diff --git "a/checkpoint-4000/trainer_state.json" "b/checkpoint-4000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-4000/trainer_state.json"
@@ -0,0 +1,7033 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 7.648183556405353,
+  "eval_steps": 500,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0076481835564053535,
+      "grad_norm": 0.294921875,
+      "learning_rate": 2.8571428571428567e-05,
+      "loss": 1.3014,
+      "step": 4
+    },
+    {
+      "epoch": 0.015296367112810707,
+      "grad_norm": 0.26953125,
+      "learning_rate": 5.7142857142857135e-05,
+      "loss": 1.3157,
+      "step": 8
+    },
+    {
+      "epoch": 0.022944550669216062,
+      "grad_norm": 0.201171875,
+      "learning_rate": 8.57142857142857e-05,
+      "loss": 1.2369,
+      "step": 12
+    },
+    {
+      "epoch": 0.030592734225621414,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00011428571428571427,
+      "loss": 1.207,
+      "step": 16
+    },
+    {
+      "epoch": 0.03824091778202677,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.00014285714285714284,
+      "loss": 1.1666,
+      "step": 20
+    },
+    {
+      "epoch": 0.045889101338432124,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0001714285714285714,
+      "loss": 1.178,
+      "step": 24
+    },
+    {
+      "epoch": 0.05353728489483748,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.00019999999999999998,
+      "loss": 1.1533,
+      "step": 28
+    },
+    {
+      "epoch": 0.06118546845124283,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.00022857142857142854,
+      "loss": 1.1508,
+      "step": 32
+    },
+    {
+      "epoch": 0.06883365200764818,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0002571428571428571,
+      "loss": 1.1415,
+      "step": 36
+    },
+    {
+      "epoch": 0.07648183556405354,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0002857142857142857,
+      "loss": 1.1255,
+      "step": 40
+    },
+    {
+      "epoch": 0.0841300191204589,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0002999998274159216,
+      "loss": 1.0581,
+      "step": 44
+    },
+    {
+      "epoch": 0.09177820267686425,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.00029999844674567734,
+      "loss": 1.0987,
+      "step": 48
+    },
+    {
+      "epoch": 0.0994263862332696,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0002999956854178972,
+      "loss": 1.089,
+      "step": 52
+    },
+    {
+      "epoch": 0.10707456978967496,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.00029999154345799773,
+      "loss": 1.0934,
+      "step": 56
+    },
+    {
+      "epoch": 0.1147227533460803,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0002999860209041035,
+      "loss": 1.0712,
+      "step": 60
+    },
+    {
+      "epoch": 0.12237093690248566,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.00029997911780704675,
+      "loss": 1.0253,
+      "step": 64
+    },
+    {
+      "epoch": 0.13001912045889102,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.00029997083423036696,
+      "loss": 1.0526,
+      "step": 68
+    },
+    {
+      "epoch": 0.13766730401529637,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.00029996117025031,
+      "loss": 1.0746,
+      "step": 72
+    },
+    {
+      "epoch": 0.14531548757170173,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.00029995012595582796,
+      "loss": 1.0502,
+      "step": 76
+    },
+    {
+      "epoch": 0.15296367112810708,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0002999377014485777,
+      "loss": 1.0461,
+      "step": 80
+    },
+    {
+      "epoch": 0.16061185468451242,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.00029992389684292025,
+      "loss": 1.0223,
+      "step": 84
+    },
+    {
+      "epoch": 0.1682600382409178,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.00029990871226591995,
+      "loss": 1.0218,
+      "step": 88
+    },
+    {
+      "epoch": 0.17590822179732313,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.00029989214785734286,
+      "loss": 1.0192,
+      "step": 92
+    },
+    {
+      "epoch": 0.1835564053537285,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.00029987420376965577,
+      "loss": 1.0243,
+      "step": 96
+    },
+    {
+      "epoch": 0.19120458891013384,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.00029985488016802457,
+      "loss": 1.0202,
+      "step": 100
+    },
+    {
+      "epoch": 0.1988527724665392,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.00029983417723031307,
+      "loss": 0.9904,
+      "step": 104
+    },
+    {
+      "epoch": 0.20650095602294455,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.0002998120951470809,
+      "loss": 1.0228,
+      "step": 108
+    },
+    {
+      "epoch": 0.21414913957934992,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.00029978863412158217,
+      "loss": 1.004,
+      "step": 112
+    },
+    {
+      "epoch": 0.22179732313575526,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0002997637943697635,
+      "loss": 0.9903,
+      "step": 116
+    },
+    {
+      "epoch": 0.2294455066921606,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.00029973757612026164,
+      "loss": 0.969,
+      "step": 120
+    },
+    {
+      "epoch": 0.23709369024856597,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.00029970997961440213,
+      "loss": 0.9869,
+      "step": 124
+    },
+    {
+      "epoch": 0.2447418738049713,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0002996810051061963,
+      "loss": 0.9758,
+      "step": 128
+    },
+    {
+      "epoch": 0.25239005736137665,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.00029965065286233943,
+      "loss": 0.9615,
+      "step": 132
+    },
+    {
+      "epoch": 0.26003824091778205,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.00029961892316220817,
+      "loss": 0.9541,
+      "step": 136
+    },
+    {
+      "epoch": 0.2676864244741874,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0002995858162978577,
+      "loss": 0.9455,
+      "step": 140
+    },
+    {
+      "epoch": 0.27533460803059273,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.0002995513325740197,
+      "loss": 0.9566,
+      "step": 144
+    },
+    {
+      "epoch": 0.2829827915869981,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00029951547230809865,
+      "loss": 0.9405,
+      "step": 148
+    },
+    {
+      "epoch": 0.29063097514340347,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.00029947823583016973,
+      "loss": 0.9119,
+      "step": 152
+    },
+    {
+      "epoch": 0.2982791586998088,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.00029943962348297535,
+      "loss": 0.9507,
+      "step": 156
+    },
+    {
+      "epoch": 0.30592734225621415,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.00029939963562192196,
+      "loss": 0.9507,
+      "step": 160
+    },
+    {
+      "epoch": 0.3135755258126195,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.000299358272615077,
+      "loss": 0.9815,
+      "step": 164
+    },
+    {
+      "epoch": 0.32122370936902483,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.0002993155348431654,
+      "loss": 0.9364,
+      "step": 168
+    },
+    {
+      "epoch": 0.32887189292543023,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0002992714226995661,
+      "loss": 0.9177,
+      "step": 172
+    },
+    {
+      "epoch": 0.3365200764818356,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.00029922593659030837,
+      "loss": 0.9224,
+      "step": 176
+    },
+    {
+      "epoch": 0.3441682600382409,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.00029917907693406817,
+      "loss": 0.9359,
+      "step": 180
+    },
+    {
+      "epoch": 0.35181644359464626,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.00029913084416216415,
+      "loss": 0.9349,
+      "step": 184
+    },
+    {
+      "epoch": 0.35946462715105165,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.00029908123871855396,
+      "loss": 0.9033,
+      "step": 188
+    },
+    {
+      "epoch": 0.367112810707457,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.0002990302610598297,
+      "loss": 0.9007,
+      "step": 192
+    },
+    {
+      "epoch": 0.37476099426386233,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00029897791165521434,
+      "loss": 0.9243,
+      "step": 196
+    },
+    {
+      "epoch": 0.3824091778202677,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.0002989241909865567,
+      "loss": 0.9095,
+      "step": 200
+    },
+    {
+      "epoch": 0.390057361376673,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0002988690995483276,
+      "loss": 0.8825,
+      "step": 204
+    },
+    {
+      "epoch": 0.3977055449330784,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.00029881263784761503,
+      "loss": 0.8929,
+      "step": 208
+    },
+    {
+      "epoch": 0.40535372848948376,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.00029875480640411957,
+      "loss": 0.9097,
+      "step": 212
+    },
+    {
+      "epoch": 0.4130019120458891,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.00029869560575014945,
+      "loss": 0.8563,
+      "step": 216
+    },
+    {
+      "epoch": 0.42065009560229444,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.00029863503643061585,
+      "loss": 0.8839,
+      "step": 220
+    },
+    {
+      "epoch": 0.42829827915869984,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002985730990030278,
+      "loss": 0.8635,
+      "step": 224
+    },
+    {
+      "epoch": 0.4359464627151052,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.00029850979403748705,
+      "loss": 0.859,
+      "step": 228
+    },
+    {
+      "epoch": 0.4435946462715105,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.00029844512211668286,
+      "loss": 0.8256,
+      "step": 232
+    },
+    {
+      "epoch": 0.45124282982791586,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.00029837908383588646,
+      "loss": 0.8282,
+      "step": 236
+    },
+    {
+      "epoch": 0.4588910133843212,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.0002983116798029459,
+      "loss": 0.8579,
+      "step": 240
+    },
+    {
+      "epoch": 0.4665391969407266,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.0002982429106382801,
+      "loss": 0.8805,
+      "step": 244
+    },
+    {
+      "epoch": 0.47418738049713194,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.00029817277697487347,
+      "loss": 0.823,
+      "step": 248
+    },
+    {
+      "epoch": 0.4818355640535373,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.0002981012794582698,
+      "loss": 0.8546,
+      "step": 252
+    },
+    {
+      "epoch": 0.4894837476099426,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0002980284187465665,
+      "loss": 0.8533,
+      "step": 256
+    },
+    {
+      "epoch": 0.497131931166348,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.00029795419551040833,
+      "loss": 0.8111,
+      "step": 260
+    },
+    {
+      "epoch": 0.5047801147227533,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002978786104329816,
+      "loss": 0.8823,
+      "step": 264
+    },
+    {
+      "epoch": 0.5124282982791587,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0002978016642100076,
+      "loss": 0.8839,
+      "step": 268
+    },
+    {
+      "epoch": 0.5200764818355641,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.00029772335754973614,
+      "loss": 0.8512,
+      "step": 272
+    },
+    {
+      "epoch": 0.5277246653919694,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.00029764369117293925,
+      "loss": 0.8557,
+      "step": 276
+    },
+    {
+      "epoch": 0.5353728489483748,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002975626658129044,
+      "loss": 0.8163,
+      "step": 280
+    },
+    {
+      "epoch": 0.5430210325047801,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0002974802822154278,
+      "loss": 0.9078,
+      "step": 284
+    },
+    {
+      "epoch": 0.5506692160611855,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.00029739654113880755,
+      "loss": 0.8652,
+      "step": 288
+    },
+    {
+      "epoch": 0.5583173996175909,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.00029731144335383663,
+      "loss": 0.8551,
+      "step": 292
+    },
+    {
+      "epoch": 0.5659655831739961,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.0002972249896437958,
+      "loss": 0.8536,
+      "step": 296
+    },
+    {
+      "epoch": 0.5736137667304015,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.0002971371808044464,
+      "loss": 0.825,
+      "step": 300
+    },
+    {
+      "epoch": 0.5812619502868069,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.000297048017644023,
+      "loss": 0.8082,
+      "step": 304
+    },
+    {
+      "epoch": 0.5889101338432122,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002969575009832261,
+      "loss": 0.8304,
+      "step": 308
+    },
+    {
+      "epoch": 0.5965583173996176,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00029686563165521435,
+      "loss": 0.8101,
+      "step": 312
+    },
+    {
+      "epoch": 0.6042065009560229,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.00029677241050559707,
+      "loss": 0.8535,
+      "step": 316
+    },
+    {
+      "epoch": 0.6118546845124283,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.00029667783839242625,
+      "loss": 0.85,
+      "step": 320
+    },
+    {
+      "epoch": 0.6195028680688337,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0002965819161861891,
+      "loss": 0.8101,
+      "step": 324
+    },
+    {
+      "epoch": 0.627151051625239,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0002964846447697994,
+      "loss": 0.8521,
+      "step": 328
+    },
+    {
+      "epoch": 0.6347992351816444,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.00029638602503858995,
+      "loss": 0.8506,
+      "step": 332
+    },
+    {
+      "epoch": 0.6424474187380497,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.00029628605790030384,
+      "loss": 0.8044,
+      "step": 336
+    },
+    {
+      "epoch": 0.6500956022944551,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.0002961847442750866,
+      "loss": 0.8311,
+      "step": 340
+    },
+    {
+      "epoch": 0.6577437858508605,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.00029608208509547735,
+      "loss": 0.8705,
+      "step": 344
+    },
+    {
+      "epoch": 0.6653919694072657,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.00029597808130640027,
+      "loss": 0.8272,
+      "step": 348
+    },
+    {
+      "epoch": 0.6730401529636711,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0002958727338651562,
+      "loss": 0.8035,
+      "step": 352
+    },
+    {
+      "epoch": 0.6806883365200764,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00029576604374141315,
+      "loss": 0.8655,
+      "step": 356
+    },
+    {
+      "epoch": 0.6883365200764818,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.00029565801191719837,
+      "loss": 0.8585,
+      "step": 360
+    },
+    {
+      "epoch": 0.6959847036328872,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002955486393868884,
+      "loss": 0.8279,
+      "step": 364
+    },
+    {
+      "epoch": 0.7036328871892925,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.0002954379271572004,
+      "loss": 0.8229,
+      "step": 368
+    },
+    {
+      "epoch": 0.7112810707456979,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002953258762471828,
+      "loss": 0.8549,
+      "step": 372
+    },
+    {
+      "epoch": 0.7189292543021033,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.0002952124876882058,
+      "loss": 0.837,
+      "step": 376
+    },
+    {
+      "epoch": 0.7265774378585086,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00029509776252395194,
+      "loss": 0.8084,
+      "step": 380
+    },
+    {
+      "epoch": 0.734225621414914,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002949817018104066,
+      "loss": 0.8454,
+      "step": 384
+    },
+    {
+      "epoch": 0.7418738049713193,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0002948643066158482,
+      "loss": 0.7603,
+      "step": 388
+    },
+    {
+      "epoch": 0.7495219885277247,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.00029474557802083834,
+      "loss": 0.8814,
+      "step": 392
+    },
+    {
+      "epoch": 0.7571701720841301,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.0002946255171182119,
+      "loss": 0.8678,
+      "step": 396
+    },
+    {
+      "epoch": 0.7648183556405354,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.00029450412501306675,
+      "loss": 0.8397,
+      "step": 400
+    },
+    {
+      "epoch": 0.7724665391969407,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.00029438140282275413,
+      "loss": 0.797,
+      "step": 404
+    },
+    {
+      "epoch": 0.780114722753346,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.0002942573516768678,
+      "loss": 0.7764,
+      "step": 408
+    },
+    {
+      "epoch": 0.7877629063097514,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00029413197271723385,
+      "loss": 0.7856,
+      "step": 412
+    },
+    {
+      "epoch": 0.7954110898661568,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0002940052670979004,
+      "loss": 0.7878,
+      "step": 416
+    },
+    {
+      "epoch": 0.8030592734225621,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0002938772359851265,
+      "loss": 0.7971,
+      "step": 420
+    },
+    {
+      "epoch": 0.8107074569789675,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.00029374788055737194,
+      "loss": 0.825,
+      "step": 424
+    },
+    {
+      "epoch": 0.8183556405353728,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.000293617202005286,
+      "loss": 0.8105,
+      "step": 428
+    },
+    {
+      "epoch": 0.8260038240917782,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00029348520153169656,
+      "loss": 0.8055,
+      "step": 432
+    },
+    {
+      "epoch": 0.8336520076481836,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0002933518803515993,
+      "loss": 0.7616,
+      "step": 436
+    },
+    {
+      "epoch": 0.8413001912045889,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00029321723969214625,
+      "loss": 0.7842,
+      "step": 440
+    },
+    {
+      "epoch": 0.8489483747609943,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.0002930812807926343,
+      "loss": 0.803,
+      "step": 444
+    },
+    {
+      "epoch": 0.8565965583173997,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0002929440049044945,
+      "loss": 0.8174,
+      "step": 448
+    },
+    {
+      "epoch": 0.864244741873805,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.0002928054132912797,
+      "loss": 0.836,
+      "step": 452
+    },
+    {
+      "epoch": 0.8718929254302104,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0002926655072286536,
+      "loss": 0.7552,
+      "step": 456
+    },
+    {
+      "epoch": 0.8795411089866156,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.00029252428800437854,
+      "loss": 0.7755,
+      "step": 460
+    },
+    {
+      "epoch": 0.887189292543021,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.00029238175691830395,
+      "loss": 0.8101,
+      "step": 464
+    },
+    {
+      "epoch": 0.8948374760994264,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.00029223791528235407,
+      "loss": 0.7662,
+      "step": 468
+    },
+    {
+      "epoch": 0.9024856596558317,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.00029209276442051634,
+      "loss": 0.7772,
+      "step": 472
+    },
+    {
+      "epoch": 0.9101338432122371,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.0002919463056688287,
+      "loss": 0.8135,
+      "step": 476
+    },
+    {
+      "epoch": 0.9177820267686424,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.00029179854037536773,
+      "loss": 0.7895,
+      "step": 480
+    },
+    {
+      "epoch": 0.9254302103250478,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0002916494699002358,
+      "loss": 0.7644,
+      "step": 484
+    },
+    {
+      "epoch": 0.9330783938814532,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.00029149909561554914,
+      "loss": 0.7804,
+      "step": 488
+    },
+    {
+      "epoch": 0.9407265774378585,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0002913474189054246,
+      "loss": 0.7779,
+      "step": 492
+    },
+    {
+      "epoch": 0.9483747609942639,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0002911944411659672,
+      "loss": 0.781,
+      "step": 496
+    },
+    {
+      "epoch": 0.9560229445506692,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002910401638052574,
+      "loss": 0.8079,
+      "step": 500
+    },
+    {
+      "epoch": 0.9636711281070746,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00029088458824333787,
+      "loss": 0.8332,
+      "step": 504
+    },
+    {
+      "epoch": 0.97131931166348,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.00029072771591220057,
+      "loss": 0.7752,
+      "step": 508
+    },
+    {
+      "epoch": 0.9789674952198852,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00029056954825577353,
+      "loss": 0.8038,
+      "step": 512
+    },
+    {
+      "epoch": 0.9866156787762906,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.0002904100867299077,
+      "loss": 0.7362,
+      "step": 516
+    },
+    {
+      "epoch": 0.994263862332696,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0002902493328023633,
+      "loss": 0.8137,
+      "step": 520
+    },
+    {
+      "epoch": 1.0019120458891013,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0002900872879527964,
+      "loss": 0.7389,
+      "step": 524
+    },
+    {
+      "epoch": 1.0095602294455066,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00028992395367274547,
+      "loss": 0.6994,
+      "step": 528
+    },
+    {
+      "epoch": 1.0172084130019121,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0002897593314656174,
+      "loss": 0.7461,
+      "step": 532
+    },
+    {
+      "epoch": 1.0248565965583174,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0002895934228466738,
+      "loss": 0.6856,
+      "step": 536
+    },
+    {
+      "epoch": 1.0325047801147227,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.0002894262293430171,
+      "loss": 0.7405,
+      "step": 540
+    },
+    {
+      "epoch": 1.0401529636711282,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0002892577524935763,
+      "loss": 0.7906,
+      "step": 544
+    },
+    {
+      "epoch": 1.0478011472275335,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00028908799384909313,
+      "loss": 0.7144,
+      "step": 548
+    },
+    {
+      "epoch": 1.0554493307839388,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0002889169549721073,
+      "loss": 0.7233,
+      "step": 552
+    },
+    {
+      "epoch": 1.063097514340344,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00028874463743694265,
+      "loss": 0.7068,
+      "step": 556
+    },
+    {
+      "epoch": 1.0707456978967496,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0002885710428296922,
+      "loss": 0.7105,
+      "step": 560
+    },
+    {
+      "epoch": 1.0783938814531548,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.00028839617274820404,
+      "loss": 0.6563,
+      "step": 564
+    },
+    {
+      "epoch": 1.0860420650095601,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.00028822002880206593,
+      "loss": 0.6956,
+      "step": 568
+    },
+    {
+      "epoch": 1.0936902485659656,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00028804261261259115,
+      "loss": 0.6669,
+      "step": 572
+    },
+    {
+      "epoch": 1.101338432122371,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00028786392581280334,
+      "loss": 0.6727,
+      "step": 576
+    },
+    {
+      "epoch": 1.1089866156787762,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00028768397004742135,
+      "loss": 0.714,
+      "step": 580
+    },
+    {
+      "epoch": 1.1166347992351817,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00028750274697284423,
+      "loss": 0.666,
+      "step": 584
+    },
+    {
+      "epoch": 1.124282982791587,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.00028732025825713587,
+      "loss": 0.6883,
+      "step": 588
+    },
+    {
+      "epoch": 1.1319311663479923,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00028713650558000983,
+      "loss": 0.7002,
+      "step": 592
+    },
+    {
+      "epoch": 1.1395793499043978,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002869514906328138,
+      "loss": 0.7038,
+      "step": 596
+    },
+    {
+      "epoch": 1.147227533460803,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00028676521511851395,
+      "loss": 0.7178,
+      "step": 600
+    },
+    {
+      "epoch": 1.1548757170172084,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002865776807516793,
+      "loss": 0.6752,
+      "step": 604
+    },
+    {
+      "epoch": 1.1625239005736137,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.0002863888892584659,
+      "loss": 0.7204,
+      "step": 608
+    },
+    {
+      "epoch": 1.1701720841300192,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00028619884237660124,
+      "loss": 0.6967,
+      "step": 612
+    },
+    {
+      "epoch": 1.1778202676864244,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0002860075418553676,
+      "loss": 0.705,
+      "step": 616
+    },
+    {
+      "epoch": 1.1854684512428297,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00028581498945558676,
+      "loss": 0.7208,
+      "step": 620
+    },
+    {
+      "epoch": 1.1931166347992352,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00028562118694960316,
+      "loss": 0.6872,
+      "step": 624
+    },
+    {
+      "epoch": 1.2007648183556405,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.0002854261361212679,
+      "loss": 0.7445,
+      "step": 628
+    },
+    {
+      "epoch": 1.2084130019120458,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.00028522983876592213,
+      "loss": 0.7267,
+      "step": 632
+    },
+    {
+      "epoch": 1.2160611854684513,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002850322966903808,
+      "loss": 0.6831,
+      "step": 636
+    },
+    {
+      "epoch": 1.2237093690248566,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00028483351171291576,
+      "loss": 0.7185,
+      "step": 640
+    },
+    {
+      "epoch": 1.231357552581262,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00028463348566323913,
+      "loss": 0.7019,
+      "step": 644
+    },
+    {
+      "epoch": 1.2390057361376674,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00028443222038248645,
+      "loss": 0.7699,
+      "step": 648
+    },
+    {
+      "epoch": 1.2466539196940727,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.00028422971772319977,
+      "loss": 0.6939,
+      "step": 652
+    },
+    {
+      "epoch": 1.254302103250478,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00028402597954931046,
+      "loss": 0.6505,
+      "step": 656
+    },
+    {
+      "epoch": 1.2619502868068833,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00028382100773612236,
+      "loss": 0.6632,
+      "step": 660
+    },
+    {
+      "epoch": 1.2695984703632888,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002836148041702941,
+      "loss": 0.7787,
+      "step": 664
+    },
+    {
+      "epoch": 1.277246653919694,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.00028340737074982207,
+      "loss": 0.7293,
+      "step": 668
+    },
+    {
+      "epoch": 1.2848948374760996,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0002831987093840229,
+      "loss": 0.7471,
+      "step": 672
+    },
+    {
+      "epoch": 1.2925430210325048,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.00028298882199351565,
+      "loss": 0.7445,
+      "step": 676
+    },
+    {
+      "epoch": 1.3001912045889101,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00028277771051020433,
+      "loss": 0.6997,
+      "step": 680
+    },
+    {
+      "epoch": 1.3078393881453154,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.00028256537687726017,
+      "loss": 0.7389,
+      "step": 684
+    },
+    {
+      "epoch": 1.3154875717017207,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002823518230491036,
+      "loss": 0.7171,
+      "step": 688
+    },
+    {
+      "epoch": 1.3231357552581262,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00028213705099138636,
+      "loss": 0.7054,
+      "step": 692
+    },
+    {
+      "epoch": 1.3307839388145315,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00028192106268097334,
+      "loss": 0.6543,
+      "step": 696
+    },
+    {
+      "epoch": 1.338432122370937,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002817038601059243,
+      "loss": 0.7012,
+      "step": 700
+    },
+    {
+      "epoch": 1.3460803059273423,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002814854452654758,
+      "loss": 0.7058,
+      "step": 704
+    },
+    {
+      "epoch": 1.3537284894837476,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00028126582017002266,
+      "loss": 0.7797,
+      "step": 708
+    },
+    {
+      "epoch": 1.3613766730401529,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002810449868410994,
+      "loss": 0.6898,
+      "step": 712
+    },
+    {
+      "epoch": 1.3690248565965584,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.00028082294731136164,
+      "loss": 0.6872,
+      "step": 716
+    },
+    {
+      "epoch": 1.3766730401529637,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00028059970362456776,
+      "loss": 0.706,
+      "step": 720
+    },
+    {
+      "epoch": 1.384321223709369,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.00028037525783555935,
+      "loss": 0.6971,
+      "step": 724
+    },
+    {
+      "epoch": 1.3919694072657744,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00028014961201024304,
+      "loss": 0.7212,
+      "step": 728
+    },
+    {
+      "epoch": 1.3996175908221797,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0002799227682255711,
+      "loss": 0.702,
+      "step": 732
+    },
+    {
+      "epoch": 1.407265774378585,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00027969472856952224,
+      "loss": 0.6604,
+      "step": 736
+    },
+    {
+      "epoch": 1.4149139579349903,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00027946549514108277,
+      "loss": 0.7,
+      "step": 740
+    },
+    {
+      "epoch": 1.4225621414913958,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00027923507005022687,
+      "loss": 0.7281,
+      "step": 744
+    },
+    {
+      "epoch": 1.430210325047801,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.00027900345541789746,
+      "loss": 0.6261,
+      "step": 748
+    },
+    {
+      "epoch": 1.4378585086042066,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0002787706533759865,
+      "loss": 0.7131,
+      "step": 752
+    },
+    {
+      "epoch": 1.445506692160612,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00027853666606731547,
+      "loss": 0.664,
+      "step": 756
+    },
+    {
+      "epoch": 1.4531548757170172,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0002783014956456157,
+      "loss": 0.6856,
+      "step": 760
+    },
+    {
+      "epoch": 1.4608030592734225,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0002780651442755083,
+      "loss": 0.7422,
+      "step": 764
+    },
+    {
+      "epoch": 1.468451242829828,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.0002778276141324844,
+      "loss": 0.6738,
+      "step": 768
+    },
+    {
+      "epoch": 1.4760994263862333,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002775889074028853,
+      "loss": 0.7595,
+      "step": 772
+    },
+    {
+      "epoch": 1.4837476099426385,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.000277349026283882,
+      "loss": 0.7132,
+      "step": 776
+    },
+    {
+      "epoch": 1.491395793499044,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0002771079729834552,
+      "loss": 0.7316,
+      "step": 780
+    },
+    {
+      "epoch": 1.4990439770554493,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002768657497203749,
+      "loss": 0.7314,
+      "step": 784
+    },
+    {
+      "epoch": 1.5066921606118546,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00027662235872418005,
+      "loss": 0.7036,
+      "step": 788
+    },
+    {
+      "epoch": 1.51434034416826,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00027637780223515793,
+      "loss": 0.7191,
+      "step": 792
+    },
+    {
+      "epoch": 1.5219885277246654,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00027613208250432353,
+      "loss": 0.712,
+      "step": 796
+    },
+    {
+      "epoch": 1.5296367112810707,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.000275885201793399,
+      "loss": 0.6918,
+      "step": 800
+    },
+    {
+      "epoch": 1.5372848948374762,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002756371623747925,
+      "loss": 0.6822,
+      "step": 804
+    },
+    {
+      "epoch": 1.5449330783938815,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002753879665315778,
+      "loss": 0.707,
+      "step": 808
+    },
+    {
+      "epoch": 1.5525812619502868,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002751376165574726,
+      "loss": 0.6921,
+      "step": 812
+    },
+    {
+      "epoch": 1.560229445506692,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002748861147568181,
+      "loss": 0.7048,
+      "step": 816
+    },
+    {
+      "epoch": 1.5678776290630974,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00027463346344455724,
+      "loss": 0.7171,
+      "step": 820
+    },
+    {
+      "epoch": 1.5755258126195029,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002743796649462137,
+      "loss": 0.6839,
+      "step": 824
+    },
+    {
+      "epoch": 1.5831739961759084,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00027412472159787037,
+      "loss": 0.6722,
+      "step": 828
+    },
+    {
+      "epoch": 1.5908221797323137,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00027386863574614803,
+      "loss": 0.6538,
+      "step": 832
+    },
+    {
+      "epoch": 1.598470363288719,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002736114097481833,
+      "loss": 0.7013,
+      "step": 836
+    },
+    {
+      "epoch": 1.6061185468451242,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002733530459716076,
+      "loss": 0.6967,
+      "step": 840
+    },
+    {
+      "epoch": 1.6137667304015295,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00027309354679452483,
+      "loss": 0.7052,
+      "step": 844
+    },
+    {
+      "epoch": 1.621414913957935,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002728329146054897,
+      "loss": 0.6677,
+      "step": 848
+    },
+    {
+      "epoch": 1.6290630975143403,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00027257115180348557,
+      "loss": 0.7128,
+      "step": 852
+    },
+    {
+      "epoch": 1.6367112810707458,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0002723082607979028,
+      "loss": 0.7085,
+      "step": 856
+    },
+    {
+      "epoch": 1.644359464627151,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00027204424400851596,
+      "loss": 0.6719,
+      "step": 860
+    },
+    {
+      "epoch": 1.6520076481835564,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00027177910386546206,
+      "loss": 0.6873,
+      "step": 864
+    },
+    {
+      "epoch": 1.6596558317399617,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.00027151284280921794,
+      "loss": 0.6919,
+      "step": 868
+    },
+    {
+      "epoch": 1.667304015296367,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0002712454632905779,
+      "loss": 0.6967,
+      "step": 872
+    },
+    {
+      "epoch": 1.6749521988527725,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00027097696777063113,
+      "loss": 0.7067,
+      "step": 876
+    },
+    {
+      "epoch": 1.682600382409178,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00027070735872073885,
+      "loss": 0.6303,
+      "step": 880
+    },
+    {
+      "epoch": 1.6902485659655833,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002704366386225119,
+      "loss": 0.6828,
+      "step": 884
+    },
+    {
+      "epoch": 1.6978967495219885,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002701648099677878,
+      "loss": 0.6876,
+      "step": 888
+    },
+    {
+      "epoch": 1.7055449330783938,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002698918752586075,
+      "loss": 0.7079,
+      "step": 892
+    },
+    {
+      "epoch": 1.7131931166347991,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00026961783700719293,
+      "loss": 0.6956,
+      "step": 896
+    },
+    {
+      "epoch": 1.7208413001912046,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002693426977359233,
+      "loss": 0.7143,
+      "step": 900
+    },
+    {
+      "epoch": 1.72848948374761,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002690664599773122,
+      "loss": 0.6973,
+      "step": 904
+    },
+    {
+      "epoch": 1.7361376673040154,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00026878912627398434,
+      "loss": 0.68,
+      "step": 908
+    },
+    {
+      "epoch": 1.7437858508604207,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.0002685106991786519,
+      "loss": 0.7011,
+      "step": 912
+    },
+    {
+      "epoch": 1.751434034416826,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00026823118125409107,
+      "loss": 0.6862,
+      "step": 916
+    },
+    {
+      "epoch": 1.7590822179732313,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002679505750731189,
+      "loss": 0.6929,
+      "step": 920
+    },
+    {
+      "epoch": 1.7667304015296366,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00026766888321856896,
+      "loss": 0.6927,
+      "step": 924
+    },
+    {
+      "epoch": 1.774378585086042,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.000267386108283268,
+      "loss": 0.6618,
+      "step": 928
+    },
+    {
+      "epoch": 1.7820267686424476,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0002671022528700118,
+      "loss": 0.7173,
+      "step": 932
+    },
+    {
+      "epoch": 1.7896749521988529,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.00026681731959154174,
+      "loss": 0.7314,
+      "step": 936
+    },
+    {
+      "epoch": 1.7973231357552581,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00026653131107052,
+      "loss": 0.7013,
+      "step": 940
+    },
+    {
+      "epoch": 1.8049713193116634,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00026624422993950603,
+      "loss": 0.7591,
+      "step": 944
+    },
+    {
+      "epoch": 1.8126195028680687,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002659560788409321,
+      "loss": 0.6398,
+      "step": 948
+    },
+    {
+      "epoch": 1.8202676864244742,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0002656668604270788,
+      "loss": 0.6778,
+      "step": 952
+    },
+    {
+      "epoch": 1.8279158699808795,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00026537657736005094,
+      "loss": 0.6543,
+      "step": 956
+    },
+    {
+      "epoch": 1.835564053537285,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.000265085232311753,
+      "loss": 0.7096,
+      "step": 960
+    },
+    {
+      "epoch": 1.8432122370936903,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.00026479282796386416,
+      "loss": 0.6939,
+      "step": 964
+    },
+    {
+      "epoch": 1.8508604206500956,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00026449936700781413,
+      "loss": 0.728,
+      "step": 968
+    },
+    {
+      "epoch": 1.8585086042065009,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002642048521447581,
+      "loss": 0.6862,
+      "step": 972
+    },
+    {
+      "epoch": 1.8661567877629062,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00026390928608555195,
+      "loss": 0.6767,
+      "step": 976
+    },
+    {
+      "epoch": 1.8738049713193117,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.0002636126715507272,
+      "loss": 0.6229,
+      "step": 980
+    },
+    {
+      "epoch": 1.8814531548757172,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.000263315011270466,
+      "loss": 0.6941,
+      "step": 984
+    },
+    {
+      "epoch": 1.8891013384321225,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00026301630798457613,
+      "loss": 0.6854,
+      "step": 988
+    },
+    {
+      "epoch": 1.8967495219885278,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00026271656444246577,
+      "loss": 0.7136,
+      "step": 992
+    },
+    {
+      "epoch": 1.904397705544933,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.000262415783403118,
+      "loss": 0.696,
+      "step": 996
+    },
+    {
+      "epoch": 1.9120458891013383,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00026211396763506546,
+      "loss": 0.6688,
+      "step": 1000
+    },
+    {
+      "epoch": 1.9196940726577438,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002618111199163651,
+      "loss": 0.6953,
+      "step": 1004
+    },
+    {
+      "epoch": 1.9273422562141491,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00026150724303457235,
+      "loss": 0.7481,
+      "step": 1008
+    },
+    {
+      "epoch": 1.9349904397705546,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002612023397867155,
+      "loss": 0.675,
+      "step": 1012
+    },
+    {
+      "epoch": 1.94263862332696,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.00026089641297927,
+      "loss": 0.6684,
+      "step": 1016
+    },
+    {
+      "epoch": 1.9502868068833652,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0002605894654281329,
+      "loss": 0.6547,
+      "step": 1020
+    },
+    {
+      "epoch": 1.9579349904397705,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002602814999585963,
+      "loss": 0.7232,
+      "step": 1024
+    },
+    {
+      "epoch": 1.9655831739961758,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0002599725194053219,
+      "loss": 0.7069,
+      "step": 1028
+    },
+    {
+      "epoch": 1.9732313575525813,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002596625266123146,
+      "loss": 0.7303,
+      "step": 1032
+    },
+    {
+      "epoch": 1.9808795411089866,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00025935152443289664,
+      "loss": 0.7096,
+      "step": 1036
+    },
+    {
+      "epoch": 1.988527724665392,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00025903951572968094,
+      "loss": 0.7055,
+      "step": 1040
+    },
+    {
+      "epoch": 1.9961759082217974,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00025872650337454504,
+      "loss": 0.7108,
+      "step": 1044
+    },
+    {
+      "epoch": 2.0038240917782026,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.00025841249024860453,
+      "loss": 0.5808,
+      "step": 1048
+    },
+    {
+      "epoch": 2.011472275334608,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00025809747924218667,
+      "loss": 0.58,
+      "step": 1052
+    },
+    {
+      "epoch": 2.019120458891013,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00025778147325480357,
+      "loss": 0.6208,
+      "step": 1056
+    },
+    {
+      "epoch": 2.026768642447419,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002574644751951256,
+      "loss": 0.5692,
+      "step": 1060
+    },
+    {
+      "epoch": 2.0344168260038242,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00025714648798095483,
+      "loss": 0.5891,
+      "step": 1064
+    },
+    {
+      "epoch": 2.0420650095602295,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.00025682751453919776,
+      "loss": 0.6214,
+      "step": 1068
+    },
+    {
+      "epoch": 2.049713193116635,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002565075578058388,
+      "loss": 0.5889,
+      "step": 1072
+    },
+    {
+      "epoch": 2.05736137667304,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0002561866207259128,
+      "loss": 0.6099,
+      "step": 1076
+    },
+    {
+      "epoch": 2.0650095602294454,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002558647062534785,
+      "loss": 0.5722,
+      "step": 1080
+    },
+    {
+      "epoch": 2.0726577437858507,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002555418173515908,
+      "loss": 0.609,
+      "step": 1084
+    },
+    {
+      "epoch": 2.0803059273422564,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0002552179569922737,
+      "loss": 0.6158,
+      "step": 1088
+    },
+    {
+      "epoch": 2.0879541108986617,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00025489312815649314,
+      "loss": 0.588,
+      "step": 1092
+    },
+    {
+      "epoch": 2.095602294455067,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00025456733383412926,
+      "loss": 0.6278,
+      "step": 1096
+    },
+    {
+      "epoch": 2.1032504780114722,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.000254240577023949,
+      "loss": 0.5915,
+      "step": 1100
+    },
+    {
+      "epoch": 2.1108986615678775,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00025391286073357856,
+      "loss": 0.5764,
+      "step": 1104
+    },
+    {
+      "epoch": 2.118546845124283,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002535841879794755,
+      "loss": 0.6146,
+      "step": 1108
+    },
+    {
+      "epoch": 2.126195028680688,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002532545617869014,
+      "loss": 0.5794,
+      "step": 1112
+    },
+    {
+      "epoch": 2.133843212237094,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0002529239851898935,
+      "loss": 0.6412,
+      "step": 1116
+    },
+    {
+      "epoch": 2.141491395793499,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00025259246123123706,
+      "loss": 0.6288,
+      "step": 1120
+    },
+    {
+      "epoch": 2.1491395793499044,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0002522599929624375,
+      "loss": 0.5644,
+      "step": 1124
+    },
+    {
+      "epoch": 2.1567877629063097,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00025192658344369193,
+      "loss": 0.6219,
+      "step": 1128
+    },
+    {
+      "epoch": 2.164435946462715,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00025159223574386114,
+      "loss": 0.6015,
+      "step": 1132
+    },
+    {
+      "epoch": 2.1720841300191203,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00025125695294044156,
+      "loss": 0.612,
+      "step": 1136
+    },
+    {
+      "epoch": 2.179732313575526,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0002509207381195366,
+      "loss": 0.5878,
+      "step": 1140
+    },
+    {
+      "epoch": 2.1873804971319313,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0002505835943758286,
+      "loss": 0.6134,
+      "step": 1144
+    },
+    {
+      "epoch": 2.1950286806883366,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00025024552481254993,
+      "loss": 0.5663,
+      "step": 1148
+    },
+    {
+      "epoch": 2.202676864244742,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002499065325414547,
+      "loss": 0.5685,
+      "step": 1152
+    },
+    {
+      "epoch": 2.210325047801147,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00024956662068279027,
+      "loss": 0.5839,
+      "step": 1156
+    },
+    {
+      "epoch": 2.2179732313575524,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00024922579236526807,
+      "loss": 0.611,
+      "step": 1160
+    },
+    {
+      "epoch": 2.2256214149139577,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00024888405072603513,
+      "loss": 0.6218,
+      "step": 1164
+    },
+    {
+      "epoch": 2.2332695984703634,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0002485413989106452,
+      "loss": 0.5743,
+      "step": 1168
+    },
+    {
+      "epoch": 2.2409177820267687,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00024819784007302966,
+      "loss": 0.5921,
+      "step": 1172
+    },
+    {
+      "epoch": 2.248565965583174,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00024785337737546863,
+      "loss": 0.5444,
+      "step": 1176
+    },
+    {
+      "epoch": 2.2562141491395793,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0002475080139885617,
+      "loss": 0.5823,
+      "step": 1180
+    },
+    {
+      "epoch": 2.2638623326959846,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00024716175309119875,
+      "loss": 0.5788,
+      "step": 1184
+    },
+    {
+      "epoch": 2.27151051625239,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00024681459787053106,
+      "loss": 0.5666,
+      "step": 1188
+    },
+    {
+      "epoch": 2.2791586998087956,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0002464665515219415,
+      "loss": 0.5988,
+      "step": 1192
+    },
+    {
+      "epoch": 2.286806883365201,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0002461176172490153,
+      "loss": 0.5851,
+      "step": 1196
+    },
+    {
+      "epoch": 2.294455066921606,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0002457677982635107,
+      "loss": 0.6328,
+      "step": 1200
+    },
+    {
+      "epoch": 2.3021032504780115,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.000245417097785329,
+      "loss": 0.6264,
+      "step": 1204
+    },
+    {
+      "epoch": 2.3097514340344167,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00024506551904248546,
+      "loss": 0.5853,
+      "step": 1208
+    },
+    {
+      "epoch": 2.317399617590822,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00024471306527107915,
+      "loss": 0.571,
+      "step": 1212
+    },
+    {
+      "epoch": 2.3250478011472273,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0002443597397152634,
+      "loss": 0.6076,
+      "step": 1216
+    },
+    {
+      "epoch": 2.332695984703633,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00024400554562721585,
+      "loss": 0.6026,
+      "step": 1220
+    },
+    {
+      "epoch": 2.3403441682600383,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00024365048626710843,
+      "loss": 0.6196,
+      "step": 1224
+    },
+    {
+      "epoch": 2.3479923518164436,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00024329456490307757,
+      "loss": 0.5704,
+      "step": 1228
+    },
+    {
+      "epoch": 2.355640535372849,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00024293778481119396,
+      "loss": 0.6217,
+      "step": 1232
+    },
+    {
+      "epoch": 2.363288718929254,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0002425801492754324,
+      "loss": 0.6238,
+      "step": 1236
+    },
+    {
+      "epoch": 2.3709369024856595,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00024222166158764161,
+      "loss": 0.5945,
+      "step": 1240
+    },
+    {
+      "epoch": 2.378585086042065,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00024186232504751397,
+      "loss": 0.5605,
+      "step": 1244
+    },
+    {
+      "epoch": 2.3862332695984705,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002415021429625551,
+      "loss": 0.6451,
+      "step": 1248
+    },
+    {
+      "epoch": 2.3938814531548758,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00024114111864805338,
+      "loss": 0.5581,
+      "step": 1252
+    },
+    {
+      "epoch": 2.401529636711281,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00024077925542704949,
+      "loss": 0.6212,
+      "step": 1256
+    },
+    {
+      "epoch": 2.4091778202676863,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002404165566303057,
+      "loss": 0.6121,
+      "step": 1260
+    },
+    {
+      "epoch": 2.4168260038240916,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00024005302559627561,
+      "loss": 0.578,
+      "step": 1264
+    },
+    {
+      "epoch": 2.424474187380497,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00023968866567107282,
+      "loss": 0.6353,
+      "step": 1268
+    },
+    {
+      "epoch": 2.4321223709369026,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00023932348020844064,
+      "loss": 0.5811,
+      "step": 1272
+    },
+    {
+      "epoch": 2.439770554493308,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.00023895747256972083,
+      "loss": 0.5834,
+      "step": 1276
+    },
+    {
+      "epoch": 2.447418738049713,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00023859064612382315,
+      "loss": 0.6234,
+      "step": 1280
+    },
+    {
+      "epoch": 2.4550669216061185,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0002382230042471938,
+      "loss": 0.6298,
+      "step": 1284
+    },
+    {
+      "epoch": 2.462715105162524,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.0002378545503237846,
+      "loss": 0.6834,
+      "step": 1288
+    },
+    {
+      "epoch": 2.470363288718929,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00023748528774502194,
+      "loss": 0.6176,
+      "step": 1292
+    },
+    {
+      "epoch": 2.478011472275335,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00023711521990977554,
+      "loss": 0.6141,
+      "step": 1296
+    },
+    {
+      "epoch": 2.48565965583174,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.00023674435022432683,
+      "loss": 0.5958,
+      "step": 1300
+    },
+    {
+      "epoch": 2.4933078393881454,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0002363726821023381,
+      "loss": 0.6249,
+      "step": 1304
+    },
+    {
+      "epoch": 2.5009560229445507,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00023600021896482063,
+      "loss": 0.5865,
+      "step": 1308
+    },
+    {
+      "epoch": 2.508604206500956,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0002356269642401036,
+      "loss": 0.5982,
+      "step": 1312
+    },
+    {
+      "epoch": 2.5162523900573612,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002352529213638022,
+      "loss": 0.5722,
+      "step": 1316
+    },
+    {
+      "epoch": 2.5239005736137665,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0002348780937787862,
+      "loss": 0.6308,
+      "step": 1320
+    },
+    {
+      "epoch": 2.5315487571701722,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00023450248493514817,
+      "loss": 0.6104,
+      "step": 1324
+    },
+    {
+      "epoch": 2.5391969407265775,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00023412609829017182,
+      "loss": 0.6053,
+      "step": 1328
+    },
+    {
+      "epoch": 2.546845124282983,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00023374893730830005,
+      "loss": 0.5447,
+      "step": 1332
+    },
+    {
+      "epoch": 2.554493307839388,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00023337100546110313,
+      "loss": 0.5987,
+      "step": 1336
+    },
+    {
+      "epoch": 2.5621414913957934,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0002329923062272468,
+      "loss": 0.6037,
+      "step": 1340
+    },
+    {
+      "epoch": 2.569789674952199,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0002326128430924602,
+      "loss": 0.6055,
+      "step": 1344
+    },
+    {
+      "epoch": 2.5774378585086044,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00023223261954950363,
+      "loss": 0.6163,
+      "step": 1348
+    },
+    {
+      "epoch": 2.5850860420650097,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00023185163909813678,
+      "loss": 0.6272,
+      "step": 1352
+    },
+    {
+      "epoch": 2.592734225621415,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00023146990524508613,
+      "loss": 0.6303,
+      "step": 1356
+    },
+    {
+      "epoch": 2.6003824091778203,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00023108742150401284,
+      "loss": 0.5862,
+      "step": 1360
+    },
+    {
+      "epoch": 2.6080305927342256,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00023070419139548044,
+      "loss": 0.6083,
+      "step": 1364
+    },
+    {
+      "epoch": 2.615678776290631,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00023032021844692242,
+      "loss": 0.6466,
+      "step": 1368
+    },
+    {
+      "epoch": 2.623326959847036,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0002299355061926096,
+      "loss": 0.6236,
+      "step": 1372
+    },
+    {
+      "epoch": 2.6309751434034414,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00022955005817361783,
+      "loss": 0.6203,
+      "step": 1376
+    },
+    {
+      "epoch": 2.638623326959847,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00022916387793779533,
+      "loss": 0.6101,
+      "step": 1380
+    },
+    {
+      "epoch": 2.6462715105162524,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00022877696903972984,
+      "loss": 0.5361,
+      "step": 1384
+    },
+    {
+      "epoch": 2.6539196940726577,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00022838933504071618,
+      "loss": 0.594,
+      "step": 1388
+    },
+    {
+      "epoch": 2.661567877629063,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0002280009795087233,
+      "loss": 0.6323,
+      "step": 1392
+    },
+    {
+      "epoch": 2.6692160611854687,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00022761190601836142,
+      "loss": 0.5863,
+      "step": 1396
+    },
+    {
+      "epoch": 2.676864244741874,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00022722211815084944,
+      "loss": 0.6109,
+      "step": 1400
+    },
+    {
+      "epoch": 2.6845124282982793,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0002268316194939815,
+      "loss": 0.6062,
+      "step": 1404
+    },
+    {
+      "epoch": 2.6921606118546846,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0002264404136420941,
+      "loss": 0.5996,
+      "step": 1408
+    },
+    {
+      "epoch": 2.69980879541109,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0002260485041960334,
+      "loss": 0.6301,
+      "step": 1412
+    },
+    {
+      "epoch": 2.707456978967495,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00022565589476312157,
+      "loss": 0.5807,
+      "step": 1416
+    },
+    {
+      "epoch": 2.7151051625239004,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00022526258895712377,
+      "loss": 0.6542,
+      "step": 1420
+    },
+    {
+      "epoch": 2.7227533460803057,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00022486859039821513,
+      "loss": 0.6001,
+      "step": 1424
+    },
+    {
+      "epoch": 2.730401529636711,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00022447390271294697,
+      "loss": 0.5997,
+      "step": 1428
+    },
+    {
+      "epoch": 2.7380497131931167,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00022407852953421382,
+      "loss": 0.6401,
+      "step": 1432
+    },
+    {
+      "epoch": 2.745697896749522,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00022368247450121965,
+      "loss": 0.5744,
+      "step": 1436
+    },
+    {
+      "epoch": 2.7533460803059273,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00022328574125944476,
+      "loss": 0.5853,
+      "step": 1440
+    },
+    {
+      "epoch": 2.7609942638623326,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.00022288833346061182,
+      "loss": 0.5861,
+      "step": 1444
+    },
+    {
+      "epoch": 2.768642447418738,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0002224902547626526,
+      "loss": 0.6194,
+      "step": 1448
+    },
+    {
+      "epoch": 2.7762906309751436,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00022209150882967398,
+      "loss": 0.604,
+      "step": 1452
+    },
+    {
+      "epoch": 2.783938814531549,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00022169209933192458,
+      "loss": 0.6312,
+      "step": 1456
+    },
+    {
+      "epoch": 2.791586998087954,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0002212920299457606,
+      "loss": 0.6312,
+      "step": 1460
+    },
+    {
+      "epoch": 2.7992351816443595,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0002208913043536123,
+      "loss": 0.6089,
+      "step": 1464
+    },
+    {
+      "epoch": 2.8068833652007648,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00022048992624394988,
+      "loss": 0.6349,
+      "step": 1468
+    },
+    {
+      "epoch": 2.81453154875717,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00022008789931124976,
+      "loss": 0.591,
+      "step": 1472
+    },
+    {
+      "epoch": 2.8221797323135753,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0002196852272559603,
+      "loss": 0.6085,
+      "step": 1476
+    },
+    {
+      "epoch": 2.8298279158699806,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00021928191378446795,
+      "loss": 0.6284,
+      "step": 1480
+    },
+    {
+      "epoch": 2.8374760994263863,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00021887796260906304,
+      "loss": 0.5796,
+      "step": 1484
+    },
+    {
+      "epoch": 2.8451242829827916,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00021847337744790562,
+      "loss": 0.5739,
+      "step": 1488
+    },
+    {
+      "epoch": 2.852772466539197,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0002180681620249913,
+      "loss": 0.6,
+      "step": 1492
+    },
+    {
+      "epoch": 2.860420650095602,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00021766232007011682,
+      "loss": 0.6113,
+      "step": 1496
+    },
+    {
+      "epoch": 2.8680688336520075,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0002172558553188459,
+      "loss": 0.5949,
+      "step": 1500
+    },
+    {
+      "epoch": 2.875717017208413,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00021684877151247485,
+      "loss": 0.5929,
+      "step": 1504
+    },
+    {
+      "epoch": 2.8833652007648185,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00021644107239799786,
+      "loss": 0.6224,
+      "step": 1508
+    },
+    {
+      "epoch": 2.891013384321224,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00021603276172807288,
+      "loss": 0.6439,
+      "step": 1512
+    },
+    {
+      "epoch": 2.898661567877629,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00021562384326098688,
+      "loss": 0.5835,
+      "step": 1516
+    },
+    {
+      "epoch": 2.9063097514340344,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0002152143207606211,
+      "loss": 0.6,
+      "step": 1520
+    },
+    {
+      "epoch": 2.9139579349904396,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00021480419799641692,
+      "loss": 0.5959,
+      "step": 1524
+    },
+    {
+      "epoch": 2.921606118546845,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0002143934787433406,
+      "loss": 0.6111,
+      "step": 1528
+    },
+    {
+      "epoch": 2.92925430210325,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00021398216678184884,
+      "loss": 0.6072,
+      "step": 1532
+    },
+    {
+      "epoch": 2.936902485659656,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00021357026589785392,
+      "loss": 0.5744,
+      "step": 1536
+    },
+    {
+      "epoch": 2.9445506692160612,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00021315777988268876,
+      "loss": 0.603,
+      "step": 1540
+    },
+    {
+      "epoch": 2.9521988527724665,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00021274471253307224,
+      "loss": 0.6364,
+      "step": 1544
+    },
+    {
+      "epoch": 2.959847036328872,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00021233106765107407,
+      "loss": 0.5766,
+      "step": 1548
+    },
+    {
+      "epoch": 2.967495219885277,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00021191684904407976,
+      "loss": 0.5715,
+      "step": 1552
+    },
+    {
+      "epoch": 2.975143403441683,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002115020605247558,
+      "loss": 0.6015,
+      "step": 1556
+    },
+    {
+      "epoch": 2.982791586998088,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00021108670591101433,
+      "loss": 0.6017,
+      "step": 1560
+    },
+    {
+      "epoch": 2.9904397705544934,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00021067078902597814,
+      "loss": 0.6201,
+      "step": 1564
+    },
+    {
+      "epoch": 2.9980879541108987,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0002102543136979454,
+      "loss": 0.6178,
+      "step": 1568
+    },
+    {
+      "epoch": 3.005736137667304,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00020983728376035448,
+      "loss": 0.464,
+      "step": 1572
+    },
+    {
+      "epoch": 3.0133843212237093,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00020941970305174862,
+      "loss": 0.4966,
+      "step": 1576
+    },
+    {
+      "epoch": 3.0210325047801145,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00020900157541574066,
+      "loss": 0.5019,
+      "step": 1580
+    },
+    {
+      "epoch": 3.0286806883365203,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00020858290470097762,
+      "loss": 0.477,
+      "step": 1584
+    },
+    {
+      "epoch": 3.0363288718929256,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00020816369476110512,
+      "loss": 0.5126,
+      "step": 1588
+    },
+    {
+      "epoch": 3.043977055449331,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0002077439494547324,
+      "loss": 0.5319,
+      "step": 1592
+    },
+    {
+      "epoch": 3.051625239005736,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.00020732367264539612,
+      "loss": 0.4976,
+      "step": 1596
+    },
+    {
+      "epoch": 3.0592734225621414,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00020690286820152534,
+      "loss": 0.4638,
+      "step": 1600
+    },
+    {
+      "epoch": 3.0669216061185467,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0002064815399964057,
+      "loss": 0.5326,
+      "step": 1604
+    },
+    {
+      "epoch": 3.0745697896749524,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00020605969190814374,
+      "loss": 0.5066,
+      "step": 1608
+    },
+    {
+      "epoch": 3.0822179732313577,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002056373278196313,
+      "loss": 0.4611,
+      "step": 1612
+    },
+    {
+      "epoch": 3.089866156787763,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0002052144516185097,
+      "loss": 0.511,
+      "step": 1616
+    },
+    {
+      "epoch": 3.0975143403441683,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00020479106719713402,
+      "loss": 0.5338,
+      "step": 1620
+    },
+    {
+      "epoch": 3.1051625239005736,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00020436717845253723,
+      "loss": 0.5384,
+      "step": 1624
+    },
+    {
+      "epoch": 3.112810707456979,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0002039427892863943,
+      "loss": 0.5188,
+      "step": 1628
+    },
+    {
+      "epoch": 3.120458891013384,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00020351790360498636,
+      "loss": 0.5045,
+      "step": 1632
+    },
+    {
+      "epoch": 3.12810707456979,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00020309252531916475,
+      "loss": 0.4894,
+      "step": 1636
+    },
+    {
+      "epoch": 3.135755258126195,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00020266665834431486,
+      "loss": 0.5241,
+      "step": 1640
+    },
+    {
+      "epoch": 3.1434034416826004,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00020224030660032023,
+      "loss": 0.4838,
+      "step": 1644
+    },
+    {
+      "epoch": 3.1510516252390057,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00020181347401152652,
+      "loss": 0.5449,
+      "step": 1648
+    },
+    {
+      "epoch": 3.158699808795411,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0002013861645067054,
+      "loss": 0.5428,
+      "step": 1652
+    },
+    {
+      "epoch": 3.1663479923518163,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00020095838201901798,
+      "loss": 0.5506,
+      "step": 1656
+    },
+    {
+      "epoch": 3.173996175908222,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00020053013048597926,
+      "loss": 0.4983,
+      "step": 1660
+    },
+    {
+      "epoch": 3.1816443594646273,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00020010141384942148,
+      "loss": 0.4941,
+      "step": 1664
+    },
+    {
+      "epoch": 3.1892925430210326,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0001996722360554577,
+      "loss": 0.5406,
+      "step": 1668
+    },
+    {
+      "epoch": 3.196940726577438,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00019924260105444602,
+      "loss": 0.484,
+      "step": 1672
+    },
+    {
+      "epoch": 3.204588910133843,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00019881251280095261,
+      "loss": 0.5374,
+      "step": 1676
+    },
+    {
+      "epoch": 3.2122370936902485,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00019838197525371583,
+      "loss": 0.4977,
+      "step": 1680
+    },
+    {
+      "epoch": 3.2198852772466537,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0001979509923756094,
+      "loss": 0.5462,
+      "step": 1684
+    },
+    {
+      "epoch": 3.2275334608030595,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0001975195681336061,
+      "loss": 0.4945,
+      "step": 1688
+    },
+    {
+      "epoch": 3.2351816443594648,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00019708770649874132,
+      "loss": 0.5042,
+      "step": 1692
+    },
+    {
+      "epoch": 3.24282982791587,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00019665541144607627,
+      "loss": 0.5369,
+      "step": 1696
+    },
+    {
+      "epoch": 3.2504780114722753,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00019622268695466166,
+      "loss": 0.5121,
+      "step": 1700
+    },
+    {
+      "epoch": 3.2581261950286806,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.000195789537007501,
+      "loss": 0.4984,
+      "step": 1704
+    },
+    {
+      "epoch": 3.265774378585086,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00019535596559151376,
+      "loss": 0.5391,
+      "step": 1708
+    },
+    {
+      "epoch": 3.2734225621414916,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00019492197669749892,
+      "loss": 0.4778,
+      "step": 1712
+    },
+    {
+      "epoch": 3.281070745697897,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00019448757432009807,
+      "loss": 0.5131,
+      "step": 1716
+    },
+    {
+      "epoch": 3.288718929254302,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00019405276245775877,
+      "loss": 0.4901,
+      "step": 1720
+    },
+    {
+      "epoch": 3.2963671128107075,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00019361754511269753,
+      "loss": 0.5426,
+      "step": 1724
+    },
+    {
+      "epoch": 3.3040152963671128,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019318192629086327,
+      "loss": 0.5105,
+      "step": 1728
+    },
+    {
+      "epoch": 3.311663479923518,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019274591000190028,
+      "loss": 0.5448,
+      "step": 1732
+    },
+    {
+      "epoch": 3.3193116634799233,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00019230950025911123,
+      "loss": 0.5079,
+      "step": 1736
+    },
+    {
+      "epoch": 3.3269598470363286,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0001918727010794204,
+      "loss": 0.5185,
+      "step": 1740
+    },
+    {
+      "epoch": 3.3346080305927344,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0001914355164833366,
+      "loss": 0.4918,
+      "step": 1744
+    },
+    {
+      "epoch": 3.3422562141491396,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019099795049491621,
+      "loss": 0.5051,
+      "step": 1748
+    },
+    {
+      "epoch": 3.349904397705545,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00019056000714172617,
+      "loss": 0.5295,
+      "step": 1752
+    },
+    {
+      "epoch": 3.35755258126195,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019012169045480676,
+      "loss": 0.5455,
+      "step": 1756
+    },
+    {
+      "epoch": 3.3652007648183555,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00018968300446863478,
+      "loss": 0.515,
+      "step": 1760
+    },
+    {
+      "epoch": 3.3728489483747612,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00018924395322108607,
+      "loss": 0.4868,
+      "step": 1764
+    },
+    {
+      "epoch": 3.3804971319311665,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00018880454075339854,
+      "loss": 0.5087,
+      "step": 1768
+    },
+    {
+      "epoch": 3.388145315487572,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00018836477111013495,
+      "loss": 0.5179,
+      "step": 1772
+    },
+    {
+      "epoch": 3.395793499043977,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00018792464833914576,
+      "loss": 0.5613,
+      "step": 1776
+    },
+    {
+      "epoch": 3.4034416826003824,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001874841764915317,
+      "loss": 0.5269,
+      "step": 1780
+    },
+    {
+      "epoch": 3.4110898661567877,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00018704335962160663,
+      "loss": 0.5024,
+      "step": 1784
+    },
+    {
+      "epoch": 3.418738049713193,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00018660220178686002,
+      "loss": 0.5272,
+      "step": 1788
+    },
+    {
+      "epoch": 3.4263862332695982,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001861607070479199,
+      "loss": 0.4757,
+      "step": 1792
+    },
+    {
+      "epoch": 3.434034416826004,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00018571887946851535,
+      "loss": 0.5167,
+      "step": 1796
+    },
+    {
+      "epoch": 3.4416826003824093,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00018527672311543887,
+      "loss": 0.5162,
+      "step": 1800
+    },
+    {
+      "epoch": 3.4493307839388145,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00018483424205850934,
+      "loss": 0.5711,
+      "step": 1804
+    },
+    {
+      "epoch": 3.45697896749522,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0001843914403705343,
+      "loss": 0.5373,
+      "step": 1808
+    },
+    {
+      "epoch": 3.464627151051625,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00018394832212727252,
+      "loss": 0.5118,
+      "step": 1812
+    },
+    {
+      "epoch": 3.472275334608031,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00018350489140739654,
+      "loss": 0.5283,
+      "step": 1816
+    },
+    {
+      "epoch": 3.479923518164436,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.00018306115229245506,
+      "loss": 0.4733,
+      "step": 1820
+    },
+    {
+      "epoch": 3.4875717017208414,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00018261710886683538,
+      "loss": 0.5091,
+      "step": 1824
+    },
+    {
+      "epoch": 3.4952198852772467,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001821727652177258,
+      "loss": 0.5055,
+      "step": 1828
+    },
+    {
+      "epoch": 3.502868068833652,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00018172812543507813,
+      "loss": 0.518,
+      "step": 1832
+    },
+    {
+      "epoch": 3.5105162523900573,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00018128319361156978,
+      "loss": 0.5309,
+      "step": 1836
+    },
+    {
+      "epoch": 3.5181644359464626,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001808379738425664,
+      "loss": 0.5381,
+      "step": 1840
+    },
+    {
+      "epoch": 3.525812619502868,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00018039247022608393,
+      "loss": 0.5596,
+      "step": 1844
+    },
+    {
+      "epoch": 3.5334608030592736,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00017994668686275092,
+      "loss": 0.5198,
+      "step": 1848
+    },
+    {
+      "epoch": 3.541108986615679,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00017950062785577104,
+      "loss": 0.542,
+      "step": 1852
+    },
+    {
+      "epoch": 3.548757170172084,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00017905429731088497,
+      "loss": 0.5138,
+      "step": 1856
+    },
+    {
+      "epoch": 3.5564053537284894,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001786076993363328,
+      "loss": 0.5144,
+      "step": 1860
+    },
+    {
+      "epoch": 3.5640535372848947,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001781608380428161,
+      "loss": 0.5127,
+      "step": 1864
+    },
+    {
+      "epoch": 3.5717017208413004,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0001777137175434602,
+      "loss": 0.5018,
+      "step": 1868
+    },
+    {
+      "epoch": 3.5793499043977057,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00017726634195377642,
+      "loss": 0.4387,
+      "step": 1872
+    },
+    {
+      "epoch": 3.586998087954111,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00017681871539162382,
+      "loss": 0.5421,
+      "step": 1876
+    },
+    {
+      "epoch": 3.5946462715105163,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00017637084197717163,
+      "loss": 0.5118,
+      "step": 1880
+    },
+    {
+      "epoch": 3.6022944550669216,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00017592272583286125,
+      "loss": 0.5017,
+      "step": 1884
+    },
+    {
+      "epoch": 3.609942638623327,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00017547437108336836,
+      "loss": 0.533,
+      "step": 1888
+    },
+    {
+      "epoch": 3.617590822179732,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00017502578185556468,
+      "loss": 0.5954,
+      "step": 1892
+    },
+    {
+      "epoch": 3.6252390057361374,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00017457696227848036,
+      "loss": 0.5098,
+      "step": 1896
+    },
+    {
+      "epoch": 3.632887189292543,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00017412791648326566,
+      "loss": 0.4859,
+      "step": 1900
+    },
+    {
+      "epoch": 3.6405353728489485,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001736786486031531,
+      "loss": 0.5466,
+      "step": 1904
+    },
+    {
+      "epoch": 3.6481835564053537,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00017322916277341945,
+      "loss": 0.5258,
+      "step": 1908
+    },
+    {
+      "epoch": 3.655831739961759,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00017277946313134758,
+      "loss": 0.5302,
+      "step": 1912
+    },
+    {
+      "epoch": 3.6634799235181643,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00017232955381618826,
+      "loss": 0.5487,
+      "step": 1916
+    },
+    {
+      "epoch": 3.67112810707457,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00017187943896912236,
+      "loss": 0.497,
+      "step": 1920
+    },
+    {
+      "epoch": 3.6787762906309753,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.0001714291227332224,
+      "loss": 0.4907,
+      "step": 1924
+    },
+    {
+      "epoch": 3.6864244741873806,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00017097860925341472,
+      "loss": 0.5322,
+      "step": 1928
+    },
+    {
+      "epoch": 3.694072657743786,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00017052790267644112,
+      "loss": 0.4859,
+      "step": 1932
+    },
+    {
+      "epoch": 3.701720841300191,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00017007700715082077,
+      "loss": 0.5101,
+      "step": 1936
+    },
+    {
+      "epoch": 3.7093690248565965,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00016962592682681206,
+      "loss": 0.5091,
+      "step": 1940
+    },
+    {
+      "epoch": 3.7170172084130018,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00016917466585637426,
+      "loss": 0.5399,
+      "step": 1944
+    },
+    {
+      "epoch": 3.724665391969407,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001687232283931294,
+      "loss": 0.4829,
+      "step": 1948
+    },
+    {
+      "epoch": 3.7323135755258128,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00016827161859232418,
+      "loss": 0.5355,
+      "step": 1952
+    },
+    {
+      "epoch": 3.739961759082218,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00016781984061079138,
+      "loss": 0.5442,
+      "step": 1956
+    },
+    {
+      "epoch": 3.7476099426386233,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00016736789860691197,
+      "loss": 0.5238,
+      "step": 1960
+    },
+    {
+      "epoch": 3.7552581261950286,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00016691579674057657,
+      "loss": 0.5004,
+      "step": 1964
+    },
+    {
+      "epoch": 3.762906309751434,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00016646353917314726,
+      "loss": 0.4718,
+      "step": 1968
+    },
+    {
+      "epoch": 3.7705544933078396,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00016601113006741916,
+      "loss": 0.4324,
+      "step": 1972
+    },
+    {
+      "epoch": 3.778202676864245,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00016555857358758252,
+      "loss": 0.457,
+      "step": 1976
+    },
+    {
+      "epoch": 3.78585086042065,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00016510587389918373,
+      "loss": 0.4992,
+      "step": 1980
+    },
+    {
+      "epoch": 3.7934990439770555,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00016465303516908762,
+      "loss": 0.4984,
+      "step": 1984
+    },
+    {
+      "epoch": 3.801147227533461,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0001642000615654387,
+      "loss": 0.5278,
+      "step": 1988
+    },
+    {
+      "epoch": 3.808795411089866,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001637469572576229,
+      "loss": 0.5272,
+      "step": 1992
+    },
+    {
+      "epoch": 3.8164435946462714,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00016329372641622934,
+      "loss": 0.5019,
+      "step": 1996
+    },
+    {
+      "epoch": 3.8240917782026767,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00016284037321301166,
+      "loss": 0.4939,
+      "step": 2000
+    },
+    {
+      "epoch": 3.8317399617590824,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00016238690182084986,
+      "loss": 0.5335,
+      "step": 2004
+    },
+    {
+      "epoch": 3.8393881453154877,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00016193331641371176,
+      "loss": 0.5396,
+      "step": 2008
+    },
+    {
+      "epoch": 3.847036328871893,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00016147962116661472,
+      "loss": 0.5078,
+      "step": 2012
+    },
+    {
+      "epoch": 3.8546845124282982,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00016102582025558703,
+      "loss": 0.5286,
+      "step": 2016
+    },
+    {
+      "epoch": 3.8623326959847035,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00016057191785762964,
+      "loss": 0.504,
+      "step": 2020
+    },
+    {
+      "epoch": 3.8699808795411093,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00016011791815067754,
+      "loss": 0.5131,
+      "step": 2024
+    },
+    {
+      "epoch": 3.8776290630975145,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00015966382531356144,
+      "loss": 0.5068,
+      "step": 2028
+    },
+    {
+      "epoch": 3.88527724665392,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00015920964352596927,
+      "loss": 0.5257,
+      "step": 2032
+    },
+    {
+      "epoch": 3.892925430210325,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00015875537696840775,
+      "loss": 0.5145,
+      "step": 2036
+    },
+    {
+      "epoch": 3.9005736137667304,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.0001583010298221638,
+      "loss": 0.5219,
+      "step": 2040
+    },
+    {
+      "epoch": 3.9082217973231357,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.0001578466062692661,
+      "loss": 0.5548,
+      "step": 2044
+    },
+    {
+      "epoch": 3.915869980879541,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00015739211049244667,
+      "loss": 0.4981,
+      "step": 2048
+    },
+    {
+      "epoch": 3.9235181644359463,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00015693754667510235,
+      "loss": 0.5184,
+      "step": 2052
+    },
+    {
+      "epoch": 3.9311663479923515,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00015648291900125609,
+      "loss": 0.5207,
+      "step": 2056
+    },
+    {
+      "epoch": 3.9388145315487573,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00015602823165551877,
+      "loss": 0.5194,
+      "step": 2060
+    },
+    {
+      "epoch": 3.9464627151051626,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0001555734888230505,
+      "loss": 0.4692,
+      "step": 2064
+    },
+    {
+      "epoch": 3.954110898661568,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00015511869468952201,
+      "loss": 0.5247,
+      "step": 2068
+    },
+    {
+      "epoch": 3.961759082217973,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0001546638534410763,
+      "loss": 0.4561,
+      "step": 2072
+    },
+    {
+      "epoch": 3.969407265774379,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00015420896926429014,
+      "loss": 0.5073,
+      "step": 2076
+    },
+    {
+      "epoch": 3.977055449330784,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00015375404634613524,
+      "loss": 0.5323,
+      "step": 2080
+    },
+    {
+      "epoch": 3.9847036328871894,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00015329908887393992,
+      "loss": 0.5223,
+      "step": 2084
+    },
+    {
+      "epoch": 3.9923518164435947,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0001528441010353508,
+      "loss": 0.5112,
+      "step": 2088
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.00015238908701829378,
+      "loss": 0.5374,
+      "step": 2092
+    },
+    {
+      "epoch": 4.007648183556405,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0001519340510109357,
+      "loss": 0.446,
+      "step": 2096
+    },
+    {
+      "epoch": 4.015296367112811,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.00015147899720164594,
+      "loss": 0.467,
+      "step": 2100
+    },
+    {
+      "epoch": 4.022944550669216,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00015102392977895765,
+      "loss": 0.4112,
+      "step": 2104
+    },
+    {
+      "epoch": 4.030592734225621,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00015056885293152932,
+      "loss": 0.4401,
+      "step": 2108
+    },
+    {
+      "epoch": 4.038240917782026,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00015011377084810624,
+      "loss": 0.406,
+      "step": 2112
+    },
+    {
+      "epoch": 4.045889101338432,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00014965868771748178,
+      "loss": 0.4528,
+      "step": 2116
+    },
+    {
+      "epoch": 4.053537284894838,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00014920360772845896,
+      "loss": 0.4345,
+      "step": 2120
+    },
+    {
+      "epoch": 4.061185468451243,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00014874853506981206,
+      "loss": 0.4349,
+      "step": 2124
+    },
+    {
+      "epoch": 4.0688336520076485,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00014829347393024764,
+      "loss": 0.4494,
+      "step": 2128
+    },
+    {
+      "epoch": 4.076481835564054,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00014783842849836644,
+      "loss": 0.4159,
+      "step": 2132
+    },
+    {
+      "epoch": 4.084130019120459,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00014738340296262443,
+      "loss": 0.372,
+      "step": 2136
+    },
+    {
+      "epoch": 4.091778202676864,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00014692840151129467,
+      "loss": 0.4621,
+      "step": 2140
+    },
+    {
+      "epoch": 4.09942638623327,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00014647342833242827,
+      "loss": 0.4095,
+      "step": 2144
+    },
+    {
+      "epoch": 4.107074569789675,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00014601848761381633,
+      "loss": 0.4172,
+      "step": 2148
+    },
+    {
+      "epoch": 4.11472275334608,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00014556358354295113,
+      "loss": 0.4197,
+      "step": 2152
+    },
+    {
+      "epoch": 4.1223709369024855,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.0001451087203069875,
+      "loss": 0.4573,
+      "step": 2156
+    },
+    {
+      "epoch": 4.130019120458891,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00014465390209270456,
+      "loss": 0.4117,
+      "step": 2160
+    },
+    {
+      "epoch": 4.137667304015296,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00014419913308646686,
+      "loss": 0.4148,
+      "step": 2164
+    },
+    {
+      "epoch": 4.145315487571701,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00014374441747418628,
+      "loss": 0.4251,
+      "step": 2168
+    },
+    {
+      "epoch": 4.1529636711281075,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00014328975944128292,
+      "loss": 0.4314,
+      "step": 2172
+    },
+    {
+      "epoch": 4.160611854684513,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00014283516317264704,
+      "loss": 0.4386,
+      "step": 2176
+    },
+    {
+      "epoch": 4.168260038240918,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00014238063285260057,
+      "loss": 0.392,
+      "step": 2180
+    },
+    {
+      "epoch": 4.175908221797323,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00014192617266485803,
+      "loss": 0.4187,
+      "step": 2184
+    },
+    {
+      "epoch": 4.183556405353729,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001414717867924888,
+      "loss": 0.4165,
+      "step": 2188
+    },
+    {
+      "epoch": 4.191204588910134,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001410174794178779,
+      "loss": 0.4191,
+      "step": 2192
+    },
+    {
+      "epoch": 4.198852772466539,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00014056325472268805,
+      "loss": 0.436,
+      "step": 2196
+    },
+    {
+      "epoch": 4.2065009560229445,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001401091168878209,
+      "loss": 0.4075,
+      "step": 2200
+    },
+    {
+      "epoch": 4.21414913957935,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00013965507009337845,
+      "loss": 0.3995,
+      "step": 2204
+    },
+    {
+      "epoch": 4.221797323135755,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00013920111851862494,
+      "loss": 0.4474,
+      "step": 2208
+    },
+    {
+      "epoch": 4.22944550669216,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00013874726634194797,
+      "loss": 0.405,
+      "step": 2212
+    },
+    {
+      "epoch": 4.237093690248566,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.0001382935177408204,
+      "loss": 0.4316,
+      "step": 2216
+    },
+    {
+      "epoch": 4.244741873804971,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00013783987689176157,
+      "loss": 0.4299,
+      "step": 2220
+    },
+    {
+      "epoch": 4.252390057361376,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00013738634797029914,
+      "loss": 0.4347,
+      "step": 2224
+    },
+    {
+      "epoch": 4.260038240917782,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00013693293515093052,
+      "loss": 0.4393,
+      "step": 2228
+    },
+    {
+      "epoch": 4.267686424474188,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00013647964260708436,
+      "loss": 0.44,
+      "step": 2232
+    },
+    {
+      "epoch": 4.275334608030593,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001360264745110824,
+      "loss": 0.4194,
+      "step": 2236
+    },
+    {
+      "epoch": 4.282982791586998,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001355734350341007,
+      "loss": 0.4259,
+      "step": 2240
+    },
+    {
+      "epoch": 4.2906309751434035,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00013512052834613165,
+      "loss": 0.4311,
+      "step": 2244
+    },
+    {
+      "epoch": 4.298279158699809,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00013466775861594523,
+      "loss": 0.4097,
+      "step": 2248
+    },
+    {
+      "epoch": 4.305927342256214,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0001342151300110509,
+      "loss": 0.4149,
+      "step": 2252
+    },
+    {
+      "epoch": 4.313575525812619,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0001337626466976591,
+      "loss": 0.3983,
+      "step": 2256
+    },
+    {
+      "epoch": 4.321223709369025,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.0001333103128406429,
+      "loss": 0.4424,
+      "step": 2260
+    },
+    {
+      "epoch": 4.32887189292543,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00013285813260349982,
+      "loss": 0.4352,
+      "step": 2264
+    },
+    {
+      "epoch": 4.336520076481835,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001324061101483132,
+      "loss": 0.4441,
+      "step": 2268
+    },
+    {
+      "epoch": 4.3441682600382405,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00013195424963571424,
+      "loss": 0.421,
+      "step": 2272
+    },
+    {
+      "epoch": 4.351816443594647,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00013150255522484345,
+      "loss": 0.4131,
+      "step": 2276
+    },
+    {
+      "epoch": 4.359464627151052,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00013105103107331255,
+      "loss": 0.3568,
+      "step": 2280
+    },
+    {
+      "epoch": 4.367112810707457,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00013059968133716606,
+      "loss": 0.445,
+      "step": 2284
+    },
+    {
+      "epoch": 4.374760994263863,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00013014851017084303,
+      "loss": 0.4267,
+      "step": 2288
+    },
+    {
+      "epoch": 4.382409177820268,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00012969752172713905,
+      "loss": 0.4458,
+      "step": 2292
+    },
+    {
+      "epoch": 4.390057361376673,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00012924672015716759,
+      "loss": 0.439,
+      "step": 2296
+    },
+    {
+      "epoch": 4.397705544933078,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00012879610961032218,
+      "loss": 0.4792,
+      "step": 2300
+    },
+    {
+      "epoch": 4.405353728489484,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0001283456942342383,
+      "loss": 0.4113,
+      "step": 2304
+    },
+    {
+      "epoch": 4.413001912045889,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001278954781747545,
+      "loss": 0.4548,
+      "step": 2308
+    },
+    {
+      "epoch": 4.420650095602294,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00012744546557587517,
+      "loss": 0.4512,
+      "step": 2312
+    },
+    {
+      "epoch": 4.4282982791587,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00012699566057973168,
+      "loss": 0.4211,
+      "step": 2316
+    },
+    {
+      "epoch": 4.435946462715105,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00012654606732654468,
+      "loss": 0.4256,
+      "step": 2320
+    },
+    {
+      "epoch": 4.44359464627151,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00012609668995458573,
+      "loss": 0.4451,
+      "step": 2324
+    },
+    {
+      "epoch": 4.451242829827915,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001256475326001394,
+      "loss": 0.4703,
+      "step": 2328
+    },
+    {
+      "epoch": 4.458891013384322,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00012519859939746504,
+      "loss": 0.4032,
+      "step": 2332
+    },
+    {
+      "epoch": 4.466539196940727,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00012474989447875886,
+      "loss": 0.4324,
+      "step": 2336
+    },
+    {
+      "epoch": 4.474187380497132,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.0001243014219741158,
+      "loss": 0.4671,
+      "step": 2340
+    },
+    {
+      "epoch": 4.4818355640535374,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00012385318601149158,
+      "loss": 0.4463,
+      "step": 2344
+    },
+    {
+      "epoch": 4.489483747609943,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00012340519071666467,
+      "loss": 0.4448,
+      "step": 2348
+    },
+    {
+      "epoch": 4.497131931166348,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.0001229574402131982,
+      "loss": 0.4345,
+      "step": 2352
+    },
+    {
+      "epoch": 4.504780114722753,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00012250993862240227,
+      "loss": 0.433,
+      "step": 2356
+    },
+    {
+      "epoch": 4.512428298279159,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00012206269006329593,
+      "loss": 0.4293,
+      "step": 2360
+    },
+    {
+      "epoch": 4.520076481835564,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00012161569865256896,
+      "loss": 0.4413,
+      "step": 2364
+    },
+    {
+      "epoch": 4.527724665391969,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00012116896850454446,
+      "loss": 0.4446,
+      "step": 2368
+    },
+    {
+      "epoch": 4.5353728489483744,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00012072250373114057,
+      "loss": 0.4642,
+      "step": 2372
+    },
+    {
+      "epoch": 4.54302103250478,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00012027630844183288,
+      "loss": 0.4825,
+      "step": 2376
+    },
+    {
+      "epoch": 4.550669216061186,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00011983038674361658,
+      "loss": 0.4303,
+      "step": 2380
+    },
+    {
+      "epoch": 4.558317399617591,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00011938474274096844,
+      "loss": 0.4013,
+      "step": 2384
+    },
+    {
+      "epoch": 4.5659655831739965,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00011893938053580933,
+      "loss": 0.4183,
+      "step": 2388
+    },
+    {
+      "epoch": 4.573613766730402,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00011849430422746624,
+      "loss": 0.4345,
+      "step": 2392
+    },
+    {
+      "epoch": 4.581261950286807,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00011804951791263466,
+      "loss": 0.4253,
+      "step": 2396
+    },
+    {
+      "epoch": 4.588910133843212,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00011760502568534081,
+      "loss": 0.473,
+      "step": 2400
+    },
+    {
+      "epoch": 4.596558317399618,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00011716083163690405,
+      "loss": 0.451,
+      "step": 2404
+    },
+    {
+      "epoch": 4.604206500956023,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00011671693985589913,
+      "loss": 0.4522,
+      "step": 2408
+    },
+    {
+      "epoch": 4.611854684512428,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00011627335442811846,
+      "loss": 0.4193,
+      "step": 2412
+    },
+    {
+      "epoch": 4.6195028680688335,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00011583007943653494,
+      "loss": 0.4616,
+      "step": 2416
+    },
+    {
+      "epoch": 4.627151051625239,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00011538711896126369,
+      "loss": 0.4549,
+      "step": 2420
+    },
+    {
+      "epoch": 4.634799235181644,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00011494447707952514,
+      "loss": 0.4119,
+      "step": 2424
+    },
+    {
+      "epoch": 4.642447418738049,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0001145021578656071,
+      "loss": 0.4203,
+      "step": 2428
+    },
+    {
+      "epoch": 4.650095602294455,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00011406016539082747,
+      "loss": 0.4199,
+      "step": 2432
+    },
+    {
+      "epoch": 4.657743785850861,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00011361850372349667,
+      "loss": 0.4791,
+      "step": 2436
+    },
+    {
+      "epoch": 4.665391969407266,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00011317717692888012,
+      "loss": 0.3904,
+      "step": 2440
+    },
+    {
+      "epoch": 4.673040152963671,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00011273618906916107,
+      "loss": 0.413,
+      "step": 2444
+    },
+    {
+      "epoch": 4.680688336520077,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00011229554420340289,
+      "loss": 0.4078,
+      "step": 2448
+    },
+    {
+      "epoch": 4.688336520076482,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00011185524638751195,
+      "loss": 0.481,
+      "step": 2452
+    },
+    {
+      "epoch": 4.695984703632887,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.0001114152996742003,
+      "loss": 0.4649,
+      "step": 2456
+    },
+    {
+      "epoch": 4.7036328871892925,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00011097570811294803,
+      "loss": 0.4758,
+      "step": 2460
+    },
+    {
+      "epoch": 4.711281070745698,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00011053647574996648,
+      "loss": 0.3909,
+      "step": 2464
+    },
+    {
+      "epoch": 4.718929254302103,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001100976066281606,
+      "loss": 0.3929,
+      "step": 2468
+    },
+    {
+      "epoch": 4.726577437858508,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00010965910478709206,
+      "loss": 0.4572,
+      "step": 2472
+    },
+    {
+      "epoch": 4.734225621414914,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00010922097426294166,
+      "loss": 0.422,
+      "step": 2476
+    },
+    {
+      "epoch": 4.741873804971319,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00010878321908847259,
+      "loss": 0.4397,
+      "step": 2480
+    },
+    {
+      "epoch": 4.749521988527725,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00010834584329299322,
+      "loss": 0.4073,
+      "step": 2484
+    },
+    {
+      "epoch": 4.75717017208413,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00010790885090231968,
+      "loss": 0.4209,
+      "step": 2488
+    },
+    {
+      "epoch": 4.764818355640536,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00010747224593873933,
+      "loss": 0.4365,
+      "step": 2492
+    },
+    {
+      "epoch": 4.772466539196941,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00010703603242097322,
+      "loss": 0.4213,
+      "step": 2496
+    },
+    {
+      "epoch": 4.780114722753346,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00010660021436413956,
+      "loss": 0.4705,
+      "step": 2500
+    },
+    {
+      "epoch": 4.7877629063097515,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00010616479577971638,
+      "loss": 0.4171,
+      "step": 2504
+    },
+    {
+      "epoch": 4.795411089866157,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00010572978067550489,
+      "loss": 0.4357,
+      "step": 2508
+    },
+    {
+      "epoch": 4.803059273422562,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00010529517305559244,
+      "loss": 0.4225,
+      "step": 2512
+    },
+    {
+      "epoch": 4.810707456978967,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00010486097692031566,
+      "loss": 0.4569,
+      "step": 2516
+    },
+    {
+      "epoch": 4.818355640535373,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00010442719626622374,
+      "loss": 0.4801,
+      "step": 2520
+    },
+    {
+      "epoch": 4.826003824091778,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001039938350860415,
+      "loss": 0.4476,
+      "step": 2524
+    },
+    {
+      "epoch": 4.833652007648183,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00010356089736863282,
+      "loss": 0.4016,
+      "step": 2528
+    },
+    {
+      "epoch": 4.8413001912045885,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0001031283870989638,
+      "loss": 0.4467,
+      "step": 2532
+    },
+    {
+      "epoch": 4.848948374760994,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00010269630825806597,
+      "loss": 0.4236,
+      "step": 2536
+    },
+    {
+      "epoch": 4.8565965583174,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00010226466482300006,
+      "loss": 0.426,
+      "step": 2540
+    },
+    {
+      "epoch": 4.864244741873805,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00010183346076681882,
+      "loss": 0.4452,
+      "step": 2544
+    },
+    {
+      "epoch": 4.871892925430211,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00010140270005853098,
+      "loss": 0.4182,
+      "step": 2548
+    },
+    {
+      "epoch": 4.879541108986616,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00010097238666306427,
+      "loss": 0.4035,
+      "step": 2552
+    },
+    {
+      "epoch": 4.887189292543021,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00010054252454122934,
+      "loss": 0.4187,
+      "step": 2556
+    },
+    {
+      "epoch": 4.894837476099426,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.000100113117649683,
+      "loss": 0.4494,
+      "step": 2560
+    },
+    {
+      "epoch": 4.902485659655832,
+      "grad_norm": 0.361328125,
+      "learning_rate": 9.968416994089189e-05,
+      "loss": 0.4461,
+      "step": 2564
+    },
+    {
+      "epoch": 4.910133843212237,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.925568536309619e-05,
+      "loss": 0.4589,
+      "step": 2568
+    },
+    {
+      "epoch": 4.917782026768642,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.88276678602731e-05,
+      "loss": 0.4329,
+      "step": 2572
+    },
+    {
+      "epoch": 4.925430210325048,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.840012137210072e-05,
+      "loss": 0.4048,
+      "step": 2576
+    },
+    {
+      "epoch": 4.933078393881453,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.797304983392164e-05,
+      "loss": 0.4689,
+      "step": 2580
+    },
+    {
+      "epoch": 4.940726577437858,
+      "grad_norm": 0.37890625,
+      "learning_rate": 9.75464571767068e-05,
+      "loss": 0.4538,
+      "step": 2584
+    },
+    {
+      "epoch": 4.948374760994264,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.712034732701942e-05,
+      "loss": 0.4251,
+      "step": 2588
+    },
+    {
+      "epoch": 4.95602294455067,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.669472420697845e-05,
+      "loss": 0.3808,
+      "step": 2592
+    },
+    {
+      "epoch": 4.963671128107075,
+      "grad_norm": 0.33984375,
+      "learning_rate": 9.626959173422306e-05,
+      "loss": 0.4249,
+      "step": 2596
+    },
+    {
+      "epoch": 4.97131931166348,
+      "grad_norm": 0.36328125,
+      "learning_rate": 9.5844953821876e-05,
+      "loss": 0.4491,
+      "step": 2600
+    },
+    {
+      "epoch": 4.9789674952198855,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.542081437850801e-05,
+      "loss": 0.3934,
+      "step": 2604
+    },
+    {
+      "epoch": 4.986615678776291,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.49971773081017e-05,
+      "loss": 0.4589,
+      "step": 2608
+    },
+    {
+      "epoch": 4.994263862332696,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.457404651001546e-05,
+      "loss": 0.4418,
+      "step": 2612
+    },
+    {
+      "epoch": 5.001912045889101,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.415142587894786e-05,
+      "loss": 0.3326,
+      "step": 2616
+    },
+    {
+      "epoch": 5.009560229445507,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.372931930490147e-05,
+      "loss": 0.3716,
+      "step": 2620
+    },
+    {
+      "epoch": 5.017208413001912,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.330773067314747e-05,
+      "loss": 0.4067,
+      "step": 2624
+    },
+    {
+      "epoch": 5.024856596558317,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.28866638641894e-05,
+      "loss": 0.3533,
+      "step": 2628
+    },
+    {
+      "epoch": 5.0325047801147225,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.246612275372786e-05,
+      "loss": 0.3446,
+      "step": 2632
+    },
+    {
+      "epoch": 5.040152963671128,
+      "grad_norm": 0.353515625,
+      "learning_rate": 9.204611121262466e-05,
+      "loss": 0.391,
+      "step": 2636
+    },
+    {
+      "epoch": 5.047801147227533,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.16266331068671e-05,
+      "loss": 0.3821,
+      "step": 2640
+    },
+    {
+      "epoch": 5.055449330783939,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.120769229753262e-05,
+      "loss": 0.3813,
+      "step": 2644
+    },
+    {
+      "epoch": 5.0630975143403445,
+      "grad_norm": 0.361328125,
+      "learning_rate": 9.078929264075293e-05,
+      "loss": 0.3493,
+      "step": 2648
+    },
+    {
+      "epoch": 5.07074569789675,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.0371437987679e-05,
+      "loss": 0.3346,
+      "step": 2652
+    },
+    {
+      "epoch": 5.078393881453155,
+      "grad_norm": 0.35546875,
+      "learning_rate": 8.995413218444502e-05,
+      "loss": 0.3753,
+      "step": 2656
+    },
+    {
+      "epoch": 5.08604206500956,
+      "grad_norm": 0.33984375,
+      "learning_rate": 8.953737907213346e-05,
+      "loss": 0.3749,
+      "step": 2660
+    },
+    {
+      "epoch": 5.093690248565966,
+      "grad_norm": 0.3359375,
+      "learning_rate": 8.912118248673966e-05,
+      "loss": 0.3673,
+      "step": 2664
+    },
+    {
+      "epoch": 5.101338432122371,
+      "grad_norm": 0.34765625,
+      "learning_rate": 8.870554625913619e-05,
+      "loss": 0.3591,
+      "step": 2668
+    },
+    {
+      "epoch": 5.108986615678776,
+      "grad_norm": 0.369140625,
+      "learning_rate": 8.8290474215038e-05,
+      "loss": 0.387,
+      "step": 2672
+    },
+    {
+      "epoch": 5.1166347992351815,
+      "grad_norm": 0.359375,
+      "learning_rate": 8.787597017496687e-05,
+      "loss": 0.3774,
+      "step": 2676
+    },
+    {
+      "epoch": 5.124282982791587,
+      "grad_norm": 0.330078125,
+      "learning_rate": 8.74620379542166e-05,
+      "loss": 0.3711,
+      "step": 2680
+    },
+    {
+      "epoch": 5.131931166347992,
+      "grad_norm": 0.33203125,
+      "learning_rate": 8.704868136281742e-05,
+      "loss": 0.3733,
+      "step": 2684
+    },
+    {
+      "epoch": 5.139579349904397,
+      "grad_norm": 0.36328125,
+      "learning_rate": 8.663590420550145e-05,
+      "loss": 0.3483,
+      "step": 2688
+    },
+    {
+      "epoch": 5.147227533460803,
+      "grad_norm": 0.369140625,
+      "learning_rate": 8.622371028166743e-05,
+      "loss": 0.3773,
+      "step": 2692
+    },
+    {
+      "epoch": 5.154875717017209,
+      "grad_norm": 0.3671875,
+      "learning_rate": 8.581210338534538e-05,
+      "loss": 0.3921,
+      "step": 2696
+    },
+    {
+      "epoch": 5.162523900573614,
+      "grad_norm": 0.365234375,
+      "learning_rate": 8.540108730516248e-05,
+      "loss": 0.355,
+      "step": 2700
+    },
+    {
+      "epoch": 5.170172084130019,
+      "grad_norm": 0.34375,
+      "learning_rate": 8.499066582430748e-05,
+      "loss": 0.3805,
+      "step": 2704
+    },
+    {
+      "epoch": 5.177820267686425,
+      "grad_norm": 0.333984375,
+      "learning_rate": 8.45808427204962e-05,
+      "loss": 0.3575,
+      "step": 2708
+    },
+    {
+      "epoch": 5.18546845124283,
+      "grad_norm": 0.373046875,
+      "learning_rate": 8.417162176593686e-05,
+      "loss": 0.3948,
+      "step": 2712
+    },
+    {
+      "epoch": 5.193116634799235,
+      "grad_norm": 0.345703125,
+      "learning_rate": 8.376300672729504e-05,
+      "loss": 0.3883,
+      "step": 2716
+    },
+    {
+      "epoch": 5.2007648183556405,
+      "grad_norm": 0.35546875,
+      "learning_rate": 8.335500136565919e-05,
+      "loss": 0.4049,
+      "step": 2720
+    },
+    {
+      "epoch": 5.208413001912046,
+      "grad_norm": 0.376953125,
+      "learning_rate": 8.294760943650605e-05,
+      "loss": 0.3689,
+      "step": 2724
+    },
+    {
+      "epoch": 5.216061185468451,
+      "grad_norm": 0.384765625,
+      "learning_rate": 8.254083468966612e-05,
+      "loss": 0.3568,
+      "step": 2728
+    },
+    {
+      "epoch": 5.223709369024856,
+      "grad_norm": 0.361328125,
+      "learning_rate": 8.213468086928891e-05,
+      "loss": 0.3961,
+      "step": 2732
+    },
+    {
+      "epoch": 5.231357552581262,
+      "grad_norm": 0.3515625,
+      "learning_rate": 8.172915171380863e-05,
+      "loss": 0.3587,
+      "step": 2736
+    },
+    {
+      "epoch": 5.239005736137667,
+      "grad_norm": 0.357421875,
+      "learning_rate": 8.132425095590999e-05,
+      "loss": 0.407,
+      "step": 2740
+    },
+    {
+      "epoch": 5.246653919694072,
+      "grad_norm": 0.369140625,
+      "learning_rate": 8.091998232249325e-05,
+      "loss": 0.4085,
+      "step": 2744
+    },
+    {
+      "epoch": 5.254302103250478,
+      "grad_norm": 0.3671875,
+      "learning_rate": 8.051634953464069e-05,
+      "loss": 0.366,
+      "step": 2748
+    },
+    {
+      "epoch": 5.261950286806884,
+      "grad_norm": 0.341796875,
+      "learning_rate": 8.011335630758169e-05,
+      "loss": 0.3572,
+      "step": 2752
+    },
+    {
+      "epoch": 5.269598470363289,
+      "grad_norm": 0.392578125,
+      "learning_rate": 7.971100635065894e-05,
+      "loss": 0.4136,
+      "step": 2756
+    },
+    {
+      "epoch": 5.277246653919694,
+      "grad_norm": 0.3671875,
+      "learning_rate": 7.930930336729406e-05,
+      "loss": 0.3311,
+      "step": 2760
+    },
+    {
+      "epoch": 5.2848948374761,
+      "grad_norm": 0.357421875,
+      "learning_rate": 7.890825105495376e-05,
+      "loss": 0.351,
+      "step": 2764
+    },
+    {
+      "epoch": 5.292543021032505,
+      "grad_norm": 0.341796875,
+      "learning_rate": 7.850785310511555e-05,
+      "loss": 0.3403,
+      "step": 2768
+    },
+    {
+      "epoch": 5.30019120458891,
+      "grad_norm": 0.3359375,
+      "learning_rate": 7.810811320323386e-05,
+      "loss": 0.2974,
+      "step": 2772
+    },
+    {
+      "epoch": 5.307839388145315,
+      "grad_norm": 0.36328125,
+      "learning_rate": 7.770903502870625e-05,
+      "loss": 0.3635,
+      "step": 2776
+    },
+    {
+      "epoch": 5.315487571701721,
+      "grad_norm": 0.341796875,
+      "learning_rate": 7.731062225483933e-05,
+      "loss": 0.3669,
+      "step": 2780
+    },
+    {
+      "epoch": 5.323135755258126,
+      "grad_norm": 0.330078125,
+      "learning_rate": 7.6912878548815e-05,
+      "loss": 0.3823,
+      "step": 2784
+    },
+    {
+      "epoch": 5.330783938814531,
+      "grad_norm": 0.37890625,
+      "learning_rate": 7.651580757165691e-05,
+      "loss": 0.3644,
+      "step": 2788
+    },
+    {
+      "epoch": 5.338432122370937,
+      "grad_norm": 0.361328125,
+      "learning_rate": 7.611941297819643e-05,
+      "loss": 0.3505,
+      "step": 2792
+    },
+    {
+      "epoch": 5.346080305927342,
+      "grad_norm": 0.361328125,
+      "learning_rate": 7.572369841703924e-05,
+      "loss": 0.3691,
+      "step": 2796
+    },
+    {
+      "epoch": 5.353728489483748,
+      "grad_norm": 0.357421875,
+      "learning_rate": 7.532866753053159e-05,
+      "loss": 0.3229,
+      "step": 2800
+    },
+    {
+      "epoch": 5.361376673040153,
+      "grad_norm": 0.39453125,
+      "learning_rate": 7.493432395472711e-05,
+      "loss": 0.3849,
+      "step": 2804
+    },
+    {
+      "epoch": 5.369024856596559,
+      "grad_norm": 0.375,
+      "learning_rate": 7.454067131935269e-05,
+      "loss": 0.4105,
+      "step": 2808
+    },
+    {
+      "epoch": 5.376673040152964,
+      "grad_norm": 0.369140625,
+      "learning_rate": 7.414771324777579e-05,
+      "loss": 0.3686,
+      "step": 2812
+    },
+    {
+      "epoch": 5.384321223709369,
+      "grad_norm": 0.357421875,
+      "learning_rate": 7.375545335697085e-05,
+      "loss": 0.3293,
+      "step": 2816
+    },
+    {
+      "epoch": 5.3919694072657744,
+      "grad_norm": 0.3828125,
+      "learning_rate": 7.336389525748548e-05,
+      "loss": 0.3892,
+      "step": 2820
+    },
+    {
+      "epoch": 5.39961759082218,
+      "grad_norm": 0.3984375,
+      "learning_rate": 7.29730425534081e-05,
+      "loss": 0.3562,
+      "step": 2824
+    },
+    {
+      "epoch": 5.407265774378585,
+      "grad_norm": 0.376953125,
+      "learning_rate": 7.258289884233417e-05,
+      "loss": 0.363,
+      "step": 2828
+    },
+    {
+      "epoch": 5.41491395793499,
+      "grad_norm": 0.35546875,
+      "learning_rate": 7.21934677153332e-05,
+      "loss": 0.3517,
+      "step": 2832
+    },
+    {
+      "epoch": 5.422562141491396,
+      "grad_norm": 0.353515625,
+      "learning_rate": 7.180475275691573e-05,
+      "loss": 0.39,
+      "step": 2836
+    },
+    {
+      "epoch": 5.430210325047801,
+      "grad_norm": 0.376953125,
+      "learning_rate": 7.141675754500049e-05,
+      "loss": 0.3898,
+      "step": 2840
+    },
+    {
+      "epoch": 5.437858508604206,
+      "grad_norm": 0.396484375,
+      "learning_rate": 7.102948565088116e-05,
+      "loss": 0.4363,
+      "step": 2844
+    },
+    {
+      "epoch": 5.4455066921606115,
+      "grad_norm": 0.373046875,
+      "learning_rate": 7.064294063919368e-05,
+      "loss": 0.3494,
+      "step": 2848
+    },
+    {
+      "epoch": 5.453154875717018,
+      "grad_norm": 0.353515625,
+      "learning_rate": 7.025712606788362e-05,
+      "loss": 0.3457,
+      "step": 2852
+    },
+    {
+      "epoch": 5.460803059273423,
+      "grad_norm": 0.369140625,
+      "learning_rate": 6.987204548817278e-05,
+      "loss": 0.3279,
+      "step": 2856
+    },
+    {
+      "epoch": 5.468451242829828,
+      "grad_norm": 0.353515625,
+      "learning_rate": 6.948770244452737e-05,
+      "loss": 0.3591,
+      "step": 2860
+    },
+    {
+      "epoch": 5.4760994263862335,
+      "grad_norm": 0.365234375,
+      "learning_rate": 6.910410047462495e-05,
+      "loss": 0.3488,
+      "step": 2864
+    },
+    {
+      "epoch": 5.483747609942639,
+      "grad_norm": 0.3671875,
+      "learning_rate": 6.87212431093215e-05,
+      "loss": 0.3466,
+      "step": 2868
+    },
+    {
+      "epoch": 5.491395793499044,
+      "grad_norm": 0.34765625,
+      "learning_rate": 6.833913387261973e-05,
+      "loss": 0.4094,
+      "step": 2872
+    },
+    {
+      "epoch": 5.499043977055449,
+      "grad_norm": 0.375,
+      "learning_rate": 6.795777628163599e-05,
+      "loss": 0.3803,
+      "step": 2876
+    },
+    {
+      "epoch": 5.506692160611855,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.757717384656817e-05,
+      "loss": 0.3832,
+      "step": 2880
+    },
+    {
+      "epoch": 5.51434034416826,
+      "grad_norm": 0.37109375,
+      "learning_rate": 6.719733007066331e-05,
+      "loss": 0.3575,
+      "step": 2884
+    },
+    {
+      "epoch": 5.521988527724665,
+      "grad_norm": 0.357421875,
+      "learning_rate": 6.68182484501855e-05,
+      "loss": 0.3435,
+      "step": 2888
+    },
+    {
+      "epoch": 5.5296367112810705,
+      "grad_norm": 0.359375,
+      "learning_rate": 6.643993247438347e-05,
+      "loss": 0.3674,
+      "step": 2892
+    },
+    {
+      "epoch": 5.537284894837476,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.606238562545859e-05,
+      "loss": 0.3507,
+      "step": 2896
+    },
+    {
+      "epoch": 5.544933078393882,
+      "grad_norm": 0.357421875,
+      "learning_rate": 6.568561137853296e-05,
+      "loss": 0.3555,
+      "step": 2900
+    },
+    {
+      "epoch": 5.552581261950287,
+      "grad_norm": 0.37890625,
+      "learning_rate": 6.530961320161712e-05,
+      "loss": 0.3786,
+      "step": 2904
+    },
+    {
+      "epoch": 5.5602294455066925,
+      "grad_norm": 0.373046875,
+      "learning_rate": 6.493439455557835e-05,
+      "loss": 0.4043,
+      "step": 2908
+    },
+    {
+      "epoch": 5.567877629063098,
+      "grad_norm": 0.3671875,
+      "learning_rate": 6.455995889410873e-05,
+      "loss": 0.4086,
+      "step": 2912
+    },
+    {
+      "epoch": 5.575525812619503,
+      "grad_norm": 0.365234375,
+      "learning_rate": 6.418630966369348e-05,
+      "loss": 0.366,
+      "step": 2916
+    },
+    {
+      "epoch": 5.583173996175908,
+      "grad_norm": 0.341796875,
+      "learning_rate": 6.381345030357899e-05,
+      "loss": 0.3801,
+      "step": 2920
+    },
+    {
+      "epoch": 5.590822179732314,
+      "grad_norm": 0.3984375,
+      "learning_rate": 6.344138424574134e-05,
+      "loss": 0.3985,
+      "step": 2924
+    },
+    {
+      "epoch": 5.598470363288719,
+      "grad_norm": 0.390625,
+      "learning_rate": 6.307011491485484e-05,
+      "loss": 0.4266,
+      "step": 2928
+    },
+    {
+      "epoch": 5.606118546845124,
+      "grad_norm": 0.359375,
+      "learning_rate": 6.269964572826001e-05,
+      "loss": 0.3317,
+      "step": 2932
+    },
+    {
+      "epoch": 5.6137667304015295,
+      "grad_norm": 0.384765625,
+      "learning_rate": 6.232998009593275e-05,
+      "loss": 0.3663,
+      "step": 2936
+    },
+    {
+      "epoch": 5.621414913957935,
+      "grad_norm": 0.392578125,
+      "learning_rate": 6.196112142045268e-05,
+      "loss": 0.3804,
+      "step": 2940
+    },
+    {
+      "epoch": 5.62906309751434,
+      "grad_norm": 0.380859375,
+      "learning_rate": 6.159307309697149e-05,
+      "loss": 0.3782,
+      "step": 2944
+    },
+    {
+      "epoch": 5.636711281070745,
+      "grad_norm": 0.400390625,
+      "learning_rate": 6.122583851318233e-05,
+      "loss": 0.4047,
+      "step": 2948
+    },
+    {
+      "epoch": 5.644359464627151,
+      "grad_norm": 0.34765625,
+      "learning_rate": 6.085942104928815e-05,
+      "loss": 0.3875,
+      "step": 2952
+    },
+    {
+      "epoch": 5.652007648183556,
+      "grad_norm": 0.388671875,
+      "learning_rate": 6.049382407797076e-05,
+      "loss": 0.3649,
+      "step": 2956
+    },
+    {
+      "epoch": 5.659655831739962,
+      "grad_norm": 0.3359375,
+      "learning_rate": 6.012905096435968e-05,
+      "loss": 0.3813,
+      "step": 2960
+    },
+    {
+      "epoch": 5.667304015296367,
+      "grad_norm": 0.396484375,
+      "learning_rate": 5.976510506600146e-05,
+      "loss": 0.3955,
+      "step": 2964
+    },
+    {
+      "epoch": 5.674952198852773,
+      "grad_norm": 0.38671875,
+      "learning_rate": 5.9401989732828384e-05,
+      "loss": 0.3574,
+      "step": 2968
+    },
+    {
+      "epoch": 5.682600382409178,
+      "grad_norm": 0.349609375,
+      "learning_rate": 5.9039708307127816e-05,
+      "loss": 0.333,
+      "step": 2972
+    },
+    {
+      "epoch": 5.690248565965583,
+      "grad_norm": 0.361328125,
+      "learning_rate": 5.8678264123511626e-05,
+      "loss": 0.3856,
+      "step": 2976
+    },
+    {
+      "epoch": 5.6978967495219885,
+      "grad_norm": 0.345703125,
+      "learning_rate": 5.8317660508885e-05,
+      "loss": 0.3931,
+      "step": 2980
+    },
+    {
+      "epoch": 5.705544933078394,
+      "grad_norm": 0.375,
+      "learning_rate": 5.795790078241641e-05,
+      "loss": 0.3364,
+      "step": 2984
+    },
+    {
+      "epoch": 5.713193116634799,
+      "grad_norm": 0.349609375,
+      "learning_rate": 5.7598988255506644e-05,
+      "loss": 0.3825,
+      "step": 2988
+    },
+    {
+      "epoch": 5.720841300191204,
+      "grad_norm": 0.376953125,
+      "learning_rate": 5.724092623175841e-05,
+      "loss": 0.3654,
+      "step": 2992
+    },
+    {
+      "epoch": 5.72848948374761,
+      "grad_norm": 0.373046875,
+      "learning_rate": 5.6883718006946146e-05,
+      "loss": 0.4175,
+      "step": 2996
+    },
+    {
+      "epoch": 5.736137667304015,
+      "grad_norm": 0.392578125,
+      "learning_rate": 5.652736686898537e-05,
+      "loss": 0.369,
+      "step": 3000
+    },
+    {
+      "epoch": 5.743785850860421,
+      "grad_norm": 0.3671875,
+      "learning_rate": 5.61718760979026e-05,
+      "loss": 0.3961,
+      "step": 3004
+    },
+    {
+      "epoch": 5.751434034416826,
+      "grad_norm": 0.353515625,
+      "learning_rate": 5.5817248965805096e-05,
+      "loss": 0.3622,
+      "step": 3008
+    },
+    {
+      "epoch": 5.759082217973232,
+      "grad_norm": 0.36328125,
+      "learning_rate": 5.546348873685089e-05,
+      "loss": 0.3369,
+      "step": 3012
+    },
+    {
+      "epoch": 5.766730401529637,
+      "grad_norm": 0.357421875,
+      "learning_rate": 5.51105986672185e-05,
+      "loss": 0.3314,
+      "step": 3016
+    },
+    {
+      "epoch": 5.774378585086042,
+      "grad_norm": 0.337890625,
+      "learning_rate": 5.475858200507708e-05,
+      "loss": 0.3212,
+      "step": 3020
+    },
+    {
+      "epoch": 5.782026768642448,
+      "grad_norm": 0.375,
+      "learning_rate": 5.440744199055663e-05,
+      "loss": 0.3978,
+      "step": 3024
+    },
+    {
+      "epoch": 5.789674952198853,
+      "grad_norm": 0.37890625,
+      "learning_rate": 5.4057181855718e-05,
+      "loss": 0.3938,
+      "step": 3028
+    },
+    {
+      "epoch": 5.797323135755258,
+      "grad_norm": 0.369140625,
+      "learning_rate": 5.370780482452317e-05,
+      "loss": 0.3613,
+      "step": 3032
+    },
+    {
+      "epoch": 5.804971319311663,
+      "grad_norm": 0.365234375,
+      "learning_rate": 5.335931411280559e-05,
+      "loss": 0.3717,
+      "step": 3036
+    },
+    {
+      "epoch": 5.812619502868069,
+      "grad_norm": 0.345703125,
+      "learning_rate": 5.3011712928240787e-05,
+      "loss": 0.3572,
+      "step": 3040
+    },
+    {
+      "epoch": 5.820267686424474,
+      "grad_norm": 0.359375,
+      "learning_rate": 5.2665004470316456e-05,
+      "loss": 0.3993,
+      "step": 3044
+    },
+    {
+      "epoch": 5.827915869980879,
+      "grad_norm": 0.365234375,
+      "learning_rate": 5.231919193030324e-05,
+      "loss": 0.3708,
+      "step": 3048
+    },
+    {
+      "epoch": 5.835564053537285,
+      "grad_norm": 0.37890625,
+      "learning_rate": 5.197427849122549e-05,
+      "loss": 0.3877,
+      "step": 3052
+    },
+    {
+      "epoch": 5.84321223709369,
+      "grad_norm": 0.34375,
+      "learning_rate": 5.1630267327831494e-05,
+      "loss": 0.3747,
+      "step": 3056
+    },
+    {
+      "epoch": 5.850860420650095,
+      "grad_norm": 0.373046875,
+      "learning_rate": 5.128716160656489e-05,
+      "loss": 0.4003,
+      "step": 3060
+    },
+    {
+      "epoch": 5.858508604206501,
+      "grad_norm": 0.3515625,
+      "learning_rate": 5.0944964485534975e-05,
+      "loss": 0.3654,
+      "step": 3064
+    },
+    {
+      "epoch": 5.866156787762907,
+      "grad_norm": 0.39453125,
+      "learning_rate": 5.06036791144879e-05,
+      "loss": 0.3837,
+      "step": 3068
+    },
+    {
+      "epoch": 5.873804971319312,
+      "grad_norm": 0.40625,
+      "learning_rate": 5.0263308634777745e-05,
+      "loss": 0.368,
+      "step": 3072
+    },
+    {
+      "epoch": 5.881453154875717,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.992385617933734e-05,
+      "loss": 0.3975,
+      "step": 3076
+    },
+    {
+      "epoch": 5.8891013384321225,
+      "grad_norm": 0.365234375,
+      "learning_rate": 4.958532487264968e-05,
+      "loss": 0.3837,
+      "step": 3080
+    },
+    {
+      "epoch": 5.896749521988528,
+      "grad_norm": 0.333984375,
+      "learning_rate": 4.924771783071895e-05,
+      "loss": 0.3744,
+      "step": 3084
+    },
+    {
+      "epoch": 5.904397705544933,
+      "grad_norm": 0.357421875,
+      "learning_rate": 4.8911038161042136e-05,
+      "loss": 0.3439,
+      "step": 3088
+    },
+    {
+      "epoch": 5.912045889101338,
+      "grad_norm": 0.37890625,
+      "learning_rate": 4.857528896258012e-05,
+      "loss": 0.4041,
+      "step": 3092
+    },
+    {
+      "epoch": 5.919694072657744,
+      "grad_norm": 0.375,
+      "learning_rate": 4.824047332572924e-05,
+      "loss": 0.3753,
+      "step": 3096
+    },
+    {
+      "epoch": 5.927342256214149,
+      "grad_norm": 0.359375,
+      "learning_rate": 4.7906594332293116e-05,
+      "loss": 0.3822,
+      "step": 3100
+    },
+    {
+      "epoch": 5.934990439770554,
+      "grad_norm": 0.3515625,
+      "learning_rate": 4.75736550554537e-05,
+      "loss": 0.3535,
+      "step": 3104
+    },
+    {
+      "epoch": 5.9426386233269595,
+      "grad_norm": 0.35546875,
+      "learning_rate": 4.724165855974367e-05,
+      "loss": 0.377,
+      "step": 3108
+    },
+    {
+      "epoch": 5.950286806883366,
+      "grad_norm": 0.365234375,
+      "learning_rate": 4.6910607901017715e-05,
+      "loss": 0.359,
+      "step": 3112
+    },
+    {
+      "epoch": 5.957934990439771,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.65805061264246e-05,
+      "loss": 0.3415,
+      "step": 3116
+    },
+    {
+      "epoch": 5.965583173996176,
+      "grad_norm": 0.390625,
+      "learning_rate": 4.625135627437922e-05,
+      "loss": 0.3968,
+      "step": 3120
+    },
+    {
+      "epoch": 5.9732313575525815,
+      "grad_norm": 0.357421875,
+      "learning_rate": 4.592316137453439e-05,
+      "loss": 0.3983,
+      "step": 3124
+    },
+    {
+      "epoch": 5.980879541108987,
+      "grad_norm": 0.357421875,
+      "learning_rate": 4.559592444775315e-05,
+      "loss": 0.3466,
+      "step": 3128
+    },
+    {
+      "epoch": 5.988527724665392,
+      "grad_norm": 0.3515625,
+      "learning_rate": 4.5269648506080816e-05,
+      "loss": 0.3442,
+      "step": 3132
+    },
+    {
+      "epoch": 5.996175908221797,
+      "grad_norm": 0.375,
+      "learning_rate": 4.4944336552717514e-05,
+      "loss": 0.3768,
+      "step": 3136
+    },
+    {
+      "epoch": 6.003824091778203,
+      "grad_norm": 0.3359375,
+      "learning_rate": 4.461999158199019e-05,
+      "loss": 0.3533,
+      "step": 3140
+    },
+    {
+      "epoch": 6.011472275334608,
+      "grad_norm": 0.34765625,
+      "learning_rate": 4.429661657932523e-05,
+      "loss": 0.3247,
+      "step": 3144
+    },
+    {
+      "epoch": 6.019120458891013,
+      "grad_norm": 0.34375,
+      "learning_rate": 4.397421452122114e-05,
+      "loss": 0.329,
+      "step": 3148
+    },
+    {
+      "epoch": 6.0267686424474185,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.3652788375220787e-05,
+      "loss": 0.3547,
+      "step": 3152
+    },
+    {
+      "epoch": 6.034416826003824,
+      "grad_norm": 0.361328125,
+      "learning_rate": 4.333234109988434e-05,
+      "loss": 0.4026,
+      "step": 3156
+    },
+    {
+      "epoch": 6.042065009560229,
+      "grad_norm": 0.3828125,
+      "learning_rate": 4.3012875644761955e-05,
+      "loss": 0.3534,
+      "step": 3160
+    },
+    {
+      "epoch": 6.049713193116634,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.269439495036678e-05,
+      "loss": 0.3035,
+      "step": 3164
+    },
+    {
+      "epoch": 6.0573613766730405,
+      "grad_norm": 0.365234375,
+      "learning_rate": 4.2376901948147465e-05,
+      "loss": 0.3553,
+      "step": 3168
+    },
+    {
+      "epoch": 6.065009560229446,
+      "grad_norm": 0.369140625,
+      "learning_rate": 4.206039956046176e-05,
+      "loss": 0.3602,
+      "step": 3172
+    },
+    {
+      "epoch": 6.072657743785851,
+      "grad_norm": 0.349609375,
+      "learning_rate": 4.174489070054927e-05,
+      "loss": 0.3606,
+      "step": 3176
+    },
+    {
+      "epoch": 6.080305927342256,
+      "grad_norm": 0.349609375,
+      "learning_rate": 4.143037827250447e-05,
+      "loss": 0.3499,
+      "step": 3180
+    },
+    {
+      "epoch": 6.087954110898662,
+      "grad_norm": 0.349609375,
+      "learning_rate": 4.1116865171250496e-05,
+      "loss": 0.3447,
+      "step": 3184
+    },
+    {
+      "epoch": 6.095602294455067,
+      "grad_norm": 0.357421875,
+      "learning_rate": 4.0804354282512016e-05,
+      "loss": 0.3541,
+      "step": 3188
+    },
+    {
+      "epoch": 6.103250478011472,
+      "grad_norm": 0.3515625,
+      "learning_rate": 4.049284848278886e-05,
+      "loss": 0.3638,
+      "step": 3192
+    },
+    {
+      "epoch": 6.1108986615678775,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.01823506393297e-05,
+      "loss": 0.362,
+      "step": 3196
+    },
+    {
+      "epoch": 6.118546845124283,
+      "grad_norm": 0.3828125,
+      "learning_rate": 3.987286361010531e-05,
+      "loss": 0.3367,
+      "step": 3200
+    },
+    {
+      "epoch": 6.126195028680688,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.9564390243782516e-05,
+      "loss": 0.3249,
+      "step": 3204
+    },
+    {
+      "epoch": 6.133843212237093,
+      "grad_norm": 0.396484375,
+      "learning_rate": 3.925693337969788e-05,
+      "loss": 0.3328,
+      "step": 3208
+    },
+    {
+      "epoch": 6.141491395793499,
+      "grad_norm": 0.35546875,
+      "learning_rate": 3.895049584783173e-05,
+      "loss": 0.3746,
+      "step": 3212
+    },
+    {
+      "epoch": 6.149139579349905,
+      "grad_norm": 0.34765625,
+      "learning_rate": 3.8645080468781676e-05,
+      "loss": 0.3702,
+      "step": 3216
+    },
+    {
+      "epoch": 6.15678776290631,
+      "grad_norm": 0.337890625,
+      "learning_rate": 3.834069005373724e-05,
+      "loss": 0.3136,
+      "step": 3220
+    },
+    {
+      "epoch": 6.164435946462715,
+      "grad_norm": 0.357421875,
+      "learning_rate": 3.8037327404453634e-05,
+      "loss": 0.3542,
+      "step": 3224
+    },
+    {
+      "epoch": 6.172084130019121,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.77349953132258e-05,
+      "loss": 0.3572,
+      "step": 3228
+    },
+    {
+      "epoch": 6.179732313575526,
+      "grad_norm": 0.37109375,
+      "learning_rate": 3.7433696562863215e-05,
+      "loss": 0.3526,
+      "step": 3232
+    },
+    {
+      "epoch": 6.187380497131931,
+      "grad_norm": 0.365234375,
+      "learning_rate": 3.7133433926663805e-05,
+      "loss": 0.331,
+      "step": 3236
+    },
+    {
+      "epoch": 6.195028680688337,
+      "grad_norm": 0.376953125,
+      "learning_rate": 3.6834210168388674e-05,
+      "loss": 0.3489,
+      "step": 3240
+    },
+    {
+      "epoch": 6.202676864244742,
+      "grad_norm": 0.359375,
+      "learning_rate": 3.653602804223656e-05,
+      "loss": 0.3329,
+      "step": 3244
+    },
+    {
+      "epoch": 6.210325047801147,
+      "grad_norm": 0.34765625,
+      "learning_rate": 3.623889029281861e-05,
+      "loss": 0.2646,
+      "step": 3248
+    },
+    {
+      "epoch": 6.217973231357552,
+      "grad_norm": 0.373046875,
+      "learning_rate": 3.5942799655132925e-05,
+      "loss": 0.3547,
+      "step": 3252
+    },
+    {
+      "epoch": 6.225621414913958,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.56477588545395e-05,
+      "loss": 0.3547,
+      "step": 3256
+    },
+    {
+      "epoch": 6.233269598470363,
+      "grad_norm": 0.341796875,
+      "learning_rate": 3.535377060673524e-05,
+      "loss": 0.3205,
+      "step": 3260
+    },
+    {
+      "epoch": 6.240917782026768,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.506083761772871e-05,
+      "loss": 0.3647,
+      "step": 3264
+    },
+    {
+      "epoch": 6.248565965583174,
+      "grad_norm": 0.33984375,
+      "learning_rate": 3.476896258381537e-05,
+      "loss": 0.3138,
+      "step": 3268
+    },
+    {
+      "epoch": 6.25621414913958,
+      "grad_norm": 0.3828125,
+      "learning_rate": 3.447814819155291e-05,
+      "loss": 0.3071,
+      "step": 3272
+    },
+    {
+      "epoch": 6.263862332695985,
+      "grad_norm": 0.349609375,
+      "learning_rate": 3.418839711773623e-05,
+      "loss": 0.3353,
+      "step": 3276
+    },
+    {
+      "epoch": 6.27151051625239,
+      "grad_norm": 0.384765625,
+      "learning_rate": 3.389971202937295e-05,
+      "loss": 0.3497,
+      "step": 3280
+    },
+    {
+      "epoch": 6.279158699808796,
+      "grad_norm": 0.3359375,
+      "learning_rate": 3.361209558365883e-05,
+      "loss": 0.3125,
+      "step": 3284
+    },
+    {
+      "epoch": 6.286806883365201,
+      "grad_norm": 0.365234375,
+      "learning_rate": 3.332555042795349e-05,
+      "loss": 0.3575,
+      "step": 3288
+    },
+    {
+      "epoch": 6.294455066921606,
+      "grad_norm": 0.359375,
+      "learning_rate": 3.304007919975563e-05,
+      "loss": 0.3554,
+      "step": 3292
+    },
+    {
+      "epoch": 6.3021032504780115,
+      "grad_norm": 0.34375,
+      "learning_rate": 3.2755684526679196e-05,
+      "loss": 0.3427,
+      "step": 3296
+    },
+    {
+      "epoch": 6.309751434034417,
+      "grad_norm": 0.35546875,
+      "learning_rate": 3.247236902642905e-05,
+      "loss": 0.3648,
+      "step": 3300
+    },
+    {
+      "epoch": 6.317399617590822,
+      "grad_norm": 0.375,
+      "learning_rate": 3.219013530677655e-05,
+      "loss": 0.3269,
+      "step": 3304
+    },
+    {
+      "epoch": 6.325047801147227,
+      "grad_norm": 0.349609375,
+      "learning_rate": 3.190898596553615e-05,
+      "loss": 0.3503,
+      "step": 3308
+    },
+    {
+      "epoch": 6.332695984703633,
+      "grad_norm": 0.33984375,
+      "learning_rate": 3.162892359054098e-05,
+      "loss": 0.3239,
+      "step": 3312
+    },
+    {
+      "epoch": 6.340344168260038,
+      "grad_norm": 0.376953125,
+      "learning_rate": 3.1349950759619255e-05,
+      "loss": 0.3253,
+      "step": 3316
+    },
+    {
+      "epoch": 6.347992351816444,
+      "grad_norm": 0.361328125,
+      "learning_rate": 3.107207004057046e-05,
+      "loss": 0.3642,
+      "step": 3320
+    },
+    {
+      "epoch": 6.355640535372849,
+      "grad_norm": 0.349609375,
+      "learning_rate": 3.079528399114189e-05,
+      "loss": 0.3058,
+      "step": 3324
+    },
+    {
+      "epoch": 6.363288718929255,
+      "grad_norm": 0.375,
+      "learning_rate": 3.0519595159004853e-05,
+      "loss": 0.3705,
+      "step": 3328
+    },
+    {
+      "epoch": 6.37093690248566,
+      "grad_norm": 0.345703125,
+      "learning_rate": 3.0245006081731367e-05,
+      "loss": 0.3444,
+      "step": 3332
+    },
+    {
+      "epoch": 6.378585086042065,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.9971519286770883e-05,
+      "loss": 0.3181,
+      "step": 3336
+    },
+    {
+      "epoch": 6.3862332695984705,
+      "grad_norm": 0.359375,
+      "learning_rate": 2.969913729142668e-05,
+      "loss": 0.3519,
+      "step": 3340
+    },
+    {
+      "epoch": 6.393881453154876,
+      "grad_norm": 0.34765625,
+      "learning_rate": 2.9427862602833165e-05,
+      "loss": 0.3261,
+      "step": 3344
+    },
+    {
+      "epoch": 6.401529636711281,
+      "grad_norm": 0.373046875,
+      "learning_rate": 2.915769771793256e-05,
+      "loss": 0.3554,
+      "step": 3348
+    },
+    {
+      "epoch": 6.409177820267686,
+      "grad_norm": 0.337890625,
+      "learning_rate": 2.8888645123451694e-05,
+      "loss": 0.3119,
+      "step": 3352
+    },
+    {
+      "epoch": 6.416826003824092,
+      "grad_norm": 0.357421875,
+      "learning_rate": 2.862070729587959e-05,
+      "loss": 0.3576,
+      "step": 3356
+    },
+    {
+      "epoch": 6.424474187380497,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.8353886701444312e-05,
+      "loss": 0.3464,
+      "step": 3360
+    },
+    {
+      "epoch": 6.432122370936902,
+      "grad_norm": 0.345703125,
+      "learning_rate": 2.808818579609037e-05,
+      "loss": 0.3362,
+      "step": 3364
+    },
+    {
+      "epoch": 6.4397705544933075,
+      "grad_norm": 0.359375,
+      "learning_rate": 2.7823607025456103e-05,
+      "loss": 0.3556,
+      "step": 3368
+    },
+    {
+      "epoch": 6.447418738049713,
+      "grad_norm": 0.353515625,
+      "learning_rate": 2.7560152824851285e-05,
+      "loss": 0.2955,
+      "step": 3372
+    },
+    {
+      "epoch": 6.455066921606119,
+      "grad_norm": 0.349609375,
+      "learning_rate": 2.7297825619234515e-05,
+      "loss": 0.3064,
+      "step": 3376
+    },
+    {
+      "epoch": 6.462715105162524,
+      "grad_norm": 0.373046875,
+      "learning_rate": 2.7036627823190994e-05,
+      "loss": 0.3553,
+      "step": 3380
+    },
+    {
+      "epoch": 6.4703632887189295,
+      "grad_norm": 0.34375,
+      "learning_rate": 2.6776561840910367e-05,
+      "loss": 0.3141,
+      "step": 3384
+    },
+    {
+      "epoch": 6.478011472275335,
+      "grad_norm": 0.3515625,
+      "learning_rate": 2.6517630066164448e-05,
+      "loss": 0.3746,
+      "step": 3388
+    },
+    {
+      "epoch": 6.48565965583174,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.6259834882285302e-05,
+      "loss": 0.3467,
+      "step": 3392
+    },
+    {
+      "epoch": 6.493307839388145,
+      "grad_norm": 0.359375,
+      "learning_rate": 2.6003178662143214e-05,
+      "loss": 0.3257,
+      "step": 3396
+    },
+    {
+      "epoch": 6.500956022944551,
+      "grad_norm": 0.36328125,
+      "learning_rate": 2.574766376812502e-05,
+      "loss": 0.3231,
+      "step": 3400
+    },
+    {
+      "epoch": 6.508604206500956,
+      "grad_norm": 0.333984375,
+      "learning_rate": 2.5493292552112128e-05,
+      "loss": 0.3102,
+      "step": 3404
+    },
+    {
+      "epoch": 6.516252390057361,
+      "grad_norm": 0.35546875,
+      "learning_rate": 2.5240067355458978e-05,
+      "loss": 0.33,
+      "step": 3408
+    },
+    {
+      "epoch": 6.5239005736137665,
+      "grad_norm": 0.373046875,
+      "learning_rate": 2.4987990508971667e-05,
+      "loss": 0.3529,
+      "step": 3412
+    },
+    {
+      "epoch": 6.531548757170172,
+      "grad_norm": 0.375,
+      "learning_rate": 2.4737064332886055e-05,
+      "loss": 0.3285,
+      "step": 3416
+    },
+    {
+      "epoch": 6.539196940726577,
+      "grad_norm": 0.361328125,
+      "learning_rate": 2.4487291136846894e-05,
+      "loss": 0.3835,
+      "step": 3420
+    },
+    {
+      "epoch": 6.546845124282983,
+      "grad_norm": 0.353515625,
+      "learning_rate": 2.4238673219886384e-05,
+      "loss": 0.3476,
+      "step": 3424
+    },
+    {
+      "epoch": 6.5544933078393885,
+      "grad_norm": 0.369140625,
+      "learning_rate": 2.399121287040275e-05,
+      "loss": 0.3209,
+      "step": 3428
+    },
+    {
+      "epoch": 6.562141491395794,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.3744912366139644e-05,
+      "loss": 0.3498,
+      "step": 3432
+    },
+    {
+      "epoch": 6.569789674952199,
+      "grad_norm": 0.359375,
+      "learning_rate": 2.3499773974164825e-05,
+      "loss": 0.348,
+      "step": 3436
+    },
+    {
+      "epoch": 6.577437858508604,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.325579995084946e-05,
+      "loss": 0.3783,
+      "step": 3440
+    },
+    {
+      "epoch": 6.58508604206501,
+      "grad_norm": 0.359375,
+      "learning_rate": 2.3012992541847254e-05,
+      "loss": 0.3855,
+      "step": 3444
+    },
+    {
+      "epoch": 6.592734225621415,
+      "grad_norm": 0.34765625,
+      "learning_rate": 2.277135398207393e-05,
+      "loss": 0.3163,
+      "step": 3448
+    },
+    {
+      "epoch": 6.60038240917782,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.2530886495686506e-05,
+      "loss": 0.3601,
+      "step": 3452
+    },
+    {
+      "epoch": 6.6080305927342256,
+      "grad_norm": 0.396484375,
+      "learning_rate": 2.229159229606281e-05,
+      "loss": 0.3425,
+      "step": 3456
+    },
+    {
+      "epoch": 6.615678776290631,
+      "grad_norm": 0.38671875,
+      "learning_rate": 2.2053473585781377e-05,
+      "loss": 0.324,
+      "step": 3460
+    },
+    {
+      "epoch": 6.623326959847036,
+      "grad_norm": 0.37890625,
+      "learning_rate": 2.181653255660072e-05,
+      "loss": 0.3386,
+      "step": 3464
+    },
+    {
+      "epoch": 6.630975143403441,
+      "grad_norm": 0.359375,
+      "learning_rate": 2.1580771389439612e-05,
+      "loss": 0.3434,
+      "step": 3468
+    },
+    {
+      "epoch": 6.638623326959847,
+      "grad_norm": 0.34765625,
+      "learning_rate": 2.1346192254356737e-05,
+      "loss": 0.331,
+      "step": 3472
+    },
+    {
+      "epoch": 6.646271510516252,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.1112797310530716e-05,
+      "loss": 0.3483,
+      "step": 3476
+    },
+    {
+      "epoch": 6.653919694072657,
+      "grad_norm": 0.376953125,
+      "learning_rate": 2.08805887062405e-05,
+      "loss": 0.3241,
+      "step": 3480
+    },
+    {
+      "epoch": 6.661567877629063,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.0649568578845205e-05,
+      "loss": 0.3269,
+      "step": 3484
+    },
+    {
+      "epoch": 6.669216061185469,
+      "grad_norm": 0.359375,
+      "learning_rate": 2.0419739054764743e-05,
+      "loss": 0.3636,
+      "step": 3488
+    },
+    {
+      "epoch": 6.676864244741874,
+      "grad_norm": 0.34375,
+      "learning_rate": 2.019110224946008e-05,
+      "loss": 0.3549,
+      "step": 3492
+    },
+    {
+      "epoch": 6.684512428298279,
+      "grad_norm": 0.3515625,
+      "learning_rate": 1.9963660267413913e-05,
+      "loss": 0.3497,
+      "step": 3496
+    },
+    {
+      "epoch": 6.692160611854685,
+      "grad_norm": 0.36328125,
+      "learning_rate": 1.9737415202111144e-05,
+      "loss": 0.3639,
+      "step": 3500
+    },
+    {
+      "epoch": 6.69980879541109,
+      "grad_norm": 0.353515625,
+      "learning_rate": 1.9512369136019663e-05,
+      "loss": 0.3487,
+      "step": 3504
+    },
+    {
+      "epoch": 6.707456978967495,
+      "grad_norm": 0.333984375,
+      "learning_rate": 1.9288524140571286e-05,
+      "loss": 0.3196,
+      "step": 3508
+    },
+    {
+      "epoch": 6.7151051625239,
+      "grad_norm": 0.361328125,
+      "learning_rate": 1.906588227614254e-05,
+      "loss": 0.3398,
+      "step": 3512
+    },
+    {
+      "epoch": 6.722753346080306,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.8844445592035767e-05,
+      "loss": 0.37,
+      "step": 3516
+    },
+    {
+      "epoch": 6.730401529636711,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.8624216126460183e-05,
+      "loss": 0.3466,
+      "step": 3520
+    },
+    {
+      "epoch": 6.738049713193116,
+      "grad_norm": 0.353515625,
+      "learning_rate": 1.8405195906513347e-05,
+      "loss": 0.3638,
+      "step": 3524
+    },
+    {
+      "epoch": 6.7456978967495225,
+      "grad_norm": 0.369140625,
+      "learning_rate": 1.8187386948162203e-05,
+      "loss": 0.3894,
+      "step": 3528
+    },
+    {
+      "epoch": 6.753346080305928,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.797079125622469e-05,
+      "loss": 0.3109,
+      "step": 3532
+    },
+    {
+      "epoch": 6.760994263862333,
+      "grad_norm": 0.345703125,
+      "learning_rate": 1.7755410824351363e-05,
+      "loss": 0.3191,
+      "step": 3536
+    },
+    {
+      "epoch": 6.768642447418738,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.7541247635006756e-05,
+      "loss": 0.3548,
+      "step": 3540
+    },
+    {
+      "epoch": 6.776290630975144,
+      "grad_norm": 0.369140625,
+      "learning_rate": 1.7328303659451477e-05,
+      "loss": 0.3518,
+      "step": 3544
+    },
+    {
+      "epoch": 6.783938814531549,
+      "grad_norm": 0.359375,
+      "learning_rate": 1.7116580857723872e-05,
+      "loss": 0.367,
+      "step": 3548
+    },
+    {
+      "epoch": 6.791586998087954,
+      "grad_norm": 0.3515625,
+      "learning_rate": 1.6906081178621917e-05,
+      "loss": 0.3502,
+      "step": 3552
+    },
+    {
+      "epoch": 6.7992351816443595,
+      "grad_norm": 0.376953125,
+      "learning_rate": 1.6696806559685553e-05,
+      "loss": 0.4081,
+      "step": 3556
+    },
+    {
+      "epoch": 6.806883365200765,
+      "grad_norm": 0.38671875,
+      "learning_rate": 1.648875892717857e-05,
+      "loss": 0.3557,
+      "step": 3560
+    },
+    {
+      "epoch": 6.81453154875717,
+      "grad_norm": 0.337890625,
+      "learning_rate": 1.628194019607099e-05,
+      "loss": 0.3765,
+      "step": 3564
+    },
+    {
+      "epoch": 6.822179732313575,
+      "grad_norm": 0.3515625,
+      "learning_rate": 1.6076352270021435e-05,
+      "loss": 0.3261,
+      "step": 3568
+    },
+    {
+      "epoch": 6.829827915869981,
+      "grad_norm": 0.40625,
+      "learning_rate": 1.587199704135973e-05,
+      "loss": 0.3734,
+      "step": 3572
+    },
+    {
+      "epoch": 6.837476099426386,
+      "grad_norm": 0.38671875,
+      "learning_rate": 1.5668876391069107e-05,
+      "loss": 0.3523,
+      "step": 3576
+    },
+    {
+      "epoch": 6.845124282982791,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.5466992188769394e-05,
+      "loss": 0.3465,
+      "step": 3580
+    },
+    {
+      "epoch": 6.8527724665391965,
+      "grad_norm": 0.357421875,
+      "learning_rate": 1.5266346292699522e-05,
+      "loss": 0.3506,
+      "step": 3584
+    },
+    {
+      "epoch": 6.860420650095603,
+      "grad_norm": 0.359375,
+      "learning_rate": 1.5066940549700285e-05,
+      "loss": 0.3112,
+      "step": 3588
+    },
+    {
+      "epoch": 6.868068833652008,
+      "grad_norm": 0.40234375,
+      "learning_rate": 1.4868776795197712e-05,
+      "loss": 0.3271,
+      "step": 3592
+    },
+    {
+      "epoch": 6.875717017208413,
+      "grad_norm": 0.35546875,
+      "learning_rate": 1.4671856853185876e-05,
+      "loss": 0.371,
+      "step": 3596
+    },
+    {
+      "epoch": 6.8833652007648185,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.4476182536210207e-05,
+      "loss": 0.3448,
+      "step": 3600
+    },
+    {
+      "epoch": 6.891013384321224,
+      "grad_norm": 0.353515625,
+      "learning_rate": 1.4281755645350873e-05,
+      "loss": 0.3605,
+      "step": 3604
+    },
+    {
+      "epoch": 6.898661567877629,
+      "grad_norm": 0.35546875,
+      "learning_rate": 1.4088577970206044e-05,
+      "loss": 0.3385,
+      "step": 3608
+    },
+    {
+      "epoch": 6.906309751434034,
+      "grad_norm": 0.35546875,
+      "learning_rate": 1.3896651288875572e-05,
+      "loss": 0.3285,
+      "step": 3612
+    },
+    {
+      "epoch": 6.91395793499044,
+      "grad_norm": 0.34765625,
+      "learning_rate": 1.3705977367944498e-05,
+      "loss": 0.3316,
+      "step": 3616
+    },
+    {
+      "epoch": 6.921606118546845,
+      "grad_norm": 0.35546875,
+      "learning_rate": 1.3516557962466978e-05,
+      "loss": 0.3636,
+      "step": 3620
+    },
+    {
+      "epoch": 6.92925430210325,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.3328394815949884e-05,
+      "loss": 0.3649,
+      "step": 3624
+    },
+    {
+      "epoch": 6.9369024856596555,
+      "grad_norm": 0.35546875,
+      "learning_rate": 1.3141489660336902e-05,
+      "loss": 0.3521,
+      "step": 3628
+    },
+    {
+      "epoch": 6.944550669216062,
+      "grad_norm": 0.34765625,
+      "learning_rate": 1.295584421599265e-05,
+      "loss": 0.3177,
+      "step": 3632
+    },
+    {
+      "epoch": 6.952198852772467,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.2771460191686656e-05,
+      "loss": 0.3148,
+      "step": 3636
+    },
+    {
+      "epoch": 6.959847036328872,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.2588339284577815e-05,
+      "loss": 0.3324,
+      "step": 3640
+    },
+    {
+      "epoch": 6.9674952198852775,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.240648318019859e-05,
+      "loss": 0.3729,
+      "step": 3644
+    },
+    {
+      "epoch": 6.975143403441683,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.2225893552439742e-05,
+      "loss": 0.3616,
+      "step": 3648
+    },
+    {
+      "epoch": 6.982791586998088,
+      "grad_norm": 0.345703125,
+      "learning_rate": 1.2046572063534587e-05,
+      "loss": 0.3261,
+      "step": 3652
+    },
+    {
+      "epoch": 6.990439770554493,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.1868520364044049e-05,
+      "loss": 0.3542,
+      "step": 3656
+    },
+    {
+      "epoch": 6.998087954110899,
+      "grad_norm": 0.33203125,
+      "learning_rate": 1.1691740092841228e-05,
+      "loss": 0.3238,
+      "step": 3660
+    },
+    {
+      "epoch": 7.005736137667304,
+      "grad_norm": 0.365234375,
+      "learning_rate": 1.151623287709636e-05,
+      "loss": 0.4048,
+      "step": 3664
+    },
+    {
+      "epoch": 7.013384321223709,
+      "grad_norm": 0.353515625,
+      "learning_rate": 1.1342000332261963e-05,
+      "loss": 0.3609,
+      "step": 3668
+    },
+    {
+      "epoch": 7.0210325047801145,
+      "grad_norm": 0.326171875,
+      "learning_rate": 1.1169044062057797e-05,
+      "loss": 0.3145,
+      "step": 3672
+    },
+    {
+      "epoch": 7.02868068833652,
+      "grad_norm": 0.35546875,
+      "learning_rate": 1.0997365658456164e-05,
+      "loss": 0.3618,
+      "step": 3676
+    },
+    {
+      "epoch": 7.036328871892925,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.082696670166736e-05,
+      "loss": 0.3298,
+      "step": 3680
+    },
+    {
+      "epoch": 7.04397705544933,
+      "grad_norm": 0.353515625,
+      "learning_rate": 1.0657848760124954e-05,
+      "loss": 0.3249,
+      "step": 3684
+    },
+    {
+      "epoch": 7.051625239005737,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.0490013390471474e-05,
+      "loss": 0.294,
+      "step": 3688
+    },
+    {
+      "epoch": 7.059273422562142,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.0323462137543998e-05,
+      "loss": 0.3264,
+      "step": 3692
+    },
+    {
+      "epoch": 7.066921606118547,
+      "grad_norm": 0.3359375,
+      "learning_rate": 1.015819653436012e-05,
+      "loss": 0.3164,
+      "step": 3696
+    },
+    {
+      "epoch": 7.074569789674952,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.994218102103468e-06,
+      "loss": 0.3317,
+      "step": 3700
+    },
+    {
+      "epoch": 7.082217973231358,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.831528350110119e-06,
+      "loss": 0.3168,
+      "step": 3704
+    },
+    {
+      "epoch": 7.089866156787763,
+      "grad_norm": 0.39453125,
+      "learning_rate": 9.670128775854513e-06,
+      "loss": 0.3901,
+      "step": 3708
+    },
+    {
+      "epoch": 7.097514340344168,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.51002086493553e-06,
+      "loss": 0.3297,
+      "step": 3712
+    },
+    {
+      "epoch": 7.105162523900574,
+      "grad_norm": 0.353515625,
+      "learning_rate": 9.351206091063107e-06,
+      "loss": 0.3484,
+      "step": 3716
+    },
+    {
+      "epoch": 7.112810707456979,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.193685916044469e-06,
+      "loss": 0.3255,
+      "step": 3720
+    },
+    {
+      "epoch": 7.120458891013384,
+      "grad_norm": 0.361328125,
+      "learning_rate": 9.03746178977074e-06,
+      "loss": 0.3837,
+      "step": 3724
+    },
+    {
+      "epoch": 7.128107074569789,
+      "grad_norm": 0.3515625,
+      "learning_rate": 8.882535150203567e-06,
+      "loss": 0.3689,
+      "step": 3728
+    },
+    {
+      "epoch": 7.135755258126195,
+      "grad_norm": 0.330078125,
+      "learning_rate": 8.728907423361991e-06,
+      "loss": 0.3209,
+      "step": 3732
+    },
+    {
+      "epoch": 7.1434034416826,
+      "grad_norm": 0.326171875,
+      "learning_rate": 8.576580023309126e-06,
+      "loss": 0.3297,
+      "step": 3736
+    },
+    {
+      "epoch": 7.151051625239006,
+      "grad_norm": 0.35546875,
+      "learning_rate": 8.425554352139313e-06,
+      "loss": 0.3432,
+      "step": 3740
+    },
+    {
+      "epoch": 7.1586998087954115,
+      "grad_norm": 0.33203125,
+      "learning_rate": 8.275831799965194e-06,
+      "loss": 0.3514,
+      "step": 3744
+    },
+    {
+      "epoch": 7.166347992351817,
+      "grad_norm": 0.349609375,
+      "learning_rate": 8.127413744904804e-06,
+      "loss": 0.3326,
+      "step": 3748
+    },
+    {
+      "epoch": 7.173996175908222,
+      "grad_norm": 0.373046875,
+      "learning_rate": 7.980301553068985e-06,
+      "loss": 0.3558,
+      "step": 3752
+    },
+    {
+      "epoch": 7.181644359464627,
+      "grad_norm": 0.353515625,
+      "learning_rate": 7.83449657854886e-06,
+      "loss": 0.3504,
+      "step": 3756
+    },
+    {
+      "epoch": 7.189292543021033,
+      "grad_norm": 0.34375,
+      "learning_rate": 7.690000163403177e-06,
+      "loss": 0.3136,
+      "step": 3760
+    },
+    {
+      "epoch": 7.196940726577438,
+      "grad_norm": 0.33984375,
+      "learning_rate": 7.546813637646182e-06,
+      "loss": 0.3292,
+      "step": 3764
+    },
+    {
+      "epoch": 7.204588910133843,
+      "grad_norm": 0.365234375,
+      "learning_rate": 7.404938319235171e-06,
+      "loss": 0.327,
+      "step": 3768
+    },
+    {
+      "epoch": 7.2122370936902485,
+      "grad_norm": 0.36328125,
+      "learning_rate": 7.264375514058607e-06,
+      "loss": 0.3526,
+      "step": 3772
+    },
+    {
+      "epoch": 7.219885277246654,
+      "grad_norm": 0.333984375,
+      "learning_rate": 7.125126515923752e-06,
+      "loss": 0.3506,
+      "step": 3776
+    },
+    {
+      "epoch": 7.227533460803059,
+      "grad_norm": 0.349609375,
+      "learning_rate": 6.987192606545156e-06,
+      "loss": 0.3355,
+      "step": 3780
+    },
+    {
+      "epoch": 7.235181644359464,
+      "grad_norm": 0.353515625,
+      "learning_rate": 6.850575055532553e-06,
+      "loss": 0.3484,
+      "step": 3784
+    },
+    {
+      "epoch": 7.24282982791587,
+      "grad_norm": 0.36328125,
+      "learning_rate": 6.715275120379271e-06,
+      "loss": 0.366,
+      "step": 3788
+    },
+    {
+      "epoch": 7.250478011472275,
+      "grad_norm": 0.33984375,
+      "learning_rate": 6.581294046450753e-06,
+      "loss": 0.322,
+      "step": 3792
+    },
+    {
+      "epoch": 7.258126195028681,
+      "grad_norm": 0.322265625,
+      "learning_rate": 6.448633066972953e-06,
+      "loss": 0.2989,
+      "step": 3796
+    },
+    {
+      "epoch": 7.265774378585086,
+      "grad_norm": 0.353515625,
+      "learning_rate": 6.317293403021029e-06,
+      "loss": 0.3377,
+      "step": 3800
+    },
+    {
+      "epoch": 7.273422562141492,
+      "grad_norm": 0.326171875,
+      "learning_rate": 6.187276263508167e-06,
+      "loss": 0.3001,
+      "step": 3804
+    },
+    {
+      "epoch": 7.281070745697897,
+      "grad_norm": 0.337890625,
+      "learning_rate": 6.0585828451743925e-06,
+      "loss": 0.3335,
+      "step": 3808
+    },
+    {
+      "epoch": 7.288718929254302,
+      "grad_norm": 0.34375,
+      "learning_rate": 5.93121433257554e-06,
+      "loss": 0.3458,
+      "step": 3812
+    },
+    {
+      "epoch": 7.2963671128107075,
+      "grad_norm": 0.3671875,
+      "learning_rate": 5.805171898072369e-06,
+      "loss": 0.2941,
+      "step": 3816
+    },
+    {
+      "epoch": 7.304015296367113,
+      "grad_norm": 0.359375,
+      "learning_rate": 5.680456701819885e-06,
+      "loss": 0.3888,
+      "step": 3820
+    },
+    {
+      "epoch": 7.311663479923518,
+      "grad_norm": 0.36328125,
+      "learning_rate": 5.5570698917563994e-06,
+      "loss": 0.3663,
+      "step": 3824
+    },
+    {
+      "epoch": 7.319311663479923,
+      "grad_norm": 0.3515625,
+      "learning_rate": 5.435012603593219e-06,
+      "loss": 0.292,
+      "step": 3828
+    },
+    {
+      "epoch": 7.326959847036329,
+      "grad_norm": 0.373046875,
+      "learning_rate": 5.3142859608041265e-06,
+      "loss": 0.3628,
+      "step": 3832
+    },
+    {
+      "epoch": 7.334608030592734,
+      "grad_norm": 0.345703125,
+      "learning_rate": 5.194891074614899e-06,
+      "loss": 0.3377,
+      "step": 3836
+    },
+    {
+      "epoch": 7.342256214149139,
+      "grad_norm": 0.353515625,
+      "learning_rate": 5.076829043993253e-06,
+      "loss": 0.3322,
+      "step": 3840
+    },
+    {
+      "epoch": 7.349904397705545,
+      "grad_norm": 0.37890625,
+      "learning_rate": 4.960100955638685e-06,
+      "loss": 0.3607,
+      "step": 3844
+    },
+    {
+      "epoch": 7.357552581261951,
+      "grad_norm": 0.345703125,
+      "learning_rate": 4.844707883972398e-06,
+      "loss": 0.3303,
+      "step": 3848
+    },
+    {
+      "epoch": 7.365200764818356,
+      "grad_norm": 0.337890625,
+      "learning_rate": 4.730650891127457e-06,
+      "loss": 0.3327,
+      "step": 3852
+    },
+    {
+      "epoch": 7.372848948374761,
+      "grad_norm": 0.357421875,
+      "learning_rate": 4.617931026939115e-06,
+      "loss": 0.3766,
+      "step": 3856
+    },
+    {
+      "epoch": 7.3804971319311665,
+      "grad_norm": 0.33203125,
+      "learning_rate": 4.506549328934916e-06,
+      "loss": 0.2884,
+      "step": 3860
+    },
+    {
+      "epoch": 7.388145315487572,
+      "grad_norm": 0.353515625,
+      "learning_rate": 4.39650682232538e-06,
+      "loss": 0.3036,
+      "step": 3864
+    },
+    {
+      "epoch": 7.395793499043977,
+      "grad_norm": 0.37890625,
+      "learning_rate": 4.287804519994431e-06,
+      "loss": 0.3398,
+      "step": 3868
+    },
+    {
+      "epoch": 7.403441682600382,
+      "grad_norm": 0.353515625,
+      "learning_rate": 4.180443422490115e-06,
+      "loss": 0.3276,
+      "step": 3872
+    },
+    {
+      "epoch": 7.411089866156788,
+      "grad_norm": 0.35546875,
+      "learning_rate": 4.074424518015384e-06,
+      "loss": 0.3319,
+      "step": 3876
+    },
+    {
+      "epoch": 7.418738049713193,
+      "grad_norm": 0.390625,
+      "learning_rate": 3.969748782418991e-06,
+      "loss": 0.3925,
+      "step": 3880
+    },
+    {
+      "epoch": 7.426386233269598,
+      "grad_norm": 0.365234375,
+      "learning_rate": 3.8664171791865765e-06,
+      "loss": 0.3556,
+      "step": 3884
+    },
+    {
+      "epoch": 7.4340344168260035,
+      "grad_norm": 0.384765625,
+      "learning_rate": 3.764430659431661e-06,
+      "loss": 0.324,
+      "step": 3888
+    },
+    {
+      "epoch": 7.441682600382409,
+      "grad_norm": 0.37109375,
+      "learning_rate": 3.6637901618870203e-06,
+      "loss": 0.3443,
+      "step": 3892
+    },
+    {
+      "epoch": 7.449330783938814,
+      "grad_norm": 0.3828125,
+      "learning_rate": 3.564496612896006e-06,
+      "loss": 0.3409,
+      "step": 3896
+    },
+    {
+      "epoch": 7.45697896749522,
+      "grad_norm": 0.322265625,
+      "learning_rate": 3.4665509264039717e-06,
+      "loss": 0.3106,
+      "step": 3900
+    },
+    {
+      "epoch": 7.4646271510516256,
+      "grad_norm": 0.35546875,
+      "learning_rate": 3.3699540039499263e-06,
+      "loss": 0.3442,
+      "step": 3904
+    },
+    {
+      "epoch": 7.472275334608031,
+      "grad_norm": 0.35546875,
+      "learning_rate": 3.274706734658228e-06,
+      "loss": 0.3649,
+      "step": 3908
+    },
+    {
+      "epoch": 7.479923518164436,
+      "grad_norm": 0.330078125,
+      "learning_rate": 3.1808099952303045e-06,
+      "loss": 0.3215,
+      "step": 3912
+    },
+    {
+      "epoch": 7.487571701720841,
+      "grad_norm": 0.37109375,
+      "learning_rate": 3.0882646499367614e-06,
+      "loss": 0.2973,
+      "step": 3916
+    },
+    {
+      "epoch": 7.495219885277247,
+      "grad_norm": 0.375,
+      "learning_rate": 2.9970715506092534e-06,
+      "loss": 0.3862,
+      "step": 3920
+    },
+    {
+      "epoch": 7.502868068833652,
+      "grad_norm": 0.3515625,
+      "learning_rate": 2.9072315366327424e-06,
+      "loss": 0.3246,
+      "step": 3924
+    },
+    {
+      "epoch": 7.510516252390057,
+      "grad_norm": 0.36328125,
+      "learning_rate": 2.8187454349377193e-06,
+      "loss": 0.3823,
+      "step": 3928
+    },
+    {
+      "epoch": 7.5181644359464626,
+      "grad_norm": 0.369140625,
+      "learning_rate": 2.731614059992676e-06,
+      "loss": 0.3697,
+      "step": 3932
+    },
+    {
+      "epoch": 7.525812619502868,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.6458382137964805e-06,
+      "loss": 0.3312,
+      "step": 3936
+    },
+    {
+      "epoch": 7.533460803059273,
+      "grad_norm": 0.353515625,
+      "learning_rate": 2.5614186858711137e-06,
+      "loss": 0.3188,
+      "step": 3940
+    },
+    {
+      "epoch": 7.541108986615678,
+      "grad_norm": 0.345703125,
+      "learning_rate": 2.47835625325436e-06,
+      "loss": 0.322,
+      "step": 3944
+    },
+    {
+      "epoch": 7.548757170172085,
+      "grad_norm": 0.369140625,
+      "learning_rate": 2.3966516804925784e-06,
+      "loss": 0.3401,
+      "step": 3948
+    },
+    {
+      "epoch": 7.55640535372849,
+      "grad_norm": 0.353515625,
+      "learning_rate": 2.3163057196338096e-06,
+      "loss": 0.3388,
+      "step": 3952
+    },
+    {
+      "epoch": 7.564053537284895,
+      "grad_norm": 0.34375,
+      "learning_rate": 2.2373191102207646e-06,
+      "loss": 0.3375,
+      "step": 3956
+    },
+    {
+      "epoch": 7.5717017208413,
+      "grad_norm": 0.322265625,
+      "learning_rate": 2.1596925792839946e-06,
+      "loss": 0.3125,
+      "step": 3960
+    },
+    {
+      "epoch": 7.579349904397706,
+      "grad_norm": 0.37890625,
+      "learning_rate": 2.083426841335284e-06,
+      "loss": 0.3436,
+      "step": 3964
+    },
+    {
+      "epoch": 7.586998087954111,
+      "grad_norm": 0.369140625,
+      "learning_rate": 2.0085225983610177e-06,
+      "loss": 0.3559,
+      "step": 3968
+    },
+    {
+      "epoch": 7.594646271510516,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.9349805398156893e-06,
+      "loss": 0.3167,
+      "step": 3972
+    },
+    {
+      "epoch": 7.602294455066922,
+      "grad_norm": 0.35546875,
+      "learning_rate": 1.8628013426156386e-06,
+      "loss": 0.3654,
+      "step": 3976
+    },
+    {
+      "epoch": 7.609942638623327,
+      "grad_norm": 0.359375,
+      "learning_rate": 1.7919856711327563e-06,
+      "loss": 0.3534,
+      "step": 3980
+    },
+    {
+      "epoch": 7.617590822179732,
+      "grad_norm": 0.36328125,
+      "learning_rate": 1.722534177188406e-06,
+      "loss": 0.3139,
+      "step": 3984
+    },
+    {
+      "epoch": 7.625239005736137,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.6544475000473957e-06,
+      "loss": 0.3578,
+      "step": 3988
+    },
+    {
+      "epoch": 7.632887189292543,
+      "grad_norm": 0.357421875,
+      "learning_rate": 1.5877262664120983e-06,
+      "loss": 0.3272,
+      "step": 3992
+    },
+    {
+      "epoch": 7.640535372848948,
+      "grad_norm": 0.373046875,
+      "learning_rate": 1.522371090416724e-06,
+      "loss": 0.3578,
+      "step": 3996
+    },
+    {
+      "epoch": 7.648183556405353,
+      "grad_norm": 0.34375,
+      "learning_rate": 1.4583825736215749e-06,
+      "loss": 0.3131,
+      "step": 4000
+    }
+  ],
+  "logging_steps": 4,
+  "max_steps": 4184,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 8,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.6138957704404664e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}