diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7105 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.862579281183932,
+  "eval_steps": 500,
+  "global_step": 11500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.826529679169359e-06,
+      "loss": 2.951,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2794598875617875e-06,
+      "loss": 2.8814,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.129369575636334e-06,
+      "loss": 2.5232,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.732390095954215e-06,
+      "loss": 1.7918,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.20012914994629e-06,
+      "loss": 1.1308,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.582299784028761e-06,
+      "loss": 0.4893,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.905420451106023e-06,
+      "loss": 0.4265,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.185320304346644e-06,
+      "loss": 0.574,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.432209472103306e-06,
+      "loss": 0.6531,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.653059358338718e-06,
+      "loss": 0.3698,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.852842381755405e-06,
+      "loss": 0.596,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999145810199027e-06,
+      "loss": 0.4422,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99060391218929e-06,
+      "loss": 0.4861,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982062014179552e-06,
+      "loss": 0.5141,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973520116169813e-06,
+      "loss": 0.4774,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964978218160076e-06,
+      "loss": 0.4923,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956436320150338e-06,
+      "loss": 0.3808,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.947894422140601e-06,
+      "loss": 0.347,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939352524130863e-06,
+      "loss": 0.6926,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930810626121124e-06,
+      "loss": 0.3291,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.922268728111387e-06,
+      "loss": 0.3808,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91372683010165e-06,
+      "loss": 0.4422,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.905184932091912e-06,
+      "loss": 0.31,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.896643034082173e-06,
+      "loss": 0.4053,
+      "step": 240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.888101136072437e-06,
+      "loss": 0.6166,
+      "step": 250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.879559238062698e-06,
+      "loss": 0.6013,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.871017340052961e-06,
+      "loss": 0.2876,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.862475442043223e-06,
+      "loss": 0.6953,
+      "step": 280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.853933544033484e-06,
+      "loss": 0.4655,
+      "step": 290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.845391646023748e-06,
+      "loss": 0.4543,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.836849748014009e-06,
+      "loss": 0.2808,
+      "step": 310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.828307850004272e-06,
+      "loss": 0.5441,
+      "step": 320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.819765951994534e-06,
+      "loss": 0.3876,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.811224053984795e-06,
+      "loss": 0.5788,
+      "step": 340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.802682155975059e-06,
+      "loss": 0.434,
+      "step": 350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.794140257965322e-06,
+      "loss": 0.4627,
+      "step": 360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.785598359955583e-06,
+      "loss": 0.4837,
+      "step": 370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.777056461945845e-06,
+      "loss": 0.2981,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.768514563936108e-06,
+      "loss": 0.3747,
+      "step": 390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.75997266592637e-06,
+      "loss": 0.4526,
+      "step": 400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.751430767916633e-06,
+      "loss": 0.3543,
+      "step": 410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.742888869906894e-06,
+      "loss": 0.3935,
+      "step": 420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.734346971897156e-06,
+      "loss": 0.2884,
+      "step": 430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.725805073887419e-06,
+      "loss": 0.2348,
+      "step": 440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.71726317587768e-06,
+      "loss": 0.4702,
+      "step": 450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.708721277867944e-06,
+      "loss": 0.4125,
+      "step": 460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.700179379858205e-06,
+      "loss": 0.4766,
+      "step": 470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.691637481848467e-06,
+      "loss": 0.3776,
+      "step": 480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.68309558383873e-06,
+      "loss": 0.0589,
+      "step": 490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.674553685828993e-06,
+      "loss": 0.3958,
+      "step": 500
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 0.3221363127231598,
+      "eval_runtime": 149.4575,
+      "eval_samples_per_second": 13.522,
+      "eval_steps_per_second": 2.255,
+      "step": 500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.666011787819254e-06,
+      "loss": 0.4896,
+      "step": 510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.657469889809516e-06,
+      "loss": 0.4172,
+      "step": 520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.64892799179978e-06,
+      "loss": 0.3452,
+      "step": 530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.64038609379004e-06,
+      "loss": 0.4153,
+      "step": 540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.631844195780304e-06,
+      "loss": 0.4012,
+      "step": 550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.623302297770565e-06,
+      "loss": 0.4108,
+      "step": 560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.614760399760827e-06,
+      "loss": 0.34,
+      "step": 570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.60621850175109e-06,
+      "loss": 0.3014,
+      "step": 580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.597676603741352e-06,
+      "loss": 0.4368,
+      "step": 590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.589134705731615e-06,
+      "loss": 0.4598,
+      "step": 600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.580592807721876e-06,
+      "loss": 0.4462,
+      "step": 610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.572050909712138e-06,
+      "loss": 0.3651,
+      "step": 620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.563509011702401e-06,
+      "loss": 0.4889,
+      "step": 630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.554967113692664e-06,
+      "loss": 0.324,
+      "step": 640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.546425215682926e-06,
+      "loss": 0.4731,
+      "step": 650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.537883317673187e-06,
+      "loss": 0.3661,
+      "step": 660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.52934141966345e-06,
+      "loss": 0.5358,
+      "step": 670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.520799521653712e-06,
+      "loss": 0.5142,
+      "step": 680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.512257623643975e-06,
+      "loss": 0.2004,
+      "step": 690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.503715725634237e-06,
+      "loss": 0.3359,
+      "step": 700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.495173827624498e-06,
+      "loss": 0.5546,
+      "step": 710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.486631929614761e-06,
+      "loss": 0.507,
+      "step": 720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.478090031605025e-06,
+      "loss": 0.3003,
+      "step": 730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.469548133595286e-06,
+      "loss": 0.303,
+      "step": 740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.461006235585548e-06,
+      "loss": 0.1986,
+      "step": 750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.452464337575809e-06,
+      "loss": 0.3847,
+      "step": 760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.443922439566072e-06,
+      "loss": 0.3421,
+      "step": 770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.435380541556335e-06,
+      "loss": 0.1399,
+      "step": 780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.426838643546597e-06,
+      "loss": 0.4816,
+      "step": 790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.418296745536858e-06,
+      "loss": 0.2678,
+      "step": 800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.409754847527122e-06,
+      "loss": 0.4857,
+      "step": 810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.401212949517383e-06,
+      "loss": 0.4605,
+      "step": 820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.392671051507646e-06,
+      "loss": 0.4682,
+      "step": 830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.384129153497908e-06,
+      "loss": 0.4602,
+      "step": 840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.37558725548817e-06,
+      "loss": 0.2615,
+      "step": 850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.367045357478433e-06,
+      "loss": 0.2202,
+      "step": 860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.358503459468696e-06,
+      "loss": 0.3829,
+      "step": 870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.349961561458957e-06,
+      "loss": 0.3233,
+      "step": 880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.341419663449219e-06,
+      "loss": 0.4153,
+      "step": 890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.33287776543948e-06,
+      "loss": 0.4387,
+      "step": 900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.324335867429743e-06,
+      "loss": 0.3553,
+      "step": 910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.315793969420007e-06,
+      "loss": 0.2822,
+      "step": 920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.307252071410268e-06,
+      "loss": 0.5052,
+      "step": 930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.29871017340053e-06,
+      "loss": 0.3052,
+      "step": 940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.290168275390793e-06,
+      "loss": 0.2768,
+      "step": 950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.281626377381054e-06,
+      "loss": 0.5307,
+      "step": 960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.273084479371318e-06,
+      "loss": 0.3067,
+      "step": 970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.264542581361579e-06,
+      "loss": 0.5421,
+      "step": 980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.25600068335184e-06,
+      "loss": 0.4828,
+      "step": 990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.247458785342104e-06,
+      "loss": 0.2279,
+      "step": 1000
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 0.31820133328437805,
+      "eval_runtime": 147.8659,
+      "eval_samples_per_second": 13.668,
+      "eval_steps_per_second": 2.279,
+      "step": 1000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.238916887332367e-06,
+      "loss": 0.3588,
+      "step": 1010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.230374989322628e-06,
+      "loss": 0.5093,
+      "step": 1020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.22183309131289e-06,
+      "loss": 0.2865,
+      "step": 1030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.213291193303153e-06,
+      "loss": 0.4917,
+      "step": 1040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.204749295293415e-06,
+      "loss": 0.2988,
+      "step": 1050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.196207397283678e-06,
+      "loss": 0.2698,
+      "step": 1060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.18766549927394e-06,
+      "loss": 0.3299,
+      "step": 1070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.179123601264201e-06,
+      "loss": 0.3823,
+      "step": 1080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.170581703254464e-06,
+      "loss": 0.3843,
+      "step": 1090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.162039805244726e-06,
+      "loss": 0.1743,
+      "step": 1100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.153497907234989e-06,
+      "loss": 0.2736,
+      "step": 1110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.14495600922525e-06,
+      "loss": 0.2336,
+      "step": 1120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.136414111215512e-06,
+      "loss": 0.393,
+      "step": 1130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.127872213205775e-06,
+      "loss": 0.3948,
+      "step": 1140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.119330315196038e-06,
+      "loss": 0.4436,
+      "step": 1150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.1107884171863e-06,
+      "loss": 0.3944,
+      "step": 1160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.102246519176561e-06,
+      "loss": 0.2986,
+      "step": 1170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.093704621166824e-06,
+      "loss": 0.622,
+      "step": 1180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.085162723157086e-06,
+      "loss": 0.4492,
+      "step": 1190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.076620825147349e-06,
+      "loss": 0.3485,
+      "step": 1200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.06807892713761e-06,
+      "loss": 0.2461,
+      "step": 1210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.059537029127872e-06,
+      "loss": 0.3418,
+      "step": 1220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.050995131118135e-06,
+      "loss": 0.3734,
+      "step": 1230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.042453233108397e-06,
+      "loss": 0.3566,
+      "step": 1240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.03391133509866e-06,
+      "loss": 0.4122,
+      "step": 1250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.025369437088922e-06,
+      "loss": 0.4349,
+      "step": 1260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.016827539079183e-06,
+      "loss": 0.3687,
+      "step": 1270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.008285641069446e-06,
+      "loss": 0.3744,
+      "step": 1280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.99974374305971e-06,
+      "loss": 0.3763,
+      "step": 1290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.991201845049971e-06,
+      "loss": 0.2757,
+      "step": 1300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.982659947040232e-06,
+      "loss": 0.1441,
+      "step": 1310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.974118049030496e-06,
+      "loss": 0.2886,
+      "step": 1320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.965576151020757e-06,
+      "loss": 0.2811,
+      "step": 1330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.95703425301102e-06,
+      "loss": 0.449,
+      "step": 1340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.948492355001282e-06,
+      "loss": 0.3085,
+      "step": 1350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.939950456991543e-06,
+      "loss": 0.3286,
+      "step": 1360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.931408558981807e-06,
+      "loss": 0.4316,
+      "step": 1370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.922866660972068e-06,
+      "loss": 0.4838,
+      "step": 1380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.914324762962331e-06,
+      "loss": 0.3533,
+      "step": 1390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.905782864952593e-06,
+      "loss": 0.2597,
+      "step": 1400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.897240966942854e-06,
+      "loss": 0.4335,
+      "step": 1410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.888699068933117e-06,
+      "loss": 0.3623,
+      "step": 1420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.88015717092338e-06,
+      "loss": 0.4639,
+      "step": 1430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.871615272913642e-06,
+      "loss": 0.2467,
+      "step": 1440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.863073374903904e-06,
+      "loss": 0.58,
+      "step": 1450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.854531476894167e-06,
+      "loss": 0.1977,
+      "step": 1460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.845989578884428e-06,
+      "loss": 0.4213,
+      "step": 1470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.837447680874692e-06,
+      "loss": 0.4523,
+      "step": 1480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.828905782864953e-06,
+      "loss": 0.3502,
+      "step": 1490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.820363884855215e-06,
+      "loss": 0.4642,
+      "step": 1500
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 0.31634676456451416,
+      "eval_runtime": 148.1948,
+      "eval_samples_per_second": 13.637,
+      "eval_steps_per_second": 2.274,
+      "step": 1500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.811821986845478e-06,
+      "loss": 0.4238,
+      "step": 1510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.80328008883574e-06,
+      "loss": 0.3259,
+      "step": 1520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.794738190826002e-06,
+      "loss": 0.2573,
+      "step": 1530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.786196292816264e-06,
+      "loss": 0.4213,
+      "step": 1540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.777654394806525e-06,
+      "loss": 0.4176,
+      "step": 1550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.769112496796789e-06,
+      "loss": 0.3298,
+      "step": 1560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.760570598787052e-06,
+      "loss": 0.3727,
+      "step": 1570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.752028700777313e-06,
+      "loss": 0.2212,
+      "step": 1580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.743486802767575e-06,
+      "loss": 0.4808,
+      "step": 1590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.734944904757838e-06,
+      "loss": 0.3223,
+      "step": 1600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.7264030067481e-06,
+      "loss": 0.4816,
+      "step": 1610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.717861108738363e-06,
+      "loss": 0.372,
+      "step": 1620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.709319210728624e-06,
+      "loss": 0.3286,
+      "step": 1630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.700777312718886e-06,
+      "loss": 0.3396,
+      "step": 1640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.692235414709149e-06,
+      "loss": 0.5711,
+      "step": 1650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.683693516699412e-06,
+      "loss": 0.4977,
+      "step": 1660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.675151618689674e-06,
+      "loss": 0.3088,
+      "step": 1670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.666609720679935e-06,
+      "loss": 0.4884,
+      "step": 1680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.658067822670197e-06,
+      "loss": 0.3061,
+      "step": 1690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.64952592466046e-06,
+      "loss": 0.4866,
+      "step": 1700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.640984026650723e-06,
+      "loss": 0.4727,
+      "step": 1710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.632442128640985e-06,
+      "loss": 0.341,
+      "step": 1720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.623900230631246e-06,
+      "loss": 0.3919,
+      "step": 1730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.61535833262151e-06,
+      "loss": 0.4861,
+      "step": 1740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.60681643461177e-06,
+      "loss": 0.3785,
+      "step": 1750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.598274536602034e-06,
+      "loss": 0.2698,
+      "step": 1760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.589732638592296e-06,
+      "loss": 0.4744,
+      "step": 1770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.581190740582557e-06,
+      "loss": 0.3691,
+      "step": 1780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.57264884257282e-06,
+      "loss": 0.4302,
+      "step": 1790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.564106944563083e-06,
+      "loss": 0.5078,
+      "step": 1800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.555565046553345e-06,
+      "loss": 0.353,
+      "step": 1810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.547023148543606e-06,
+      "loss": 0.3028,
+      "step": 1820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.538481250533868e-06,
+      "loss": 0.3568,
+      "step": 1830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.529939352524131e-06,
+      "loss": 0.3159,
+      "step": 1840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.521397454514394e-06,
+      "loss": 0.3822,
+      "step": 1850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.512855556504656e-06,
+      "loss": 0.2627,
+      "step": 1860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.504313658494917e-06,
+      "loss": 0.2919,
+      "step": 1870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.49577176048518e-06,
+      "loss": 0.3402,
+      "step": 1880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.487229862475442e-06,
+      "loss": 0.1855,
+      "step": 1890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.478687964465705e-06,
+      "loss": 0.3334,
+      "step": 1900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.470146066455967e-06,
+      "loss": 0.3978,
+      "step": 1910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.461604168446228e-06,
+      "loss": 0.3597,
+      "step": 1920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.453062270436491e-06,
+      "loss": 0.2911,
+      "step": 1930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.444520372426755e-06,
+      "loss": 0.5741,
+      "step": 1940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.435978474417016e-06,
+      "loss": 0.3767,
+      "step": 1950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.427436576407278e-06,
+      "loss": 0.2925,
+      "step": 1960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.418894678397541e-06,
+      "loss": 0.3217,
+      "step": 1970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.410352780387802e-06,
+      "loss": 0.4271,
+      "step": 1980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.401810882378066e-06,
+      "loss": 0.363,
+      "step": 1990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.393268984368327e-06,
+      "loss": 0.3269,
+      "step": 2000
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.31494140625,
+      "eval_runtime": 147.7201,
+      "eval_samples_per_second": 13.681,
+      "eval_steps_per_second": 2.281,
+      "step": 2000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.38472708635859e-06,
+      "loss": 0.3319,
+      "step": 2010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.376185188348852e-06,
+      "loss": 0.3003,
+      "step": 2020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.367643290339113e-06,
+      "loss": 0.3746,
+      "step": 2030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.359101392329376e-06,
+      "loss": 0.4951,
+      "step": 2040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.35055949431964e-06,
+      "loss": 0.4965,
+      "step": 2050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.342017596309901e-06,
+      "loss": 0.4376,
+      "step": 2060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.333475698300163e-06,
+      "loss": 0.294,
+      "step": 2070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.324933800290426e-06,
+      "loss": 0.3379,
+      "step": 2080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.316391902280687e-06,
+      "loss": 0.2212,
+      "step": 2090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.30785000427095e-06,
+      "loss": 0.3702,
+      "step": 2100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.299308106261212e-06,
+      "loss": 0.4595,
+      "step": 2110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.290766208251474e-06,
+      "loss": 0.1368,
+      "step": 2120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.282224310241737e-06,
+      "loss": 0.2244,
+      "step": 2130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.273682412231998e-06,
+      "loss": 0.4319,
+      "step": 2140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.265140514222262e-06,
+      "loss": 0.4145,
+      "step": 2150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.256598616212523e-06,
+      "loss": 0.6454,
+      "step": 2160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.248056718202785e-06,
+      "loss": 0.402,
+      "step": 2170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.239514820193048e-06,
+      "loss": 0.3575,
+      "step": 2180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.230972922183311e-06,
+      "loss": 0.3174,
+      "step": 2190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.222431024173572e-06,
+      "loss": 0.3201,
+      "step": 2200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.213889126163834e-06,
+      "loss": 0.4354,
+      "step": 2210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.205347228154097e-06,
+      "loss": 0.3313,
+      "step": 2220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.196805330144359e-06,
+      "loss": 0.3215,
+      "step": 2230
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.188263432134622e-06,
+      "loss": 0.4446,
+      "step": 2240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.179721534124883e-06,
+      "loss": 0.2536,
+      "step": 2250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.171179636115145e-06,
+      "loss": 0.3579,
+      "step": 2260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.162637738105408e-06,
+      "loss": 0.3149,
+      "step": 2270
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.154095840095671e-06,
+      "loss": 0.4549,
+      "step": 2280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.145553942085933e-06,
+      "loss": 0.4434,
+      "step": 2290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.137012044076194e-06,
+      "loss": 0.3353,
+      "step": 2300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.128470146066456e-06,
+      "loss": 0.3522,
+      "step": 2310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.119928248056719e-06,
+      "loss": 0.4549,
+      "step": 2320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.111386350046982e-06,
+      "loss": 0.2115,
+      "step": 2330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.102844452037244e-06,
+      "loss": 0.4747,
+      "step": 2340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.094302554027505e-06,
+      "loss": 0.5261,
+      "step": 2350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.085760656017768e-06,
+      "loss": 0.2563,
+      "step": 2360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.07721875800803e-06,
+      "loss": 0.3753,
+      "step": 2370
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.068676859998293e-06,
+      "loss": 0.4599,
+      "step": 2380
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.060134961988555e-06,
+      "loss": 0.5983,
+      "step": 2390
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.051593063978816e-06,
+      "loss": 0.3679,
+      "step": 2400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.04305116596908e-06,
+      "loss": 0.4962,
+      "step": 2410
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.034509267959342e-06,
+      "loss": 0.3391,
+      "step": 2420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.025967369949604e-06,
+      "loss": 0.4035,
+      "step": 2430
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.017425471939865e-06,
+      "loss": 0.3439,
+      "step": 2440
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.008883573930127e-06,
+      "loss": 0.3525,
+      "step": 2450
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.00034167592039e-06,
+      "loss": 0.4039,
+      "step": 2460
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.991799777910653e-06,
+      "loss": 0.4068,
+      "step": 2470
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.983257879900915e-06,
+      "loss": 0.3456,
+      "step": 2480
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.974715981891176e-06,
+      "loss": 0.5975,
+      "step": 2490
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.96617408388144e-06,
+      "loss": 0.3192,
+      "step": 2500
+    },
+    {
+      "epoch": 1.06,
+      "eval_loss": 0.3140031695365906,
+      "eval_runtime": 147.4502,
+      "eval_samples_per_second": 13.706,
+      "eval_steps_per_second": 2.286,
+      "step": 2500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.957632185871701e-06,
+      "loss": 0.3395,
+      "step": 2510
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.949090287861964e-06,
+      "loss": 0.3258,
+      "step": 2520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.940548389852226e-06,
+      "loss": 0.2365,
+      "step": 2530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.932006491842487e-06,
+      "loss": 0.3901,
+      "step": 2540
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.92346459383275e-06,
+      "loss": 0.2342,
+      "step": 2550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.914922695823014e-06,
+      "loss": 0.4549,
+      "step": 2560
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.906380797813275e-06,
+      "loss": 0.1577,
+      "step": 2570
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.897838899803537e-06,
+      "loss": 0.3662,
+      "step": 2580
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.8892970017938e-06,
+      "loss": 0.4641,
+      "step": 2590
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.880755103784061e-06,
+      "loss": 0.4122,
+      "step": 2600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.872213205774325e-06,
+      "loss": 0.4164,
+      "step": 2610
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.863671307764586e-06,
+      "loss": 0.4609,
+      "step": 2620
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.855129409754848e-06,
+      "loss": 0.2847,
+      "step": 2630
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.84658751174511e-06,
+      "loss": 0.5949,
+      "step": 2640
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.838045613735372e-06,
+      "loss": 0.3877,
+      "step": 2650
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.829503715725636e-06,
+      "loss": 0.4248,
+      "step": 2660
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.820961817715897e-06,
+      "loss": 0.4988,
+      "step": 2670
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.812419919706159e-06,
+      "loss": 0.2621,
+      "step": 2680
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.803878021696422e-06,
+      "loss": 0.4065,
+      "step": 2690
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.795336123686685e-06,
+      "loss": 0.3675,
+      "step": 2700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.786794225676946e-06,
+      "loss": 0.2646,
+      "step": 2710
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.778252327667208e-06,
+      "loss": 0.4343,
+      "step": 2720
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.769710429657471e-06,
+      "loss": 0.4538,
+      "step": 2730
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.761168531647733e-06,
+      "loss": 0.522,
+      "step": 2740
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.752626633637996e-06,
+      "loss": 0.2705,
+      "step": 2750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.744084735628257e-06,
+      "loss": 0.4023,
+      "step": 2760
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.735542837618519e-06,
+      "loss": 0.2753,
+      "step": 2770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.727000939608782e-06,
+      "loss": 0.3735,
+      "step": 2780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.718459041599044e-06,
+      "loss": 0.3162,
+      "step": 2790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.709917143589307e-06,
+      "loss": 0.374,
+      "step": 2800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.701375245579568e-06,
+      "loss": 0.2146,
+      "step": 2810
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.69283334756983e-06,
+      "loss": 0.4826,
+      "step": 2820
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.684291449560093e-06,
+      "loss": 0.4694,
+      "step": 2830
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.675749551550356e-06,
+      "loss": 0.242,
+      "step": 2840
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.667207653540618e-06,
+      "loss": 0.4113,
+      "step": 2850
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.65866575553088e-06,
+      "loss": 0.5189,
+      "step": 2860
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.650123857521142e-06,
+      "loss": 0.2997,
+      "step": 2870
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.641581959511404e-06,
+      "loss": 0.3923,
+      "step": 2880
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.633040061501667e-06,
+      "loss": 0.3352,
+      "step": 2890
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.6244981634919294e-06,
+      "loss": 0.5519,
+      "step": 2900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.615956265482191e-06,
+      "loss": 0.3687,
+      "step": 2910
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.607414367472453e-06,
+      "loss": 0.5134,
+      "step": 2920
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.598872469462715e-06,
+      "loss": 0.3172,
+      "step": 2930
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.590330571452977e-06,
+      "loss": 0.5343,
+      "step": 2940
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.58178867344324e-06,
+      "loss": 0.3346,
+      "step": 2950
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.573246775433502e-06,
+      "loss": 0.3423,
+      "step": 2960
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.564704877423764e-06,
+      "loss": 0.2869,
+      "step": 2970
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.5561629794140266e-06,
+      "loss": 0.3877,
+      "step": 2980
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.547621081404288e-06,
+      "loss": 0.3519,
+      "step": 2990
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.539079183394551e-06,
+      "loss": 0.3819,
+      "step": 3000
+    },
+    {
+      "epoch": 1.27,
+      "eval_loss": 0.3138478100299835,
+      "eval_runtime": 147.5308,
+      "eval_samples_per_second": 13.699,
+      "eval_steps_per_second": 2.284,
+      "step": 3000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.530537285384814e-06,
+      "loss": 0.1818,
+      "step": 3010
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.521995387375075e-06,
+      "loss": 0.3404,
+      "step": 3020
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.5134534893653375e-06,
+      "loss": 0.2653,
+      "step": 3030
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.504911591355601e-06,
+      "loss": 0.3424,
+      "step": 3040
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.496369693345862e-06,
+      "loss": 0.3157,
+      "step": 3050
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.4878277953361245e-06,
+      "loss": 0.466,
+      "step": 3060
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.479285897326386e-06,
+      "loss": 0.3224,
+      "step": 3070
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.470743999316648e-06,
+      "loss": 0.2007,
+      "step": 3080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.462202101306912e-06,
+      "loss": 0.2543,
+      "step": 3090
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.453660203297173e-06,
+      "loss": 0.3866,
+      "step": 3100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.4451183052874354e-06,
+      "loss": 0.3445,
+      "step": 3110
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.436576407277698e-06,
+      "loss": 0.2159,
+      "step": 3120
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.428034509267959e-06,
+      "loss": 0.1272,
+      "step": 3130
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.4194926112582225e-06,
+      "loss": 0.4201,
+      "step": 3140
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.410950713248485e-06,
+      "loss": 0.4086,
+      "step": 3150
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.402408815238746e-06,
+      "loss": 0.3111,
+      "step": 3160
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.393866917229009e-06,
+      "loss": 0.3086,
+      "step": 3170
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.385325019219272e-06,
+      "loss": 0.3362,
+      "step": 3180
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.376783121209533e-06,
+      "loss": 0.3311,
+      "step": 3190
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.368241223199796e-06,
+      "loss": 0.427,
+      "step": 3200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.359699325190058e-06,
+      "loss": 0.4626,
+      "step": 3210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.35115742718032e-06,
+      "loss": 0.3225,
+      "step": 3220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.342615529170583e-06,
+      "loss": 0.3641,
+      "step": 3230
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.334073631160844e-06,
+      "loss": 0.2543,
+      "step": 3240
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.325531733151107e-06,
+      "loss": 0.4455,
+      "step": 3250
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.316989835141369e-06,
+      "loss": 0.6165,
+      "step": 3260
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.3084479371316305e-06,
+      "loss": 0.3889,
+      "step": 3270
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.299906039121894e-06,
+      "loss": 0.4556,
+      "step": 3280
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.291364141112156e-06,
+      "loss": 0.2785,
+      "step": 3290
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.282822243102418e-06,
+      "loss": 0.3028,
+      "step": 3300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.27428034509268e-06,
+      "loss": 0.4229,
+      "step": 3310
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.265738447082943e-06,
+      "loss": 0.2957,
+      "step": 3320
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.257196549073205e-06,
+      "loss": 0.3546,
+      "step": 3330
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.248654651063467e-06,
+      "loss": 0.2534,
+      "step": 3340
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.240112753053729e-06,
+      "loss": 0.1781,
+      "step": 3350
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.231570855043991e-06,
+      "loss": 0.3613,
+      "step": 3360
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.223028957034254e-06,
+      "loss": 0.4137,
+      "step": 3370
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.2144870590245156e-06,
+      "loss": 0.2967,
+      "step": 3380
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.205945161014778e-06,
+      "loss": 0.3548,
+      "step": 3390
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.19740326300504e-06,
+      "loss": 0.2941,
+      "step": 3400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.188861364995302e-06,
+      "loss": 0.3029,
+      "step": 3410
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.180319466985565e-06,
+      "loss": 0.2385,
+      "step": 3420
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.171777568975827e-06,
+      "loss": 0.3762,
+      "step": 3430
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.163235670966089e-06,
+      "loss": 0.4625,
+      "step": 3440
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.154693772956351e-06,
+      "loss": 0.5495,
+      "step": 3450
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.146151874946614e-06,
+      "loss": 0.2959,
+      "step": 3460
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.137609976936876e-06,
+      "loss": 0.3331,
+      "step": 3470
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.129068078927138e-06,
+      "loss": 0.3718,
+      "step": 3480
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.1205261809174006e-06,
+      "loss": 0.4167,
+      "step": 3490
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.111984282907662e-06,
+      "loss": 0.3922,
+      "step": 3500
+    },
+    {
+      "epoch": 1.48,
+      "eval_loss": 0.31320881843566895,
+      "eval_runtime": 148.0958,
+      "eval_samples_per_second": 13.647,
+      "eval_steps_per_second": 2.276,
+      "step": 3500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.103442384897925e-06,
+      "loss": 0.3721,
+      "step": 3510
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.094900486888188e-06,
+      "loss": 0.4192,
+      "step": 3520
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.086358588878449e-06,
+      "loss": 0.4409,
+      "step": 3530
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.0778166908687115e-06,
+      "loss": 0.4261,
+      "step": 3540
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.069274792858973e-06,
+      "loss": 0.1757,
+      "step": 3550
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.060732894849236e-06,
+      "loss": 0.3679,
+      "step": 3560
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.0521909968394985e-06,
+      "loss": 0.2991,
+      "step": 3570
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.04364909882976e-06,
+      "loss": 0.2238,
+      "step": 3580
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.035107200820022e-06,
+      "loss": 0.5698,
+      "step": 3590
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.026565302810286e-06,
+      "loss": 0.3213,
+      "step": 3600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.018023404800547e-06,
+      "loss": 0.149,
+      "step": 3610
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.0094815067908095e-06,
+      "loss": 0.4403,
+      "step": 3620
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.000939608781072e-06,
+      "loss": 0.3872,
+      "step": 3630
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.992397710771333e-06,
+      "loss": 0.4537,
+      "step": 3640
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.9838558127615965e-06,
+      "loss": 0.1784,
+      "step": 3650
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.975313914751859e-06,
+      "loss": 0.5157,
+      "step": 3660
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.96677201674212e-06,
+      "loss": 0.5058,
+      "step": 3670
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.958230118732383e-06,
+      "loss": 0.3691,
+      "step": 3680
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.949688220722644e-06,
+      "loss": 0.2682,
+      "step": 3690
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.941146322712907e-06,
+      "loss": 0.3153,
+      "step": 3700
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.93260442470317e-06,
+      "loss": 0.3217,
+      "step": 3710
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.924062526693431e-06,
+      "loss": 0.3731,
+      "step": 3720
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.915520628683694e-06,
+      "loss": 0.4139,
+      "step": 3730
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.906978730673957e-06,
+      "loss": 0.2913,
+      "step": 3740
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.898436832664218e-06,
+      "loss": 0.2702,
+      "step": 3750
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.889894934654481e-06,
+      "loss": 0.2258,
+      "step": 3760
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.881353036644743e-06,
+      "loss": 0.3999,
+      "step": 3770
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.8728111386350045e-06,
+      "loss": 0.56,
+      "step": 3780
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.864269240625268e-06,
+      "loss": 0.3391,
+      "step": 3790
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.85572734261553e-06,
+      "loss": 0.3139,
+      "step": 3800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.847185444605792e-06,
+      "loss": 0.3657,
+      "step": 3810
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.838643546596054e-06,
+      "loss": 0.3958,
+      "step": 3820
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.830101648586317e-06,
+      "loss": 0.3364,
+      "step": 3830
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 6.821559750576579e-06,
+      "loss": 0.4123,
+      "step": 3840
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.813017852566841e-06,
+      "loss": 0.3019,
+      "step": 3850
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 6.8044759545571025e-06,
+      "loss": 0.4308,
+      "step": 3860
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.795934056547365e-06,
+      "loss": 0.4005,
+      "step": 3870
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.787392158537628e-06,
+      "loss": 0.5024,
+      "step": 3880
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.7788502605278896e-06,
+      "loss": 0.3102,
+      "step": 3890
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.770308362518152e-06,
+      "loss": 0.514,
+      "step": 3900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.761766464508414e-06,
+      "loss": 0.4315,
+      "step": 3910
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.753224566498676e-06,
+      "loss": 0.3066,
+      "step": 3920
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.744682668488939e-06,
+      "loss": 0.4766,
+      "step": 3930
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.736140770479201e-06,
+      "loss": 0.482,
+      "step": 3940
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.727598872469463e-06,
+      "loss": 0.4122,
+      "step": 3950
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.719056974459725e-06,
+      "loss": 0.4521,
+      "step": 3960
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.710515076449988e-06,
+      "loss": 0.3859,
+      "step": 3970
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.70197317844025e-06,
+      "loss": 0.3144,
+      "step": 3980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.693431280430512e-06,
+      "loss": 0.1476,
+      "step": 3990
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.684889382420774e-06,
+      "loss": 0.3443,
+      "step": 4000
+    },
+    {
+      "epoch": 1.69,
+      "eval_loss": 0.31326767802238464,
+      "eval_runtime": 148.1653,
+      "eval_samples_per_second": 13.64,
+      "eval_steps_per_second": 2.274,
+      "step": 4000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.676347484411037e-06,
+      "loss": 0.6021,
+      "step": 4010
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.667805586401299e-06,
+      "loss": 0.2874,
+      "step": 4020
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.659263688391561e-06,
+      "loss": 0.2902,
+      "step": 4030
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.650721790381823e-06,
+      "loss": 0.3469,
+      "step": 4040
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.642179892372086e-06,
+      "loss": 0.2126,
+      "step": 4050
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.633637994362348e-06,
+      "loss": 0.6042,
+      "step": 4060
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 6.62509609635261e-06,
+      "loss": 0.3642,
+      "step": 4070
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.6165541983428726e-06,
+      "loss": 0.3103,
+      "step": 4080
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.608012300333134e-06,
+      "loss": 0.4791,
+      "step": 4090
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 6.599470402323397e-06,
+      "loss": 0.3973,
+      "step": 4100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.59092850431366e-06,
+      "loss": 0.2553,
+      "step": 4110
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.582386606303921e-06,
+      "loss": 0.4041,
+      "step": 4120
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.5738447082941835e-06,
+      "loss": 0.3603,
+      "step": 4130
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.565302810284447e-06,
+      "loss": 0.4866,
+      "step": 4140
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 6.556760912274708e-06,
+      "loss": 0.4017,
+      "step": 4150
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.5482190142649705e-06,
+      "loss": 0.3596,
+      "step": 4160
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 6.539677116255232e-06,
+      "loss": 0.2408,
+      "step": 4170
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.531135218245494e-06,
+      "loss": 0.1687,
+      "step": 4180
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.522593320235758e-06,
+      "loss": 0.3384,
+      "step": 4190
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.514051422226019e-06,
+      "loss": 0.3191,
+      "step": 4200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.5055095242162814e-06,
+      "loss": 0.2992,
+      "step": 4210
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.496967626206544e-06,
+      "loss": 0.3559,
+      "step": 4220
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.488425728196805e-06,
+      "loss": 0.287,
+      "step": 4230
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.4798838301870685e-06,
+      "loss": 0.4084,
+      "step": 4240
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.471341932177331e-06,
+      "loss": 0.4548,
+      "step": 4250
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.462800034167592e-06,
+      "loss": 0.4089,
+      "step": 4260
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.454258136157855e-06,
+      "loss": 0.525,
+      "step": 4270
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.445716238148118e-06,
+      "loss": 0.395,
+      "step": 4280
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.437174340138379e-06,
+      "loss": 0.4474,
+      "step": 4290
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.428632442128642e-06,
+      "loss": 0.1524,
+      "step": 4300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.420090544118903e-06,
+      "loss": 0.3649,
+      "step": 4310
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.411548646109166e-06,
+      "loss": 0.2933,
+      "step": 4320
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.403006748099429e-06,
+      "loss": 0.4789,
+      "step": 4330
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.39446485008969e-06,
+      "loss": 0.3489,
+      "step": 4340
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.385922952079953e-06,
+      "loss": 0.2916,
+      "step": 4350
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.377381054070215e-06,
+      "loss": 0.1772,
+      "step": 4360
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.3688391560604765e-06,
+      "loss": 0.3849,
+      "step": 4370
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.36029725805074e-06,
+      "loss": 0.4892,
+      "step": 4380
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.351755360041002e-06,
+      "loss": 0.4328,
+      "step": 4390
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.343213462031264e-06,
+      "loss": 0.3038,
+      "step": 4400
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.334671564021526e-06,
+      "loss": 0.2119,
+      "step": 4410
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.326129666011789e-06,
+      "loss": 0.3328,
+      "step": 4420
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.317587768002051e-06,
+      "loss": 0.3124,
+      "step": 4430
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.309045869992313e-06,
+      "loss": 0.2876,
+      "step": 4440
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.300503971982575e-06,
+      "loss": 0.2042,
+      "step": 4450
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.291962073972837e-06,
+      "loss": 0.2934,
+      "step": 4460
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.2834201759631e-06,
+      "loss": 0.4138,
+      "step": 4470
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.2748782779533615e-06,
+      "loss": 0.5214,
+      "step": 4480
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.266336379943624e-06,
+      "loss": 0.3396,
+      "step": 4490
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.257794481933886e-06,
+      "loss": 0.3827,
+      "step": 4500
+    },
+    {
+      "epoch": 1.9,
+      "eval_loss": 0.3127913177013397,
+      "eval_runtime": 148.3308,
+      "eval_samples_per_second": 13.625,
+      "eval_steps_per_second": 2.272,
+      "step": 4500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.249252583924148e-06,
+      "loss": 0.3716,
+      "step": 4510
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.240710685914411e-06,
+      "loss": 0.2787,
+      "step": 4520
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.232168787904673e-06,
+      "loss": 0.3925,
+      "step": 4530
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.223626889894935e-06,
+      "loss": 0.3131,
+      "step": 4540
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.215084991885197e-06,
+      "loss": 0.2891,
+      "step": 4550
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.20654309387546e-06,
+      "loss": 0.3638,
+      "step": 4560
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.198001195865722e-06,
+      "loss": 0.3896,
+      "step": 4570
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.189459297855984e-06,
+      "loss": 0.5276,
+      "step": 4580
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.1809173998462466e-06,
+      "loss": 0.3169,
+      "step": 4590
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.172375501836508e-06,
+      "loss": 0.336,
+      "step": 4600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.163833603826771e-06,
+      "loss": 0.3627,
+      "step": 4610
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.155291705817033e-06,
+      "loss": 0.3236,
+      "step": 4620
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.146749807807295e-06,
+      "loss": 0.3226,
+      "step": 4630
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.1382079097975575e-06,
+      "loss": 0.3447,
+      "step": 4640
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.129666011787819e-06,
+      "loss": 0.419,
+      "step": 4650
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.121124113778082e-06,
+      "loss": 0.446,
+      "step": 4660
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.1125822157683445e-06,
+      "loss": 0.3607,
+      "step": 4670
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.104040317758606e-06,
+      "loss": 0.268,
+      "step": 4680
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.095498419748868e-06,
+      "loss": 0.155,
+      "step": 4690
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.086956521739132e-06,
+      "loss": 0.4208,
+      "step": 4700
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.078414623729393e-06,
+      "loss": 0.3829,
+      "step": 4710
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.0698727257196555e-06,
+      "loss": 0.4763,
+      "step": 4720
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.061330827709918e-06,
+      "loss": 0.3541,
+      "step": 4730
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.052788929700179e-06,
+      "loss": 0.395,
+      "step": 4740
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.0442470316904425e-06,
+      "loss": 0.4241,
+      "step": 4750
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.035705133680705e-06,
+      "loss": 0.2576,
+      "step": 4760
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.027163235670966e-06,
+      "loss": 0.5861,
+      "step": 4770
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.018621337661229e-06,
+      "loss": 0.2823,
+      "step": 4780
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.01007943965149e-06,
+      "loss": 0.3957,
+      "step": 4790
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.001537541641753e-06,
+      "loss": 0.2409,
+      "step": 4800
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.992995643632016e-06,
+      "loss": 0.4057,
+      "step": 4810
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.984453745622277e-06,
+      "loss": 0.3656,
+      "step": 4820
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.97591184761254e-06,
+      "loss": 0.2622,
+      "step": 4830
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.967369949602803e-06,
+      "loss": 0.4196,
+      "step": 4840
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.958828051593064e-06,
+      "loss": 0.3719,
+      "step": 4850
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.950286153583327e-06,
+      "loss": 0.4678,
+      "step": 4860
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.941744255573589e-06,
+      "loss": 0.372,
+      "step": 4870
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.9332023575638505e-06,
+      "loss": 0.3476,
+      "step": 4880
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.924660459554114e-06,
+      "loss": 0.3638,
+      "step": 4890
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.916118561544376e-06,
+      "loss": 0.3632,
+      "step": 4900
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.907576663534638e-06,
+      "loss": 0.2015,
+      "step": 4910
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.8990347655249e-06,
+      "loss": 0.3838,
+      "step": 4920
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.8904928675151615e-06,
+      "loss": 0.2182,
+      "step": 4930
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.881950969505425e-06,
+      "loss": 0.3247,
+      "step": 4940
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.873409071495687e-06,
+      "loss": 0.3435,
+      "step": 4950
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.8648671734859485e-06,
+      "loss": 0.2764,
+      "step": 4960
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.856325275476211e-06,
+      "loss": 0.4463,
+      "step": 4970
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.847783377466474e-06,
+      "loss": 0.3739,
+      "step": 4980
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.8392414794567356e-06,
+      "loss": 0.3779,
+      "step": 4990
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.830699581446998e-06,
+      "loss": 0.4169,
+      "step": 5000
+    },
+    {
+      "epoch": 2.11,
+      "eval_loss": 0.31448912620544434,
+      "eval_runtime": 148.8038,
+      "eval_samples_per_second": 13.582,
+      "eval_steps_per_second": 2.265,
+      "step": 5000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.82215768343726e-06,
+      "loss": 0.4081,
+      "step": 5010
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.813615785427522e-06,
+      "loss": 0.3377,
+      "step": 5020
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.805073887417785e-06,
+      "loss": 0.38,
+      "step": 5030
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.796531989408047e-06,
+      "loss": 0.3641,
+      "step": 5040
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.787990091398309e-06,
+      "loss": 0.435,
+      "step": 5050
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.779448193388571e-06,
+      "loss": 0.3914,
+      "step": 5060
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.770906295378834e-06,
+      "loss": 0.1876,
+      "step": 5070
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.762364397369096e-06,
+      "loss": 0.2638,
+      "step": 5080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.753822499359358e-06,
+      "loss": 0.4779,
+      "step": 5090
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.74528060134962e-06,
+      "loss": 0.2917,
+      "step": 5100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.736738703339882e-06,
+      "loss": 0.324,
+      "step": 5110
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.728196805330145e-06,
+      "loss": 0.2366,
+      "step": 5120
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.719654907320407e-06,
+      "loss": 0.4861,
+      "step": 5130
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.711113009310669e-06,
+      "loss": 0.4288,
+      "step": 5140
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.7025711113009315e-06,
+      "loss": 0.2315,
+      "step": 5150
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.694029213291193e-06,
+      "loss": 0.2426,
+      "step": 5160
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.685487315281456e-06,
+      "loss": 0.4719,
+      "step": 5170
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.6769454172717186e-06,
+      "loss": 0.4703,
+      "step": 5180
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.66840351926198e-06,
+      "loss": 0.3042,
+      "step": 5190
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.659861621252242e-06,
+      "loss": 0.2986,
+      "step": 5200
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.651319723242506e-06,
+      "loss": 0.3161,
+      "step": 5210
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.642777825232767e-06,
+      "loss": 0.4217,
+      "step": 5220
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.6342359272230295e-06,
+      "loss": 0.2574,
+      "step": 5230
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.625694029213291e-06,
+      "loss": 0.3925,
+      "step": 5240
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.617152131203553e-06,
+      "loss": 0.277,
+      "step": 5250
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.6086102331938165e-06,
+      "loss": 0.448,
+      "step": 5260
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.600068335184078e-06,
+      "loss": 0.4412,
+      "step": 5270
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.59152643717434e-06,
+      "loss": 0.2734,
+      "step": 5280
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.582984539164603e-06,
+      "loss": 0.304,
+      "step": 5290
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.574442641154865e-06,
+      "loss": 0.2944,
+      "step": 5300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.5659007431451274e-06,
+      "loss": 0.3414,
+      "step": 5310
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.55735884513539e-06,
+      "loss": 0.2897,
+      "step": 5320
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.548816947125651e-06,
+      "loss": 0.5415,
+      "step": 5330
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.5402750491159145e-06,
+      "loss": 0.284,
+      "step": 5340
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.531733151106177e-06,
+      "loss": 0.3498,
+      "step": 5350
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.523191253096438e-06,
+      "loss": 0.4256,
+      "step": 5360
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.514649355086701e-06,
+      "loss": 0.3603,
+      "step": 5370
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.506107457076964e-06,
+      "loss": 0.2654,
+      "step": 5380
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.497565559067225e-06,
+      "loss": 0.1896,
+      "step": 5390
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.489023661057488e-06,
+      "loss": 0.3347,
+      "step": 5400
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.480481763047749e-06,
+      "loss": 0.3013,
+      "step": 5410
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.471939865038012e-06,
+      "loss": 0.4533,
+      "step": 5420
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.463397967028275e-06,
+      "loss": 0.2693,
+      "step": 5430
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.454856069018536e-06,
+      "loss": 0.4373,
+      "step": 5440
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.446314171008799e-06,
+      "loss": 0.3536,
+      "step": 5450
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.437772272999061e-06,
+      "loss": 0.2906,
+      "step": 5460
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.4292303749893225e-06,
+      "loss": 0.3749,
+      "step": 5470
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.420688476979586e-06,
+      "loss": 0.3229,
+      "step": 5480
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.412146578969848e-06,
+      "loss": 0.382,
+      "step": 5490
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.40360468096011e-06,
+      "loss": 0.2356,
+      "step": 5500
+    },
+    {
+      "epoch": 2.33,
+      "eval_loss": 0.31541574001312256,
+      "eval_runtime": 148.0605,
+      "eval_samples_per_second": 13.65,
+      "eval_steps_per_second": 2.276,
+      "step": 5500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.395062782950372e-06,
+      "loss": 0.3956,
+      "step": 5510
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.386520884940635e-06,
+      "loss": 0.3128,
+      "step": 5520
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.377978986930897e-06,
+      "loss": 0.3933,
+      "step": 5530
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.369437088921159e-06,
+      "loss": 0.5467,
+      "step": 5540
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.3608951909114205e-06,
+      "loss": 0.132,
+      "step": 5550
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.352353292901683e-06,
+      "loss": 0.4206,
+      "step": 5560
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.343811394891946e-06,
+      "loss": 0.2647,
+      "step": 5570
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.3352694968822075e-06,
+      "loss": 0.3597,
+      "step": 5580
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.32672759887247e-06,
+      "loss": 0.2996,
+      "step": 5590
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.318185700862732e-06,
+      "loss": 0.3868,
+      "step": 5600
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.309643802852994e-06,
+      "loss": 0.2282,
+      "step": 5610
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.301101904843257e-06,
+      "loss": 0.1808,
+      "step": 5620
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.292560006833519e-06,
+      "loss": 0.2549,
+      "step": 5630
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.284018108823781e-06,
+      "loss": 0.1823,
+      "step": 5640
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.275476210814043e-06,
+      "loss": 0.3847,
+      "step": 5650
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.266934312804306e-06,
+      "loss": 0.4476,
+      "step": 5660
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.258392414794568e-06,
+      "loss": 0.1905,
+      "step": 5670
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.24985051678483e-06,
+      "loss": 0.3908,
+      "step": 5680
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.2413086187750926e-06,
+      "loss": 0.3519,
+      "step": 5690
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.232766720765354e-06,
+      "loss": 0.3762,
+      "step": 5700
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.224224822755617e-06,
+      "loss": 0.3823,
+      "step": 5710
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.215682924745879e-06,
+      "loss": 0.2805,
+      "step": 5720
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.207141026736141e-06,
+      "loss": 0.3178,
+      "step": 5730
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.1985991287264035e-06,
+      "loss": 0.4441,
+      "step": 5740
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.190057230716665e-06,
+      "loss": 0.4681,
+      "step": 5750
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.181515332706928e-06,
+      "loss": 0.2889,
+      "step": 5760
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.1729734346971905e-06,
+      "loss": 0.2544,
+      "step": 5770
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.164431536687452e-06,
+      "loss": 0.4712,
+      "step": 5780
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.155889638677714e-06,
+      "loss": 0.4941,
+      "step": 5790
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.147347740667978e-06,
+      "loss": 0.1824,
+      "step": 5800
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.138805842658239e-06,
+      "loss": 0.3744,
+      "step": 5810
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.1302639446485014e-06,
+      "loss": 0.3031,
+      "step": 5820
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.121722046638764e-06,
+      "loss": 0.3723,
+      "step": 5830
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.113180148629025e-06,
+      "loss": 0.2414,
+      "step": 5840
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.1046382506192885e-06,
+      "loss": 0.4484,
+      "step": 5850
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.09609635260955e-06,
+      "loss": 0.3328,
+      "step": 5860
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.087554454599812e-06,
+      "loss": 0.3509,
+      "step": 5870
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.079012556590075e-06,
+      "loss": 0.3446,
+      "step": 5880
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.070470658580336e-06,
+      "loss": 0.3094,
+      "step": 5890
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.061928760570599e-06,
+      "loss": 0.3308,
+      "step": 5900
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.053386862560862e-06,
+      "loss": 0.3024,
+      "step": 5910
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.044844964551123e-06,
+      "loss": 0.3678,
+      "step": 5920
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.036303066541386e-06,
+      "loss": 0.3632,
+      "step": 5930
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.027761168531649e-06,
+      "loss": 0.2619,
+      "step": 5940
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.01921927052191e-06,
+      "loss": 0.3098,
+      "step": 5950
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.010677372512173e-06,
+      "loss": 0.4114,
+      "step": 5960
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.002135474502435e-06,
+      "loss": 0.4709,
+      "step": 5970
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.9935935764926965e-06,
+      "loss": 0.5214,
+      "step": 5980
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.98505167848296e-06,
+      "loss": 0.2079,
+      "step": 5990
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.976509780473221e-06,
+      "loss": 0.4347,
+      "step": 6000
+    },
+    {
+      "epoch": 2.54,
+      "eval_loss": 0.3158428966999054,
+      "eval_runtime": 148.2261,
+      "eval_samples_per_second": 13.635,
+      "eval_steps_per_second": 2.274,
+      "step": 6000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.967967882463484e-06,
+      "loss": 0.3363,
+      "step": 6010
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.959425984453746e-06,
+      "loss": 0.4161,
+      "step": 6020
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.950884086444008e-06,
+      "loss": 0.3139,
+      "step": 6030
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.942342188434271e-06,
+      "loss": 0.1947,
+      "step": 6040
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.933800290424533e-06,
+      "loss": 0.3568,
+      "step": 6050
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.925258392414795e-06,
+      "loss": 0.4489,
+      "step": 6060
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.916716494405057e-06,
+      "loss": 0.4417,
+      "step": 6070
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.908174596395319e-06,
+      "loss": 0.336,
+      "step": 6080
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.8996326983855816e-06,
+      "loss": 0.5577,
+      "step": 6090
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.891090800375844e-06,
+      "loss": 0.4839,
+      "step": 6100
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.882548902366106e-06,
+      "loss": 0.3746,
+      "step": 6110
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.874007004356369e-06,
+      "loss": 0.3279,
+      "step": 6120
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.865465106346631e-06,
+      "loss": 0.418,
+      "step": 6130
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.8569232083368925e-06,
+      "loss": 0.6271,
+      "step": 6140
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.848381310327155e-06,
+      "loss": 0.3411,
+      "step": 6150
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.839839412317417e-06,
+      "loss": 0.319,
+      "step": 6160
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.8312975143076795e-06,
+      "loss": 0.2534,
+      "step": 6170
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.822755616297942e-06,
+      "loss": 0.3032,
+      "step": 6180
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.814213718288204e-06,
+      "loss": 0.2322,
+      "step": 6190
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.805671820278467e-06,
+      "loss": 0.2349,
+      "step": 6200
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.797129922268728e-06,
+      "loss": 0.3819,
+      "step": 6210
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.7885880242589904e-06,
+      "loss": 0.4986,
+      "step": 6220
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.780046126249253e-06,
+      "loss": 0.2811,
+      "step": 6230
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.771504228239515e-06,
+      "loss": 0.3729,
+      "step": 6240
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.7629623302297775e-06,
+      "loss": 0.2989,
+      "step": 6250
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.75442043222004e-06,
+      "loss": 0.265,
+      "step": 6260
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.745878534210302e-06,
+      "loss": 0.1843,
+      "step": 6270
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.737336636200564e-06,
+      "loss": 0.4455,
+      "step": 6280
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.728794738190826e-06,
+      "loss": 0.2162,
+      "step": 6290
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.720252840181088e-06,
+      "loss": 0.3711,
+      "step": 6300
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.711710942171351e-06,
+      "loss": 0.3361,
+      "step": 6310
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.703169044161613e-06,
+      "loss": 0.3901,
+      "step": 6320
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.6946271461518755e-06,
+      "loss": 0.3042,
+      "step": 6330
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.686085248142138e-06,
+      "loss": 0.3939,
+      "step": 6340
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.677543350132399e-06,
+      "loss": 0.2726,
+      "step": 6350
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6690014521226625e-06,
+      "loss": 0.3253,
+      "step": 6360
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.660459554112924e-06,
+      "loss": 0.3218,
+      "step": 6370
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.651917656103186e-06,
+      "loss": 0.4015,
+      "step": 6380
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.643375758093449e-06,
+      "loss": 0.2483,
+      "step": 6390
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.634833860083711e-06,
+      "loss": 0.2528,
+      "step": 6400
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6262919620739734e-06,
+      "loss": 0.2913,
+      "step": 6410
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.617750064064235e-06,
+      "loss": 0.4788,
+      "step": 6420
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.609208166054498e-06,
+      "loss": 0.2941,
+      "step": 6430
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.60066626804476e-06,
+      "loss": 0.4911,
+      "step": 6440
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.592124370035022e-06,
+      "loss": 0.3398,
+      "step": 6450
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.583582472025284e-06,
+      "loss": 0.35,
+      "step": 6460
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.575040574015547e-06,
+      "loss": 0.3149,
+      "step": 6470
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.566498676005809e-06,
+      "loss": 0.3246,
+      "step": 6480
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5579567779960706e-06,
+      "loss": 0.3306,
+      "step": 6490
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.549414879986334e-06,
+      "loss": 0.2695,
+      "step": 6500
+    },
+    {
+      "epoch": 2.75,
+      "eval_loss": 0.3164987862110138,
+      "eval_runtime": 147.7906,
+      "eval_samples_per_second": 13.675,
+      "eval_steps_per_second": 2.28,
+      "step": 6500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.540872981976595e-06,
+      "loss": 0.2815,
+      "step": 6510
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.532331083966858e-06,
+      "loss": 0.4566,
+      "step": 6520
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.52378918595712e-06,
+      "loss": 0.3873,
+      "step": 6530
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.515247287947382e-06,
+      "loss": 0.2694,
+      "step": 6540
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.506705389937645e-06,
+      "loss": 0.2986,
+      "step": 6550
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.498163491927906e-06,
+      "loss": 0.3569,
+      "step": 6560
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.489621593918169e-06,
+      "loss": 0.3491,
+      "step": 6570
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.481079695908431e-06,
+      "loss": 0.4444,
+      "step": 6580
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.472537797898693e-06,
+      "loss": 0.2079,
+      "step": 6590
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.4639958998889556e-06,
+      "loss": 0.3359,
+      "step": 6600
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.455454001879218e-06,
+      "loss": 0.1888,
+      "step": 6610
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.44691210386948e-06,
+      "loss": 0.3059,
+      "step": 6620
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.438370205859743e-06,
+      "loss": 0.2553,
+      "step": 6630
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.429828307850005e-06,
+      "loss": 0.3415,
+      "step": 6640
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.421286409840267e-06,
+      "loss": 0.3051,
+      "step": 6650
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.412744511830529e-06,
+      "loss": 0.2773,
+      "step": 6660
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.404202613820792e-06,
+      "loss": 0.3695,
+      "step": 6670
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.3956607158110535e-06,
+      "loss": 0.3404,
+      "step": 6680
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.387118817801316e-06,
+      "loss": 0.1903,
+      "step": 6690
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.378576919791578e-06,
+      "loss": 0.2607,
+      "step": 6700
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.370035021781841e-06,
+      "loss": 0.2597,
+      "step": 6710
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.361493123772103e-06,
+      "loss": 0.2355,
+      "step": 6720
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.3529512257623645e-06,
+      "loss": 0.1798,
+      "step": 6730
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.344409327752628e-06,
+      "loss": 0.5509,
+      "step": 6740
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.335867429742889e-06,
+      "loss": 0.3129,
+      "step": 6750
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.3273255317331515e-06,
+      "loss": 0.2288,
+      "step": 6760
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.318783633723414e-06,
+      "loss": 0.3235,
+      "step": 6770
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.310241735713676e-06,
+      "loss": 0.2354,
+      "step": 6780
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.3016998377039386e-06,
+      "loss": 0.2814,
+      "step": 6790
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.2931579396942e-06,
+      "loss": 0.3052,
+      "step": 6800
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.284616041684463e-06,
+      "loss": 0.271,
+      "step": 6810
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.276074143674725e-06,
+      "loss": 0.5061,
+      "step": 6820
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.267532245664987e-06,
+      "loss": 0.2946,
+      "step": 6830
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.2589903476552495e-06,
+      "loss": 0.3785,
+      "step": 6840
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.250448449645512e-06,
+      "loss": 0.258,
+      "step": 6850
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.241906551635774e-06,
+      "loss": 0.3433,
+      "step": 6860
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.233364653626036e-06,
+      "loss": 0.3728,
+      "step": 6870
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.224822755616299e-06,
+      "loss": 0.3265,
+      "step": 6880
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.21628085760656e-06,
+      "loss": 0.3782,
+      "step": 6890
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.207738959596823e-06,
+      "loss": 0.4276,
+      "step": 6900
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.199197061587085e-06,
+      "loss": 0.3275,
+      "step": 6910
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.1906551635773474e-06,
+      "loss": 0.4285,
+      "step": 6920
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.18211326556761e-06,
+      "loss": 0.2002,
+      "step": 6930
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.173571367557871e-06,
+      "loss": 0.28,
+      "step": 6940
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.1650294695481345e-06,
+      "loss": 0.4632,
+      "step": 6950
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.156487571538396e-06,
+      "loss": 0.5316,
+      "step": 6960
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.147945673528658e-06,
+      "loss": 0.304,
+      "step": 6970
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.139403775518921e-06,
+      "loss": 0.2603,
+      "step": 6980
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.130861877509183e-06,
+      "loss": 0.1586,
+      "step": 6990
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.122319979499445e-06,
+      "loss": 0.425,
+      "step": 7000
+    },
+    {
+      "epoch": 2.96,
+      "eval_loss": 0.31690141558647156,
+      "eval_runtime": 147.8924,
+      "eval_samples_per_second": 13.665,
+      "eval_steps_per_second": 2.279,
+      "step": 7000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.113778081489707e-06,
+      "loss": 0.4468,
+      "step": 7010
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.10523618347997e-06,
+      "loss": 0.2205,
+      "step": 7020
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.096694285470232e-06,
+      "loss": 0.2492,
+      "step": 7030
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.088152387460494e-06,
+      "loss": 0.3075,
+      "step": 7040
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.079610489450756e-06,
+      "loss": 0.382,
+      "step": 7050
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.071068591441019e-06,
+      "loss": 0.3731,
+      "step": 7060
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.062526693431281e-06,
+      "loss": 0.3225,
+      "step": 7070
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.0539847954215425e-06,
+      "loss": 0.4533,
+      "step": 7080
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.045442897411806e-06,
+      "loss": 0.2833,
+      "step": 7090
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.036900999402067e-06,
+      "loss": 0.3202,
+      "step": 7100
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.02835910139233e-06,
+      "loss": 0.2674,
+      "step": 7110
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.019817203382592e-06,
+      "loss": 0.2685,
+      "step": 7120
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.011275305372854e-06,
+      "loss": 0.4526,
+      "step": 7130
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.002733407363117e-06,
+      "loss": 0.1943,
+      "step": 7140
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.994191509353378e-06,
+      "loss": 0.1707,
+      "step": 7150
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.985649611343641e-06,
+      "loss": 0.457,
+      "step": 7160
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.977107713333903e-06,
+      "loss": 0.3018,
+      "step": 7170
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.968565815324165e-06,
+      "loss": 0.2961,
+      "step": 7180
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.9600239173144276e-06,
+      "loss": 0.451,
+      "step": 7190
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.95148201930469e-06,
+      "loss": 0.293,
+      "step": 7200
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.942940121294952e-06,
+      "loss": 0.3328,
+      "step": 7210
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.934398223285214e-06,
+      "loss": 0.2717,
+      "step": 7220
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.925856325275477e-06,
+      "loss": 0.296,
+      "step": 7230
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.9173144272657385e-06,
+      "loss": 0.2992,
+      "step": 7240
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.908772529256001e-06,
+      "loss": 0.1454,
+      "step": 7250
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.900230631246263e-06,
+      "loss": 0.2253,
+      "step": 7260
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.8916887332365255e-06,
+      "loss": 0.2319,
+      "step": 7270
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.883146835226788e-06,
+      "loss": 0.3923,
+      "step": 7280
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.87460493721705e-06,
+      "loss": 0.3627,
+      "step": 7290
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.866063039207313e-06,
+      "loss": 0.3542,
+      "step": 7300
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.857521141197574e-06,
+      "loss": 0.1871,
+      "step": 7310
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.8489792431878364e-06,
+      "loss": 0.2915,
+      "step": 7320
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.840437345178099e-06,
+      "loss": 0.2174,
+      "step": 7330
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.831895447168361e-06,
+      "loss": 0.2377,
+      "step": 7340
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.8233535491586235e-06,
+      "loss": 0.1499,
+      "step": 7350
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.814811651148886e-06,
+      "loss": 0.3904,
+      "step": 7360
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.8062697531391478e-06,
+      "loss": 0.2644,
+      "step": 7370
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.79772785512941e-06,
+      "loss": 0.3754,
+      "step": 7380
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.789185957119672e-06,
+      "loss": 0.3259,
+      "step": 7390
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.780644059109935e-06,
+      "loss": 0.2098,
+      "step": 7400
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.7721021611001968e-06,
+      "loss": 0.3464,
+      "step": 7410
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.7635602630904587e-06,
+      "loss": 0.3925,
+      "step": 7420
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.7550183650807215e-06,
+      "loss": 0.3838,
+      "step": 7430
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.7464764670709834e-06,
+      "loss": 0.3914,
+      "step": 7440
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.7379345690612457e-06,
+      "loss": 0.3022,
+      "step": 7450
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.7293926710515077e-06,
+      "loss": 0.1867,
+      "step": 7460
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.7208507730417704e-06,
+      "loss": 0.2706,
+      "step": 7470
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.7123088750320324e-06,
+      "loss": 0.2583,
+      "step": 7480
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.7037669770222943e-06,
+      "loss": 0.2944,
+      "step": 7490
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.695225079012557e-06,
+      "loss": 0.3686,
+      "step": 7500
+    },
+    {
+      "epoch": 3.17,
+      "eval_loss": 0.3217047452926636,
+      "eval_runtime": 154.1964,
+      "eval_samples_per_second": 13.107,
+      "eval_steps_per_second": 2.186,
+      "step": 7500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.686683181002819e-06,
+      "loss": 0.2531,
+      "step": 7510
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.6781412829930814e-06,
+      "loss": 0.2616,
+      "step": 7520
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.6695993849833433e-06,
+      "loss": 0.2927,
+      "step": 7530
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.661057486973606e-06,
+      "loss": 0.2863,
+      "step": 7540
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.652515588963868e-06,
+      "loss": 0.4489,
+      "step": 7550
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.64397369095413e-06,
+      "loss": 0.3503,
+      "step": 7560
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.6354317929443927e-06,
+      "loss": 0.2711,
+      "step": 7570
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.6268898949346546e-06,
+      "loss": 0.3376,
+      "step": 7580
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.618347996924917e-06,
+      "loss": 0.283,
+      "step": 7590
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.6098060989151793e-06,
+      "loss": 0.2893,
+      "step": 7600
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.6012642009054417e-06,
+      "loss": 0.2331,
+      "step": 7610
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.5927223028957036e-06,
+      "loss": 0.5075,
+      "step": 7620
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.584180404885966e-06,
+      "loss": 0.2439,
+      "step": 7630
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.5756385068762283e-06,
+      "loss": 0.3249,
+      "step": 7640
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.5670966088664907e-06,
+      "loss": 0.3276,
+      "step": 7650
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.5585547108567526e-06,
+      "loss": 0.3269,
+      "step": 7660
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.5500128128470154e-06,
+      "loss": 0.4029,
+      "step": 7670
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.5414709148372773e-06,
+      "loss": 0.3574,
+      "step": 7680
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.5329290168275392e-06,
+      "loss": 0.1384,
+      "step": 7690
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.5243871188178016e-06,
+      "loss": 0.4162,
+      "step": 7700
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.515845220808064e-06,
+      "loss": 0.4534,
+      "step": 7710
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.5073033227983263e-06,
+      "loss": 0.269,
+      "step": 7720
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.498761424788588e-06,
+      "loss": 0.3813,
+      "step": 7730
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.490219526778851e-06,
+      "loss": 0.421,
+      "step": 7740
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.481677628769113e-06,
+      "loss": 0.2597,
+      "step": 7750
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.473135730759375e-06,
+      "loss": 0.3672,
+      "step": 7760
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.464593832749637e-06,
+      "loss": 0.3144,
+      "step": 7770
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.4560519347398995e-06,
+      "loss": 0.3438,
+      "step": 7780
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.447510036730162e-06,
+      "loss": 0.3665,
+      "step": 7790
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.438968138720424e-06,
+      "loss": 0.1775,
+      "step": 7800
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.4304262407106866e-06,
+      "loss": 0.4987,
+      "step": 7810
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.4218843427009485e-06,
+      "loss": 0.282,
+      "step": 7820
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.4133424446912105e-06,
+      "loss": 0.2632,
+      "step": 7830
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.404800546681473e-06,
+      "loss": 0.2824,
+      "step": 7840
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.396258648671735e-06,
+      "loss": 0.3774,
+      "step": 7850
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3877167506619975e-06,
+      "loss": 0.2894,
+      "step": 7860
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3791748526522594e-06,
+      "loss": 0.145,
+      "step": 7870
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.370632954642522e-06,
+      "loss": 0.3516,
+      "step": 7880
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.362091056632784e-06,
+      "loss": 0.3059,
+      "step": 7890
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.353549158623046e-06,
+      "loss": 0.3699,
+      "step": 7900
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.345007260613309e-06,
+      "loss": 0.4026,
+      "step": 7910
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3364653626035708e-06,
+      "loss": 0.3167,
+      "step": 7920
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.327923464593833e-06,
+      "loss": 0.3744,
+      "step": 7930
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.319381566584095e-06,
+      "loss": 0.3112,
+      "step": 7940
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.310839668574358e-06,
+      "loss": 0.4122,
+      "step": 7950
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3022977705646198e-06,
+      "loss": 0.2248,
+      "step": 7960
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2937558725548817e-06,
+      "loss": 0.421,
+      "step": 7970
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.2852139745451445e-06,
+      "loss": 0.3756,
+      "step": 7980
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.2766720765354064e-06,
+      "loss": 0.3177,
+      "step": 7990
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.2681301785256687e-06,
+      "loss": 0.258,
+      "step": 8000
+    },
+    {
+      "epoch": 3.38,
+      "eval_loss": 0.3234722316265106,
+      "eval_runtime": 147.8131,
+      "eval_samples_per_second": 13.673,
+      "eval_steps_per_second": 2.28,
+      "step": 8000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.2595882805159307e-06,
+      "loss": 0.4607,
+      "step": 8010
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.2510463825061934e-06,
+      "loss": 0.3055,
+      "step": 8020
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2425044844964554e-06,
+      "loss": 0.2157,
+      "step": 8030
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2339625864867173e-06,
+      "loss": 0.4536,
+      "step": 8040
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.22542068847698e-06,
+      "loss": 0.4621,
+      "step": 8050
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.216878790467242e-06,
+      "loss": 0.3938,
+      "step": 8060
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2083368924575044e-06,
+      "loss": 0.3495,
+      "step": 8070
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.1997949944477663e-06,
+      "loss": 0.2466,
+      "step": 8080
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.191253096438029e-06,
+      "loss": 0.3841,
+      "step": 8090
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.182711198428291e-06,
+      "loss": 0.1754,
+      "step": 8100
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.174169300418553e-06,
+      "loss": 0.286,
+      "step": 8110
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.1656274024088157e-06,
+      "loss": 0.2455,
+      "step": 8120
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.1570855043990776e-06,
+      "loss": 0.344,
+      "step": 8130
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.14854360638934e-06,
+      "loss": 0.3695,
+      "step": 8140
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.140001708379602e-06,
+      "loss": 0.1437,
+      "step": 8150
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.1314598103698647e-06,
+      "loss": 0.3331,
+      "step": 8160
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.1229179123601266e-06,
+      "loss": 0.3586,
+      "step": 8170
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.1143760143503885e-06,
+      "loss": 0.1924,
+      "step": 8180
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.1058341163406513e-06,
+      "loss": 0.3175,
+      "step": 8190
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0972922183309132e-06,
+      "loss": 0.3194,
+      "step": 8200
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0887503203211756e-06,
+      "loss": 0.2796,
+      "step": 8210
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.080208422311438e-06,
+      "loss": 0.4855,
+      "step": 8220
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0716665243017003e-06,
+      "loss": 0.3277,
+      "step": 8230
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0631246262919622e-06,
+      "loss": 0.2622,
+      "step": 8240
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.054582728282224e-06,
+      "loss": 0.258,
+      "step": 8250
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.046040830272487e-06,
+      "loss": 0.2123,
+      "step": 8260
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.037498932262749e-06,
+      "loss": 0.3804,
+      "step": 8270
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.028957034253011e-06,
+      "loss": 0.2365,
+      "step": 8280
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0204151362432736e-06,
+      "loss": 0.3523,
+      "step": 8290
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.011873238233536e-06,
+      "loss": 0.3232,
+      "step": 8300
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.003331340223798e-06,
+      "loss": 0.2693,
+      "step": 8310
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.99478944221406e-06,
+      "loss": 0.3266,
+      "step": 8320
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.9862475442043225e-06,
+      "loss": 0.3768,
+      "step": 8330
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.977705646194585e-06,
+      "loss": 0.2142,
+      "step": 8340
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.969163748184847e-06,
+      "loss": 0.2571,
+      "step": 8350
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.9606218501751096e-06,
+      "loss": 0.3081,
+      "step": 8360
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9520799521653715e-06,
+      "loss": 0.1718,
+      "step": 8370
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9435380541556335e-06,
+      "loss": 0.3054,
+      "step": 8380
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.934996156145896e-06,
+      "loss": 0.2853,
+      "step": 8390
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.926454258136158e-06,
+      "loss": 0.3432,
+      "step": 8400
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9179123601264205e-06,
+      "loss": 0.1891,
+      "step": 8410
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9093704621166824e-06,
+      "loss": 0.4322,
+      "step": 8420
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.900828564106945e-06,
+      "loss": 0.2226,
+      "step": 8430
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.892286666097207e-06,
+      "loss": 0.5497,
+      "step": 8440
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.883744768087469e-06,
+      "loss": 0.3176,
+      "step": 8450
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.8752028700777314e-06,
+      "loss": 0.315,
+      "step": 8460
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.8666609720679938e-06,
+      "loss": 0.3701,
+      "step": 8470
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.858119074058256e-06,
+      "loss": 0.4904,
+      "step": 8480
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.849577176048518e-06,
+      "loss": 0.4895,
+      "step": 8490
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.841035278038781e-06,
+      "loss": 0.2017,
+      "step": 8500
+    },
+    {
+      "epoch": 3.59,
+      "eval_loss": 0.3225812315940857,
+      "eval_runtime": 148.6185,
+      "eval_samples_per_second": 13.599,
+      "eval_steps_per_second": 2.268,
+      "step": 8500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.8324933800290428e-06,
+      "loss": 0.5415,
+      "step": 8510
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.8239514820193047e-06,
+      "loss": 0.3701,
+      "step": 8520
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.8154095840095675e-06,
+      "loss": 0.2505,
+      "step": 8530
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.8068676859998294e-06,
+      "loss": 0.2222,
+      "step": 8540
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.7983257879900917e-06,
+      "loss": 0.2643,
+      "step": 8550
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.7897838899803537e-06,
+      "loss": 0.2694,
+      "step": 8560
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.7812419919706164e-06,
+      "loss": 0.2564,
+      "step": 8570
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.7727000939608784e-06,
+      "loss": 0.2611,
+      "step": 8580
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.7641581959511403e-06,
+      "loss": 0.3023,
+      "step": 8590
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.755616297941403e-06,
+      "loss": 0.3328,
+      "step": 8600
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.747074399931665e-06,
+      "loss": 0.2286,
+      "step": 8610
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.7385325019219274e-06,
+      "loss": 0.3566,
+      "step": 8620
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.7299906039121893e-06,
+      "loss": 0.1548,
+      "step": 8630
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.721448705902452e-06,
+      "loss": 0.3871,
+      "step": 8640
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.712906807892714e-06,
+      "loss": 0.2635,
+      "step": 8650
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.704364909882976e-06,
+      "loss": 0.3074,
+      "step": 8660
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.6958230118732387e-06,
+      "loss": 0.2853,
+      "step": 8670
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.6872811138635006e-06,
+      "loss": 0.4648,
+      "step": 8680
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.678739215853763e-06,
+      "loss": 0.2805,
+      "step": 8690
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.670197317844025e-06,
+      "loss": 0.3591,
+      "step": 8700
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.6616554198342877e-06,
+      "loss": 0.3443,
+      "step": 8710
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.6531135218245496e-06,
+      "loss": 0.3672,
+      "step": 8720
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.6445716238148115e-06,
+      "loss": 0.2658,
+      "step": 8730
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.6360297258050743e-06,
+      "loss": 0.3545,
+      "step": 8740
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.6274878277953362e-06,
+      "loss": 0.2131,
+      "step": 8750
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.6189459297855986e-06,
+      "loss": 0.3093,
+      "step": 8760
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.6104040317758605e-06,
+      "loss": 0.2516,
+      "step": 8770
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.6018621337661233e-06,
+      "loss": 0.351,
+      "step": 8780
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.5933202357563852e-06,
+      "loss": 0.4154,
+      "step": 8790
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.584778337746647e-06,
+      "loss": 0.2662,
+      "step": 8800
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.57623643973691e-06,
+      "loss": 0.2682,
+      "step": 8810
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.567694541727172e-06,
+      "loss": 0.2932,
+      "step": 8820
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.559152643717434e-06,
+      "loss": 0.3601,
+      "step": 8830
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.5506107457076966e-06,
+      "loss": 0.3081,
+      "step": 8840
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.542068847697959e-06,
+      "loss": 0.4105,
+      "step": 8850
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.533526949688221e-06,
+      "loss": 0.1959,
+      "step": 8860
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.5249850516784828e-06,
+      "loss": 0.2514,
+      "step": 8870
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.5164431536687455e-06,
+      "loss": 0.4048,
+      "step": 8880
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.5079012556590075e-06,
+      "loss": 0.348,
+      "step": 8890
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.49935935764927e-06,
+      "loss": 0.4546,
+      "step": 8900
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.490817459639532e-06,
+      "loss": 0.1874,
+      "step": 8910
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.4822755616297945e-06,
+      "loss": 0.3567,
+      "step": 8920
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.4737336636200564e-06,
+      "loss": 0.4323,
+      "step": 8930
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.465191765610319e-06,
+      "loss": 0.3178,
+      "step": 8940
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.4566498676005807e-06,
+      "loss": 0.3025,
+      "step": 8950
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.448107969590843e-06,
+      "loss": 0.3172,
+      "step": 8960
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.4395660715811054e-06,
+      "loss": 0.2824,
+      "step": 8970
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.4310241735713678e-06,
+      "loss": 0.2669,
+      "step": 8980
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.42248227556163e-06,
+      "loss": 0.2742,
+      "step": 8990
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.413940377551892e-06,
+      "loss": 0.2043,
+      "step": 9000
+    },
+    {
+      "epoch": 3.81,
+      "eval_loss": 0.3242332935333252,
+      "eval_runtime": 148.8211,
+      "eval_samples_per_second": 13.58,
+      "eval_steps_per_second": 2.264,
+      "step": 9000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.4053984795421544e-06,
+      "loss": 0.3298,
+      "step": 9010
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.3968565815324168e-06,
+      "loss": 0.4304,
+      "step": 9020
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.388314683522679e-06,
+      "loss": 0.2878,
+      "step": 9030
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.3797727855129415e-06,
+      "loss": 0.381,
+      "step": 9040
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3712308875032034e-06,
+      "loss": 0.4256,
+      "step": 9050
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3626889894934658e-06,
+      "loss": 0.3008,
+      "step": 9060
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.3541470914837277e-06,
+      "loss": 0.1974,
+      "step": 9070
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.34560519347399e-06,
+      "loss": 0.2593,
+      "step": 9080
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.3370632954642524e-06,
+      "loss": 0.469,
+      "step": 9090
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.3285213974545147e-06,
+      "loss": 0.3363,
+      "step": 9100
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.319979499444777e-06,
+      "loss": 0.3765,
+      "step": 9110
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.311437601435039e-06,
+      "loss": 0.2657,
+      "step": 9120
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.3028957034253014e-06,
+      "loss": 0.2082,
+      "step": 9130
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.2943538054155633e-06,
+      "loss": 0.3357,
+      "step": 9140
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.2858119074058256e-06,
+      "loss": 0.2493,
+      "step": 9150
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.277270009396088e-06,
+      "loss": 0.4436,
+      "step": 9160
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.2687281113863504e-06,
+      "loss": 0.2447,
+      "step": 9170
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.2601862133766127e-06,
+      "loss": 0.3259,
+      "step": 9180
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.2516443153668746e-06,
+      "loss": 0.2412,
+      "step": 9190
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.243102417357137e-06,
+      "loss": 0.4121,
+      "step": 9200
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.234560519347399e-06,
+      "loss": 0.3135,
+      "step": 9210
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.2260186213376613e-06,
+      "loss": 0.2737,
+      "step": 9220
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.2174767233279236e-06,
+      "loss": 0.3063,
+      "step": 9230
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.208934825318186e-06,
+      "loss": 0.2151,
+      "step": 9240
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.2003929273084483e-06,
+      "loss": 0.3822,
+      "step": 9250
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.1918510292987102e-06,
+      "loss": 0.1962,
+      "step": 9260
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.1833091312889726e-06,
+      "loss": 0.3872,
+      "step": 9270
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.174767233279235e-06,
+      "loss": 0.3354,
+      "step": 9280
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.166225335269497e-06,
+      "loss": 0.3195,
+      "step": 9290
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.1576834372597592e-06,
+      "loss": 0.4004,
+      "step": 9300
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.1491415392500216e-06,
+      "loss": 0.1636,
+      "step": 9310
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.140599641240284e-06,
+      "loss": 0.3894,
+      "step": 9320
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.132057743230546e-06,
+      "loss": 0.1734,
+      "step": 9330
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.1235158452208082e-06,
+      "loss": 0.3858,
+      "step": 9340
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.1149739472110706e-06,
+      "loss": 0.3019,
+      "step": 9350
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1064320492013325e-06,
+      "loss": 0.3738,
+      "step": 9360
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.097890151191595e-06,
+      "loss": 0.3204,
+      "step": 9370
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.089348253181857e-06,
+      "loss": 0.3476,
+      "step": 9380
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.0808063551721196e-06,
+      "loss": 0.2087,
+      "step": 9390
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.072264457162382e-06,
+      "loss": 0.3081,
+      "step": 9400
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.063722559152644e-06,
+      "loss": 0.4049,
+      "step": 9410
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.055180661142906e-06,
+      "loss": 0.2562,
+      "step": 9420
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.046638763133168e-06,
+      "loss": 0.3464,
+      "step": 9430
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0380968651234305e-06,
+      "loss": 0.2826,
+      "step": 9440
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.029554967113693e-06,
+      "loss": 0.3741,
+      "step": 9450
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.021013069103955e-06,
+      "loss": 0.297,
+      "step": 9460
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0124711710942175e-06,
+      "loss": 0.21,
+      "step": 9470
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0039292730844794e-06,
+      "loss": 0.4022,
+      "step": 9480
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.995387375074742e-06,
+      "loss": 0.3468,
+      "step": 9490
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9868454770650037e-06,
+      "loss": 0.3147,
+      "step": 9500
+    },
+    {
+      "epoch": 4.02,
+      "eval_loss": 0.32756686210632324,
+      "eval_runtime": 147.5942,
+      "eval_samples_per_second": 13.693,
+      "eval_steps_per_second": 2.283,
+      "step": 9500
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.978303579055266e-06,
+      "loss": 0.1465,
+      "step": 9510
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9697616810455284e-06,
+      "loss": 0.2636,
+      "step": 9520
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9612197830357908e-06,
+      "loss": 0.2951,
+      "step": 9530
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.952677885026053e-06,
+      "loss": 0.2382,
+      "step": 9540
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.944135987016315e-06,
+      "loss": 0.2929,
+      "step": 9550
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.9355940890065774e-06,
+      "loss": 0.2791,
+      "step": 9560
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.9270521909968393e-06,
+      "loss": 0.2678,
+      "step": 9570
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.9185102929871017e-06,
+      "loss": 0.2825,
+      "step": 9580
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.909968394977364e-06,
+      "loss": 0.2533,
+      "step": 9590
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.9014264969676264e-06,
+      "loss": 0.0906,
+      "step": 9600
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8928845989578888e-06,
+      "loss": 0.3084,
+      "step": 9610
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.8843427009481507e-06,
+      "loss": 0.2555,
+      "step": 9620
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.875800802938413e-06,
+      "loss": 0.3383,
+      "step": 9630
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.8672589049286752e-06,
+      "loss": 0.4268,
+      "step": 9640
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.8587170069189375e-06,
+      "loss": 0.3699,
+      "step": 9650
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.8501751089091999e-06,
+      "loss": 0.2969,
+      "step": 9660
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.841633210899462e-06,
+      "loss": 0.1516,
+      "step": 9670
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.8330913128897244e-06,
+      "loss": 0.3962,
+      "step": 9680
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.8245494148799863e-06,
+      "loss": 0.1523,
+      "step": 9690
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.8160075168702486e-06,
+      "loss": 0.334,
+      "step": 9700
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.807465618860511e-06,
+      "loss": 0.2674,
+      "step": 9710
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7989237208507731e-06,
+      "loss": 0.2575,
+      "step": 9720
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7903818228410355e-06,
+      "loss": 0.2724,
+      "step": 9730
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7818399248312976e-06,
+      "loss": 0.2849,
+      "step": 9740
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.77329802682156e-06,
+      "loss": 0.2505,
+      "step": 9750
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.764756128811822e-06,
+      "loss": 0.2649,
+      "step": 9760
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7562142308020843e-06,
+      "loss": 0.3912,
+      "step": 9770
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7476723327923466e-06,
+      "loss": 0.2389,
+      "step": 9780
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7391304347826088e-06,
+      "loss": 0.2256,
+      "step": 9790
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.7305885367728711e-06,
+      "loss": 0.2882,
+      "step": 9800
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7220466387631332e-06,
+      "loss": 0.3467,
+      "step": 9810
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.7135047407533956e-06,
+      "loss": 0.1719,
+      "step": 9820
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.7049628427436577e-06,
+      "loss": 0.1348,
+      "step": 9830
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.69642094473392e-06,
+      "loss": 0.2768,
+      "step": 9840
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.6878790467241824e-06,
+      "loss": 0.2471,
+      "step": 9850
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.6793371487144444e-06,
+      "loss": 0.2372,
+      "step": 9860
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.6707952507047067e-06,
+      "loss": 0.202,
+      "step": 9870
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.6622533526949689e-06,
+      "loss": 0.1912,
+      "step": 9880
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.6537114546852312e-06,
+      "loss": 0.2944,
+      "step": 9890
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.6451695566754936e-06,
+      "loss": 0.2321,
+      "step": 9900
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.6366276586657557e-06,
+      "loss": 0.2251,
+      "step": 9910
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.628085760656018e-06,
+      "loss": 0.1805,
+      "step": 9920
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.61954386264628e-06,
+      "loss": 0.3134,
+      "step": 9930
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.6110019646365423e-06,
+      "loss": 0.2548,
+      "step": 9940
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.6024600666268045e-06,
+      "loss": 0.233,
+      "step": 9950
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.5939181686170668e-06,
+      "loss": 0.2322,
+      "step": 9960
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.5853762706073292e-06,
+      "loss": 0.433,
+      "step": 9970
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.5768343725975913e-06,
+      "loss": 0.1856,
+      "step": 9980
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.5682924745878537e-06,
+      "loss": 0.1677,
+      "step": 9990
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.5597505765781156e-06,
+      "loss": 0.4449,
+      "step": 10000
+    },
+    {
+      "epoch": 4.23,
+      "eval_loss": 0.3288760483264923,
+      "eval_runtime": 148.5461,
+      "eval_samples_per_second": 13.605,
+      "eval_steps_per_second": 2.269,
+      "step": 10000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.551208678568378e-06,
+      "loss": 0.2478,
+      "step": 10010
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.5426667805586403e-06,
+      "loss": 0.1186,
+      "step": 10020
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.5341248825489024e-06,
+      "loss": 0.2686,
+      "step": 10030
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.5255829845391648e-06,
+      "loss": 0.3326,
+      "step": 10040
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.517041086529427e-06,
+      "loss": 0.2495,
+      "step": 10050
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.5084991885196893e-06,
+      "loss": 0.2794,
+      "step": 10060
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.4999572905099512e-06,
+      "loss": 0.4192,
+      "step": 10070
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.4914153925002136e-06,
+      "loss": 0.1891,
+      "step": 10080
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.482873494490476e-06,
+      "loss": 0.2966,
+      "step": 10090
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.474331596480738e-06,
+      "loss": 0.3602,
+      "step": 10100
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.4657896984710004e-06,
+      "loss": 0.2245,
+      "step": 10110
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.4572478004612626e-06,
+      "loss": 0.3007,
+      "step": 10120
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.448705902451525e-06,
+      "loss": 0.3375,
+      "step": 10130
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.440164004441787e-06,
+      "loss": 0.4571,
+      "step": 10140
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4316221064320494e-06,
+      "loss": 0.2544,
+      "step": 10150
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4230802084223117e-06,
+      "loss": 0.1462,
+      "step": 10160
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4145383104125737e-06,
+      "loss": 0.478,
+      "step": 10170
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.405996412402836e-06,
+      "loss": 0.1972,
+      "step": 10180
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3974545143930982e-06,
+      "loss": 0.3709,
+      "step": 10190
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3889126163833605e-06,
+      "loss": 0.493,
+      "step": 10200
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3803707183736229e-06,
+      "loss": 0.293,
+      "step": 10210
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.371828820363885e-06,
+      "loss": 0.2344,
+      "step": 10220
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3632869223541474e-06,
+      "loss": 0.2663,
+      "step": 10230
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3547450243444093e-06,
+      "loss": 0.2691,
+      "step": 10240
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3462031263346716e-06,
+      "loss": 0.2603,
+      "step": 10250
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3376612283249338e-06,
+      "loss": 0.2077,
+      "step": 10260
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3291193303151961e-06,
+      "loss": 0.357,
+      "step": 10270
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.3205774323054585e-06,
+      "loss": 0.3537,
+      "step": 10280
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.3120355342957206e-06,
+      "loss": 0.3919,
+      "step": 10290
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.303493636285983e-06,
+      "loss": 0.3016,
+      "step": 10300
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.294951738276245e-06,
+      "loss": 0.342,
+      "step": 10310
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2864098402665073e-06,
+      "loss": 0.2113,
+      "step": 10320
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.2778679422567696e-06,
+      "loss": 0.1965,
+      "step": 10330
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.2693260442470318e-06,
+      "loss": 0.4485,
+      "step": 10340
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.260784146237294e-06,
+      "loss": 0.3297,
+      "step": 10350
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.2522422482275562e-06,
+      "loss": 0.206,
+      "step": 10360
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.2437003502178186e-06,
+      "loss": 0.355,
+      "step": 10370
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.2351584522080807e-06,
+      "loss": 0.3375,
+      "step": 10380
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.2266165541983429e-06,
+      "loss": 0.2723,
+      "step": 10390
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.2180746561886052e-06,
+      "loss": 0.2809,
+      "step": 10400
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.2095327581788674e-06,
+      "loss": 0.2849,
+      "step": 10410
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.2009908601691297e-06,
+      "loss": 0.2704,
+      "step": 10420
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1924489621593919e-06,
+      "loss": 0.2916,
+      "step": 10430
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1839070641496542e-06,
+      "loss": 0.2689,
+      "step": 10440
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1753651661399164e-06,
+      "loss": 0.3276,
+      "step": 10450
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1668232681301785e-06,
+      "loss": 0.3632,
+      "step": 10460
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.1582813701204408e-06,
+      "loss": 0.3089,
+      "step": 10470
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.149739472110703e-06,
+      "loss": 0.3956,
+      "step": 10480
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1411975741009653e-06,
+      "loss": 0.4161,
+      "step": 10490
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1326556760912277e-06,
+      "loss": 0.372,
+      "step": 10500
+    },
+    {
+      "epoch": 4.44,
+      "eval_loss": 0.33047744631767273,
+      "eval_runtime": 147.853,
+      "eval_samples_per_second": 13.669,
+      "eval_steps_per_second": 2.279,
+      "step": 10500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.1241137780814898e-06,
+      "loss": 0.3199,
+      "step": 10510
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.115571880071752e-06,
+      "loss": 0.2258,
+      "step": 10520
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.1070299820620143e-06,
+      "loss": 0.2715,
+      "step": 10530
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0984880840522765e-06,
+      "loss": 0.1483,
+      "step": 10540
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0899461860425388e-06,
+      "loss": 0.3261,
+      "step": 10550
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.081404288032801e-06,
+      "loss": 0.4206,
+      "step": 10560
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0728623900230633e-06,
+      "loss": 0.2714,
+      "step": 10570
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0643204920133254e-06,
+      "loss": 0.214,
+      "step": 10580
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0557785940035876e-06,
+      "loss": 0.3961,
+      "step": 10590
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.04723669599385e-06,
+      "loss": 0.2667,
+      "step": 10600
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0386947979841123e-06,
+      "loss": 0.3939,
+      "step": 10610
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0301528999743744e-06,
+      "loss": 0.2613,
+      "step": 10620
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0216110019646366e-06,
+      "loss": 0.3627,
+      "step": 10630
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.013069103954899e-06,
+      "loss": 0.2011,
+      "step": 10640
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.004527205945161e-06,
+      "loss": 0.2729,
+      "step": 10650
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 9.959853079354232e-07,
+      "loss": 0.2348,
+      "step": 10660
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 9.874434099256856e-07,
+      "loss": 0.39,
+      "step": 10670
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 9.78901511915948e-07,
+      "loss": 0.201,
+      "step": 10680
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 9.7035961390621e-07,
+      "loss": 0.2851,
+      "step": 10690
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 9.618177158964722e-07,
+      "loss": 0.1529,
+      "step": 10700
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 9.532758178867345e-07,
+      "loss": 0.1532,
+      "step": 10710
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 9.447339198769967e-07,
+      "loss": 0.4729,
+      "step": 10720
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 9.36192021867259e-07,
+      "loss": 0.3227,
+      "step": 10730
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 9.276501238575213e-07,
+      "loss": 0.2037,
+      "step": 10740
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 9.191082258477835e-07,
+      "loss": 0.1925,
+      "step": 10750
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 9.105663278380457e-07,
+      "loss": 0.3124,
+      "step": 10760
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 9.020244298283079e-07,
+      "loss": 0.4045,
+      "step": 10770
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 8.934825318185702e-07,
+      "loss": 0.3472,
+      "step": 10780
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 8.849406338088323e-07,
+      "loss": 0.5171,
+      "step": 10790
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 8.763987357990946e-07,
+      "loss": 0.3393,
+      "step": 10800
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 8.678568377893569e-07,
+      "loss": 0.2787,
+      "step": 10810
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.593149397796191e-07,
+      "loss": 0.2752,
+      "step": 10820
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.507730417698813e-07,
+      "loss": 0.4029,
+      "step": 10830
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 8.422311437601435e-07,
+      "loss": 0.1633,
+      "step": 10840
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.336892457504058e-07,
+      "loss": 0.2118,
+      "step": 10850
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.251473477406681e-07,
+      "loss": 0.2321,
+      "step": 10860
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.166054497309303e-07,
+      "loss": 0.3549,
+      "step": 10870
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.080635517211925e-07,
+      "loss": 0.2597,
+      "step": 10880
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 7.995216537114548e-07,
+      "loss": 0.4471,
+      "step": 10890
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.90979755701717e-07,
+      "loss": 0.2494,
+      "step": 10900
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.824378576919791e-07,
+      "loss": 0.3454,
+      "step": 10910
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.738959596822415e-07,
+      "loss": 0.3058,
+      "step": 10920
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.653540616725037e-07,
+      "loss": 0.2049,
+      "step": 10930
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.56812163662766e-07,
+      "loss": 0.3407,
+      "step": 10940
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.482702656530281e-07,
+      "loss": 0.4908,
+      "step": 10950
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.397283676432904e-07,
+      "loss": 0.4501,
+      "step": 10960
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.311864696335526e-07,
+      "loss": 0.3257,
+      "step": 10970
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.22644571623815e-07,
+      "loss": 0.3559,
+      "step": 10980
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.141026736140771e-07,
+      "loss": 0.3225,
+      "step": 10990
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 7.055607756043394e-07,
+      "loss": 0.1853,
+      "step": 11000
+    },
+    {
+      "epoch": 4.65,
+      "eval_loss": 0.3304537832736969,
+      "eval_runtime": 148.0543,
+      "eval_samples_per_second": 13.65,
+      "eval_steps_per_second": 2.276,
+      "step": 11000
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.970188775946016e-07,
+      "loss": 0.3078,
+      "step": 11010
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.884769795848637e-07,
+      "loss": 0.2939,
+      "step": 11020
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.79935081575126e-07,
+      "loss": 0.1934,
+      "step": 11030
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.713931835653883e-07,
+      "loss": 0.2136,
+      "step": 11040
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 6.628512855556506e-07,
+      "loss": 0.4001,
+      "step": 11050
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.543093875459128e-07,
+      "loss": 0.2579,
+      "step": 11060
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.45767489536175e-07,
+      "loss": 0.3025,
+      "step": 11070
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 6.372255915264372e-07,
+      "loss": 0.1499,
+      "step": 11080
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.286836935166995e-07,
+      "loss": 0.3261,
+      "step": 11090
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 6.201417955069617e-07,
+      "loss": 0.2846,
+      "step": 11100
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.11599897497224e-07,
+      "loss": 0.2776,
+      "step": 11110
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 6.030579994874862e-07,
+      "loss": 0.3738,
+      "step": 11120
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.945161014777484e-07,
+      "loss": 0.1723,
+      "step": 11130
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.859742034680106e-07,
+      "loss": 0.3108,
+      "step": 11140
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.774323054582729e-07,
+      "loss": 0.1867,
+      "step": 11150
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.688904074485351e-07,
+      "loss": 0.2446,
+      "step": 11160
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.603485094387973e-07,
+      "loss": 0.306,
+      "step": 11170
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.518066114290596e-07,
+      "loss": 0.4434,
+      "step": 11180
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.432647134193218e-07,
+      "loss": 0.2455,
+      "step": 11190
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.347228154095841e-07,
+      "loss": 0.2104,
+      "step": 11200
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.261809173998463e-07,
+      "loss": 0.3635,
+      "step": 11210
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.176390193901086e-07,
+      "loss": 0.3346,
+      "step": 11220
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.090971213803708e-07,
+      "loss": 0.3388,
+      "step": 11230
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 5.005552233706329e-07,
+      "loss": 0.2691,
+      "step": 11240
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.920133253608953e-07,
+      "loss": 0.3303,
+      "step": 11250
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.834714273511574e-07,
+      "loss": 0.2713,
+      "step": 11260
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.749295293414197e-07,
+      "loss": 0.3958,
+      "step": 11270
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.66387631331682e-07,
+      "loss": 0.1392,
+      "step": 11280
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.5784573332194417e-07,
+      "loss": 0.3297,
+      "step": 11290
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.4930383531220636e-07,
+      "loss": 0.2355,
+      "step": 11300
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.4076193730246866e-07,
+      "loss": 0.33,
+      "step": 11310
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.3222003929273085e-07,
+      "loss": 0.3648,
+      "step": 11320
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.236781412829931e-07,
+      "loss": 0.2548,
+      "step": 11330
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.151362432732554e-07,
+      "loss": 0.268,
+      "step": 11340
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.065943452635176e-07,
+      "loss": 0.3429,
+      "step": 11350
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.980524472537798e-07,
+      "loss": 0.3105,
+      "step": 11360
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.895105492440421e-07,
+      "loss": 0.304,
+      "step": 11370
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.809686512343043e-07,
+      "loss": 0.2001,
+      "step": 11380
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.724267532245665e-07,
+      "loss": 0.2684,
+      "step": 11390
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.6388485521482877e-07,
+      "loss": 0.3093,
+      "step": 11400
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.55342957205091e-07,
+      "loss": 0.3993,
+      "step": 11410
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.468010591953532e-07,
+      "loss": 0.2772,
+      "step": 11420
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.382591611856155e-07,
+      "loss": 0.1927,
+      "step": 11430
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.297172631758777e-07,
+      "loss": 0.4071,
+      "step": 11440
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.2117536516613994e-07,
+      "loss": 0.3036,
+      "step": 11450
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.126334671564022e-07,
+      "loss": 0.3697,
+      "step": 11460
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.0409156914666444e-07,
+      "loss": 0.3055,
+      "step": 11470
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.9554967113692663e-07,
+      "loss": 0.2913,
+      "step": 11480
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.870077731271889e-07,
+      "loss": 0.1646,
+      "step": 11490
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.784658751174511e-07,
+      "loss": 0.4051,
+      "step": 11500
+    },
+    {
+      "epoch": 4.86,
+      "eval_loss": 0.3311156928539276,
+      "eval_runtime": 147.2126,
+      "eval_samples_per_second": 13.728,
+      "eval_steps_per_second": 2.289,
+      "step": 11500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 11825,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "total_flos": 656496252190720.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}