diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,982 +1,5688 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.999627004848937,
+  "epoch": 2.9988810145468108,
   "eval_steps": 500,
-  "global_step": 670,
+  "global_step": 4020,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "grad_norm": 0.337890625,
-      "learning_rate": 9.477611940298507e-07,
-      "loss": 2.637,
+      "grad_norm": 0.353515625,
+      "learning_rate": 9.950248756218907e-07,
+      "loss": 2.6395,
       "step": 1
     },
     {
-      "epoch": 0.01,
-      "grad_norm": 0.349609375,
-      "learning_rate": 4.738805970149253e-06,
-      "loss": 2.6461,
+      "epoch": 0.0,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.975124378109453e-06,
+      "loss": 2.6482,
       "step": 5
     },
     {
       "epoch": 0.01,
-      "grad_norm": 0.32421875,
-      "learning_rate": 9.477611940298506e-06,
-      "loss": 2.6439,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.950248756218906e-06,
+      "loss": 2.6389,
       "step": 10
     },
     {
-      "epoch": 0.02,
-      "grad_norm": 0.35546875,
-      "learning_rate": 1.4216417910447761e-05,
-      "loss": 2.642,
+      "epoch": 0.01,
+      "grad_norm": 0.3203125,
+      "learning_rate": 1.4925373134328357e-05,
+      "loss": 2.629,
       "step": 15
     },
     {
-      "epoch": 0.03,
-      "grad_norm": 0.365234375,
-      "learning_rate": 1.895522388059701e-05,
-      "loss": 2.6207,
+      "epoch": 0.01,
+      "grad_norm": 0.328125,
+      "learning_rate": 1.990049751243781e-05,
+      "loss": 2.6301,
       "step": 20
     },
     {
-      "epoch": 0.04,
-      "grad_norm": 0.353515625,
-      "learning_rate": 2.369402985074627e-05,
-      "loss": 2.6062,
+      "epoch": 0.02,
+      "grad_norm": 0.36328125,
+      "learning_rate": 2.4875621890547266e-05,
+      "loss": 2.6048,
       "step": 25
     },
     {
-      "epoch": 0.04,
-      "grad_norm": 0.291015625,
-      "learning_rate": 2.8432835820895522e-05,
-      "loss": 2.5733,
+      "epoch": 0.02,
+      "grad_norm": 0.298828125,
+      "learning_rate": 2.9850746268656714e-05,
+      "loss": 2.5818,
       "step": 30
     },
     {
-      "epoch": 0.05,
-      "grad_norm": 0.255859375,
-      "learning_rate": 3.317164179104477e-05,
-      "loss": 2.531,
+      "epoch": 0.03,
+      "grad_norm": 0.26171875,
+      "learning_rate": 3.4825870646766175e-05,
+      "loss": 2.5372,
       "step": 35
     },
     {
-      "epoch": 0.06,
-      "grad_norm": 0.259765625,
-      "learning_rate": 3.791044776119402e-05,
-      "loss": 2.495,
+      "epoch": 0.03,
+      "grad_norm": 0.271484375,
+      "learning_rate": 3.980099502487562e-05,
+      "loss": 2.507,
       "step": 40
     },
     {
-      "epoch": 0.07,
-      "grad_norm": 0.177734375,
-      "learning_rate": 4.2649253731343286e-05,
-      "loss": 2.4618,
+      "epoch": 0.03,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 4.477611940298508e-05,
+      "loss": 2.4718,
       "step": 45
     },
     {
-      "epoch": 0.07,
-      "grad_norm": 0.1494140625,
-      "learning_rate": 4.738805970149254e-05,
-      "loss": 2.4374,
+      "epoch": 0.04,
+      "grad_norm": 0.1640625,
+      "learning_rate": 4.975124378109453e-05,
+      "loss": 2.4398,
       "step": 50
     },
     {
-      "epoch": 0.08,
-      "grad_norm": 0.1513671875,
-      "learning_rate": 5.2126865671641794e-05,
-      "loss": 2.4205,
+      "epoch": 0.04,
+      "grad_norm": 0.16015625,
+      "learning_rate": 5.472636815920398e-05,
+      "loss": 2.4118,
       "step": 55
     },
     {
-      "epoch": 0.09,
-      "grad_norm": 0.15234375,
-      "learning_rate": 5.6865671641791044e-05,
-      "loss": 2.3846,
+      "epoch": 0.04,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 5.970149253731343e-05,
+      "loss": 2.3949,
       "step": 60
     },
     {
-      "epoch": 0.1,
-      "grad_norm": 0.1298828125,
-      "learning_rate": 6.16044776119403e-05,
-      "loss": 2.3675,
+      "epoch": 0.05,
+      "grad_norm": 0.13671875,
+      "learning_rate": 6.46766169154229e-05,
+      "loss": 2.3556,
       "step": 65
     },
     {
-      "epoch": 0.1,
-      "grad_norm": 0.10888671875,
-      "learning_rate": 6.349612195786573e-05,
-      "loss": 2.3648,
+      "epoch": 0.05,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 6.965174129353235e-05,
+      "loss": 2.3644,
       "step": 70
     },
     {
-      "epoch": 0.11,
-      "grad_norm": 0.138671875,
-      "learning_rate": 6.3472426242142e-05,
-      "loss": 2.3394,
+      "epoch": 0.06,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 7.46268656716418e-05,
+      "loss": 2.3262,
       "step": 75
     },
     {
-      "epoch": 0.12,
-      "grad_norm": 0.09912109375,
-      "learning_rate": 6.342720533773385e-05,
-      "loss": 2.3241,
+      "epoch": 0.06,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 7.960199004975125e-05,
+      "loss": 2.3181,
       "step": 80
     },
     {
-      "epoch": 0.13,
-      "grad_norm": 0.083984375,
-      "learning_rate": 6.336048992919527e-05,
-      "loss": 2.3123,
+      "epoch": 0.06,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 8.45771144278607e-05,
+      "loss": 2.3195,
       "step": 85
     },
     {
-      "epoch": 0.13,
-      "grad_norm": 0.09228515625,
-      "learning_rate": 6.327232528613285e-05,
-      "loss": 2.3201,
+      "epoch": 0.07,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 8.955223880597016e-05,
+      "loss": 2.2922,
       "step": 90
     },
     {
-      "epoch": 0.14,
-      "grad_norm": 0.0908203125,
-      "learning_rate": 6.316277123248829e-05,
-      "loss": 2.2999,
+      "epoch": 0.07,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 9.452736318407961e-05,
+      "loss": 2.2996,
       "step": 95
     },
     {
-      "epoch": 0.15,
-      "grad_norm": 0.07958984375,
-      "learning_rate": 6.303190210594489e-05,
-      "loss": 2.3114,
+      "epoch": 0.07,
+      "grad_norm": 0.095703125,
+      "learning_rate": 9.950248756218906e-05,
+      "loss": 2.2923,
       "step": 100
     },
     {
-      "epoch": 0.16,
-      "grad_norm": 0.0712890625,
-      "learning_rate": 6.287980670748592e-05,
-      "loss": 2.3002,
+      "epoch": 0.08,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0001044776119402985,
+      "loss": 2.2908,
       "step": 105
     },
     {
-      "epoch": 0.16,
-      "grad_norm": 0.072265625,
-      "learning_rate": 6.270658824113884e-05,
-      "loss": 2.2732,
+      "epoch": 0.08,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.00010945273631840796,
+      "loss": 2.2892,
       "step": 110
     },
     {
-      "epoch": 0.17,
-      "grad_norm": 0.07275390625,
-      "learning_rate": 6.251236424394651e-05,
-      "loss": 2.2867,
+      "epoch": 0.09,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.00011442786069651741,
+      "loss": 2.2772,
       "step": 115
     },
     {
-      "epoch": 0.18,
-      "grad_norm": 0.0732421875,
-      "learning_rate": 6.229726650621257e-05,
-      "loss": 2.2718,
+      "epoch": 0.09,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.00011940298507462686,
+      "loss": 2.2492,
       "step": 120
     },
     {
-      "epoch": 0.19,
-      "grad_norm": 0.0751953125,
-      "learning_rate": 6.20614409820754e-05,
-      "loss": 2.274,
+      "epoch": 0.09,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0001243781094527363,
+      "loss": 2.2669,
       "step": 125
     },
     {
-      "epoch": 0.19,
-      "grad_norm": 0.072265625,
-      "learning_rate": 6.180504769047129e-05,
-      "loss": 2.2688,
+      "epoch": 0.1,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0001293532338308458,
+      "loss": 2.2486,
       "step": 130
     },
     {
-      "epoch": 0.2,
-      "grad_norm": 0.0830078125,
-      "learning_rate": 6.152826060655387e-05,
-      "loss": 2.272,
+      "epoch": 0.1,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.00013432835820895525,
+      "loss": 2.2655,
       "step": 135
     },
     {
-      "epoch": 0.21,
-      "grad_norm": 0.07861328125,
-      "learning_rate": 6.123126754364366e-05,
-      "loss": 2.2807,
+      "epoch": 0.1,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0001393034825870647,
+      "loss": 2.2695,
       "step": 140
     },
     {
-      "epoch": 0.22,
-      "grad_norm": 0.072265625,
-      "learning_rate": 6.091427002578765e-05,
-      "loss": 2.2519,
+      "epoch": 0.11,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.00014427860696517416,
+      "loss": 2.2627,
       "step": 145
     },
     {
-      "epoch": 0.22,
-      "grad_norm": 0.0732421875,
-      "learning_rate": 6.057748315101562e-05,
-      "loss": 2.2648,
+      "epoch": 0.11,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0001492537313432836,
+      "loss": 2.2453,
       "step": 150
     },
     {
-      "epoch": 0.23,
-      "grad_norm": 0.07177734375,
-      "learning_rate": 6.0221135445385774e-05,
-      "loss": 2.2739,
+      "epoch": 0.12,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.00015422885572139304,
+      "loss": 2.2462,
       "step": 155
     },
     {
-      "epoch": 0.24,
-      "grad_norm": 0.07470703125,
-      "learning_rate": 5.984546870791885e-05,
-      "loss": 2.2764,
+      "epoch": 0.12,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0001592039800995025,
+      "loss": 2.2574,
       "step": 160
     },
     {
-      "epoch": 0.25,
-      "grad_norm": 0.0751953125,
-      "learning_rate": 5.945073784652589e-05,
-      "loss": 2.2514,
+      "epoch": 0.12,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.00016417910447761195,
+      "loss": 2.2522,
       "step": 165
     },
     {
-      "epoch": 0.25,
-      "grad_norm": 0.078125,
-      "learning_rate": 5.9037210705040984e-05,
-      "loss": 2.2544,
+      "epoch": 0.13,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0001691542288557214,
+      "loss": 2.2428,
       "step": 170
     },
     {
-      "epoch": 0.26,
-      "grad_norm": 0.07470703125,
-      "learning_rate": 5.860516788147634e-05,
-      "loss": 2.2479,
+      "epoch": 0.13,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.00017412935323383086,
+      "loss": 2.2617,
       "step": 175
     },
     {
-      "epoch": 0.27,
-      "grad_norm": 0.07763671875,
-      "learning_rate": 5.815490253762313e-05,
-      "loss": 2.253,
+      "epoch": 0.13,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0001791044776119403,
+      "loss": 2.2604,
       "step": 180
     },
     {
-      "epoch": 0.28,
-      "grad_norm": 0.080078125,
-      "learning_rate": 5.7686720200127084e-05,
-      "loss": 2.2528,
+      "epoch": 0.14,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.00018407960199004977,
+      "loss": 2.2437,
       "step": 185
     },
     {
-      "epoch": 0.28,
-      "grad_norm": 0.0791015625,
-      "learning_rate": 5.7200938553174043e-05,
-      "loss": 2.253,
+      "epoch": 0.14,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.00018905472636815922,
+      "loss": 2.2489,
       "step": 190
     },
     {
-      "epoch": 0.29,
-      "grad_norm": 0.078125,
-      "learning_rate": 5.669788722292595e-05,
-      "loss": 2.2571,
+      "epoch": 0.15,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.00019402985074626867,
+      "loss": 2.2788,
       "step": 195
     },
     {
-      "epoch": 0.3,
-      "grad_norm": 0.0791015625,
-      "learning_rate": 5.617790755385372e-05,
-      "loss": 2.2666,
+      "epoch": 0.15,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.00019900497512437813,
+      "loss": 2.2438,
       "step": 200
     },
     {
-      "epoch": 0.31,
-      "grad_norm": 0.07958984375,
-      "learning_rate": 5.5641352377118605e-05,
-      "loss": 2.2601,
+      "epoch": 0.15,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.00020398009950248756,
+      "loss": 2.2603,
       "step": 205
     },
     {
-      "epoch": 0.31,
-      "grad_norm": 0.07861328125,
-      "learning_rate": 5.508858577115933e-05,
-      "loss": 2.2471,
+      "epoch": 0.16,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.000208955223880597,
+      "loss": 2.242,
       "step": 210
     },
     {
-      "epoch": 0.32,
-      "grad_norm": 0.080078125,
-      "learning_rate": 5.451998281464741e-05,
-      "loss": 2.2523,
+      "epoch": 0.16,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.00021393034825870647,
+      "loss": 2.2205,
       "step": 215
     },
     {
-      "epoch": 0.33,
-      "grad_norm": 0.07958984375,
-      "learning_rate": 5.393592933197822e-05,
-      "loss": 2.244,
+      "epoch": 0.16,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.00021890547263681592,
+      "loss": 2.2329,
       "step": 220
     },
     {
-      "epoch": 0.34,
-      "grad_norm": 0.0791015625,
-      "learning_rate": 5.333682163147071e-05,
-      "loss": 2.242,
+      "epoch": 0.17,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.00022388059701492538,
+      "loss": 2.2511,
       "step": 225
     },
     {
-      "epoch": 0.34,
-      "grad_norm": 0.0771484375,
-      "learning_rate": 5.2723066236453086e-05,
-      "loss": 2.2519,
+      "epoch": 0.17,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.00022885572139303483,
+      "loss": 2.2369,
       "step": 230
     },
     {
-      "epoch": 0.35,
-      "grad_norm": 0.08349609375,
-      "learning_rate": 5.209507960941733e-05,
-      "loss": 2.2521,
+      "epoch": 0.18,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.00023383084577114426,
+      "loss": 2.2116,
       "step": 235
     },
     {
-      "epoch": 0.36,
-      "grad_norm": 0.0830078125,
-      "learning_rate": 5.145328786942933e-05,
-      "loss": 2.2525,
+      "epoch": 0.18,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0002388059701492537,
+      "loss": 2.2502,
       "step": 240
     },
     {
-      "epoch": 0.37,
-      "grad_norm": 0.07958984375,
-      "learning_rate": 5.0798126502986685e-05,
-      "loss": 2.2423,
+      "epoch": 0.18,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.00024378109452736317,
+      "loss": 2.2344,
       "step": 245
     },
     {
-      "epoch": 0.37,
-      "grad_norm": 0.080078125,
-      "learning_rate": 5.013004006852019e-05,
-      "loss": 2.2444,
+      "epoch": 0.19,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0002487562189054726,
+      "loss": 2.2336,
       "step": 250
     },
     {
-      "epoch": 0.38,
-      "grad_norm": 0.08154296875,
-      "learning_rate": 4.944948189473962e-05,
-      "loss": 2.2562,
+      "epoch": 0.19,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0002537313432835821,
+      "loss": 2.2274,
       "step": 255
     },
     {
-      "epoch": 0.39,
-      "grad_norm": 0.08203125,
-      "learning_rate": 4.875691377302846e-05,
-      "loss": 2.2443,
+      "epoch": 0.19,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0002587064676616916,
+      "loss": 2.2365,
       "step": 260
     },
     {
-      "epoch": 0.4,
-      "grad_norm": 0.0859375,
-      "learning_rate": 4.805280564409623e-05,
-      "loss": 2.2492,
+      "epoch": 0.2,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.000263681592039801,
+      "loss": 2.2357,
       "step": 265
     },
     {
-      "epoch": 0.4,
-      "grad_norm": 0.08154296875,
-      "learning_rate": 4.7337635279101233e-05,
-      "loss": 2.2402,
+      "epoch": 0.2,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0002686567164179105,
+      "loss": 2.2314,
       "step": 270
     },
     {
-      "epoch": 0.41,
-      "grad_norm": 0.08203125,
-      "learning_rate": 4.661188795545985e-05,
-      "loss": 2.2402,
+      "epoch": 0.21,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0002736318407960199,
+      "loss": 2.2381,
       "step": 275
     },
     {
-      "epoch": 0.42,
-      "grad_norm": 0.0869140625,
-      "learning_rate": 4.5876056127562524e-05,
-      "loss": 2.2464,
+      "epoch": 0.21,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0002786069651741294,
+      "loss": 2.2481,
       "step": 280
     },
     {
-      "epoch": 0.43,
-      "grad_norm": 0.0859375,
-      "learning_rate": 4.5130639092619825e-05,
-      "loss": 2.2518,
+      "epoch": 0.21,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.00028358208955223883,
+      "loss": 2.2104,
       "step": 285
     },
     {
-      "epoch": 0.43,
-      "grad_norm": 0.083984375,
-      "learning_rate": 4.437614265186536e-05,
-      "loss": 2.2491,
+      "epoch": 0.22,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0002885572139303483,
+      "loss": 2.2203,
       "step": 290
     },
     {
-      "epoch": 0.44,
-      "grad_norm": 0.08154296875,
-      "learning_rate": 4.361307876734529e-05,
-      "loss": 2.2474,
+      "epoch": 0.22,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.00029353233830845774,
+      "loss": 2.2426,
       "step": 295
     },
     {
-      "epoch": 0.45,
-      "grad_norm": 0.083984375,
-      "learning_rate": 4.2841965214527606e-05,
-      "loss": 2.2634,
+      "epoch": 0.22,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0002985074626865672,
+      "loss": 2.2137,
       "step": 300
     },
     {
-      "epoch": 0.46,
-      "grad_norm": 0.0849609375,
-      "learning_rate": 4.206332523096655e-05,
-      "loss": 2.2433,
+      "epoch": 0.23,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.00030348258706467665,
+      "loss": 2.2393,
       "step": 305
     },
     {
-      "epoch": 0.46,
-      "grad_norm": 0.083984375,
-      "learning_rate": 4.127768716126082e-05,
-      "loss": 2.2468,
+      "epoch": 0.23,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0003084577114427861,
+      "loss": 2.2369,
       "step": 310
     },
     {
-      "epoch": 0.47,
-      "grad_norm": 0.0869140625,
-      "learning_rate": 4.0485584098546456e-05,
-      "loss": 2.2399,
+      "epoch": 0.23,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.00031343283582089556,
+      "loss": 2.2205,
       "step": 315
     },
     {
-      "epoch": 0.48,
-      "grad_norm": 0.08544921875,
-      "learning_rate": 3.968755352276755e-05,
-      "loss": 2.2431,
+      "epoch": 0.24,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.000318407960199005,
+      "loss": 2.2607,
       "step": 320
     },
     {
-      "epoch": 0.48,
-      "grad_norm": 0.08349609375,
-      "learning_rate": 3.888413693597025e-05,
-      "loss": 2.2309,
+      "epoch": 0.24,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.00032338308457711447,
+      "loss": 2.201,
       "step": 325
     },
     {
-      "epoch": 0.49,
-      "grad_norm": 0.08740234375,
-      "learning_rate": 3.8075879494867705e-05,
-      "loss": 2.2298,
+      "epoch": 0.25,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0003283582089552239,
+      "loss": 2.2304,
       "step": 330
     },
     {
-      "epoch": 0.5,
-      "grad_norm": 0.08349609375,
-      "learning_rate": 3.726332964092504e-05,
-      "loss": 2.2443,
+      "epoch": 0.25,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0003333333333333334,
+      "loss": 2.214,
       "step": 335
     },
     {
-      "epoch": 0.51,
-      "grad_norm": 0.09033203125,
-      "learning_rate": 3.644703872821547e-05,
-      "loss": 2.2481,
+      "epoch": 0.25,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0003383084577114428,
+      "loss": 2.2251,
       "step": 340
     },
     {
-      "epoch": 0.51,
-      "grad_norm": 0.08642578125,
-      "learning_rate": 3.5627560649300175e-05,
-      "loss": 2.2375,
+      "epoch": 0.26,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0003432835820895523,
+      "loss": 2.2159,
       "step": 345
     },
     {
-      "epoch": 0.52,
-      "grad_norm": 0.087890625,
-      "learning_rate": 3.4805451459385544e-05,
-      "loss": 2.2368,
+      "epoch": 0.26,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0003482587064676617,
+      "loss": 2.2085,
       "step": 350
     },
     {
-      "epoch": 0.53,
-      "grad_norm": 0.0830078125,
-      "learning_rate": 3.398126899901305e-05,
-      "loss": 2.2359,
+      "epoch": 0.26,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0003532338308457712,
+      "loss": 2.2165,
       "step": 355
     },
     {
-      "epoch": 0.54,
-      "grad_norm": 0.0830078125,
-      "learning_rate": 3.31555725155377e-05,
-      "loss": 2.2472,
+      "epoch": 0.27,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0003582089552238806,
+      "loss": 2.2196,
       "step": 360
     },
     {
-      "epoch": 0.54,
-      "grad_norm": 0.0849609375,
-      "learning_rate": 3.232892228365181e-05,
-      "loss": 2.2511,
+      "epoch": 0.27,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0003631840796019901,
+      "loss": 2.212,
       "step": 365
     },
     {
-      "epoch": 0.55,
-      "grad_norm": 0.0849609375,
-      "learning_rate": 3.15018792252118e-05,
-      "loss": 2.2273,
+      "epoch": 0.28,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.00036815920398009953,
+      "loss": 2.2246,
       "step": 370
     },
     {
-      "epoch": 0.56,
-      "grad_norm": 0.08740234375,
-      "learning_rate": 3.067500452862575e-05,
-      "loss": 2.2391,
+      "epoch": 0.28,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.00037313432835820896,
+      "loss": 2.2166,
       "step": 375
     },
     {
-      "epoch": 0.57,
-      "grad_norm": 0.08447265625,
-      "learning_rate": 2.984885926806012e-05,
-      "loss": 2.2424,
+      "epoch": 0.28,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.00037810945273631844,
+      "loss": 2.2207,
       "step": 380
     },
     {
-      "epoch": 0.57,
-      "grad_norm": 0.0830078125,
-      "learning_rate": 2.9024004022724027e-05,
-      "loss": 2.2435,
+      "epoch": 0.29,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.00038308457711442787,
+      "loss": 2.2283,
       "step": 385
     },
     {
-      "epoch": 0.58,
-      "grad_norm": 0.087890625,
-      "learning_rate": 2.8200998496489373e-05,
-      "loss": 2.223,
+      "epoch": 0.29,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.00038805970149253735,
+      "loss": 2.2168,
       "step": 390
     },
     {
-      "epoch": 0.59,
-      "grad_norm": 0.08447265625,
-      "learning_rate": 2.7380401138104845e-05,
-      "loss": 2.2235,
+      "epoch": 0.29,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0003930348258706468,
+      "loss": 2.2358,
       "step": 395
     },
     {
-      "epoch": 0.6,
-      "grad_norm": 0.08984375,
-      "learning_rate": 2.656276876226166e-05,
-      "loss": 2.2379,
+      "epoch": 0.3,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.00039800995024875626,
+      "loss": 2.2289,
       "step": 400
     },
     {
-      "epoch": 0.6,
-      "grad_norm": 0.0869140625,
-      "learning_rate": 2.5748656171768127e-05,
-      "loss": 2.2232,
+      "epoch": 0.3,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.00039999932141401753,
+      "loss": 2.2386,
       "step": 405
     },
     {
-      "epoch": 0.61,
-      "grad_norm": 0.0888671875,
-      "learning_rate": 2.4938615781089216e-05,
-      "loss": 2.2492,
+      "epoch": 0.31,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.000399995174516356,
+      "loss": 2.2128,
       "step": 410
     },
     {
-      "epoch": 0.62,
-      "grad_norm": 0.08447265625,
-      "learning_rate": 2.413319724150689e-05,
-      "loss": 2.2429,
+      "epoch": 0.31,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.00039998725779131805,
+      "loss": 2.2145,
       "step": 415
     },
     {
-      "epoch": 0.63,
-      "grad_norm": 0.0849609375,
-      "learning_rate": 2.3332947068155364e-05,
-      "loss": 2.2486,
+      "epoch": 0.31,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0003999755713881306,
+      "loss": 2.2056,
       "step": 420
     },
     {
-      "epoch": 0.63,
-      "grad_norm": 0.08349609375,
-      "learning_rate": 2.2538408269184268e-05,
-      "loss": 2.2415,
+      "epoch": 0.32,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0003999601155270777,
+      "loss": 2.2149,
       "step": 425
     },
     {
-      "epoch": 0.64,
-      "grad_norm": 0.0869140625,
-      "learning_rate": 2.1750119977301616e-05,
-      "loss": 2.2568,
+      "epoch": 0.32,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.00039994089049949597,
+      "loss": 2.2165,
       "step": 430
     },
     {
-      "epoch": 0.65,
-      "grad_norm": 0.0859375,
-      "learning_rate": 2.096861708394641e-05,
-      "loss": 2.235,
+      "epoch": 0.32,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0003999178966677693,
+      "loss": 2.2031,
       "step": 435
     },
     {
-      "epoch": 0.66,
-      "grad_norm": 0.0849609375,
-      "learning_rate": 2.0194429876339054e-05,
-      "loss": 2.2364,
+      "epoch": 0.33,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.00039989113446532205,
+      "loss": 2.2134,
       "step": 440
     },
     {
-      "epoch": 0.66,
-      "grad_norm": 0.0859375,
-      "learning_rate": 1.9428083677656066e-05,
-      "loss": 2.2439,
+      "epoch": 0.33,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0003998606043966108,
+      "loss": 2.1896,
       "step": 445
     },
     {
-      "epoch": 0.67,
-      "grad_norm": 0.0859375,
-      "learning_rate": 1.8670098490573132e-05,
-      "loss": 2.238,
+      "epoch": 0.34,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.00039982630703711496,
+      "loss": 2.2205,
       "step": 450
     },
     {
-      "epoch": 0.68,
-      "grad_norm": 0.0869140625,
-      "learning_rate": 1.792098864441825e-05,
-      "loss": 2.2325,
+      "epoch": 0.34,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0003997882430333256,
+      "loss": 2.2203,
       "step": 455
     },
     {
-      "epoch": 0.69,
-      "grad_norm": 0.08544921875,
-      "learning_rate": 1.7181262446174615e-05,
-      "loss": 2.2342,
+      "epoch": 0.34,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.00039974641310273386,
+      "loss": 2.2134,
       "step": 460
     },
     {
-      "epoch": 0.69,
-      "grad_norm": 0.08740234375,
-      "learning_rate": 1.6451421835570044e-05,
-      "loss": 2.2262,
+      "epoch": 0.35,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0003997008180338166,
+      "loss": 2.2344,
       "step": 465
     },
     {
-      "epoch": 0.7,
-      "grad_norm": 0.08642578125,
-      "learning_rate": 1.5731962044486608e-05,
-      "loss": 2.2577,
+      "epoch": 0.35,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.00039965145868602243,
+      "loss": 2.1976,
       "step": 470
     },
     {
-      "epoch": 0.71,
-      "grad_norm": 0.08544921875,
-      "learning_rate": 1.5023371260922157e-05,
-      "loss": 2.2481,
+      "epoch": 0.35,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0003995983359897548,
+      "loss": 2.2175,
       "step": 475
     },
     {
-      "epoch": 0.72,
-      "grad_norm": 0.083984375,
-      "learning_rate": 1.4326130297731294e-05,
-      "loss": 2.229,
+      "epoch": 0.36,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.000399541450946355,
+      "loss": 2.2145,
       "step": 480
     },
     {
-      "epoch": 0.72,
-      "grad_norm": 0.08251953125,
-      "learning_rate": 1.3640712266370778e-05,
-      "loss": 2.2365,
+      "epoch": 0.36,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.00039948080462808266,
+      "loss": 2.2186,
       "step": 485
     },
     {
-      "epoch": 0.73,
-      "grad_norm": 0.08349609375,
-      "learning_rate": 1.2967582255870662e-05,
-      "loss": 2.2276,
+      "epoch": 0.37,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0003994163981780963,
+      "loss": 2.1911,
       "step": 490
     },
     {
-      "epoch": 0.74,
-      "grad_norm": 0.08203125,
-      "learning_rate": 1.2307197017249163e-05,
-      "loss": 2.245,
+      "epoch": 0.37,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.00039934823281043103,
+      "loss": 2.2123,
       "step": 495
     },
     {
-      "epoch": 0.75,
-      "grad_norm": 0.08544921875,
-      "learning_rate": 1.166000465358504e-05,
-      "loss": 2.2327,
+      "epoch": 0.37,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.00039927630980997634,
+      "loss": 2.2019,
       "step": 500
     },
     {
-      "epoch": 0.75,
-      "grad_norm": 0.08349609375,
-      "learning_rate": 1.1026444315958248e-05,
-      "loss": 2.2257,
+      "epoch": 0.38,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.00039920063053245145,
+      "loss": 2.2251,
       "step": 505
     },
     {
-      "epoch": 0.76,
-      "grad_norm": 0.08642578125,
-      "learning_rate": 1.0406945905464832e-05,
-      "loss": 2.2355,
+      "epoch": 0.38,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.00039912119640437963,
+      "loss": 2.2116,
       "step": 510
     },
     {
-      "epoch": 0.77,
-      "grad_norm": 0.08349609375,
-      "learning_rate": 9.801929781508377e-06,
-      "loss": 2.2366,
+      "epoch": 0.38,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.00039903800892306194,
+      "loss": 2.2142,
       "step": 515
     },
     {
-      "epoch": 0.78,
-      "grad_norm": 0.0849609375,
-      "learning_rate": 9.211806476565995e-06,
-      "loss": 2.2307,
+      "epoch": 0.39,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.00039895106965654836,
+      "loss": 2.2006,
       "step": 520
     },
     {
-      "epoch": 0.78,
-      "grad_norm": 0.0869140625,
-      "learning_rate": 8.63697641762235e-06,
-      "loss": 2.2298,
+      "epoch": 0.39,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0003988603802436086,
+      "loss": 2.2168,
       "step": 525
     },
     {
-      "epoch": 0.79,
-      "grad_norm": 0.0830078125,
-      "learning_rate": 8.077829654460684e-06,
-      "loss": 2.24,
+      "epoch": 0.4,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0003987659423937011,
+      "loss": 2.2054,
       "step": 530
     },
     {
-      "epoch": 0.8,
-      "grad_norm": 0.0869140625,
-      "learning_rate": 7.534745594995376e-06,
-      "loss": 2.2349,
+      "epoch": 0.4,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0003986677578869407,
+      "loss": 2.2115,
       "step": 535
     },
     {
-      "epoch": 0.81,
-      "grad_norm": 0.08544921875,
-      "learning_rate": 7.00809274782543e-06,
-      "loss": 2.2522,
+      "epoch": 0.4,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.00039856582857406524,
+      "loss": 2.1906,
       "step": 540
     },
     {
-      "epoch": 0.81,
-      "grad_norm": 0.087890625,
-      "learning_rate": 6.498228472183709e-06,
-      "loss": 2.2253,
+      "epoch": 0.41,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0003984601563764007,
+      "loss": 2.2139,
       "step": 545
     },
     {
-      "epoch": 0.82,
-      "grad_norm": 0.08544921875,
-      "learning_rate": 6.00549873545155e-06,
-      "loss": 2.2319,
+      "epoch": 0.41,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0003983507432858249,
+      "loss": 2.1912,
       "step": 550
     },
     {
-      "epoch": 0.83,
-      "grad_norm": 0.08544921875,
-      "learning_rate": 5.530237878403297e-06,
-      "loss": 2.23,
+      "epoch": 0.41,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0003982375913647298,
+      "loss": 2.2201,
       "step": 555
     },
     {
-      "epoch": 0.84,
-      "grad_norm": 0.08203125,
-      "learning_rate": 5.0727683883399965e-06,
-      "loss": 2.2366,
+      "epoch": 0.42,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.000398120702745983,
+      "loss": 2.1941,
       "step": 560
     },
     {
-      "epoch": 0.84,
-      "grad_norm": 0.08642578125,
-      "learning_rate": 4.63340068026638e-06,
-      "loss": 2.238,
+      "epoch": 0.42,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0003980000796328872,
+      "loss": 2.2206,
       "step": 565
     },
     {
-      "epoch": 0.85,
-      "grad_norm": 0.087890625,
-      "learning_rate": 4.212432886259269e-06,
-      "loss": 2.2238,
+      "epoch": 0.43,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0003978757242991389,
+      "loss": 2.2062,
       "step": 570
     },
     {
-      "epoch": 0.86,
-      "grad_norm": 0.0859375,
-      "learning_rate": 3.8101506531707373e-06,
-      "loss": 2.2308,
+      "epoch": 0.43,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.00039774763908878525,
+      "loss": 2.2098,
       "step": 575
     },
     {
-      "epoch": 0.87,
-      "grad_norm": 0.0859375,
-      "learning_rate": 3.426826948802979e-06,
-      "loss": 2.2305,
+      "epoch": 0.43,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0003976158264161802,
+      "loss": 2.2109,
       "step": 580
     },
     {
-      "epoch": 0.87,
-      "grad_norm": 0.0849609375,
-      "learning_rate": 3.0627218766865635e-06,
-      "loss": 2.2364,
+      "epoch": 0.44,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0003974802887659389,
+      "loss": 2.2148,
       "step": 585
     },
     {
-      "epoch": 0.88,
-      "grad_norm": 0.08642578125,
-      "learning_rate": 2.7180824995877015e-06,
-      "loss": 2.2435,
+      "epoch": 0.44,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0003973410286928906,
+      "loss": 2.199,
       "step": 590
     },
     {
-      "epoch": 0.89,
-      "grad_norm": 0.08642578125,
-      "learning_rate": 2.3931426718643005e-06,
-      "loss": 2.2347,
+      "epoch": 0.44,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0003971980488220308,
+      "loss": 2.2271,
       "step": 595
     },
     {
-      "epoch": 0.9,
-      "grad_norm": 0.08251953125,
-      "learning_rate": 2.0881228807845587e-06,
-      "loss": 2.2589,
+      "epoch": 0.45,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0003970513518484718,
+      "loss": 2.2221,
       "step": 600
     },
     {
-      "epoch": 0.9,
-      "grad_norm": 0.0849609375,
-      "learning_rate": 1.8032300969157738e-06,
-      "loss": 2.2323,
+      "epoch": 0.45,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.00039690094053739157,
+      "loss": 2.1961,
       "step": 605
     },
     {
-      "epoch": 0.91,
-      "grad_norm": 0.08642578125,
-      "learning_rate": 1.5386576336848828e-06,
-      "loss": 2.2285,
+      "epoch": 0.46,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0003967468177239819,
+      "loss": 2.2078,
       "step": 610
     },
     {
-      "epoch": 0.92,
-      "grad_norm": 0.0849609375,
-      "learning_rate": 1.2945850162060271e-06,
-      "loss": 2.2368,
+      "epoch": 0.46,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.00039658898631339496,
+      "loss": 2.2026,
       "step": 615
     },
     {
-      "epoch": 0.93,
-      "grad_norm": 0.08251953125,
-      "learning_rate": 1.0711778594641267e-06,
-      "loss": 2.2417,
+      "epoch": 0.46,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0003964274492806883,
+      "loss": 2.2094,
       "step": 620
     },
     {
-      "epoch": 0.93,
-      "grad_norm": 0.08642578125,
-      "learning_rate": 8.685877559371869e-07,
-      "loss": 2.2355,
+      "epoch": 0.47,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.00039626220967076917,
+      "loss": 2.2022,
       "step": 625
     },
     {
-      "epoch": 0.94,
-      "grad_norm": 0.08544921875,
-      "learning_rate": 6.869521727335145e-07,
-      "loss": 2.2285,
+      "epoch": 0.47,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0003960932705983365,
+      "loss": 2.1984,
       "step": 630
     },
     {
-      "epoch": 0.95,
-      "grad_norm": 0.08447265625,
-      "learning_rate": 5.26394358313689e-07,
-      "loss": 2.2262,
+      "epoch": 0.47,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.00039592063524782306,
+      "loss": 2.1981,
       "step": 635
     },
     {
-      "epoch": 0.95,
-      "grad_norm": 0.08154296875,
-      "learning_rate": 3.8702325886057816e-07,
-      "loss": 2.2289,
+      "epoch": 0.48,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.00039574430687333464,
+      "loss": 2.2084,
       "step": 640
     },
     {
-      "epoch": 0.96,
-      "grad_norm": 0.08642578125,
-      "learning_rate": 2.6893344435409695e-07,
-      "loss": 2.2428,
+      "epoch": 0.48,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.00039556428879858904,
+      "loss": 2.1912,
       "step": 645
     },
     {
-      "epoch": 0.97,
-      "grad_norm": 0.08740234375,
-      "learning_rate": 1.722050444009702e-07,
-      "loss": 2.2248,
+      "epoch": 0.48,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.00039538058441685353,
+      "loss": 2.1871,
       "step": 650
     },
     {
-      "epoch": 0.98,
-      "grad_norm": 0.08349609375,
-      "learning_rate": 9.690369386293054e-08,
-      "loss": 2.2297,
+      "epoch": 0.49,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0003951931971908807,
+      "loss": 2.187,
       "step": 655
     },
     {
-      "epoch": 0.98,
-      "grad_norm": 0.0849609375,
-      "learning_rate": 4.308048832030403e-08,
-      "loss": 2.2344,
+      "epoch": 0.49,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0003950021306528432,
+      "loss": 2.1915,
       "step": 660
     },
     {
-      "epoch": 0.99,
-      "grad_norm": 0.0859375,
-      "learning_rate": 1.0771949401241265e-08,
-      "loss": 2.2296,
+      "epoch": 0.5,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0003948073884042673,
+      "loss": 2.1892,
       "step": 665
     },
     {
-      "epoch": 1.0,
-      "grad_norm": 0.087890625,
-      "learning_rate": 0.0,
-      "loss": 2.2321,
+      "epoch": 0.5,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.00039460897411596477,
+      "loss": 2.2194,
       "step": 670
     },
     {
-      "epoch": 1.0,
-      "eval_loss": 2.2356441020965576,
-      "eval_runtime": 186.994,
-      "eval_samples_per_second": 25.787,
-      "eval_steps_per_second": 3.225,
-      "step": 670
+      "epoch": 0.5,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.00039440689152796406,
+      "loss": 2.2103,
+      "step": 675
     },
     {
-      "epoch": 1.0,
-      "step": 670,
-      "total_flos": 3.038889647317975e+17,
-      "train_loss": 2.275535515884855,
-      "train_runtime": 10859.5951,
-      "train_samples_per_second": 7.898,
-      "train_steps_per_second": 0.062
+      "epoch": 0.51,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.00039420114444943934,
+      "loss": 2.2032,
+      "step": 680
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0003939917367586391,
+      "loss": 2.1989,
+      "step": 685
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.00039377867240281275,
+      "loss": 2.1929,
+      "step": 690
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0003935619553981364,
+      "loss": 2.1961,
+      "step": 695
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0003933415898296372,
+      "loss": 2.1941,
+      "step": 700
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.000393117579851116,
+      "loss": 2.1983,
+      "step": 705
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0003928899296850695,
+      "loss": 2.1912,
+      "step": 710
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0003926586436226103,
+      "loss": 2.2096,
+      "step": 715
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0003924237260233863,
+      "loss": 2.2007,
+      "step": 720
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0003921851813154983,
+      "loss": 2.2171,
+      "step": 725
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0003919430139954167,
+      "loss": 2.2002,
+      "step": 730
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.00039169722862789644,
+      "loss": 2.1913,
+      "step": 735
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0003914478298458916,
+      "loss": 2.1765,
+      "step": 740
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.00039119482235046716,
+      "loss": 2.1971,
+      "step": 745
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.00039093821091071117,
+      "loss": 2.1978,
+      "step": 750
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.00039067800036364443,
+      "loss": 2.1954,
+      "step": 755
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0003904141956141297,
+      "loss": 2.2025,
+      "step": 760
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0003901468016347786,
+      "loss": 2.2045,
+      "step": 765
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.00038987582346585847,
+      "loss": 2.1939,
+      "step": 770
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0003896012662151972,
+      "loss": 2.175,
+      "step": 775
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.00038932313505808685,
+      "loss": 2.185,
+      "step": 780
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.00038904143523718615,
+      "loss": 2.1939,
+      "step": 785
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.00038875617206242174,
+      "loss": 2.1634,
+      "step": 790
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0003884673509108879,
+      "loss": 2.1947,
+      "step": 795
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.00038817497722674546,
+      "loss": 2.1925,
+      "step": 800
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.000387879056521119,
+      "loss": 2.1809,
+      "step": 805
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0003875795943719929,
+      "loss": 2.1763,
+      "step": 810
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.00038727659642410654,
+      "loss": 2.2132,
+      "step": 815
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0003869700683888474,
+      "loss": 2.194,
+      "step": 820
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0003866600160441438,
+      "loss": 2.1861,
+      "step": 825
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.00038634644523435587,
+      "loss": 2.2093,
+      "step": 830
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0003860293618701653,
+      "loss": 2.1894,
+      "step": 835
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0003857087719284641,
+      "loss": 2.2219,
+      "step": 840
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.00038538468145224165,
+      "loss": 2.1982,
+      "step": 845
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.00038505709655047113,
+      "loss": 2.1954,
+      "step": 850
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.00038472602339799427,
+      "loss": 2.2064,
+      "step": 855
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0003843914682354047,
+      "loss": 2.2166,
+      "step": 860
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.00038405343736893065,
+      "loss": 2.2001,
+      "step": 865
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.000383711937170316,
+      "loss": 2.1765,
+      "step": 870
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.00038336697407669994,
+      "loss": 2.1826,
+      "step": 875
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.000383018554590496,
+      "loss": 2.1964,
+      "step": 880
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0003826666852792692,
+      "loss": 2.1954,
+      "step": 885
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.00038231137277561244,
+      "loss": 2.2015,
+      "step": 890
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0003819526237770212,
+      "loss": 2.1932,
+      "step": 895
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0003815904450457677,
+      "loss": 2.1906,
+      "step": 900
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0003812248434087732,
+      "loss": 2.1776,
+      "step": 905
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.00038085582575747914,
+      "loss": 2.1936,
+      "step": 910
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0003804833990477177,
+      "loss": 2.1819,
+      "step": 915
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.00038010757029958016,
+      "loss": 2.1918,
+      "step": 920
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0003797283465972851,
+      "loss": 2.1776,
+      "step": 925
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0003793457350890443,
+      "loss": 2.1786,
+      "step": 930
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0003789597429869286,
+      "loss": 2.2086,
+      "step": 935
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0003785703775667314,
+      "loss": 2.2138,
+      "step": 940
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.00037817764616783196,
+      "loss": 2.2123,
+      "step": 945
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0003777815561930568,
+      "loss": 2.1889,
+      "step": 950
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0003773821151085401,
+      "loss": 2.1946,
+      "step": 955
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.00037697933044358335,
+      "loss": 2.1688,
+      "step": 960
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0003765732097905129,
+      "loss": 2.1795,
+      "step": 965
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.00037616376080453737,
+      "loss": 2.1966,
+      "step": 970
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0003757509912036028,
+      "loss": 2.1901,
+      "step": 975
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0003753349087682477,
+      "loss": 2.1685,
+      "step": 980
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.000374915521341456,
+      "loss": 2.1923,
+      "step": 985
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.00037449283682850957,
+      "loss": 2.1995,
+      "step": 990
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.00037406686319683887,
+      "loss": 2.1921,
+      "step": 995
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.00037363760847587284,
+      "loss": 2.178,
+      "step": 1000
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.00037320508075688776,
+      "loss": 2.1711,
+      "step": 1005
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.00037276928819285446,
+      "loss": 2.1825,
+      "step": 1010
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0003723302389982849,
+      "loss": 2.1925,
+      "step": 1015
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0003718879414490771,
+      "loss": 2.1758,
+      "step": 1020
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0003714424038823592,
+      "loss": 2.1837,
+      "step": 1025
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.00037099363469633245,
+      "loss": 2.1914,
+      "step": 1030
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0003705416423501128,
+      "loss": 2.1667,
+      "step": 1035
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0003700864353635714,
+      "loss": 2.1911,
+      "step": 1040
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.00036962802231717403,
+      "loss": 2.1867,
+      "step": 1045
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0003691664118518195,
+      "loss": 2.1717,
+      "step": 1050
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0003687016126686765,
+      "loss": 2.1794,
+      "step": 1055
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.00036823363352901997,
+      "loss": 2.2012,
+      "step": 1060
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0003677624832540655,
+      "loss": 2.1957,
+      "step": 1065
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0003672881707248034,
+      "loss": 2.1762,
+      "step": 1070
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0003668107048818312,
+      "loss": 2.2005,
+      "step": 1075
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0003663300947251851,
+      "loss": 2.2066,
+      "step": 1080
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0003658463493141703,
+      "loss": 2.1813,
+      "step": 1085
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.00036535947776719017,
+      "loss": 2.1659,
+      "step": 1090
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0003648694892615747,
+      "loss": 2.191,
+      "step": 1095
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0003643763930334071,
+      "loss": 2.1706,
+      "step": 1100
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.00036388019837734994,
+      "loss": 2.1885,
+      "step": 1105
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.00036338091464646984,
+      "loss": 2.1711,
+      "step": 1110
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0003628785512520613,
+      "loss": 2.1687,
+      "step": 1115
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0003623731176634691,
+      "loss": 2.2004,
+      "step": 1120
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.00036186462340791014,
+      "loss": 2.188,
+      "step": 1125
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0003613530780702934,
+      "loss": 2.1822,
+      "step": 1130
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.00036083849129303966,
+      "loss": 2.1848,
+      "step": 1135
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0003603208727758995,
+      "loss": 2.1607,
+      "step": 1140
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.00035980023227577063,
+      "loss": 2.1863,
+      "step": 1145
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.00035927657960651394,
+      "loss": 2.1711,
+      "step": 1150
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0003587499246387684,
+      "loss": 2.1806,
+      "step": 1155
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.00035822027729976504,
+      "loss": 2.1735,
+      "step": 1160
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.00035768764757314,
+      "loss": 2.1989,
+      "step": 1165
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.00035715204549874617,
+      "loss": 2.1728,
+      "step": 1170
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0003566134811724639,
+      "loss": 2.1933,
+      "step": 1175
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.00035607196474601074,
+      "loss": 2.1886,
+      "step": 1180
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.00035552750642675043,
+      "loss": 2.1829,
+      "step": 1185
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.00035498011647749976,
+      "loss": 2.1755,
+      "step": 1190
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.00035442980521633595,
+      "loss": 2.2011,
+      "step": 1195
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.00035387658301640136,
+      "loss": 2.2043,
+      "step": 1200
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0003533204603057088,
+      "loss": 2.1782,
+      "step": 1205
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.00035276144756694406,
+      "loss": 2.179,
+      "step": 1210
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.00035219955533726915,
+      "loss": 2.1841,
+      "step": 1215
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0003516347942081232,
+      "loss": 2.1646,
+      "step": 1220
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.00035106717482502267,
+      "loss": 2.1878,
+      "step": 1225
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0003504967078873613,
+      "loss": 2.1753,
+      "step": 1230
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.000349923404148208,
+      "loss": 2.1854,
+      "step": 1235
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0003493472744141041,
+      "loss": 2.1878,
+      "step": 1240
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0003487683295448598,
+      "loss": 2.1675,
+      "step": 1245
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0003481865804533494,
+      "loss": 2.1902,
+      "step": 1250
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.00034760203810530594,
+      "loss": 2.1848,
+      "step": 1255
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.00034701471351911395,
+      "loss": 2.1638,
+      "step": 1260
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.000346424617765602,
+      "loss": 2.1763,
+      "step": 1265
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.000345831761967834,
+      "loss": 2.165,
+      "step": 1270
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.00034523615730089986,
+      "loss": 2.1875,
+      "step": 1275
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0003446378149917042,
+      "loss": 2.1595,
+      "step": 1280
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0003440367463187553,
+      "loss": 2.1841,
+      "step": 1285
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.00034343296261195224,
+      "loss": 2.1882,
+      "step": 1290
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0003428264752523712,
+      "loss": 2.1671,
+      "step": 1295
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0003422172956720514,
+      "loss": 2.1671,
+      "step": 1300
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.00034160543535377926,
+      "loss": 2.1607,
+      "step": 1305
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0003409909058308718,
+      "loss": 2.1827,
+      "step": 1310
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0003403737186869596,
+      "loss": 2.183,
+      "step": 1315
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.00033975388555576835,
+      "loss": 2.1692,
+      "step": 1320
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0003391314181208995,
+      "loss": 2.1786,
+      "step": 1325
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.00033850632811561,
+      "loss": 2.166,
+      "step": 1330
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.00033787862732259123,
+      "loss": 2.1767,
+      "step": 1335
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0003372483275737468,
+      "loss": 2.1729,
+      "step": 1340
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 2.1778452396392822,
+      "eval_runtime": 189.5201,
+      "eval_samples_per_second": 25.443,
+      "eval_steps_per_second": 3.182,
+      "step": 1340
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0003366154407499695,
+      "loss": 2.1415,
+      "step": 1345
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0003359799787809179,
+      "loss": 2.1377,
+      "step": 1350
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0003353419536447902,
+      "loss": 2.1349,
+      "step": 1355
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.00033470137736809995,
+      "loss": 2.1339,
+      "step": 1360
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0003340582620254484,
+      "loss": 2.1364,
+      "step": 1365
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.000333412619739297,
+      "loss": 2.1484,
+      "step": 1370
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0003327644626797394,
+      "loss": 2.1753,
+      "step": 1375
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0003321138030642714,
+      "loss": 2.165,
+      "step": 1380
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.00033146065315756113,
+      "loss": 2.1622,
+      "step": 1385
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.00033080502527121756,
+      "loss": 2.1704,
+      "step": 1390
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0003301469317635587,
+      "loss": 2.162,
+      "step": 1395
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.00032948638503937846,
+      "loss": 2.1461,
+      "step": 1400
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0003288233975497128,
+      "loss": 2.1515,
+      "step": 1405
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.00032815798179160524,
+      "loss": 2.1306,
+      "step": 1410
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0003274901503078711,
+      "loss": 2.1343,
+      "step": 1415
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0003268199156868612,
+      "loss": 2.1235,
+      "step": 1420
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0003261472905622244,
+      "loss": 2.1496,
+      "step": 1425
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0003254722876126697,
+      "loss": 2.1503,
+      "step": 1430
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.00032479491956172705,
+      "loss": 2.1566,
+      "step": 1435
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0003241151991775076,
+      "loss": 2.1638,
+      "step": 1440
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0003234331392724631,
+      "loss": 2.1497,
+      "step": 1445
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.00032274875270314426,
+      "loss": 2.1478,
+      "step": 1450
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.00032206205236995843,
+      "loss": 2.149,
+      "step": 1455
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.00032137305121692655,
+      "loss": 2.1514,
+      "step": 1460
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.00032068176223143884,
+      "loss": 2.1817,
+      "step": 1465
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0003199881984440106,
+      "loss": 2.1721,
+      "step": 1470
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.000319292372928036,
+      "loss": 2.1319,
+      "step": 1475
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0003185942987995418,
+      "loss": 2.15,
+      "step": 1480
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0003178939892169403,
+      "loss": 2.1696,
+      "step": 1485
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0003171914573807813,
+      "loss": 2.1567,
+      "step": 1490
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.0003164867165335029,
+      "loss": 2.1454,
+      "step": 1495
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0003157797799591823,
+      "loss": 2.1482,
+      "step": 1500
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0003150706609832854,
+      "loss": 2.1297,
+      "step": 1505
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.00031435937297241527,
+      "loss": 2.1142,
+      "step": 1510
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0003136459293340605,
+      "loss": 2.1509,
+      "step": 1515
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.00031293034351634227,
+      "loss": 2.1382,
+      "step": 1520
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.00031221262900776116,
+      "loss": 2.1552,
+      "step": 1525
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0003114927993369425,
+      "loss": 2.1434,
+      "step": 1530
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.00031077086807238175,
+      "loss": 2.1364,
+      "step": 1535
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.00031004684882218845,
+      "loss": 2.1569,
+      "step": 1540
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0003093207552338298,
+      "loss": 2.1756,
+      "step": 1545
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0003085926009938735,
+      "loss": 2.1448,
+      "step": 1550
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.0003078623998277296,
+      "loss": 2.1403,
+      "step": 1555
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0003071301654993919,
+      "loss": 2.1391,
+      "step": 1560
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0003063959118111785,
+      "loss": 2.154,
+      "step": 1565
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0003056596526034717,
+      "loss": 2.1494,
+      "step": 1570
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0003049214017544569,
+      "loss": 2.1319,
+      "step": 1575
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0003041811731798611,
+      "loss": 2.1384,
+      "step": 1580
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.0003034389808326907,
+      "loss": 2.1642,
+      "step": 1585
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0003026948387029684,
+      "loss": 2.1557,
+      "step": 1590
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0003019487608174695,
+      "loss": 2.1392,
+      "step": 1595
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0003012007612394575,
+      "loss": 2.1431,
+      "step": 1600
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.000300450854068419,
+      "loss": 2.1507,
+      "step": 1605
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.000299699053439798,
+      "loss": 2.147,
+      "step": 1610
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.00029894537352472927,
+      "loss": 2.1361,
+      "step": 1615
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.00029818982852977157,
+      "loss": 2.1514,
+      "step": 1620
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.00029743243269663957,
+      "loss": 2.1597,
+      "step": 1625
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0002966732003019353,
+      "loss": 2.1449,
+      "step": 1630
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0002959121456568796,
+      "loss": 2.1392,
+      "step": 1635
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.00029514928310704164,
+      "loss": 2.1396,
+      "step": 1640
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.000294384627032069,
+      "loss": 2.1509,
+      "step": 1645
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0002936181918454164,
+      "loss": 2.1538,
+      "step": 1650
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0002928499919940743,
+      "loss": 2.1337,
+      "step": 1655
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0002920800419582961,
+      "loss": 2.1475,
+      "step": 1660
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0002913083562513257,
+      "loss": 2.1293,
+      "step": 1665
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0002905349494191235,
+      "loss": 2.1539,
+      "step": 1670
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.00028975983604009246,
+      "loss": 2.1215,
+      "step": 1675
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0002889830307248033,
+      "loss": 2.1578,
+      "step": 1680
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.00028820454811571907,
+      "loss": 2.1492,
+      "step": 1685
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0002874244028869191,
+      "loss": 2.1536,
+      "step": 1690
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0002866426097438222,
+      "loss": 2.1584,
+      "step": 1695
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0002858591834229102,
+      "loss": 2.163,
+      "step": 1700
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.000285074138691449,
+      "loss": 2.1423,
+      "step": 1705
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0002842874903472115,
+      "loss": 2.1499,
+      "step": 1710
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.00028349925321819776,
+      "loss": 2.1552,
+      "step": 1715
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.00028270944216235574,
+      "loss": 2.1471,
+      "step": 1720
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0002819180720673013,
+      "loss": 2.1362,
+      "step": 1725
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0002811251578500377,
+      "loss": 2.173,
+      "step": 1730
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0002803307144566741,
+      "loss": 2.1381,
+      "step": 1735
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.00027953475686214404,
+      "loss": 2.1409,
+      "step": 1740
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.0002787373000699232,
+      "loss": 2.1486,
+      "step": 1745
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.00027793835911174656,
+      "loss": 2.1659,
+      "step": 1750
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.00027713794904732483,
+      "loss": 2.1639,
+      "step": 1755
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.00027633608496406103,
+      "loss": 2.1667,
+      "step": 1760
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.00027553278197676567,
+      "loss": 2.1442,
+      "step": 1765
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.00027472805522737195,
+      "loss": 2.117,
+      "step": 1770
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.00027392191988465065,
+      "loss": 2.1499,
+      "step": 1775
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0002731143911439237,
+      "loss": 2.15,
+      "step": 1780
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.00027230548422677817,
+      "loss": 2.1542,
+      "step": 1785
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0002714952143807792,
+      "loss": 2.1437,
+      "step": 1790
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0002706835968791824,
+      "loss": 2.1627,
+      "step": 1795
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0002698706470206464,
+      "loss": 2.1453,
+      "step": 1800
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.00026905638012894405,
+      "loss": 2.1482,
+      "step": 1805
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.00026824081155267374,
+      "loss": 2.1707,
+      "step": 1810
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.00026742395666497015,
+      "loss": 2.1583,
+      "step": 1815
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0002666058308632144,
+      "loss": 2.1528,
+      "step": 1820
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0002657864495687437,
+      "loss": 2.1546,
+      "step": 1825
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.00026496582822656094,
+      "loss": 2.145,
+      "step": 1830
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.00026414398230504335,
+      "loss": 2.1581,
+      "step": 1835
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0002633209272956509,
+      "loss": 2.1352,
+      "step": 1840
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0002624966787126345,
+      "loss": 2.1661,
+      "step": 1845
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0002616712520927434,
+      "loss": 2.15,
+      "step": 1850
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00026084466299493227,
+      "loss": 2.1389,
+      "step": 1855
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0002600169270000682,
+      "loss": 2.1566,
+      "step": 1860
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0002591880597106365,
+      "loss": 2.155,
+      "step": 1865
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0002583580767504474,
+      "loss": 2.1491,
+      "step": 1870
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0002575269937643406,
+      "loss": 2.1399,
+      "step": 1875
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.00025669482641789106,
+      "loss": 2.1316,
+      "step": 1880
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0002558615903971135,
+      "loss": 2.1265,
+      "step": 1885
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.00025502730140816666,
+      "loss": 2.1384,
+      "step": 1890
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0002541919751770574,
+      "loss": 2.1535,
+      "step": 1895
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.00025335562744934403,
+      "loss": 2.1292,
+      "step": 1900
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0002525182739898397,
+      "loss": 2.1489,
+      "step": 1905
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.00025167993058231524,
+      "loss": 2.1454,
+      "step": 1910
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.00025084061302920146,
+      "loss": 2.1436,
+      "step": 1915
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0002500003371512917,
+      "loss": 2.1461,
+      "step": 1920
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.000249159118787443,
+      "loss": 2.1515,
+      "step": 1925
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.00024831697379427807,
+      "loss": 2.1204,
+      "step": 1930
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0002474739180458863,
+      "loss": 2.1579,
+      "step": 1935
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0002466299674335241,
+      "loss": 2.1382,
+      "step": 1940
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.00024578513786531605,
+      "loss": 2.1551,
+      "step": 1945
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0002449394452659544,
+      "loss": 2.1509,
+      "step": 1950
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.00024409290557639947,
+      "loss": 2.1462,
+      "step": 1955
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.00024324553475357866,
+      "loss": 2.1404,
+      "step": 1960
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.00024239734877008604,
+      "loss": 2.1677,
+      "step": 1965
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.000241548363613881,
+      "loss": 2.1602,
+      "step": 1970
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.00024069859528798714,
+      "loss": 2.1534,
+      "step": 1975
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.0002398480598101903,
+      "loss": 2.1448,
+      "step": 1980
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.00023899677321273714,
+      "loss": 2.1356,
+      "step": 1985
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.00023814475154203222,
+      "loss": 2.154,
+      "step": 1990
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00023729201085833626,
+      "loss": 2.1383,
+      "step": 1995
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.00023643856723546295,
+      "loss": 2.1611,
+      "step": 2000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.00023558443676047596,
+      "loss": 2.1302,
+      "step": 2005
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.00023472963553338613,
+      "loss": 2.1535,
+      "step": 2010
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.00023387417966684742,
+      "loss": 2.1414,
+      "step": 2015
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.00023301808528585375,
+      "loss": 2.1352,
+      "step": 2020
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0002321613685274346,
+      "loss": 2.152,
+      "step": 2025
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00023130404554035102,
+      "loss": 2.142,
+      "step": 2030
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.00023044613248479116,
+      "loss": 2.1588,
+      "step": 2035
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.000229587645532066,
+      "loss": 2.1475,
+      "step": 2040
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.00022872860086430393,
+      "loss": 2.1593,
+      "step": 2045
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.00022786901467414619,
+      "loss": 2.1467,
+      "step": 2050
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0002270089031644415,
+      "loss": 2.1364,
+      "step": 2055
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.00022614828254794055,
+      "loss": 2.1384,
+      "step": 2060
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.00022528716904699056,
+      "loss": 2.1428,
+      "step": 2065
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.00022442557889322946,
+      "loss": 2.1517,
+      "step": 2070
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.00022356352832727985,
+      "loss": 2.1474,
+      "step": 2075
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.00022270103359844283,
+      "loss": 2.1684,
+      "step": 2080
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.00022183811096439194,
+      "loss": 2.1616,
+      "step": 2085
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.00022097477669086638,
+      "loss": 2.1468,
+      "step": 2090
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.00022011104705136475,
+      "loss": 2.1374,
+      "step": 2095
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.00021924693832683806,
+      "loss": 2.1539,
+      "step": 2100
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.00021838246680538293,
+      "loss": 2.1514,
+      "step": 2105
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.00021751764878193459,
+      "loss": 2.1407,
+      "step": 2110
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.00021665250055795957,
+      "loss": 2.1485,
+      "step": 2115
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0002157870384411487,
+      "loss": 2.1496,
+      "step": 2120
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.00021492127874510946,
+      "loss": 2.143,
+      "step": 2125
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0002140552377890586,
+      "loss": 2.1498,
+      "step": 2130
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.00021318893189751457,
+      "loss": 2.1586,
+      "step": 2135
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.00021232237739998965,
+      "loss": 2.139,
+      "step": 2140
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00021145559063068223,
+      "loss": 2.1481,
+      "step": 2145
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.00021058858792816904,
+      "loss": 2.1449,
+      "step": 2150
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00020972138563509708,
+      "loss": 2.1629,
+      "step": 2155
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.00020885400009787528,
+      "loss": 2.1458,
+      "step": 2160
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.000207986447666367,
+      "loss": 2.1531,
+      "step": 2165
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0002071187446935813,
+      "loss": 2.1339,
+      "step": 2170
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.00020625090753536492,
+      "loss": 2.1525,
+      "step": 2175
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.00020538295255009384,
+      "loss": 2.1522,
+      "step": 2180
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0002045148960983652,
+      "loss": 2.1358,
+      "step": 2185
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0002036467545426886,
+      "loss": 2.149,
+      "step": 2190
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.00020277854424717803,
+      "loss": 2.1394,
+      "step": 2195
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.00020191028157724294,
+      "loss": 2.1424,
+      "step": 2200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0002010419828992801,
+      "loss": 2.1615,
+      "step": 2205
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.00020017366458036513,
+      "loss": 2.1549,
+      "step": 2210
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.00019930534298794365,
+      "loss": 2.1115,
+      "step": 2215
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0001984370344895232,
+      "loss": 2.1267,
+      "step": 2220
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.00019756875545236453,
+      "loss": 2.1387,
+      "step": 2225
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.00019670052224317274,
+      "loss": 2.1365,
+      "step": 2230
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0001958323512277895,
+      "loss": 2.1511,
+      "step": 2235
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0001949642587708838,
+      "loss": 2.1503,
+      "step": 2240
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.00019409626123564403,
+      "loss": 2.1469,
+      "step": 2245
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.00019322837498346934,
+      "loss": 2.125,
+      "step": 2250
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00019236061637366124,
+      "loss": 2.1331,
+      "step": 2255
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00019149300176311504,
+      "loss": 2.1354,
+      "step": 2260
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.00019062554750601198,
+      "loss": 2.1512,
+      "step": 2265
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0001897582699535107,
+      "loss": 2.1151,
+      "step": 2270
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.00018889118545343877,
+      "loss": 2.1239,
+      "step": 2275
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.000188024310349985,
+      "loss": 2.1381,
+      "step": 2280
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.00018715766098339117,
+      "loss": 2.1306,
+      "step": 2285
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.00018629125368964405,
+      "loss": 2.1489,
+      "step": 2290
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.00018542510480016713,
+      "loss": 2.1547,
+      "step": 2295
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.00018455923064151342,
+      "loss": 2.1343,
+      "step": 2300
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.00018369364753505728,
+      "loss": 2.144,
+      "step": 2305
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.00018282837179668679,
+      "loss": 2.1494,
+      "step": 2310
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.00018196341973649637,
+      "loss": 2.1511,
+      "step": 2315
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.00018109880765847906,
+      "loss": 2.148,
+      "step": 2320
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.00018023455186021961,
+      "loss": 2.142,
+      "step": 2325
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0001793706686325868,
+      "loss": 2.1445,
+      "step": 2330
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0001785071742594268,
+      "loss": 2.1344,
+      "step": 2335
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.00017764408501725593,
+      "loss": 2.1214,
+      "step": 2340
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00017678141717495394,
+      "loss": 2.1232,
+      "step": 2345
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.00017591918699345755,
+      "loss": 2.129,
+      "step": 2350
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.00017505741072545346,
+      "loss": 2.1462,
+      "step": 2355
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.00017419610461507254,
+      "loss": 2.1401,
+      "step": 2360
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0001733352848975832,
+      "loss": 2.1497,
+      "step": 2365
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.00017247496779908565,
+      "loss": 2.1356,
+      "step": 2370
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0001716151695362059,
+      "loss": 2.1436,
+      "step": 2375
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.00017075590631579019,
+      "loss": 2.1538,
+      "step": 2380
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00016989719433459924,
+      "loss": 2.1497,
+      "step": 2385
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.00016903904977900333,
+      "loss": 2.1333,
+      "step": 2390
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.000168181488824677,
+      "loss": 2.1542,
+      "step": 2395
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.00016732452763629395,
+      "loss": 2.1197,
+      "step": 2400
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.00016646818236722282,
+      "loss": 2.1151,
+      "step": 2405
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.00016561246915922204,
+      "loss": 2.1505,
+      "step": 2410
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00016475740414213642,
+      "loss": 2.1501,
+      "step": 2415
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.00016390300343359216,
+      "loss": 2.1556,
+      "step": 2420
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0001630492831386939,
+      "loss": 2.133,
+      "step": 2425
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.0001621962593497205,
+      "loss": 2.162,
+      "step": 2430
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0001613439481458221,
+      "loss": 2.1333,
+      "step": 2435
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.000160492365592717,
+      "loss": 2.1566,
+      "step": 2440
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00015964152774238842,
+      "loss": 2.1692,
+      "step": 2445
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.00015879145063278256,
+      "loss": 2.1413,
+      "step": 2450
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.00015794215028750567,
+      "loss": 2.1564,
+      "step": 2455
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.00015709364271552262,
+      "loss": 2.1305,
+      "step": 2460
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.00015624594391085457,
+      "loss": 2.1526,
+      "step": 2465
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.00015539906985227798,
+      "loss": 2.138,
+      "step": 2470
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0001545530365030229,
+      "loss": 2.1432,
+      "step": 2475
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.00015370785981047252,
+      "loss": 2.1508,
+      "step": 2480
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.00015286355570586255,
+      "loss": 2.1347,
+      "step": 2485
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.00015202014010398042,
+      "loss": 2.1497,
+      "step": 2490
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.00015117762890286602,
+      "loss": 2.1377,
+      "step": 2495
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0001503360379835113,
+      "loss": 2.1337,
+      "step": 2500
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00014949538320956158,
+      "loss": 2.156,
+      "step": 2505
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.00014865568042701592,
+      "loss": 2.1386,
+      "step": 2510
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0001478169454639291,
+      "loss": 2.1468,
+      "step": 2515
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.00014697919413011253,
+      "loss": 2.1566,
+      "step": 2520
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.00014614244221683686,
+      "loss": 2.1441,
+      "step": 2525
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.00014530670549653407,
+      "loss": 2.1473,
+      "step": 2530
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.00014447199972249987,
+      "loss": 2.1481,
+      "step": 2535
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.00014363834062859748,
+      "loss": 2.1546,
+      "step": 2540
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.00014280574392896032,
+      "loss": 2.1314,
+      "step": 2545
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0001419742253176962,
+      "loss": 2.1129,
+      "step": 2550
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.00014114380046859138,
+      "loss": 2.1353,
+      "step": 2555
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.00014031448503481532,
+      "loss": 2.1423,
+      "step": 2560
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.00013948629464862516,
+      "loss": 2.1173,
+      "step": 2565
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.00013865924492107153,
+      "loss": 2.1468,
+      "step": 2570
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.00013783335144170418,
+      "loss": 2.1517,
+      "step": 2575
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0001370086297782779,
+      "loss": 2.153,
+      "step": 2580
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0001361850954764594,
+      "loss": 2.1427,
+      "step": 2585
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0001353627640595338,
+      "loss": 2.1477,
+      "step": 2590
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.00013454165102811272,
+      "loss": 2.1414,
+      "step": 2595
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.00013372177185984134,
+      "loss": 2.1579,
+      "step": 2600
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00013290314200910735,
+      "loss": 2.123,
+      "step": 2605
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.00013208577690674905,
+      "loss": 2.1311,
+      "step": 2610
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.00013126969195976495,
+      "loss": 2.1314,
+      "step": 2615
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00013045490255102316,
+      "loss": 2.1374,
+      "step": 2620
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.00012964142403897112,
+      "loss": 2.1489,
+      "step": 2625
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0001288292717573468,
+      "loss": 2.145,
+      "step": 2630
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.00012801846101488898,
+      "loss": 2.1288,
+      "step": 2635
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.00012720900709504917,
+      "loss": 2.1468,
+      "step": 2640
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.00012640092525570312,
+      "loss": 2.1201,
+      "step": 2645
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0001255942307288637,
+      "loss": 2.1523,
+      "step": 2650
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.00012478893872039314,
+      "loss": 2.146,
+      "step": 2655
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.00012398506440971713,
+      "loss": 2.1387,
+      "step": 2660
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.00012318262294953815,
+      "loss": 2.1272,
+      "step": 2665
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.00012238162946555002,
+      "loss": 2.1488,
+      "step": 2670
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.00012158209905615301,
+      "loss": 2.1445,
+      "step": 2675
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.00012078404679216864,
+      "loss": 2.1327,
+      "step": 2680
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 2.1632144451141357,
+      "eval_runtime": 186.9566,
+      "eval_samples_per_second": 25.792,
+      "eval_steps_per_second": 3.225,
+      "step": 2681
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0001199874877165564,
+      "loss": 2.1156,
+      "step": 2685
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.00011919243684412948,
+      "loss": 2.115,
+      "step": 2690
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.00011839890916127228,
+      "loss": 2.0971,
+      "step": 2695
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.00011760691962565752,
+      "loss": 2.119,
+      "step": 2700
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.00011681648316596461,
+      "loss": 2.1282,
+      "step": 2705
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.00011602761468159813,
+      "loss": 2.1151,
+      "step": 2710
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.00011524032904240671,
+      "loss": 2.101,
+      "step": 2715
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.00011445464108840345,
+      "loss": 2.113,
+      "step": 2720
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.0001136705656294851,
+      "loss": 2.118,
+      "step": 2725
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.00011288811744515433,
+      "loss": 2.1045,
+      "step": 2730
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.0001121073112842395,
+      "loss": 2.108,
+      "step": 2735
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.00011132816186461821,
+      "loss": 2.0919,
+      "step": 2740
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.00011055068387293879,
+      "loss": 2.1063,
+      "step": 2745
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.00010977489196434381,
+      "loss": 2.1175,
+      "step": 2750
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.00010900080076219426,
+      "loss": 2.1103,
+      "step": 2755
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.00010822842485779285,
+      "loss": 2.1111,
+      "step": 2760
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.00010745777881011027,
+      "loss": 2.0899,
+      "step": 2765
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.00010668887714550974,
+      "loss": 2.0935,
+      "step": 2770
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0001059217343574737,
+      "loss": 2.0919,
+      "step": 2775
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.00010515636490633043,
+      "loss": 2.1157,
+      "step": 2780
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.00010439278321898153,
+      "loss": 2.1024,
+      "step": 2785
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.00010363100368863021,
+      "loss": 2.1038,
+      "step": 2790
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.00010287104067450928,
+      "loss": 2.1088,
+      "step": 2795
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.00010211290850161144,
+      "loss": 2.0861,
+      "step": 2800
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.00010135662146041855,
+      "loss": 2.1215,
+      "step": 2805
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.0001006021938066325,
+      "loss": 2.1062,
+      "step": 2810
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 9.984963976090651e-05,
+      "loss": 2.1014,
+      "step": 2815
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.123046875,
+      "learning_rate": 9.909897350857706e-05,
+      "loss": 2.1023,
+      "step": 2820
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 9.83502091993965e-05,
+      "loss": 2.1025,
+      "step": 2825
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 9.760336094726624e-05,
+      "loss": 2.1241,
+      "step": 2830
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 9.6858442829971e-05,
+      "loss": 2.1155,
+      "step": 2835
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.123046875,
+      "learning_rate": 9.611546888891307e-05,
+      "loss": 2.1012,
+      "step": 2840
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 9.537445312884788e-05,
+      "loss": 2.1058,
+      "step": 2845
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 9.463540951761989e-05,
+      "loss": 2.0876,
+      "step": 2850
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 9.389835198589944e-05,
+      "loss": 2.1222,
+      "step": 2855
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.126953125,
+      "learning_rate": 9.316329442691995e-05,
+      "loss": 2.1107,
+      "step": 2860
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 9.243025069621649e-05,
+      "loss": 2.1065,
+      "step": 2865
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 9.169923461136376e-05,
+      "loss": 2.1193,
+      "step": 2870
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.125,
+      "learning_rate": 9.097025995171669e-05,
+      "loss": 2.1154,
+      "step": 2875
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 9.024334045814988e-05,
+      "loss": 2.1055,
+      "step": 2880
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 8.951848983279898e-05,
+      "loss": 2.1039,
+      "step": 2885
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.123046875,
+      "learning_rate": 8.87957217388023e-05,
+      "loss": 2.1249,
+      "step": 2890
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.125,
+      "learning_rate": 8.80750498000432e-05,
+      "loss": 2.1231,
+      "step": 2895
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 8.735648760089367e-05,
+      "loss": 2.1346,
+      "step": 2900
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 8.66400486859575e-05,
+      "loss": 2.107,
+      "step": 2905
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 8.592574655981594e-05,
+      "loss": 2.1189,
+      "step": 2910
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 8.521359468677214e-05,
+      "loss": 2.1061,
+      "step": 2915
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 8.450360649059834e-05,
+      "loss": 2.1297,
+      "step": 2920
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 8.379579535428203e-05,
+      "loss": 2.1119,
+      "step": 2925
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 8.309017461977409e-05,
+      "loss": 2.0947,
+      "step": 2930
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 8.23867575877374e-05,
+      "loss": 2.1072,
+      "step": 2935
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.123046875,
+      "learning_rate": 8.168555751729551e-05,
+      "loss": 2.106,
+      "step": 2940
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 8.098658762578369e-05,
+      "loss": 2.1183,
+      "step": 2945
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.125,
+      "learning_rate": 8.028986108849887e-05,
+      "loss": 2.1103,
+      "step": 2950
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 7.959539103845184e-05,
+      "loss": 2.1414,
+      "step": 2955
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.125,
+      "learning_rate": 7.890319056611942e-05,
+      "loss": 2.11,
+      "step": 2960
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 7.82132727191978e-05,
+      "loss": 2.1178,
+      "step": 2965
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 7.752565050235694e-05,
+      "loss": 2.1018,
+      "step": 2970
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 7.684033687699455e-05,
+      "loss": 2.1184,
+      "step": 2975
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 7.615734476099284e-05,
+      "loss": 2.1208,
+      "step": 2980
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 7.547668702847421e-05,
+      "loss": 2.1201,
+      "step": 2985
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 7.479837650955906e-05,
+      "loss": 2.123,
+      "step": 2990
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 7.412242599012366e-05,
+      "loss": 2.1324,
+      "step": 2995
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.123046875,
+      "learning_rate": 7.34488482115593e-05,
+      "loss": 2.1275,
+      "step": 3000
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 7.277765587053206e-05,
+      "loss": 2.1193,
+      "step": 3005
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 7.210886161874344e-05,
+      "loss": 2.1165,
+      "step": 3010
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 7.144247806269213e-05,
+      "loss": 2.1136,
+      "step": 3015
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 7.0778517763436e-05,
+      "loss": 2.0813,
+      "step": 3020
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 7.011699323635559e-05,
+      "loss": 2.0982,
+      "step": 3025
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 6.94579169509181e-05,
+      "loss": 2.1135,
+      "step": 3030
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 6.88013013304424e-05,
+      "loss": 2.1057,
+      "step": 3035
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 6.814715875186475e-05,
+      "loss": 2.1319,
+      "step": 3040
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 6.749550154550585e-05,
+      "loss": 2.1206,
+      "step": 3045
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.126953125,
+      "learning_rate": 6.684634199483773e-05,
+      "loss": 2.123,
+      "step": 3050
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.126953125,
+      "learning_rate": 6.619969233625298e-05,
+      "loss": 2.1197,
+      "step": 3055
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 6.55555647588336e-05,
+      "loss": 2.1075,
+      "step": 3060
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.125,
+      "learning_rate": 6.491397140412139e-05,
+      "loss": 2.1185,
+      "step": 3065
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 6.42749243658891e-05,
+      "loss": 2.1114,
+      "step": 3070
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 6.363843568991243e-05,
+      "loss": 2.0937,
+      "step": 3075
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.123046875,
+      "learning_rate": 6.300451737374322e-05,
+      "loss": 2.0954,
+      "step": 3080
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.125,
+      "learning_rate": 6.237318136648258e-05,
+      "loss": 2.1127,
+      "step": 3085
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 6.174443956855671e-05,
+      "loss": 2.1174,
+      "step": 3090
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.1328125,
+      "learning_rate": 6.111830383149164e-05,
+      "loss": 2.1148,
+      "step": 3095
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 6.04947859576904e-05,
+      "loss": 2.1382,
+      "step": 3100
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.126953125,
+      "learning_rate": 5.9873897700210304e-05,
+      "loss": 2.1125,
+      "step": 3105
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.125,
+      "learning_rate": 5.92556507625414e-05,
+      "loss": 2.1068,
+      "step": 3110
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 5.86400567983862e-05,
+      "loss": 2.1204,
+      "step": 3115
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 5.802712741143934e-05,
+      "loss": 2.1046,
+      "step": 3120
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.126953125,
+      "learning_rate": 5.741687415516968e-05,
+      "loss": 2.1179,
+      "step": 3125
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.125,
+      "learning_rate": 5.680930853260182e-05,
+      "loss": 2.1184,
+      "step": 3130
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 5.6204441996099686e-05,
+      "loss": 2.1204,
+      "step": 3135
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.125,
+      "learning_rate": 5.560228594715049e-05,
+      "loss": 2.1097,
+      "step": 3140
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.1328125,
+      "learning_rate": 5.500285173614985e-05,
+      "loss": 2.1141,
+      "step": 3145
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.125,
+      "learning_rate": 5.4406150662188035e-05,
+      "loss": 2.1096,
+      "step": 3150
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 5.3812193972836436e-05,
+      "loss": 2.1134,
+      "step": 3155
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.126953125,
+      "learning_rate": 5.322099286393625e-05,
+      "loss": 2.1132,
+      "step": 3160
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.126953125,
+      "learning_rate": 5.263255847938693e-05,
+      "loss": 2.1083,
+      "step": 3165
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.126953125,
+      "learning_rate": 5.204690191093635e-05,
+      "loss": 2.1093,
+      "step": 3170
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 5.1464034197971726e-05,
+      "loss": 2.1123,
+      "step": 3175
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 5.08839663273114e-05,
+      "loss": 2.1115,
+      "step": 3180
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 5.030670923299785e-05,
+      "loss": 2.1129,
+      "step": 3185
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.125,
+      "learning_rate": 4.9732273796091685e-05,
+      "loss": 2.1259,
+      "step": 3190
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 4.916067084446632e-05,
+      "loss": 2.1305,
+      "step": 3195
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 4.859191115260393e-05,
+      "loss": 2.098,
+      "step": 3200
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.126953125,
+      "learning_rate": 4.8026005441392505e-05,
+      "loss": 2.1109,
+      "step": 3205
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 4.7462964377923635e-05,
+      "loss": 2.1043,
+      "step": 3210
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 4.690279857529145e-05,
+      "loss": 2.0896,
+      "step": 3215
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.125,
+      "learning_rate": 4.634551859239254e-05,
+      "loss": 2.1266,
+      "step": 3220
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.125,
+      "learning_rate": 4.57911349337272e-05,
+      "loss": 2.0982,
+      "step": 3225
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 4.523965804920078e-05,
+      "loss": 2.1275,
+      "step": 3230
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 4.469109833392759e-05,
+      "loss": 2.1323,
+      "step": 3235
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 4.414546612803421e-05,
+      "loss": 2.1026,
+      "step": 3240
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 4.3602771716464874e-05,
+      "loss": 2.085,
+      "step": 3245
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.126953125,
+      "learning_rate": 4.3063025328787676e-05,
+      "loss": 2.1026,
+      "step": 3250
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 4.252623713900159e-05,
+      "loss": 2.1093,
+      "step": 3255
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.125,
+      "learning_rate": 4.199241726534495e-05,
+      "loss": 2.1026,
+      "step": 3260
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 4.146157577010421e-05,
+      "loss": 2.1192,
+      "step": 3265
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 4.0933722659424945e-05,
+      "loss": 2.1114,
+      "step": 3270
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 4.040886788312268e-05,
+      "loss": 2.1239,
+      "step": 3275
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 3.9887021334495625e-05,
+      "loss": 2.0952,
+      "step": 3280
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 3.936819285013826e-05,
+      "loss": 2.114,
+      "step": 3285
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.12890625,
+      "learning_rate": 3.885239220975547e-05,
+      "loss": 2.1189,
+      "step": 3290
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.125,
+      "learning_rate": 3.833962913597893e-05,
+      "loss": 2.0974,
+      "step": 3295
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 3.7829913294183e-05,
+      "loss": 2.1085,
+      "step": 3300
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 3.73232542923033e-05,
+      "loss": 2.1023,
+      "step": 3305
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.125,
+      "learning_rate": 3.681966168065509e-05,
+      "loss": 2.1213,
+      "step": 3310
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 3.6319144951753436e-05,
+      "loss": 2.1312,
+      "step": 3315
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 3.582171354013444e-05,
+      "loss": 2.1268,
+      "step": 3320
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.126953125,
+      "learning_rate": 3.5327376822176885e-05,
+      "loss": 2.1068,
+      "step": 3325
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.126953125,
+      "learning_rate": 3.483614411592628e-05,
+      "loss": 2.1057,
+      "step": 3330
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.12890625,
+      "learning_rate": 3.434802468091836e-05,
+      "loss": 2.103,
+      "step": 3335
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 3.386302771800527e-05,
+      "loss": 2.1166,
+      "step": 3340
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 3.3381162369181717e-05,
+      "loss": 2.1176,
+      "step": 3345
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.125,
+      "learning_rate": 3.290243771741275e-05,
+      "loss": 2.1137,
+      "step": 3350
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 3.2426862786462565e-05,
+      "loss": 2.1017,
+      "step": 3355
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 3.195444654072439e-05,
+      "loss": 2.1049,
+      "step": 3360
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.125,
+      "learning_rate": 3.148519788505166e-05,
+      "loss": 2.1144,
+      "step": 3365
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.125,
+      "learning_rate": 3.101912566458989e-05,
+      "loss": 2.0956,
+      "step": 3370
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 3.0556238664610105e-05,
+      "loss": 2.1077,
+      "step": 3375
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.126953125,
+      "learning_rate": 3.009654561034323e-05,
+      "loss": 2.1178,
+      "step": 3380
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 2.9640055166815673e-05,
+      "loss": 2.105,
+      "step": 3385
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 2.918677593868586e-05,
+      "loss": 2.1051,
+      "step": 3390
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 2.8736716470082204e-05,
+      "loss": 2.0967,
+      "step": 3395
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 2.8289885244441803e-05,
+      "loss": 2.1174,
+      "step": 3400
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.12890625,
+      "learning_rate": 2.7846290684350963e-05,
+      "loss": 2.1216,
+      "step": 3405
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 2.740594115138595e-05,
+      "loss": 2.1199,
+      "step": 3410
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 2.6968844945955617e-05,
+      "loss": 2.1112,
+      "step": 3415
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 2.6535010307145002e-05,
+      "loss": 2.1374,
+      "step": 3420
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 2.6104445412559876e-05,
+      "loss": 2.1233,
+      "step": 3425
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 2.5677158378172707e-05,
+      "loss": 2.1049,
+      "step": 3430
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 2.5253157258169567e-05,
+      "loss": 2.1058,
+      "step": 3435
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.125,
+      "learning_rate": 2.4832450044798573e-05,
+      "loss": 2.1154,
+      "step": 3440
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.126953125,
+      "learning_rate": 2.4415044668218735e-05,
+      "loss": 2.1126,
+      "step": 3445
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 2.4000948996351104e-05,
+      "loss": 2.132,
+      "step": 3450
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 2.359017083472994e-05,
+      "loss": 2.1093,
+      "step": 3455
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 2.3182717926355845e-05,
+      "loss": 2.0929,
+      "step": 3460
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 2.277859795154986e-05,
+      "loss": 2.1068,
+      "step": 3465
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 2.237781852780838e-05,
+      "loss": 2.1095,
+      "step": 3470
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.125,
+      "learning_rate": 2.1980387209660026e-05,
+      "loss": 2.1148,
+      "step": 3475
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.126953125,
+      "learning_rate": 2.1586311488522702e-05,
+      "loss": 2.1104,
+      "step": 3480
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 2.1195598792562964e-05,
+      "loss": 2.1175,
+      "step": 3485
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 2.0808256486555554e-05,
+      "loss": 2.1094,
+      "step": 3490
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.126953125,
+      "learning_rate": 2.042429187174475e-05,
+      "loss": 2.121,
+      "step": 3495
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 2.0043712185706863e-05,
+      "loss": 2.1047,
+      "step": 3500
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.126953125,
+      "learning_rate": 1.966652460221341e-05,
+      "loss": 2.1098,
+      "step": 3505
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 1.9292736231096464e-05,
+      "loss": 2.1114,
+      "step": 3510
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.12890625,
+      "learning_rate": 1.8922354118114138e-05,
+      "loss": 2.1267,
+      "step": 3515
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.125,
+      "learning_rate": 1.8555385244818035e-05,
+      "loss": 2.0916,
+      "step": 3520
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.126953125,
+      "learning_rate": 1.8191836528421558e-05,
+      "loss": 2.0985,
+      "step": 3525
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.123046875,
+      "learning_rate": 1.7831714821669588e-05,
+      "loss": 2.1129,
+      "step": 3530
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 1.7475026912709235e-05,
+      "loss": 2.0889,
+      "step": 3535
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.12890625,
+      "learning_rate": 1.71217795249619e-05,
+      "loss": 2.1067,
+      "step": 3540
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 1.6771979316996677e-05,
+      "loss": 2.0987,
+      "step": 3545
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 1.6425632882404618e-05,
+      "loss": 2.099,
+      "step": 3550
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 1.6082746749674604e-05,
+      "loss": 2.1211,
+      "step": 3555
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 1.5743327382070206e-05,
+      "loss": 2.1099,
+      "step": 3560
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.125,
+      "learning_rate": 1.540738117750793e-05,
+      "loss": 2.1109,
+      "step": 3565
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.126953125,
+      "learning_rate": 1.507491446843654e-05,
+      "loss": 2.1064,
+      "step": 3570
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 1.4745933521717781e-05,
+      "loss": 2.0908,
+      "step": 3575
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 1.4420444538508083e-05,
+      "loss": 2.119,
+      "step": 3580
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.125,
+      "learning_rate": 1.4098453654141975e-05,
+      "loss": 2.1008,
+      "step": 3585
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.130859375,
+      "learning_rate": 1.377996693801611e-05,
+      "loss": 2.1134,
+      "step": 3590
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.126953125,
+      "learning_rate": 1.346499039347504e-05,
+      "loss": 2.1147,
+      "step": 3595
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 1.3153529957698008e-05,
+      "loss": 2.1141,
+      "step": 3600
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 1.2845591501587017e-05,
+      "loss": 2.0835,
+      "step": 3605
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.126953125,
+      "learning_rate": 1.254118082965634e-05,
+      "loss": 2.1109,
+      "step": 3610
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 1.2240303679922727e-05,
+      "loss": 2.1165,
+      "step": 3615
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 1.1942965723797671e-05,
+      "loss": 2.1035,
+      "step": 3620
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 1.164917256598017e-05,
+      "loss": 2.1112,
+      "step": 3625
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 1.1358929744351332e-05,
+      "loss": 2.1051,
+      "step": 3630
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 1.1072242729869819e-05,
+      "loss": 2.1133,
+      "step": 3635
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.123046875,
+      "learning_rate": 1.0789116926468756e-05,
+      "loss": 2.1097,
+      "step": 3640
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.125,
+      "learning_rate": 1.050955767095403e-05,
+      "loss": 2.1082,
+      "step": 3645
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 1.0233570232903323e-05,
+      "loss": 2.1208,
+      "step": 3650
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 9.961159814567267e-06,
+      "loss": 2.097,
+      "step": 3655
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 9.692331550770918e-06,
+      "loss": 2.1102,
+      "step": 3660
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.126953125,
+      "learning_rate": 9.42709050881736e-06,
+      "loss": 2.0946,
+      "step": 3665
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.125,
+      "learning_rate": 9.165441688391885e-06,
+      "loss": 2.1129,
+      "step": 3670
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 8.907390021467921e-06,
+      "loss": 2.1016,
+      "step": 3675
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 8.652940372214069e-06,
+      "loss": 2.0817,
+      "step": 3680
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.125,
+      "learning_rate": 8.40209753690222e-06,
+      "loss": 2.1327,
+      "step": 3685
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.125,
+      "learning_rate": 8.154866243817494e-06,
+      "loss": 2.1231,
+      "step": 3690
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.125,
+      "learning_rate": 7.911251153168752e-06,
+      "loss": 2.1175,
+      "step": 3695
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 7.67125685700103e-06,
+      "loss": 2.113,
+      "step": 3700
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 7.434887879108776e-06,
+      "loss": 2.0957,
+      "step": 3705
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.126953125,
+      "learning_rate": 7.202148674950704e-06,
+      "loss": 2.1117,
+      "step": 3710
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 6.97304363156579e-06,
+      "loss": 2.1021,
+      "step": 3715
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 6.747577067490563e-06,
+      "loss": 2.1233,
+      "step": 3720
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.126953125,
+      "learning_rate": 6.525753232677678e-06,
+      "loss": 2.1289,
+      "step": 3725
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.125,
+      "learning_rate": 6.307576308415852e-06,
+      "loss": 2.1155,
+      "step": 3730
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 6.093050407251033e-06,
+      "loss": 2.108,
+      "step": 3735
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.130859375,
+      "learning_rate": 5.882179572908841e-06,
+      "loss": 2.1112,
+      "step": 3740
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.125,
+      "learning_rate": 5.6749677802184095e-06,
+      "loss": 2.118,
+      "step": 3745
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 5.471418935037398e-06,
+      "loss": 2.096,
+      "step": 3750
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 5.271536874178451e-06,
+      "loss": 2.1087,
+      "step": 3755
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 5.075325365336791e-06,
+      "loss": 2.1044,
+      "step": 3760
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 4.882788107019231e-06,
+      "loss": 2.127,
+      "step": 3765
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 4.693928728474517e-06,
+      "loss": 2.0874,
+      "step": 3770
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 4.5087507896247605e-06,
+      "loss": 2.1114,
+      "step": 3775
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.12890625,
+      "learning_rate": 4.327257780998517e-06,
+      "loss": 2.1198,
+      "step": 3780
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 4.149453123664881e-06,
+      "loss": 2.1174,
+      "step": 3785
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 3.975340169169095e-06,
+      "loss": 2.0976,
+      "step": 3790
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.125,
+      "learning_rate": 3.804922199469174e-06,
+      "loss": 2.0964,
+      "step": 3795
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 3.6382024268743153e-06,
+      "loss": 2.1217,
+      "step": 3800
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.126953125,
+      "learning_rate": 3.4751839939841435e-06,
+      "loss": 2.1284,
+      "step": 3805
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 3.3158699736295375e-06,
+      "loss": 2.1031,
+      "step": 3810
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 3.160263368814764e-06,
+      "loss": 2.1215,
+      "step": 3815
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 3.0083671126607484e-06,
+      "loss": 2.1315,
+      "step": 3820
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.126953125,
+      "learning_rate": 2.860184068349958e-06,
+      "loss": 2.1137,
+      "step": 3825
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 2.7157170290721625e-06,
+      "loss": 2.1069,
+      "step": 3830
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 2.5749687179721815e-06,
+      "loss": 2.1111,
+      "step": 3835
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 2.4379417880981304e-06,
+      "loss": 2.0955,
+      "step": 3840
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 2.304638822351701e-06,
+      "loss": 2.1026,
+      "step": 3845
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.125,
+      "learning_rate": 2.1750623334393816e-06,
+      "loss": 2.0882,
+      "step": 3850
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.125,
+      "learning_rate": 2.049214763825069e-06,
+      "loss": 2.1075,
+      "step": 3855
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.126953125,
+      "learning_rate": 1.9270984856840867e-06,
+      "loss": 2.1132,
+      "step": 3860
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.125,
+      "learning_rate": 1.8087158008583515e-06,
+      "loss": 2.1085,
+      "step": 3865
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.125,
+      "learning_rate": 1.6940689408132092e-06,
+      "loss": 2.0978,
+      "step": 3870
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.126953125,
+      "learning_rate": 1.583160066595113e-06,
+      "loss": 2.1095,
+      "step": 3875
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 1.4759912687910771e-06,
+      "loss": 2.1079,
+      "step": 3880
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 1.3725645674891762e-06,
+      "loss": 2.1186,
+      "step": 3885
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.12890625,
+      "learning_rate": 1.2728819122404646e-06,
+      "loss": 2.1165,
+      "step": 3890
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 1.1769451820223376e-06,
+      "loss": 2.1125,
+      "step": 3895
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.126953125,
+      "learning_rate": 1.084756185202962e-06,
+      "loss": 2.1132,
+      "step": 3900
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.126953125,
+      "learning_rate": 9.963166595073014e-07,
+      "loss": 2.1062,
+      "step": 3905
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 9.116282719842772e-07,
+      "loss": 2.1068,
+      "step": 3910
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 8.306926189754372e-07,
+      "loss": 2.1134,
+      "step": 3915
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 7.535112260847799e-07,
+      "loss": 2.1227,
+      "step": 3920
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 6.800855481500445e-07,
+      "loss": 2.1204,
+      "step": 3925
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.126953125,
+      "learning_rate": 6.104169692153105e-07,
+      "loss": 2.1101,
+      "step": 3930
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 5.44506802504774e-07,
+      "loss": 2.1111,
+      "step": 3935
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.125,
+      "learning_rate": 4.823562903982337e-07,
+      "loss": 2.1068,
+      "step": 3940
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 4.239666044074442e-07,
+      "loss": 2.1416,
+      "step": 3945
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 3.693388451541102e-07,
+      "loss": 2.1217,
+      "step": 3950
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 3.1847404234923715e-07,
+      "loss": 2.1201,
+      "step": 3955
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 2.713731547735687e-07,
+      "loss": 2.1135,
+      "step": 3960
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.125,
+      "learning_rate": 2.280370702596013e-07,
+      "loss": 2.0992,
+      "step": 3965
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 1.8846660567484186e-07,
+      "loss": 2.1167,
+      "step": 3970
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 1.5266250690635363e-07,
+      "loss": 2.0971,
+      "step": 3975
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 1.2062544884683391e-07,
+      "loss": 2.1277,
+      "step": 3980
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.125,
+      "learning_rate": 9.235603538171322e-08,
+      "loss": 2.1004,
+      "step": 3985
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 6.785479937789773e-08,
+      "loss": 2.0967,
+      "step": 3990
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 4.712220267366618e-08,
+      "loss": 2.1211,
+      "step": 3995
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.126953125,
+      "learning_rate": 3.015863607003233e-08,
+      "loss": 2.1347,
+      "step": 4000
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.125,
+      "learning_rate": 1.69644193232843e-08,
+      "loss": 2.1306,
+      "step": 4005
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 7.539801139011538e-09,
+      "loss": 2.1013,
+      "step": 4010
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 1.884959167419709e-09,
+      "loss": 2.099,
+      "step": 4015
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0,
+      "loss": 2.0989,
+      "step": 4020
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 2.1637070178985596,
+      "eval_runtime": 187.0083,
+      "eval_samples_per_second": 25.785,
+      "eval_steps_per_second": 3.224,
+      "step": 4020
+    },
+    {
+      "epoch": 3.0,
+      "step": 4020,
+      "total_flos": 9.114826330351862e+17,
+      "train_loss": 2.1598364234563725,
+      "train_runtime": 32657.7671,
+      "train_samples_per_second": 7.879,
+      "train_steps_per_second": 0.123
     }
   ],
   "logging_steps": 5,
-  "max_steps": 670,
+  "max_steps": 4020,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 1,
+  "num_train_epochs": 3,
   "save_steps": 100,
-  "total_flos": 3.038889647317975e+17,
+  "total_flos": 9.114826330351862e+17,
   "train_batch_size": 32,
   "trial_name": null,
   "trial_params": null