learning_rate": 2.4328026529390174e-05,
+      "loss": 0.221,
+      "step": 3213
+    },
+    {
+      "epoch": 0.6787750791974657,
+      "grad_norm": 0.3440711200237274,
+      "learning_rate": 2.4298950640563156e-05,
+      "loss": 0.2232,
+      "step": 3214
+    },
+    {
+      "epoch": 0.6789862724392819,
+      "grad_norm": 0.3760956823825836,
+      "learning_rate": 2.4269886559049993e-05,
+      "loss": 0.2266,
+      "step": 3215
+    },
+    {
+      "epoch": 0.6791974656810982,
+      "grad_norm": 0.43470877408981323,
+      "learning_rate": 2.4240834298203004e-05,
+      "loss": 0.224,
+      "step": 3216
+    },
+    {
+      "epoch": 0.6794086589229145,
+      "grad_norm": 0.3479612469673157,
+      "learning_rate": 2.4211793871369086e-05,
+      "loss": 0.2285,
+      "step": 3217
+    },
+    {
+      "epoch": 0.6796198521647308,
+      "grad_norm": 0.3493119478225708,
+      "learning_rate": 2.4182765291889697e-05,
+      "loss": 0.2331,
+      "step": 3218
+    },
+    {
+      "epoch": 0.679831045406547,
+      "grad_norm": 0.3561949133872986,
+      "learning_rate": 2.415374857310085e-05,
+      "loss": 0.2272,
+      "step": 3219
+    },
+    {
+      "epoch": 0.6800422386483632,
+      "grad_norm": 0.3391503393650055,
+      "learning_rate": 2.4124743728333105e-05,
+      "loss": 0.2268,
+      "step": 3220
+    },
+    {
+      "epoch": 0.6802534318901795,
+      "grad_norm": 0.36419743299484253,
+      "learning_rate": 2.409575077091158e-05,
+      "loss": 0.2262,
+      "step": 3221
+    },
+    {
+      "epoch": 0.6804646251319958,
+      "grad_norm": 0.37757381796836853,
+      "learning_rate": 2.4066769714155914e-05,
+      "loss": 0.23,
+      "step": 3222
+    },
+    {
+      "epoch": 0.6806758183738121,
+      "grad_norm": 0.38468441367149353,
+      "learning_rate": 2.4037800571380297e-05,
+      "loss": 0.2286,
+      "step": 3223
+    },
+    {
+      "epoch": 0.6808870116156283,
+      "grad_norm": 0.27444544434547424,
+      "learning_rate": 2.400884335589343e-05,
+      "loss": 0.2291,
+      "step": 3224
+    },
+    {
+      "epoch": 0.6810982048574445,
+      "grad_norm": 0.3002014756202698,
+      "learning_rate": 2.3979898080998547e-05,
+      "loss": 0.2281,
+      "step": 3225
+    },
+    {
+      "epoch": 0.6813093980992608,
+      "grad_norm": 0.4866010844707489,
+      "learning_rate": 2.3950964759993393e-05,
+      "loss": 0.2334,
+      "step": 3226
+    },
+    {
+      "epoch": 0.6815205913410771,
+      "grad_norm": 0.33552074432373047,
+      "learning_rate": 2.392204340617019e-05,
+      "loss": 0.2232,
+      "step": 3227
+    },
+    {
+      "epoch": 0.6817317845828933,
+      "grad_norm": 0.45047619938850403,
+      "learning_rate": 2.3893134032815757e-05,
+      "loss": 0.2299,
+      "step": 3228
+    },
+    {
+      "epoch": 0.6819429778247096,
+      "grad_norm": 0.3571729063987732,
+      "learning_rate": 2.3864236653211286e-05,
+      "loss": 0.2302,
+      "step": 3229
+    },
+    {
+      "epoch": 0.6821541710665259,
+      "grad_norm": 0.33792027831077576,
+      "learning_rate": 2.3835351280632513e-05,
+      "loss": 0.2297,
+      "step": 3230
+    },
+    {
+      "epoch": 0.6823653643083422,
+      "grad_norm": 0.3133138418197632,
+      "learning_rate": 2.3806477928349713e-05,
+      "loss": 0.2256,
+      "step": 3231
+    },
+    {
+      "epoch": 0.6825765575501584,
+      "grad_norm": 0.33933839201927185,
+      "learning_rate": 2.377761660962754e-05,
+      "loss": 0.2211,
+      "step": 3232
+    },
+    {
+      "epoch": 0.6827877507919746,
+      "grad_norm": 0.3781851828098297,
+      "learning_rate": 2.3748767337725163e-05,
+      "loss": 0.2258,
+      "step": 3233
+    },
+    {
+      "epoch": 0.6829989440337909,
+      "grad_norm": 0.3086962401866913,
+      "learning_rate": 2.371993012589626e-05,
+      "loss": 0.2301,
+      "step": 3234
+    },
+    {
+      "epoch": 0.6832101372756072,
+      "grad_norm": 0.3407095670700073,
+      "learning_rate": 2.3691104987388924e-05,
+      "loss": 0.2329,
+      "step": 3235
+    },
+    {
+      "epoch": 0.6834213305174235,
+      "grad_norm": 0.3484911024570465,
+      "learning_rate": 2.366229193544565e-05,
+      "loss": 0.2305,
+      "step": 3236
+    },
+    {
+      "epoch": 0.6836325237592397,
+      "grad_norm": 0.3641394376754761,
+      "learning_rate": 2.3633490983303495e-05,
+      "loss": 0.2251,
+      "step": 3237
+    },
+    {
+      "epoch": 0.683843717001056,
+      "grad_norm": 0.37976381182670593,
+      "learning_rate": 2.3604702144193895e-05,
+      "loss": 0.2305,
+      "step": 3238
+    },
+    {
+      "epoch": 0.6840549102428722,
+      "grad_norm": 0.3571474254131317,
+      "learning_rate": 2.3575925431342676e-05,
+      "loss": 0.2248,
+      "step": 3239
+    },
+    {
+      "epoch": 0.6842661034846885,
+      "grad_norm": 0.41629183292388916,
+      "learning_rate": 2.3547160857970197e-05,
+      "loss": 0.2347,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6844772967265047,
+      "grad_norm": 0.40144386887550354,
+      "learning_rate": 2.3518408437291157e-05,
+      "loss": 0.2283,
+      "step": 3241
+    },
+    {
+      "epoch": 0.684688489968321,
+      "grad_norm": 0.35487326979637146,
+      "learning_rate": 2.34896681825147e-05,
+      "loss": 0.2253,
+      "step": 3242
+    },
+    {
+      "epoch": 0.6848996832101373,
+      "grad_norm": 0.318873792886734,
+      "learning_rate": 2.346094010684439e-05,
+      "loss": 0.2251,
+      "step": 3243
+    },
+    {
+      "epoch": 0.6851108764519536,
+      "grad_norm": 0.28845155239105225,
+      "learning_rate": 2.343222422347819e-05,
+      "loss": 0.2304,
+      "step": 3244
+    },
+    {
+      "epoch": 0.6853220696937699,
+      "grad_norm": 0.30928361415863037,
+      "learning_rate": 2.340352054560844e-05,
+      "loss": 0.2283,
+      "step": 3245
+    },
+    {
+      "epoch": 0.685533262935586,
+      "grad_norm": 0.3428705632686615,
+      "learning_rate": 2.3374829086421905e-05,
+      "loss": 0.2225,
+      "step": 3246
+    },
+    {
+      "epoch": 0.6857444561774023,
+      "grad_norm": 0.38881027698516846,
+      "learning_rate": 2.334614985909972e-05,
+      "loss": 0.228,
+      "step": 3247
+    },
+    {
+      "epoch": 0.6859556494192186,
+      "grad_norm": 0.3388351798057556,
+      "learning_rate": 2.33174828768174e-05,
+      "loss": 0.2308,
+      "step": 3248
+    },
+    {
+      "epoch": 0.6861668426610349,
+      "grad_norm": 0.36002230644226074,
+      "learning_rate": 2.3288828152744834e-05,
+      "loss": 0.2291,
+      "step": 3249
+    },
+    {
+      "epoch": 0.6863780359028511,
+      "grad_norm": 0.2930295765399933,
+      "learning_rate": 2.3260185700046294e-05,
+      "loss": 0.2232,
+      "step": 3250
+    },
+    {
+      "epoch": 0.6865892291446674,
+      "grad_norm": 0.3325492739677429,
+      "learning_rate": 2.3231555531880394e-05,
+      "loss": 0.2284,
+      "step": 3251
+    },
+    {
+      "epoch": 0.6868004223864836,
+      "grad_norm": 0.25538885593414307,
+      "learning_rate": 2.320293766140012e-05,
+      "loss": 0.2288,
+      "step": 3252
+    },
+    {
+      "epoch": 0.6870116156282999,
+      "grad_norm": 0.2882811725139618,
+      "learning_rate": 2.3174332101752798e-05,
+      "loss": 0.2262,
+      "step": 3253
+    },
+    {
+      "epoch": 0.6872228088701161,
+      "grad_norm": 0.5493389964103699,
+      "learning_rate": 2.31457388660801e-05,
+      "loss": 0.2292,
+      "step": 3254
+    },
+    {
+      "epoch": 0.6874340021119324,
+      "grad_norm": 0.3884214162826538,
+      "learning_rate": 2.311715796751805e-05,
+      "loss": 0.2293,
+      "step": 3255
+    },
+    {
+      "epoch": 0.6876451953537487,
+      "grad_norm": 0.3268096148967743,
+      "learning_rate": 2.3088589419196983e-05,
+      "loss": 0.2276,
+      "step": 3256
+    },
+    {
+      "epoch": 0.687856388595565,
+      "grad_norm": 0.379686564207077,
+      "learning_rate": 2.3060033234241567e-05,
+      "loss": 0.2322,
+      "step": 3257
+    },
+    {
+      "epoch": 0.6880675818373813,
+      "grad_norm": 0.30940645933151245,
+      "learning_rate": 2.3031489425770782e-05,
+      "loss": 0.2281,
+      "step": 3258
+    },
+    {
+      "epoch": 0.6882787750791974,
+      "grad_norm": 0.29148492217063904,
+      "learning_rate": 2.3002958006897984e-05,
+      "loss": 0.2286,
+      "step": 3259
+    },
+    {
+      "epoch": 0.6884899683210137,
+      "grad_norm": 0.41504964232444763,
+      "learning_rate": 2.2974438990730734e-05,
+      "loss": 0.2248,
+      "step": 3260
+    },
+    {
+      "epoch": 0.68870116156283,
+      "grad_norm": 0.29221606254577637,
+      "learning_rate": 2.2945932390370944e-05,
+      "loss": 0.2271,
+      "step": 3261
+    },
+    {
+      "epoch": 0.6889123548046463,
+      "grad_norm": 0.3474777638912201,
+      "learning_rate": 2.291743821891489e-05,
+      "loss": 0.2274,
+      "step": 3262
+    },
+    {
+      "epoch": 0.6891235480464625,
+      "grad_norm": 0.36578837037086487,
+      "learning_rate": 2.2888956489453005e-05,
+      "loss": 0.2349,
+      "step": 3263
+    },
+    {
+      "epoch": 0.6893347412882788,
+      "grad_norm": 0.5404615998268127,
+      "learning_rate": 2.286048721507008e-05,
+      "loss": 0.2295,
+      "step": 3264
+    },
+    {
+      "epoch": 0.689545934530095,
+      "grad_norm": 0.3386414051055908,
+      "learning_rate": 2.283203040884524e-05,
+      "loss": 0.2351,
+      "step": 3265
+    },
+    {
+      "epoch": 0.6897571277719113,
+      "grad_norm": 0.4688723683357239,
+      "learning_rate": 2.2803586083851757e-05,
+      "loss": 0.2305,
+      "step": 3266
+    },
+    {
+      "epoch": 0.6899683210137275,
+      "grad_norm": 0.3480318784713745,
+      "learning_rate": 2.277515425315724e-05,
+      "loss": 0.2283,
+      "step": 3267
+    },
+    {
+      "epoch": 0.6901795142555438,
+      "grad_norm": 0.3644017279148102,
+      "learning_rate": 2.2746734929823594e-05,
+      "loss": 0.2287,
+      "step": 3268
+    },
+    {
+      "epoch": 0.6903907074973601,
+      "grad_norm": 0.37339574098587036,
+      "learning_rate": 2.271832812690689e-05,
+      "loss": 0.2271,
+      "step": 3269
+    },
+    {
+      "epoch": 0.6906019007391764,
+      "grad_norm": 0.4077439606189728,
+      "learning_rate": 2.268993385745749e-05,
+      "loss": 0.2227,
+      "step": 3270
+    },
+    {
+      "epoch": 0.6908130939809926,
+      "grad_norm": 0.4288909137248993,
+      "learning_rate": 2.2661552134520052e-05,
+      "loss": 0.2294,
+      "step": 3271
+    },
+    {
+      "epoch": 0.6910242872228088,
+      "grad_norm": 0.3805524706840515,
+      "learning_rate": 2.2633182971133343e-05,
+      "loss": 0.2354,
+      "step": 3272
+    },
+    {
+      "epoch": 0.6912354804646251,
+      "grad_norm": 0.39135491847991943,
+      "learning_rate": 2.260482638033049e-05,
+      "loss": 0.229,
+      "step": 3273
+    },
+    {
+      "epoch": 0.6914466737064414,
+      "grad_norm": 0.3169142007827759,
+      "learning_rate": 2.2576482375138797e-05,
+      "loss": 0.2249,
+      "step": 3274
+    },
+    {
+      "epoch": 0.6916578669482577,
+      "grad_norm": 0.33123263716697693,
+      "learning_rate": 2.254815096857971e-05,
+      "loss": 0.2275,
+      "step": 3275
+    },
+    {
+      "epoch": 0.6918690601900739,
+      "grad_norm": 0.3194729685783386,
+      "learning_rate": 2.251983217366902e-05,
+      "loss": 0.229,
+      "step": 3276
+    },
+    {
+      "epoch": 0.6920802534318902,
+      "grad_norm": 0.3362564444541931,
+      "learning_rate": 2.249152600341665e-05,
+      "loss": 0.2293,
+      "step": 3277
+    },
+    {
+      "epoch": 0.6922914466737065,
+      "grad_norm": 0.3883498013019562,
+      "learning_rate": 2.2463232470826685e-05,
+      "loss": 0.2283,
+      "step": 3278
+    },
+    {
+      "epoch": 0.6925026399155227,
+      "grad_norm": 0.33738094568252563,
+      "learning_rate": 2.2434951588897502e-05,
+      "loss": 0.2341,
+      "step": 3279
+    },
+    {
+      "epoch": 0.6927138331573389,
+      "grad_norm": 0.33029043674468994,
+      "learning_rate": 2.2406683370621618e-05,
+      "loss": 0.2356,
+      "step": 3280
+    },
+    {
+      "epoch": 0.6929250263991552,
+      "grad_norm": 0.454550564289093,
+      "learning_rate": 2.2378427828985666e-05,
+      "loss": 0.2329,
+      "step": 3281
+    },
+    {
+      "epoch": 0.6931362196409715,
+      "grad_norm": 0.331429660320282,
+      "learning_rate": 2.2350184976970584e-05,
+      "loss": 0.2277,
+      "step": 3282
+    },
+    {
+      "epoch": 0.6933474128827878,
+      "grad_norm": 0.3342190086841583,
+      "learning_rate": 2.2321954827551405e-05,
+      "loss": 0.2261,
+      "step": 3283
+    },
+    {
+      "epoch": 0.693558606124604,
+      "grad_norm": 0.27948811650276184,
+      "learning_rate": 2.2293737393697296e-05,
+      "loss": 0.2256,
+      "step": 3284
+    },
+    {
+      "epoch": 0.6937697993664202,
+      "grad_norm": 0.3886413872241974,
+      "learning_rate": 2.2265532688371658e-05,
+      "loss": 0.2224,
+      "step": 3285
+    },
+    {
+      "epoch": 0.6939809926082365,
+      "grad_norm": 0.3375164270401001,
+      "learning_rate": 2.2237340724532007e-05,
+      "loss": 0.2244,
+      "step": 3286
+    },
+    {
+      "epoch": 0.6941921858500528,
+      "grad_norm": 0.2964361310005188,
+      "learning_rate": 2.220916151513e-05,
+      "loss": 0.2269,
+      "step": 3287
+    },
+    {
+      "epoch": 0.6944033790918691,
+      "grad_norm": 0.33201149106025696,
+      "learning_rate": 2.2180995073111445e-05,
+      "loss": 0.2297,
+      "step": 3288
+    },
+    {
+      "epoch": 0.6946145723336853,
+      "grad_norm": 0.3483050465583801,
+      "learning_rate": 2.2152841411416274e-05,
+      "loss": 0.2242,
+      "step": 3289
+    },
+    {
+      "epoch": 0.6948257655755016,
+      "grad_norm": 0.34519803524017334,
+      "learning_rate": 2.2124700542978567e-05,
+      "loss": 0.2277,
+      "step": 3290
+    },
+    {
+      "epoch": 0.6950369588173179,
+      "grad_norm": 0.28365299105644226,
+      "learning_rate": 2.2096572480726503e-05,
+      "loss": 0.2261,
+      "step": 3291
+    },
+    {
+      "epoch": 0.6952481520591342,
+      "grad_norm": 0.2706265151500702,
+      "learning_rate": 2.2068457237582397e-05,
+      "loss": 0.2281,
+      "step": 3292
+    },
+    {
+      "epoch": 0.6954593453009503,
+      "grad_norm": 0.44888627529144287,
+      "learning_rate": 2.2040354826462668e-05,
+      "loss": 0.2265,
+      "step": 3293
+    },
+    {
+      "epoch": 0.6956705385427666,
+      "grad_norm": 0.2860288918018341,
+      "learning_rate": 2.2012265260277833e-05,
+      "loss": 0.2278,
+      "step": 3294
+    },
+    {
+      "epoch": 0.6958817317845829,
+      "grad_norm": 0.34534579515457153,
+      "learning_rate": 2.1984188551932512e-05,
+      "loss": 0.2307,
+      "step": 3295
+    },
+    {
+      "epoch": 0.6960929250263992,
+      "grad_norm": 0.3213375210762024,
+      "learning_rate": 2.1956124714325426e-05,
+      "loss": 0.2298,
+      "step": 3296
+    },
+    {
+      "epoch": 0.6963041182682154,
+      "grad_norm": 0.30781665444374084,
+      "learning_rate": 2.1928073760349378e-05,
+      "loss": 0.2256,
+      "step": 3297
+    },
+    {
+      "epoch": 0.6965153115100317,
+      "grad_norm": 0.2810186445713043,
+      "learning_rate": 2.1900035702891246e-05,
+      "loss": 0.2273,
+      "step": 3298
+    },
+    {
+      "epoch": 0.6967265047518479,
+      "grad_norm": 0.32521557807922363,
+      "learning_rate": 2.187201055483199e-05,
+      "loss": 0.2282,
+      "step": 3299
+    },
+    {
+      "epoch": 0.6969376979936642,
+      "grad_norm": 0.2792641222476959,
+      "learning_rate": 2.1843998329046622e-05,
+      "loss": 0.2233,
+      "step": 3300
+    },
+    {
+      "epoch": 0.6971488912354805,
+      "grad_norm": 0.31173834204673767,
+      "learning_rate": 2.1815999038404284e-05,
+      "loss": 0.2261,
+      "step": 3301
+    },
+    {
+      "epoch": 0.6973600844772967,
+      "grad_norm": 0.3326112627983093,
+      "learning_rate": 2.1788012695768074e-05,
+      "loss": 0.2282,
+      "step": 3302
+    },
+    {
+      "epoch": 0.697571277719113,
+      "grad_norm": 0.28016573190689087,
+      "learning_rate": 2.176003931399519e-05,
+      "loss": 0.2271,
+      "step": 3303
+    },
+    {
+      "epoch": 0.6977824709609293,
+      "grad_norm": 0.32224032282829285,
+      "learning_rate": 2.1732078905936924e-05,
+      "loss": 0.2275,
+      "step": 3304
+    },
+    {
+      "epoch": 0.6979936642027456,
+      "grad_norm": 0.289597749710083,
+      "learning_rate": 2.170413148443852e-05,
+      "loss": 0.2241,
+      "step": 3305
+    },
+    {
+      "epoch": 0.6982048574445617,
+      "grad_norm": 0.3067176043987274,
+      "learning_rate": 2.1676197062339293e-05,
+      "loss": 0.2243,
+      "step": 3306
+    },
+    {
+      "epoch": 0.698416050686378,
+      "grad_norm": 0.29144445061683655,
+      "learning_rate": 2.164827565247264e-05,
+      "loss": 0.2244,
+      "step": 3307
+    },
+    {
+      "epoch": 0.6986272439281943,
+      "grad_norm": 0.32186275720596313,
+      "learning_rate": 2.162036726766588e-05,
+      "loss": 0.2244,
+      "step": 3308
+    },
+    {
+      "epoch": 0.6988384371700106,
+      "grad_norm": 0.31100934743881226,
+      "learning_rate": 2.1592471920740397e-05,
+      "loss": 0.2283,
+      "step": 3309
+    },
+    {
+      "epoch": 0.6990496304118268,
+      "grad_norm": 0.2642686367034912,
+      "learning_rate": 2.156458962451164e-05,
+      "loss": 0.2245,
+      "step": 3310
+    },
+    {
+      "epoch": 0.6992608236536431,
+      "grad_norm": 0.2819663882255554,
+      "learning_rate": 2.1536720391788955e-05,
+      "loss": 0.2224,
+      "step": 3311
+    },
+    {
+      "epoch": 0.6994720168954593,
+      "grad_norm": 0.3471440076828003,
+      "learning_rate": 2.1508864235375746e-05,
+      "loss": 0.2247,
+      "step": 3312
+    },
+    {
+      "epoch": 0.6996832101372756,
+      "grad_norm": 0.27738261222839355,
+      "learning_rate": 2.1481021168069437e-05,
+      "loss": 0.227,
+      "step": 3313
+    },
+    {
+      "epoch": 0.6998944033790919,
+      "grad_norm": 0.35405483841896057,
+      "learning_rate": 2.1453191202661383e-05,
+      "loss": 0.2267,
+      "step": 3314
+    },
+    {
+      "epoch": 0.7001055966209081,
+      "grad_norm": 0.32576900720596313,
+      "learning_rate": 2.1425374351936943e-05,
+      "loss": 0.2262,
+      "step": 3315
+    },
+    {
+      "epoch": 0.7003167898627244,
+      "grad_norm": 0.37225332856178284,
+      "learning_rate": 2.139757062867546e-05,
+      "loss": 0.228,
+      "step": 3316
+    },
+    {
+      "epoch": 0.7005279831045407,
+      "grad_norm": 0.31554368138313293,
+      "learning_rate": 2.136978004565023e-05,
+      "loss": 0.2334,
+      "step": 3317
+    },
+    {
+      "epoch": 0.700739176346357,
+      "grad_norm": 0.27602481842041016,
+      "learning_rate": 2.134200261562852e-05,
+      "loss": 0.2303,
+      "step": 3318
+    },
+    {
+      "epoch": 0.7009503695881731,
+      "grad_norm": 0.26708468794822693,
+      "learning_rate": 2.131423835137155e-05,
+      "loss": 0.2268,
+      "step": 3319
+    },
+    {
+      "epoch": 0.7011615628299894,
+      "grad_norm": 0.350301593542099,
+      "learning_rate": 2.1286487265634504e-05,
+      "loss": 0.2281,
+      "step": 3320
+    },
+    {
+      "epoch": 0.7013727560718057,
+      "grad_norm": 0.308232843875885,
+      "learning_rate": 2.125874937116649e-05,
+      "loss": 0.2289,
+      "step": 3321
+    },
+    {
+      "epoch": 0.701583949313622,
+      "grad_norm": 0.32449913024902344,
+      "learning_rate": 2.123102468071058e-05,
+      "loss": 0.2249,
+      "step": 3322
+    },
+    {
+      "epoch": 0.7017951425554382,
+      "grad_norm": 0.30075007677078247,
+      "learning_rate": 2.1203313207003766e-05,
+      "loss": 0.2222,
+      "step": 3323
+    },
+    {
+      "epoch": 0.7020063357972545,
+      "grad_norm": 0.3285229504108429,
+      "learning_rate": 2.117561496277697e-05,
+      "loss": 0.232,
+      "step": 3324
+    },
+    {
+      "epoch": 0.7022175290390708,
+      "grad_norm": 0.5353978276252747,
+      "learning_rate": 2.1147929960755032e-05,
+      "loss": 0.2281,
+      "step": 3325
+    },
+    {
+      "epoch": 0.702428722280887,
+      "grad_norm": 0.47648850083351135,
+      "learning_rate": 2.1120258213656712e-05,
+      "loss": 0.2326,
+      "step": 3326
+    },
+    {
+      "epoch": 0.7026399155227033,
+      "grad_norm": 0.2756692171096802,
+      "learning_rate": 2.109259973419469e-05,
+      "loss": 0.227,
+      "step": 3327
+    },
+    {
+      "epoch": 0.7028511087645195,
+      "grad_norm": 0.2810112535953522,
+      "learning_rate": 2.1064954535075544e-05,
+      "loss": 0.2329,
+      "step": 3328
+    },
+    {
+      "epoch": 0.7030623020063358,
+      "grad_norm": 0.305154025554657,
+      "learning_rate": 2.1037322628999734e-05,
+      "loss": 0.2275,
+      "step": 3329
+    },
+    {
+      "epoch": 0.7032734952481521,
+      "grad_norm": 0.2902466654777527,
+      "learning_rate": 2.100970402866164e-05,
+      "loss": 0.2273,
+      "step": 3330
+    },
+    {
+      "epoch": 0.7034846884899684,
+      "grad_norm": 0.2681199312210083,
+      "learning_rate": 2.0982098746749523e-05,
+      "loss": 0.2284,
+      "step": 3331
+    },
+    {
+      "epoch": 0.7036958817317845,
+      "grad_norm": 0.2978092432022095,
+      "learning_rate": 2.095450679594551e-05,
+      "loss": 0.2286,
+      "step": 3332
+    },
+    {
+      "epoch": 0.7039070749736008,
+      "grad_norm": 0.35419952869415283,
+      "learning_rate": 2.0926928188925614e-05,
+      "loss": 0.2231,
+      "step": 3333
+    },
+    {
+      "epoch": 0.7041182682154171,
+      "grad_norm": 0.28435319662094116,
+      "learning_rate": 2.089936293835972e-05,
+      "loss": 0.226,
+      "step": 3334
+    },
+    {
+      "epoch": 0.7043294614572334,
+      "grad_norm": 0.28456398844718933,
+      "learning_rate": 2.0871811056911574e-05,
+      "loss": 0.2267,
+      "step": 3335
+    },
+    {
+      "epoch": 0.7045406546990496,
+      "grad_norm": 0.29990655183792114,
+      "learning_rate": 2.0844272557238774e-05,
+      "loss": 0.2273,
+      "step": 3336
+    },
+    {
+      "epoch": 0.7047518479408659,
+      "grad_norm": 0.26797109842300415,
+      "learning_rate": 2.0816747451992775e-05,
+      "loss": 0.2283,
+      "step": 3337
+    },
+    {
+      "epoch": 0.7049630411826822,
+      "grad_norm": 0.28372693061828613,
+      "learning_rate": 2.0789235753818913e-05,
+      "loss": 0.2248,
+      "step": 3338
+    },
+    {
+      "epoch": 0.7051742344244984,
+      "grad_norm": 0.29108428955078125,
+      "learning_rate": 2.0761737475356295e-05,
+      "loss": 0.2255,
+      "step": 3339
+    },
+    {
+      "epoch": 0.7053854276663146,
+      "grad_norm": 0.2891122102737427,
+      "learning_rate": 2.0734252629237894e-05,
+      "loss": 0.2255,
+      "step": 3340
+    },
+    {
+      "epoch": 0.7055966209081309,
+      "grad_norm": 0.31336307525634766,
+      "learning_rate": 2.0706781228090565e-05,
+      "loss": 0.2313,
+      "step": 3341
+    },
+    {
+      "epoch": 0.7058078141499472,
+      "grad_norm": 0.3052317500114441,
+      "learning_rate": 2.0679323284534874e-05,
+      "loss": 0.2271,
+      "step": 3342
+    },
+    {
+      "epoch": 0.7060190073917635,
+      "grad_norm": 0.2701479494571686,
+      "learning_rate": 2.065187881118532e-05,
+      "loss": 0.2244,
+      "step": 3343
+    },
+    {
+      "epoch": 0.7062302006335798,
+      "grad_norm": 0.2991526126861572,
+      "learning_rate": 2.0624447820650162e-05,
+      "loss": 0.2254,
+      "step": 3344
+    },
+    {
+      "epoch": 0.706441393875396,
+      "grad_norm": 0.5561180114746094,
+      "learning_rate": 2.059703032553142e-05,
+      "loss": 0.2248,
+      "step": 3345
+    },
+    {
+      "epoch": 0.7066525871172122,
+      "grad_norm": 0.3200213313102722,
+      "learning_rate": 2.0569626338425002e-05,
+      "loss": 0.2255,
+      "step": 3346
+    },
+    {
+      "epoch": 0.7068637803590285,
+      "grad_norm": 0.30933451652526855,
+      "learning_rate": 2.0542235871920575e-05,
+      "loss": 0.2268,
+      "step": 3347
+    },
+    {
+      "epoch": 0.7070749736008448,
+      "grad_norm": 0.34984466433525085,
+      "learning_rate": 2.0514858938601533e-05,
+      "loss": 0.2309,
+      "step": 3348
+    },
+    {
+      "epoch": 0.707286166842661,
+      "grad_norm": 0.30555152893066406,
+      "learning_rate": 2.048749555104516e-05,
+      "loss": 0.223,
+      "step": 3349
+    },
+    {
+      "epoch": 0.7074973600844773,
+      "grad_norm": 0.5592265725135803,
+      "learning_rate": 2.0460145721822464e-05,
+      "loss": 0.2219,
+      "step": 3350
+    },
+    {
+      "epoch": 0.7077085533262936,
+      "grad_norm": 0.3397500216960907,
+      "learning_rate": 2.0432809463498177e-05,
+      "loss": 0.2238,
+      "step": 3351
+    },
+    {
+      "epoch": 0.7079197465681099,
+      "grad_norm": 0.29968661069869995,
+      "learning_rate": 2.040548678863089e-05,
+      "loss": 0.2248,
+      "step": 3352
+    },
+    {
+      "epoch": 0.708130939809926,
+      "grad_norm": 0.6405219435691833,
+      "learning_rate": 2.037817770977291e-05,
+      "loss": 0.2263,
+      "step": 3353
+    },
+    {
+      "epoch": 0.7083421330517423,
+      "grad_norm": 0.28838637471199036,
+      "learning_rate": 2.0350882239470244e-05,
+      "loss": 0.2214,
+      "step": 3354
+    },
+    {
+      "epoch": 0.7085533262935586,
+      "grad_norm": 0.3443657159805298,
+      "learning_rate": 2.0323600390262742e-05,
+      "loss": 0.2228,
+      "step": 3355
+    },
+    {
+      "epoch": 0.7087645195353749,
+      "grad_norm": 0.305437833070755,
+      "learning_rate": 2.0296332174683962e-05,
+      "loss": 0.2232,
+      "step": 3356
+    },
+    {
+      "epoch": 0.7089757127771912,
+      "grad_norm": 0.28101813793182373,
+      "learning_rate": 2.0269077605261137e-05,
+      "loss": 0.2255,
+      "step": 3357
+    },
+    {
+      "epoch": 0.7091869060190074,
+      "grad_norm": 0.3140544295310974,
+      "learning_rate": 2.0241836694515336e-05,
+      "loss": 0.2312,
+      "step": 3358
+    },
+    {
+      "epoch": 0.7093980992608236,
+      "grad_norm": 0.5144098401069641,
+      "learning_rate": 2.021460945496128e-05,
+      "loss": 0.2237,
+      "step": 3359
+    },
+    {
+      "epoch": 0.7096092925026399,
+      "grad_norm": 0.355684369802475,
+      "learning_rate": 2.0187395899107424e-05,
+      "loss": 0.2332,
+      "step": 3360
+    },
+    {
+      "epoch": 0.7098204857444562,
+      "grad_norm": 0.3378099203109741,
+      "learning_rate": 2.016019603945596e-05,
+      "loss": 0.2337,
+      "step": 3361
+    },
+    {
+      "epoch": 0.7100316789862724,
+      "grad_norm": 0.3646467328071594,
+      "learning_rate": 2.0133009888502745e-05,
+      "loss": 0.2313,
+      "step": 3362
+    },
+    {
+      "epoch": 0.7102428722280887,
+      "grad_norm": 0.36807510256767273,
+      "learning_rate": 2.0105837458737382e-05,
+      "loss": 0.2302,
+      "step": 3363
+    },
+    {
+      "epoch": 0.710454065469905,
+      "grad_norm": 0.5399953722953796,
+      "learning_rate": 2.0078678762643144e-05,
+      "loss": 0.2245,
+      "step": 3364
+    },
+    {
+      "epoch": 0.7106652587117213,
+      "grad_norm": 0.36841413378715515,
+      "learning_rate": 2.005153381269701e-05,
+      "loss": 0.2272,
+      "step": 3365
+    },
+    {
+      "epoch": 0.7108764519535374,
+      "grad_norm": 0.3450266420841217,
+      "learning_rate": 2.002440262136962e-05,
+      "loss": 0.2324,
+      "step": 3366
+    },
+    {
+      "epoch": 0.7110876451953537,
+      "grad_norm": 0.2901979684829712,
+      "learning_rate": 1.9997285201125325e-05,
+      "loss": 0.2273,
+      "step": 3367
+    },
+    {
+      "epoch": 0.71129883843717,
+      "grad_norm": 0.3182978630065918,
+      "learning_rate": 1.9970181564422124e-05,
+      "loss": 0.228,
+      "step": 3368
+    },
+    {
+      "epoch": 0.7115100316789863,
+      "grad_norm": 0.3460440933704376,
+      "learning_rate": 1.99430917237117e-05,
+      "loss": 0.2309,
+      "step": 3369
+    },
+    {
+      "epoch": 0.7117212249208026,
+      "grad_norm": 0.3363269865512848,
+      "learning_rate": 1.9916015691439383e-05,
+      "loss": 0.2256,
+      "step": 3370
+    },
+    {
+      "epoch": 0.7119324181626188,
+      "grad_norm": 0.5596103668212891,
+      "learning_rate": 1.9888953480044176e-05,
+      "loss": 0.2213,
+      "step": 3371
+    },
+    {
+      "epoch": 0.712143611404435,
+      "grad_norm": 0.31031540036201477,
+      "learning_rate": 1.9861905101958715e-05,
+      "loss": 0.2295,
+      "step": 3372
+    },
+    {
+      "epoch": 0.7123548046462513,
+      "grad_norm": 0.3063788115978241,
+      "learning_rate": 1.983487056960928e-05,
+      "loss": 0.2283,
+      "step": 3373
+    },
+    {
+      "epoch": 0.7125659978880676,
+      "grad_norm": 0.30296722054481506,
+      "learning_rate": 1.9807849895415853e-05,
+      "loss": 0.2295,
+      "step": 3374
+    },
+    {
+      "epoch": 0.7127771911298838,
+      "grad_norm": 0.3785286843776703,
+      "learning_rate": 1.978084309179194e-05,
+      "loss": 0.2269,
+      "step": 3375
+    },
+    {
+      "epoch": 0.7129883843717001,
+      "grad_norm": 0.31033799052238464,
+      "learning_rate": 1.975385017114473e-05,
+      "loss": 0.2252,
+      "step": 3376
+    },
+    {
+      "epoch": 0.7131995776135164,
+      "grad_norm": 0.27585312724113464,
+      "learning_rate": 1.9726871145875092e-05,
+      "loss": 0.228,
+      "step": 3377
+    },
+    {
+      "epoch": 0.7134107708553327,
+      "grad_norm": 0.3780837059020996,
+      "learning_rate": 1.9699906028377397e-05,
+      "loss": 0.2207,
+      "step": 3378
+    },
+    {
+      "epoch": 0.7136219640971488,
+      "grad_norm": 0.3636508882045746,
+      "learning_rate": 1.9672954831039685e-05,
+      "loss": 0.2242,
+      "step": 3379
+    },
+    {
+      "epoch": 0.7138331573389651,
+      "grad_norm": 0.34049418568611145,
+      "learning_rate": 1.9646017566243658e-05,
+      "loss": 0.2281,
+      "step": 3380
+    },
+    {
+      "epoch": 0.7140443505807814,
+      "grad_norm": 0.883463442325592,
+      "learning_rate": 1.9619094246364507e-05,
+      "loss": 0.2289,
+      "step": 3381
+    },
+    {
+      "epoch": 0.7142555438225977,
+      "grad_norm": 0.52713942527771,
+      "learning_rate": 1.9592184883771058e-05,
+      "loss": 0.2268,
+      "step": 3382
+    },
+    {
+      "epoch": 0.714466737064414,
+      "grad_norm": 0.29950177669525146,
+      "learning_rate": 1.95652894908258e-05,
+      "loss": 0.2269,
+      "step": 3383
+    },
+    {
+      "epoch": 0.7146779303062302,
+      "grad_norm": 0.4057658910751343,
+      "learning_rate": 1.9538408079884678e-05,
+      "loss": 0.2277,
+      "step": 3384
+    },
+    {
+      "epoch": 0.7148891235480465,
+      "grad_norm": 0.42365044355392456,
+      "learning_rate": 1.9511540663297285e-05,
+      "loss": 0.225,
+      "step": 3385
+    },
+    {
+      "epoch": 0.7151003167898627,
+      "grad_norm": 0.2908778190612793,
+      "learning_rate": 1.9484687253406824e-05,
+      "loss": 0.2257,
+      "step": 3386
+    },
+    {
+      "epoch": 0.715311510031679,
+      "grad_norm": 0.2917545735836029,
+      "learning_rate": 1.9457847862549944e-05,
+      "loss": 0.2233,
+      "step": 3387
+    },
+    {
+      "epoch": 0.7155227032734952,
+      "grad_norm": 0.3306361138820648,
+      "learning_rate": 1.943102250305697e-05,
+      "loss": 0.2233,
+      "step": 3388
+    },
+    {
+      "epoch": 0.7157338965153115,
+      "grad_norm": 0.37727826833724976,
+      "learning_rate": 1.9404211187251747e-05,
+      "loss": 0.227,
+      "step": 3389
+    },
+    {
+      "epoch": 0.7159450897571278,
+      "grad_norm": 0.28608351945877075,
+      "learning_rate": 1.9377413927451598e-05,
+      "loss": 0.2212,
+      "step": 3390
+    },
+    {
+      "epoch": 0.7161562829989441,
+      "grad_norm": 0.3448997139930725,
+      "learning_rate": 1.93506307359675e-05,
+      "loss": 0.224,
+      "step": 3391
+    },
+    {
+      "epoch": 0.7163674762407602,
+      "grad_norm": 0.3086482286453247,
+      "learning_rate": 1.9323861625103894e-05,
+      "loss": 0.2251,
+      "step": 3392
+    },
+    {
+      "epoch": 0.7165786694825765,
+      "grad_norm": 0.31621503829956055,
+      "learning_rate": 1.929710660715877e-05,
+      "loss": 0.2231,
+      "step": 3393
+    },
+    {
+      "epoch": 0.7167898627243928,
+      "grad_norm": 0.3351730704307556,
+      "learning_rate": 1.9270365694423653e-05,
+      "loss": 0.229,
+      "step": 3394
+    },
+    {
+      "epoch": 0.7170010559662091,
+      "grad_norm": 0.4270564019680023,
+      "learning_rate": 1.9243638899183574e-05,
+      "loss": 0.229,
+      "step": 3395
+    },
+    {
+      "epoch": 0.7172122492080254,
+      "grad_norm": 0.3122486472129822,
+      "learning_rate": 1.9216926233717085e-05,
+      "loss": 0.2216,
+      "step": 3396
+    },
+    {
+      "epoch": 0.7174234424498416,
+      "grad_norm": 0.360963374376297,
+      "learning_rate": 1.9190227710296254e-05,
+      "loss": 0.2249,
+      "step": 3397
+    },
+    {
+      "epoch": 0.7176346356916579,
+      "grad_norm": 0.4788248836994171,
+      "learning_rate": 1.916354334118663e-05,
+      "loss": 0.2299,
+      "step": 3398
+    },
+    {
+      "epoch": 0.7178458289334742,
+      "grad_norm": 0.7046374678611755,
+      "learning_rate": 1.913687313864729e-05,
+      "loss": 0.2248,
+      "step": 3399
+    },
+    {
+      "epoch": 0.7180570221752904,
+      "grad_norm": 0.33398836851119995,
+      "learning_rate": 1.911021711493077e-05,
+      "loss": 0.2256,
+      "step": 3400
+    },
+    {
+      "epoch": 0.7180570221752904,
+      "eval_train_split_loss": 0.3368355631828308,
+      "eval_train_split_runtime": 10.9786,
+      "eval_train_split_samples_per_second": 0.911,
+      "eval_train_split_steps_per_second": 0.091,
+      "step": 3400
+    },
+    {
+      "epoch": 0.7182682154171066,
+      "grad_norm": 0.3316670060157776,
+      "learning_rate": 1.9083575282283118e-05,
+      "loss": 0.2307,
+      "step": 3401
+    },
+    {
+      "epoch": 0.7184794086589229,
+      "grad_norm": 0.31191369891166687,
+      "learning_rate": 1.9056947652943847e-05,
+      "loss": 0.2231,
+      "step": 3402
+    },
+    {
+      "epoch": 0.7186906019007392,
+      "grad_norm": 0.4847922921180725,
+      "learning_rate": 1.903033423914596e-05,
+      "loss": 0.2206,
+      "step": 3403
+    },
+    {
+      "epoch": 0.7189017951425555,
+      "grad_norm": 0.32678699493408203,
+      "learning_rate": 1.9003735053115913e-05,
+      "loss": 0.22,
+      "step": 3404
+    },
+    {
+      "epoch": 0.7191129883843717,
+      "grad_norm": 0.34094828367233276,
+      "learning_rate": 1.8977150107073633e-05,
+      "loss": 0.2256,
+      "step": 3405
+    },
+    {
+      "epoch": 0.7193241816261879,
+      "grad_norm": 0.358064204454422,
+      "learning_rate": 1.895057941323251e-05,
+      "loss": 0.2329,
+      "step": 3406
+    },
+    {
+      "epoch": 0.7195353748680042,
+      "grad_norm": 0.8223753571510315,
+      "learning_rate": 1.8924022983799373e-05,
+      "loss": 0.2258,
+      "step": 3407
+    },
+    {
+      "epoch": 0.7197465681098205,
+      "grad_norm": 0.3488786518573761,
+      "learning_rate": 1.8897480830974512e-05,
+      "loss": 0.2264,
+      "step": 3408
+    },
+    {
+      "epoch": 0.7199577613516367,
+      "grad_norm": 0.29888269305229187,
+      "learning_rate": 1.887095296695165e-05,
+      "loss": 0.2267,
+      "step": 3409
+    },
+    {
+      "epoch": 0.720168954593453,
+      "grad_norm": 0.3433454930782318,
+      "learning_rate": 1.8844439403917946e-05,
+      "loss": 0.223,
+      "step": 3410
+    },
+    {
+      "epoch": 0.7203801478352693,
+      "grad_norm": 0.43471768498420715,
+      "learning_rate": 1.8817940154053993e-05,
+      "loss": 0.2267,
+      "step": 3411
+    },
+    {
+      "epoch": 0.7205913410770856,
+      "grad_norm": 0.3326507806777954,
+      "learning_rate": 1.8791455229533804e-05,
+      "loss": 0.2184,
+      "step": 3412
+    },
+    {
+      "epoch": 0.7208025343189018,
+      "grad_norm": 0.27561625838279724,
+      "learning_rate": 1.8764984642524818e-05,
+      "loss": 0.2211,
+      "step": 3413
+    },
+    {
+      "epoch": 0.721013727560718,
+      "grad_norm": 0.31727227568626404,
+      "learning_rate": 1.8738528405187873e-05,
+      "loss": 0.2222,
+      "step": 3414
+    },
+    {
+      "epoch": 0.7212249208025343,
+      "grad_norm": 0.28655457496643066,
+      "learning_rate": 1.8712086529677215e-05,
+      "loss": 0.221,
+      "step": 3415
+    },
+    {
+      "epoch": 0.7214361140443506,
+      "grad_norm": 0.30089282989501953,
+      "learning_rate": 1.868565902814053e-05,
+      "loss": 0.227,
+      "step": 3416
+    },
+    {
+      "epoch": 0.7216473072861669,
+      "grad_norm": 0.27638015151023865,
+      "learning_rate": 1.865924591271887e-05,
+      "loss": 0.2236,
+      "step": 3417
+    },
+    {
+      "epoch": 0.7218585005279831,
+      "grad_norm": 0.31500768661499023,
+      "learning_rate": 1.8632847195546633e-05,
+      "loss": 0.2265,
+      "step": 3418
+    },
+    {
+      "epoch": 0.7220696937697993,
+      "grad_norm": 0.3215233087539673,
+      "learning_rate": 1.8606462888751697e-05,
+      "loss": 0.221,
+      "step": 3419
+    },
+    {
+      "epoch": 0.7222808870116156,
+      "grad_norm": 0.2757122218608856,
+      "learning_rate": 1.8580093004455267e-05,
+      "loss": 0.2267,
+      "step": 3420
+    },
+    {
+      "epoch": 0.7224920802534319,
+      "grad_norm": 0.3711574673652649,
+      "learning_rate": 1.8553737554771882e-05,
+      "loss": 0.2172,
+      "step": 3421
+    },
+    {
+      "epoch": 0.7227032734952481,
+      "grad_norm": 0.2876644730567932,
+      "learning_rate": 1.8527396551809538e-05,
+      "loss": 0.2252,
+      "step": 3422
+    },
+    {
+      "epoch": 0.7229144667370644,
+      "grad_norm": 0.3296727240085602,
+      "learning_rate": 1.8501070007669557e-05,
+      "loss": 0.2206,
+      "step": 3423
+    },
+    {
+      "epoch": 0.7231256599788807,
+      "grad_norm": 0.4337751269340515,
+      "learning_rate": 1.8474757934446558e-05,
+      "loss": 0.2261,
+      "step": 3424
+    },
+    {
+      "epoch": 0.723336853220697,
+      "grad_norm": 0.366911381483078,
+      "learning_rate": 1.8448460344228608e-05,
+      "loss": 0.2306,
+      "step": 3425
+    },
+    {
+      "epoch": 0.7235480464625133,
+      "grad_norm": 0.3579188883304596,
+      "learning_rate": 1.842217724909709e-05,
+      "loss": 0.2265,
+      "step": 3426
+    },
+    {
+      "epoch": 0.7237592397043294,
+      "grad_norm": 0.2933277487754822,
+      "learning_rate": 1.8395908661126654e-05,
+      "loss": 0.2283,
+      "step": 3427
+    },
+    {
+      "epoch": 0.7239704329461457,
+      "grad_norm": 0.32366108894348145,
+      "learning_rate": 1.836965459238541e-05,
+      "loss": 0.2252,
+      "step": 3428
+    },
+    {
+      "epoch": 0.724181626187962,
+      "grad_norm": 0.40035563707351685,
+      "learning_rate": 1.8343415054934704e-05,
+      "loss": 0.2264,
+      "step": 3429
+    },
+    {
+      "epoch": 0.7243928194297783,
+      "grad_norm": 0.513067364692688,
+      "learning_rate": 1.831719006082924e-05,
+      "loss": 0.2225,
+      "step": 3430
+    },
+    {
+      "epoch": 0.7246040126715945,
+      "grad_norm": 0.34338557720184326,
+      "learning_rate": 1.8290979622117044e-05,
+      "loss": 0.2277,
+      "step": 3431
+    },
+    {
+      "epoch": 0.7248152059134108,
+      "grad_norm": 0.3353792130947113,
+      "learning_rate": 1.8264783750839442e-05,
+      "loss": 0.2326,
+      "step": 3432
+    },
+    {
+      "epoch": 0.725026399155227,
+      "grad_norm": 0.3195915222167969,
+      "learning_rate": 1.823860245903108e-05,
+      "loss": 0.2288,
+      "step": 3433
+    },
+    {
+      "epoch": 0.7252375923970433,
+      "grad_norm": 0.2891676723957062,
+      "learning_rate": 1.8212435758719894e-05,
+      "loss": 0.2285,
+      "step": 3434
+    },
+    {
+      "epoch": 0.7254487856388595,
+      "grad_norm": 0.28953856229782104,
+      "learning_rate": 1.8186283661927116e-05,
+      "loss": 0.2322,
+      "step": 3435
+    },
+    {
+      "epoch": 0.7256599788806758,
+      "grad_norm": 0.3118908703327179,
+      "learning_rate": 1.816014618066729e-05,
+      "loss": 0.2289,
+      "step": 3436
+    },
+    {
+      "epoch": 0.7258711721224921,
+      "grad_norm": 0.2881069481372833,
+      "learning_rate": 1.8134023326948225e-05,
+      "loss": 0.2208,
+      "step": 3437
+    },
+    {
+      "epoch": 0.7260823653643084,
+      "grad_norm": 0.25665053725242615,
+      "learning_rate": 1.8107915112771016e-05,
+      "loss": 0.2124,
+      "step": 3438
+    },
+    {
+      "epoch": 0.7262935586061247,
+      "grad_norm": 0.3071800172328949,
+      "learning_rate": 1.8081821550130028e-05,
+      "loss": 0.2232,
+      "step": 3439
+    },
+    {
+      "epoch": 0.7265047518479408,
+      "grad_norm": 0.29547667503356934,
+      "learning_rate": 1.8055742651012907e-05,
+      "loss": 0.2268,
+      "step": 3440
+    },
+    {
+      "epoch": 0.7267159450897571,
+      "grad_norm": 0.3428885042667389,
+      "learning_rate": 1.8029678427400548e-05,
+      "loss": 0.2345,
+      "step": 3441
+    },
+    {
+      "epoch": 0.7269271383315734,
+      "grad_norm": 0.4504431486129761,
+      "learning_rate": 1.8003628891267117e-05,
+      "loss": 0.2264,
+      "step": 3442
+    },
+    {
+      "epoch": 0.7271383315733897,
+      "grad_norm": 0.2877937853336334,
+      "learning_rate": 1.7977594054580004e-05,
+      "loss": 0.2305,
+      "step": 3443
+    },
+    {
+      "epoch": 0.7273495248152059,
+      "grad_norm": 0.289638876914978,
+      "learning_rate": 1.795157392929992e-05,
+      "loss": 0.2374,
+      "step": 3444
+    },
+    {
+      "epoch": 0.7275607180570222,
+      "grad_norm": 0.30132749676704407,
+      "learning_rate": 1.7925568527380717e-05,
+      "loss": 0.2349,
+      "step": 3445
+    },
+    {
+      "epoch": 0.7277719112988384,
+      "grad_norm": 0.28801459074020386,
+      "learning_rate": 1.7899577860769522e-05,
+      "loss": 0.2257,
+      "step": 3446
+    },
+    {
+      "epoch": 0.7279831045406547,
+      "grad_norm": 1.1799837350845337,
+      "learning_rate": 1.7873601941406763e-05,
+      "loss": 0.2232,
+      "step": 3447
+    },
+    {
+      "epoch": 0.7281942977824709,
+      "grad_norm": 0.28765809535980225,
+      "learning_rate": 1.7847640781225983e-05,
+      "loss": 0.2298,
+      "step": 3448
+    },
+    {
+      "epoch": 0.7284054910242872,
+      "grad_norm": 0.30512821674346924,
+      "learning_rate": 1.7821694392153987e-05,
+      "loss": 0.2289,
+      "step": 3449
+    },
+    {
+      "epoch": 0.7286166842661035,
+      "grad_norm": 0.26941195130348206,
+      "learning_rate": 1.7795762786110854e-05,
+      "loss": 0.2203,
+      "step": 3450
+    },
+    {
+      "epoch": 0.7288278775079198,
+      "grad_norm": 0.3571496605873108,
+      "learning_rate": 1.7769845975009763e-05,
+      "loss": 0.2249,
+      "step": 3451
+    },
+    {
+      "epoch": 0.7290390707497361,
+      "grad_norm": 0.30578774213790894,
+      "learning_rate": 1.7743943970757155e-05,
+      "loss": 0.2336,
+      "step": 3452
+    },
+    {
+      "epoch": 0.7292502639915522,
+      "grad_norm": 0.26777347922325134,
+      "learning_rate": 1.7718056785252714e-05,
+      "loss": 0.2221,
+      "step": 3453
+    },
+    {
+      "epoch": 0.7294614572333685,
+      "grad_norm": 0.3003236651420593,
+      "learning_rate": 1.7692184430389214e-05,
+      "loss": 0.2248,
+      "step": 3454
+    },
+    {
+      "epoch": 0.7296726504751848,
+      "grad_norm": 0.3459048271179199,
+      "learning_rate": 1.7666326918052667e-05,
+      "loss": 0.2289,
+      "step": 3455
+    },
+    {
+      "epoch": 0.7298838437170011,
+      "grad_norm": 0.2721727192401886,
+      "learning_rate": 1.7640484260122325e-05,
+      "loss": 0.2224,
+      "step": 3456
+    },
+    {
+      "epoch": 0.7300950369588173,
+      "grad_norm": 0.2590268850326538,
+      "learning_rate": 1.761465646847051e-05,
+      "loss": 0.2283,
+      "step": 3457
+    },
+    {
+      "epoch": 0.7303062302006336,
+      "grad_norm": 0.25758421421051025,
+      "learning_rate": 1.7588843554962747e-05,
+      "loss": 0.2235,
+      "step": 3458
+    },
+    {
+      "epoch": 0.7305174234424499,
+      "grad_norm": 0.24941475689411163,
+      "learning_rate": 1.7563045531457806e-05,
+      "loss": 0.2193,
+      "step": 3459
+    },
+    {
+      "epoch": 0.7307286166842661,
+      "grad_norm": 0.26503661274909973,
+      "learning_rate": 1.7537262409807474e-05,
+      "loss": 0.2241,
+      "step": 3460
+    },
+    {
+      "epoch": 0.7309398099260823,
+      "grad_norm": 0.27962255477905273,
+      "learning_rate": 1.7511494201856832e-05,
+      "loss": 0.2269,
+      "step": 3461
+    },
+    {
+      "epoch": 0.7311510031678986,
+      "grad_norm": 0.2737715244293213,
+      "learning_rate": 1.748574091944404e-05,
+      "loss": 0.2291,
+      "step": 3462
+    },
+    {
+      "epoch": 0.7313621964097149,
+      "grad_norm": 0.27879372239112854,
+      "learning_rate": 1.746000257440035e-05,
+      "loss": 0.2233,
+      "step": 3463
+    },
+    {
+      "epoch": 0.7315733896515312,
+      "grad_norm": 0.30133575201034546,
+      "learning_rate": 1.743427917855028e-05,
+      "loss": 0.2321,
+      "step": 3464
+    },
+    {
+      "epoch": 0.7317845828933475,
+      "grad_norm": 0.30003416538238525,
+      "learning_rate": 1.7408570743711393e-05,
+      "loss": 0.2285,
+      "step": 3465
+    },
+    {
+      "epoch": 0.7319957761351636,
+      "grad_norm": 0.30838868021965027,
+      "learning_rate": 1.7382877281694355e-05,
+      "loss": 0.2223,
+      "step": 3466
+    },
+    {
+      "epoch": 0.7322069693769799,
+      "grad_norm": 0.3065990209579468,
+      "learning_rate": 1.7357198804303037e-05,
+      "loss": 0.2321,
+      "step": 3467
+    },
+    {
+      "epoch": 0.7324181626187962,
+      "grad_norm": 0.3021054267883301,
+      "learning_rate": 1.7331535323334374e-05,
+      "loss": 0.2192,
+      "step": 3468
+    },
+    {
+      "epoch": 0.7326293558606125,
+      "grad_norm": 0.3533285856246948,
+      "learning_rate": 1.7305886850578412e-05,
+      "loss": 0.2209,
+      "step": 3469
+    },
+    {
+      "epoch": 0.7328405491024287,
+      "grad_norm": 0.3450993001461029,
+      "learning_rate": 1.728025339781832e-05,
+      "loss": 0.2229,
+      "step": 3470
+    },
+    {
+      "epoch": 0.733051742344245,
+      "grad_norm": 0.27879026532173157,
+      "learning_rate": 1.7254634976830354e-05,
+      "loss": 0.2232,
+      "step": 3471
+    },
+    {
+      "epoch": 0.7332629355860613,
+      "grad_norm": 0.3102920651435852,
+      "learning_rate": 1.722903159938386e-05,
+      "loss": 0.2242,
+      "step": 3472
+    },
+    {
+      "epoch": 0.7334741288278775,
+      "grad_norm": 0.33628642559051514,
+      "learning_rate": 1.7203443277241293e-05,
+      "loss": 0.224,
+      "step": 3473
+    },
+    {
+      "epoch": 0.7336853220696937,
+      "grad_norm": 0.2820534110069275,
+      "learning_rate": 1.7177870022158167e-05,
+      "loss": 0.2205,
+      "step": 3474
+    },
+    {
+      "epoch": 0.73389651531151,
+      "grad_norm": 0.31778448820114136,
+      "learning_rate": 1.7152311845883095e-05,
+      "loss": 0.223,
+      "step": 3475
+    },
+    {
+      "epoch": 0.7341077085533263,
+      "grad_norm": 0.3031570315361023,
+      "learning_rate": 1.712676876015775e-05,
+      "loss": 0.2264,
+      "step": 3476
+    },
+    {
+      "epoch": 0.7343189017951426,
+      "grad_norm": 0.28334155678749084,
+      "learning_rate": 1.7101240776716877e-05,
+      "loss": 0.2174,
+      "step": 3477
+    },
+    {
+      "epoch": 0.7345300950369588,
+      "grad_norm": 0.2844255566596985,
+      "learning_rate": 1.7075727907288276e-05,
+      "loss": 0.2242,
+      "step": 3478
+    },
+    {
+      "epoch": 0.734741288278775,
+      "grad_norm": 0.31761232018470764,
+      "learning_rate": 1.7050230163592816e-05,
+      "loss": 0.2227,
+      "step": 3479
+    },
+    {
+      "epoch": 0.7349524815205913,
+      "grad_norm": 0.3152781128883362,
+      "learning_rate": 1.702474755734441e-05,
+      "loss": 0.2217,
+      "step": 3480
+    },
+    {
+      "epoch": 0.7351636747624076,
+      "grad_norm": 0.27593106031417847,
+      "learning_rate": 1.6999280100250016e-05,
+      "loss": 0.217,
+      "step": 3481
+    },
+    {
+      "epoch": 0.7353748680042239,
+      "grad_norm": 0.2733933925628662,
+      "learning_rate": 1.697382780400963e-05,
+      "loss": 0.221,
+      "step": 3482
+    },
+    {
+      "epoch": 0.7355860612460401,
+      "grad_norm": 0.27853918075561523,
+      "learning_rate": 1.6948390680316285e-05,
+      "loss": 0.2288,
+      "step": 3483
+    },
+    {
+      "epoch": 0.7357972544878564,
+      "grad_norm": 0.3034326434135437,
+      "learning_rate": 1.692296874085605e-05,
+      "loss": 0.2227,
+      "step": 3484
+    },
+    {
+      "epoch": 0.7360084477296727,
+      "grad_norm": 0.27381789684295654,
+      "learning_rate": 1.6897561997308016e-05,
+      "loss": 0.2202,
+      "step": 3485
+    },
+    {
+      "epoch": 0.736219640971489,
+      "grad_norm": 0.30238571763038635,
+      "learning_rate": 1.6872170461344276e-05,
+      "loss": 0.2254,
+      "step": 3486
+    },
+    {
+      "epoch": 0.7364308342133051,
+      "grad_norm": 0.2654973864555359,
+      "learning_rate": 1.6846794144629967e-05,
+      "loss": 0.2247,
+      "step": 3487
+    },
+    {
+      "epoch": 0.7366420274551214,
+      "grad_norm": 0.2743845582008362,
+      "learning_rate": 1.6821433058823187e-05,
+      "loss": 0.2226,
+      "step": 3488
+    },
+    {
+      "epoch": 0.7368532206969377,
+      "grad_norm": 0.35168927907943726,
+      "learning_rate": 1.6796087215575128e-05,
+      "loss": 0.2219,
+      "step": 3489
+    },
+    {
+      "epoch": 0.737064413938754,
+      "grad_norm": 0.31954526901245117,
+      "learning_rate": 1.6770756626529866e-05,
+      "loss": 0.2236,
+      "step": 3490
+    },
+    {
+      "epoch": 0.7372756071805702,
+      "grad_norm": 0.3953140079975128,
+      "learning_rate": 1.674544130332452e-05,
+      "loss": 0.2245,
+      "step": 3491
+    },
+    {
+      "epoch": 0.7374868004223865,
+      "grad_norm": 0.49411696195602417,
+      "learning_rate": 1.672014125758923e-05,
+      "loss": 0.2261,
+      "step": 3492
+    },
+    {
+      "epoch": 0.7376979936642027,
+      "grad_norm": 0.3120756149291992,
+      "learning_rate": 1.669485650094708e-05,
+      "loss": 0.2274,
+      "step": 3493
+    },
+    {
+      "epoch": 0.737909186906019,
+      "grad_norm": 0.3104393482208252,
+      "learning_rate": 1.6669587045014097e-05,
+      "loss": 0.2233,
+      "step": 3494
+    },
+    {
+      "epoch": 0.7381203801478353,
+      "grad_norm": 0.292866975069046,
+      "learning_rate": 1.6644332901399357e-05,
+      "loss": 0.2192,
+      "step": 3495
+    },
+    {
+      "epoch": 0.7383315733896515,
+      "grad_norm": 0.3420705497264862,
+      "learning_rate": 1.6619094081704868e-05,
+      "loss": 0.2202,
+      "step": 3496
+    },
+    {
+      "epoch": 0.7385427666314678,
+      "grad_norm": 0.3003863990306854,
+      "learning_rate": 1.6593870597525534e-05,
+      "loss": 0.2189,
+      "step": 3497
+    },
+    {
+      "epoch": 0.7387539598732841,
+      "grad_norm": 0.30948859453201294,
+      "learning_rate": 1.6568662460449325e-05,
+      "loss": 0.2238,
+      "step": 3498
+    },
+    {
+      "epoch": 0.7389651531151004,
+      "grad_norm": 0.3255822956562042,
+      "learning_rate": 1.6543469682057106e-05,
+      "loss": 0.2221,
+      "step": 3499
+    },
+    {
+      "epoch": 0.7391763463569165,
+      "grad_norm": 0.3010782301425934,
+      "learning_rate": 1.651829227392263e-05,
+      "loss": 0.2262,
+      "step": 3500
+    },
+    {
+      "epoch": 0.7393875395987328,
+      "grad_norm": 0.312837690114975,
+      "learning_rate": 1.649313024761271e-05,
+      "loss": 0.2251,
+      "step": 3501
+    },
+    {
+      "epoch": 0.7395987328405491,
+      "grad_norm": 0.29938238859176636,
+      "learning_rate": 1.6467983614686994e-05,
+      "loss": 0.229,
+      "step": 3502
+    },
+    {
+      "epoch": 0.7398099260823654,
+      "grad_norm": 0.4016336500644684,
+      "learning_rate": 1.64428523866981e-05,
+      "loss": 0.2247,
+      "step": 3503
+    },
+    {
+      "epoch": 0.7400211193241816,
+      "grad_norm": 0.28733065724372864,
+      "learning_rate": 1.6417736575191555e-05,
+      "loss": 0.2218,
+      "step": 3504
+    },
+    {
+      "epoch": 0.7402323125659979,
+      "grad_norm": 0.277647465467453,
+      "learning_rate": 1.6392636191705817e-05,
+      "loss": 0.2289,
+      "step": 3505
+    },
+    {
+      "epoch": 0.7404435058078142,
+      "grad_norm": 0.2954343557357788,
+      "learning_rate": 1.636755124777224e-05,
+      "loss": 0.2294,
+      "step": 3506
+    },
+    {
+      "epoch": 0.7406546990496304,
+      "grad_norm": 0.27808597683906555,
+      "learning_rate": 1.63424817549151e-05,
+      "loss": 0.2216,
+      "step": 3507
+    },
+    {
+      "epoch": 0.7408658922914467,
+      "grad_norm": 0.3359050750732422,
+      "learning_rate": 1.631742772465156e-05,
+      "loss": 0.2225,
+      "step": 3508
+    },
+    {
+      "epoch": 0.7410770855332629,
+      "grad_norm": 0.3105338215827942,
+      "learning_rate": 1.6292389168491702e-05,
+      "loss": 0.2323,
+      "step": 3509
+    },
+    {
+      "epoch": 0.7412882787750792,
+      "grad_norm": 0.3035981357097626,
+      "learning_rate": 1.6267366097938462e-05,
+      "loss": 0.2272,
+      "step": 3510
+    },
+    {
+      "epoch": 0.7414994720168955,
+      "grad_norm": 0.5048880577087402,
+      "learning_rate": 1.62423585244877e-05,
+      "loss": 0.2197,
+      "step": 3511
+    },
+    {
+      "epoch": 0.7417106652587118,
+      "grad_norm": 0.3435819149017334,
+      "learning_rate": 1.6217366459628136e-05,
+      "loss": 0.2301,
+      "step": 3512
+    },
+    {
+      "epoch": 0.7419218585005279,
+      "grad_norm": 0.35293152928352356,
+      "learning_rate": 1.6192389914841364e-05,
+      "loss": 0.2201,
+      "step": 3513
+    },
+    {
+      "epoch": 0.7421330517423442,
+      "grad_norm": 0.3004331886768341,
+      "learning_rate": 1.616742890160186e-05,
+      "loss": 0.2228,
+      "step": 3514
+    },
+    {
+      "epoch": 0.7423442449841605,
+      "grad_norm": 0.3247804343700409,
+      "learning_rate": 1.614248343137696e-05,
+      "loss": 0.2244,
+      "step": 3515
+    },
+    {
+      "epoch": 0.7425554382259768,
+      "grad_norm": 0.39112597703933716,
+      "learning_rate": 1.611755351562685e-05,
+      "loss": 0.2253,
+      "step": 3516
+    },
+    {
+      "epoch": 0.742766631467793,
+      "grad_norm": 0.397832989692688,
+      "learning_rate": 1.609263916580458e-05,
+      "loss": 0.2243,
+      "step": 3517
+    },
+    {
+      "epoch": 0.7429778247096093,
+      "grad_norm": 0.2784580588340759,
+      "learning_rate": 1.6067740393356044e-05,
+      "loss": 0.2245,
+      "step": 3518
+    },
+    {
+      "epoch": 0.7431890179514256,
+      "grad_norm": 0.466871052980423,
+      "learning_rate": 1.6042857209719975e-05,
+      "loss": 0.2192,
+      "step": 3519
+    },
+    {
+      "epoch": 0.7434002111932418,
+      "grad_norm": 0.29696890711784363,
+      "learning_rate": 1.601798962632799e-05,
+      "loss": 0.221,
+      "step": 3520
+    },
+    {
+      "epoch": 0.7436114044350581,
+      "grad_norm": 0.2720375657081604,
+      "learning_rate": 1.5993137654604455e-05,
+      "loss": 0.2278,
+      "step": 3521
+    },
+    {
+      "epoch": 0.7438225976768743,
+      "grad_norm": 0.33629485964775085,
+      "learning_rate": 1.5968301305966614e-05,
+      "loss": 0.2257,
+      "step": 3522
+    },
+    {
+      "epoch": 0.7440337909186906,
+      "grad_norm": 0.2771477997303009,
+      "learning_rate": 1.594348059182458e-05,
+      "loss": 0.2198,
+      "step": 3523
+    },
+    {
+      "epoch": 0.7442449841605069,
+      "grad_norm": 0.2998025715351105,
+      "learning_rate": 1.5918675523581167e-05,
+      "loss": 0.2291,
+      "step": 3524
+    },
+    {
+      "epoch": 0.7444561774023232,
+      "grad_norm": 0.2593635618686676,
+      "learning_rate": 1.5893886112632077e-05,
+      "loss": 0.224,
+      "step": 3525
+    },
+    {
+      "epoch": 0.7446673706441393,
+      "grad_norm": 0.35546278953552246,
+      "learning_rate": 1.5869112370365857e-05,
+      "loss": 0.2254,
+      "step": 3526
+    },
+    {
+      "epoch": 0.7448785638859556,
+      "grad_norm": 0.363542377948761,
+      "learning_rate": 1.584435430816376e-05,
+      "loss": 0.2233,
+      "step": 3527
+    },
+    {
+      "epoch": 0.7450897571277719,
+      "grad_norm": 0.2731059789657593,
+      "learning_rate": 1.581961193739988e-05,
+      "loss": 0.2246,
+      "step": 3528
+    },
+    {
+      "epoch": 0.7453009503695882,
+      "grad_norm": 0.35688555240631104,
+      "learning_rate": 1.5794885269441153e-05,
+      "loss": 0.2277,
+      "step": 3529
+    },
+    {
+      "epoch": 0.7455121436114044,
+      "grad_norm": 0.31173327565193176,
+      "learning_rate": 1.5770174315647186e-05,
+      "loss": 0.2296,
+      "step": 3530
+    },
+    {
+      "epoch": 0.7457233368532207,
+      "grad_norm": 0.34857887029647827,
+      "learning_rate": 1.5745479087370486e-05,
+      "loss": 0.2345,
+      "step": 3531
+    },
+    {
+      "epoch": 0.745934530095037,
+      "grad_norm": 0.2730456292629242,
+      "learning_rate": 1.5720799595956275e-05,
+      "loss": 0.2286,
+      "step": 3532
+    },
+    {
+      "epoch": 0.7461457233368533,
+      "grad_norm": 0.3208880126476288,
+      "learning_rate": 1.5696135852742515e-05,
+      "loss": 0.225,
+      "step": 3533
+    },
+    {
+      "epoch": 0.7463569165786695,
+      "grad_norm": 0.3132641017436981,
+      "learning_rate": 1.5671487869060014e-05,
+      "loss": 0.2289,
+      "step": 3534
+    },
+    {
+      "epoch": 0.7465681098204857,
+      "grad_norm": 0.28669142723083496,
+      "learning_rate": 1.5646855656232296e-05,
+      "loss": 0.2277,
+      "step": 3535
+    },
+    {
+      "epoch": 0.746779303062302,
+      "grad_norm": 0.2914336621761322,
+      "learning_rate": 1.5622239225575598e-05,
+      "loss": 0.2278,
+      "step": 3536
+    },
+    {
+      "epoch": 0.7469904963041183,
+      "grad_norm": 0.2772876024246216,
+      "learning_rate": 1.5597638588398987e-05,
+      "loss": 0.2252,
+      "step": 3537
+    },
+    {
+      "epoch": 0.7472016895459346,
+      "grad_norm": 0.412874311208725,
+      "learning_rate": 1.557305375600425e-05,
+      "loss": 0.223,
+      "step": 3538
+    },
+    {
+      "epoch": 0.7474128827877508,
+      "grad_norm": 0.4259602129459381,
+      "learning_rate": 1.5548484739685848e-05,
+      "loss": 0.2236,
+      "step": 3539
+    },
+    {
+      "epoch": 0.747624076029567,
+      "grad_norm": 0.29791224002838135,
+      "learning_rate": 1.5523931550731068e-05,
+      "loss": 0.2254,
+      "step": 3540
+    },
+    {
+      "epoch": 0.7478352692713833,
+      "grad_norm": 0.2956344187259674,
+      "learning_rate": 1.549939420041989e-05,
+      "loss": 0.2294,
+      "step": 3541
+    },
+    {
+      "epoch": 0.7480464625131996,
+      "grad_norm": 0.26207178831100464,
+      "learning_rate": 1.5474872700024972e-05,
+      "loss": 0.2223,
+      "step": 3542
+    },
+    {
+      "epoch": 0.7482576557550158,
+      "grad_norm": 0.3007482886314392,
+      "learning_rate": 1.545036706081177e-05,
+      "loss": 0.2294,
+      "step": 3543
+    },
+    {
+      "epoch": 0.7484688489968321,
+      "grad_norm": 0.35405242443084717,
+      "learning_rate": 1.54258772940384e-05,
+      "loss": 0.2216,
+      "step": 3544
+    },
+    {
+      "epoch": 0.7486800422386484,
+      "grad_norm": 0.2643187642097473,
+      "learning_rate": 1.5401403410955706e-05,
+      "loss": 0.2237,
+      "step": 3545
+    },
+    {
+      "epoch": 0.7488912354804647,
+      "grad_norm": 0.2909862995147705,
+      "learning_rate": 1.5376945422807227e-05,
+      "loss": 0.2272,
+      "step": 3546
+    },
+    {
+      "epoch": 0.7491024287222808,
+      "grad_norm": 0.33378252387046814,
+      "learning_rate": 1.53525033408292e-05,
+      "loss": 0.2262,
+      "step": 3547
+    },
+    {
+      "epoch": 0.7493136219640971,
+      "grad_norm": 0.41008588671684265,
+      "learning_rate": 1.5328077176250562e-05,
+      "loss": 0.2311,
+      "step": 3548
+    },
+    {
+      "epoch": 0.7495248152059134,
+      "grad_norm": 0.30998528003692627,
+      "learning_rate": 1.5303666940292926e-05,
+      "loss": 0.2281,
+      "step": 3549
+    },
+    {
+      "epoch": 0.7497360084477297,
+      "grad_norm": 0.28377652168273926,
+      "learning_rate": 1.52792726441706e-05,
+      "loss": 0.2244,
+      "step": 3550
+    },
+    {
+      "epoch": 0.749947201689546,
+      "grad_norm": 0.27934324741363525,
+      "learning_rate": 1.5254894299090555e-05,
+      "loss": 0.2229,
+      "step": 3551
+    },
+    {
+      "epoch": 0.7501583949313622,
+      "grad_norm": 0.274677574634552,
+      "learning_rate": 1.523053191625245e-05,
+      "loss": 0.224,
+      "step": 3552
+    },
+    {
+      "epoch": 0.7503695881731784,
+      "grad_norm": 0.3279375731945038,
+      "learning_rate": 1.5206185506848597e-05,
+      "loss": 0.2215,
+      "step": 3553
+    },
+    {
+      "epoch": 0.7505807814149947,
+      "grad_norm": 0.33308112621307373,
+      "learning_rate": 1.5181855082063972e-05,
+      "loss": 0.2234,
+      "step": 3554
+    },
+    {
+      "epoch": 0.750791974656811,
+      "grad_norm": 0.36448001861572266,
+      "learning_rate": 1.5157540653076219e-05,
+      "loss": 0.2193,
+      "step": 3555
+    },
+    {
+      "epoch": 0.7510031678986272,
+      "grad_norm": 0.27325108647346497,
+      "learning_rate": 1.5133242231055622e-05,
+      "loss": 0.2198,
+      "step": 3556
+    },
+    {
+      "epoch": 0.7512143611404435,
+      "grad_norm": 0.26204001903533936,
+      "learning_rate": 1.5108959827165108e-05,
+      "loss": 0.2226,
+      "step": 3557
+    },
+    {
+      "epoch": 0.7514255543822598,
+      "grad_norm": 0.3190080225467682,
+      "learning_rate": 1.5084693452560262e-05,
+      "loss": 0.2215,
+      "step": 3558
+    },
+    {
+      "epoch": 0.7516367476240761,
+      "grad_norm": 0.2986411154270172,
+      "learning_rate": 1.5060443118389283e-05,
+      "loss": 0.2273,
+      "step": 3559
+    },
+    {
+      "epoch": 0.7518479408658922,
+      "grad_norm": 0.38574182987213135,
+      "learning_rate": 1.503620883579302e-05,
+      "loss": 0.2236,
+      "step": 3560
+    },
+    {
+      "epoch": 0.7520591341077085,
+      "grad_norm": 0.29530057311058044,
+      "learning_rate": 1.5011990615904919e-05,
+      "loss": 0.2187,
+      "step": 3561
+    },
+    {
+      "epoch": 0.7522703273495248,
+      "grad_norm": 0.5279195308685303,
+      "learning_rate": 1.4987788469851117e-05,
+      "loss": 0.2203,
+      "step": 3562
+    },
+    {
+      "epoch": 0.7524815205913411,
+      "grad_norm": 0.27077093720436096,
+      "learning_rate": 1.4963602408750271e-05,
+      "loss": 0.226,
+      "step": 3563
+    },
+    {
+      "epoch": 0.7526927138331574,
+      "grad_norm": 0.3102589249610901,
+      "learning_rate": 1.4939432443713697e-05,
+      "loss": 0.2196,
+      "step": 3564
+    },
+    {
+      "epoch": 0.7529039070749736,
+      "grad_norm": 0.31125330924987793,
+      "learning_rate": 1.491527858584535e-05,
+      "loss": 0.2206,
+      "step": 3565
+    },
+    {
+      "epoch": 0.7531151003167899,
+      "grad_norm": 0.2739287316799164,
+      "learning_rate": 1.4891140846241713e-05,
+      "loss": 0.2208,
+      "step": 3566
+    },
+    {
+      "epoch": 0.7533262935586061,
+      "grad_norm": 0.401899129152298,
+      "learning_rate": 1.4867019235991897e-05,
+      "loss": 0.2213,
+      "step": 3567
+    },
+    {
+      "epoch": 0.7535374868004224,
+      "grad_norm": 0.2940559387207031,
+      "learning_rate": 1.4842913766177652e-05,
+      "loss": 0.2195,
+      "step": 3568
+    },
+    {
+      "epoch": 0.7537486800422386,
+      "grad_norm": 0.33487170934677124,
+      "learning_rate": 1.4818824447873214e-05,
+      "loss": 0.2258,
+      "step": 3569
+    },
+    {
+      "epoch": 0.7539598732840549,
+      "grad_norm": 0.2845116853713989,
+      "learning_rate": 1.4794751292145465e-05,
+      "loss": 0.2205,
+      "step": 3570
+    },
+    {
+      "epoch": 0.7541710665258712,
+      "grad_norm": 0.2983340919017792,
+      "learning_rate": 1.4770694310053867e-05,
+      "loss": 0.2171,
+      "step": 3571
+    },
+    {
+      "epoch": 0.7543822597676875,
+      "grad_norm": 0.309155136346817,
+      "learning_rate": 1.4746653512650444e-05,
+      "loss": 0.2221,
+      "step": 3572
+    },
+    {
+      "epoch": 0.7545934530095036,
+      "grad_norm": 0.3143276274204254,
+      "learning_rate": 1.4722628910979718e-05,
+      "loss": 0.2195,
+      "step": 3573
+    },
+    {
+      "epoch": 0.7548046462513199,
+      "grad_norm": 0.3242114186286926,
+      "learning_rate": 1.4698620516078882e-05,
+      "loss": 0.2233,
+      "step": 3574
+    },
+    {
+      "epoch": 0.7550158394931362,
+      "grad_norm": 0.4509166181087494,
+      "learning_rate": 1.4674628338977603e-05,
+      "loss": 0.2173,
+      "step": 3575
+    },
+    {
+      "epoch": 0.7552270327349525,
+      "grad_norm": 0.63100665807724,
+      "learning_rate": 1.4650652390698127e-05,
+      "loss": 0.2222,
+      "step": 3576
+    },
+    {
+      "epoch": 0.7554382259767688,
+      "grad_norm": 0.3597918748855591,
+      "learning_rate": 1.4626692682255239e-05,
+      "loss": 0.2156,
+      "step": 3577
+    },
+    {
+      "epoch": 0.755649419218585,
+      "grad_norm": 0.3552287817001343,
+      "learning_rate": 1.4602749224656264e-05,
+      "loss": 0.2243,
+      "step": 3578
+    },
+    {
+      "epoch": 0.7558606124604013,
+      "grad_norm": 0.28725680708885193,
+      "learning_rate": 1.4578822028901063e-05,
+      "loss": 0.2233,
+      "step": 3579
+    },
+    {
+      "epoch": 0.7560718057022175,
+      "grad_norm": 0.326539546251297,
+      "learning_rate": 1.4554911105982021e-05,
+      "loss": 0.2186,
+      "step": 3580
+    },
+    {
+      "epoch": 0.7562829989440338,
+      "grad_norm": 0.33115634322166443,
+      "learning_rate": 1.4531016466884051e-05,
+      "loss": 0.2225,
+      "step": 3581
+    },
+    {
+      "epoch": 0.75649419218585,
+      "grad_norm": 0.2883807420730591,
+      "learning_rate": 1.450713812258458e-05,
+      "loss": 0.2204,
+      "step": 3582
+    },
+    {
+      "epoch": 0.7567053854276663,
+      "grad_norm": 0.2897742986679077,
+      "learning_rate": 1.4483276084053565e-05,
+      "loss": 0.223,
+      "step": 3583
+    },
+    {
+      "epoch": 0.7569165786694826,
+      "grad_norm": 0.28932350873947144,
+      "learning_rate": 1.4459430362253445e-05,
+      "loss": 0.2262,
+      "step": 3584
+    },
+    {
+      "epoch": 0.7571277719112989,
+      "grad_norm": 0.33162614703178406,
+      "learning_rate": 1.443560096813919e-05,
+      "loss": 0.224,
+      "step": 3585
+    },
+    {
+      "epoch": 0.757338965153115,
+      "grad_norm": 0.3268282115459442,
+      "learning_rate": 1.4411787912658259e-05,
+      "loss": 0.2296,
+      "step": 3586
+    },
+    {
+      "epoch": 0.7575501583949313,
+      "grad_norm": 0.3075575530529022,
+      "learning_rate": 1.4387991206750589e-05,
+      "loss": 0.2258,
+      "step": 3587
+    },
+    {
+      "epoch": 0.7577613516367476,
+      "grad_norm": 0.30517885088920593,
+      "learning_rate": 1.4364210861348632e-05,
+      "loss": 0.2153,
+      "step": 3588
+    },
+    {
+      "epoch": 0.7579725448785639,
+      "grad_norm": 0.3361615240573883,
+      "learning_rate": 1.4340446887377306e-05,
+      "loss": 0.2196,
+      "step": 3589
+    },
+    {
+      "epoch": 0.7581837381203802,
+      "grad_norm": 0.31835293769836426,
+      "learning_rate": 1.4316699295754015e-05,
+      "loss": 0.2204,
+      "step": 3590
+    },
+    {
+      "epoch": 0.7583949313621964,
+      "grad_norm": 0.2958146333694458,
+      "learning_rate": 1.4292968097388637e-05,
+      "loss": 0.2217,
+      "step": 3591
+    },
+    {
+      "epoch": 0.7586061246040127,
+      "grad_norm": 0.3093109130859375,
+      "learning_rate": 1.4269253303183516e-05,
+      "loss": 0.2289,
+      "step": 3592
+    },
+    {
+      "epoch": 0.758817317845829,
+      "grad_norm": 0.6568298935890198,
+      "learning_rate": 1.424555492403346e-05,
+      "loss": 0.2283,
+      "step": 3593
+    },
+    {
+      "epoch": 0.7590285110876452,
+      "grad_norm": 0.2708004415035248,
+      "learning_rate": 1.4221872970825734e-05,
+      "loss": 0.2198,
+      "step": 3594
+    },
+    {
+      "epoch": 0.7592397043294614,
+      "grad_norm": 0.29651880264282227,
+      "learning_rate": 1.4198207454440049e-05,
+      "loss": 0.2283,
+      "step": 3595
+    },
+    {
+      "epoch": 0.7594508975712777,
+      "grad_norm": 0.4528253376483917,
+      "learning_rate": 1.4174558385748616e-05,
+      "loss": 0.225,
+      "step": 3596
+    },
+    {
+      "epoch": 0.759662090813094,
+      "grad_norm": 0.3122156858444214,
+      "learning_rate": 1.4150925775616008e-05,
+      "loss": 0.2213,
+      "step": 3597
+    },
+    {
+      "epoch": 0.7598732840549103,
+      "grad_norm": 0.30818402767181396,
+      "learning_rate": 1.4127309634899277e-05,
+      "loss": 0.2197,
+      "step": 3598
+    },
+    {
+      "epoch": 0.7600844772967265,
+      "grad_norm": 0.3199005722999573,
+      "learning_rate": 1.4103709974447948e-05,
+      "loss": 0.2171,
+      "step": 3599
+    },
+    {
+      "epoch": 0.7602956705385427,
+      "grad_norm": 0.2860449552536011,
+      "learning_rate": 1.4080126805103899e-05,
+      "loss": 0.2238,
+      "step": 3600
+    },
+    {
+      "epoch": 0.7602956705385427,
+      "eval_train_split_loss": 0.33796408772468567,
+      "eval_train_split_runtime": 10.0355,
+      "eval_train_split_samples_per_second": 0.996,
+      "eval_train_split_steps_per_second": 0.1,
+      "step": 3600
+    },
+    {
+      "epoch": 0.760506863780359,
+      "grad_norm": 0.29735130071640015,
+      "learning_rate": 1.405656013770147e-05,
+      "loss": 0.2286,
+      "step": 3601
+    },
+    {
+      "epoch": 0.7607180570221753,
+      "grad_norm": 0.3449653685092926,
+      "learning_rate": 1.4033009983067452e-05,
+      "loss": 0.224,
+      "step": 3602
+    },
+    {
+      "epoch": 0.7609292502639915,
+      "grad_norm": 1.8111618757247925,
+      "learning_rate": 1.4009476352020973e-05,
+      "loss": 0.2207,
+      "step": 3603
+    },
+    {
+      "epoch": 0.7611404435058078,
+      "grad_norm": 0.33743923902511597,
+      "learning_rate": 1.3985959255373643e-05,
+      "loss": 0.22,
+      "step": 3604
+    },
+    {
+      "epoch": 0.7613516367476241,
+      "grad_norm": 0.25553837418556213,
+      "learning_rate": 1.3962458703929459e-05,
+      "loss": 0.2218,
+      "step": 3605
+    },
+    {
+      "epoch": 0.7615628299894404,
+      "grad_norm": 0.43794170022010803,
+      "learning_rate": 1.3938974708484753e-05,
+      "loss": 0.2186,
+      "step": 3606
+    },
+    {
+      "epoch": 0.7617740232312566,
+      "grad_norm": 0.33731117844581604,
+      "learning_rate": 1.3915507279828349e-05,
+      "loss": 0.2207,
+      "step": 3607
+    },
+    {
+      "epoch": 0.7619852164730728,
+      "grad_norm": 0.3402393162250519,
+      "learning_rate": 1.3892056428741412e-05,
+      "loss": 0.2241,
+      "step": 3608
+    },
+    {
+      "epoch": 0.7621964097148891,
+      "grad_norm": 0.3905806541442871,
+      "learning_rate": 1.3868622165997453e-05,
+      "loss": 0.2192,
+      "step": 3609
+    },
+    {
+      "epoch": 0.7624076029567054,
+      "grad_norm": 0.30946120619773865,
+      "learning_rate": 1.384520450236244e-05,
+      "loss": 0.2207,
+      "step": 3610
+    },
+    {
+      "epoch": 0.7626187961985217,
+      "grad_norm": 0.2980096936225891,
+      "learning_rate": 1.3821803448594683e-05,
+      "loss": 0.2194,
+      "step": 3611
+    },
+    {
+      "epoch": 0.7628299894403379,
+      "grad_norm": 0.3370390832424164,
+      "learning_rate": 1.3798419015444807e-05,
+      "loss": 0.2241,
+      "step": 3612
+    },
+    {
+      "epoch": 0.7630411826821542,
+      "grad_norm": 0.3493973910808563,
+      "learning_rate": 1.3775051213655893e-05,
+      "loss": 0.2212,
+      "step": 3613
+    },
+    {
+      "epoch": 0.7632523759239704,
+      "grad_norm": 0.3412136733531952,
+      "learning_rate": 1.3751700053963345e-05,
+      "loss": 0.2155,
+      "step": 3614
+    },
+    {
+      "epoch": 0.7634635691657867,
+      "grad_norm": 0.4158650040626526,
+      "learning_rate": 1.3728365547094862e-05,
+      "loss": 0.2207,
+      "step": 3615
+    },
+    {
+      "epoch": 0.7636747624076029,
+      "grad_norm": 0.3107655346393585,
+      "learning_rate": 1.370504770377059e-05,
+      "loss": 0.228,
+      "step": 3616
+    },
+    {
+      "epoch": 0.7638859556494192,
+      "grad_norm": 0.3422705829143524,
+      "learning_rate": 1.3681746534702972e-05,
+      "loss": 0.2253,
+      "step": 3617
+    },
+    {
+      "epoch": 0.7640971488912355,
+      "grad_norm": 0.31944483518600464,
+      "learning_rate": 1.3658462050596777e-05,
+      "loss": 0.2247,
+      "step": 3618
+    },
+    {
+      "epoch": 0.7643083421330518,
+      "grad_norm": 0.328176885843277,
+      "learning_rate": 1.3635194262149131e-05,
+      "loss": 0.2159,
+      "step": 3619
+    },
+    {
+      "epoch": 0.764519535374868,
+      "grad_norm": 0.2856539487838745,
+      "learning_rate": 1.3611943180049491e-05,
+      "loss": 0.223,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7647307286166842,
+      "grad_norm": 0.30189183354377747,
+      "learning_rate": 1.3588708814979622e-05,
+      "loss": 0.2225,
+      "step": 3621
+    },
+    {
+      "epoch": 0.7649419218585005,
+      "grad_norm": 0.24742412567138672,
+      "learning_rate": 1.356549117761362e-05,
+      "loss": 0.2206,
+      "step": 3622
+    },
+    {
+      "epoch": 0.7651531151003168,
+      "grad_norm": 0.2826240360736847,
+      "learning_rate": 1.3542290278617903e-05,
+      "loss": 0.2304,
+      "step": 3623
+    },
+    {
+      "epoch": 0.7653643083421331,
+      "grad_norm": 0.34409356117248535,
+      "learning_rate": 1.3519106128651188e-05,
+      "loss": 0.2234,
+      "step": 3624
+    },
+    {
+      "epoch": 0.7655755015839493,
+      "grad_norm": 0.3377842307090759,
+      "learning_rate": 1.3495938738364495e-05,
+      "loss": 0.2365,
+      "step": 3625
+    },
+    {
+      "epoch": 0.7657866948257656,
+      "grad_norm": 0.3474664092063904,
+      "learning_rate": 1.3472788118401153e-05,
+      "loss": 0.2251,
+      "step": 3626
+    },
+    {
+      "epoch": 0.7659978880675818,
+      "grad_norm": 0.28234872221946716,
+      "learning_rate": 1.344965427939679e-05,
+      "loss": 0.2183,
+      "step": 3627
+    },
+    {
+      "epoch": 0.7662090813093981,
+      "grad_norm": 0.3007965683937073,
+      "learning_rate": 1.3426537231979307e-05,
+      "loss": 0.2253,
+      "step": 3628
+    },
+    {
+      "epoch": 0.7664202745512143,
+      "grad_norm": 0.3345962464809418,
+      "learning_rate": 1.3403436986768907e-05,
+      "loss": 0.2266,
+      "step": 3629
+    },
+    {
+      "epoch": 0.7666314677930306,
+      "grad_norm": 0.3023263216018677,
+      "learning_rate": 1.3380353554378073e-05,
+      "loss": 0.2295,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7668426610348469,
+      "grad_norm": 0.3497113287448883,
+      "learning_rate": 1.3357286945411545e-05,
+      "loss": 0.2253,
+      "step": 3631
+    },
+    {
+      "epoch": 0.7670538542766632,
+      "grad_norm": 0.28914695978164673,
+      "learning_rate": 1.3334237170466385e-05,
+      "loss": 0.2213,
+      "step": 3632
+    },
+    {
+      "epoch": 0.7672650475184795,
+      "grad_norm": 0.3095831871032715,
+      "learning_rate": 1.3311204240131846e-05,
+      "loss": 0.2225,
+      "step": 3633
+    },
+    {
+      "epoch": 0.7674762407602956,
+      "grad_norm": 0.3422868549823761,
+      "learning_rate": 1.328818816498949e-05,
+      "loss": 0.2227,
+      "step": 3634
+    },
+    {
+      "epoch": 0.7676874340021119,
+      "grad_norm": 0.25861942768096924,
+      "learning_rate": 1.3265188955613156e-05,
+      "loss": 0.2308,
+      "step": 3635
+    },
+    {
+      "epoch": 0.7678986272439282,
+      "grad_norm": 0.3195645809173584,
+      "learning_rate": 1.3242206622568875e-05,
+      "loss": 0.2195,
+      "step": 3636
+    },
+    {
+      "epoch": 0.7681098204857445,
+      "grad_norm": 0.3015667498111725,
+      "learning_rate": 1.3219241176414948e-05,
+      "loss": 0.2272,
+      "step": 3637
+    },
+    {
+      "epoch": 0.7683210137275607,
+      "grad_norm": 0.27908793091773987,
+      "learning_rate": 1.3196292627701972e-05,
+      "loss": 0.2239,
+      "step": 3638
+    },
+    {
+      "epoch": 0.768532206969377,
+      "grad_norm": 0.27370062470436096,
+      "learning_rate": 1.31733609869727e-05,
+      "loss": 0.2243,
+      "step": 3639
+    },
+    {
+      "epoch": 0.7687434002111933,
+      "grad_norm": 0.29777437448501587,
+      "learning_rate": 1.3150446264762134e-05,
+      "loss": 0.2212,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7689545934530095,
+      "grad_norm": 0.3053264617919922,
+      "learning_rate": 1.3127548471597585e-05,
+      "loss": 0.2258,
+      "step": 3641
+    },
+    {
+      "epoch": 0.7691657866948257,
+      "grad_norm": 0.28448060154914856,
+      "learning_rate": 1.3104667617998462e-05,
+      "loss": 0.2287,
+      "step": 3642
+    },
+    {
+      "epoch": 0.769376979936642,
+      "grad_norm": 0.3041512966156006,
+      "learning_rate": 1.3081803714476466e-05,
+      "loss": 0.2231,
+      "step": 3643
+    },
+    {
+      "epoch": 0.7695881731784583,
+      "grad_norm": 0.25535285472869873,
+      "learning_rate": 1.3058956771535541e-05,
+      "loss": 0.2211,
+      "step": 3644
+    },
+    {
+      "epoch": 0.7697993664202746,
+      "grad_norm": 0.3355608880519867,
+      "learning_rate": 1.3036126799671733e-05,
+      "loss": 0.2185,
+      "step": 3645
+    },
+    {
+      "epoch": 0.7700105596620909,
+      "grad_norm": 0.3039676547050476,
+      "learning_rate": 1.3013313809373395e-05,
+      "loss": 0.2237,
+      "step": 3646
+    },
+    {
+      "epoch": 0.770221752903907,
+      "grad_norm": 0.30452316999435425,
+      "learning_rate": 1.299051781112105e-05,
+      "loss": 0.2234,
+      "step": 3647
+    },
+    {
+      "epoch": 0.7704329461457233,
+      "grad_norm": 0.26513028144836426,
+      "learning_rate": 1.2967738815387354e-05,
+      "loss": 0.2256,
+      "step": 3648
+    },
+    {
+      "epoch": 0.7706441393875396,
+      "grad_norm": 0.4958110451698303,
+      "learning_rate": 1.294497683263724e-05,
+      "loss": 0.219,
+      "step": 3649
+    },
+    {
+      "epoch": 0.7708553326293559,
+      "grad_norm": 0.5657695531845093,
+      "learning_rate": 1.2922231873327778e-05,
+      "loss": 0.2222,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7710665258711721,
+      "grad_norm": 0.3007160425186157,
+      "learning_rate": 1.289950394790822e-05,
+      "loss": 0.2171,
+      "step": 3651
+    },
+    {
+      "epoch": 0.7712777191129884,
+      "grad_norm": 0.24289099872112274,
+      "learning_rate": 1.2876793066820003e-05,
+      "loss": 0.2227,
+      "step": 3652
+    },
+    {
+      "epoch": 0.7714889123548047,
+      "grad_norm": 0.34365901350975037,
+      "learning_rate": 1.285409924049672e-05,
+      "loss": 0.223,
+      "step": 3653
+    },
+    {
+      "epoch": 0.7717001055966209,
+      "grad_norm": 0.2881329357624054,
+      "learning_rate": 1.2831422479364142e-05,
+      "loss": 0.2217,
+      "step": 3654
+    },
+    {
+      "epoch": 0.7719112988384371,
+      "grad_norm": 0.27679798007011414,
+      "learning_rate": 1.2808762793840201e-05,
+      "loss": 0.2268,
+      "step": 3655
+    },
+    {
+      "epoch": 0.7721224920802534,
+      "grad_norm": 0.27743399143218994,
+      "learning_rate": 1.2786120194334966e-05,
+      "loss": 0.2278,
+      "step": 3656
+    },
+    {
+      "epoch": 0.7723336853220697,
+      "grad_norm": 0.43033382296562195,
+      "learning_rate": 1.2763494691250677e-05,
+      "loss": 0.2323,
+      "step": 3657
+    },
+    {
+      "epoch": 0.772544878563886,
+      "grad_norm": 0.28780096769332886,
+      "learning_rate": 1.2740886294981713e-05,
+      "loss": 0.2182,
+      "step": 3658
+    },
+    {
+      "epoch": 0.7727560718057023,
+      "grad_norm": 0.32196471095085144,
+      "learning_rate": 1.2718295015914588e-05,
+      "loss": 0.2261,
+      "step": 3659
+    },
+    {
+      "epoch": 0.7729672650475184,
+      "grad_norm": 0.5061303973197937,
+      "learning_rate": 1.2695720864427962e-05,
+      "loss": 0.2248,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7731784582893347,
+      "grad_norm": 0.2920447587966919,
+      "learning_rate": 1.2673163850892617e-05,
+      "loss": 0.2207,
+      "step": 3661
+    },
+    {
+      "epoch": 0.773389651531151,
+      "grad_norm": 0.2918713390827179,
+      "learning_rate": 1.2650623985671466e-05,
+      "loss": 0.2235,
+      "step": 3662
+    },
+    {
+      "epoch": 0.7736008447729673,
+      "grad_norm": 0.28495314717292786,
+      "learning_rate": 1.2628101279119548e-05,
+      "loss": 0.2227,
+      "step": 3663
+    },
+    {
+      "epoch": 0.7738120380147835,
+      "grad_norm": 0.26784253120422363,
+      "learning_rate": 1.2605595741584015e-05,
+      "loss": 0.2222,
+      "step": 3664
+    },
+    {
+      "epoch": 0.7740232312565998,
+      "grad_norm": 0.2865445613861084,
+      "learning_rate": 1.2583107383404125e-05,
+      "loss": 0.2273,
+      "step": 3665
+    },
+    {
+      "epoch": 0.7742344244984161,
+      "grad_norm": 0.28624555468559265,
+      "learning_rate": 1.2560636214911254e-05,
+      "loss": 0.2285,
+      "step": 3666
+    },
+    {
+      "epoch": 0.7744456177402324,
+      "grad_norm": 0.26315784454345703,
+      "learning_rate": 1.2538182246428876e-05,
+      "loss": 0.2228,
+      "step": 3667
+    },
+    {
+      "epoch": 0.7746568109820485,
+      "grad_norm": 0.34696194529533386,
+      "learning_rate": 1.2515745488272562e-05,
+      "loss": 0.2231,
+      "step": 3668
+    },
+    {
+      "epoch": 0.7748680042238648,
+      "grad_norm": 0.28931328654289246,
+      "learning_rate": 1.2493325950749985e-05,
+      "loss": 0.228,
+      "step": 3669
+    },
+    {
+      "epoch": 0.7750791974656811,
+      "grad_norm": 0.2860826253890991,
+      "learning_rate": 1.2470923644160898e-05,
+      "loss": 0.216,
+      "step": 3670
+    },
+    {
+      "epoch": 0.7752903907074974,
+      "grad_norm": 0.2673848569393158,
+      "learning_rate": 1.2448538578797137e-05,
+      "loss": 0.2224,
+      "step": 3671
+    },
+    {
+      "epoch": 0.7755015839493136,
+      "grad_norm": 0.25385138392448425,
+      "learning_rate": 1.2426170764942625e-05,
+      "loss": 0.2216,
+      "step": 3672
+    },
+    {
+      "epoch": 0.7757127771911299,
+      "grad_norm": 0.27441370487213135,
+      "learning_rate": 1.2403820212873347e-05,
+      "loss": 0.2213,
+      "step": 3673
+    },
+    {
+      "epoch": 0.7759239704329461,
+      "grad_norm": 0.26668065786361694,
+      "learning_rate": 1.2381486932857356e-05,
+      "loss": 0.2236,
+      "step": 3674
+    },
+    {
+      "epoch": 0.7761351636747624,
+      "grad_norm": 0.2837276756763458,
+      "learning_rate": 1.235917093515483e-05,
+      "loss": 0.2239,
+      "step": 3675
+    },
+    {
+      "epoch": 0.7763463569165787,
+      "grad_norm": 0.376228004693985,
+      "learning_rate": 1.2336872230017882e-05,
+      "loss": 0.2189,
+      "step": 3676
+    },
+    {
+      "epoch": 0.7765575501583949,
+      "grad_norm": 0.28172555565834045,
+      "learning_rate": 1.2314590827690814e-05,
+      "loss": 0.2228,
+      "step": 3677
+    },
+    {
+      "epoch": 0.7767687434002112,
+      "grad_norm": 0.2666279375553131,
+      "learning_rate": 1.2292326738409915e-05,
+      "loss": 0.224,
+      "step": 3678
+    },
+    {
+      "epoch": 0.7769799366420275,
+      "grad_norm": 0.30366265773773193,
+      "learning_rate": 1.2270079972403487e-05,
+      "loss": 0.2187,
+      "step": 3679
+    },
+    {
+      "epoch": 0.7771911298838438,
+      "grad_norm": 0.2981114983558655,
+      "learning_rate": 1.2247850539891948e-05,
+      "loss": 0.2197,
+      "step": 3680
+    },
+    {
+      "epoch": 0.7774023231256599,
+      "grad_norm": 0.3134240210056305,
+      "learning_rate": 1.2225638451087717e-05,
+      "loss": 0.2195,
+      "step": 3681
+    },
+    {
+      "epoch": 0.7776135163674762,
+      "grad_norm": 0.3291384279727936,
+      "learning_rate": 1.2203443716195212e-05,
+      "loss": 0.2198,
+      "step": 3682
+    },
+    {
+      "epoch": 0.7778247096092925,
+      "grad_norm": 0.2621394693851471,
+      "learning_rate": 1.218126634541094e-05,
+      "loss": 0.2231,
+      "step": 3683
+    },
+    {
+      "epoch": 0.7780359028511088,
+      "grad_norm": 0.3253099322319031,
+      "learning_rate": 1.2159106348923422e-05,
+      "loss": 0.2195,
+      "step": 3684
+    },
+    {
+      "epoch": 0.778247096092925,
+      "grad_norm": 0.3076798617839813,
+      "learning_rate": 1.2136963736913116e-05,
+      "loss": 0.223,
+      "step": 3685
+    },
+    {
+      "epoch": 0.7784582893347413,
+      "grad_norm": 0.29127275943756104,
+      "learning_rate": 1.2114838519552602e-05,
+      "loss": 0.2199,
+      "step": 3686
+    },
+    {
+      "epoch": 0.7786694825765575,
+      "grad_norm": 0.2575564980506897,
+      "learning_rate": 1.2092730707006428e-05,
+      "loss": 0.2241,
+      "step": 3687
+    },
+    {
+      "epoch": 0.7788806758183738,
+      "grad_norm": 0.3233320415019989,
+      "learning_rate": 1.2070640309431097e-05,
+      "loss": 0.2217,
+      "step": 3688
+    },
+    {
+      "epoch": 0.7790918690601901,
+      "grad_norm": 0.30449554324150085,
+      "learning_rate": 1.2048567336975187e-05,
+      "loss": 0.2188,
+      "step": 3689
+    },
+    {
+      "epoch": 0.7793030623020063,
+      "grad_norm": 0.2536589503288269,
+      "learning_rate": 1.2026511799779233e-05,
+      "loss": 0.2207,
+      "step": 3690
+    },
+    {
+      "epoch": 0.7795142555438226,
+      "grad_norm": 0.28331613540649414,
+      "learning_rate": 1.2004473707975761e-05,
+      "loss": 0.2218,
+      "step": 3691
+    },
+    {
+      "epoch": 0.7797254487856389,
+      "grad_norm": 0.30778369307518005,
+      "learning_rate": 1.1982453071689286e-05,
+      "loss": 0.2203,
+      "step": 3692
+    },
+    {
+      "epoch": 0.7799366420274552,
+      "grad_norm": 0.29477658867836,
+      "learning_rate": 1.1960449901036308e-05,
+      "loss": 0.22,
+      "step": 3693
+    },
+    {
+      "epoch": 0.7801478352692713,
+      "grad_norm": 0.26436418294906616,
+      "learning_rate": 1.1938464206125294e-05,
+      "loss": 0.2213,
+      "step": 3694
+    },
+    {
+      "epoch": 0.7803590285110876,
+      "grad_norm": 0.26844093203544617,
+      "learning_rate": 1.1916495997056692e-05,
+      "loss": 0.222,
+      "step": 3695
+    },
+    {
+      "epoch": 0.7805702217529039,
+      "grad_norm": 0.4809677302837372,
+      "learning_rate": 1.1894545283922909e-05,
+      "loss": 0.2194,
+      "step": 3696
+    },
+    {
+      "epoch": 0.7807814149947202,
+      "grad_norm": 0.28438395261764526,
+      "learning_rate": 1.1872612076808316e-05,
+      "loss": 0.2199,
+      "step": 3697
+    },
+    {
+      "epoch": 0.7809926082365364,
+      "grad_norm": 0.26844537258148193,
+      "learning_rate": 1.185069638578925e-05,
+      "loss": 0.2197,
+      "step": 3698
+    },
+    {
+      "epoch": 0.7812038014783527,
+      "grad_norm": 0.27767565846443176,
+      "learning_rate": 1.1828798220933984e-05,
+      "loss": 0.2187,
+      "step": 3699
+    },
+    {
+      "epoch": 0.781414994720169,
+      "grad_norm": 0.28150296211242676,
+      "learning_rate": 1.1806917592302762e-05,
+      "loss": 0.2182,
+      "step": 3700
+    },
+    {
+      "epoch": 0.7816261879619852,
+      "grad_norm": 0.3027787506580353,
+      "learning_rate": 1.1785054509947752e-05,
+      "loss": 0.2186,
+      "step": 3701
+    },
+    {
+      "epoch": 0.7818373812038015,
+      "grad_norm": 0.27628862857818604,
+      "learning_rate": 1.1763208983913065e-05,
+      "loss": 0.2182,
+      "step": 3702
+    },
+    {
+      "epoch": 0.7820485744456177,
+      "grad_norm": 0.27323129773139954,
+      "learning_rate": 1.1741381024234755e-05,
+      "loss": 0.221,
+      "step": 3703
+    },
+    {
+      "epoch": 0.782259767687434,
+      "grad_norm": 0.27808889746665955,
+      "learning_rate": 1.1719570640940785e-05,
+      "loss": 0.2197,
+      "step": 3704
+    },
+    {
+      "epoch": 0.7824709609292503,
+      "grad_norm": 0.37524423003196716,
+      "learning_rate": 1.1697777844051105e-05,
+      "loss": 0.2239,
+      "step": 3705
+    },
+    {
+      "epoch": 0.7826821541710666,
+      "grad_norm": 0.2600768506526947,
+      "learning_rate": 1.1676002643577488e-05,
+      "loss": 0.2215,
+      "step": 3706
+    },
+    {
+      "epoch": 0.7828933474128827,
+      "grad_norm": 0.2768322229385376,
+      "learning_rate": 1.1654245049523683e-05,
+      "loss": 0.2198,
+      "step": 3707
+    },
+    {
+      "epoch": 0.783104540654699,
+      "grad_norm": 0.26683509349823,
+      "learning_rate": 1.163250507188538e-05,
+      "loss": 0.2197,
+      "step": 3708
+    },
+    {
+      "epoch": 0.7833157338965153,
+      "grad_norm": 0.2741868197917938,
+      "learning_rate": 1.1610782720650098e-05,
+      "loss": 0.2196,
+      "step": 3709
+    },
+    {
+      "epoch": 0.7835269271383316,
+      "grad_norm": 0.3654620349407196,
+      "learning_rate": 1.1589078005797294e-05,
+      "loss": 0.222,
+      "step": 3710
+    },
+    {
+      "epoch": 0.7837381203801478,
+      "grad_norm": 0.38217106461524963,
+      "learning_rate": 1.1567390937298372e-05,
+      "loss": 0.2228,
+      "step": 3711
+    },
+    {
+      "epoch": 0.7839493136219641,
+      "grad_norm": 0.30361416935920715,
+      "learning_rate": 1.1545721525116537e-05,
+      "loss": 0.2202,
+      "step": 3712
+    },
+    {
+      "epoch": 0.7841605068637804,
+      "grad_norm": 0.3072085976600647,
+      "learning_rate": 1.1524069779206937e-05,
+      "loss": 0.2233,
+      "step": 3713
+    },
+    {
+      "epoch": 0.7843717001055966,
+      "grad_norm": 0.3204714059829712,
+      "learning_rate": 1.1502435709516629e-05,
+      "loss": 0.2204,
+      "step": 3714
+    },
+    {
+      "epoch": 0.7845828933474129,
+      "grad_norm": 0.31630611419677734,
+      "learning_rate": 1.1480819325984488e-05,
+      "loss": 0.2215,
+      "step": 3715
+    },
+    {
+      "epoch": 0.7847940865892291,
+      "grad_norm": 0.31686869263648987,
+      "learning_rate": 1.1459220638541274e-05,
+      "loss": 0.2201,
+      "step": 3716
+    },
+    {
+      "epoch": 0.7850052798310454,
+      "grad_norm": 0.2705281376838684,
+      "learning_rate": 1.1437639657109694e-05,
+      "loss": 0.2254,
+      "step": 3717
+    },
+    {
+      "epoch": 0.7852164730728617,
+      "grad_norm": 0.2708519399166107,
+      "learning_rate": 1.1416076391604197e-05,
+      "loss": 0.2185,
+      "step": 3718
+    },
+    {
+      "epoch": 0.785427666314678,
+      "grad_norm": 0.29281097650527954,
+      "learning_rate": 1.1394530851931206e-05,
+      "loss": 0.2165,
+      "step": 3719
+    },
+    {
+      "epoch": 0.7856388595564942,
+      "grad_norm": 0.29911696910858154,
+      "learning_rate": 1.137300304798895e-05,
+      "loss": 0.2184,
+      "step": 3720
+    },
+    {
+      "epoch": 0.7858500527983104,
+      "grad_norm": 0.2319680005311966,
+      "learning_rate": 1.1351492989667467e-05,
+      "loss": 0.2157,
+      "step": 3721
+    },
+    {
+      "epoch": 0.7860612460401267,
+      "grad_norm": 0.4123140871524811,
+      "learning_rate": 1.133000068684874e-05,
+      "loss": 0.2192,
+      "step": 3722
+    },
+    {
+      "epoch": 0.786272439281943,
+      "grad_norm": 0.2689760625362396,
+      "learning_rate": 1.1308526149406534e-05,
+      "loss": 0.221,
+      "step": 3723
+    },
+    {
+      "epoch": 0.7864836325237592,
+      "grad_norm": 0.28222358226776123,
+      "learning_rate": 1.1287069387206429e-05,
+      "loss": 0.2225,
+      "step": 3724
+    },
+    {
+      "epoch": 0.7866948257655755,
+      "grad_norm": 0.5241146683692932,
+      "learning_rate": 1.1265630410105904e-05,
+      "loss": 0.2275,
+      "step": 3725
+    },
+    {
+      "epoch": 0.7869060190073918,
+      "grad_norm": 0.35710373520851135,
+      "learning_rate": 1.1244209227954238e-05,
+      "loss": 0.2219,
+      "step": 3726
+    },
+    {
+      "epoch": 0.787117212249208,
+      "grad_norm": 0.2734037935733795,
+      "learning_rate": 1.1222805850592499e-05,
+      "loss": 0.2234,
+      "step": 3727
+    },
+    {
+      "epoch": 0.7873284054910243,
+      "grad_norm": 0.3271646201610565,
+      "learning_rate": 1.1201420287853636e-05,
+      "loss": 0.2287,
+      "step": 3728
+    },
+    {
+      "epoch": 0.7875395987328405,
+      "grad_norm": 0.24987001717090607,
+      "learning_rate": 1.1180052549562381e-05,
+      "loss": 0.2207,
+      "step": 3729
+    },
+    {
+      "epoch": 0.7877507919746568,
+      "grad_norm": 0.28393831849098206,
+      "learning_rate": 1.1158702645535286e-05,
+      "loss": 0.2165,
+      "step": 3730
+    },
+    {
+      "epoch": 0.7879619852164731,
+      "grad_norm": 0.25528639554977417,
+      "learning_rate": 1.1137370585580698e-05,
+      "loss": 0.2186,
+      "step": 3731
+    },
+    {
+      "epoch": 0.7881731784582894,
+      "grad_norm": 0.29122254252433777,
+      "learning_rate": 1.1116056379498773e-05,
+      "loss": 0.221,
+      "step": 3732
+    },
+    {
+      "epoch": 0.7883843717001056,
+      "grad_norm": 0.273124098777771,
+      "learning_rate": 1.109476003708147e-05,
+      "loss": 0.223,
+      "step": 3733
+    },
+    {
+      "epoch": 0.7885955649419218,
+      "grad_norm": 0.27969440817832947,
+      "learning_rate": 1.1073481568112543e-05,
+      "loss": 0.2197,
+      "step": 3734
+    },
+    {
+      "epoch": 0.7888067581837381,
+      "grad_norm": 0.2827796936035156,
+      "learning_rate": 1.1052220982367516e-05,
+      "loss": 0.2283,
+      "step": 3735
+    },
+    {
+      "epoch": 0.7890179514255544,
+      "grad_norm": 0.25983762741088867,
+      "learning_rate": 1.1030978289613726e-05,
+      "loss": 0.2268,
+      "step": 3736
+    },
+    {
+      "epoch": 0.7892291446673706,
+      "grad_norm": 0.34274354577064514,
+      "learning_rate": 1.1009753499610258e-05,
+      "loss": 0.2221,
+      "step": 3737
+    },
+    {
+      "epoch": 0.7894403379091869,
+      "grad_norm": 0.33697566390037537,
+      "learning_rate": 1.0988546622107997e-05,
+      "loss": 0.2166,
+      "step": 3738
+    },
+    {
+      "epoch": 0.7896515311510032,
+      "grad_norm": 0.3219049274921417,
+      "learning_rate": 1.0967357666849593e-05,
+      "loss": 0.2169,
+      "step": 3739
+    },
+    {
+      "epoch": 0.7898627243928195,
+      "grad_norm": 0.26265716552734375,
+      "learning_rate": 1.0946186643569455e-05,
+      "loss": 0.2194,
+      "step": 3740
+    },
+    {
+      "epoch": 0.7900739176346356,
+      "grad_norm": 0.29123547673225403,
+      "learning_rate": 1.0925033561993758e-05,
+      "loss": 0.2166,
+      "step": 3741
+    },
+    {
+      "epoch": 0.7902851108764519,
+      "grad_norm": 0.29279303550720215,
+      "learning_rate": 1.0903898431840443e-05,
+      "loss": 0.2222,
+      "step": 3742
+    },
+    {
+      "epoch": 0.7904963041182682,
+      "grad_norm": 0.26037994027137756,
+      "learning_rate": 1.088278126281918e-05,
+      "loss": 0.226,
+      "step": 3743
+    },
+    {
+      "epoch": 0.7907074973600845,
+      "grad_norm": 0.29777008295059204,
+      "learning_rate": 1.0861682064631423e-05,
+      "loss": 0.2253,
+      "step": 3744
+    },
+    {
+      "epoch": 0.7909186906019008,
+      "grad_norm": 0.2514497935771942,
+      "learning_rate": 1.0840600846970334e-05,
+      "loss": 0.2183,
+      "step": 3745
+    },
+    {
+      "epoch": 0.791129883843717,
+      "grad_norm": 0.29592713713645935,
+      "learning_rate": 1.0819537619520819e-05,
+      "loss": 0.2226,
+      "step": 3746
+    },
+    {
+      "epoch": 0.7913410770855333,
+      "grad_norm": 0.28066590428352356,
+      "learning_rate": 1.0798492391959587e-05,
+      "loss": 0.2183,
+      "step": 3747
+    },
+    {
+      "epoch": 0.7915522703273495,
+      "grad_norm": 0.2791113555431366,
+      "learning_rate": 1.0777465173954959e-05,
+      "loss": 0.2235,
+      "step": 3748
+    },
+    {
+      "epoch": 0.7917634635691658,
+      "grad_norm": 0.38841208815574646,
+      "learning_rate": 1.0756455975167063e-05,
+      "loss": 0.2189,
+      "step": 3749
+    },
+    {
+      "epoch": 0.791974656810982,
+      "grad_norm": 0.27778786420822144,
+      "learning_rate": 1.0735464805247764e-05,
+      "loss": 0.2169,
+      "step": 3750
+    },
+    {
+      "epoch": 0.7921858500527983,
+      "grad_norm": 0.2839658260345459,
+      "learning_rate": 1.071449167384057e-05,
+      "loss": 0.2203,
+      "step": 3751
+    },
+    {
+      "epoch": 0.7923970432946146,
+      "grad_norm": 0.33445149660110474,
+      "learning_rate": 1.0693536590580745e-05,
+      "loss": 0.2223,
+      "step": 3752
+    },
+    {
+      "epoch": 0.7926082365364309,
+      "grad_norm": 0.2822965383529663,
+      "learning_rate": 1.0672599565095282e-05,
+      "loss": 0.2228,
+      "step": 3753
+    },
+    {
+      "epoch": 0.792819429778247,
+      "grad_norm": 0.27931153774261475,
+      "learning_rate": 1.0651680607002862e-05,
+      "loss": 0.2218,
+      "step": 3754
+    },
+    {
+      "epoch": 0.7930306230200633,
+      "grad_norm": 0.3273766040802002,
+      "learning_rate": 1.063077972591382e-05,
+      "loss": 0.2203,
+      "step": 3755
+    },
+    {
+      "epoch": 0.7932418162618796,
+      "grad_norm": 0.2704015374183655,
+      "learning_rate": 1.0609896931430257e-05,
+      "loss": 0.2206,
+      "step": 3756
+    },
+    {
+      "epoch": 0.7934530095036959,
+      "grad_norm": 0.27211397886276245,
+      "learning_rate": 1.058903223314594e-05,
+      "loss": 0.2231,
+      "step": 3757
+    },
+    {
+      "epoch": 0.7936642027455122,
+      "grad_norm": 0.36250758171081543,
+      "learning_rate": 1.0568185640646272e-05,
+      "loss": 0.2241,
+      "step": 3758
+    },
+    {
+      "epoch": 0.7938753959873284,
+      "grad_norm": 0.3082680404186249,
+      "learning_rate": 1.0547357163508425e-05,
+      "loss": 0.2266,
+      "step": 3759
+    },
+    {
+      "epoch": 0.7940865892291447,
+      "grad_norm": 0.2691916227340698,
+      "learning_rate": 1.0526546811301202e-05,
+      "loss": 0.2217,
+      "step": 3760
+    },
+    {
+      "epoch": 0.794297782470961,
+      "grad_norm": 0.25225189328193665,
+      "learning_rate": 1.050575459358505e-05,
+      "loss": 0.2225,
+      "step": 3761
+    },
+    {
+      "epoch": 0.7945089757127772,
+      "grad_norm": 0.3065657913684845,
+      "learning_rate": 1.0484980519912146e-05,
+      "loss": 0.2241,
+      "step": 3762
+    },
+    {
+      "epoch": 0.7947201689545934,
+      "grad_norm": 0.36688101291656494,
+      "learning_rate": 1.04642245998263e-05,
+      "loss": 0.2241,
+      "step": 3763
+    },
+    {
+      "epoch": 0.7949313621964097,
+      "grad_norm": 0.26096591353416443,
+      "learning_rate": 1.0443486842862981e-05,
+      "loss": 0.2239,
+      "step": 3764
+    },
+    {
+      "epoch": 0.795142555438226,
+      "grad_norm": 0.2687482237815857,
+      "learning_rate": 1.0422767258549315e-05,
+      "loss": 0.2269,
+      "step": 3765
+    },
+    {
+      "epoch": 0.7953537486800423,
+      "grad_norm": 0.34550273418426514,
+      "learning_rate": 1.0402065856404087e-05,
+      "loss": 0.2201,
+      "step": 3766
+    },
+    {
+      "epoch": 0.7955649419218584,
+      "grad_norm": 0.27149704098701477,
+      "learning_rate": 1.0381382645937721e-05,
+      "loss": 0.2208,
+      "step": 3767
+    },
+    {
+      "epoch": 0.7957761351636747,
+      "grad_norm": 0.30042722821235657,
+      "learning_rate": 1.036071763665229e-05,
+      "loss": 0.2258,
+      "step": 3768
+    },
+    {
+      "epoch": 0.795987328405491,
+      "grad_norm": 0.2944454848766327,
+      "learning_rate": 1.0340070838041498e-05,
+      "loss": 0.2181,
+      "step": 3769
+    },
+    {
+      "epoch": 0.7961985216473073,
+      "grad_norm": 0.2650624215602875,
+      "learning_rate": 1.0319442259590683e-05,
+      "loss": 0.2163,
+      "step": 3770
+    },
+    {
+      "epoch": 0.7964097148891236,
+      "grad_norm": 0.27900776267051697,
+      "learning_rate": 1.0298831910776824e-05,
+      "loss": 0.2254,
+      "step": 3771
+    },
+    {
+      "epoch": 0.7966209081309398,
+      "grad_norm": 0.3331044018268585,
+      "learning_rate": 1.0278239801068517e-05,
+      "loss": 0.2221,
+      "step": 3772
+    },
+    {
+      "epoch": 0.7968321013727561,
+      "grad_norm": 0.27012506127357483,
+      "learning_rate": 1.025766593992597e-05,
+      "loss": 0.2188,
+      "step": 3773
+    },
+    {
+      "epoch": 0.7970432946145724,
+      "grad_norm": 0.30857330560684204,
+      "learning_rate": 1.0237110336801026e-05,
+      "loss": 0.2192,
+      "step": 3774
+    },
+    {
+      "epoch": 0.7972544878563886,
+      "grad_norm": 0.30294448137283325,
+      "learning_rate": 1.0216573001137126e-05,
+      "loss": 0.2214,
+      "step": 3775
+    },
+    {
+      "epoch": 0.7974656810982048,
+      "grad_norm": 0.2993299663066864,
+      "learning_rate": 1.0196053942369321e-05,
+      "loss": 0.219,
+      "step": 3776
+    },
+    {
+      "epoch": 0.7976768743400211,
+      "grad_norm": 0.28119978308677673,
+      "learning_rate": 1.0175553169924256e-05,
+      "loss": 0.229,
+      "step": 3777
+    },
+    {
+      "epoch": 0.7978880675818374,
+      "grad_norm": 0.3257310092449188,
+      "learning_rate": 1.0155070693220231e-05,
+      "loss": 0.2272,
+      "step": 3778
+    },
+    {
+      "epoch": 0.7980992608236537,
+      "grad_norm": 0.2899830639362335,
+      "learning_rate": 1.0134606521667051e-05,
+      "loss": 0.2242,
+      "step": 3779
+    },
+    {
+      "epoch": 0.7983104540654699,
+      "grad_norm": 0.33668455481529236,
+      "learning_rate": 1.0114160664666155e-05,
+      "loss": 0.2259,
+      "step": 3780
+    },
+    {
+      "epoch": 0.7985216473072861,
+      "grad_norm": 0.3063999116420746,
+      "learning_rate": 1.0093733131610622e-05,
+      "loss": 0.2218,
+      "step": 3781
+    },
+    {
+      "epoch": 0.7987328405491024,
+      "grad_norm": 0.3060685992240906,
+      "learning_rate": 1.007332393188501e-05,
+      "loss": 0.222,
+      "step": 3782
+    },
+    {
+      "epoch": 0.7989440337909187,
+      "grad_norm": 0.39808642864227295,
+      "learning_rate": 1.0052933074865505e-05,
+      "loss": 0.2221,
+      "step": 3783
+    },
+    {
+      "epoch": 0.799155227032735,
+      "grad_norm": 0.2952180802822113,
+      "learning_rate": 1.0032560569919913e-05,
+      "loss": 0.2232,
+      "step": 3784
+    },
+    {
+      "epoch": 0.7993664202745512,
+      "grad_norm": 0.28649866580963135,
+      "learning_rate": 1.0012206426407516e-05,
+      "loss": 0.2217,
+      "step": 3785
+    },
+    {
+      "epoch": 0.7995776135163675,
+      "grad_norm": 0.307332307100296,
+      "learning_rate": 9.991870653679215e-06,
+      "loss": 0.2232,
+      "step": 3786
+    },
+    {
+      "epoch": 0.7997888067581838,
+      "grad_norm": 0.2866005301475525,
+      "learning_rate": 9.971553261077493e-06,
+      "loss": 0.2158,
+      "step": 3787
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3303745687007904,
+      "learning_rate": 9.951254257936322e-06,
+      "loss": 0.2265,
+      "step": 3788
+    },
+    {
+      "epoch": 0.8002111932418162,
+      "grad_norm": 0.30812233686447144,
+      "learning_rate": 9.93097365358127e-06,
+      "loss": 0.2201,
+      "step": 3789
+    },
+    {
+      "epoch": 0.8004223864836325,
+      "grad_norm": 0.24219447374343872,
+      "learning_rate": 9.91071145732948e-06,
+      "loss": 0.2207,
+      "step": 3790
+    },
+    {
+      "epoch": 0.8006335797254488,
+      "grad_norm": 0.37459608912467957,
+      "learning_rate": 9.890467678489546e-06,
+      "loss": 0.2216,
+      "step": 3791
+    },
+    {
+      "epoch": 0.8008447729672651,
+      "grad_norm": 0.25979873538017273,
+      "learning_rate": 9.870242326361705e-06,
+      "loss": 0.2244,
+      "step": 3792
+    },
+    {
+      "epoch": 0.8010559662090813,
+      "grad_norm": 0.4562368094921112,
+      "learning_rate": 9.850035410237685e-06,
+      "loss": 0.2206,
+      "step": 3793
+    },
+    {
+      "epoch": 0.8012671594508975,
+      "grad_norm": 0.29072609543800354,
+      "learning_rate": 9.829846939400689e-06,
+      "loss": 0.2204,
+      "step": 3794
+    },
+    {
+      "epoch": 0.8014783526927138,
+      "grad_norm": 0.28391319513320923,
+      "learning_rate": 9.809676923125549e-06,
+      "loss": 0.2249,
+      "step": 3795
+    },
+    {
+      "epoch": 0.8016895459345301,
+      "grad_norm": 0.2969712018966675,
+      "learning_rate": 9.78952537067856e-06,
+      "loss": 0.2269,
+      "step": 3796
+    },
+    {
+      "epoch": 0.8019007391763464,
+      "grad_norm": 0.2981444001197815,
+      "learning_rate": 9.769392291317514e-06,
+      "loss": 0.222,
+      "step": 3797
+    },
+    {
+      "epoch": 0.8021119324181626,
+      "grad_norm": 0.2914969027042389,
+      "learning_rate": 9.74927769429177e-06,
+      "loss": 0.2172,
+      "step": 3798
+    },
+    {
+      "epoch": 0.8023231256599789,
+      "grad_norm": 0.2950490713119507,
+      "learning_rate": 9.729181588842185e-06,
+      "loss": 0.2171,
+      "step": 3799
+    },
+    {
+      "epoch": 0.8025343189017952,
+      "grad_norm": 0.3009367883205414,
+      "learning_rate": 9.709103984201056e-06,
+      "loss": 0.2145,
+      "step": 3800
+    },
+    {
+      "epoch": 0.8025343189017952,
+      "eval_train_split_loss": 0.38565295934677124,
+      "eval_train_split_runtime": 6.9945,
+      "eval_train_split_samples_per_second": 1.43,
+      "eval_train_split_steps_per_second": 0.143,
+      "step": 3800
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 4735,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 200,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 9.235974288663013e+20,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}