diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4742 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 200,
+  "global_step": 668,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0029940119760479044,
+      "grad_norm": 6.117997680045978,
+      "learning_rate": 9.999944704978835e-06,
+      "loss": 0.3833,
+      "step": 1
+    },
+    {
+      "epoch": 0.005988023952095809,
+      "grad_norm": 3.1975245454800962,
+      "learning_rate": 9.999778821138357e-06,
+      "loss": 0.2683,
+      "step": 2
+    },
+    {
+      "epoch": 0.008982035928143712,
+      "grad_norm": 3.2522961944549746,
+      "learning_rate": 9.999502352147583e-06,
+      "loss": 0.2628,
+      "step": 3
+    },
+    {
+      "epoch": 0.011976047904191617,
+      "grad_norm": 3.0562451155008103,
+      "learning_rate": 9.999115304121459e-06,
+      "loss": 0.2176,
+      "step": 4
+    },
+    {
+      "epoch": 0.014970059880239521,
+      "grad_norm": 3.390219170595323,
+      "learning_rate": 9.998617685620715e-06,
+      "loss": 0.2365,
+      "step": 5
+    },
+    {
+      "epoch": 0.017964071856287425,
+      "grad_norm": 2.958463130835355,
+      "learning_rate": 9.998009507651683e-06,
+      "loss": 0.2183,
+      "step": 6
+    },
+    {
+      "epoch": 0.020958083832335328,
+      "grad_norm": 2.836726535613556,
+      "learning_rate": 9.997290783666048e-06,
+      "loss": 0.2358,
+      "step": 7
+    },
+    {
+      "epoch": 0.023952095808383235,
+      "grad_norm": 2.7930130687626953,
+      "learning_rate": 9.996461529560553e-06,
+      "loss": 0.2369,
+      "step": 8
+    },
+    {
+      "epoch": 0.02694610778443114,
+      "grad_norm": 2.7223358618217057,
+      "learning_rate": 9.995521763676645e-06,
+      "loss": 0.274,
+      "step": 9
+    },
+    {
+      "epoch": 0.029940119760479042,
+      "grad_norm": 2.5905866254162473,
+      "learning_rate": 9.994471506800078e-06,
+      "loss": 0.2188,
+      "step": 10
+    },
+    {
+      "epoch": 0.03293413173652695,
+      "grad_norm": 2.1380941439974803,
+      "learning_rate": 9.993310782160439e-06,
+      "loss": 0.2355,
+      "step": 11
+    },
+    {
+      "epoch": 0.03592814371257485,
+      "grad_norm": 2.3265707275100627,
+      "learning_rate": 9.992039615430648e-06,
+      "loss": 0.2407,
+      "step": 12
+    },
+    {
+      "epoch": 0.038922155688622756,
+      "grad_norm": 2.336076725373038,
+      "learning_rate": 9.99065803472638e-06,
+      "loss": 0.2455,
+      "step": 13
+    },
+    {
+      "epoch": 0.041916167664670656,
+      "grad_norm": 2.533114628685348,
+      "learning_rate": 9.989166070605447e-06,
+      "loss": 0.2292,
+      "step": 14
+    },
+    {
+      "epoch": 0.04491017964071856,
+      "grad_norm": 2.645365490927225,
+      "learning_rate": 9.98756375606713e-06,
+      "loss": 0.2427,
+      "step": 15
+    },
+    {
+      "epoch": 0.04790419161676647,
+      "grad_norm": 2.7143236569768407,
+      "learning_rate": 9.985851126551428e-06,
+      "loss": 0.2502,
+      "step": 16
+    },
+    {
+      "epoch": 0.05089820359281437,
+      "grad_norm": 2.083879111799519,
+      "learning_rate": 9.9840282199383e-06,
+      "loss": 0.2037,
+      "step": 17
+    },
+    {
+      "epoch": 0.05389221556886228,
+      "grad_norm": 2.861576377218155,
+      "learning_rate": 9.982095076546806e-06,
+      "loss": 0.2663,
+      "step": 18
+    },
+    {
+      "epoch": 0.05688622754491018,
+      "grad_norm": 3.0011630876552737,
+      "learning_rate": 9.980051739134235e-06,
+      "loss": 0.28,
+      "step": 19
+    },
+    {
+      "epoch": 0.059880239520958084,
+      "grad_norm": 2.276848689330839,
+      "learning_rate": 9.977898252895133e-06,
+      "loss": 0.2504,
+      "step": 20
+    },
+    {
+      "epoch": 0.06287425149700598,
+      "grad_norm": 2.2156073937150587,
+      "learning_rate": 9.975634665460333e-06,
+      "loss": 0.2504,
+      "step": 21
+    },
+    {
+      "epoch": 0.0658682634730539,
+      "grad_norm": 2.429004314466581,
+      "learning_rate": 9.973261026895878e-06,
+      "loss": 0.2548,
+      "step": 22
+    },
+    {
+      "epoch": 0.0688622754491018,
+      "grad_norm": 4.26629618129125,
+      "learning_rate": 9.970777389701927e-06,
+      "loss": 0.2513,
+      "step": 23
+    },
+    {
+      "epoch": 0.0718562874251497,
+      "grad_norm": 2.265323486541379,
+      "learning_rate": 9.968183808811586e-06,
+      "loss": 0.2354,
+      "step": 24
+    },
+    {
+      "epoch": 0.0748502994011976,
+      "grad_norm": 2.5362989276922536,
+      "learning_rate": 9.965480341589702e-06,
+      "loss": 0.2456,
+      "step": 25
+    },
+    {
+      "epoch": 0.07784431137724551,
+      "grad_norm": 3.0650672030555213,
+      "learning_rate": 9.962667047831585e-06,
+      "loss": 0.2628,
+      "step": 26
+    },
+    {
+      "epoch": 0.08083832335329341,
+      "grad_norm": 2.3354543514578197,
+      "learning_rate": 9.95974398976169e-06,
+      "loss": 0.257,
+      "step": 27
+    },
+    {
+      "epoch": 0.08383233532934131,
+      "grad_norm": 2.5211750182428183,
+      "learning_rate": 9.95671123203224e-06,
+      "loss": 0.2539,
+      "step": 28
+    },
+    {
+      "epoch": 0.08682634730538923,
+      "grad_norm": 2.3774553044534894,
+      "learning_rate": 9.953568841721796e-06,
+      "loss": 0.2442,
+      "step": 29
+    },
+    {
+      "epoch": 0.08982035928143713,
+      "grad_norm": 2.243412039133249,
+      "learning_rate": 9.950316888333775e-06,
+      "loss": 0.2448,
+      "step": 30
+    },
+    {
+      "epoch": 0.09281437125748503,
+      "grad_norm": 2.1016873719908538,
+      "learning_rate": 9.946955443794908e-06,
+      "loss": 0.2296,
+      "step": 31
+    },
+    {
+      "epoch": 0.09580838323353294,
+      "grad_norm": 2.685551882732653,
+      "learning_rate": 9.943484582453653e-06,
+      "loss": 0.2499,
+      "step": 32
+    },
+    {
+      "epoch": 0.09880239520958084,
+      "grad_norm": 2.5505305638437905,
+      "learning_rate": 9.939904381078553e-06,
+      "loss": 0.2292,
+      "step": 33
+    },
+    {
+      "epoch": 0.10179640718562874,
+      "grad_norm": 2.6599812616819767,
+      "learning_rate": 9.93621491885653e-06,
+      "loss": 0.2615,
+      "step": 34
+    },
+    {
+      "epoch": 0.10479041916167664,
+      "grad_norm": 3.2234811959576635,
+      "learning_rate": 9.932416277391144e-06,
+      "loss": 0.2222,
+      "step": 35
+    },
+    {
+      "epoch": 0.10778443113772455,
+      "grad_norm": 3.0156641432384617,
+      "learning_rate": 9.928508540700775e-06,
+      "loss": 0.2747,
+      "step": 36
+    },
+    {
+      "epoch": 0.11077844311377245,
+      "grad_norm": 2.576757327929195,
+      "learning_rate": 9.924491795216777e-06,
+      "loss": 0.2784,
+      "step": 37
+    },
+    {
+      "epoch": 0.11377245508982035,
+      "grad_norm": 2.2486559994912536,
+      "learning_rate": 9.920366129781564e-06,
+      "loss": 0.235,
+      "step": 38
+    },
+    {
+      "epoch": 0.11676646706586827,
+      "grad_norm": 2.404758924937487,
+      "learning_rate": 9.916131635646635e-06,
+      "loss": 0.2211,
+      "step": 39
+    },
+    {
+      "epoch": 0.11976047904191617,
+      "grad_norm": 2.9139181018855793,
+      "learning_rate": 9.91178840647057e-06,
+      "loss": 0.2637,
+      "step": 40
+    },
+    {
+      "epoch": 0.12275449101796407,
+      "grad_norm": 2.328332716917134,
+      "learning_rate": 9.907336538316946e-06,
+      "loss": 0.2466,
+      "step": 41
+    },
+    {
+      "epoch": 0.12574850299401197,
+      "grad_norm": 1.9568988576476365,
+      "learning_rate": 9.902776129652223e-06,
+      "loss": 0.1933,
+      "step": 42
+    },
+    {
+      "epoch": 0.12874251497005987,
+      "grad_norm": 1.952019224391899,
+      "learning_rate": 9.898107281343557e-06,
+      "loss": 0.2246,
+      "step": 43
+    },
+    {
+      "epoch": 0.1317365269461078,
+      "grad_norm": 1.9707959674254527,
+      "learning_rate": 9.893330096656576e-06,
+      "loss": 0.1931,
+      "step": 44
+    },
+    {
+      "epoch": 0.1347305389221557,
+      "grad_norm": 2.077723999022485,
+      "learning_rate": 9.888444681253087e-06,
+      "loss": 0.2064,
+      "step": 45
+    },
+    {
+      "epoch": 0.1377245508982036,
+      "grad_norm": 2.3940840441042224,
+      "learning_rate": 9.883451143188753e-06,
+      "loss": 0.2367,
+      "step": 46
+    },
+    {
+      "epoch": 0.1407185628742515,
+      "grad_norm": 2.0046109515640596,
+      "learning_rate": 9.878349592910694e-06,
+      "loss": 0.2025,
+      "step": 47
+    },
+    {
+      "epoch": 0.1437125748502994,
+      "grad_norm": 2.7338838515360293,
+      "learning_rate": 9.873140143255035e-06,
+      "loss": 0.2669,
+      "step": 48
+    },
+    {
+      "epoch": 0.1467065868263473,
+      "grad_norm": 2.5363253348947423,
+      "learning_rate": 9.867822909444435e-06,
+      "loss": 0.2403,
+      "step": 49
+    },
+    {
+      "epoch": 0.1497005988023952,
+      "grad_norm": 2.1880739780161234,
+      "learning_rate": 9.862398009085511e-06,
+      "loss": 0.2172,
+      "step": 50
+    },
+    {
+      "epoch": 0.15269461077844312,
+      "grad_norm": 2.338108135734929,
+      "learning_rate": 9.856865562166256e-06,
+      "loss": 0.2377,
+      "step": 51
+    },
+    {
+      "epoch": 0.15568862275449102,
+      "grad_norm": 2.1926445321637265,
+      "learning_rate": 9.851225691053382e-06,
+      "loss": 0.2168,
+      "step": 52
+    },
+    {
+      "epoch": 0.15868263473053892,
+      "grad_norm": 2.555721102748371,
+      "learning_rate": 9.8454785204896e-06,
+      "loss": 0.2544,
+      "step": 53
+    },
+    {
+      "epoch": 0.16167664670658682,
+      "grad_norm": 2.1621440865534236,
+      "learning_rate": 9.83962417759088e-06,
+      "loss": 0.2054,
+      "step": 54
+    },
+    {
+      "epoch": 0.16467065868263472,
+      "grad_norm": 2.4738711157695144,
+      "learning_rate": 9.833662791843628e-06,
+      "loss": 0.2365,
+      "step": 55
+    },
+    {
+      "epoch": 0.16766467065868262,
+      "grad_norm": 2.212910536300774,
+      "learning_rate": 9.827594495101824e-06,
+      "loss": 0.2351,
+      "step": 56
+    },
+    {
+      "epoch": 0.17065868263473055,
+      "grad_norm": 2.253513003544616,
+      "learning_rate": 9.821419421584108e-06,
+      "loss": 0.2411,
+      "step": 57
+    },
+    {
+      "epoch": 0.17365269461077845,
+      "grad_norm": 2.4339035524598382,
+      "learning_rate": 9.815137707870806e-06,
+      "loss": 0.2832,
+      "step": 58
+    },
+    {
+      "epoch": 0.17664670658682635,
+      "grad_norm": 2.4150049198503343,
+      "learning_rate": 9.808749492900917e-06,
+      "loss": 0.2543,
+      "step": 59
+    },
+    {
+      "epoch": 0.17964071856287425,
+      "grad_norm": 2.175000897091873,
+      "learning_rate": 9.802254917969033e-06,
+      "loss": 0.2311,
+      "step": 60
+    },
+    {
+      "epoch": 0.18263473053892215,
+      "grad_norm": 2.6132017007557593,
+      "learning_rate": 9.795654126722218e-06,
+      "loss": 0.2435,
+      "step": 61
+    },
+    {
+      "epoch": 0.18562874251497005,
+      "grad_norm": 2.272149002800371,
+      "learning_rate": 9.788947265156828e-06,
+      "loss": 0.2298,
+      "step": 62
+    },
+    {
+      "epoch": 0.18862275449101795,
+      "grad_norm": 2.068890167925924,
+      "learning_rate": 9.782134481615282e-06,
+      "loss": 0.196,
+      "step": 63
+    },
+    {
+      "epoch": 0.19161676646706588,
+      "grad_norm": 2.1546640316835264,
+      "learning_rate": 9.775215926782788e-06,
+      "loss": 0.2332,
+      "step": 64
+    },
+    {
+      "epoch": 0.19461077844311378,
+      "grad_norm": 2.4348961669228513,
+      "learning_rate": 9.768191753683997e-06,
+      "loss": 0.2271,
+      "step": 65
+    },
+    {
+      "epoch": 0.19760479041916168,
+      "grad_norm": 2.324206636855764,
+      "learning_rate": 9.761062117679632e-06,
+      "loss": 0.2209,
+      "step": 66
+    },
+    {
+      "epoch": 0.20059880239520958,
+      "grad_norm": 2.245460252023205,
+      "learning_rate": 9.75382717646304e-06,
+      "loss": 0.2347,
+      "step": 67
+    },
+    {
+      "epoch": 0.20359281437125748,
+      "grad_norm": 2.0296344750151376,
+      "learning_rate": 9.746487090056712e-06,
+      "loss": 0.2083,
+      "step": 68
+    },
+    {
+      "epoch": 0.20658682634730538,
+      "grad_norm": 1.9162096792791554,
+      "learning_rate": 9.739042020808746e-06,
+      "loss": 0.1965,
+      "step": 69
+    },
+    {
+      "epoch": 0.20958083832335328,
+      "grad_norm": 2.32454226237422,
+      "learning_rate": 9.73149213338924e-06,
+      "loss": 0.2225,
+      "step": 70
+    },
+    {
+      "epoch": 0.2125748502994012,
+      "grad_norm": 2.282715046904575,
+      "learning_rate": 9.72383759478667e-06,
+      "loss": 0.2523,
+      "step": 71
+    },
+    {
+      "epoch": 0.2155688622754491,
+      "grad_norm": 2.011540257513772,
+      "learning_rate": 9.71607857430419e-06,
+      "loss": 0.1903,
+      "step": 72
+    },
+    {
+      "epoch": 0.218562874251497,
+      "grad_norm": 2.4224201423781713,
+      "learning_rate": 9.708215243555875e-06,
+      "loss": 0.2553,
+      "step": 73
+    },
+    {
+      "epoch": 0.2215568862275449,
+      "grad_norm": 2.082486040466063,
+      "learning_rate": 9.700247776462944e-06,
+      "loss": 0.2233,
+      "step": 74
+    },
+    {
+      "epoch": 0.2245508982035928,
+      "grad_norm": 2.6097490240776593,
+      "learning_rate": 9.6921763492499e-06,
+      "loss": 0.2358,
+      "step": 75
+    },
+    {
+      "epoch": 0.2275449101796407,
+      "grad_norm": 1.92989037092677,
+      "learning_rate": 9.68400114044064e-06,
+      "loss": 0.1991,
+      "step": 76
+    },
+    {
+      "epoch": 0.23053892215568864,
+      "grad_norm": 2.09339461489786,
+      "learning_rate": 9.6757223308545e-06,
+      "loss": 0.191,
+      "step": 77
+    },
+    {
+      "epoch": 0.23353293413173654,
+      "grad_norm": 2.1425948449278738,
+      "learning_rate": 9.667340103602263e-06,
+      "loss": 0.2261,
+      "step": 78
+    },
+    {
+      "epoch": 0.23652694610778444,
+      "grad_norm": 2.0308665771819,
+      "learning_rate": 9.658854644082099e-06,
+      "loss": 0.211,
+      "step": 79
+    },
+    {
+      "epoch": 0.23952095808383234,
+      "grad_norm": 2.795787212534617,
+      "learning_rate": 9.650266139975474e-06,
+      "loss": 0.2668,
+      "step": 80
+    },
+    {
+      "epoch": 0.24251497005988024,
+      "grad_norm": 1.9992301893061497,
+      "learning_rate": 9.641574781242999e-06,
+      "loss": 0.2127,
+      "step": 81
+    },
+    {
+      "epoch": 0.24550898203592814,
+      "grad_norm": 2.0010692121312137,
+      "learning_rate": 9.632780760120217e-06,
+      "loss": 0.196,
+      "step": 82
+    },
+    {
+      "epoch": 0.24850299401197604,
+      "grad_norm": 2.079998789530112,
+      "learning_rate": 9.62388427111336e-06,
+      "loss": 0.2072,
+      "step": 83
+    },
+    {
+      "epoch": 0.25149700598802394,
+      "grad_norm": 1.9260878395983723,
+      "learning_rate": 9.614885510995047e-06,
+      "loss": 0.1971,
+      "step": 84
+    },
+    {
+      "epoch": 0.25449101796407186,
+      "grad_norm": 1.78552012579504,
+      "learning_rate": 9.605784678799934e-06,
+      "loss": 0.1706,
+      "step": 85
+    },
+    {
+      "epoch": 0.25748502994011974,
+      "grad_norm": 2.5964864962564222,
+      "learning_rate": 9.596581975820304e-06,
+      "loss": 0.2088,
+      "step": 86
+    },
+    {
+      "epoch": 0.26047904191616766,
+      "grad_norm": 2.1114710742991902,
+      "learning_rate": 9.587277605601617e-06,
+      "loss": 0.2192,
+      "step": 87
+    },
+    {
+      "epoch": 0.2634730538922156,
+      "grad_norm": 2.5667976658069005,
+      "learning_rate": 9.577871773938013e-06,
+      "loss": 0.2317,
+      "step": 88
+    },
+    {
+      "epoch": 0.26646706586826346,
+      "grad_norm": 2.349654868760893,
+      "learning_rate": 9.568364688867757e-06,
+      "loss": 0.2683,
+      "step": 89
+    },
+    {
+      "epoch": 0.2694610778443114,
+      "grad_norm": 1.9674800153835732,
+      "learning_rate": 9.558756560668637e-06,
+      "loss": 0.2057,
+      "step": 90
+    },
+    {
+      "epoch": 0.27245508982035926,
+      "grad_norm": 2.3236868805418163,
+      "learning_rate": 9.549047601853313e-06,
+      "loss": 0.2248,
+      "step": 91
+    },
+    {
+      "epoch": 0.2754491017964072,
+      "grad_norm": 2.1653238463144704,
+      "learning_rate": 9.539238027164618e-06,
+      "loss": 0.2295,
+      "step": 92
+    },
+    {
+      "epoch": 0.27844311377245506,
+      "grad_norm": 2.018840627990619,
+      "learning_rate": 9.52932805357081e-06,
+      "loss": 0.1964,
+      "step": 93
+    },
+    {
+      "epoch": 0.281437125748503,
+      "grad_norm": 2.3183638369692883,
+      "learning_rate": 9.519317900260769e-06,
+      "loss": 0.2222,
+      "step": 94
+    },
+    {
+      "epoch": 0.2844311377245509,
+      "grad_norm": 1.8093584958311555,
+      "learning_rate": 9.509207788639148e-06,
+      "loss": 0.1991,
+      "step": 95
+    },
+    {
+      "epoch": 0.2874251497005988,
+      "grad_norm": 1.8942095874597822,
+      "learning_rate": 9.498997942321484e-06,
+      "loss": 0.2204,
+      "step": 96
+    },
+    {
+      "epoch": 0.2904191616766467,
+      "grad_norm": 1.803734648495183,
+      "learning_rate": 9.488688587129243e-06,
+      "loss": 0.1993,
+      "step": 97
+    },
+    {
+      "epoch": 0.2934131736526946,
+      "grad_norm": 1.7787760345188386,
+      "learning_rate": 9.47827995108483e-06,
+      "loss": 0.1823,
+      "step": 98
+    },
+    {
+      "epoch": 0.2964071856287425,
+      "grad_norm": 2.075994350667922,
+      "learning_rate": 9.467772264406545e-06,
+      "loss": 0.2171,
+      "step": 99
+    },
+    {
+      "epoch": 0.2994011976047904,
+      "grad_norm": 2.197915358130509,
+      "learning_rate": 9.457165759503492e-06,
+      "loss": 0.2117,
+      "step": 100
+    },
+    {
+      "epoch": 0.3023952095808383,
+      "grad_norm": 2.048805744821835,
+      "learning_rate": 9.446460670970436e-06,
+      "loss": 0.1853,
+      "step": 101
+    },
+    {
+      "epoch": 0.30538922155688625,
+      "grad_norm": 2.032178792813892,
+      "learning_rate": 9.435657235582616e-06,
+      "loss": 0.1825,
+      "step": 102
+    },
+    {
+      "epoch": 0.3083832335329341,
+      "grad_norm": 2.332771226287244,
+      "learning_rate": 9.424755692290507e-06,
+      "loss": 0.2167,
+      "step": 103
+    },
+    {
+      "epoch": 0.31137724550898205,
+      "grad_norm": 2.184502188058908,
+      "learning_rate": 9.413756282214538e-06,
+      "loss": 0.2215,
+      "step": 104
+    },
+    {
+      "epoch": 0.3143712574850299,
+      "grad_norm": 2.0604465245760406,
+      "learning_rate": 9.402659248639749e-06,
+      "loss": 0.2192,
+      "step": 105
+    },
+    {
+      "epoch": 0.31736526946107785,
+      "grad_norm": 2.1226990068644516,
+      "learning_rate": 9.391464837010428e-06,
+      "loss": 0.2288,
+      "step": 106
+    },
+    {
+      "epoch": 0.3203592814371258,
+      "grad_norm": 2.4481247861707085,
+      "learning_rate": 9.380173294924661e-06,
+      "loss": 0.251,
+      "step": 107
+    },
+    {
+      "epoch": 0.32335329341317365,
+      "grad_norm": 2.2351771333662476,
+      "learning_rate": 9.368784872128877e-06,
+      "loss": 0.2432,
+      "step": 108
+    },
+    {
+      "epoch": 0.3263473053892216,
+      "grad_norm": 2.2410637672786757,
+      "learning_rate": 9.357299820512305e-06,
+      "loss": 0.2261,
+      "step": 109
+    },
+    {
+      "epoch": 0.32934131736526945,
+      "grad_norm": 2.128137603151087,
+      "learning_rate": 9.345718394101412e-06,
+      "loss": 0.2248,
+      "step": 110
+    },
+    {
+      "epoch": 0.3323353293413174,
+      "grad_norm": 2.30863891503116,
+      "learning_rate": 9.334040849054288e-06,
+      "loss": 0.2685,
+      "step": 111
+    },
+    {
+      "epoch": 0.33532934131736525,
+      "grad_norm": 2.2772295215097453,
+      "learning_rate": 9.322267443654974e-06,
+      "loss": 0.2304,
+      "step": 112
+    },
+    {
+      "epoch": 0.3383233532934132,
+      "grad_norm": 2.2632160965994275,
+      "learning_rate": 9.310398438307747e-06,
+      "loss": 0.2569,
+      "step": 113
+    },
+    {
+      "epoch": 0.3413173652694611,
+      "grad_norm": 1.988970082016937,
+      "learning_rate": 9.29843409553137e-06,
+      "loss": 0.2033,
+      "step": 114
+    },
+    {
+      "epoch": 0.344311377245509,
+      "grad_norm": 1.9623496808226022,
+      "learning_rate": 9.286374679953278e-06,
+      "loss": 0.219,
+      "step": 115
+    },
+    {
+      "epoch": 0.3473053892215569,
+      "grad_norm": 1.9173816017459178,
+      "learning_rate": 9.274220458303727e-06,
+      "loss": 0.2047,
+      "step": 116
+    },
+    {
+      "epoch": 0.3502994011976048,
+      "grad_norm": 2.5076140499550528,
+      "learning_rate": 9.261971699409893e-06,
+      "loss": 0.2552,
+      "step": 117
+    },
+    {
+      "epoch": 0.3532934131736527,
+      "grad_norm": 2.253503988850195,
+      "learning_rate": 9.249628674189928e-06,
+      "loss": 0.2265,
+      "step": 118
+    },
+    {
+      "epoch": 0.3562874251497006,
+      "grad_norm": 2.0852729401535424,
+      "learning_rate": 9.237191655646972e-06,
+      "loss": 0.2304,
+      "step": 119
+    },
+    {
+      "epoch": 0.3592814371257485,
+      "grad_norm": 2.1842188928521895,
+      "learning_rate": 9.224660918863104e-06,
+      "loss": 0.215,
+      "step": 120
+    },
+    {
+      "epoch": 0.36227544910179643,
+      "grad_norm": 2.074645444866756,
+      "learning_rate": 9.212036740993265e-06,
+      "loss": 0.2222,
+      "step": 121
+    },
+    {
+      "epoch": 0.3652694610778443,
+      "grad_norm": 2.2896726560732117,
+      "learning_rate": 9.199319401259132e-06,
+      "loss": 0.202,
+      "step": 122
+    },
+    {
+      "epoch": 0.36826347305389223,
+      "grad_norm": 2.024384563609649,
+      "learning_rate": 9.186509180942928e-06,
+      "loss": 0.1854,
+      "step": 123
+    },
+    {
+      "epoch": 0.3712574850299401,
+      "grad_norm": 2.171289891749186,
+      "learning_rate": 9.173606363381218e-06,
+      "loss": 0.2313,
+      "step": 124
+    },
+    {
+      "epoch": 0.37425149700598803,
+      "grad_norm": 2.0820059516896157,
+      "learning_rate": 9.16061123395863e-06,
+      "loss": 0.1897,
+      "step": 125
+    },
+    {
+      "epoch": 0.3772455089820359,
+      "grad_norm": 2.046948741309251,
+      "learning_rate": 9.147524080101543e-06,
+      "loss": 0.1974,
+      "step": 126
+    },
+    {
+      "epoch": 0.38023952095808383,
+      "grad_norm": 2.242555237414818,
+      "learning_rate": 9.134345191271742e-06,
+      "loss": 0.2242,
+      "step": 127
+    },
+    {
+      "epoch": 0.38323353293413176,
+      "grad_norm": 2.1750997069746063,
+      "learning_rate": 9.121074858959997e-06,
+      "loss": 0.253,
+      "step": 128
+    },
+    {
+      "epoch": 0.38622754491017963,
+      "grad_norm": 2.204135579311514,
+      "learning_rate": 9.107713376679634e-06,
+      "loss": 0.2426,
+      "step": 129
+    },
+    {
+      "epoch": 0.38922155688622756,
+      "grad_norm": 2.095720109192354,
+      "learning_rate": 9.094261039960028e-06,
+      "loss": 0.2091,
+      "step": 130
+    },
+    {
+      "epoch": 0.39221556886227543,
+      "grad_norm": 2.248651957463789,
+      "learning_rate": 9.08071814634008e-06,
+      "loss": 0.227,
+      "step": 131
+    },
+    {
+      "epoch": 0.39520958083832336,
+      "grad_norm": 2.204245096267591,
+      "learning_rate": 9.067084995361623e-06,
+      "loss": 0.1979,
+      "step": 132
+    },
+    {
+      "epoch": 0.39820359281437123,
+      "grad_norm": 2.1708347520768627,
+      "learning_rate": 9.053361888562807e-06,
+      "loss": 0.2433,
+      "step": 133
+    },
+    {
+      "epoch": 0.40119760479041916,
+      "grad_norm": 2.294015025457841,
+      "learning_rate": 9.039549129471423e-06,
+      "loss": 0.2158,
+      "step": 134
+    },
+    {
+      "epoch": 0.4041916167664671,
+      "grad_norm": 1.7437178376062743,
+      "learning_rate": 9.025647023598196e-06,
+      "loss": 0.1821,
+      "step": 135
+    },
+    {
+      "epoch": 0.40718562874251496,
+      "grad_norm": 1.9622075743388179,
+      "learning_rate": 9.011655878430018e-06,
+      "loss": 0.1803,
+      "step": 136
+    },
+    {
+      "epoch": 0.4101796407185629,
+      "grad_norm": 2.4118812741006566,
+      "learning_rate": 8.99757600342316e-06,
+      "loss": 0.2561,
+      "step": 137
+    },
+    {
+      "epoch": 0.41317365269461076,
+      "grad_norm": 2.211471950792459,
+      "learning_rate": 8.983407709996415e-06,
+      "loss": 0.2041,
+      "step": 138
+    },
+    {
+      "epoch": 0.4161676646706587,
+      "grad_norm": 1.984614908355809,
+      "learning_rate": 8.969151311524215e-06,
+      "loss": 0.1807,
+      "step": 139
+    },
+    {
+      "epoch": 0.41916167664670656,
+      "grad_norm": 2.2029762445942525,
+      "learning_rate": 8.954807123329703e-06,
+      "loss": 0.2267,
+      "step": 140
+    },
+    {
+      "epoch": 0.4221556886227545,
+      "grad_norm": 1.92215912110175,
+      "learning_rate": 8.940375462677758e-06,
+      "loss": 0.2134,
+      "step": 141
+    },
+    {
+      "epoch": 0.4251497005988024,
+      "grad_norm": 2.0197811390799756,
+      "learning_rate": 8.92585664876797e-06,
+      "loss": 0.2332,
+      "step": 142
+    },
+    {
+      "epoch": 0.4281437125748503,
+      "grad_norm": 1.9835840809937593,
+      "learning_rate": 8.911251002727588e-06,
+      "loss": 0.1906,
+      "step": 143
+    },
+    {
+      "epoch": 0.4311377245508982,
+      "grad_norm": 1.8395381200838579,
+      "learning_rate": 8.896558847604414e-06,
+      "loss": 0.1979,
+      "step": 144
+    },
+    {
+      "epoch": 0.4341317365269461,
+      "grad_norm": 2.036758389671924,
+      "learning_rate": 8.881780508359661e-06,
+      "loss": 0.2285,
+      "step": 145
+    },
+    {
+      "epoch": 0.437125748502994,
+      "grad_norm": 2.056400623646841,
+      "learning_rate": 8.86691631186076e-06,
+      "loss": 0.2158,
+      "step": 146
+    },
+    {
+      "epoch": 0.44011976047904194,
+      "grad_norm": 1.7694623059290855,
+      "learning_rate": 8.851966586874138e-06,
+      "loss": 0.1794,
+      "step": 147
+    },
+    {
+      "epoch": 0.4431137724550898,
+      "grad_norm": 2.1977807470605004,
+      "learning_rate": 8.836931664057935e-06,
+      "loss": 0.2348,
+      "step": 148
+    },
+    {
+      "epoch": 0.44610778443113774,
+      "grad_norm": 2.063170256876057,
+      "learning_rate": 8.821811875954705e-06,
+      "loss": 0.1948,
+      "step": 149
+    },
+    {
+      "epoch": 0.4491017964071856,
+      "grad_norm": 2.0789895629985096,
+      "learning_rate": 8.806607556984045e-06,
+      "loss": 0.2339,
+      "step": 150
+    },
+    {
+      "epoch": 0.45209580838323354,
+      "grad_norm": 2.5559310668328585,
+      "learning_rate": 8.791319043435213e-06,
+      "loss": 0.2081,
+      "step": 151
+    },
+    {
+      "epoch": 0.4550898203592814,
+      "grad_norm": 2.221677168145311,
+      "learning_rate": 8.775946673459682e-06,
+      "loss": 0.2186,
+      "step": 152
+    },
+    {
+      "epoch": 0.45808383233532934,
+      "grad_norm": 1.879829857972592,
+      "learning_rate": 8.76049078706366e-06,
+      "loss": 0.185,
+      "step": 153
+    },
+    {
+      "epoch": 0.46107784431137727,
+      "grad_norm": 1.9676399694599847,
+      "learning_rate": 8.744951726100572e-06,
+      "loss": 0.1975,
+      "step": 154
+    },
+    {
+      "epoch": 0.46407185628742514,
+      "grad_norm": 1.979489474056044,
+      "learning_rate": 8.729329834263503e-06,
+      "loss": 0.1842,
+      "step": 155
+    },
+    {
+      "epoch": 0.46706586826347307,
+      "grad_norm": 2.3272382462041583,
+      "learning_rate": 8.713625457077585e-06,
+      "loss": 0.2452,
+      "step": 156
+    },
+    {
+      "epoch": 0.47005988023952094,
+      "grad_norm": 2.171193662603986,
+      "learning_rate": 8.697838941892371e-06,
+      "loss": 0.2316,
+      "step": 157
+    },
+    {
+      "epoch": 0.47305389221556887,
+      "grad_norm": 1.9438264169997759,
+      "learning_rate": 8.681970637874131e-06,
+      "loss": 0.2143,
+      "step": 158
+    },
+    {
+      "epoch": 0.47604790419161674,
+      "grad_norm": 2.127297885012091,
+      "learning_rate": 8.666020895998154e-06,
+      "loss": 0.2279,
+      "step": 159
+    },
+    {
+      "epoch": 0.47904191616766467,
+      "grad_norm": 1.9998858365843528,
+      "learning_rate": 8.64999006904096e-06,
+      "loss": 0.1903,
+      "step": 160
+    },
+    {
+      "epoch": 0.4820359281437126,
+      "grad_norm": 2.175444508451546,
+      "learning_rate": 8.63387851157252e-06,
+      "loss": 0.208,
+      "step": 161
+    },
+    {
+      "epoch": 0.48502994011976047,
+      "grad_norm": 2.133342028323775,
+      "learning_rate": 8.617686579948396e-06,
+      "loss": 0.2031,
+      "step": 162
+    },
+    {
+      "epoch": 0.4880239520958084,
+      "grad_norm": 2.1755162419708447,
+      "learning_rate": 8.60141463230187e-06,
+      "loss": 0.2097,
+      "step": 163
+    },
+    {
+      "epoch": 0.49101796407185627,
+      "grad_norm": 2.3455284072748936,
+      "learning_rate": 8.585063028536015e-06,
+      "loss": 0.2204,
+      "step": 164
+    },
+    {
+      "epoch": 0.4940119760479042,
+      "grad_norm": 2.3963397391538575,
+      "learning_rate": 8.568632130315747e-06,
+      "loss": 0.2293,
+      "step": 165
+    },
+    {
+      "epoch": 0.49700598802395207,
+      "grad_norm": 1.8910044508297472,
+      "learning_rate": 8.552122301059807e-06,
+      "loss": 0.1883,
+      "step": 166
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0621076187287057,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 0.1873,
+      "step": 167
+    },
+    {
+      "epoch": 0.5029940119760479,
+      "grad_norm": 2.1666475498344786,
+      "learning_rate": 8.518867311836808e-06,
+      "loss": 0.2282,
+      "step": 168
+    },
+    {
+      "epoch": 0.5059880239520959,
+      "grad_norm": 1.9379278378686429,
+      "learning_rate": 8.502122887403882e-06,
+      "loss": 0.1749,
+      "step": 169
+    },
+    {
+      "epoch": 0.5089820359281437,
+      "grad_norm": 2.0651511811905543,
+      "learning_rate": 8.485301002987285e-06,
+      "loss": 0.1793,
+      "step": 170
+    },
+    {
+      "epoch": 0.5119760479041916,
+      "grad_norm": 2.2532217549885103,
+      "learning_rate": 8.468402030653598e-06,
+      "loss": 0.2314,
+      "step": 171
+    },
+    {
+      "epoch": 0.5149700598802395,
+      "grad_norm": 1.9556320841919537,
+      "learning_rate": 8.451426344174433e-06,
+      "loss": 0.2067,
+      "step": 172
+    },
+    {
+      "epoch": 0.5179640718562875,
+      "grad_norm": 2.0777928140678577,
+      "learning_rate": 8.434374319018165e-06,
+      "loss": 0.2085,
+      "step": 173
+    },
+    {
+      "epoch": 0.5209580838323353,
+      "grad_norm": 2.2119655050446125,
+      "learning_rate": 8.417246332341638e-06,
+      "loss": 0.1987,
+      "step": 174
+    },
+    {
+      "epoch": 0.5239520958083832,
+      "grad_norm": 2.0810011626186573,
+      "learning_rate": 8.4000427629818e-06,
+      "loss": 0.21,
+      "step": 175
+    },
+    {
+      "epoch": 0.5269461077844312,
+      "grad_norm": 1.7890061739740544,
+      "learning_rate": 8.382763991447344e-06,
+      "loss": 0.1906,
+      "step": 176
+    },
+    {
+      "epoch": 0.5299401197604791,
+      "grad_norm": 2.103678829134735,
+      "learning_rate": 8.365410399910287e-06,
+      "loss": 0.2118,
+      "step": 177
+    },
+    {
+      "epoch": 0.5329341317365269,
+      "grad_norm": 2.0999921791623763,
+      "learning_rate": 8.347982372197515e-06,
+      "loss": 0.1965,
+      "step": 178
+    },
+    {
+      "epoch": 0.5359281437125748,
+      "grad_norm": 2.227470699291841,
+      "learning_rate": 8.33048029378229e-06,
+      "loss": 0.2102,
+      "step": 179
+    },
+    {
+      "epoch": 0.5389221556886228,
+      "grad_norm": 2.146446769686846,
+      "learning_rate": 8.312904551775731e-06,
+      "loss": 0.2381,
+      "step": 180
+    },
+    {
+      "epoch": 0.5419161676646707,
+      "grad_norm": 2.1293939434561096,
+      "learning_rate": 8.295255534918249e-06,
+      "loss": 0.252,
+      "step": 181
+    },
+    {
+      "epoch": 0.5449101796407185,
+      "grad_norm": 2.2571143080362135,
+      "learning_rate": 8.277533633570948e-06,
+      "loss": 0.2122,
+      "step": 182
+    },
+    {
+      "epoch": 0.5479041916167665,
+      "grad_norm": 2.328443169499519,
+      "learning_rate": 8.25973923970699e-06,
+      "loss": 0.2272,
+      "step": 183
+    },
+    {
+      "epoch": 0.5508982035928144,
+      "grad_norm": 1.956494942475171,
+      "learning_rate": 8.241872746902934e-06,
+      "loss": 0.209,
+      "step": 184
+    },
+    {
+      "epoch": 0.5538922155688623,
+      "grad_norm": 1.9122027895224285,
+      "learning_rate": 8.223934550330015e-06,
+      "loss": 0.2103,
+      "step": 185
+    },
+    {
+      "epoch": 0.5568862275449101,
+      "grad_norm": 2.0181863447827375,
+      "learning_rate": 8.20592504674542e-06,
+      "loss": 0.2136,
+      "step": 186
+    },
+    {
+      "epoch": 0.5598802395209581,
+      "grad_norm": 2.464145315075048,
+      "learning_rate": 8.187844634483495e-06,
+      "loss": 0.2451,
+      "step": 187
+    },
+    {
+      "epoch": 0.562874251497006,
+      "grad_norm": 2.1568906432402537,
+      "learning_rate": 8.16969371344696e-06,
+      "loss": 0.2163,
+      "step": 188
+    },
+    {
+      "epoch": 0.5658682634730539,
+      "grad_norm": 1.9590870778135598,
+      "learning_rate": 8.151472685098037e-06,
+      "loss": 0.2056,
+      "step": 189
+    },
+    {
+      "epoch": 0.5688622754491018,
+      "grad_norm": 1.924584617553309,
+      "learning_rate": 8.13318195244958e-06,
+      "loss": 0.2107,
+      "step": 190
+    },
+    {
+      "epoch": 0.5718562874251497,
+      "grad_norm": 2.0716958517354014,
+      "learning_rate": 8.114821920056177e-06,
+      "loss": 0.2112,
+      "step": 191
+    },
+    {
+      "epoch": 0.5748502994011976,
+      "grad_norm": 2.30933527486489,
+      "learning_rate": 8.096392994005177e-06,
+      "loss": 0.2426,
+      "step": 192
+    },
+    {
+      "epoch": 0.5778443113772455,
+      "grad_norm": 1.765612742327491,
+      "learning_rate": 8.077895581907719e-06,
+      "loss": 0.1987,
+      "step": 193
+    },
+    {
+      "epoch": 0.5808383233532934,
+      "grad_norm": 1.8946212573236976,
+      "learning_rate": 8.059330092889724e-06,
+      "loss": 0.2072,
+      "step": 194
+    },
+    {
+      "epoch": 0.5838323353293413,
+      "grad_norm": 2.244560260670403,
+      "learning_rate": 8.040696937582833e-06,
+      "loss": 0.2367,
+      "step": 195
+    },
+    {
+      "epoch": 0.5868263473053892,
+      "grad_norm": 1.8231889991559906,
+      "learning_rate": 8.021996528115335e-06,
+      "loss": 0.1878,
+      "step": 196
+    },
+    {
+      "epoch": 0.5898203592814372,
+      "grad_norm": 1.9591547345350766,
+      "learning_rate": 8.003229278103044e-06,
+      "loss": 0.1762,
+      "step": 197
+    },
+    {
+      "epoch": 0.592814371257485,
+      "grad_norm": 2.033994466208927,
+      "learning_rate": 7.984395602640153e-06,
+      "loss": 0.2184,
+      "step": 198
+    },
+    {
+      "epoch": 0.5958083832335329,
+      "grad_norm": 1.7971757316367645,
+      "learning_rate": 7.96549591829006e-06,
+      "loss": 0.192,
+      "step": 199
+    },
+    {
+      "epoch": 0.5988023952095808,
+      "grad_norm": 2.0645123392078095,
+      "learning_rate": 7.946530643076138e-06,
+      "loss": 0.2115,
+      "step": 200
+    },
+    {
+      "epoch": 0.5988023952095808,
+      "eval_loss": 0.22190050780773163,
+      "eval_runtime": 8.2334,
+      "eval_samples_per_second": 6.559,
+      "eval_steps_per_second": 1.7,
+      "step": 200
+    },
+    {
+      "epoch": 0.6017964071856288,
+      "grad_norm": 1.7026761204702943,
+      "learning_rate": 7.927500196472506e-06,
+      "loss": 0.1909,
+      "step": 201
+    },
+    {
+      "epoch": 0.6047904191616766,
+      "grad_norm": 2.2246643763566403,
+      "learning_rate": 7.908404999394747e-06,
+      "loss": 0.1991,
+      "step": 202
+    },
+    {
+      "epoch": 0.6077844311377245,
+      "grad_norm": 2.153865556205914,
+      "learning_rate": 7.889245474190588e-06,
+      "loss": 0.2275,
+      "step": 203
+    },
+    {
+      "epoch": 0.6107784431137725,
+      "grad_norm": 2.664761056305188,
+      "learning_rate": 7.870022044630569e-06,
+      "loss": 0.2479,
+      "step": 204
+    },
+    {
+      "epoch": 0.6137724550898204,
+      "grad_norm": 1.836557392956106,
+      "learning_rate": 7.85073513589867e-06,
+      "loss": 0.2078,
+      "step": 205
+    },
+    {
+      "epoch": 0.6167664670658682,
+      "grad_norm": 1.9520492409217767,
+      "learning_rate": 7.831385174582901e-06,
+      "loss": 0.2033,
+      "step": 206
+    },
+    {
+      "epoch": 0.6197604790419161,
+      "grad_norm": 1.9241755033844297,
+      "learning_rate": 7.81197258866587e-06,
+      "loss": 0.1815,
+      "step": 207
+    },
+    {
+      "epoch": 0.6227544910179641,
+      "grad_norm": 2.015073899879637,
+      "learning_rate": 7.792497807515317e-06,
+      "loss": 0.1959,
+      "step": 208
+    },
+    {
+      "epoch": 0.625748502994012,
+      "grad_norm": 1.8502927170801848,
+      "learning_rate": 7.772961261874615e-06,
+      "loss": 0.1968,
+      "step": 209
+    },
+    {
+      "epoch": 0.6287425149700598,
+      "grad_norm": 2.3118898678514754,
+      "learning_rate": 7.75336338385325e-06,
+      "loss": 0.2099,
+      "step": 210
+    },
+    {
+      "epoch": 0.6317365269461078,
+      "grad_norm": 2.048337580951677,
+      "learning_rate": 7.733704606917248e-06,
+      "loss": 0.2359,
+      "step": 211
+    },
+    {
+      "epoch": 0.6347305389221557,
+      "grad_norm": 1.8843145559725893,
+      "learning_rate": 7.713985365879607e-06,
+      "loss": 0.1939,
+      "step": 212
+    },
+    {
+      "epoch": 0.6377245508982036,
+      "grad_norm": 2.1117913593510247,
+      "learning_rate": 7.694206096890667e-06,
+      "loss": 0.2638,
+      "step": 213
+    },
+    {
+      "epoch": 0.6407185628742516,
+      "grad_norm": 1.9019309926013162,
+      "learning_rate": 7.674367237428467e-06,
+      "loss": 0.1899,
+      "step": 214
+    },
+    {
+      "epoch": 0.6437125748502994,
+      "grad_norm": 1.8477338518482065,
+      "learning_rate": 7.654469226289068e-06,
+      "loss": 0.1945,
+      "step": 215
+    },
+    {
+      "epoch": 0.6467065868263473,
+      "grad_norm": 2.0646421699217234,
+      "learning_rate": 7.63451250357685e-06,
+      "loss": 0.2139,
+      "step": 216
+    },
+    {
+      "epoch": 0.6497005988023952,
+      "grad_norm": 1.7506425168653987,
+      "learning_rate": 7.614497510694774e-06,
+      "loss": 0.1836,
+      "step": 217
+    },
+    {
+      "epoch": 0.6526946107784432,
+      "grad_norm": 1.9524525600584752,
+      "learning_rate": 7.5944246903346204e-06,
+      "loss": 0.1944,
+      "step": 218
+    },
+    {
+      "epoch": 0.655688622754491,
+      "grad_norm": 1.8658516690182516,
+      "learning_rate": 7.574294486467204e-06,
+      "loss": 0.1669,
+      "step": 219
+    },
+    {
+      "epoch": 0.6586826347305389,
+      "grad_norm": 1.856391869494492,
+      "learning_rate": 7.55410734433254e-06,
+      "loss": 0.1794,
+      "step": 220
+    },
+    {
+      "epoch": 0.6616766467065869,
+      "grad_norm": 1.9959325168714759,
+      "learning_rate": 7.533863710430011e-06,
+      "loss": 0.1999,
+      "step": 221
+    },
+    {
+      "epoch": 0.6646706586826348,
+      "grad_norm": 1.822065705071486,
+      "learning_rate": 7.513564032508484e-06,
+      "loss": 0.1606,
+      "step": 222
+    },
+    {
+      "epoch": 0.6676646706586826,
+      "grad_norm": 2.39046855796897,
+      "learning_rate": 7.493208759556406e-06,
+      "loss": 0.2347,
+      "step": 223
+    },
+    {
+      "epoch": 0.6706586826347305,
+      "grad_norm": 2.041270670541334,
+      "learning_rate": 7.472798341791877e-06,
+      "loss": 0.2126,
+      "step": 224
+    },
+    {
+      "epoch": 0.6736526946107785,
+      "grad_norm": 2.0090859329660473,
+      "learning_rate": 7.452333230652688e-06,
+      "loss": 0.1591,
+      "step": 225
+    },
+    {
+      "epoch": 0.6766467065868264,
+      "grad_norm": 2.0956402431423604,
+      "learning_rate": 7.431813878786343e-06,
+      "loss": 0.2227,
+      "step": 226
+    },
+    {
+      "epoch": 0.6796407185628742,
+      "grad_norm": 1.7407865476065647,
+      "learning_rate": 7.4112407400400395e-06,
+      "loss": 0.1598,
+      "step": 227
+    },
+    {
+      "epoch": 0.6826347305389222,
+      "grad_norm": 1.7650797378120815,
+      "learning_rate": 7.390614269450633e-06,
+      "loss": 0.1699,
+      "step": 228
+    },
+    {
+      "epoch": 0.6856287425149701,
+      "grad_norm": 1.8939791067028862,
+      "learning_rate": 7.369934923234577e-06,
+      "loss": 0.172,
+      "step": 229
+    },
+    {
+      "epoch": 0.688622754491018,
+      "grad_norm": 2.156249580721313,
+      "learning_rate": 7.349203158777826e-06,
+      "loss": 0.1976,
+      "step": 230
+    },
+    {
+      "epoch": 0.6916167664670658,
+      "grad_norm": 1.760885691458116,
+      "learning_rate": 7.32841943462572e-06,
+      "loss": 0.1877,
+      "step": 231
+    },
+    {
+      "epoch": 0.6946107784431138,
+      "grad_norm": 2.107304013543223,
+      "learning_rate": 7.3075842104728445e-06,
+      "loss": 0.2124,
+      "step": 232
+    },
+    {
+      "epoch": 0.6976047904191617,
+      "grad_norm": 2.2791362744540278,
+      "learning_rate": 7.286697947152868e-06,
+      "loss": 0.1979,
+      "step": 233
+    },
+    {
+      "epoch": 0.7005988023952096,
+      "grad_norm": 1.8786046579469293,
+      "learning_rate": 7.265761106628338e-06,
+      "loss": 0.1784,
+      "step": 234
+    },
+    {
+      "epoch": 0.7035928143712575,
+      "grad_norm": 1.9625568781606961,
+      "learning_rate": 7.244774151980466e-06,
+      "loss": 0.2029,
+      "step": 235
+    },
+    {
+      "epoch": 0.7065868263473054,
+      "grad_norm": 2.030460885688198,
+      "learning_rate": 7.223737547398898e-06,
+      "loss": 0.1967,
+      "step": 236
+    },
+    {
+      "epoch": 0.7095808383233533,
+      "grad_norm": 1.8107535514428748,
+      "learning_rate": 7.20265175817143e-06,
+      "loss": 0.1928,
+      "step": 237
+    },
+    {
+      "epoch": 0.7125748502994012,
+      "grad_norm": 2.388886552974491,
+      "learning_rate": 7.181517250673729e-06,
+      "loss": 0.2577,
+      "step": 238
+    },
+    {
+      "epoch": 0.7155688622754491,
+      "grad_norm": 2.0722344649941267,
+      "learning_rate": 7.1603344923590065e-06,
+      "loss": 0.2043,
+      "step": 239
+    },
+    {
+      "epoch": 0.718562874251497,
+      "grad_norm": 2.229117942854137,
+      "learning_rate": 7.139103951747694e-06,
+      "loss": 0.2462,
+      "step": 240
+    },
+    {
+      "epoch": 0.7215568862275449,
+      "grad_norm": 2.075263551266433,
+      "learning_rate": 7.1178260984170675e-06,
+      "loss": 0.1934,
+      "step": 241
+    },
+    {
+      "epoch": 0.7245508982035929,
+      "grad_norm": 2.056989786044168,
+      "learning_rate": 7.0965014029908654e-06,
+      "loss": 0.1881,
+      "step": 242
+    },
+    {
+      "epoch": 0.7275449101796407,
+      "grad_norm": 2.012708636981654,
+      "learning_rate": 7.075130337128883e-06,
+      "loss": 0.2344,
+      "step": 243
+    },
+    {
+      "epoch": 0.7305389221556886,
+      "grad_norm": 2.064420611976832,
+      "learning_rate": 7.053713373516538e-06,
+      "loss": 0.2426,
+      "step": 244
+    },
+    {
+      "epoch": 0.7335329341317365,
+      "grad_norm": 1.961110213836023,
+      "learning_rate": 7.03225098585441e-06,
+      "loss": 0.223,
+      "step": 245
+    },
+    {
+      "epoch": 0.7365269461077845,
+      "grad_norm": 1.988029283700158,
+      "learning_rate": 7.0107436488477694e-06,
+      "loss": 0.2402,
+      "step": 246
+    },
+    {
+      "epoch": 0.7395209580838323,
+      "grad_norm": 2.0250344729622185,
+      "learning_rate": 6.989191838196083e-06,
+      "loss": 0.2129,
+      "step": 247
+    },
+    {
+      "epoch": 0.7425149700598802,
+      "grad_norm": 1.6647311426773186,
+      "learning_rate": 6.9675960305824785e-06,
+      "loss": 0.1811,
+      "step": 248
+    },
+    {
+      "epoch": 0.7455089820359282,
+      "grad_norm": 2.0222991297486694,
+      "learning_rate": 6.945956703663212e-06,
+      "loss": 0.202,
+      "step": 249
+    },
+    {
+      "epoch": 0.7485029940119761,
+      "grad_norm": 2.007631237455473,
+      "learning_rate": 6.9242743360570985e-06,
+      "loss": 0.2134,
+      "step": 250
+    },
+    {
+      "epoch": 0.7514970059880239,
+      "grad_norm": 1.832435389431675,
+      "learning_rate": 6.9025494073349284e-06,
+      "loss": 0.2019,
+      "step": 251
+    },
+    {
+      "epoch": 0.7544910179640718,
+      "grad_norm": 1.7551292568679946,
+      "learning_rate": 6.880782398008862e-06,
+      "loss": 0.206,
+      "step": 252
+    },
+    {
+      "epoch": 0.7574850299401198,
+      "grad_norm": 1.8627773936019942,
+      "learning_rate": 6.858973789521792e-06,
+      "loss": 0.185,
+      "step": 253
+    },
+    {
+      "epoch": 0.7604790419161677,
+      "grad_norm": 1.8972178218802618,
+      "learning_rate": 6.837124064236709e-06,
+      "loss": 0.1871,
+      "step": 254
+    },
+    {
+      "epoch": 0.7634730538922155,
+      "grad_norm": 1.6529718300812177,
+      "learning_rate": 6.815233705426019e-06,
+      "loss": 0.1711,
+      "step": 255
+    },
+    {
+      "epoch": 0.7664670658682635,
+      "grad_norm": 1.8473322887342682,
+      "learning_rate": 6.7933031972608644e-06,
+      "loss": 0.1789,
+      "step": 256
+    },
+    {
+      "epoch": 0.7694610778443114,
+      "grad_norm": 2.0382922713283977,
+      "learning_rate": 6.771333024800411e-06,
+      "loss": 0.2,
+      "step": 257
+    },
+    {
+      "epoch": 0.7724550898203593,
+      "grad_norm": 2.1221191580118126,
+      "learning_rate": 6.74932367398112e-06,
+      "loss": 0.2009,
+      "step": 258
+    },
+    {
+      "epoch": 0.7754491017964071,
+      "grad_norm": 1.758064508917816,
+      "learning_rate": 6.727275631605996e-06,
+      "loss": 0.1668,
+      "step": 259
+    },
+    {
+      "epoch": 0.7784431137724551,
+      "grad_norm": 2.3952537397381954,
+      "learning_rate": 6.70518938533383e-06,
+      "loss": 0.2034,
+      "step": 260
+    },
+    {
+      "epoch": 0.781437125748503,
+      "grad_norm": 2.9075455590007495,
+      "learning_rate": 6.683065423668403e-06,
+      "loss": 0.1716,
+      "step": 261
+    },
+    {
+      "epoch": 0.7844311377245509,
+      "grad_norm": 2.0043118931686434,
+      "learning_rate": 6.660904235947687e-06,
+      "loss": 0.1942,
+      "step": 262
+    },
+    {
+      "epoch": 0.7874251497005988,
+      "grad_norm": 1.9587332504376953,
+      "learning_rate": 6.638706312333018e-06,
+      "loss": 0.1964,
+      "step": 263
+    },
+    {
+      "epoch": 0.7904191616766467,
+      "grad_norm": 2.074887023022505,
+      "learning_rate": 6.61647214379826e-06,
+      "loss": 0.1999,
+      "step": 264
+    },
+    {
+      "epoch": 0.7934131736526946,
+      "grad_norm": 2.4500273717965895,
+      "learning_rate": 6.594202222118941e-06,
+      "loss": 0.2172,
+      "step": 265
+    },
+    {
+      "epoch": 0.7964071856287425,
+      "grad_norm": 1.9090529047590825,
+      "learning_rate": 6.571897039861377e-06,
+      "loss": 0.1785,
+      "step": 266
+    },
+    {
+      "epoch": 0.7994011976047904,
+      "grad_norm": 2.069812540101633,
+      "learning_rate": 6.549557090371775e-06,
+      "loss": 0.2083,
+      "step": 267
+    },
+    {
+      "epoch": 0.8023952095808383,
+      "grad_norm": 1.7599883485970342,
+      "learning_rate": 6.527182867765333e-06,
+      "loss": 0.1629,
+      "step": 268
+    },
+    {
+      "epoch": 0.8053892215568862,
+      "grad_norm": 2.1678457750535602,
+      "learning_rate": 6.504774866915291e-06,
+      "loss": 0.2213,
+      "step": 269
+    },
+    {
+      "epoch": 0.8083832335329342,
+      "grad_norm": 1.7785853251497423,
+      "learning_rate": 6.482333583442002e-06,
+      "loss": 0.1664,
+      "step": 270
+    },
+    {
+      "epoch": 0.811377245508982,
+      "grad_norm": 1.970971514686613,
+      "learning_rate": 6.459859513701967e-06,
+      "loss": 0.2077,
+      "step": 271
+    },
+    {
+      "epoch": 0.8143712574850299,
+      "grad_norm": 2.234340016375132,
+      "learning_rate": 6.437353154776848e-06,
+      "loss": 0.2048,
+      "step": 272
+    },
+    {
+      "epoch": 0.8173652694610778,
+      "grad_norm": 1.9168279054779973,
+      "learning_rate": 6.414815004462483e-06,
+      "loss": 0.1701,
+      "step": 273
+    },
+    {
+      "epoch": 0.8203592814371258,
+      "grad_norm": 2.003283050177111,
+      "learning_rate": 6.3922455612578715e-06,
+      "loss": 0.219,
+      "step": 274
+    },
+    {
+      "epoch": 0.8233532934131736,
+      "grad_norm": 1.8224187009093122,
+      "learning_rate": 6.369645324354149e-06,
+      "loss": 0.1724,
+      "step": 275
+    },
+    {
+      "epoch": 0.8263473053892215,
+      "grad_norm": 2.4795272213190724,
+      "learning_rate": 6.3470147936235485e-06,
+      "loss": 0.2252,
+      "step": 276
+    },
+    {
+      "epoch": 0.8293413173652695,
+      "grad_norm": 1.6118512666724374,
+      "learning_rate": 6.3243544696083355e-06,
+      "loss": 0.1751,
+      "step": 277
+    },
+    {
+      "epoch": 0.8323353293413174,
+      "grad_norm": 2.251096717061321,
+      "learning_rate": 6.301664853509755e-06,
+      "loss": 0.207,
+      "step": 278
+    },
+    {
+      "epoch": 0.8353293413173652,
+      "grad_norm": 1.8765563120289703,
+      "learning_rate": 6.278946447176924e-06,
+      "loss": 0.1627,
+      "step": 279
+    },
+    {
+      "epoch": 0.8383233532934131,
+      "grad_norm": 1.9020379733169448,
+      "learning_rate": 6.256199753095745e-06,
+      "loss": 0.2002,
+      "step": 280
+    },
+    {
+      "epoch": 0.8413173652694611,
+      "grad_norm": 1.7332807462579276,
+      "learning_rate": 6.233425274377793e-06,
+      "loss": 0.1731,
+      "step": 281
+    },
+    {
+      "epoch": 0.844311377245509,
+      "grad_norm": 1.867279006893392,
+      "learning_rate": 6.21062351474918e-06,
+      "loss": 0.178,
+      "step": 282
+    },
+    {
+      "epoch": 0.8473053892215568,
+      "grad_norm": 1.8578749380870647,
+      "learning_rate": 6.18779497853942e-06,
+      "loss": 0.1894,
+      "step": 283
+    },
+    {
+      "epoch": 0.8502994011976048,
+      "grad_norm": 1.8796926463237162,
+      "learning_rate": 6.164940170670266e-06,
+      "loss": 0.198,
+      "step": 284
+    },
+    {
+      "epoch": 0.8532934131736527,
+      "grad_norm": 2.0206642686222174,
+      "learning_rate": 6.142059596644557e-06,
+      "loss": 0.1851,
+      "step": 285
+    },
+    {
+      "epoch": 0.8562874251497006,
+      "grad_norm": 1.7279118559465598,
+      "learning_rate": 6.11915376253502e-06,
+      "loss": 0.1599,
+      "step": 286
+    },
+    {
+      "epoch": 0.8592814371257484,
+      "grad_norm": 1.9879261719146308,
+      "learning_rate": 6.096223174973091e-06,
+      "loss": 0.1701,
+      "step": 287
+    },
+    {
+      "epoch": 0.8622754491017964,
+      "grad_norm": 1.8513200108072538,
+      "learning_rate": 6.073268341137694e-06,
+      "loss": 0.1638,
+      "step": 288
+    },
+    {
+      "epoch": 0.8652694610778443,
+      "grad_norm": 1.9197483492036942,
+      "learning_rate": 6.050289768744042e-06,
+      "loss": 0.1778,
+      "step": 289
+    },
+    {
+      "epoch": 0.8682634730538922,
+      "grad_norm": 1.9074371287702236,
+      "learning_rate": 6.0272879660323936e-06,
+      "loss": 0.1962,
+      "step": 290
+    },
+    {
+      "epoch": 0.8712574850299402,
+      "grad_norm": 1.9734342233017201,
+      "learning_rate": 6.004263441756815e-06,
+      "loss": 0.1907,
+      "step": 291
+    },
+    {
+      "epoch": 0.874251497005988,
+      "grad_norm": 2.187499188364905,
+      "learning_rate": 5.98121670517393e-06,
+      "loss": 0.2084,
+      "step": 292
+    },
+    {
+      "epoch": 0.8772455089820359,
+      "grad_norm": 1.7886245273797357,
+      "learning_rate": 5.958148266031654e-06,
+      "loss": 0.1968,
+      "step": 293
+    },
+    {
+      "epoch": 0.8802395209580839,
+      "grad_norm": 1.6714498675009575,
+      "learning_rate": 5.935058634557917e-06,
+      "loss": 0.1707,
+      "step": 294
+    },
+    {
+      "epoch": 0.8832335329341318,
+      "grad_norm": 1.9202259942564088,
+      "learning_rate": 5.911948321449384e-06,
+      "loss": 0.1852,
+      "step": 295
+    },
+    {
+      "epoch": 0.8862275449101796,
+      "grad_norm": 2.0449566828000103,
+      "learning_rate": 5.8888178378601565e-06,
+      "loss": 0.2032,
+      "step": 296
+    },
+    {
+      "epoch": 0.8892215568862275,
+      "grad_norm": 1.8358349527752167,
+      "learning_rate": 5.865667695390468e-06,
+      "loss": 0.1849,
+      "step": 297
+    },
+    {
+      "epoch": 0.8922155688622755,
+      "grad_norm": 1.5755225849697878,
+      "learning_rate": 5.842498406075363e-06,
+      "loss": 0.1559,
+      "step": 298
+    },
+    {
+      "epoch": 0.8952095808383234,
+      "grad_norm": 1.6807446640909467,
+      "learning_rate": 5.819310482373381e-06,
+      "loss": 0.1411,
+      "step": 299
+    },
+    {
+      "epoch": 0.8982035928143712,
+      "grad_norm": 1.8158144553482063,
+      "learning_rate": 5.796104437155213e-06,
+      "loss": 0.1943,
+      "step": 300
+    },
+    {
+      "epoch": 0.9011976047904192,
+      "grad_norm": 1.7866471392783534,
+      "learning_rate": 5.772880783692363e-06,
+      "loss": 0.1561,
+      "step": 301
+    },
+    {
+      "epoch": 0.9041916167664671,
+      "grad_norm": 2.046505298297432,
+      "learning_rate": 5.749640035645798e-06,
+      "loss": 0.1876,
+      "step": 302
+    },
+    {
+      "epoch": 0.907185628742515,
+      "grad_norm": 2.114758374154537,
+      "learning_rate": 5.726382707054578e-06,
+      "loss": 0.1586,
+      "step": 303
+    },
+    {
+      "epoch": 0.9101796407185628,
+      "grad_norm": 1.947411771932179,
+      "learning_rate": 5.703109312324493e-06,
+      "loss": 0.1812,
+      "step": 304
+    },
+    {
+      "epoch": 0.9131736526946108,
+      "grad_norm": 1.9216832034823756,
+      "learning_rate": 5.679820366216684e-06,
+      "loss": 0.2075,
+      "step": 305
+    },
+    {
+      "epoch": 0.9161676646706587,
+      "grad_norm": 1.5335056880145521,
+      "learning_rate": 5.656516383836263e-06,
+      "loss": 0.1443,
+      "step": 306
+    },
+    {
+      "epoch": 0.9191616766467066,
+      "grad_norm": 1.9859073845188653,
+      "learning_rate": 5.6331978806209044e-06,
+      "loss": 0.2084,
+      "step": 307
+    },
+    {
+      "epoch": 0.9221556886227545,
+      "grad_norm": 1.9976572010587028,
+      "learning_rate": 5.609865372329461e-06,
+      "loss": 0.1817,
+      "step": 308
+    },
+    {
+      "epoch": 0.9251497005988024,
+      "grad_norm": 2.541418228379657,
+      "learning_rate": 5.586519375030549e-06,
+      "loss": 0.213,
+      "step": 309
+    },
+    {
+      "epoch": 0.9281437125748503,
+      "grad_norm": 2.0632938129398455,
+      "learning_rate": 5.5631604050911354e-06,
+      "loss": 0.1999,
+      "step": 310
+    },
+    {
+      "epoch": 0.9311377245508982,
+      "grad_norm": 1.8707542903435745,
+      "learning_rate": 5.539788979165115e-06,
+      "loss": 0.1779,
+      "step": 311
+    },
+    {
+      "epoch": 0.9341317365269461,
+      "grad_norm": 1.6657841258195458,
+      "learning_rate": 5.516405614181883e-06,
+      "loss": 0.1822,
+      "step": 312
+    },
+    {
+      "epoch": 0.937125748502994,
+      "grad_norm": 1.7216645603962142,
+      "learning_rate": 5.4930108273349034e-06,
+      "loss": 0.1644,
+      "step": 313
+    },
+    {
+      "epoch": 0.9401197604790419,
+      "grad_norm": 1.8692836757629316,
+      "learning_rate": 5.4696051360702725e-06,
+      "loss": 0.1765,
+      "step": 314
+    },
+    {
+      "epoch": 0.9431137724550899,
+      "grad_norm": 2.0347548127365824,
+      "learning_rate": 5.446189058075265e-06,
+      "loss": 0.2043,
+      "step": 315
+    },
+    {
+      "epoch": 0.9461077844311377,
+      "grad_norm": 2.192389551817286,
+      "learning_rate": 5.4227631112668955e-06,
+      "loss": 0.2199,
+      "step": 316
+    },
+    {
+      "epoch": 0.9491017964071856,
+      "grad_norm": 1.9152238501777001,
+      "learning_rate": 5.39932781378045e-06,
+      "loss": 0.1615,
+      "step": 317
+    },
+    {
+      "epoch": 0.9520958083832335,
+      "grad_norm": 1.9917172382115225,
+      "learning_rate": 5.375883683958041e-06,
+      "loss": 0.2012,
+      "step": 318
+    },
+    {
+      "epoch": 0.9550898203592815,
+      "grad_norm": 1.9002788140909717,
+      "learning_rate": 5.3524312403371255e-06,
+      "loss": 0.1683,
+      "step": 319
+    },
+    {
+      "epoch": 0.9580838323353293,
+      "grad_norm": 1.994006525280562,
+      "learning_rate": 5.328971001639054e-06,
+      "loss": 0.1868,
+      "step": 320
+    },
+    {
+      "epoch": 0.9610778443113772,
+      "grad_norm": 2.0305212698887987,
+      "learning_rate": 5.3055034867575825e-06,
+      "loss": 0.2036,
+      "step": 321
+    },
+    {
+      "epoch": 0.9640718562874252,
+      "grad_norm": 2.0744337459207207,
+      "learning_rate": 5.282029214747404e-06,
+      "loss": 0.2092,
+      "step": 322
+    },
+    {
+      "epoch": 0.9670658682634731,
+      "grad_norm": 2.1908722280120627,
+      "learning_rate": 5.258548704812667e-06,
+      "loss": 0.186,
+      "step": 323
+    },
+    {
+      "epoch": 0.9700598802395209,
+      "grad_norm": 2.159699350040178,
+      "learning_rate": 5.235062476295488e-06,
+      "loss": 0.1946,
+      "step": 324
+    },
+    {
+      "epoch": 0.9730538922155688,
+      "grad_norm": 2.0869700847375414,
+      "learning_rate": 5.211571048664469e-06,
+      "loss": 0.1727,
+      "step": 325
+    },
+    {
+      "epoch": 0.9760479041916168,
+      "grad_norm": 1.818549729765939,
+      "learning_rate": 5.188074941503203e-06,
+      "loss": 0.1819,
+      "step": 326
+    },
+    {
+      "epoch": 0.9790419161676647,
+      "grad_norm": 1.9637923904453976,
+      "learning_rate": 5.164574674498788e-06,
+      "loss": 0.1819,
+      "step": 327
+    },
+    {
+      "epoch": 0.9820359281437125,
+      "grad_norm": 2.0808470949479148,
+      "learning_rate": 5.141070767430331e-06,
+      "loss": 0.1775,
+      "step": 328
+    },
+    {
+      "epoch": 0.9850299401197605,
+      "grad_norm": 2.2781209927002903,
+      "learning_rate": 5.117563740157444e-06,
+      "loss": 0.1799,
+      "step": 329
+    },
+    {
+      "epoch": 0.9880239520958084,
+      "grad_norm": 2.1100507826545214,
+      "learning_rate": 5.094054112608758e-06,
+      "loss": 0.1885,
+      "step": 330
+    },
+    {
+      "epoch": 0.9910179640718563,
+      "grad_norm": 1.9795383553504626,
+      "learning_rate": 5.070542404770413e-06,
+      "loss": 0.1901,
+      "step": 331
+    },
+    {
+      "epoch": 0.9940119760479041,
+      "grad_norm": 1.6734485479260668,
+      "learning_rate": 5.047029136674563e-06,
+      "loss": 0.15,
+      "step": 332
+    },
+    {
+      "epoch": 0.9970059880239521,
+      "grad_norm": 2.3038994241955244,
+      "learning_rate": 5.023514828387868e-06,
+      "loss": 0.2157,
+      "step": 333
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.644626702266248,
+      "learning_rate": 5e-06,
+      "loss": 0.121,
+      "step": 334
+    },
+    {
+      "epoch": 1.0029940119760479,
+      "grad_norm": 1.381616242516181,
+      "learning_rate": 4.976485171612134e-06,
+      "loss": 0.0794,
+      "step": 335
+    },
+    {
+      "epoch": 1.0059880239520957,
+      "grad_norm": 1.5026916182698242,
+      "learning_rate": 4.95297086332544e-06,
+      "loss": 0.0926,
+      "step": 336
+    },
+    {
+      "epoch": 1.0089820359281436,
+      "grad_norm": 1.4189987227807108,
+      "learning_rate": 4.9294575952295896e-06,
+      "loss": 0.0749,
+      "step": 337
+    },
+    {
+      "epoch": 1.0119760479041917,
+      "grad_norm": 1.3721951569021762,
+      "learning_rate": 4.905945887391242e-06,
+      "loss": 0.0808,
+      "step": 338
+    },
+    {
+      "epoch": 1.0149700598802396,
+      "grad_norm": 1.2797987792642362,
+      "learning_rate": 4.882436259842556e-06,
+      "loss": 0.056,
+      "step": 339
+    },
+    {
+      "epoch": 1.0179640718562875,
+      "grad_norm": 1.3350644750863863,
+      "learning_rate": 4.858929232569671e-06,
+      "loss": 0.0666,
+      "step": 340
+    },
+    {
+      "epoch": 1.0209580838323353,
+      "grad_norm": 1.8338903167573628,
+      "learning_rate": 4.835425325501214e-06,
+      "loss": 0.089,
+      "step": 341
+    },
+    {
+      "epoch": 1.0239520958083832,
+      "grad_norm": 1.8899191934070858,
+      "learning_rate": 4.811925058496799e-06,
+      "loss": 0.0797,
+      "step": 342
+    },
+    {
+      "epoch": 1.026946107784431,
+      "grad_norm": 1.7797674841587086,
+      "learning_rate": 4.788428951335534e-06,
+      "loss": 0.0793,
+      "step": 343
+    },
+    {
+      "epoch": 1.029940119760479,
+      "grad_norm": 2.0435109008087062,
+      "learning_rate": 4.7649375237045135e-06,
+      "loss": 0.0798,
+      "step": 344
+    },
+    {
+      "epoch": 1.032934131736527,
+      "grad_norm": 1.8300085342521888,
+      "learning_rate": 4.741451295187333e-06,
+      "loss": 0.0676,
+      "step": 345
+    },
+    {
+      "epoch": 1.035928143712575,
+      "grad_norm": 2.1756459221145734,
+      "learning_rate": 4.717970785252596e-06,
+      "loss": 0.0798,
+      "step": 346
+    },
+    {
+      "epoch": 1.0389221556886228,
+      "grad_norm": 2.257897071558614,
+      "learning_rate": 4.694496513242418e-06,
+      "loss": 0.0815,
+      "step": 347
+    },
+    {
+      "epoch": 1.0419161676646707,
+      "grad_norm": 1.7903397638684464,
+      "learning_rate": 4.671028998360947e-06,
+      "loss": 0.063,
+      "step": 348
+    },
+    {
+      "epoch": 1.0449101796407185,
+      "grad_norm": 1.618977959965445,
+      "learning_rate": 4.647568759662876e-06,
+      "loss": 0.0748,
+      "step": 349
+    },
+    {
+      "epoch": 1.0479041916167664,
+      "grad_norm": 1.9004553346724073,
+      "learning_rate": 4.624116316041962e-06,
+      "loss": 0.0841,
+      "step": 350
+    },
+    {
+      "epoch": 1.0508982035928143,
+      "grad_norm": 1.7504510185536164,
+      "learning_rate": 4.600672186219551e-06,
+      "loss": 0.0688,
+      "step": 351
+    },
+    {
+      "epoch": 1.0538922155688624,
+      "grad_norm": 2.2955147522975765,
+      "learning_rate": 4.5772368887331044e-06,
+      "loss": 0.0764,
+      "step": 352
+    },
+    {
+      "epoch": 1.0568862275449102,
+      "grad_norm": 1.8834254593619675,
+      "learning_rate": 4.553810941924735e-06,
+      "loss": 0.0879,
+      "step": 353
+    },
+    {
+      "epoch": 1.0598802395209581,
+      "grad_norm": 1.6549099259412252,
+      "learning_rate": 4.530394863929728e-06,
+      "loss": 0.0781,
+      "step": 354
+    },
+    {
+      "epoch": 1.062874251497006,
+      "grad_norm": 1.5687806397775854,
+      "learning_rate": 4.506989172665097e-06,
+      "loss": 0.0815,
+      "step": 355
+    },
+    {
+      "epoch": 1.0658682634730539,
+      "grad_norm": 1.4740133299510603,
+      "learning_rate": 4.483594385818119e-06,
+      "loss": 0.069,
+      "step": 356
+    },
+    {
+      "epoch": 1.0688622754491017,
+      "grad_norm": 1.9357269176012717,
+      "learning_rate": 4.460211020834887e-06,
+      "loss": 0.0956,
+      "step": 357
+    },
+    {
+      "epoch": 1.0718562874251496,
+      "grad_norm": 1.7382980066520093,
+      "learning_rate": 4.436839594908866e-06,
+      "loss": 0.0778,
+      "step": 358
+    },
+    {
+      "epoch": 1.0748502994011977,
+      "grad_norm": 1.4623277516867463,
+      "learning_rate": 4.4134806249694514e-06,
+      "loss": 0.0628,
+      "step": 359
+    },
+    {
+      "epoch": 1.0778443113772456,
+      "grad_norm": 1.997482468132072,
+      "learning_rate": 4.39013462767054e-06,
+      "loss": 0.0927,
+      "step": 360
+    },
+    {
+      "epoch": 1.0808383233532934,
+      "grad_norm": 1.5524284093544982,
+      "learning_rate": 4.366802119379098e-06,
+      "loss": 0.0811,
+      "step": 361
+    },
+    {
+      "epoch": 1.0838323353293413,
+      "grad_norm": 1.5941832920879642,
+      "learning_rate": 4.34348361616374e-06,
+      "loss": 0.076,
+      "step": 362
+    },
+    {
+      "epoch": 1.0868263473053892,
+      "grad_norm": 1.874708732385176,
+      "learning_rate": 4.3201796337833165e-06,
+      "loss": 0.0773,
+      "step": 363
+    },
+    {
+      "epoch": 1.089820359281437,
+      "grad_norm": 1.4799139616296857,
+      "learning_rate": 4.29689068767551e-06,
+      "loss": 0.0688,
+      "step": 364
+    },
+    {
+      "epoch": 1.092814371257485,
+      "grad_norm": 1.404758432897779,
+      "learning_rate": 4.273617292945425e-06,
+      "loss": 0.078,
+      "step": 365
+    },
+    {
+      "epoch": 1.095808383233533,
+      "grad_norm": 1.4715833008654857,
+      "learning_rate": 4.250359964354203e-06,
+      "loss": 0.0716,
+      "step": 366
+    },
+    {
+      "epoch": 1.098802395209581,
+      "grad_norm": 1.5617044021992046,
+      "learning_rate": 4.227119216307637e-06,
+      "loss": 0.0646,
+      "step": 367
+    },
+    {
+      "epoch": 1.1017964071856288,
+      "grad_norm": 1.4837419483466212,
+      "learning_rate": 4.203895562844789e-06,
+      "loss": 0.0687,
+      "step": 368
+    },
+    {
+      "epoch": 1.1047904191616766,
+      "grad_norm": 1.840312186349095,
+      "learning_rate": 4.18068951762662e-06,
+      "loss": 0.0692,
+      "step": 369
+    },
+    {
+      "epoch": 1.1077844311377245,
+      "grad_norm": 1.6949570633686344,
+      "learning_rate": 4.157501593924638e-06,
+      "loss": 0.0799,
+      "step": 370
+    },
+    {
+      "epoch": 1.1107784431137724,
+      "grad_norm": 1.9652904098006017,
+      "learning_rate": 4.134332304609533e-06,
+      "loss": 0.0958,
+      "step": 371
+    },
+    {
+      "epoch": 1.1137724550898203,
+      "grad_norm": 1.6606617819963352,
+      "learning_rate": 4.111182162139844e-06,
+      "loss": 0.0745,
+      "step": 372
+    },
+    {
+      "epoch": 1.1167664670658684,
+      "grad_norm": 1.520830367645309,
+      "learning_rate": 4.088051678550617e-06,
+      "loss": 0.0591,
+      "step": 373
+    },
+    {
+      "epoch": 1.1197604790419162,
+      "grad_norm": 1.712419097605987,
+      "learning_rate": 4.064941365442084e-06,
+      "loss": 0.0783,
+      "step": 374
+    },
+    {
+      "epoch": 1.122754491017964,
+      "grad_norm": 2.1582106832047914,
+      "learning_rate": 4.041851733968348e-06,
+      "loss": 0.0797,
+      "step": 375
+    },
+    {
+      "epoch": 1.125748502994012,
+      "grad_norm": 1.7501771880679808,
+      "learning_rate": 4.018783294826071e-06,
+      "loss": 0.0728,
+      "step": 376
+    },
+    {
+      "epoch": 1.1287425149700598,
+      "grad_norm": 2.204665204357101,
+      "learning_rate": 3.995736558243186e-06,
+      "loss": 0.0891,
+      "step": 377
+    },
+    {
+      "epoch": 1.1317365269461077,
+      "grad_norm": 1.6816019368680264,
+      "learning_rate": 3.972712033967608e-06,
+      "loss": 0.0791,
+      "step": 378
+    },
+    {
+      "epoch": 1.1347305389221556,
+      "grad_norm": 1.7592135314721509,
+      "learning_rate": 3.949710231255961e-06,
+      "loss": 0.0875,
+      "step": 379
+    },
+    {
+      "epoch": 1.1377245508982037,
+      "grad_norm": 1.7843416729759316,
+      "learning_rate": 3.926731658862307e-06,
+      "loss": 0.0741,
+      "step": 380
+    },
+    {
+      "epoch": 1.1407185628742516,
+      "grad_norm": 1.5241929382191561,
+      "learning_rate": 3.903776825026912e-06,
+      "loss": 0.0695,
+      "step": 381
+    },
+    {
+      "epoch": 1.1437125748502994,
+      "grad_norm": 1.7941445508860325,
+      "learning_rate": 3.8808462374649805e-06,
+      "loss": 0.086,
+      "step": 382
+    },
+    {
+      "epoch": 1.1467065868263473,
+      "grad_norm": 1.5190035907041612,
+      "learning_rate": 3.857940403355444e-06,
+      "loss": 0.0736,
+      "step": 383
+    },
+    {
+      "epoch": 1.1497005988023952,
+      "grad_norm": 1.9045880699760642,
+      "learning_rate": 3.8350598293297345e-06,
+      "loss": 0.0792,
+      "step": 384
+    },
+    {
+      "epoch": 1.152694610778443,
+      "grad_norm": 1.817049384734948,
+      "learning_rate": 3.8122050214605822e-06,
+      "loss": 0.0757,
+      "step": 385
+    },
+    {
+      "epoch": 1.1556886227544911,
+      "grad_norm": 1.9706654149884935,
+      "learning_rate": 3.7893764852508207e-06,
+      "loss": 0.099,
+      "step": 386
+    },
+    {
+      "epoch": 1.158682634730539,
+      "grad_norm": 1.4983694826524019,
+      "learning_rate": 3.766574725622208e-06,
+      "loss": 0.0615,
+      "step": 387
+    },
+    {
+      "epoch": 1.1616766467065869,
+      "grad_norm": 1.482082132611248,
+      "learning_rate": 3.7438002469042567e-06,
+      "loss": 0.064,
+      "step": 388
+    },
+    {
+      "epoch": 1.1646706586826348,
+      "grad_norm": 1.6334637047926808,
+      "learning_rate": 3.721053552823078e-06,
+      "loss": 0.0805,
+      "step": 389
+    },
+    {
+      "epoch": 1.1676646706586826,
+      "grad_norm": 1.6369641513062678,
+      "learning_rate": 3.698335146490246e-06,
+      "loss": 0.0717,
+      "step": 390
+    },
+    {
+      "epoch": 1.1706586826347305,
+      "grad_norm": 1.6607679847012171,
+      "learning_rate": 3.675645530391665e-06,
+      "loss": 0.0677,
+      "step": 391
+    },
+    {
+      "epoch": 1.1736526946107784,
+      "grad_norm": 1.6795086730378275,
+      "learning_rate": 3.652985206376455e-06,
+      "loss": 0.0723,
+      "step": 392
+    },
+    {
+      "epoch": 1.1766467065868262,
+      "grad_norm": 1.5817763161987946,
+      "learning_rate": 3.630354675645853e-06,
+      "loss": 0.0797,
+      "step": 393
+    },
+    {
+      "epoch": 1.1796407185628743,
+      "grad_norm": 1.6741475397894603,
+      "learning_rate": 3.6077544387421293e-06,
+      "loss": 0.0749,
+      "step": 394
+    },
+    {
+      "epoch": 1.1826347305389222,
+      "grad_norm": 1.468731716725861,
+      "learning_rate": 3.5851849955375177e-06,
+      "loss": 0.0683,
+      "step": 395
+    },
+    {
+      "epoch": 1.18562874251497,
+      "grad_norm": 1.5966620497838973,
+      "learning_rate": 3.5626468452231534e-06,
+      "loss": 0.0662,
+      "step": 396
+    },
+    {
+      "epoch": 1.188622754491018,
+      "grad_norm": 1.4988458840519034,
+      "learning_rate": 3.540140486298035e-06,
+      "loss": 0.0688,
+      "step": 397
+    },
+    {
+      "epoch": 1.1916167664670658,
+      "grad_norm": 1.897868761474463,
+      "learning_rate": 3.517666416557999e-06,
+      "loss": 0.0873,
+      "step": 398
+    },
+    {
+      "epoch": 1.1946107784431137,
+      "grad_norm": 1.5705474517548206,
+      "learning_rate": 3.495225133084712e-06,
+      "loss": 0.0584,
+      "step": 399
+    },
+    {
+      "epoch": 1.1976047904191618,
+      "grad_norm": 1.3344210140134862,
+      "learning_rate": 3.472817132234669e-06,
+      "loss": 0.0572,
+      "step": 400
+    },
+    {
+      "epoch": 1.1976047904191618,
+      "eval_loss": 0.22033333778381348,
+      "eval_runtime": 8.224,
+      "eval_samples_per_second": 6.566,
+      "eval_steps_per_second": 1.702,
+      "step": 400
+    },
+    {
+      "epoch": 1.2005988023952097,
+      "grad_norm": 1.509203481241487,
+      "learning_rate": 3.4504429096282246e-06,
+      "loss": 0.0649,
+      "step": 401
+    },
+    {
+      "epoch": 1.2035928143712575,
+      "grad_norm": 1.663988063233224,
+      "learning_rate": 3.428102960138625e-06,
+      "loss": 0.0805,
+      "step": 402
+    },
+    {
+      "epoch": 1.2065868263473054,
+      "grad_norm": 1.9314945868291045,
+      "learning_rate": 3.405797777881059e-06,
+      "loss": 0.0813,
+      "step": 403
+    },
+    {
+      "epoch": 1.2095808383233533,
+      "grad_norm": 1.4719718836020617,
+      "learning_rate": 3.3835278562017405e-06,
+      "loss": 0.063,
+      "step": 404
+    },
+    {
+      "epoch": 1.2125748502994012,
+      "grad_norm": 1.499613775700233,
+      "learning_rate": 3.3612936876669834e-06,
+      "loss": 0.0649,
+      "step": 405
+    },
+    {
+      "epoch": 1.215568862275449,
+      "grad_norm": 1.6790152034578139,
+      "learning_rate": 3.3390957640523147e-06,
+      "loss": 0.0799,
+      "step": 406
+    },
+    {
+      "epoch": 1.218562874251497,
+      "grad_norm": 1.4293726118465215,
+      "learning_rate": 3.3169345763315986e-06,
+      "loss": 0.0643,
+      "step": 407
+    },
+    {
+      "epoch": 1.221556886227545,
+      "grad_norm": 1.6923162381802592,
+      "learning_rate": 3.29481061466617e-06,
+      "loss": 0.0787,
+      "step": 408
+    },
+    {
+      "epoch": 1.2245508982035929,
+      "grad_norm": 1.9540770399329328,
+      "learning_rate": 3.2727243683940045e-06,
+      "loss": 0.0798,
+      "step": 409
+    },
+    {
+      "epoch": 1.2275449101796407,
+      "grad_norm": 1.53124631972842,
+      "learning_rate": 3.2506763260188824e-06,
+      "loss": 0.07,
+      "step": 410
+    },
+    {
+      "epoch": 1.2305389221556886,
+      "grad_norm": 1.8956502589856736,
+      "learning_rate": 3.2286669751995905e-06,
+      "loss": 0.0631,
+      "step": 411
+    },
+    {
+      "epoch": 1.2335329341317365,
+      "grad_norm": 1.5963855425169526,
+      "learning_rate": 3.2066968027391377e-06,
+      "loss": 0.061,
+      "step": 412
+    },
+    {
+      "epoch": 1.2365269461077844,
+      "grad_norm": 1.4497407072319592,
+      "learning_rate": 3.1847662945739833e-06,
+      "loss": 0.0654,
+      "step": 413
+    },
+    {
+      "epoch": 1.2395209580838324,
+      "grad_norm": 1.5941666688708278,
+      "learning_rate": 3.1628759357632943e-06,
+      "loss": 0.0637,
+      "step": 414
+    },
+    {
+      "epoch": 1.2425149700598803,
+      "grad_norm": 1.632544551391709,
+      "learning_rate": 3.1410262104782086e-06,
+      "loss": 0.066,
+      "step": 415
+    },
+    {
+      "epoch": 1.2455089820359282,
+      "grad_norm": 1.7637405015085608,
+      "learning_rate": 3.119217601991139e-06,
+      "loss": 0.0723,
+      "step": 416
+    },
+    {
+      "epoch": 1.248502994011976,
+      "grad_norm": 1.6598101724135699,
+      "learning_rate": 3.0974505926650724e-06,
+      "loss": 0.0738,
+      "step": 417
+    },
+    {
+      "epoch": 1.251497005988024,
+      "grad_norm": 1.688140140076491,
+      "learning_rate": 3.0757256639429027e-06,
+      "loss": 0.0746,
+      "step": 418
+    },
+    {
+      "epoch": 1.2544910179640718,
+      "grad_norm": 1.5828127729692618,
+      "learning_rate": 3.0540432963367907e-06,
+      "loss": 0.0689,
+      "step": 419
+    },
+    {
+      "epoch": 1.2574850299401197,
+      "grad_norm": 1.4519798068768506,
+      "learning_rate": 3.032403969417523e-06,
+      "loss": 0.0564,
+      "step": 420
+    },
+    {
+      "epoch": 1.2604790419161676,
+      "grad_norm": 1.4632343777217751,
+      "learning_rate": 3.010808161803917e-06,
+      "loss": 0.0658,
+      "step": 421
+    },
+    {
+      "epoch": 1.2634730538922156,
+      "grad_norm": 1.8516168540542934,
+      "learning_rate": 2.9892563511522305e-06,
+      "loss": 0.0789,
+      "step": 422
+    },
+    {
+      "epoch": 1.2664670658682635,
+      "grad_norm": 1.686450569359416,
+      "learning_rate": 2.9677490141455915e-06,
+      "loss": 0.0723,
+      "step": 423
+    },
+    {
+      "epoch": 1.2694610778443114,
+      "grad_norm": 1.4595915410943603,
+      "learning_rate": 2.946286626483463e-06,
+      "loss": 0.0646,
+      "step": 424
+    },
+    {
+      "epoch": 1.2724550898203593,
+      "grad_norm": 1.6997011233873018,
+      "learning_rate": 2.924869662871117e-06,
+      "loss": 0.0697,
+      "step": 425
+    },
+    {
+      "epoch": 1.2754491017964071,
+      "grad_norm": 1.4184165876798536,
+      "learning_rate": 2.903498597009136e-06,
+      "loss": 0.0732,
+      "step": 426
+    },
+    {
+      "epoch": 1.278443113772455,
+      "grad_norm": 1.6603658072095377,
+      "learning_rate": 2.8821739015829338e-06,
+      "loss": 0.0681,
+      "step": 427
+    },
+    {
+      "epoch": 1.281437125748503,
+      "grad_norm": 1.330088931854646,
+      "learning_rate": 2.8608960482523058e-06,
+      "loss": 0.0523,
+      "step": 428
+    },
+    {
+      "epoch": 1.284431137724551,
+      "grad_norm": 1.6805566750189211,
+      "learning_rate": 2.839665507640992e-06,
+      "loss": 0.0586,
+      "step": 429
+    },
+    {
+      "epoch": 1.2874251497005988,
+      "grad_norm": 1.4883063566543024,
+      "learning_rate": 2.818482749326272e-06,
+      "loss": 0.0625,
+      "step": 430
+    },
+    {
+      "epoch": 1.2904191616766467,
+      "grad_norm": 1.621974029335736,
+      "learning_rate": 2.797348241828569e-06,
+      "loss": 0.0666,
+      "step": 431
+    },
+    {
+      "epoch": 1.2934131736526946,
+      "grad_norm": 1.4140428833956944,
+      "learning_rate": 2.776262452601104e-06,
+      "loss": 0.0583,
+      "step": 432
+    },
+    {
+      "epoch": 1.2964071856287425,
+      "grad_norm": 1.77569250600627,
+      "learning_rate": 2.7552258480195348e-06,
+      "loss": 0.0702,
+      "step": 433
+    },
+    {
+      "epoch": 1.2994011976047903,
+      "grad_norm": 1.7623532468035936,
+      "learning_rate": 2.734238893371667e-06,
+      "loss": 0.065,
+      "step": 434
+    },
+    {
+      "epoch": 1.3023952095808382,
+      "grad_norm": 1.7274081363054548,
+      "learning_rate": 2.7133020528471322e-06,
+      "loss": 0.0733,
+      "step": 435
+    },
+    {
+      "epoch": 1.3053892215568863,
+      "grad_norm": 1.5374566937974774,
+      "learning_rate": 2.6924157895271563e-06,
+      "loss": 0.065,
+      "step": 436
+    },
+    {
+      "epoch": 1.3083832335329342,
+      "grad_norm": 1.702719988886982,
+      "learning_rate": 2.671580565374282e-06,
+      "loss": 0.0802,
+      "step": 437
+    },
+    {
+      "epoch": 1.311377245508982,
+      "grad_norm": 1.5430538852855165,
+      "learning_rate": 2.6507968412221763e-06,
+      "loss": 0.0599,
+      "step": 438
+    },
+    {
+      "epoch": 1.31437125748503,
+      "grad_norm": 1.6353808988183587,
+      "learning_rate": 2.6300650767654234e-06,
+      "loss": 0.0663,
+      "step": 439
+    },
+    {
+      "epoch": 1.3173652694610778,
+      "grad_norm": 1.6835699495900533,
+      "learning_rate": 2.6093857305493666e-06,
+      "loss": 0.07,
+      "step": 440
+    },
+    {
+      "epoch": 1.3203592814371259,
+      "grad_norm": 1.74763591664886,
+      "learning_rate": 2.588759259959962e-06,
+      "loss": 0.0701,
+      "step": 441
+    },
+    {
+      "epoch": 1.3233532934131738,
+      "grad_norm": 1.5464446376200454,
+      "learning_rate": 2.568186121213658e-06,
+      "loss": 0.066,
+      "step": 442
+    },
+    {
+      "epoch": 1.3263473053892216,
+      "grad_norm": 1.6855674534648784,
+      "learning_rate": 2.547666769347312e-06,
+      "loss": 0.0715,
+      "step": 443
+    },
+    {
+      "epoch": 1.3293413173652695,
+      "grad_norm": 1.957856453570676,
+      "learning_rate": 2.5272016582081236e-06,
+      "loss": 0.0768,
+      "step": 444
+    },
+    {
+      "epoch": 1.3323353293413174,
+      "grad_norm": 1.6878344723594567,
+      "learning_rate": 2.5067912404435952e-06,
+      "loss": 0.0742,
+      "step": 445
+    },
+    {
+      "epoch": 1.3353293413173652,
+      "grad_norm": 1.9563961195759916,
+      "learning_rate": 2.486435967491516e-06,
+      "loss": 0.0823,
+      "step": 446
+    },
+    {
+      "epoch": 1.3383233532934131,
+      "grad_norm": 1.6427157264989363,
+      "learning_rate": 2.4661362895699903e-06,
+      "loss": 0.073,
+      "step": 447
+    },
+    {
+      "epoch": 1.341317365269461,
+      "grad_norm": 1.7794560678013864,
+      "learning_rate": 2.445892655667462e-06,
+      "loss": 0.0838,
+      "step": 448
+    },
+    {
+      "epoch": 1.3443113772455089,
+      "grad_norm": 1.6943204928496665,
+      "learning_rate": 2.425705513532798e-06,
+      "loss": 0.0707,
+      "step": 449
+    },
+    {
+      "epoch": 1.347305389221557,
+      "grad_norm": 1.7476502460564978,
+      "learning_rate": 2.4055753096653795e-06,
+      "loss": 0.0806,
+      "step": 450
+    },
+    {
+      "epoch": 1.3502994011976048,
+      "grad_norm": 1.542490279360059,
+      "learning_rate": 2.3855024893052286e-06,
+      "loss": 0.0669,
+      "step": 451
+    },
+    {
+      "epoch": 1.3532934131736527,
+      "grad_norm": 1.521902280634556,
+      "learning_rate": 2.365487496423152e-06,
+      "loss": 0.0749,
+      "step": 452
+    },
+    {
+      "epoch": 1.3562874251497006,
+      "grad_norm": 1.5498938795735708,
+      "learning_rate": 2.3455307737109338e-06,
+      "loss": 0.0611,
+      "step": 453
+    },
+    {
+      "epoch": 1.3592814371257484,
+      "grad_norm": 1.6724684061742812,
+      "learning_rate": 2.3256327625715345e-06,
+      "loss": 0.0691,
+      "step": 454
+    },
+    {
+      "epoch": 1.3622754491017965,
+      "grad_norm": 1.4529750183483368,
+      "learning_rate": 2.3057939031093346e-06,
+      "loss": 0.0583,
+      "step": 455
+    },
+    {
+      "epoch": 1.3652694610778444,
+      "grad_norm": 1.3052078589756266,
+      "learning_rate": 2.2860146341203936e-06,
+      "loss": 0.052,
+      "step": 456
+    },
+    {
+      "epoch": 1.3682634730538923,
+      "grad_norm": 2.074943975842417,
+      "learning_rate": 2.2662953930827546e-06,
+      "loss": 0.0927,
+      "step": 457
+    },
+    {
+      "epoch": 1.3712574850299402,
+      "grad_norm": 1.4289271585538827,
+      "learning_rate": 2.2466366161467528e-06,
+      "loss": 0.0549,
+      "step": 458
+    },
+    {
+      "epoch": 1.374251497005988,
+      "grad_norm": 1.5901826623996778,
+      "learning_rate": 2.227038738125385e-06,
+      "loss": 0.0627,
+      "step": 459
+    },
+    {
+      "epoch": 1.377245508982036,
+      "grad_norm": 2.1513825844652428,
+      "learning_rate": 2.207502192484685e-06,
+      "loss": 0.0799,
+      "step": 460
+    },
+    {
+      "epoch": 1.3802395209580838,
+      "grad_norm": 1.4365014532880382,
+      "learning_rate": 2.188027411334131e-06,
+      "loss": 0.061,
+      "step": 461
+    },
+    {
+      "epoch": 1.3832335329341316,
+      "grad_norm": 1.6088196264830839,
+      "learning_rate": 2.1686148254171012e-06,
+      "loss": 0.0726,
+      "step": 462
+    },
+    {
+      "epoch": 1.3862275449101795,
+      "grad_norm": 1.417572694624261,
+      "learning_rate": 2.1492648641013305e-06,
+      "loss": 0.0584,
+      "step": 463
+    },
+    {
+      "epoch": 1.3892215568862276,
+      "grad_norm": 1.1909084265659045,
+      "learning_rate": 2.1299779553694323e-06,
+      "loss": 0.047,
+      "step": 464
+    },
+    {
+      "epoch": 1.3922155688622755,
+      "grad_norm": 1.6061851340192603,
+      "learning_rate": 2.1107545258094135e-06,
+      "loss": 0.0701,
+      "step": 465
+    },
+    {
+      "epoch": 1.3952095808383234,
+      "grad_norm": 1.5194746123801195,
+      "learning_rate": 2.0915950006052555e-06,
+      "loss": 0.0555,
+      "step": 466
+    },
+    {
+      "epoch": 1.3982035928143712,
+      "grad_norm": 1.591247848532409,
+      "learning_rate": 2.0724998035274947e-06,
+      "loss": 0.059,
+      "step": 467
+    },
+    {
+      "epoch": 1.401197604790419,
+      "grad_norm": 1.526627533837993,
+      "learning_rate": 2.053469356923865e-06,
+      "loss": 0.061,
+      "step": 468
+    },
+    {
+      "epoch": 1.4041916167664672,
+      "grad_norm": 1.5731828869635374,
+      "learning_rate": 2.0345040817099433e-06,
+      "loss": 0.0573,
+      "step": 469
+    },
+    {
+      "epoch": 1.407185628742515,
+      "grad_norm": 1.3960369569397377,
+      "learning_rate": 2.0156043973598475e-06,
+      "loss": 0.0569,
+      "step": 470
+    },
+    {
+      "epoch": 1.410179640718563,
+      "grad_norm": 1.7112322975360998,
+      "learning_rate": 1.996770721896957e-06,
+      "loss": 0.0652,
+      "step": 471
+    },
+    {
+      "epoch": 1.4131736526946108,
+      "grad_norm": 1.359460170906636,
+      "learning_rate": 1.9780034718846653e-06,
+      "loss": 0.0512,
+      "step": 472
+    },
+    {
+      "epoch": 1.4161676646706587,
+      "grad_norm": 1.7530123447068766,
+      "learning_rate": 1.9593030624171683e-06,
+      "loss": 0.0751,
+      "step": 473
+    },
+    {
+      "epoch": 1.4191616766467066,
+      "grad_norm": 1.6318021870040988,
+      "learning_rate": 1.9406699071102774e-06,
+      "loss": 0.0652,
+      "step": 474
+    },
+    {
+      "epoch": 1.4221556886227544,
+      "grad_norm": 1.3604726886408112,
+      "learning_rate": 1.9221044180922833e-06,
+      "loss": 0.0495,
+      "step": 475
+    },
+    {
+      "epoch": 1.4251497005988023,
+      "grad_norm": 1.642141522002382,
+      "learning_rate": 1.9036070059948253e-06,
+      "loss": 0.0692,
+      "step": 476
+    },
+    {
+      "epoch": 1.4281437125748502,
+      "grad_norm": 2.150202208561037,
+      "learning_rate": 1.885178079943823e-06,
+      "loss": 0.0712,
+      "step": 477
+    },
+    {
+      "epoch": 1.4311377245508983,
+      "grad_norm": 1.707885082211808,
+      "learning_rate": 1.866818047550419e-06,
+      "loss": 0.0786,
+      "step": 478
+    },
+    {
+      "epoch": 1.4341317365269461,
+      "grad_norm": 1.4590788670860722,
+      "learning_rate": 1.8485273149019655e-06,
+      "loss": 0.0629,
+      "step": 479
+    },
+    {
+      "epoch": 1.437125748502994,
+      "grad_norm": 1.4846552788026863,
+      "learning_rate": 1.8303062865530407e-06,
+      "loss": 0.0585,
+      "step": 480
+    },
+    {
+      "epoch": 1.4401197604790419,
+      "grad_norm": 1.6667782606534387,
+      "learning_rate": 1.8121553655165058e-06,
+      "loss": 0.0613,
+      "step": 481
+    },
+    {
+      "epoch": 1.4431137724550898,
+      "grad_norm": 1.6148047544966386,
+      "learning_rate": 1.7940749532545832e-06,
+      "loss": 0.0619,
+      "step": 482
+    },
+    {
+      "epoch": 1.4461077844311379,
+      "grad_norm": 1.7788847742716485,
+      "learning_rate": 1.7760654496699876e-06,
+      "loss": 0.0619,
+      "step": 483
+    },
+    {
+      "epoch": 1.4491017964071857,
+      "grad_norm": 1.702765040079345,
+      "learning_rate": 1.7581272530970666e-06,
+      "loss": 0.0615,
+      "step": 484
+    },
+    {
+      "epoch": 1.4520958083832336,
+      "grad_norm": 1.6617274436982556,
+      "learning_rate": 1.7402607602930106e-06,
+      "loss": 0.0658,
+      "step": 485
+    },
+    {
+      "epoch": 1.4550898203592815,
+      "grad_norm": 1.7734416612744945,
+      "learning_rate": 1.7224663664290537e-06,
+      "loss": 0.0653,
+      "step": 486
+    },
+    {
+      "epoch": 1.4580838323353293,
+      "grad_norm": 1.7349459216554302,
+      "learning_rate": 1.7047444650817518e-06,
+      "loss": 0.0726,
+      "step": 487
+    },
+    {
+      "epoch": 1.4610778443113772,
+      "grad_norm": 1.5134802529028657,
+      "learning_rate": 1.6870954482242707e-06,
+      "loss": 0.0585,
+      "step": 488
+    },
+    {
+      "epoch": 1.464071856287425,
+      "grad_norm": 1.5725749928492743,
+      "learning_rate": 1.669519706217711e-06,
+      "loss": 0.0547,
+      "step": 489
+    },
+    {
+      "epoch": 1.467065868263473,
+      "grad_norm": 1.9001228452023615,
+      "learning_rate": 1.652017627802487e-06,
+      "loss": 0.0828,
+      "step": 490
+    },
+    {
+      "epoch": 1.4700598802395208,
+      "grad_norm": 1.4623304753019872,
+      "learning_rate": 1.6345896000897122e-06,
+      "loss": 0.0602,
+      "step": 491
+    },
+    {
+      "epoch": 1.473053892215569,
+      "grad_norm": 1.6330915295329298,
+      "learning_rate": 1.6172360085526567e-06,
+      "loss": 0.073,
+      "step": 492
+    },
+    {
+      "epoch": 1.4760479041916168,
+      "grad_norm": 1.5163127084117827,
+      "learning_rate": 1.5999572370182016e-06,
+      "loss": 0.0669,
+      "step": 493
+    },
+    {
+      "epoch": 1.4790419161676647,
+      "grad_norm": 1.4492306006483444,
+      "learning_rate": 1.5827536676583643e-06,
+      "loss": 0.0617,
+      "step": 494
+    },
+    {
+      "epoch": 1.4820359281437125,
+      "grad_norm": 1.4422482047382028,
+      "learning_rate": 1.5656256809818343e-06,
+      "loss": 0.0632,
+      "step": 495
+    },
+    {
+      "epoch": 1.4850299401197604,
+      "grad_norm": 1.4155915997064654,
+      "learning_rate": 1.54857365582557e-06,
+      "loss": 0.056,
+      "step": 496
+    },
+    {
+      "epoch": 1.4880239520958085,
+      "grad_norm": 1.6685703008998483,
+      "learning_rate": 1.5315979693464039e-06,
+      "loss": 0.07,
+      "step": 497
+    },
+    {
+      "epoch": 1.4910179640718564,
+      "grad_norm": 1.3574935609472154,
+      "learning_rate": 1.5146989970127158e-06,
+      "loss": 0.0536,
+      "step": 498
+    },
+    {
+      "epoch": 1.4940119760479043,
+      "grad_norm": 1.7351475189027918,
+      "learning_rate": 1.4978771125961177e-06,
+      "loss": 0.0628,
+      "step": 499
+    },
+    {
+      "epoch": 1.4970059880239521,
+      "grad_norm": 1.501798000111715,
+      "learning_rate": 1.4811326881631937e-06,
+      "loss": 0.0644,
+      "step": 500
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.9380122568782592,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 0.0691,
+      "step": 501
+    },
+    {
+      "epoch": 1.5029940119760479,
+      "grad_norm": 1.8677400097817,
+      "learning_rate": 1.4478776989401949e-06,
+      "loss": 0.068,
+      "step": 502
+    },
+    {
+      "epoch": 1.5059880239520957,
+      "grad_norm": 1.5306475941608269,
+      "learning_rate": 1.4313678696842559e-06,
+      "loss": 0.0667,
+      "step": 503
+    },
+    {
+      "epoch": 1.5089820359281436,
+      "grad_norm": 1.4740425732387699,
+      "learning_rate": 1.4149369714639856e-06,
+      "loss": 0.0596,
+      "step": 504
+    },
+    {
+      "epoch": 1.5119760479041915,
+      "grad_norm": 1.7108366077763746,
+      "learning_rate": 1.3985853676981316e-06,
+      "loss": 0.0711,
+      "step": 505
+    },
+    {
+      "epoch": 1.5149700598802394,
+      "grad_norm": 1.5348864966453744,
+      "learning_rate": 1.3823134200516043e-06,
+      "loss": 0.056,
+      "step": 506
+    },
+    {
+      "epoch": 1.5179640718562875,
+      "grad_norm": 1.6173070700082506,
+      "learning_rate": 1.366121488427481e-06,
+      "loss": 0.0635,
+      "step": 507
+    },
+    {
+      "epoch": 1.5209580838323353,
+      "grad_norm": 1.4877631156195827,
+      "learning_rate": 1.3500099309590397e-06,
+      "loss": 0.0544,
+      "step": 508
+    },
+    {
+      "epoch": 1.5239520958083832,
+      "grad_norm": 1.7098095270857783,
+      "learning_rate": 1.3339791040018479e-06,
+      "loss": 0.062,
+      "step": 509
+    },
+    {
+      "epoch": 1.5269461077844313,
+      "grad_norm": 1.5871159646614867,
+      "learning_rate": 1.3180293621258694e-06,
+      "loss": 0.065,
+      "step": 510
+    },
+    {
+      "epoch": 1.5299401197604792,
+      "grad_norm": 1.5818163043797837,
+      "learning_rate": 1.3021610581076316e-06,
+      "loss": 0.0599,
+      "step": 511
+    },
+    {
+      "epoch": 1.532934131736527,
+      "grad_norm": 1.6961991774421827,
+      "learning_rate": 1.2863745429224145e-06,
+      "loss": 0.0675,
+      "step": 512
+    },
+    {
+      "epoch": 1.535928143712575,
+      "grad_norm": 1.353481536813635,
+      "learning_rate": 1.270670165736499e-06,
+      "loss": 0.0587,
+      "step": 513
+    },
+    {
+      "epoch": 1.5389221556886228,
+      "grad_norm": 2.075413541723379,
+      "learning_rate": 1.2550482738994284e-06,
+      "loss": 0.0708,
+      "step": 514
+    },
+    {
+      "epoch": 1.5419161676646707,
+      "grad_norm": 2.156517580216609,
+      "learning_rate": 1.239509212936343e-06,
+      "loss": 0.0759,
+      "step": 515
+    },
+    {
+      "epoch": 1.5449101796407185,
+      "grad_norm": 1.5563800059790636,
+      "learning_rate": 1.22405332654032e-06,
+      "loss": 0.0552,
+      "step": 516
+    },
+    {
+      "epoch": 1.5479041916167664,
+      "grad_norm": 1.6339895381730996,
+      "learning_rate": 1.2086809565647877e-06,
+      "loss": 0.0717,
+      "step": 517
+    },
+    {
+      "epoch": 1.5508982035928143,
+      "grad_norm": 1.5880394889003904,
+      "learning_rate": 1.1933924430159571e-06,
+      "loss": 0.0594,
+      "step": 518
+    },
+    {
+      "epoch": 1.5538922155688621,
+      "grad_norm": 1.6646502631772824,
+      "learning_rate": 1.1781881240452958e-06,
+      "loss": 0.0576,
+      "step": 519
+    },
+    {
+      "epoch": 1.55688622754491,
+      "grad_norm": 1.7047181958205797,
+      "learning_rate": 1.1630683359420653e-06,
+      "loss": 0.0619,
+      "step": 520
+    },
+    {
+      "epoch": 1.5598802395209581,
+      "grad_norm": 1.5259457125093634,
+      "learning_rate": 1.1480334131258626e-06,
+      "loss": 0.0656,
+      "step": 521
+    },
+    {
+      "epoch": 1.562874251497006,
+      "grad_norm": 1.8796790886306889,
+      "learning_rate": 1.1330836881392405e-06,
+      "loss": 0.0687,
+      "step": 522
+    },
+    {
+      "epoch": 1.5658682634730539,
+      "grad_norm": 1.4783310524557831,
+      "learning_rate": 1.11821949164034e-06,
+      "loss": 0.0619,
+      "step": 523
+    },
+    {
+      "epoch": 1.568862275449102,
+      "grad_norm": 1.4369349853682416,
+      "learning_rate": 1.103441152395588e-06,
+      "loss": 0.0507,
+      "step": 524
+    },
+    {
+      "epoch": 1.5718562874251498,
+      "grad_norm": 1.71495106324488,
+      "learning_rate": 1.088748997272414e-06,
+      "loss": 0.0704,
+      "step": 525
+    },
+    {
+      "epoch": 1.5748502994011977,
+      "grad_norm": 1.727863326675934,
+      "learning_rate": 1.0741433512320316e-06,
+      "loss": 0.0565,
+      "step": 526
+    },
+    {
+      "epoch": 1.5778443113772456,
+      "grad_norm": 1.5096858499910297,
+      "learning_rate": 1.0596245373222424e-06,
+      "loss": 0.0544,
+      "step": 527
+    },
+    {
+      "epoch": 1.5808383233532934,
+      "grad_norm": 1.5896944203875445,
+      "learning_rate": 1.045192876670298e-06,
+      "loss": 0.0654,
+      "step": 528
+    },
+    {
+      "epoch": 1.5838323353293413,
+      "grad_norm": 1.6133500073019824,
+      "learning_rate": 1.0308486884757868e-06,
+      "loss": 0.0598,
+      "step": 529
+    },
+    {
+      "epoch": 1.5868263473053892,
+      "grad_norm": 1.520674500521713,
+      "learning_rate": 1.0165922900035886e-06,
+      "loss": 0.0601,
+      "step": 530
+    },
+    {
+      "epoch": 1.589820359281437,
+      "grad_norm": 1.3249572670003777,
+      "learning_rate": 1.0024239965768417e-06,
+      "loss": 0.0555,
+      "step": 531
+    },
+    {
+      "epoch": 1.592814371257485,
+      "grad_norm": 1.5004920862618154,
+      "learning_rate": 9.883441215699824e-07,
+      "loss": 0.0507,
+      "step": 532
+    },
+    {
+      "epoch": 1.5958083832335328,
+      "grad_norm": 1.635842670927333,
+      "learning_rate": 9.74352976401805e-07,
+      "loss": 0.0571,
+      "step": 533
+    },
+    {
+      "epoch": 1.5988023952095807,
+      "grad_norm": 1.920114894092032,
+      "learning_rate": 9.604508705285765e-07,
+      "loss": 0.0741,
+      "step": 534
+    },
+    {
+      "epoch": 1.6017964071856288,
+      "grad_norm": 1.5320152804916445,
+      "learning_rate": 9.466381114371942e-07,
+      "loss": 0.0631,
+      "step": 535
+    },
+    {
+      "epoch": 1.6047904191616766,
+      "grad_norm": 1.7970259875176704,
+      "learning_rate": 9.329150046383773e-07,
+      "loss": 0.0721,
+      "step": 536
+    },
+    {
+      "epoch": 1.6077844311377245,
+      "grad_norm": 1.3984386223935281,
+      "learning_rate": 9.192818536599213e-07,
+      "loss": 0.0635,
+      "step": 537
+    },
+    {
+      "epoch": 1.6107784431137726,
+      "grad_norm": 1.643344266454895,
+      "learning_rate": 9.057389600399719e-07,
+      "loss": 0.0641,
+      "step": 538
+    },
+    {
+      "epoch": 1.6137724550898205,
+      "grad_norm": 1.581458222608595,
+      "learning_rate": 8.922866233203681e-07,
+      "loss": 0.0672,
+      "step": 539
+    },
+    {
+      "epoch": 1.6167664670658684,
+      "grad_norm": 1.4549179718026544,
+      "learning_rate": 8.789251410400024e-07,
+      "loss": 0.061,
+      "step": 540
+    },
+    {
+      "epoch": 1.6197604790419162,
+      "grad_norm": 1.4262995809700199,
+      "learning_rate": 8.65654808728259e-07,
+      "loss": 0.0586,
+      "step": 541
+    },
+    {
+      "epoch": 1.622754491017964,
+      "grad_norm": 1.7193508626218372,
+      "learning_rate": 8.524759198984567e-07,
+      "loss": 0.0591,
+      "step": 542
+    },
+    {
+      "epoch": 1.625748502994012,
+      "grad_norm": 1.7517355021439103,
+      "learning_rate": 8.393887660413719e-07,
+      "loss": 0.0545,
+      "step": 543
+    },
+    {
+      "epoch": 1.6287425149700598,
+      "grad_norm": 1.8983971281903382,
+      "learning_rate": 8.263936366187825e-07,
+      "loss": 0.0867,
+      "step": 544
+    },
+    {
+      "epoch": 1.6317365269461077,
+      "grad_norm": 1.6468897821944162,
+      "learning_rate": 8.134908190570723e-07,
+      "loss": 0.0696,
+      "step": 545
+    },
+    {
+      "epoch": 1.6347305389221556,
+      "grad_norm": 1.5818721044261885,
+      "learning_rate": 8.006805987408705e-07,
+      "loss": 0.0659,
+      "step": 546
+    },
+    {
+      "epoch": 1.6377245508982035,
+      "grad_norm": 1.3110735414379466,
+      "learning_rate": 7.879632590067354e-07,
+      "loss": 0.0536,
+      "step": 547
+    },
+    {
+      "epoch": 1.6407185628742516,
+      "grad_norm": 1.3314149501781605,
+      "learning_rate": 7.753390811368972e-07,
+      "loss": 0.047,
+      "step": 548
+    },
+    {
+      "epoch": 1.6437125748502994,
+      "grad_norm": 1.663574670238363,
+      "learning_rate": 7.628083443530287e-07,
+      "loss": 0.0607,
+      "step": 549
+    },
+    {
+      "epoch": 1.6467065868263473,
+      "grad_norm": 1.8582278422369445,
+      "learning_rate": 7.503713258100726e-07,
+      "loss": 0.0685,
+      "step": 550
+    },
+    {
+      "epoch": 1.6497005988023952,
+      "grad_norm": 1.5800917114821824,
+      "learning_rate": 7.380283005901084e-07,
+      "loss": 0.0587,
+      "step": 551
+    },
+    {
+      "epoch": 1.6526946107784433,
+      "grad_norm": 1.4489853041205063,
+      "learning_rate": 7.257795416962754e-07,
+      "loss": 0.0585,
+      "step": 552
+    },
+    {
+      "epoch": 1.6556886227544911,
+      "grad_norm": 1.299351649210388,
+      "learning_rate": 7.136253200467231e-07,
+      "loss": 0.0431,
+      "step": 553
+    },
+    {
+      "epoch": 1.658682634730539,
+      "grad_norm": 1.4842731624193612,
+      "learning_rate": 7.015659044686307e-07,
+      "loss": 0.0531,
+      "step": 554
+    },
+    {
+      "epoch": 1.6616766467065869,
+      "grad_norm": 1.9063767406849013,
+      "learning_rate": 6.896015616922535e-07,
+      "loss": 0.0644,
+      "step": 555
+    },
+    {
+      "epoch": 1.6646706586826348,
+      "grad_norm": 2.082571136222972,
+      "learning_rate": 6.777325563450282e-07,
+      "loss": 0.0706,
+      "step": 556
+    },
+    {
+      "epoch": 1.6676646706586826,
+      "grad_norm": 1.7589007160444798,
+      "learning_rate": 6.659591509457125e-07,
+      "loss": 0.0763,
+      "step": 557
+    },
+    {
+      "epoch": 1.6706586826347305,
+      "grad_norm": 1.483075316617874,
+      "learning_rate": 6.542816058985896e-07,
+      "loss": 0.0521,
+      "step": 558
+    },
+    {
+      "epoch": 1.6736526946107784,
+      "grad_norm": 1.6332101703050352,
+      "learning_rate": 6.427001794876974e-07,
+      "loss": 0.0653,
+      "step": 559
+    },
+    {
+      "epoch": 1.6766467065868262,
+      "grad_norm": 1.4442167325958006,
+      "learning_rate": 6.312151278711237e-07,
+      "loss": 0.0485,
+      "step": 560
+    },
+    {
+      "epoch": 1.6796407185628741,
+      "grad_norm": 1.514813612775749,
+      "learning_rate": 6.198267050753387e-07,
+      "loss": 0.0554,
+      "step": 561
+    },
+    {
+      "epoch": 1.6826347305389222,
+      "grad_norm": 1.6382022048089737,
+      "learning_rate": 6.085351629895736e-07,
+      "loss": 0.0603,
+      "step": 562
+    },
+    {
+      "epoch": 1.68562874251497,
+      "grad_norm": 1.5714022576441489,
+      "learning_rate": 5.973407513602514e-07,
+      "loss": 0.0544,
+      "step": 563
+    },
+    {
+      "epoch": 1.688622754491018,
+      "grad_norm": 1.8055501858109513,
+      "learning_rate": 5.862437177854629e-07,
+      "loss": 0.0717,
+      "step": 564
+    },
+    {
+      "epoch": 1.6916167664670658,
+      "grad_norm": 1.608256899704192,
+      "learning_rate": 5.752443077094927e-07,
+      "loss": 0.0574,
+      "step": 565
+    },
+    {
+      "epoch": 1.694610778443114,
+      "grad_norm": 1.4995644661849035,
+      "learning_rate": 5.643427644173838e-07,
+      "loss": 0.0513,
+      "step": 566
+    },
+    {
+      "epoch": 1.6976047904191618,
+      "grad_norm": 1.6831528283313466,
+      "learning_rate": 5.535393290295643e-07,
+      "loss": 0.0629,
+      "step": 567
+    },
+    {
+      "epoch": 1.7005988023952097,
+      "grad_norm": 1.3734123044522035,
+      "learning_rate": 5.428342404965076e-07,
+      "loss": 0.0526,
+      "step": 568
+    },
+    {
+      "epoch": 1.7035928143712575,
+      "grad_norm": 1.4903440673739854,
+      "learning_rate": 5.322277355934557e-07,
+      "loss": 0.0547,
+      "step": 569
+    },
+    {
+      "epoch": 1.7065868263473054,
+      "grad_norm": 1.257226268391359,
+      "learning_rate": 5.217200489151714e-07,
+      "loss": 0.0455,
+      "step": 570
+    },
+    {
+      "epoch": 1.7095808383233533,
+      "grad_norm": 1.3605821816369508,
+      "learning_rate": 5.113114128707592e-07,
+      "loss": 0.0533,
+      "step": 571
+    },
+    {
+      "epoch": 1.7125748502994012,
+      "grad_norm": 1.7276287914519548,
+      "learning_rate": 5.010020576785174e-07,
+      "loss": 0.061,
+      "step": 572
+    },
+    {
+      "epoch": 1.715568862275449,
+      "grad_norm": 1.6012933524777972,
+      "learning_rate": 4.907922113608532e-07,
+      "loss": 0.0588,
+      "step": 573
+    },
+    {
+      "epoch": 1.718562874251497,
+      "grad_norm": 1.8702103539983497,
+      "learning_rate": 4.806820997392325e-07,
+      "loss": 0.0707,
+      "step": 574
+    },
+    {
+      "epoch": 1.7215568862275448,
+      "grad_norm": 1.5639881283678851,
+      "learning_rate": 4.7067194642919036e-07,
+      "loss": 0.0549,
+      "step": 575
+    },
+    {
+      "epoch": 1.7245508982035929,
+      "grad_norm": 1.6568223855345963,
+      "learning_rate": 4.607619728353818e-07,
+      "loss": 0.0579,
+      "step": 576
+    },
+    {
+      "epoch": 1.7275449101796407,
+      "grad_norm": 1.3924708342660808,
+      "learning_rate": 4.50952398146689e-07,
+      "loss": 0.0513,
+      "step": 577
+    },
+    {
+      "epoch": 1.7305389221556886,
+      "grad_norm": 1.3497858725454825,
+      "learning_rate": 4.4124343933136525e-07,
+      "loss": 0.0463,
+      "step": 578
+    },
+    {
+      "epoch": 1.7335329341317365,
+      "grad_norm": 1.6751999199017242,
+      "learning_rate": 4.3163531113224466e-07,
+      "loss": 0.0632,
+      "step": 579
+    },
+    {
+      "epoch": 1.7365269461077846,
+      "grad_norm": 2.0164349380544118,
+      "learning_rate": 4.221282260619891e-07,
+      "loss": 0.0777,
+      "step": 580
+    },
+    {
+      "epoch": 1.7395209580838324,
+      "grad_norm": 1.4684191296833098,
+      "learning_rate": 4.127223943983849e-07,
+      "loss": 0.0505,
+      "step": 581
+    },
+    {
+      "epoch": 1.7425149700598803,
+      "grad_norm": 1.6091853946582984,
+      "learning_rate": 4.03418024179697e-07,
+      "loss": 0.0545,
+      "step": 582
+    },
+    {
+      "epoch": 1.7455089820359282,
+      "grad_norm": 1.7103743105282496,
+      "learning_rate": 3.9421532120006544e-07,
+      "loss": 0.0722,
+      "step": 583
+    },
+    {
+      "epoch": 1.748502994011976,
+      "grad_norm": 1.427598042133789,
+      "learning_rate": 3.851144890049535e-07,
+      "loss": 0.0536,
+      "step": 584
+    },
+    {
+      "epoch": 1.751497005988024,
+      "grad_norm": 1.3356408381228415,
+      "learning_rate": 3.761157288866418e-07,
+      "loss": 0.055,
+      "step": 585
+    },
+    {
+      "epoch": 1.7544910179640718,
+      "grad_norm": 1.3939578983440781,
+      "learning_rate": 3.672192398797858e-07,
+      "loss": 0.0476,
+      "step": 586
+    },
+    {
+      "epoch": 1.7574850299401197,
+      "grad_norm": 1.8005851973320695,
+      "learning_rate": 3.58425218757002e-07,
+      "loss": 0.0582,
+      "step": 587
+    },
+    {
+      "epoch": 1.7604790419161676,
+      "grad_norm": 1.632372481728219,
+      "learning_rate": 3.497338600245254e-07,
+      "loss": 0.0549,
+      "step": 588
+    },
+    {
+      "epoch": 1.7634730538922154,
+      "grad_norm": 1.5473578192489055,
+      "learning_rate": 3.4114535591790233e-07,
+      "loss": 0.0575,
+      "step": 589
+    },
+    {
+      "epoch": 1.7664670658682635,
+      "grad_norm": 1.7149284898171917,
+      "learning_rate": 3.326598963977395e-07,
+      "loss": 0.0508,
+      "step": 590
+    },
+    {
+      "epoch": 1.7694610778443114,
+      "grad_norm": 1.6395651857406088,
+      "learning_rate": 3.242776691455013e-07,
+      "loss": 0.0608,
+      "step": 591
+    },
+    {
+      "epoch": 1.7724550898203593,
+      "grad_norm": 1.4938451494811882,
+      "learning_rate": 3.159988595593616e-07,
+      "loss": 0.0535,
+      "step": 592
+    },
+    {
+      "epoch": 1.7754491017964071,
+      "grad_norm": 2.04134320009435,
+      "learning_rate": 3.078236507501015e-07,
+      "loss": 0.0586,
+      "step": 593
+    },
+    {
+      "epoch": 1.7784431137724552,
+      "grad_norm": 1.5009690428650782,
+      "learning_rate": 2.9975222353705757e-07,
+      "loss": 0.0612,
+      "step": 594
+    },
+    {
+      "epoch": 1.781437125748503,
+      "grad_norm": 1.7393120503070636,
+      "learning_rate": 2.917847564441256e-07,
+      "loss": 0.064,
+      "step": 595
+    },
+    {
+      "epoch": 1.784431137724551,
+      "grad_norm": 1.5191902309825036,
+      "learning_rate": 2.839214256958106e-07,
+      "loss": 0.0649,
+      "step": 596
+    },
+    {
+      "epoch": 1.7874251497005988,
+      "grad_norm": 1.6503437822779496,
+      "learning_rate": 2.7616240521332884e-07,
+      "loss": 0.0599,
+      "step": 597
+    },
+    {
+      "epoch": 1.7904191616766467,
+      "grad_norm": 1.886551676901496,
+      "learning_rate": 2.6850786661076047e-07,
+      "loss": 0.0611,
+      "step": 598
+    },
+    {
+      "epoch": 1.7934131736526946,
+      "grad_norm": 1.5576694502333521,
+      "learning_rate": 2.6095797919125533e-07,
+      "loss": 0.0507,
+      "step": 599
+    },
+    {
+      "epoch": 1.7964071856287425,
+      "grad_norm": 1.6671309194647306,
+      "learning_rate": 2.5351290994328703e-07,
+      "loss": 0.0592,
+      "step": 600
+    },
+    {
+      "epoch": 1.7964071856287425,
+      "eval_loss": 0.2049744874238968,
+      "eval_runtime": 8.2236,
+      "eval_samples_per_second": 6.566,
+      "eval_steps_per_second": 1.702,
+      "step": 600
+    },
+    {
+      "epoch": 1.7994011976047903,
+      "grad_norm": 1.42046737003675,
+      "learning_rate": 2.4617282353696093e-07,
+      "loss": 0.0629,
+      "step": 601
+    },
+    {
+      "epoch": 1.8023952095808382,
+      "grad_norm": 1.9601183051624707,
+      "learning_rate": 2.3893788232036807e-07,
+      "loss": 0.068,
+      "step": 602
+    },
+    {
+      "epoch": 1.805389221556886,
+      "grad_norm": 1.4968907093062147,
+      "learning_rate": 2.318082463160032e-07,
+      "loss": 0.0534,
+      "step": 603
+    },
+    {
+      "epoch": 1.8083832335329342,
+      "grad_norm": 1.3461659295069583,
+      "learning_rate": 2.2478407321721295e-07,
+      "loss": 0.0498,
+      "step": 604
+    },
+    {
+      "epoch": 1.811377245508982,
+      "grad_norm": 1.6749244971081318,
+      "learning_rate": 2.1786551838471892e-07,
+      "loss": 0.0689,
+      "step": 605
+    },
+    {
+      "epoch": 1.81437125748503,
+      "grad_norm": 1.4810755450092639,
+      "learning_rate": 2.1105273484317402e-07,
+      "loss": 0.0515,
+      "step": 606
+    },
+    {
+      "epoch": 1.8173652694610778,
+      "grad_norm": 1.4449149226798588,
+      "learning_rate": 2.043458732777831e-07,
+      "loss": 0.062,
+      "step": 607
+    },
+    {
+      "epoch": 1.8203592814371259,
+      "grad_norm": 1.8025238520031246,
+      "learning_rate": 1.9774508203096843e-07,
+      "loss": 0.0663,
+      "step": 608
+    },
+    {
+      "epoch": 1.8233532934131738,
+      "grad_norm": 1.2273228667520284,
+      "learning_rate": 1.9125050709908388e-07,
+      "loss": 0.0435,
+      "step": 609
+    },
+    {
+      "epoch": 1.8263473053892216,
+      "grad_norm": 1.4499924954550212,
+      "learning_rate": 1.8486229212919482e-07,
+      "loss": 0.0534,
+      "step": 610
+    },
+    {
+      "epoch": 1.8293413173652695,
+      "grad_norm": 1.465773836541318,
+      "learning_rate": 1.7858057841589281e-07,
+      "loss": 0.055,
+      "step": 611
+    },
+    {
+      "epoch": 1.8323353293413174,
+      "grad_norm": 1.304323092343546,
+      "learning_rate": 1.7240550489817652e-07,
+      "loss": 0.0444,
+      "step": 612
+    },
+    {
+      "epoch": 1.8353293413173652,
+      "grad_norm": 1.6460706219702073,
+      "learning_rate": 1.66337208156373e-07,
+      "loss": 0.0508,
+      "step": 613
+    },
+    {
+      "epoch": 1.8383233532934131,
+      "grad_norm": 2.0206845247000342,
+      "learning_rate": 1.6037582240912175e-07,
+      "loss": 0.0713,
+      "step": 614
+    },
+    {
+      "epoch": 1.841317365269461,
+      "grad_norm": 1.7297996523540646,
+      "learning_rate": 1.5452147951040165e-07,
+      "loss": 0.0699,
+      "step": 615
+    },
+    {
+      "epoch": 1.8443113772455089,
+      "grad_norm": 1.6902179062955793,
+      "learning_rate": 1.4877430894662037e-07,
+      "loss": 0.0653,
+      "step": 616
+    },
+    {
+      "epoch": 1.8473053892215567,
+      "grad_norm": 1.4180847333764885,
+      "learning_rate": 1.4313443783374405e-07,
+      "loss": 0.0558,
+      "step": 617
+    },
+    {
+      "epoch": 1.8502994011976048,
+      "grad_norm": 1.613880574947737,
+      "learning_rate": 1.3760199091449045e-07,
+      "loss": 0.0518,
+      "step": 618
+    },
+    {
+      "epoch": 1.8532934131736527,
+      "grad_norm": 1.5222918132810281,
+      "learning_rate": 1.3217709055556638e-07,
+      "loss": 0.0615,
+      "step": 619
+    },
+    {
+      "epoch": 1.8562874251497006,
+      "grad_norm": 1.7537919129636534,
+      "learning_rate": 1.268598567449647e-07,
+      "loss": 0.0639,
+      "step": 620
+    },
+    {
+      "epoch": 1.8592814371257484,
+      "grad_norm": 1.5643181124610306,
+      "learning_rate": 1.2165040708930763e-07,
+      "loss": 0.0605,
+      "step": 621
+    },
+    {
+      "epoch": 1.8622754491017965,
+      "grad_norm": 1.2338770998129176,
+      "learning_rate": 1.1654885681124661e-07,
+      "loss": 0.0495,
+      "step": 622
+    },
+    {
+      "epoch": 1.8652694610778444,
+      "grad_norm": 1.6192141590352853,
+      "learning_rate": 1.1155531874691372e-07,
+      "loss": 0.0582,
+      "step": 623
+    },
+    {
+      "epoch": 1.8682634730538923,
+      "grad_norm": 1.4548372115239796,
+      "learning_rate": 1.0666990334342708e-07,
+      "loss": 0.0523,
+      "step": 624
+    },
+    {
+      "epoch": 1.8712574850299402,
+      "grad_norm": 1.6290907465563864,
+      "learning_rate": 1.0189271865644445e-07,
+      "loss": 0.0516,
+      "step": 625
+    },
+    {
+      "epoch": 1.874251497005988,
+      "grad_norm": 1.8971382386168782,
+      "learning_rate": 9.722387034777847e-08,
+      "loss": 0.0636,
+      "step": 626
+    },
+    {
+      "epoch": 1.877245508982036,
+      "grad_norm": 1.8259464193024688,
+      "learning_rate": 9.266346168305518e-08,
+      "loss": 0.0651,
+      "step": 627
+    },
+    {
+      "epoch": 1.8802395209580838,
+      "grad_norm": 1.3645764319651372,
+      "learning_rate": 8.821159352943142e-08,
+      "loss": 0.0501,
+      "step": 628
+    },
+    {
+      "epoch": 1.8832335329341316,
+      "grad_norm": 1.3245565924784979,
+      "learning_rate": 8.38683643533661e-08,
+      "loss": 0.0432,
+      "step": 629
+    },
+    {
+      "epoch": 1.8862275449101795,
+      "grad_norm": 1.638277284307031,
+      "learning_rate": 7.963387021843683e-08,
+      "loss": 0.0529,
+      "step": 630
+    },
+    {
+      "epoch": 1.8892215568862274,
+      "grad_norm": 1.681646470464429,
+      "learning_rate": 7.550820478322285e-08,
+      "loss": 0.0648,
+      "step": 631
+    },
+    {
+      "epoch": 1.8922155688622755,
+      "grad_norm": 1.5314514593041408,
+      "learning_rate": 7.149145929922607e-08,
+      "loss": 0.0695,
+      "step": 632
+    },
+    {
+      "epoch": 1.8952095808383234,
+      "grad_norm": 1.8459088063539417,
+      "learning_rate": 6.758372260885714e-08,
+      "loss": 0.0701,
+      "step": 633
+    },
+    {
+      "epoch": 1.8982035928143712,
+      "grad_norm": 1.8817410292601389,
+      "learning_rate": 6.378508114346982e-08,
+      "loss": 0.072,
+      "step": 634
+    },
+    {
+      "epoch": 1.9011976047904193,
+      "grad_norm": 1.3344188445796707,
+      "learning_rate": 6.009561892144744e-08,
+      "loss": 0.0477,
+      "step": 635
+    },
+    {
+      "epoch": 1.9041916167664672,
+      "grad_norm": 1.40612017046323,
+      "learning_rate": 5.651541754634726e-08,
+      "loss": 0.0518,
+      "step": 636
+    },
+    {
+      "epoch": 1.907185628742515,
+      "grad_norm": 1.5489439449823093,
+      "learning_rate": 5.304455620509297e-08,
+      "loss": 0.0604,
+      "step": 637
+    },
+    {
+      "epoch": 1.910179640718563,
+      "grad_norm": 1.4362223214265843,
+      "learning_rate": 4.968311166622553e-08,
+      "loss": 0.0556,
+      "step": 638
+    },
+    {
+      "epoch": 1.9131736526946108,
+      "grad_norm": 1.6139689467696063,
+      "learning_rate": 4.643115827820399e-08,
+      "loss": 0.0584,
+      "step": 639
+    },
+    {
+      "epoch": 1.9161676646706587,
+      "grad_norm": 1.8740951658995462,
+      "learning_rate": 4.328876796776071e-08,
+      "loss": 0.0552,
+      "step": 640
+    },
+    {
+      "epoch": 1.9191616766467066,
+      "grad_norm": 1.4020102008899353,
+      "learning_rate": 4.0256010238310936e-08,
+      "loss": 0.0542,
+      "step": 641
+    },
+    {
+      "epoch": 1.9221556886227544,
+      "grad_norm": 1.6961109456474293,
+      "learning_rate": 3.733295216841626e-08,
+      "loss": 0.0649,
+      "step": 642
+    },
+    {
+      "epoch": 1.9251497005988023,
+      "grad_norm": 2.010152362901947,
+      "learning_rate": 3.451965841029914e-08,
+      "loss": 0.0605,
+      "step": 643
+    },
+    {
+      "epoch": 1.9281437125748502,
+      "grad_norm": 1.5148530579152966,
+      "learning_rate": 3.181619118841517e-08,
+      "loss": 0.0517,
+      "step": 644
+    },
+    {
+      "epoch": 1.931137724550898,
+      "grad_norm": 1.4448653465366044,
+      "learning_rate": 2.9222610298074717e-08,
+      "loss": 0.0462,
+      "step": 645
+    },
+    {
+      "epoch": 1.9341317365269461,
+      "grad_norm": 1.5973742580021506,
+      "learning_rate": 2.673897310412288e-08,
+      "loss": 0.0596,
+      "step": 646
+    },
+    {
+      "epoch": 1.937125748502994,
+      "grad_norm": 1.4296532769921837,
+      "learning_rate": 2.4365334539667717e-08,
+      "loss": 0.0494,
+      "step": 647
+    },
+    {
+      "epoch": 1.9401197604790419,
+      "grad_norm": 1.5941868347924208,
+      "learning_rate": 2.210174710486679e-08,
+      "loss": 0.0654,
+      "step": 648
+    },
+    {
+      "epoch": 1.94311377245509,
+      "grad_norm": 1.1181196705839183,
+      "learning_rate": 1.99482608657664e-08,
+      "loss": 0.0429,
+      "step": 649
+    },
+    {
+      "epoch": 1.9461077844311379,
+      "grad_norm": 1.4470443433272222,
+      "learning_rate": 1.7904923453193056e-08,
+      "loss": 0.0559,
+      "step": 650
+    },
+    {
+      "epoch": 1.9491017964071857,
+      "grad_norm": 1.5345978442798223,
+      "learning_rate": 1.5971780061701524e-08,
+      "loss": 0.0521,
+      "step": 651
+    },
+    {
+      "epoch": 1.9520958083832336,
+      "grad_norm": 1.47348560568937,
+      "learning_rate": 1.4148873448573408e-08,
+      "loss": 0.0519,
+      "step": 652
+    },
+    {
+      "epoch": 1.9550898203592815,
+      "grad_norm": 1.6046762221186293,
+      "learning_rate": 1.2436243932872349e-08,
+      "loss": 0.066,
+      "step": 653
+    },
+    {
+      "epoch": 1.9580838323353293,
+      "grad_norm": 1.5430542959251832,
+      "learning_rate": 1.0833929394552523e-08,
+      "loss": 0.0494,
+      "step": 654
+    },
+    {
+      "epoch": 1.9610778443113772,
+      "grad_norm": 1.5220196570983686,
+      "learning_rate": 9.341965273621522e-09,
+      "loss": 0.0503,
+      "step": 655
+    },
+    {
+      "epoch": 1.964071856287425,
+      "grad_norm": 1.7773430508356551,
+      "learning_rate": 7.96038456935322e-09,
+      "loss": 0.0651,
+      "step": 656
+    },
+    {
+      "epoch": 1.967065868263473,
+      "grad_norm": 1.5939425569567203,
+      "learning_rate": 6.6892178395611125e-09,
+      "loss": 0.0595,
+      "step": 657
+    },
+    {
+      "epoch": 1.9700598802395208,
+      "grad_norm": 1.554632242573711,
+      "learning_rate": 5.528493199922769e-09,
+      "loss": 0.0627,
+      "step": 658
+    },
+    {
+      "epoch": 1.9730538922155687,
+      "grad_norm": 1.5709006723057153,
+      "learning_rate": 4.478236323355312e-09,
+      "loss": 0.0611,
+      "step": 659
+    },
+    {
+      "epoch": 1.9760479041916168,
+      "grad_norm": 1.5584355086710606,
+      "learning_rate": 3.538470439448105e-09,
+      "loss": 0.0535,
+      "step": 660
+    },
+    {
+      "epoch": 1.9790419161676647,
+      "grad_norm": 1.687008625880879,
+      "learning_rate": 2.709216333952602e-09,
+      "loss": 0.0657,
+      "step": 661
+    },
+    {
+      "epoch": 1.9820359281437125,
+      "grad_norm": 1.3448086229610081,
+      "learning_rate": 1.9904923483171632e-09,
+      "loss": 0.0553,
+      "step": 662
+    },
+    {
+      "epoch": 1.9850299401197606,
+      "grad_norm": 1.8818836222312525,
+      "learning_rate": 1.3823143792851545e-09,
+      "loss": 0.0673,
+      "step": 663
+    },
+    {
+      "epoch": 1.9880239520958085,
+      "grad_norm": 1.821735153978393,
+      "learning_rate": 8.846958785418969e-10,
+      "loss": 0.0699,
+      "step": 664
+    },
+    {
+      "epoch": 1.9910179640718564,
+      "grad_norm": 1.4509044498946764,
+      "learning_rate": 4.97647852417682e-10,
+      "loss": 0.0534,
+      "step": 665
+    },
+    {
+      "epoch": 1.9940119760479043,
+      "grad_norm": 1.437951915262999,
+      "learning_rate": 2.2117886164407797e-10,
+      "loss": 0.0519,
+      "step": 666
+    },
+    {
+      "epoch": 1.9970059880239521,
+      "grad_norm": 1.4081653127767364,
+      "learning_rate": 5.529502116519148e-11,
+      "loss": 0.0505,
+      "step": 667
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.0092826618782436,
+      "learning_rate": 0.0,
+      "loss": 0.0238,
+      "step": 668
+    },
+    {
+      "epoch": 2.0,
+      "step": 668,
+      "total_flos": 12704325304320.0,
+      "train_loss": 0.13785797523747303,
+      "train_runtime": 1825.1094,
+      "train_samples_per_second": 5.84,
+      "train_steps_per_second": 0.366
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 668,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 2000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 12704325304320.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}