diff --git "a/checkpoint-6139/trainer_state.json" "b/checkpoint-6139/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-6139/trainer_state.json"
@@ -0,0 +1,4428 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.999959278413487,
+  "eval_steps": 500,
+  "global_step": 6139,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0001628863460520422,
+      "eval_loss": 2.4029905796051025,
+      "eval_runtime": 101.0475,
+      "eval_samples_per_second": 25.582,
+      "eval_steps_per_second": 25.582,
+      "step": 1
+    },
+    {
+      "epoch": 0.0016288634605204219,
+      "grad_norm": 24.467580795288086,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.7479,
+      "step": 10
+    },
+    {
+      "epoch": 0.0032577269210408437,
+      "grad_norm": 25.53505516052246,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.7216,
+      "step": 20
+    },
+    {
+      "epoch": 0.004886590381561266,
+      "grad_norm": 18.23355484008789,
+      "learning_rate": 6e-06,
+      "loss": 1.5159,
+      "step": 30
+    },
+    {
+      "epoch": 0.006515453842081687,
+      "grad_norm": 12.464448928833008,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.4322,
+      "step": 40
+    },
+    {
+      "epoch": 0.00814431730260211,
+      "grad_norm": 10.985918998718262,
+      "learning_rate": 1e-05,
+      "loss": 1.3426,
+      "step": 50
+    },
+    {
+      "epoch": 0.009773180763122532,
+      "grad_norm": 9.040105819702148,
+      "learning_rate": 1.2e-05,
+      "loss": 1.314,
+      "step": 60
+    },
+    {
+      "epoch": 0.011402044223642953,
+      "grad_norm": 10.273420333862305,
+      "learning_rate": 1.4e-05,
+      "loss": 1.2463,
+      "step": 70
+    },
+    {
+      "epoch": 0.013030907684163375,
+      "grad_norm": 11.754436492919922,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.2561,
+      "step": 80
+    },
+    {
+      "epoch": 0.014659771144683796,
+      "grad_norm": 12.387304306030273,
+      "learning_rate": 1.8e-05,
+      "loss": 1.275,
+      "step": 90
+    },
+    {
+      "epoch": 0.01628863460520422,
+      "grad_norm": 9.28602409362793,
+      "learning_rate": 2e-05,
+      "loss": 1.3315,
+      "step": 100
+    },
+    {
+      "epoch": 0.01791749806572464,
+      "grad_norm": 9.231125831604004,
+      "learning_rate": 1.9999864687252914e-05,
+      "loss": 1.3106,
+      "step": 110
+    },
+    {
+      "epoch": 0.019546361526245063,
+      "grad_norm": 19.837526321411133,
+      "learning_rate": 1.9999458752673554e-05,
+      "loss": 1.342,
+      "step": 120
+    },
+    {
+      "epoch": 0.021175224986765485,
+      "grad_norm": 11.052694320678711,
+      "learning_rate": 1.9998782207247556e-05,
+      "loss": 1.2678,
+      "step": 130
+    },
+    {
+      "epoch": 0.022804088447285906,
+      "grad_norm": 14.448543548583984,
+      "learning_rate": 1.9997835069283954e-05,
+      "loss": 1.262,
+      "step": 140
+    },
+    {
+      "epoch": 0.024432951907806328,
+      "grad_norm": 8.338994979858398,
+      "learning_rate": 1.999661736441472e-05,
+      "loss": 1.2932,
+      "step": 150
+    },
+    {
+      "epoch": 0.02606181536832675,
+      "grad_norm": 10.267782211303711,
+      "learning_rate": 1.9995129125594058e-05,
+      "loss": 1.2789,
+      "step": 160
+    },
+    {
+      "epoch": 0.02769067882884717,
+      "grad_norm": 17.692113876342773,
+      "learning_rate": 1.999337039309749e-05,
+      "loss": 1.2546,
+      "step": 170
+    },
+    {
+      "epoch": 0.029319542289367593,
+      "grad_norm": 11.050065040588379,
+      "learning_rate": 1.9991341214520816e-05,
+      "loss": 1.2649,
+      "step": 180
+    },
+    {
+      "epoch": 0.030948405749888015,
+      "grad_norm": 11.198172569274902,
+      "learning_rate": 1.9989041644778773e-05,
+      "loss": 1.2018,
+      "step": 190
+    },
+    {
+      "epoch": 0.03257726921040844,
+      "grad_norm": 8.730816841125488,
+      "learning_rate": 1.9986471746103584e-05,
+      "loss": 1.3364,
+      "step": 200
+    },
+    {
+      "epoch": 0.03420613267092886,
+      "grad_norm": 12.193893432617188,
+      "learning_rate": 1.998363158804326e-05,
+      "loss": 1.3358,
+      "step": 210
+    },
+    {
+      "epoch": 0.03583499613144928,
+      "grad_norm": 12.82657241821289,
+      "learning_rate": 1.9980521247459714e-05,
+      "loss": 1.3936,
+      "step": 220
+    },
+    {
+      "epoch": 0.037463859591969705,
+      "grad_norm": 10.334324836730957,
+      "learning_rate": 1.9977140808526696e-05,
+      "loss": 1.2811,
+      "step": 230
+    },
+    {
+      "epoch": 0.039092723052490126,
+      "grad_norm": 10.822787284851074,
+      "learning_rate": 1.99734903627275e-05,
+      "loss": 1.3964,
+      "step": 240
+    },
+    {
+      "epoch": 0.04072158651301055,
+      "grad_norm": 10.038861274719238,
+      "learning_rate": 1.9969570008852498e-05,
+      "loss": 1.3229,
+      "step": 250
+    },
+    {
+      "epoch": 0.04235044997353097,
+      "grad_norm": 10.05347728729248,
+      "learning_rate": 1.9965379852996455e-05,
+      "loss": 1.2529,
+      "step": 260
+    },
+    {
+      "epoch": 0.04397931343405139,
+      "grad_norm": 15.111456871032715,
+      "learning_rate": 1.996092000855568e-05,
+      "loss": 1.3372,
+      "step": 270
+    },
+    {
+      "epoch": 0.04560817689457181,
+      "grad_norm": 8.95826244354248,
+      "learning_rate": 1.9956190596224923e-05,
+      "loss": 1.3154,
+      "step": 280
+    },
+    {
+      "epoch": 0.047237040355092234,
+      "grad_norm": 8.862021446228027,
+      "learning_rate": 1.9951191743994146e-05,
+      "loss": 1.2578,
+      "step": 290
+    },
+    {
+      "epoch": 0.048865903815612656,
+      "grad_norm": 12.974872589111328,
+      "learning_rate": 1.9945923587145032e-05,
+      "loss": 1.2848,
+      "step": 300
+    },
+    {
+      "epoch": 0.05049476727613308,
+      "grad_norm": 8.246659278869629,
+      "learning_rate": 1.994038626824734e-05,
+      "loss": 1.2349,
+      "step": 310
+    },
+    {
+      "epoch": 0.0521236307366535,
+      "grad_norm": 10.3037748336792,
+      "learning_rate": 1.993457993715503e-05,
+      "loss": 1.2731,
+      "step": 320
+    },
+    {
+      "epoch": 0.05375249419717392,
+      "grad_norm": 9.140987396240234,
+      "learning_rate": 1.992850475100223e-05,
+      "loss": 1.2673,
+      "step": 330
+    },
+    {
+      "epoch": 0.05538135765769434,
+      "grad_norm": 9.048047065734863,
+      "learning_rate": 1.992216087419896e-05,
+      "loss": 1.2842,
+      "step": 340
+    },
+    {
+      "epoch": 0.057010221118214764,
+      "grad_norm": 8.832723617553711,
+      "learning_rate": 1.99155484784267e-05,
+      "loss": 1.227,
+      "step": 350
+    },
+    {
+      "epoch": 0.058639084578735186,
+      "grad_norm": 8.928304672241211,
+      "learning_rate": 1.9908667742633742e-05,
+      "loss": 1.3681,
+      "step": 360
+    },
+    {
+      "epoch": 0.06026794803925561,
+      "grad_norm": 9.978281021118164,
+      "learning_rate": 1.990151885303034e-05,
+      "loss": 1.1123,
+      "step": 370
+    },
+    {
+      "epoch": 0.06189681149977603,
+      "grad_norm": 8.430088996887207,
+      "learning_rate": 1.989410200308366e-05,
+      "loss": 1.263,
+      "step": 380
+    },
+    {
+      "epoch": 0.06352567496029646,
+      "grad_norm": 11.750625610351562,
+      "learning_rate": 1.9886417393512584e-05,
+      "loss": 1.255,
+      "step": 390
+    },
+    {
+      "epoch": 0.06515453842081688,
+      "grad_norm": 10.856910705566406,
+      "learning_rate": 1.987846523228223e-05,
+      "loss": 1.2889,
+      "step": 400
+    },
+    {
+      "epoch": 0.0667834018813373,
+      "grad_norm": 10.42456340789795,
+      "learning_rate": 1.9870245734598358e-05,
+      "loss": 1.3936,
+      "step": 410
+    },
+    {
+      "epoch": 0.06841226534185772,
+      "grad_norm": 9.490607261657715,
+      "learning_rate": 1.986175912290153e-05,
+      "loss": 1.3335,
+      "step": 420
+    },
+    {
+      "epoch": 0.07004112880237814,
+      "grad_norm": 8.353849411010742,
+      "learning_rate": 1.985300562686109e-05,
+      "loss": 1.2302,
+      "step": 430
+    },
+    {
+      "epoch": 0.07166999226289857,
+      "grad_norm": 13.212517738342285,
+      "learning_rate": 1.9843985483368968e-05,
+      "loss": 1.38,
+      "step": 440
+    },
+    {
+      "epoch": 0.07329885572341899,
+      "grad_norm": 6.969715118408203,
+      "learning_rate": 1.9834698936533226e-05,
+      "loss": 1.2775,
+      "step": 450
+    },
+    {
+      "epoch": 0.07492771918393941,
+      "grad_norm": 7.460076332092285,
+      "learning_rate": 1.9825146237671513e-05,
+      "loss": 1.4045,
+      "step": 460
+    },
+    {
+      "epoch": 0.07655658264445983,
+      "grad_norm": 8.584343910217285,
+      "learning_rate": 1.9815327645304204e-05,
+      "loss": 1.2884,
+      "step": 470
+    },
+    {
+      "epoch": 0.07818544610498025,
+      "grad_norm": 8.771677017211914,
+      "learning_rate": 1.9805243425147448e-05,
+      "loss": 1.3839,
+      "step": 480
+    },
+    {
+      "epoch": 0.07981430956550067,
+      "grad_norm": 10.828780174255371,
+      "learning_rate": 1.9794893850105942e-05,
+      "loss": 1.3274,
+      "step": 490
+    },
+    {
+      "epoch": 0.0814431730260211,
+      "grad_norm": 8.709300994873047,
+      "learning_rate": 1.978427920026558e-05,
+      "loss": 1.2572,
+      "step": 500
+    },
+    {
+      "epoch": 0.0814431730260211,
+      "eval_loss": 1.1935244798660278,
+      "eval_runtime": 101.9438,
+      "eval_samples_per_second": 25.357,
+      "eval_steps_per_second": 25.357,
+      "step": 500
+    },
+    {
+      "epoch": 0.08307203648654152,
+      "grad_norm": 9.961617469787598,
+      "learning_rate": 1.977339976288584e-05,
+      "loss": 1.382,
+      "step": 510
+    },
+    {
+      "epoch": 0.08470089994706194,
+      "grad_norm": 9.85478401184082,
+      "learning_rate": 1.9762255832392046e-05,
+      "loss": 1.3768,
+      "step": 520
+    },
+    {
+      "epoch": 0.08632976340758236,
+      "grad_norm": 9.125312805175781,
+      "learning_rate": 1.975084771036736e-05,
+      "loss": 1.2088,
+      "step": 530
+    },
+    {
+      "epoch": 0.08795862686810278,
+      "grad_norm": 7.915559768676758,
+      "learning_rate": 1.973917570554464e-05,
+      "loss": 1.2951,
+      "step": 540
+    },
+    {
+      "epoch": 0.0895874903286232,
+      "grad_norm": 9.664942741394043,
+      "learning_rate": 1.9727240133798106e-05,
+      "loss": 1.3305,
+      "step": 550
+    },
+    {
+      "epoch": 0.09121635378914363,
+      "grad_norm": 10.48803997039795,
+      "learning_rate": 1.9715041318134756e-05,
+      "loss": 1.4256,
+      "step": 560
+    },
+    {
+      "epoch": 0.09284521724966405,
+      "grad_norm": 10.634197235107422,
+      "learning_rate": 1.9702579588685634e-05,
+      "loss": 1.2912,
+      "step": 570
+    },
+    {
+      "epoch": 0.09447408071018447,
+      "grad_norm": 9.283407211303711,
+      "learning_rate": 1.968985528269692e-05,
+      "loss": 1.3337,
+      "step": 580
+    },
+    {
+      "epoch": 0.09610294417070489,
+      "grad_norm": 9.688226699829102,
+      "learning_rate": 1.9676868744520768e-05,
+      "loss": 1.3231,
+      "step": 590
+    },
+    {
+      "epoch": 0.09773180763122531,
+      "grad_norm": 10.292962074279785,
+      "learning_rate": 1.9663620325606005e-05,
+      "loss": 1.2227,
+      "step": 600
+    },
+    {
+      "epoch": 0.09936067109174573,
+      "grad_norm": 5.677855014801025,
+      "learning_rate": 1.965011038448863e-05,
+      "loss": 1.4159,
+      "step": 610
+    },
+    {
+      "epoch": 0.10098953455226616,
+      "grad_norm": 7.153832912445068,
+      "learning_rate": 1.963633928678209e-05,
+      "loss": 1.3664,
+      "step": 620
+    },
+    {
+      "epoch": 0.10261839801278658,
+      "grad_norm": 12.980141639709473,
+      "learning_rate": 1.9622307405167395e-05,
+      "loss": 1.2955,
+      "step": 630
+    },
+    {
+      "epoch": 0.104247261473307,
+      "grad_norm": 10.464163780212402,
+      "learning_rate": 1.9608015119383036e-05,
+      "loss": 1.2868,
+      "step": 640
+    },
+    {
+      "epoch": 0.10587612493382742,
+      "grad_norm": 11.611323356628418,
+      "learning_rate": 1.9593462816214698e-05,
+      "loss": 1.2963,
+      "step": 650
+    },
+    {
+      "epoch": 0.10750498839434784,
+      "grad_norm": 6.969916820526123,
+      "learning_rate": 1.9578650889484815e-05,
+      "loss": 1.2765,
+      "step": 660
+    },
+    {
+      "epoch": 0.10913385185486826,
+      "grad_norm": 10.346363067626953,
+      "learning_rate": 1.9563579740041884e-05,
+      "loss": 1.3072,
+      "step": 670
+    },
+    {
+      "epoch": 0.11076271531538869,
+      "grad_norm": 10.077250480651855,
+      "learning_rate": 1.9548249775749623e-05,
+      "loss": 1.3099,
+      "step": 680
+    },
+    {
+      "epoch": 0.1123915787759091,
+      "grad_norm": 10.486137390136719,
+      "learning_rate": 1.9532661411475955e-05,
+      "loss": 1.3504,
+      "step": 690
+    },
+    {
+      "epoch": 0.11402044223642953,
+      "grad_norm": 6.885483264923096,
+      "learning_rate": 1.9516815069081758e-05,
+      "loss": 1.2268,
+      "step": 700
+    },
+    {
+      "epoch": 0.11564930569694995,
+      "grad_norm": 7.814432144165039,
+      "learning_rate": 1.9500711177409456e-05,
+      "loss": 1.394,
+      "step": 710
+    },
+    {
+      "epoch": 0.11727816915747037,
+      "grad_norm": 13.40298080444336,
+      "learning_rate": 1.948435017227141e-05,
+      "loss": 1.2634,
+      "step": 720
+    },
+    {
+      "epoch": 0.1189070326179908,
+      "grad_norm": 8.984935760498047,
+      "learning_rate": 1.9467732496438137e-05,
+      "loss": 1.2645,
+      "step": 730
+    },
+    {
+      "epoch": 0.12053589607851121,
+      "grad_norm": 8.895575523376465,
+      "learning_rate": 1.9450858599626304e-05,
+      "loss": 1.2724,
+      "step": 740
+    },
+    {
+      "epoch": 0.12216475953903164,
+      "grad_norm": 9.734760284423828,
+      "learning_rate": 1.9433728938486576e-05,
+      "loss": 1.3362,
+      "step": 750
+    },
+    {
+      "epoch": 0.12379362299955206,
+      "grad_norm": 11.392585754394531,
+      "learning_rate": 1.941634397659126e-05,
+      "loss": 1.2574,
+      "step": 760
+    },
+    {
+      "epoch": 0.12542248646007248,
+      "grad_norm": 7.931302070617676,
+      "learning_rate": 1.9398704184421745e-05,
+      "loss": 1.3003,
+      "step": 770
+    },
+    {
+      "epoch": 0.12705134992059292,
+      "grad_norm": 7.866138458251953,
+      "learning_rate": 1.9380810039355776e-05,
+      "loss": 1.2672,
+      "step": 780
+    },
+    {
+      "epoch": 0.12868021338111332,
+      "grad_norm": 12.06697940826416,
+      "learning_rate": 1.936266202565454e-05,
+      "loss": 1.4121,
+      "step": 790
+    },
+    {
+      "epoch": 0.13030907684163376,
+      "grad_norm": 10.089153289794922,
+      "learning_rate": 1.9344260634449556e-05,
+      "loss": 1.3314,
+      "step": 800
+    },
+    {
+      "epoch": 0.13193794030215417,
+      "grad_norm": 7.83293342590332,
+      "learning_rate": 1.9325606363729378e-05,
+      "loss": 1.332,
+      "step": 810
+    },
+    {
+      "epoch": 0.1335668037626746,
+      "grad_norm": 18.291975021362305,
+      "learning_rate": 1.930669971832613e-05,
+      "loss": 1.2154,
+      "step": 820
+    },
+    {
+      "epoch": 0.135195667223195,
+      "grad_norm": 9.173296928405762,
+      "learning_rate": 1.9287541209901842e-05,
+      "loss": 1.3371,
+      "step": 830
+    },
+    {
+      "epoch": 0.13682453068371545,
+      "grad_norm": 7.5841875076293945,
+      "learning_rate": 1.9268131356934592e-05,
+      "loss": 1.3549,
+      "step": 840
+    },
+    {
+      "epoch": 0.13845339414423585,
+      "grad_norm": 9.693364143371582,
+      "learning_rate": 1.924847068470449e-05,
+      "loss": 1.2978,
+      "step": 850
+    },
+    {
+      "epoch": 0.1400822576047563,
+      "grad_norm": 9.127781867980957,
+      "learning_rate": 1.9228559725279444e-05,
+      "loss": 1.261,
+      "step": 860
+    },
+    {
+      "epoch": 0.1417111210652767,
+      "grad_norm": 10.408056259155273,
+      "learning_rate": 1.9208399017500773e-05,
+      "loss": 1.2478,
+      "step": 870
+    },
+    {
+      "epoch": 0.14333998452579713,
+      "grad_norm": 9.875020980834961,
+      "learning_rate": 1.918798910696864e-05,
+      "loss": 1.3205,
+      "step": 880
+    },
+    {
+      "epoch": 0.14496884798631754,
+      "grad_norm": 8.818717956542969,
+      "learning_rate": 1.916733054602725e-05,
+      "loss": 1.389,
+      "step": 890
+    },
+    {
+      "epoch": 0.14659771144683797,
+      "grad_norm": 13.227215766906738,
+      "learning_rate": 1.9146423893749924e-05,
+      "loss": 1.3243,
+      "step": 900
+    },
+    {
+      "epoch": 0.14822657490735838,
+      "grad_norm": 7.599000930786133,
+      "learning_rate": 1.9125269715923983e-05,
+      "loss": 1.1711,
+      "step": 910
+    },
+    {
+      "epoch": 0.14985543836787882,
+      "grad_norm": 9.84504508972168,
+      "learning_rate": 1.910386858503541e-05,
+      "loss": 1.2145,
+      "step": 920
+    },
+    {
+      "epoch": 0.15148430182839923,
+      "grad_norm": 7.403562068939209,
+      "learning_rate": 1.908222108025336e-05,
+      "loss": 1.2809,
+      "step": 930
+    },
+    {
+      "epoch": 0.15311316528891966,
+      "grad_norm": 12.983716011047363,
+      "learning_rate": 1.9060327787414498e-05,
+      "loss": 1.2949,
+      "step": 940
+    },
+    {
+      "epoch": 0.15474202874944007,
+      "grad_norm": 8.185487747192383,
+      "learning_rate": 1.9038189299007154e-05,
+      "loss": 1.3662,
+      "step": 950
+    },
+    {
+      "epoch": 0.1563708922099605,
+      "grad_norm": 9.041784286499023,
+      "learning_rate": 1.901580621415526e-05,
+      "loss": 1.2835,
+      "step": 960
+    },
+    {
+      "epoch": 0.1579997556704809,
+      "grad_norm": 7.92235803604126,
+      "learning_rate": 1.899317913860215e-05,
+      "loss": 1.3625,
+      "step": 970
+    },
+    {
+      "epoch": 0.15962861913100135,
+      "grad_norm": 15.111337661743164,
+      "learning_rate": 1.8970308684694186e-05,
+      "loss": 1.2816,
+      "step": 980
+    },
+    {
+      "epoch": 0.16125748259152176,
+      "grad_norm": 10.149383544921875,
+      "learning_rate": 1.894719547136415e-05,
+      "loss": 1.2481,
+      "step": 990
+    },
+    {
+      "epoch": 0.1628863460520422,
+      "grad_norm": 10.364059448242188,
+      "learning_rate": 1.8923840124114517e-05,
+      "loss": 1.3061,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1628863460520422,
+      "eval_loss": 1.1865288019180298,
+      "eval_runtime": 101.3159,
+      "eval_samples_per_second": 25.514,
+      "eval_steps_per_second": 25.514,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1645152095125626,
+      "grad_norm": 8.624238014221191,
+      "learning_rate": 1.8900243275000532e-05,
+      "loss": 1.2933,
+      "step": 1010
+    },
+    {
+      "epoch": 0.16614407297308303,
+      "grad_norm": 9.401992797851562,
+      "learning_rate": 1.8876405562613088e-05,
+      "loss": 1.244,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16777293643360344,
+      "grad_norm": 10.441336631774902,
+      "learning_rate": 1.8852327632061457e-05,
+      "loss": 1.277,
+      "step": 1030
+    },
+    {
+      "epoch": 0.16940179989412388,
+      "grad_norm": 11.799003601074219,
+      "learning_rate": 1.8828010134955822e-05,
+      "loss": 1.33,
+      "step": 1040
+    },
+    {
+      "epoch": 0.17103066335464429,
+      "grad_norm": 9.582467079162598,
+      "learning_rate": 1.8803453729389648e-05,
+      "loss": 1.4002,
+      "step": 1050
+    },
+    {
+      "epoch": 0.17265952681516472,
+      "grad_norm": 8.40396499633789,
+      "learning_rate": 1.8778659079921877e-05,
+      "loss": 1.3458,
+      "step": 1060
+    },
+    {
+      "epoch": 0.17428839027568513,
+      "grad_norm": 10.204269409179688,
+      "learning_rate": 1.8753626857558935e-05,
+      "loss": 1.3259,
+      "step": 1070
+    },
+    {
+      "epoch": 0.17591725373620556,
+      "grad_norm": 6.9873948097229,
+      "learning_rate": 1.8728357739736578e-05,
+      "loss": 1.3199,
+      "step": 1080
+    },
+    {
+      "epoch": 0.17754611719672597,
+      "grad_norm": 9.196039199829102,
+      "learning_rate": 1.8702852410301556e-05,
+      "loss": 1.337,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1791749806572464,
+      "grad_norm": 8.873251914978027,
+      "learning_rate": 1.86771115594931e-05,
+      "loss": 1.2909,
+      "step": 1100
+    },
+    {
+      "epoch": 0.18080384411776682,
+      "grad_norm": 9.222848892211914,
+      "learning_rate": 1.865113588392427e-05,
+      "loss": 1.2325,
+      "step": 1110
+    },
+    {
+      "epoch": 0.18243270757828725,
+      "grad_norm": 7.872706413269043,
+      "learning_rate": 1.8624926086563057e-05,
+      "loss": 1.2469,
+      "step": 1120
+    },
+    {
+      "epoch": 0.18406157103880766,
+      "grad_norm": 9.13894271850586,
+      "learning_rate": 1.85984828767134e-05,
+      "loss": 1.2889,
+      "step": 1130
+    },
+    {
+      "epoch": 0.1856904344993281,
+      "grad_norm": 7.499070167541504,
+      "learning_rate": 1.8571806969995982e-05,
+      "loss": 1.3696,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1873192979598485,
+      "grad_norm": 10.489768028259277,
+      "learning_rate": 1.854489908832884e-05,
+      "loss": 1.2415,
+      "step": 1150
+    },
+    {
+      "epoch": 0.18894816142036894,
+      "grad_norm": 9.07557201385498,
+      "learning_rate": 1.8517759959907845e-05,
+      "loss": 1.307,
+      "step": 1160
+    },
+    {
+      "epoch": 0.19057702488088937,
+      "grad_norm": 8.078954696655273,
+      "learning_rate": 1.849039031918701e-05,
+      "loss": 1.2854,
+      "step": 1170
+    },
+    {
+      "epoch": 0.19220588834140978,
+      "grad_norm": 8.329951286315918,
+      "learning_rate": 1.846279090685859e-05,
+      "loss": 1.2446,
+      "step": 1180
+    },
+    {
+      "epoch": 0.19383475180193022,
+      "grad_norm": 9.939054489135742,
+      "learning_rate": 1.8434962469833036e-05,
+      "loss": 1.4163,
+      "step": 1190
+    },
+    {
+      "epoch": 0.19546361526245062,
+      "grad_norm": 9.757793426513672,
+      "learning_rate": 1.8406905761218815e-05,
+      "loss": 1.1905,
+      "step": 1200
+    },
+    {
+      "epoch": 0.19709247872297106,
+      "grad_norm": 7.9107489585876465,
+      "learning_rate": 1.8378621540301976e-05,
+      "loss": 1.2955,
+      "step": 1210
+    },
+    {
+      "epoch": 0.19872134218349147,
+      "grad_norm": 9.854551315307617,
+      "learning_rate": 1.835011057252565e-05,
+      "loss": 1.2293,
+      "step": 1220
+    },
+    {
+      "epoch": 0.2003502056440119,
+      "grad_norm": 10.388916969299316,
+      "learning_rate": 1.8321373629469313e-05,
+      "loss": 1.337,
+      "step": 1230
+    },
+    {
+      "epoch": 0.2019790691045323,
+      "grad_norm": 9.081336975097656,
+      "learning_rate": 1.8292411488827906e-05,
+      "loss": 1.2708,
+      "step": 1240
+    },
+    {
+      "epoch": 0.20360793256505275,
+      "grad_norm": 10.85607624053955,
+      "learning_rate": 1.826322493439079e-05,
+      "loss": 1.2249,
+      "step": 1250
+    },
+    {
+      "epoch": 0.20523679602557315,
+      "grad_norm": 10.35145378112793,
+      "learning_rate": 1.823381475602054e-05,
+      "loss": 1.3433,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2068656594860936,
+      "grad_norm": 12.942462921142578,
+      "learning_rate": 1.8204181749631557e-05,
+      "loss": 1.3062,
+      "step": 1270
+    },
+    {
+      "epoch": 0.208494522946614,
+      "grad_norm": 8.48989200592041,
+      "learning_rate": 1.8174326717168547e-05,
+      "loss": 1.2474,
+      "step": 1280
+    },
+    {
+      "epoch": 0.21012338640713443,
+      "grad_norm": 7.585806846618652,
+      "learning_rate": 1.8144250466584794e-05,
+      "loss": 1.3294,
+      "step": 1290
+    },
+    {
+      "epoch": 0.21175224986765484,
+      "grad_norm": 9.783578872680664,
+      "learning_rate": 1.8113953811820322e-05,
+      "loss": 1.308,
+      "step": 1300
+    },
+    {
+      "epoch": 0.21338111332817528,
+      "grad_norm": 8.406821250915527,
+      "learning_rate": 1.8083437572779842e-05,
+      "loss": 1.2757,
+      "step": 1310
+    },
+    {
+      "epoch": 0.21500997678869568,
+      "grad_norm": 11.778732299804688,
+      "learning_rate": 1.8052702575310588e-05,
+      "loss": 1.2255,
+      "step": 1320
+    },
+    {
+      "epoch": 0.21663884024921612,
+      "grad_norm": 8.81125259399414,
+      "learning_rate": 1.802174965117994e-05,
+      "loss": 1.2558,
+      "step": 1330
+    },
+    {
+      "epoch": 0.21826770370973653,
+      "grad_norm": 8.534640312194824,
+      "learning_rate": 1.7990579638052944e-05,
+      "loss": 1.3359,
+      "step": 1340
+    },
+    {
+      "epoch": 0.21989656717025696,
+      "grad_norm": 7.685283184051514,
+      "learning_rate": 1.795919337946962e-05,
+      "loss": 1.2347,
+      "step": 1350
+    },
+    {
+      "epoch": 0.22152543063077737,
+      "grad_norm": 9.883221626281738,
+      "learning_rate": 1.7927591724822132e-05,
+      "loss": 1.2605,
+      "step": 1360
+    },
+    {
+      "epoch": 0.2231542940912978,
+      "grad_norm": 6.772229194641113,
+      "learning_rate": 1.7895775529331835e-05,
+      "loss": 1.2475,
+      "step": 1370
+    },
+    {
+      "epoch": 0.2247831575518182,
+      "grad_norm": 6.82761287689209,
+      "learning_rate": 1.7863745654026078e-05,
+      "loss": 1.2686,
+      "step": 1380
+    },
+    {
+      "epoch": 0.22641202101233865,
+      "grad_norm": 13.5228853225708,
+      "learning_rate": 1.7831502965714958e-05,
+      "loss": 1.2544,
+      "step": 1390
+    },
+    {
+      "epoch": 0.22804088447285906,
+      "grad_norm": 7.083252429962158,
+      "learning_rate": 1.779904833696781e-05,
+      "loss": 1.2435,
+      "step": 1400
+    },
+    {
+      "epoch": 0.2296697479333795,
+      "grad_norm": 9.345582962036133,
+      "learning_rate": 1.7766382646089635e-05,
+      "loss": 1.3229,
+      "step": 1410
+    },
+    {
+      "epoch": 0.2312986113938999,
+      "grad_norm": 10.011900901794434,
+      "learning_rate": 1.77335067770973e-05,
+      "loss": 1.3074,
+      "step": 1420
+    },
+    {
+      "epoch": 0.23292747485442034,
+      "grad_norm": 8.366225242614746,
+      "learning_rate": 1.770042161969564e-05,
+      "loss": 1.1657,
+      "step": 1430
+    },
+    {
+      "epoch": 0.23455633831494074,
+      "grad_norm": 13.709329605102539,
+      "learning_rate": 1.7667128069253362e-05,
+      "loss": 1.3257,
+      "step": 1440
+    },
+    {
+      "epoch": 0.23618520177546118,
+      "grad_norm": 9.527036666870117,
+      "learning_rate": 1.763362702677882e-05,
+      "loss": 1.1838,
+      "step": 1450
+    },
+    {
+      "epoch": 0.2378140652359816,
+      "grad_norm": 10.506624221801758,
+      "learning_rate": 1.759991939889562e-05,
+      "loss": 1.3324,
+      "step": 1460
+    },
+    {
+      "epoch": 0.23944292869650202,
+      "grad_norm": 9.029693603515625,
+      "learning_rate": 1.7566006097818123e-05,
+      "loss": 1.2102,
+      "step": 1470
+    },
+    {
+      "epoch": 0.24107179215702243,
+      "grad_norm": 9.59033203125,
+      "learning_rate": 1.7531888041326715e-05,
+      "loss": 1.3553,
+      "step": 1480
+    },
+    {
+      "epoch": 0.24270065561754287,
+      "grad_norm": 9.097341537475586,
+      "learning_rate": 1.7497566152742975e-05,
+      "loss": 1.3165,
+      "step": 1490
+    },
+    {
+      "epoch": 0.24432951907806327,
+      "grad_norm": 19.68447494506836,
+      "learning_rate": 1.7463041360904714e-05,
+      "loss": 1.2733,
+      "step": 1500
+    },
+    {
+      "epoch": 0.24432951907806327,
+      "eval_loss": 1.1864172220230103,
+      "eval_runtime": 101.8915,
+      "eval_samples_per_second": 25.37,
+      "eval_steps_per_second": 25.37,
+      "step": 1500
+    },
+    {
+      "epoch": 0.2459583825385837,
+      "grad_norm": 9.446989059448242,
+      "learning_rate": 1.742831460014082e-05,
+      "loss": 1.1618,
+      "step": 1510
+    },
+    {
+      "epoch": 0.24758724599910412,
+      "grad_norm": 7.566542148590088,
+      "learning_rate": 1.7393386810245968e-05,
+      "loss": 1.271,
+      "step": 1520
+    },
+    {
+      "epoch": 0.24921610945962455,
+      "grad_norm": 12.757163047790527,
+      "learning_rate": 1.7358258936455203e-05,
+      "loss": 1.2472,
+      "step": 1530
+    },
+    {
+      "epoch": 0.25084497292014496,
+      "grad_norm": 6.86335563659668,
+      "learning_rate": 1.7322931929418338e-05,
+      "loss": 1.3589,
+      "step": 1540
+    },
+    {
+      "epoch": 0.2524738363806654,
+      "grad_norm": 8.989468574523926,
+      "learning_rate": 1.7287406745174253e-05,
+      "loss": 1.246,
+      "step": 1550
+    },
+    {
+      "epoch": 0.25410269984118583,
+      "grad_norm": 9.14529037475586,
+      "learning_rate": 1.7251684345125e-05,
+      "loss": 1.1678,
+      "step": 1560
+    },
+    {
+      "epoch": 0.2557315633017062,
+      "grad_norm": 9.417140007019043,
+      "learning_rate": 1.7215765696009795e-05,
+      "loss": 1.2109,
+      "step": 1570
+    },
+    {
+      "epoch": 0.25736042676222665,
+      "grad_norm": 8.230705261230469,
+      "learning_rate": 1.7179651769878854e-05,
+      "loss": 1.2815,
+      "step": 1580
+    },
+    {
+      "epoch": 0.2589892902227471,
+      "grad_norm": 7.2801289558410645,
+      "learning_rate": 1.7143343544067094e-05,
+      "loss": 1.3492,
+      "step": 1590
+    },
+    {
+      "epoch": 0.2606181536832675,
+      "grad_norm": 6.567676544189453,
+      "learning_rate": 1.7106842001167664e-05,
+      "loss": 1.2145,
+      "step": 1600
+    },
+    {
+      "epoch": 0.2622470171437879,
+      "grad_norm": 7.0790910720825195,
+      "learning_rate": 1.7070148129005373e-05,
+      "loss": 1.2142,
+      "step": 1610
+    },
+    {
+      "epoch": 0.26387588060430833,
+      "grad_norm": 11.59786319732666,
+      "learning_rate": 1.7033262920609947e-05,
+      "loss": 1.2607,
+      "step": 1620
+    },
+    {
+      "epoch": 0.26550474406482877,
+      "grad_norm": 7.409655570983887,
+      "learning_rate": 1.699618737418917e-05,
+      "loss": 1.246,
+      "step": 1630
+    },
+    {
+      "epoch": 0.2671336075253492,
+      "grad_norm": 7.234554767608643,
+      "learning_rate": 1.6958922493101844e-05,
+      "loss": 1.2376,
+      "step": 1640
+    },
+    {
+      "epoch": 0.2687624709858696,
+      "grad_norm": 8.587789535522461,
+      "learning_rate": 1.6921469285830654e-05,
+      "loss": 1.312,
+      "step": 1650
+    },
+    {
+      "epoch": 0.27039133444639,
+      "grad_norm": 8.513998031616211,
+      "learning_rate": 1.688382876595487e-05,
+      "loss": 1.3469,
+      "step": 1660
+    },
+    {
+      "epoch": 0.27202019790691045,
+      "grad_norm": 7.039551258087158,
+      "learning_rate": 1.684600195212293e-05,
+      "loss": 1.3468,
+      "step": 1670
+    },
+    {
+      "epoch": 0.2736490613674309,
+      "grad_norm": 11.176895141601562,
+      "learning_rate": 1.6807989868024845e-05,
+      "loss": 1.419,
+      "step": 1680
+    },
+    {
+      "epoch": 0.27527792482795127,
+      "grad_norm": 8.219011306762695,
+      "learning_rate": 1.676979354236452e-05,
+      "loss": 1.3015,
+      "step": 1690
+    },
+    {
+      "epoch": 0.2769067882884717,
+      "grad_norm": 10.750848770141602,
+      "learning_rate": 1.673141400883191e-05,
+      "loss": 1.2978,
+      "step": 1700
+    },
+    {
+      "epoch": 0.27853565174899214,
+      "grad_norm": 11.351160049438477,
+      "learning_rate": 1.6692852306075033e-05,
+      "loss": 1.2648,
+      "step": 1710
+    },
+    {
+      "epoch": 0.2801645152095126,
+      "grad_norm": 7.423020362854004,
+      "learning_rate": 1.665410947767188e-05,
+      "loss": 1.2352,
+      "step": 1720
+    },
+    {
+      "epoch": 0.28179337867003296,
+      "grad_norm": 7.212891101837158,
+      "learning_rate": 1.6615186572102154e-05,
+      "loss": 1.2899,
+      "step": 1730
+    },
+    {
+      "epoch": 0.2834222421305534,
+      "grad_norm": 8.066819190979004,
+      "learning_rate": 1.6576084642718915e-05,
+      "loss": 1.2763,
+      "step": 1740
+    },
+    {
+      "epoch": 0.28505110559107383,
+      "grad_norm": 8.984209060668945,
+      "learning_rate": 1.653680474772006e-05,
+      "loss": 1.3167,
+      "step": 1750
+    },
+    {
+      "epoch": 0.28667996905159426,
+      "grad_norm": 17.02054786682129,
+      "learning_rate": 1.6497347950119687e-05,
+      "loss": 1.2149,
+      "step": 1760
+    },
+    {
+      "epoch": 0.2883088325121147,
+      "grad_norm": 10.05130386352539,
+      "learning_rate": 1.645771531771933e-05,
+      "loss": 1.1194,
+      "step": 1770
+    },
+    {
+      "epoch": 0.2899376959726351,
+      "grad_norm": 7.705258369445801,
+      "learning_rate": 1.6417907923079057e-05,
+      "loss": 1.2395,
+      "step": 1780
+    },
+    {
+      "epoch": 0.2915665594331555,
+      "grad_norm": 10.932085037231445,
+      "learning_rate": 1.6377926843488462e-05,
+      "loss": 1.3887,
+      "step": 1790
+    },
+    {
+      "epoch": 0.29319542289367595,
+      "grad_norm": 12.111505508422852,
+      "learning_rate": 1.633777316093748e-05,
+      "loss": 1.2902,
+      "step": 1800
+    },
+    {
+      "epoch": 0.2948242863541964,
+      "grad_norm": 7.594025135040283,
+      "learning_rate": 1.6297447962087133e-05,
+      "loss": 1.3156,
+      "step": 1810
+    },
+    {
+      "epoch": 0.29645314981471677,
+      "grad_norm": 6.600276470184326,
+      "learning_rate": 1.625695233824011e-05,
+      "loss": 1.2334,
+      "step": 1820
+    },
+    {
+      "epoch": 0.2980820132752372,
+      "grad_norm": 8.619099617004395,
+      "learning_rate": 1.621628738531123e-05,
+      "loss": 1.2654,
+      "step": 1830
+    },
+    {
+      "epoch": 0.29971087673575764,
+      "grad_norm": 8.581014633178711,
+      "learning_rate": 1.6175454203797786e-05,
+      "loss": 1.3179,
+      "step": 1840
+    },
+    {
+      "epoch": 0.30133974019627807,
+      "grad_norm": 11.676705360412598,
+      "learning_rate": 1.6134453898749778e-05,
+      "loss": 1.311,
+      "step": 1850
+    },
+    {
+      "epoch": 0.30296860365679845,
+      "grad_norm": 8.072620391845703,
+      "learning_rate": 1.6093287579739983e-05,
+      "loss": 1.1559,
+      "step": 1860
+    },
+    {
+      "epoch": 0.3045974671173189,
+      "grad_norm": 8.07971477508545,
+      "learning_rate": 1.605195636083395e-05,
+      "loss": 1.3077,
+      "step": 1870
+    },
+    {
+      "epoch": 0.3062263305778393,
+      "grad_norm": 8.255348205566406,
+      "learning_rate": 1.6010461360559823e-05,
+      "loss": 1.2889,
+      "step": 1880
+    },
+    {
+      "epoch": 0.30785519403835976,
+      "grad_norm": 9.381460189819336,
+      "learning_rate": 1.5968803701878107e-05,
+      "loss": 1.2049,
+      "step": 1890
+    },
+    {
+      "epoch": 0.30948405749888014,
+      "grad_norm": 10.263215065002441,
+      "learning_rate": 1.5926984512151243e-05,
+      "loss": 1.2165,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3111129209594006,
+      "grad_norm": 7.69245719909668,
+      "learning_rate": 1.588500492311312e-05,
+      "loss": 1.2796,
+      "step": 1910
+    },
+    {
+      "epoch": 0.312741784419921,
+      "grad_norm": 6.975790023803711,
+      "learning_rate": 1.5842866070838444e-05,
+      "loss": 1.306,
+      "step": 1920
+    },
+    {
+      "epoch": 0.31437064788044145,
+      "grad_norm": 5.779184818267822,
+      "learning_rate": 1.5800569095711983e-05,
+      "loss": 1.233,
+      "step": 1930
+    },
+    {
+      "epoch": 0.3159995113409618,
+      "grad_norm": 8.1918306350708,
+      "learning_rate": 1.575811514239772e-05,
+      "loss": 1.2872,
+      "step": 1940
+    },
+    {
+      "epoch": 0.31762837480148226,
+      "grad_norm": 10.501468658447266,
+      "learning_rate": 1.5715505359807862e-05,
+      "loss": 1.1615,
+      "step": 1950
+    },
+    {
+      "epoch": 0.3192572382620027,
+      "grad_norm": 5.401251792907715,
+      "learning_rate": 1.567274090107176e-05,
+      "loss": 1.2808,
+      "step": 1960
+    },
+    {
+      "epoch": 0.32088610172252313,
+      "grad_norm": 7.17600154876709,
+      "learning_rate": 1.5629822923504692e-05,
+      "loss": 1.3381,
+      "step": 1970
+    },
+    {
+      "epoch": 0.3225149651830435,
+      "grad_norm": 7.486255645751953,
+      "learning_rate": 1.558675258857654e-05,
+      "loss": 1.2904,
+      "step": 1980
+    },
+    {
+      "epoch": 0.32414382864356395,
+      "grad_norm": 11.746787071228027,
+      "learning_rate": 1.5543531061880374e-05,
+      "loss": 1.2446,
+      "step": 1990
+    },
+    {
+      "epoch": 0.3257726921040844,
+      "grad_norm": 8.739921569824219,
+      "learning_rate": 1.55001595131009e-05,
+      "loss": 1.265,
+      "step": 2000
+    },
+    {
+      "epoch": 0.3257726921040844,
+      "eval_loss": 1.175255537033081,
+      "eval_runtime": 102.6201,
+      "eval_samples_per_second": 25.19,
+      "eval_steps_per_second": 25.19,
+      "step": 2000
+    },
+    {
+      "epoch": 0.3274015555646048,
+      "grad_norm": 7.848710536956787,
+      "learning_rate": 1.5456639115982795e-05,
+      "loss": 1.3146,
+      "step": 2010
+    },
+    {
+      "epoch": 0.3290304190251252,
+      "grad_norm": 11.283839225769043,
+      "learning_rate": 1.5412971048298964e-05,
+      "loss": 1.0946,
+      "step": 2020
+    },
+    {
+      "epoch": 0.33065928248564563,
+      "grad_norm": 11.305326461791992,
+      "learning_rate": 1.536915649181864e-05,
+      "loss": 1.2178,
+      "step": 2030
+    },
+    {
+      "epoch": 0.33228814594616607,
+      "grad_norm": 7.214547157287598,
+      "learning_rate": 1.5325196632275424e-05,
+      "loss": 1.2246,
+      "step": 2040
+    },
+    {
+      "epoch": 0.3339170094066865,
+      "grad_norm": 8.374602317810059,
+      "learning_rate": 1.528109265933519e-05,
+      "loss": 1.2537,
+      "step": 2050
+    },
+    {
+      "epoch": 0.3355458728672069,
+      "grad_norm": 7.756241321563721,
+      "learning_rate": 1.5236845766563881e-05,
+      "loss": 1.2351,
+      "step": 2060
+    },
+    {
+      "epoch": 0.3371747363277273,
+      "grad_norm": 13.376108169555664,
+      "learning_rate": 1.5192457151395226e-05,
+      "loss": 1.2366,
+      "step": 2070
+    },
+    {
+      "epoch": 0.33880359978824776,
+      "grad_norm": 8.468932151794434,
+      "learning_rate": 1.5147928015098309e-05,
+      "loss": 1.2931,
+      "step": 2080
+    },
+    {
+      "epoch": 0.3404324632487682,
+      "grad_norm": 7.511579513549805,
+      "learning_rate": 1.5103259562745084e-05,
+      "loss": 1.3374,
+      "step": 2090
+    },
+    {
+      "epoch": 0.34206132670928857,
+      "grad_norm": 8.611263275146484,
+      "learning_rate": 1.5058453003177756e-05,
+      "loss": 1.3561,
+      "step": 2100
+    },
+    {
+      "epoch": 0.343690190169809,
+      "grad_norm": 8.382079124450684,
+      "learning_rate": 1.5013509548976049e-05,
+      "loss": 1.34,
+      "step": 2110
+    },
+    {
+      "epoch": 0.34531905363032944,
+      "grad_norm": 6.7042365074157715,
+      "learning_rate": 1.4968430416424417e-05,
+      "loss": 1.2588,
+      "step": 2120
+    },
+    {
+      "epoch": 0.3469479170908499,
+      "grad_norm": 8.887292861938477,
+      "learning_rate": 1.4923216825479115e-05,
+      "loss": 1.3223,
+      "step": 2130
+    },
+    {
+      "epoch": 0.34857678055137026,
+      "grad_norm": 6.505458831787109,
+      "learning_rate": 1.4877869999735175e-05,
+      "loss": 1.3147,
+      "step": 2140
+    },
+    {
+      "epoch": 0.3502056440118907,
+      "grad_norm": 12.044365882873535,
+      "learning_rate": 1.4832391166393316e-05,
+      "loss": 1.2917,
+      "step": 2150
+    },
+    {
+      "epoch": 0.35183450747241113,
+      "grad_norm": 7.598618507385254,
+      "learning_rate": 1.4786781556226713e-05,
+      "loss": 1.1821,
+      "step": 2160
+    },
+    {
+      "epoch": 0.35346337093293156,
+      "grad_norm": 8.489738464355469,
+      "learning_rate": 1.4741042403547692e-05,
+      "loss": 1.2837,
+      "step": 2170
+    },
+    {
+      "epoch": 0.35509223439345194,
+      "grad_norm": 12.408584594726562,
+      "learning_rate": 1.4695174946174334e-05,
+      "loss": 1.2392,
+      "step": 2180
+    },
+    {
+      "epoch": 0.3567210978539724,
+      "grad_norm": 9.772265434265137,
+      "learning_rate": 1.4649180425396972e-05,
+      "loss": 1.3963,
+      "step": 2190
+    },
+    {
+      "epoch": 0.3583499613144928,
+      "grad_norm": 8.034818649291992,
+      "learning_rate": 1.4603060085944594e-05,
+      "loss": 1.2304,
+      "step": 2200
+    },
+    {
+      "epoch": 0.35997882477501325,
+      "grad_norm": 8.509257316589355,
+      "learning_rate": 1.455681517595117e-05,
+      "loss": 1.1302,
+      "step": 2210
+    },
+    {
+      "epoch": 0.36160768823553363,
+      "grad_norm": 7.198489189147949,
+      "learning_rate": 1.4510446946921857e-05,
+      "loss": 1.3122,
+      "step": 2220
+    },
+    {
+      "epoch": 0.36323655169605407,
+      "grad_norm": 13.541308403015137,
+      "learning_rate": 1.4463956653699148e-05,
+      "loss": 1.2654,
+      "step": 2230
+    },
+    {
+      "epoch": 0.3648654151565745,
+      "grad_norm": 9.754889488220215,
+      "learning_rate": 1.4417345554428898e-05,
+      "loss": 1.2936,
+      "step": 2240
+    },
+    {
+      "epoch": 0.36649427861709494,
+      "grad_norm": 11.386918067932129,
+      "learning_rate": 1.437061491052629e-05,
+      "loss": 1.2025,
+      "step": 2250
+    },
+    {
+      "epoch": 0.3681231420776153,
+      "grad_norm": 9.887473106384277,
+      "learning_rate": 1.4323765986641681e-05,
+      "loss": 1.3471,
+      "step": 2260
+    },
+    {
+      "epoch": 0.36975200553813575,
+      "grad_norm": 11.245658874511719,
+      "learning_rate": 1.4276800050626385e-05,
+      "loss": 1.3279,
+      "step": 2270
+    },
+    {
+      "epoch": 0.3713808689986562,
+      "grad_norm": 5.411240100860596,
+      "learning_rate": 1.4229718373498371e-05,
+      "loss": 1.2211,
+      "step": 2280
+    },
+    {
+      "epoch": 0.3730097324591766,
+      "grad_norm": 9.82564926147461,
+      "learning_rate": 1.4182522229407854e-05,
+      "loss": 1.0828,
+      "step": 2290
+    },
+    {
+      "epoch": 0.374638595919697,
+      "grad_norm": 10.895565032958984,
+      "learning_rate": 1.413521289560281e-05,
+      "loss": 1.2662,
+      "step": 2300
+    },
+    {
+      "epoch": 0.37626745938021744,
+      "grad_norm": 8.565170288085938,
+      "learning_rate": 1.4087791652394427e-05,
+      "loss": 1.1959,
+      "step": 2310
+    },
+    {
+      "epoch": 0.3778963228407379,
+      "grad_norm": 8.679253578186035,
+      "learning_rate": 1.404025978312244e-05,
+      "loss": 1.3137,
+      "step": 2320
+    },
+    {
+      "epoch": 0.3795251863012583,
+      "grad_norm": 7.74443244934082,
+      "learning_rate": 1.3992618574120415e-05,
+      "loss": 1.2227,
+      "step": 2330
+    },
+    {
+      "epoch": 0.38115404976177875,
+      "grad_norm": 6.578245639801025,
+      "learning_rate": 1.3944869314680922e-05,
+      "loss": 1.2857,
+      "step": 2340
+    },
+    {
+      "epoch": 0.3827829132222991,
+      "grad_norm": 7.3876237869262695,
+      "learning_rate": 1.3897013297020651e-05,
+      "loss": 1.3174,
+      "step": 2350
+    },
+    {
+      "epoch": 0.38441177668281956,
+      "grad_norm": 7.339137077331543,
+      "learning_rate": 1.3849051816245451e-05,
+      "loss": 1.281,
+      "step": 2360
+    },
+    {
+      "epoch": 0.38604064014334,
+      "grad_norm": 9.710099220275879,
+      "learning_rate": 1.3800986170315263e-05,
+      "loss": 1.3175,
+      "step": 2370
+    },
+    {
+      "epoch": 0.38766950360386043,
+      "grad_norm": 8.572495460510254,
+      "learning_rate": 1.3752817660009004e-05,
+      "loss": 1.3693,
+      "step": 2380
+    },
+    {
+      "epoch": 0.3892983670643808,
+      "grad_norm": 8.108867645263672,
+      "learning_rate": 1.3704547588889368e-05,
+      "loss": 1.3073,
+      "step": 2390
+    },
+    {
+      "epoch": 0.39092723052490125,
+      "grad_norm": 6.642714500427246,
+      "learning_rate": 1.3656177263267534e-05,
+      "loss": 1.2876,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3925560939854217,
+      "grad_norm": 8.35530948638916,
+      "learning_rate": 1.3607707992167836e-05,
+      "loss": 1.3244,
+      "step": 2410
+    },
+    {
+      "epoch": 0.3941849574459421,
+      "grad_norm": 13.95084285736084,
+      "learning_rate": 1.3559141087292313e-05,
+      "loss": 1.2599,
+      "step": 2420
+    },
+    {
+      "epoch": 0.3958138209064625,
+      "grad_norm": 9.53144645690918,
+      "learning_rate": 1.3510477862985233e-05,
+      "loss": 1.1835,
+      "step": 2430
+    },
+    {
+      "epoch": 0.39744268436698293,
+      "grad_norm": 7.759304046630859,
+      "learning_rate": 1.3461719636197503e-05,
+      "loss": 1.2536,
+      "step": 2440
+    },
+    {
+      "epoch": 0.39907154782750337,
+      "grad_norm": 12.239033699035645,
+      "learning_rate": 1.3412867726451051e-05,
+      "loss": 1.2358,
+      "step": 2450
+    },
+    {
+      "epoch": 0.4007004112880238,
+      "grad_norm": 8.611287117004395,
+      "learning_rate": 1.3363923455803098e-05,
+      "loss": 1.273,
+      "step": 2460
+    },
+    {
+      "epoch": 0.4023292747485442,
+      "grad_norm": 9.023387908935547,
+      "learning_rate": 1.3314888148810381e-05,
+      "loss": 1.2195,
+      "step": 2470
+    },
+    {
+      "epoch": 0.4039581382090646,
+      "grad_norm": 10.45922565460205,
+      "learning_rate": 1.3265763132493325e-05,
+      "loss": 1.2142,
+      "step": 2480
+    },
+    {
+      "epoch": 0.40558700166958506,
+      "grad_norm": 9.23582649230957,
+      "learning_rate": 1.3216549736300108e-05,
+      "loss": 1.3208,
+      "step": 2490
+    },
+    {
+      "epoch": 0.4072158651301055,
+      "grad_norm": 12.40174674987793,
+      "learning_rate": 1.3167249292070701e-05,
+      "loss": 1.2436,
+      "step": 2500
+    },
+    {
+      "epoch": 0.4072158651301055,
+      "eval_loss": 1.1542352437973022,
+      "eval_runtime": 102.4621,
+      "eval_samples_per_second": 25.229,
+      "eval_steps_per_second": 25.229,
+      "step": 2500
+    },
+    {
+      "epoch": 0.40884472859062587,
+      "grad_norm": 11.083693504333496,
+      "learning_rate": 1.311786313400081e-05,
+      "loss": 1.1723,
+      "step": 2510
+    },
+    {
+      "epoch": 0.4104735920511463,
+      "grad_norm": 6.602702617645264,
+      "learning_rate": 1.3068392598605775e-05,
+      "loss": 1.3112,
+      "step": 2520
+    },
+    {
+      "epoch": 0.41210245551166674,
+      "grad_norm": 5.926374912261963,
+      "learning_rate": 1.3018839024684407e-05,
+      "loss": 1.1096,
+      "step": 2530
+    },
+    {
+      "epoch": 0.4137313189721872,
+      "grad_norm": 8.42911148071289,
+      "learning_rate": 1.296920375328275e-05,
+      "loss": 1.3634,
+      "step": 2540
+    },
+    {
+      "epoch": 0.41536018243270756,
+      "grad_norm": 7.443211555480957,
+      "learning_rate": 1.2919488127657788e-05,
+      "loss": 1.1775,
+      "step": 2550
+    },
+    {
+      "epoch": 0.416989045893228,
+      "grad_norm": 7.255467414855957,
+      "learning_rate": 1.28696934932411e-05,
+      "loss": 1.2215,
+      "step": 2560
+    },
+    {
+      "epoch": 0.41861790935374843,
+      "grad_norm": 7.436680316925049,
+      "learning_rate": 1.2819821197602434e-05,
+      "loss": 1.3608,
+      "step": 2570
+    },
+    {
+      "epoch": 0.42024677281426887,
+      "grad_norm": 9.471477508544922,
+      "learning_rate": 1.2769872590413262e-05,
+      "loss": 1.2015,
+      "step": 2580
+    },
+    {
+      "epoch": 0.42187563627478925,
+      "grad_norm": 10.229029655456543,
+      "learning_rate": 1.271984902341023e-05,
+      "loss": 1.1578,
+      "step": 2590
+    },
+    {
+      "epoch": 0.4235044997353097,
+      "grad_norm": 9.242801666259766,
+      "learning_rate": 1.2669751850358593e-05,
+      "loss": 1.2569,
+      "step": 2600
+    },
+    {
+      "epoch": 0.4251333631958301,
+      "grad_norm": 7.492494106292725,
+      "learning_rate": 1.2619582427015575e-05,
+      "loss": 1.1613,
+      "step": 2610
+    },
+    {
+      "epoch": 0.42676222665635055,
+      "grad_norm": 9.39388370513916,
+      "learning_rate": 1.256934211109367e-05,
+      "loss": 1.2344,
+      "step": 2620
+    },
+    {
+      "epoch": 0.42839109011687093,
+      "grad_norm": 17.988927841186523,
+      "learning_rate": 1.2519032262223913e-05,
+      "loss": 1.2001,
+      "step": 2630
+    },
+    {
+      "epoch": 0.43001995357739137,
+      "grad_norm": 8.884842872619629,
+      "learning_rate": 1.2468654241919077e-05,
+      "loss": 1.3394,
+      "step": 2640
+    },
+    {
+      "epoch": 0.4316488170379118,
+      "grad_norm": 5.588969707489014,
+      "learning_rate": 1.2418209413536822e-05,
+      "loss": 1.2306,
+      "step": 2650
+    },
+    {
+      "epoch": 0.43327768049843224,
+      "grad_norm": 9.964385032653809,
+      "learning_rate": 1.2367699142242808e-05,
+      "loss": 1.3004,
+      "step": 2660
+    },
+    {
+      "epoch": 0.4349065439589526,
+      "grad_norm": 9.903732299804688,
+      "learning_rate": 1.2317124794973757e-05,
+      "loss": 1.2649,
+      "step": 2670
+    },
+    {
+      "epoch": 0.43653540741947305,
+      "grad_norm": 8.422842979431152,
+      "learning_rate": 1.2266487740400432e-05,
+      "loss": 1.3842,
+      "step": 2680
+    },
+    {
+      "epoch": 0.4381642708799935,
+      "grad_norm": 6.042238235473633,
+      "learning_rate": 1.2215789348890627e-05,
+      "loss": 1.1659,
+      "step": 2690
+    },
+    {
+      "epoch": 0.4397931343405139,
+      "grad_norm": 10.695905685424805,
+      "learning_rate": 1.216503099247207e-05,
+      "loss": 1.1943,
+      "step": 2700
+    },
+    {
+      "epoch": 0.4414219978010343,
+      "grad_norm": 7.222139835357666,
+      "learning_rate": 1.2114214044795287e-05,
+      "loss": 1.1867,
+      "step": 2710
+    },
+    {
+      "epoch": 0.44305086126155474,
+      "grad_norm": 8.005696296691895,
+      "learning_rate": 1.206333988109644e-05,
+      "loss": 1.2714,
+      "step": 2720
+    },
+    {
+      "epoch": 0.4446797247220752,
+      "grad_norm": 10.262436866760254,
+      "learning_rate": 1.2012409878160093e-05,
+      "loss": 1.2536,
+      "step": 2730
+    },
+    {
+      "epoch": 0.4463085881825956,
+      "grad_norm": 5.630941867828369,
+      "learning_rate": 1.196142541428197e-05,
+      "loss": 1.2511,
+      "step": 2740
+    },
+    {
+      "epoch": 0.447937451643116,
+      "grad_norm": 5.890045642852783,
+      "learning_rate": 1.1910387869231646e-05,
+      "loss": 1.204,
+      "step": 2750
+    },
+    {
+      "epoch": 0.4495663151036364,
+      "grad_norm": 7.118770122528076,
+      "learning_rate": 1.1859298624215202e-05,
+      "loss": 1.2963,
+      "step": 2760
+    },
+    {
+      "epoch": 0.45119517856415686,
+      "grad_norm": 6.118231773376465,
+      "learning_rate": 1.180815906183786e-05,
+      "loss": 1.1031,
+      "step": 2770
+    },
+    {
+      "epoch": 0.4528240420246773,
+      "grad_norm": 9.652909278869629,
+      "learning_rate": 1.175697056606655e-05,
+      "loss": 1.2406,
+      "step": 2780
+    },
+    {
+      "epoch": 0.4544529054851977,
+      "grad_norm": 8.817192077636719,
+      "learning_rate": 1.170573452219247e-05,
+      "loss": 1.1656,
+      "step": 2790
+    },
+    {
+      "epoch": 0.4560817689457181,
+      "grad_norm": 7.891348838806152,
+      "learning_rate": 1.1654452316793592e-05,
+      "loss": 1.1508,
+      "step": 2800
+    },
+    {
+      "epoch": 0.45771063240623855,
+      "grad_norm": 7.836014270782471,
+      "learning_rate": 1.1603125337697129e-05,
+      "loss": 1.2084,
+      "step": 2810
+    },
+    {
+      "epoch": 0.459339495866759,
+      "grad_norm": 7.247401714324951,
+      "learning_rate": 1.1551754973941996e-05,
+      "loss": 1.2001,
+      "step": 2820
+    },
+    {
+      "epoch": 0.46096835932727936,
+      "grad_norm": 7.9783830642700195,
+      "learning_rate": 1.1500342615741193e-05,
+      "loss": 1.2263,
+      "step": 2830
+    },
+    {
+      "epoch": 0.4625972227877998,
+      "grad_norm": 8.071614265441895,
+      "learning_rate": 1.144888965444421e-05,
+      "loss": 1.1689,
+      "step": 2840
+    },
+    {
+      "epoch": 0.46422608624832024,
+      "grad_norm": 13.862833976745605,
+      "learning_rate": 1.1397397482499352e-05,
+      "loss": 1.2704,
+      "step": 2850
+    },
+    {
+      "epoch": 0.46585494970884067,
+      "grad_norm": 10.272599220275879,
+      "learning_rate": 1.1345867493416067e-05,
+      "loss": 1.3094,
+      "step": 2860
+    },
+    {
+      "epoch": 0.46748381316936105,
+      "grad_norm": 8.56863784790039,
+      "learning_rate": 1.1294301081727235e-05,
+      "loss": 1.2395,
+      "step": 2870
+    },
+    {
+      "epoch": 0.4691126766298815,
+      "grad_norm": 9.376766204833984,
+      "learning_rate": 1.1242699642951411e-05,
+      "loss": 1.2032,
+      "step": 2880
+    },
+    {
+      "epoch": 0.4707415400904019,
+      "grad_norm": 10.573633193969727,
+      "learning_rate": 1.1191064573555094e-05,
+      "loss": 1.2421,
+      "step": 2890
+    },
+    {
+      "epoch": 0.47237040355092236,
+      "grad_norm": 10.285879135131836,
+      "learning_rate": 1.1139397270914893e-05,
+      "loss": 1.3089,
+      "step": 2900
+    },
+    {
+      "epoch": 0.4739992670114428,
+      "grad_norm": 6.976883888244629,
+      "learning_rate": 1.1087699133279743e-05,
+      "loss": 1.1944,
+      "step": 2910
+    },
+    {
+      "epoch": 0.4756281304719632,
+      "grad_norm": 8.282888412475586,
+      "learning_rate": 1.1035971559733047e-05,
+      "loss": 1.2164,
+      "step": 2920
+    },
+    {
+      "epoch": 0.4772569939324836,
+      "grad_norm": 12.169346809387207,
+      "learning_rate": 1.0984215950154821e-05,
+      "loss": 1.2526,
+      "step": 2930
+    },
+    {
+      "epoch": 0.47888585739300404,
+      "grad_norm": 10.577465057373047,
+      "learning_rate": 1.0932433705183806e-05,
+      "loss": 1.252,
+      "step": 2940
+    },
+    {
+      "epoch": 0.4805147208535245,
+      "grad_norm": 10.231258392333984,
+      "learning_rate": 1.0880626226179566e-05,
+      "loss": 1.2123,
+      "step": 2950
+    },
+    {
+      "epoch": 0.48214358431404486,
+      "grad_norm": 9.532476425170898,
+      "learning_rate": 1.0828794915184556e-05,
+      "loss": 1.0514,
+      "step": 2960
+    },
+    {
+      "epoch": 0.4837724477745653,
+      "grad_norm": 7.888957500457764,
+      "learning_rate": 1.0776941174886204e-05,
+      "loss": 1.1282,
+      "step": 2970
+    },
+    {
+      "epoch": 0.48540131123508573,
+      "grad_norm": 13.633064270019531,
+      "learning_rate": 1.072506640857891e-05,
+      "loss": 1.2246,
+      "step": 2980
+    },
+    {
+      "epoch": 0.48703017469560617,
+      "grad_norm": 10.267495155334473,
+      "learning_rate": 1.06731720201261e-05,
+      "loss": 1.2691,
+      "step": 2990
+    },
+    {
+      "epoch": 0.48865903815612655,
+      "grad_norm": 11.644023895263672,
+      "learning_rate": 1.0621259413922234e-05,
+      "loss": 1.2935,
+      "step": 3000
+    },
+    {
+      "epoch": 0.48865903815612655,
+      "eval_loss": 1.1448081731796265,
+      "eval_runtime": 102.8493,
+      "eval_samples_per_second": 25.134,
+      "eval_steps_per_second": 25.134,
+      "step": 3000
+    },
+    {
+      "epoch": 0.490287901616647,
+      "grad_norm": 9.024466514587402,
+      "learning_rate": 1.056932999485477e-05,
+      "loss": 1.1484,
+      "step": 3010
+    },
+    {
+      "epoch": 0.4919167650771674,
+      "grad_norm": 5.3878655433654785,
+      "learning_rate": 1.0517385168266193e-05,
+      "loss": 1.1989,
+      "step": 3020
+    },
+    {
+      "epoch": 0.49354562853768785,
+      "grad_norm": 7.960078716278076,
+      "learning_rate": 1.0465426339915927e-05,
+      "loss": 1.28,
+      "step": 3030
+    },
+    {
+      "epoch": 0.49517449199820823,
+      "grad_norm": 9.044086456298828,
+      "learning_rate": 1.041345491594234e-05,
+      "loss": 1.1462,
+      "step": 3040
+    },
+    {
+      "epoch": 0.49680335545872867,
+      "grad_norm": 9.06221866607666,
+      "learning_rate": 1.0361472302824656e-05,
+      "loss": 1.2558,
+      "step": 3050
+    },
+    {
+      "epoch": 0.4984322189192491,
+      "grad_norm": 7.99545955657959,
+      "learning_rate": 1.0309479907344915e-05,
+      "loss": 1.3108,
+      "step": 3060
+    },
+    {
+      "epoch": 0.5000610823797695,
+      "grad_norm": 11.91790771484375,
+      "learning_rate": 1.0257479136549889e-05,
+      "loss": 1.3106,
+      "step": 3070
+    },
+    {
+      "epoch": 0.5016899458402899,
+      "grad_norm": 9.534984588623047,
+      "learning_rate": 1.0205471397713002e-05,
+      "loss": 1.2752,
+      "step": 3080
+    },
+    {
+      "epoch": 0.5033188093008104,
+      "grad_norm": 8.188611030578613,
+      "learning_rate": 1.0153458098296265e-05,
+      "loss": 1.2704,
+      "step": 3090
+    },
+    {
+      "epoch": 0.5049476727613308,
+      "grad_norm": 10.754870414733887,
+      "learning_rate": 1.0101440645912156e-05,
+      "loss": 1.2822,
+      "step": 3100
+    },
+    {
+      "epoch": 0.5065765362218512,
+      "grad_norm": 9.690641403198242,
+      "learning_rate": 1.0049420448285554e-05,
+      "loss": 1.1982,
+      "step": 3110
+    },
+    {
+      "epoch": 0.5082053996823717,
+      "grad_norm": 7.106996059417725,
+      "learning_rate": 9.997398913215629e-06,
+      "loss": 1.1991,
+      "step": 3120
+    },
+    {
+      "epoch": 0.509834263142892,
+      "grad_norm": 9.254938125610352,
+      "learning_rate": 9.945377448537744e-06,
+      "loss": 1.356,
+      "step": 3130
+    },
+    {
+      "epoch": 0.5114631266034124,
+      "grad_norm": 8.23971939086914,
+      "learning_rate": 9.893357462085355e-06,
+      "loss": 1.1756,
+      "step": 3140
+    },
+    {
+      "epoch": 0.5130919900639329,
+      "grad_norm": 7.241464614868164,
+      "learning_rate": 9.841340361651921e-06,
+      "loss": 1.1899,
+      "step": 3150
+    },
+    {
+      "epoch": 0.5147208535244533,
+      "grad_norm": 10.916207313537598,
+      "learning_rate": 9.78932755495279e-06,
+      "loss": 1.261,
+      "step": 3160
+    },
+    {
+      "epoch": 0.5163497169849738,
+      "grad_norm": 10.14999771118164,
+      "learning_rate": 9.737320449587113e-06,
+      "loss": 1.244,
+      "step": 3170
+    },
+    {
+      "epoch": 0.5179785804454942,
+      "grad_norm": 10.535028457641602,
+      "learning_rate": 9.68532045299975e-06,
+      "loss": 1.2962,
+      "step": 3180
+    },
+    {
+      "epoch": 0.5196074439060145,
+      "grad_norm": 12.43609619140625,
+      "learning_rate": 9.63332897244318e-06,
+      "loss": 1.2515,
+      "step": 3190
+    },
+    {
+      "epoch": 0.521236307366535,
+      "grad_norm": 8.834859848022461,
+      "learning_rate": 9.581347414939416e-06,
+      "loss": 1.2559,
+      "step": 3200
+    },
+    {
+      "epoch": 0.5228651708270554,
+      "grad_norm": 9.219295501708984,
+      "learning_rate": 9.529377187241921e-06,
+      "loss": 1.0892,
+      "step": 3210
+    },
+    {
+      "epoch": 0.5244940342875758,
+      "grad_norm": 6.687707424163818,
+      "learning_rate": 9.477419695797551e-06,
+      "loss": 1.2412,
+      "step": 3220
+    },
+    {
+      "epoch": 0.5261228977480963,
+      "grad_norm": 7.913536548614502,
+      "learning_rate": 9.425476346708489e-06,
+      "loss": 1.3012,
+      "step": 3230
+    },
+    {
+      "epoch": 0.5277517612086167,
+      "grad_norm": 10.74738597869873,
+      "learning_rate": 9.373548545694189e-06,
+      "loss": 1.2319,
+      "step": 3240
+    },
+    {
+      "epoch": 0.5293806246691372,
+      "grad_norm": 8.231693267822266,
+      "learning_rate": 9.321637698053327e-06,
+      "loss": 1.1856,
+      "step": 3250
+    },
+    {
+      "epoch": 0.5310094881296575,
+      "grad_norm": 10.926920890808105,
+      "learning_rate": 9.269745208625784e-06,
+      "loss": 1.3557,
+      "step": 3260
+    },
+    {
+      "epoch": 0.5326383515901779,
+      "grad_norm": 7.84801721572876,
+      "learning_rate": 9.217872481754619e-06,
+      "loss": 1.2767,
+      "step": 3270
+    },
+    {
+      "epoch": 0.5342672150506984,
+      "grad_norm": 9.671969413757324,
+      "learning_rate": 9.16602092124807e-06,
+      "loss": 1.2934,
+      "step": 3280
+    },
+    {
+      "epoch": 0.5358960785112188,
+      "grad_norm": 11.278433799743652,
+      "learning_rate": 9.11419193034155e-06,
+      "loss": 1.2475,
+      "step": 3290
+    },
+    {
+      "epoch": 0.5375249419717392,
+      "grad_norm": 8.938894271850586,
+      "learning_rate": 9.062386911659692e-06,
+      "loss": 1.2476,
+      "step": 3300
+    },
+    {
+      "epoch": 0.5391538054322597,
+      "grad_norm": 10.644549369812012,
+      "learning_rate": 9.010607267178372e-06,
+      "loss": 1.2407,
+      "step": 3310
+    },
+    {
+      "epoch": 0.54078266889278,
+      "grad_norm": 10.27429485321045,
+      "learning_rate": 8.958854398186774e-06,
+      "loss": 1.3306,
+      "step": 3320
+    },
+    {
+      "epoch": 0.5424115323533005,
+      "grad_norm": 5.996701717376709,
+      "learning_rate": 8.90712970524948e-06,
+      "loss": 1.162,
+      "step": 3330
+    },
+    {
+      "epoch": 0.5440403958138209,
+      "grad_norm": 11.066450119018555,
+      "learning_rate": 8.855434588168543e-06,
+      "loss": 1.16,
+      "step": 3340
+    },
+    {
+      "epoch": 0.5456692592743413,
+      "grad_norm": 8.821560859680176,
+      "learning_rate": 8.803770445945626e-06,
+      "loss": 1.2471,
+      "step": 3350
+    },
+    {
+      "epoch": 0.5472981227348618,
+      "grad_norm": 8.159625053405762,
+      "learning_rate": 8.752138676744128e-06,
+      "loss": 1.1295,
+      "step": 3360
+    },
+    {
+      "epoch": 0.5489269861953822,
+      "grad_norm": 10.010625839233398,
+      "learning_rate": 8.70054067785136e-06,
+      "loss": 1.1393,
+      "step": 3370
+    },
+    {
+      "epoch": 0.5505558496559025,
+      "grad_norm": 8.413679122924805,
+      "learning_rate": 8.648977845640713e-06,
+      "loss": 1.3249,
+      "step": 3380
+    },
+    {
+      "epoch": 0.552184713116423,
+      "grad_norm": 7.034830570220947,
+      "learning_rate": 8.597451575533884e-06,
+      "loss": 1.1537,
+      "step": 3390
+    },
+    {
+      "epoch": 0.5538135765769434,
+      "grad_norm": 7.142354488372803,
+      "learning_rate": 8.545963261963102e-06,
+      "loss": 1.3551,
+      "step": 3400
+    },
+    {
+      "epoch": 0.5554424400374639,
+      "grad_norm": 11.520668029785156,
+      "learning_rate": 8.494514298333401e-06,
+      "loss": 1.2437,
+      "step": 3410
+    },
+    {
+      "epoch": 0.5570713034979843,
+      "grad_norm": 9.558789253234863,
+      "learning_rate": 8.443106076984895e-06,
+      "loss": 1.3416,
+      "step": 3420
+    },
+    {
+      "epoch": 0.5587001669585047,
+      "grad_norm": 7.987767696380615,
+      "learning_rate": 8.39173998915512e-06,
+      "loss": 1.1504,
+      "step": 3430
+    },
+    {
+      "epoch": 0.5603290304190252,
+      "grad_norm": 8.003718376159668,
+      "learning_rate": 8.340417424941363e-06,
+      "loss": 1.1578,
+      "step": 3440
+    },
+    {
+      "epoch": 0.5619578938795455,
+      "grad_norm": 9.900299072265625,
+      "learning_rate": 8.289139773263057e-06,
+      "loss": 1.2571,
+      "step": 3450
+    },
+    {
+      "epoch": 0.5635867573400659,
+      "grad_norm": 7.391176700592041,
+      "learning_rate": 8.237908421824186e-06,
+      "loss": 1.3128,
+      "step": 3460
+    },
+    {
+      "epoch": 0.5652156208005864,
+      "grad_norm": 7.9699387550354,
+      "learning_rate": 8.186724757075725e-06,
+      "loss": 1.2942,
+      "step": 3470
+    },
+    {
+      "epoch": 0.5668444842611068,
+      "grad_norm": 13.502717018127441,
+      "learning_rate": 8.135590164178136e-06,
+      "loss": 1.157,
+      "step": 3480
+    },
+    {
+      "epoch": 0.5684733477216273,
+      "grad_norm": 10.46247673034668,
+      "learning_rate": 8.084506026963859e-06,
+      "loss": 1.1876,
+      "step": 3490
+    },
+    {
+      "epoch": 0.5701022111821477,
+      "grad_norm": 6.954629898071289,
+      "learning_rate": 8.033473727899889e-06,
+      "loss": 1.2595,
+      "step": 3500
+    },
+    {
+      "epoch": 0.5701022111821477,
+      "eval_loss": 1.13480544090271,
+      "eval_runtime": 102.3333,
+      "eval_samples_per_second": 25.261,
+      "eval_steps_per_second": 25.261,
+      "step": 3500
+    },
+    {
+      "epoch": 0.571731074642668,
+      "grad_norm": 9.53747272491455,
+      "learning_rate": 7.982494648050341e-06,
+      "loss": 1.3107,
+      "step": 3510
+    },
+    {
+      "epoch": 0.5733599381031885,
+      "grad_norm": 9.204121589660645,
+      "learning_rate": 7.93157016703908e-06,
+      "loss": 1.3048,
+      "step": 3520
+    },
+    {
+      "epoch": 0.5749888015637089,
+      "grad_norm": 8.356864929199219,
+      "learning_rate": 7.880701663012387e-06,
+      "loss": 1.1239,
+      "step": 3530
+    },
+    {
+      "epoch": 0.5766176650242294,
+      "grad_norm": 6.577971935272217,
+      "learning_rate": 7.829890512601672e-06,
+      "loss": 1.2206,
+      "step": 3540
+    },
+    {
+      "epoch": 0.5782465284847498,
+      "grad_norm": 7.473880767822266,
+      "learning_rate": 7.779138090886202e-06,
+      "loss": 1.1229,
+      "step": 3550
+    },
+    {
+      "epoch": 0.5798753919452702,
+      "grad_norm": 10.256245613098145,
+      "learning_rate": 7.728445771355897e-06,
+      "loss": 1.2466,
+      "step": 3560
+    },
+    {
+      "epoch": 0.5815042554057906,
+      "grad_norm": 3.6034979820251465,
+      "learning_rate": 7.677814925874159e-06,
+      "loss": 1.1838,
+      "step": 3570
+    },
+    {
+      "epoch": 0.583133118866311,
+      "grad_norm": 13.721611976623535,
+      "learning_rate": 7.627246924640744e-06,
+      "loss": 1.2497,
+      "step": 3580
+    },
+    {
+      "epoch": 0.5847619823268314,
+      "grad_norm": 12.126471519470215,
+      "learning_rate": 7.57674313615469e-06,
+      "loss": 1.2542,
+      "step": 3590
+    },
+    {
+      "epoch": 0.5863908457873519,
+      "grad_norm": 9.544647216796875,
+      "learning_rate": 7.5263049271772645e-06,
+      "loss": 1.2369,
+      "step": 3600
+    },
+    {
+      "epoch": 0.5880197092478723,
+      "grad_norm": 9.008516311645508,
+      "learning_rate": 7.475933662694993e-06,
+      "loss": 1.1323,
+      "step": 3610
+    },
+    {
+      "epoch": 0.5896485727083928,
+      "grad_norm": 14.236581802368164,
+      "learning_rate": 7.425630705882707e-06,
+      "loss": 1.1859,
+      "step": 3620
+    },
+    {
+      "epoch": 0.5912774361689132,
+      "grad_norm": 7.743088722229004,
+      "learning_rate": 7.375397418066665e-06,
+      "loss": 1.2496,
+      "step": 3630
+    },
+    {
+      "epoch": 0.5929062996294335,
+      "grad_norm": 10.839030265808105,
+      "learning_rate": 7.3252351586876955e-06,
+      "loss": 1.2483,
+      "step": 3640
+    },
+    {
+      "epoch": 0.594535163089954,
+      "grad_norm": 10.632163047790527,
+      "learning_rate": 7.275145285264424e-06,
+      "loss": 1.2619,
+      "step": 3650
+    },
+    {
+      "epoch": 0.5961640265504744,
+      "grad_norm": 8.51111125946045,
+      "learning_rate": 7.2251291533565245e-06,
+      "loss": 1.2082,
+      "step": 3660
+    },
+    {
+      "epoch": 0.5977928900109948,
+      "grad_norm": 8.234833717346191,
+      "learning_rate": 7.175188116528044e-06,
+      "loss": 1.2497,
+      "step": 3670
+    },
+    {
+      "epoch": 0.5994217534715153,
+      "grad_norm": 12.51526927947998,
+      "learning_rate": 7.125323526310752e-06,
+      "loss": 1.2674,
+      "step": 3680
+    },
+    {
+      "epoch": 0.6010506169320357,
+      "grad_norm": 8.174943923950195,
+      "learning_rate": 7.0755367321675915e-06,
+      "loss": 1.1269,
+      "step": 3690
+    },
+    {
+      "epoch": 0.6026794803925561,
+      "grad_norm": 8.4658842086792,
+      "learning_rate": 7.025829081456137e-06,
+      "loss": 1.2499,
+      "step": 3700
+    },
+    {
+      "epoch": 0.6043083438530765,
+      "grad_norm": 7.7957987785339355,
+      "learning_rate": 6.976201919392138e-06,
+      "loss": 1.2735,
+      "step": 3710
+    },
+    {
+      "epoch": 0.6059372073135969,
+      "grad_norm": 7.680945873260498,
+      "learning_rate": 6.926656589013127e-06,
+      "loss": 1.2044,
+      "step": 3720
+    },
+    {
+      "epoch": 0.6075660707741174,
+      "grad_norm": 8.13155460357666,
+      "learning_rate": 6.877194431142055e-06,
+      "loss": 1.2299,
+      "step": 3730
+    },
+    {
+      "epoch": 0.6091949342346378,
+      "grad_norm": 7.943605899810791,
+      "learning_rate": 6.827816784351011e-06,
+      "loss": 1.1769,
+      "step": 3740
+    },
+    {
+      "epoch": 0.6108237976951582,
+      "grad_norm": 8.881424903869629,
+      "learning_rate": 6.778524984924999e-06,
+      "loss": 1.1524,
+      "step": 3750
+    },
+    {
+      "epoch": 0.6124526611556786,
+      "grad_norm": 9.679656028747559,
+      "learning_rate": 6.729320366825785e-06,
+      "loss": 1.2131,
+      "step": 3760
+    },
+    {
+      "epoch": 0.614081524616199,
+      "grad_norm": 8.224056243896484,
+      "learning_rate": 6.68020426165577e-06,
+      "loss": 1.254,
+      "step": 3770
+    },
+    {
+      "epoch": 0.6157103880767195,
+      "grad_norm": 7.536296844482422,
+      "learning_rate": 6.631177998621982e-06,
+      "loss": 1.2487,
+      "step": 3780
+    },
+    {
+      "epoch": 0.6173392515372399,
+      "grad_norm": 8.117711067199707,
+      "learning_rate": 6.582242904500085e-06,
+      "loss": 1.2872,
+      "step": 3790
+    },
+    {
+      "epoch": 0.6189681149977603,
+      "grad_norm": 9.820115089416504,
+      "learning_rate": 6.53340030359848e-06,
+      "loss": 1.3597,
+      "step": 3800
+    },
+    {
+      "epoch": 0.6205969784582808,
+      "grad_norm": 8.39965534210205,
+      "learning_rate": 6.4846515177224735e-06,
+      "loss": 1.2167,
+      "step": 3810
+    },
+    {
+      "epoch": 0.6222258419188011,
+      "grad_norm": 14.166945457458496,
+      "learning_rate": 6.435997866138488e-06,
+      "loss": 1.347,
+      "step": 3820
+    },
+    {
+      "epoch": 0.6238547053793215,
+      "grad_norm": 7.368659496307373,
+      "learning_rate": 6.3874406655383755e-06,
+      "loss": 1.1975,
+      "step": 3830
+    },
+    {
+      "epoch": 0.625483568839842,
+      "grad_norm": 6.8791913986206055,
+      "learning_rate": 6.3389812300037774e-06,
+      "loss": 1.3515,
+      "step": 3840
+    },
+    {
+      "epoch": 0.6271124323003624,
+      "grad_norm": 8.488605499267578,
+      "learning_rate": 6.290620870970561e-06,
+      "loss": 1.2868,
+      "step": 3850
+    },
+    {
+      "epoch": 0.6287412957608829,
+      "grad_norm": 11.543606758117676,
+      "learning_rate": 6.242360897193331e-06,
+      "loss": 1.1796,
+      "step": 3860
+    },
+    {
+      "epoch": 0.6303701592214033,
+      "grad_norm": 7.279272556304932,
+      "learning_rate": 6.194202614710015e-06,
+      "loss": 1.1563,
+      "step": 3870
+    },
+    {
+      "epoch": 0.6319990226819237,
+      "grad_norm": 6.415475368499756,
+      "learning_rate": 6.146147326806509e-06,
+      "loss": 1.246,
+      "step": 3880
+    },
+    {
+      "epoch": 0.6336278861424441,
+      "grad_norm": 8.384225845336914,
+      "learning_rate": 6.098196333981421e-06,
+      "loss": 1.2252,
+      "step": 3890
+    },
+    {
+      "epoch": 0.6352567496029645,
+      "grad_norm": 7.170252323150635,
+      "learning_rate": 6.050350933910865e-06,
+      "loss": 1.0904,
+      "step": 3900
+    },
+    {
+      "epoch": 0.6368856130634849,
+      "grad_norm": 11.731266021728516,
+      "learning_rate": 6.002612421413341e-06,
+      "loss": 1.196,
+      "step": 3910
+    },
+    {
+      "epoch": 0.6385144765240054,
+      "grad_norm": 8.852503776550293,
+      "learning_rate": 5.954982088414701e-06,
+      "loss": 1.3241,
+      "step": 3920
+    },
+    {
+      "epoch": 0.6401433399845258,
+      "grad_norm": 8.035179138183594,
+      "learning_rate": 5.9074612239131915e-06,
+      "loss": 1.2826,
+      "step": 3930
+    },
+    {
+      "epoch": 0.6417722034450463,
+      "grad_norm": 8.706337928771973,
+      "learning_rate": 5.8600511139445536e-06,
+      "loss": 1.1557,
+      "step": 3940
+    },
+    {
+      "epoch": 0.6434010669055666,
+      "grad_norm": 14.4612398147583,
+      "learning_rate": 5.81275304154723e-06,
+      "loss": 1.1653,
+      "step": 3950
+    },
+    {
+      "epoch": 0.645029930366087,
+      "grad_norm": 7.372416973114014,
+      "learning_rate": 5.765568286727646e-06,
+      "loss": 1.1933,
+      "step": 3960
+    },
+    {
+      "epoch": 0.6466587938266075,
+      "grad_norm": 8.995341300964355,
+      "learning_rate": 5.718498126425556e-06,
+      "loss": 1.1356,
+      "step": 3970
+    },
+    {
+      "epoch": 0.6482876572871279,
+      "grad_norm": 12.190003395080566,
+      "learning_rate": 5.671543834479503e-06,
+      "loss": 1.3165,
+      "step": 3980
+    },
+    {
+      "epoch": 0.6499165207476483,
+      "grad_norm": 6.376560211181641,
+      "learning_rate": 5.624706681592329e-06,
+      "loss": 1.2403,
+      "step": 3990
+    },
+    {
+      "epoch": 0.6515453842081688,
+      "grad_norm": 8.215283393859863,
+      "learning_rate": 5.5779879352968e-06,
+      "loss": 1.2896,
+      "step": 4000
+    },
+    {
+      "epoch": 0.6515453842081688,
+      "eval_loss": 1.1294902563095093,
+      "eval_runtime": 102.4055,
+      "eval_samples_per_second": 25.243,
+      "eval_steps_per_second": 25.243,
+      "step": 4000
+    },
+    {
+      "epoch": 0.6531742476686891,
+      "grad_norm": 7.186392307281494,
+      "learning_rate": 5.531388859921303e-06,
+      "loss": 1.2025,
+      "step": 4010
+    },
+    {
+      "epoch": 0.6548031111292096,
+      "grad_norm": 11.444113731384277,
+      "learning_rate": 5.484910716555607e-06,
+      "loss": 1.3191,
+      "step": 4020
+    },
+    {
+      "epoch": 0.65643197458973,
+      "grad_norm": 4.889548301696777,
+      "learning_rate": 5.438554763016775e-06,
+      "loss": 1.1232,
+      "step": 4030
+    },
+    {
+      "epoch": 0.6580608380502504,
+      "grad_norm": 12.841841697692871,
+      "learning_rate": 5.392322253815079e-06,
+      "loss": 1.2834,
+      "step": 4040
+    },
+    {
+      "epoch": 0.6596897015107709,
+      "grad_norm": 7.367560386657715,
+      "learning_rate": 5.3462144401200945e-06,
+      "loss": 1.3042,
+      "step": 4050
+    },
+    {
+      "epoch": 0.6613185649712913,
+      "grad_norm": 7.350122451782227,
+      "learning_rate": 5.300232569726805e-06,
+      "loss": 1.2925,
+      "step": 4060
+    },
+    {
+      "epoch": 0.6629474284318118,
+      "grad_norm": 6.530559539794922,
+      "learning_rate": 5.254377887021842e-06,
+      "loss": 1.2089,
+      "step": 4070
+    },
+    {
+      "epoch": 0.6645762918923321,
+      "grad_norm": 8.514747619628906,
+      "learning_rate": 5.20865163294983e-06,
+      "loss": 1.1635,
+      "step": 4080
+    },
+    {
+      "epoch": 0.6662051553528525,
+      "grad_norm": 10.842570304870605,
+      "learning_rate": 5.163055044979783e-06,
+      "loss": 1.1896,
+      "step": 4090
+    },
+    {
+      "epoch": 0.667834018813373,
+      "grad_norm": 10.257299423217773,
+      "learning_rate": 5.1175893570716075e-06,
+      "loss": 1.2857,
+      "step": 4100
+    },
+    {
+      "epoch": 0.6694628822738934,
+      "grad_norm": 7.264848709106445,
+      "learning_rate": 5.072255799642737e-06,
+      "loss": 1.2011,
+      "step": 4110
+    },
+    {
+      "epoch": 0.6710917457344138,
+      "grad_norm": 7.874101161956787,
+      "learning_rate": 5.027055599534802e-06,
+      "loss": 1.2601,
+      "step": 4120
+    },
+    {
+      "epoch": 0.6727206091949343,
+      "grad_norm": 9.67297077178955,
+      "learning_rate": 4.981989979980457e-06,
+      "loss": 1.2038,
+      "step": 4130
+    },
+    {
+      "epoch": 0.6743494726554546,
+      "grad_norm": 8.36242389678955,
+      "learning_rate": 4.93706016057026e-06,
+      "loss": 1.2007,
+      "step": 4140
+    },
+    {
+      "epoch": 0.6759783361159751,
+      "grad_norm": 8.81531810760498,
+      "learning_rate": 4.8922673572196625e-06,
+      "loss": 1.2002,
+      "step": 4150
+    },
+    {
+      "epoch": 0.6776071995764955,
+      "grad_norm": 13.88326358795166,
+      "learning_rate": 4.847612782136127e-06,
+      "loss": 1.2879,
+      "step": 4160
+    },
+    {
+      "epoch": 0.6792360630370159,
+      "grad_norm": 8.99789810180664,
+      "learning_rate": 4.803097643786289e-06,
+      "loss": 1.0767,
+      "step": 4170
+    },
+    {
+      "epoch": 0.6808649264975364,
+      "grad_norm": 9.136383056640625,
+      "learning_rate": 4.758723146863285e-06,
+      "loss": 1.2943,
+      "step": 4180
+    },
+    {
+      "epoch": 0.6824937899580568,
+      "grad_norm": 11.534010887145996,
+      "learning_rate": 4.714490492254134e-06,
+      "loss": 1.2505,
+      "step": 4190
+    },
+    {
+      "epoch": 0.6841226534185771,
+      "grad_norm": 7.77902889251709,
+      "learning_rate": 4.670400877007229e-06,
+      "loss": 1.3522,
+      "step": 4200
+    },
+    {
+      "epoch": 0.6857515168790976,
+      "grad_norm": 9.129291534423828,
+      "learning_rate": 4.6264554942999685e-06,
+      "loss": 1.2585,
+      "step": 4210
+    },
+    {
+      "epoch": 0.687380380339618,
+      "grad_norm": 9.164624214172363,
+      "learning_rate": 4.582655533406445e-06,
+      "loss": 1.2429,
+      "step": 4220
+    },
+    {
+      "epoch": 0.6890092438001385,
+      "grad_norm": 10.26278018951416,
+      "learning_rate": 4.539002179665256e-06,
+      "loss": 1.3034,
+      "step": 4230
+    },
+    {
+      "epoch": 0.6906381072606589,
+      "grad_norm": 6.888890743255615,
+      "learning_rate": 4.495496614447455e-06,
+      "loss": 1.2,
+      "step": 4240
+    },
+    {
+      "epoch": 0.6922669707211793,
+      "grad_norm": 6.9961090087890625,
+      "learning_rate": 4.452140015124539e-06,
+      "loss": 1.2924,
+      "step": 4250
+    },
+    {
+      "epoch": 0.6938958341816998,
+      "grad_norm": 9.418280601501465,
+      "learning_rate": 4.4089335550366275e-06,
+      "loss": 1.2678,
+      "step": 4260
+    },
+    {
+      "epoch": 0.6955246976422201,
+      "grad_norm": 9.817927360534668,
+      "learning_rate": 4.365878403460687e-06,
+      "loss": 1.1866,
+      "step": 4270
+    },
+    {
+      "epoch": 0.6971535611027405,
+      "grad_norm": 7.495052337646484,
+      "learning_rate": 4.322975725578871e-06,
+      "loss": 1.2452,
+      "step": 4280
+    },
+    {
+      "epoch": 0.698782424563261,
+      "grad_norm": 7.096460342407227,
+      "learning_rate": 4.280226682447026e-06,
+      "loss": 1.2315,
+      "step": 4290
+    },
+    {
+      "epoch": 0.7004112880237814,
+      "grad_norm": 8.440934181213379,
+      "learning_rate": 4.23763243096325e-06,
+      "loss": 1.1404,
+      "step": 4300
+    },
+    {
+      "epoch": 0.7020401514843019,
+      "grad_norm": 10.936495780944824,
+      "learning_rate": 4.195194123836569e-06,
+      "loss": 1.303,
+      "step": 4310
+    },
+    {
+      "epoch": 0.7036690149448223,
+      "grad_norm": 10.077718734741211,
+      "learning_rate": 4.152912909555775e-06,
+      "loss": 1.2253,
+      "step": 4320
+    },
+    {
+      "epoch": 0.7052978784053426,
+      "grad_norm": 10.520235061645508,
+      "learning_rate": 4.110789932358312e-06,
+      "loss": 1.1638,
+      "step": 4330
+    },
+    {
+      "epoch": 0.7069267418658631,
+      "grad_norm": 8.664091110229492,
+      "learning_rate": 4.068826332199336e-06,
+      "loss": 1.2585,
+      "step": 4340
+    },
+    {
+      "epoch": 0.7085556053263835,
+      "grad_norm": 9.869935989379883,
+      "learning_rate": 4.027023244720853e-06,
+      "loss": 1.2998,
+      "step": 4350
+    },
+    {
+      "epoch": 0.7101844687869039,
+      "grad_norm": 7.815759658813477,
+      "learning_rate": 3.985381801220975e-06,
+      "loss": 1.1673,
+      "step": 4360
+    },
+    {
+      "epoch": 0.7118133322474244,
+      "grad_norm": 9.876256942749023,
+      "learning_rate": 3.943903128623336e-06,
+      "loss": 1.1876,
+      "step": 4370
+    },
+    {
+      "epoch": 0.7134421957079448,
+      "grad_norm": 10.636608123779297,
+      "learning_rate": 3.902588349446551e-06,
+      "loss": 1.2142,
+      "step": 4380
+    },
+    {
+      "epoch": 0.7150710591684653,
+      "grad_norm": 8.32884693145752,
+      "learning_rate": 3.86143858177388e-06,
+      "loss": 1.3103,
+      "step": 4390
+    },
+    {
+      "epoch": 0.7166999226289856,
+      "grad_norm": 13.757806777954102,
+      "learning_rate": 3.820454939222946e-06,
+      "loss": 1.1283,
+      "step": 4400
+    },
+    {
+      "epoch": 0.718328786089506,
+      "grad_norm": 19.816259384155273,
+      "learning_rate": 3.7796385309155948e-06,
+      "loss": 1.2626,
+      "step": 4410
+    },
+    {
+      "epoch": 0.7199576495500265,
+      "grad_norm": 8.810699462890625,
+      "learning_rate": 3.7389904614479e-06,
+      "loss": 1.3254,
+      "step": 4420
+    },
+    {
+      "epoch": 0.7215865130105469,
+      "grad_norm": 7.040134429931641,
+      "learning_rate": 3.698511830860243e-06,
+      "loss": 1.3012,
+      "step": 4430
+    },
+    {
+      "epoch": 0.7232153764710673,
+      "grad_norm": 8.462788581848145,
+      "learning_rate": 3.658203734607567e-06,
+      "loss": 1.1578,
+      "step": 4440
+    },
+    {
+      "epoch": 0.7248442399315878,
+      "grad_norm": 8.182144165039062,
+      "learning_rate": 3.6180672635297243e-06,
+      "loss": 1.3664,
+      "step": 4450
+    },
+    {
+      "epoch": 0.7264731033921081,
+      "grad_norm": 11.486858367919922,
+      "learning_rate": 3.578103503821939e-06,
+      "loss": 1.192,
+      "step": 4460
+    },
+    {
+      "epoch": 0.7281019668526286,
+      "grad_norm": 8.237690925598145,
+      "learning_rate": 3.53831353700544e-06,
+      "loss": 1.1809,
+      "step": 4470
+    },
+    {
+      "epoch": 0.729730830313149,
+      "grad_norm": 6.926926612854004,
+      "learning_rate": 3.4986984398981662e-06,
+      "loss": 1.1828,
+      "step": 4480
+    },
+    {
+      "epoch": 0.7313596937736694,
+      "grad_norm": 8.473318099975586,
+      "learning_rate": 3.4592592845856388e-06,
+      "loss": 1.3035,
+      "step": 4490
+    },
+    {
+      "epoch": 0.7329885572341899,
+      "grad_norm": 9.865799903869629,
+      "learning_rate": 3.4199971383919538e-06,
+      "loss": 1.2081,
+      "step": 4500
+    },
+    {
+      "epoch": 0.7329885572341899,
+      "eval_loss": 1.1236326694488525,
+      "eval_runtime": 102.8937,
+      "eval_samples_per_second": 25.123,
+      "eval_steps_per_second": 25.123,
+      "step": 4500
+    },
+    {
+      "epoch": 0.7346174206947103,
+      "grad_norm": 9.886106491088867,
+      "learning_rate": 3.380913063850877e-06,
+      "loss": 1.2866,
+      "step": 4510
+    },
+    {
+      "epoch": 0.7362462841552306,
+      "grad_norm": 7.137485504150391,
+      "learning_rate": 3.342008118677108e-06,
+      "loss": 1.0974,
+      "step": 4520
+    },
+    {
+      "epoch": 0.7378751476157511,
+      "grad_norm": 9.091876029968262,
+      "learning_rate": 3.303283355737653e-06,
+      "loss": 1.2417,
+      "step": 4530
+    },
+    {
+      "epoch": 0.7395040110762715,
+      "grad_norm": 7.459966659545898,
+      "learning_rate": 3.2647398230233175e-06,
+      "loss": 1.2105,
+      "step": 4540
+    },
+    {
+      "epoch": 0.741132874536792,
+      "grad_norm": 7.54026460647583,
+      "learning_rate": 3.2263785636203635e-06,
+      "loss": 1.1231,
+      "step": 4550
+    },
+    {
+      "epoch": 0.7427617379973124,
+      "grad_norm": 9.739063262939453,
+      "learning_rate": 3.188200615682265e-06,
+      "loss": 1.1882,
+      "step": 4560
+    },
+    {
+      "epoch": 0.7443906014578328,
+      "grad_norm": 11.615285873413086,
+      "learning_rate": 3.150207012401629e-06,
+      "loss": 1.1598,
+      "step": 4570
+    },
+    {
+      "epoch": 0.7460194649183532,
+      "grad_norm": 7.322878837585449,
+      "learning_rate": 3.1123987819822234e-06,
+      "loss": 1.198,
+      "step": 4580
+    },
+    {
+      "epoch": 0.7476483283788736,
+      "grad_norm": 7.064319133758545,
+      "learning_rate": 3.0747769476111454e-06,
+      "loss": 1.1921,
+      "step": 4590
+    },
+    {
+      "epoch": 0.749277191839394,
+      "grad_norm": 9.456534385681152,
+      "learning_rate": 3.037342527431152e-06,
+      "loss": 1.263,
+      "step": 4600
+    },
+    {
+      "epoch": 0.7509060552999145,
+      "grad_norm": 7.1396098136901855,
+      "learning_rate": 3.0000965345130904e-06,
+      "loss": 1.2136,
+      "step": 4610
+    },
+    {
+      "epoch": 0.7525349187604349,
+      "grad_norm": 7.965095520019531,
+      "learning_rate": 2.96303997682848e-06,
+      "loss": 1.2013,
+      "step": 4620
+    },
+    {
+      "epoch": 0.7541637822209554,
+      "grad_norm": 6.984764575958252,
+      "learning_rate": 2.9261738572222487e-06,
+      "loss": 1.2054,
+      "step": 4630
+    },
+    {
+      "epoch": 0.7557926456814757,
+      "grad_norm": 9.820785522460938,
+      "learning_rate": 2.889499173385576e-06,
+      "loss": 1.2689,
+      "step": 4640
+    },
+    {
+      "epoch": 0.7574215091419961,
+      "grad_norm": 12.76016616821289,
+      "learning_rate": 2.8530169178289068e-06,
+      "loss": 1.2673,
+      "step": 4650
+    },
+    {
+      "epoch": 0.7590503726025166,
+      "grad_norm": 6.978170871734619,
+      "learning_rate": 2.8167280778550897e-06,
+      "loss": 1.3873,
+      "step": 4660
+    },
+    {
+      "epoch": 0.760679236063037,
+      "grad_norm": 5.620461463928223,
+      "learning_rate": 2.7806336355326434e-06,
+      "loss": 1.1573,
+      "step": 4670
+    },
+    {
+      "epoch": 0.7623080995235575,
+      "grad_norm": 6.476634979248047,
+      "learning_rate": 2.744734567669203e-06,
+      "loss": 1.1334,
+      "step": 4680
+    },
+    {
+      "epoch": 0.7639369629840779,
+      "grad_norm": 6.739288330078125,
+      "learning_rate": 2.709031845785062e-06,
+      "loss": 1.2286,
+      "step": 4690
+    },
+    {
+      "epoch": 0.7655658264445983,
+      "grad_norm": 8.93143367767334,
+      "learning_rate": 2.673526436086894e-06,
+      "loss": 1.2131,
+      "step": 4700
+    },
+    {
+      "epoch": 0.7671946899051187,
+      "grad_norm": 17.24730110168457,
+      "learning_rate": 2.63821929944161e-06,
+      "loss": 1.2658,
+      "step": 4710
+    },
+    {
+      "epoch": 0.7688235533656391,
+      "grad_norm": 11.340949058532715,
+      "learning_rate": 2.6031113913503337e-06,
+      "loss": 1.2794,
+      "step": 4720
+    },
+    {
+      "epoch": 0.7704524168261595,
+      "grad_norm": 10.817314147949219,
+      "learning_rate": 2.5682036619225657e-06,
+      "loss": 1.3443,
+      "step": 4730
+    },
+    {
+      "epoch": 0.77208128028668,
+      "grad_norm": 17.411333084106445,
+      "learning_rate": 2.5334970558504613e-06,
+      "loss": 1.1532,
+      "step": 4740
+    },
+    {
+      "epoch": 0.7737101437472004,
+      "grad_norm": 9.042006492614746,
+      "learning_rate": 2.4989925123832583e-06,
+      "loss": 1.2526,
+      "step": 4750
+    },
+    {
+      "epoch": 0.7753390072077209,
+      "grad_norm": 9.995912551879883,
+      "learning_rate": 2.4646909653018724e-06,
+      "loss": 1.0986,
+      "step": 4760
+    },
+    {
+      "epoch": 0.7769678706682412,
+      "grad_norm": 11.739777565002441,
+      "learning_rate": 2.4305933428936137e-06,
+      "loss": 1.2693,
+      "step": 4770
+    },
+    {
+      "epoch": 0.7785967341287616,
+      "grad_norm": 11.440888404846191,
+      "learning_rate": 2.3967005679270736e-06,
+      "loss": 1.2691,
+      "step": 4780
+    },
+    {
+      "epoch": 0.7802255975892821,
+      "grad_norm": 9.173127174377441,
+      "learning_rate": 2.3630135576271563e-06,
+      "loss": 1.137,
+      "step": 4790
+    },
+    {
+      "epoch": 0.7818544610498025,
+      "grad_norm": 11.49284839630127,
+      "learning_rate": 2.329533223650233e-06,
+      "loss": 1.2192,
+      "step": 4800
+    },
+    {
+      "epoch": 0.7834833245103229,
+      "grad_norm": 6.268970012664795,
+      "learning_rate": 2.296260472059505e-06,
+      "loss": 1.1762,
+      "step": 4810
+    },
+    {
+      "epoch": 0.7851121879708434,
+      "grad_norm": 7.918376445770264,
+      "learning_rate": 2.2631962033004486e-06,
+      "loss": 1.2459,
+      "step": 4820
+    },
+    {
+      "epoch": 0.7867410514313637,
+      "grad_norm": 9.789480209350586,
+      "learning_rate": 2.230341312176476e-06,
+      "loss": 1.2437,
+      "step": 4830
+    },
+    {
+      "epoch": 0.7883699148918842,
+      "grad_norm": 6.269535064697266,
+      "learning_rate": 2.197696687824703e-06,
+      "loss": 1.2836,
+      "step": 4840
+    },
+    {
+      "epoch": 0.7899987783524046,
+      "grad_norm": 8.587770462036133,
+      "learning_rate": 2.165263213691885e-06,
+      "loss": 1.3312,
+      "step": 4850
+    },
+    {
+      "epoch": 0.791627641812925,
+      "grad_norm": 7.60506010055542,
+      "learning_rate": 2.133041767510523e-06,
+      "loss": 1.285,
+      "step": 4860
+    },
+    {
+      "epoch": 0.7932565052734455,
+      "grad_norm": 8.978716850280762,
+      "learning_rate": 2.1010332212750926e-06,
+      "loss": 1.2393,
+      "step": 4870
+    },
+    {
+      "epoch": 0.7948853687339659,
+      "grad_norm": 8.022439956665039,
+      "learning_rate": 2.0692384412184587e-06,
+      "loss": 1.1718,
+      "step": 4880
+    },
+    {
+      "epoch": 0.7965142321944862,
+      "grad_norm": 11.286605834960938,
+      "learning_rate": 2.0376582877884322e-06,
+      "loss": 1.1298,
+      "step": 4890
+    },
+    {
+      "epoch": 0.7981430956550067,
+      "grad_norm": 8.459285736083984,
+      "learning_rate": 2.0062936156244695e-06,
+      "loss": 1.3101,
+      "step": 4900
+    },
+    {
+      "epoch": 0.7997719591155271,
+      "grad_norm": 8.629647254943848,
+      "learning_rate": 1.9751452735345677e-06,
+      "loss": 1.3471,
+      "step": 4910
+    },
+    {
+      "epoch": 0.8014008225760476,
+      "grad_norm": 9.93689250946045,
+      "learning_rate": 1.9442141044722694e-06,
+      "loss": 1.2816,
+      "step": 4920
+    },
+    {
+      "epoch": 0.803029686036568,
+      "grad_norm": 13.70149040222168,
+      "learning_rate": 1.9135009455138643e-06,
+      "loss": 1.1864,
+      "step": 4930
+    },
+    {
+      "epoch": 0.8046585494970884,
+      "grad_norm": 11.720361709594727,
+      "learning_rate": 1.8830066278357395e-06,
+      "loss": 1.2885,
+      "step": 4940
+    },
+    {
+      "epoch": 0.8062874129576089,
+      "grad_norm": 9.665403366088867,
+      "learning_rate": 1.8527319766918694e-06,
+      "loss": 1.2245,
+      "step": 4950
+    },
+    {
+      "epoch": 0.8079162764181292,
+      "grad_norm": 8.559013366699219,
+      "learning_rate": 1.8226778113914989e-06,
+      "loss": 1.1385,
+      "step": 4960
+    },
+    {
+      "epoch": 0.8095451398786496,
+      "grad_norm": 9.008288383483887,
+      "learning_rate": 1.7928449452769636e-06,
+      "loss": 1.1462,
+      "step": 4970
+    },
+    {
+      "epoch": 0.8111740033391701,
+      "grad_norm": 7.702124118804932,
+      "learning_rate": 1.7632341857016733e-06,
+      "loss": 1.2371,
+      "step": 4980
+    },
+    {
+      "epoch": 0.8128028667996905,
+      "grad_norm": 7.910833835601807,
+      "learning_rate": 1.7338463340082734e-06,
+      "loss": 1.2431,
+      "step": 4990
+    },
+    {
+      "epoch": 0.814431730260211,
+      "grad_norm": 10.374570846557617,
+      "learning_rate": 1.7046821855069562e-06,
+      "loss": 1.2451,
+      "step": 5000
+    },
+    {
+      "epoch": 0.814431730260211,
+      "eval_loss": 1.1212321519851685,
+      "eval_runtime": 102.729,
+      "eval_samples_per_second": 25.163,
+      "eval_steps_per_second": 25.163,
+      "step": 5000
+    },
+    {
+      "epoch": 0.8160605937207314,
+      "grad_norm": 12.469367027282715,
+      "learning_rate": 1.6757425294539266e-06,
+      "loss": 1.3257,
+      "step": 5010
+    },
+    {
+      "epoch": 0.8176894571812517,
+      "grad_norm": 8.322937965393066,
+      "learning_rate": 1.647028149030061e-06,
+      "loss": 1.2494,
+      "step": 5020
+    },
+    {
+      "epoch": 0.8193183206417722,
+      "grad_norm": 10.362074851989746,
+      "learning_rate": 1.6185398213196935e-06,
+      "loss": 1.1441,
+      "step": 5030
+    },
+    {
+      "epoch": 0.8209471841022926,
+      "grad_norm": 10.657358169555664,
+      "learning_rate": 1.5902783172896042e-06,
+      "loss": 1.2038,
+      "step": 5040
+    },
+    {
+      "epoch": 0.822576047562813,
+      "grad_norm": 6.7474517822265625,
+      "learning_rate": 1.5622444017681438e-06,
+      "loss": 1.1308,
+      "step": 5050
+    },
+    {
+      "epoch": 0.8242049110233335,
+      "grad_norm": 10.932724952697754,
+      "learning_rate": 1.534438833424533e-06,
+      "loss": 1.1732,
+      "step": 5060
+    },
+    {
+      "epoch": 0.8258337744838539,
+      "grad_norm": 7.627405643463135,
+      "learning_rate": 1.5068623647483428e-06,
+      "loss": 1.1552,
+      "step": 5070
+    },
+    {
+      "epoch": 0.8274626379443744,
+      "grad_norm": 13.693827629089355,
+      "learning_rate": 1.479515742029115e-06,
+      "loss": 1.1808,
+      "step": 5080
+    },
+    {
+      "epoch": 0.8290915014048947,
+      "grad_norm": 7.652097702026367,
+      "learning_rate": 1.4523997053361805e-06,
+      "loss": 1.2478,
+      "step": 5090
+    },
+    {
+      "epoch": 0.8307203648654151,
+      "grad_norm": 6.031581401824951,
+      "learning_rate": 1.4255149884986253e-06,
+      "loss": 1.1577,
+      "step": 5100
+    },
+    {
+      "epoch": 0.8323492283259356,
+      "grad_norm": 7.7093729972839355,
+      "learning_rate": 1.3988623190854233e-06,
+      "loss": 1.1844,
+      "step": 5110
+    },
+    {
+      "epoch": 0.833978091786456,
+      "grad_norm": 9.999074935913086,
+      "learning_rate": 1.3724424183857599e-06,
+      "loss": 1.1921,
+      "step": 5120
+    },
+    {
+      "epoch": 0.8356069552469764,
+      "grad_norm": 6.947312355041504,
+      "learning_rate": 1.3462560013895031e-06,
+      "loss": 1.273,
+      "step": 5130
+    },
+    {
+      "epoch": 0.8372358187074969,
+      "grad_norm": 9.171476364135742,
+      "learning_rate": 1.320303776767855e-06,
+      "loss": 1.1454,
+      "step": 5140
+    },
+    {
+      "epoch": 0.8388646821680172,
+      "grad_norm": 7.364622592926025,
+      "learning_rate": 1.2945864468541792e-06,
+      "loss": 1.2112,
+      "step": 5150
+    },
+    {
+      "epoch": 0.8404935456285377,
+      "grad_norm": 6.328592777252197,
+      "learning_rate": 1.2691047076249852e-06,
+      "loss": 1.2982,
+      "step": 5160
+    },
+    {
+      "epoch": 0.8421224090890581,
+      "grad_norm": 11.03521728515625,
+      "learning_rate": 1.2438592486811007e-06,
+      "loss": 1.2062,
+      "step": 5170
+    },
+    {
+      "epoch": 0.8437512725495785,
+      "grad_norm": 7.168650150299072,
+      "learning_rate": 1.2188507532290094e-06,
+      "loss": 1.2764,
+      "step": 5180
+    },
+    {
+      "epoch": 0.845380136010099,
+      "grad_norm": 7.8710784912109375,
+      "learning_rate": 1.194079898062349e-06,
+      "loss": 1.1388,
+      "step": 5190
+    },
+    {
+      "epoch": 0.8470089994706194,
+      "grad_norm": 9.251112937927246,
+      "learning_rate": 1.1695473535436187e-06,
+      "loss": 1.1258,
+      "step": 5200
+    },
+    {
+      "epoch": 0.8486378629311399,
+      "grad_norm": 10.411331176757812,
+      "learning_rate": 1.145253783586011e-06,
+      "loss": 1.2281,
+      "step": 5210
+    },
+    {
+      "epoch": 0.8502667263916602,
+      "grad_norm": 13.52040958404541,
+      "learning_rate": 1.1211998456354656e-06,
+      "loss": 1.2075,
+      "step": 5220
+    },
+    {
+      "epoch": 0.8518955898521806,
+      "grad_norm": 8.210290908813477,
+      "learning_rate": 1.0973861906528692e-06,
+      "loss": 1.3148,
+      "step": 5230
+    },
+    {
+      "epoch": 0.8535244533127011,
+      "grad_norm": 7.864938259124756,
+      "learning_rate": 1.0738134630964326e-06,
+      "loss": 1.3147,
+      "step": 5240
+    },
+    {
+      "epoch": 0.8551533167732215,
+      "grad_norm": 8.113484382629395,
+      "learning_rate": 1.050482300904264e-06,
+      "loss": 1.2167,
+      "step": 5250
+    },
+    {
+      "epoch": 0.8567821802337419,
+      "grad_norm": 10.562095642089844,
+      "learning_rate": 1.0273933354770894e-06,
+      "loss": 1.3086,
+      "step": 5260
+    },
+    {
+      "epoch": 0.8584110436942624,
+      "grad_norm": 10.281449317932129,
+      "learning_rate": 1.004547191661178e-06,
+      "loss": 1.21,
+      "step": 5270
+    },
+    {
+      "epoch": 0.8600399071547827,
+      "grad_norm": 9.31655216217041,
+      "learning_rate": 9.819444877314299e-07,
+      "loss": 1.3514,
+      "step": 5280
+    },
+    {
+      "epoch": 0.8616687706153032,
+      "grad_norm": 8.388626098632812,
+      "learning_rate": 9.5958583537463e-07,
+      "loss": 1.2307,
+      "step": 5290
+    },
+    {
+      "epoch": 0.8632976340758236,
+      "grad_norm": 8.135517120361328,
+      "learning_rate": 9.374718396729188e-07,
+      "loss": 1.2003,
+      "step": 5300
+    },
+    {
+      "epoch": 0.864926497536344,
+      "grad_norm": 7.857133388519287,
+      "learning_rate": 9.156030990873932e-07,
+      "loss": 1.1086,
+      "step": 5310
+    },
+    {
+      "epoch": 0.8665553609968645,
+      "grad_norm": 10.863335609436035,
+      "learning_rate": 8.939802054419289e-07,
+      "loss": 1.2312,
+      "step": 5320
+    },
+    {
+      "epoch": 0.8681842244573849,
+      "grad_norm": 8.662337303161621,
+      "learning_rate": 8.726037439071555e-07,
+      "loss": 1.2288,
+      "step": 5330
+    },
+    {
+      "epoch": 0.8698130879179052,
+      "grad_norm": 8.542695045471191,
+      "learning_rate": 8.514742929846142e-07,
+      "loss": 1.1772,
+      "step": 5340
+    },
+    {
+      "epoch": 0.8714419513784257,
+      "grad_norm": 8.269294738769531,
+      "learning_rate": 8.305924244911178e-07,
+      "loss": 1.1143,
+      "step": 5350
+    },
+    {
+      "epoch": 0.8730708148389461,
+      "grad_norm": 7.3109354972839355,
+      "learning_rate": 8.099587035432654e-07,
+      "loss": 1.128,
+      "step": 5360
+    },
+    {
+      "epoch": 0.8746996782994666,
+      "grad_norm": 7.124019622802734,
+      "learning_rate": 7.895736885421468e-07,
+      "loss": 1.3185,
+      "step": 5370
+    },
+    {
+      "epoch": 0.876328541759987,
+      "grad_norm": 10.167950630187988,
+      "learning_rate": 7.694379311582401e-07,
+      "loss": 1.2222,
+      "step": 5380
+    },
+    {
+      "epoch": 0.8779574052205074,
+      "grad_norm": 9.3758544921875,
+      "learning_rate": 7.49551976316475e-07,
+      "loss": 1.23,
+      "step": 5390
+    },
+    {
+      "epoch": 0.8795862686810278,
+      "grad_norm": 10.248433113098145,
+      "learning_rate": 7.299163621814853e-07,
+      "loss": 1.2133,
+      "step": 5400
+    },
+    {
+      "epoch": 0.8812151321415482,
+      "grad_norm": 13.722402572631836,
+      "learning_rate": 7.105316201430512e-07,
+      "loss": 1.2697,
+      "step": 5410
+    },
+    {
+      "epoch": 0.8828439956020686,
+      "grad_norm": 8.881364822387695,
+      "learning_rate": 6.91398274801709e-07,
+      "loss": 1.18,
+      "step": 5420
+    },
+    {
+      "epoch": 0.8844728590625891,
+      "grad_norm": 6.859842300415039,
+      "learning_rate": 6.725168439545637e-07,
+      "loss": 1.1595,
+      "step": 5430
+    },
+    {
+      "epoch": 0.8861017225231095,
+      "grad_norm": 8.630757331848145,
+      "learning_rate": 6.53887838581273e-07,
+      "loss": 1.2345,
+      "step": 5440
+    },
+    {
+      "epoch": 0.88773058598363,
+      "grad_norm": 11.166703224182129,
+      "learning_rate": 6.355117628302121e-07,
+      "loss": 1.2953,
+      "step": 5450
+    },
+    {
+      "epoch": 0.8893594494441504,
+      "grad_norm": 7.578332901000977,
+      "learning_rate": 6.173891140048427e-07,
+      "loss": 1.1959,
+      "step": 5460
+    },
+    {
+      "epoch": 0.8909883129046707,
+      "grad_norm": 18.070478439331055,
+      "learning_rate": 5.995203825502393e-07,
+      "loss": 1.2072,
+      "step": 5470
+    },
+    {
+      "epoch": 0.8926171763651912,
+      "grad_norm": 9.904817581176758,
+      "learning_rate": 5.819060520398345e-07,
+      "loss": 1.2266,
+      "step": 5480
+    },
+    {
+      "epoch": 0.8942460398257116,
+      "grad_norm": 8.371089935302734,
+      "learning_rate": 5.645465991623167e-07,
+      "loss": 1.2003,
+      "step": 5490
+    },
+    {
+      "epoch": 0.895874903286232,
+      "grad_norm": 9.28614616394043,
+      "learning_rate": 5.474424937087353e-07,
+      "loss": 1.2134,
+      "step": 5500
+    },
+    {
+      "epoch": 0.895874903286232,
+      "eval_loss": 1.1204986572265625,
+      "eval_runtime": 102.9981,
+      "eval_samples_per_second": 25.098,
+      "eval_steps_per_second": 25.098,
+      "step": 5500
+    },
+    {
+      "epoch": 0.8975037667467525,
+      "grad_norm": 8.397842407226562,
+      "learning_rate": 5.305941985597929e-07,
+      "loss": 1.1799,
+      "step": 5510
+    },
+    {
+      "epoch": 0.8991326302072729,
+      "grad_norm": 5.709924697875977,
+      "learning_rate": 5.140021696733066e-07,
+      "loss": 1.233,
+      "step": 5520
+    },
+    {
+      "epoch": 0.9007614936677933,
+      "grad_norm": 8.849600791931152,
+      "learning_rate": 4.97666856071879e-07,
+      "loss": 1.3359,
+      "step": 5530
+    },
+    {
+      "epoch": 0.9023903571283137,
+      "grad_norm": 8.294559478759766,
+      "learning_rate": 4.815886998307439e-07,
+      "loss": 1.2746,
+      "step": 5540
+    },
+    {
+      "epoch": 0.9040192205888341,
+      "grad_norm": 8.917488098144531,
+      "learning_rate": 4.657681360657962e-07,
+      "loss": 1.1755,
+      "step": 5550
+    },
+    {
+      "epoch": 0.9056480840493546,
+      "grad_norm": 6.684938907623291,
+      "learning_rate": 4.502055929218241e-07,
+      "loss": 1.2818,
+      "step": 5560
+    },
+    {
+      "epoch": 0.907276947509875,
+      "grad_norm": 6.304566383361816,
+      "learning_rate": 4.34901491560924e-07,
+      "loss": 1.3162,
+      "step": 5570
+    },
+    {
+      "epoch": 0.9089058109703954,
+      "grad_norm": 16.449485778808594,
+      "learning_rate": 4.1985624615109134e-07,
+      "loss": 1.0536,
+      "step": 5580
+    },
+    {
+      "epoch": 0.9105346744309158,
+      "grad_norm": 9.711615562438965,
+      "learning_rate": 4.0507026385502747e-07,
+      "loss": 1.1989,
+      "step": 5590
+    },
+    {
+      "epoch": 0.9121635378914362,
+      "grad_norm": 9.28388786315918,
+      "learning_rate": 3.9054394481910507e-07,
+      "loss": 1.3322,
+      "step": 5600
+    },
+    {
+      "epoch": 0.9137924013519567,
+      "grad_norm": 7.323754787445068,
+      "learning_rate": 3.7627768216255244e-07,
+      "loss": 1.1938,
+      "step": 5610
+    },
+    {
+      "epoch": 0.9154212648124771,
+      "grad_norm": 7.1576385498046875,
+      "learning_rate": 3.6227186196680976e-07,
+      "loss": 1.2738,
+      "step": 5620
+    },
+    {
+      "epoch": 0.9170501282729975,
+      "grad_norm": 8.460196495056152,
+      "learning_rate": 3.485268632650751e-07,
+      "loss": 1.2285,
+      "step": 5630
+    },
+    {
+      "epoch": 0.918678991733518,
+      "grad_norm": 8.773188591003418,
+      "learning_rate": 3.350430580320574e-07,
+      "loss": 1.2892,
+      "step": 5640
+    },
+    {
+      "epoch": 0.9203078551940383,
+      "grad_norm": 7.650882720947266,
+      "learning_rate": 3.218208111738996e-07,
+      "loss": 1.2367,
+      "step": 5650
+    },
+    {
+      "epoch": 0.9219367186545587,
+      "grad_norm": 12.023070335388184,
+      "learning_rate": 3.088604805183126e-07,
+      "loss": 1.2383,
+      "step": 5660
+    },
+    {
+      "epoch": 0.9235655821150792,
+      "grad_norm": 10.934995651245117,
+      "learning_rate": 2.9616241680488713e-07,
+      "loss": 1.2227,
+      "step": 5670
+    },
+    {
+      "epoch": 0.9251944455755996,
+      "grad_norm": 8.503402709960938,
+      "learning_rate": 2.837269636755946e-07,
+      "loss": 1.3065,
+      "step": 5680
+    },
+    {
+      "epoch": 0.9268233090361201,
+      "grad_norm": 8.271093368530273,
+      "learning_rate": 2.7155445766550605e-07,
+      "loss": 1.0851,
+      "step": 5690
+    },
+    {
+      "epoch": 0.9284521724966405,
+      "grad_norm": 8.876206398010254,
+      "learning_rate": 2.5964522819366125e-07,
+      "loss": 1.2316,
+      "step": 5700
+    },
+    {
+      "epoch": 0.9300810359571609,
+      "grad_norm": 11.97021484375,
+      "learning_rate": 2.479995975541749e-07,
+      "loss": 1.2201,
+      "step": 5710
+    },
+    {
+      "epoch": 0.9317098994176813,
+      "grad_norm": 9.127492904663086,
+      "learning_rate": 2.3661788090750038e-07,
+      "loss": 1.2073,
+      "step": 5720
+    },
+    {
+      "epoch": 0.9333387628782017,
+      "grad_norm": 7.851465702056885,
+      "learning_rate": 2.255003862719074e-07,
+      "loss": 1.2395,
+      "step": 5730
+    },
+    {
+      "epoch": 0.9349676263387221,
+      "grad_norm": 11.95673942565918,
+      "learning_rate": 2.1464741451514447e-07,
+      "loss": 1.2248,
+      "step": 5740
+    },
+    {
+      "epoch": 0.9365964897992426,
+      "grad_norm": 7.73248291015625,
+      "learning_rate": 2.0405925934629423e-07,
+      "loss": 1.3249,
+      "step": 5750
+    },
+    {
+      "epoch": 0.938225353259763,
+      "grad_norm": 10.03739070892334,
+      "learning_rate": 1.9373620730783082e-07,
+      "loss": 1.1815,
+      "step": 5760
+    },
+    {
+      "epoch": 0.9398542167202835,
+      "grad_norm": 8.619077682495117,
+      "learning_rate": 1.836785377678596e-07,
+      "loss": 1.244,
+      "step": 5770
+    },
+    {
+      "epoch": 0.9414830801808038,
+      "grad_norm": 8.094244956970215,
+      "learning_rate": 1.738865229125597e-07,
+      "loss": 1.2881,
+      "step": 5780
+    },
+    {
+      "epoch": 0.9431119436413242,
+      "grad_norm": 9.411540031433105,
+      "learning_rate": 1.6436042773881666e-07,
+      "loss": 1.2496,
+      "step": 5790
+    },
+    {
+      "epoch": 0.9447408071018447,
+      "grad_norm": 9.164313316345215,
+      "learning_rate": 1.5510051004705263e-07,
+      "loss": 1.2901,
+      "step": 5800
+    },
+    {
+      "epoch": 0.9463696705623651,
+      "grad_norm": 10.367000579833984,
+      "learning_rate": 1.4610702043424628e-07,
+      "loss": 1.1452,
+      "step": 5810
+    },
+    {
+      "epoch": 0.9479985340228856,
+      "grad_norm": 8.201516151428223,
+      "learning_rate": 1.373802022871551e-07,
+      "loss": 1.2266,
+      "step": 5820
+    },
+    {
+      "epoch": 0.949627397483406,
+      "grad_norm": 8.908825874328613,
+      "learning_rate": 1.2892029177572817e-07,
+      "loss": 1.2799,
+      "step": 5830
+    },
+    {
+      "epoch": 0.9512562609439263,
+      "grad_norm": 7.841489315032959,
+      "learning_rate": 1.2072751784671043e-07,
+      "loss": 1.2983,
+      "step": 5840
+    },
+    {
+      "epoch": 0.9528851244044468,
+      "grad_norm": 9.294130325317383,
+      "learning_rate": 1.1280210221745192e-07,
+      "loss": 1.2792,
+      "step": 5850
+    },
+    {
+      "epoch": 0.9545139878649672,
+      "grad_norm": 7.930861473083496,
+      "learning_rate": 1.0514425936990369e-07,
+      "loss": 1.1413,
+      "step": 5860
+    },
+    {
+      "epoch": 0.9561428513254876,
+      "grad_norm": 9.744686126708984,
+      "learning_rate": 9.775419654481588e-08,
+      "loss": 1.1266,
+      "step": 5870
+    },
+    {
+      "epoch": 0.9577717147860081,
+      "grad_norm": 7.7475905418396,
+      "learning_rate": 9.063211373613102e-08,
+      "loss": 1.4206,
+      "step": 5880
+    },
+    {
+      "epoch": 0.9594005782465285,
+      "grad_norm": 5.828322887420654,
+      "learning_rate": 8.3778203685565e-08,
+      "loss": 1.2971,
+      "step": 5890
+    },
+    {
+      "epoch": 0.961029441707049,
+      "grad_norm": 10.001614570617676,
+      "learning_rate": 7.71926518773991e-08,
+      "loss": 1.2998,
+      "step": 5900
+    },
+    {
+      "epoch": 0.9626583051675693,
+      "grad_norm": 9.513165473937988,
+      "learning_rate": 7.087563653345286e-08,
+      "loss": 1.1007,
+      "step": 5910
+    },
+    {
+      "epoch": 0.9642871686280897,
+      "grad_norm": 9.425087928771973,
+      "learning_rate": 6.482732860826679e-08,
+      "loss": 1.245,
+      "step": 5920
+    },
+    {
+      "epoch": 0.9659160320886102,
+      "grad_norm": 8.666535377502441,
+      "learning_rate": 5.90478917844739e-08,
+      "loss": 1.2323,
+      "step": 5930
+    },
+    {
+      "epoch": 0.9675448955491306,
+      "grad_norm": 8.544173240661621,
+      "learning_rate": 5.3537482468366544e-08,
+      "loss": 1.2152,
+      "step": 5940
+    },
+    {
+      "epoch": 0.969173759009651,
+      "grad_norm": 9.873296737670898,
+      "learning_rate": 4.829624978567204e-08,
+      "loss": 1.216,
+      "step": 5950
+    },
+    {
+      "epoch": 0.9708026224701715,
+      "grad_norm": 6.528816223144531,
+      "learning_rate": 4.332433557750704e-08,
+      "loss": 1.36,
+      "step": 5960
+    },
+    {
+      "epoch": 0.9724314859306918,
+      "grad_norm": 13.494250297546387,
+      "learning_rate": 3.862187439654608e-08,
+      "loss": 1.1844,
+      "step": 5970
+    },
+    {
+      "epoch": 0.9740603493912123,
+      "grad_norm": 10.320327758789062,
+      "learning_rate": 3.41889935033779e-08,
+      "loss": 1.3135,
+      "step": 5980
+    },
+    {
+      "epoch": 0.9756892128517327,
+      "grad_norm": 8.843477249145508,
+      "learning_rate": 3.002581286305817e-08,
+      "loss": 1.2222,
+      "step": 5990
+    },
+    {
+      "epoch": 0.9773180763122531,
+      "grad_norm": 7.343149662017822,
+      "learning_rate": 2.6132445141872074e-08,
+      "loss": 1.2437,
+      "step": 6000
+    },
+    {
+      "epoch": 0.9773180763122531,
+      "eval_loss": 1.1204884052276611,
+      "eval_runtime": 102.4877,
+      "eval_samples_per_second": 25.223,
+      "eval_steps_per_second": 25.223,
+      "step": 6000
+    },
+    {
+      "epoch": 0.9789469397727736,
+      "grad_norm": 10.39957332611084,
+      "learning_rate": 2.250899570427345e-08,
+      "loss": 1.174,
+      "step": 6010
+    },
+    {
+      "epoch": 0.980575803233294,
+      "grad_norm": 7.477624416351318,
+      "learning_rate": 1.91555626100437e-08,
+      "loss": 1.3035,
+      "step": 6020
+    },
+    {
+      "epoch": 0.9822046666938143,
+      "grad_norm": 7.673905849456787,
+      "learning_rate": 1.6072236611629487e-08,
+      "loss": 1.2888,
+      "step": 6030
+    },
+    {
+      "epoch": 0.9838335301543348,
+      "grad_norm": 8.837902069091797,
+      "learning_rate": 1.325910115169471e-08,
+      "loss": 1.3579,
+      "step": 6040
+    },
+    {
+      "epoch": 0.9854623936148552,
+      "grad_norm": 9.265447616577148,
+      "learning_rate": 1.0716232360856726e-08,
+      "loss": 1.2222,
+      "step": 6050
+    },
+    {
+      "epoch": 0.9870912570753757,
+      "grad_norm": 11.45450210571289,
+      "learning_rate": 8.44369905562692e-09,
+      "loss": 1.2786,
+      "step": 6060
+    },
+    {
+      "epoch": 0.9887201205358961,
+      "grad_norm": 9.006537437438965,
+      "learning_rate": 6.441562736551054e-09,
+      "loss": 1.2923,
+      "step": 6070
+    },
+    {
+      "epoch": 0.9903489839964165,
+      "grad_norm": 7.819212913513184,
+      "learning_rate": 4.709877586540623e-09,
+      "loss": 1.1388,
+      "step": 6080
+    },
+    {
+      "epoch": 0.991977847456937,
+      "grad_norm": 10.596216201782227,
+      "learning_rate": 3.2486904694128963e-09,
+      "loss": 1.1399,
+      "step": 6090
+    },
+    {
+      "epoch": 0.9936067109174573,
+      "grad_norm": 9.980060577392578,
+      "learning_rate": 2.0580409286152792e-09,
+      "loss": 1.1898,
+      "step": 6100
+    },
+    {
+      "epoch": 0.9952355743779777,
+      "grad_norm": 7.423088073730469,
+      "learning_rate": 1.1379611861594974e-09,
+      "loss": 1.234,
+      "step": 6110
+    },
+    {
+      "epoch": 0.9968644378384982,
+      "grad_norm": 8.370770454406738,
+      "learning_rate": 4.884761417489614e-10,
+      "loss": 1.2935,
+      "step": 6120
+    },
+    {
+      "epoch": 0.9984933012990186,
+      "grad_norm": 8.747970581054688,
+      "learning_rate": 1.0960337210597083e-10,
+      "loss": 1.2299,
+      "step": 6130
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 6139,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.1514993505940275e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}