diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5046 @@
+{
+  "best_metric": 0.7773267737033928,
+  "best_model_checkpoint": "models/combined_dist_10.0-esm2_t6_8M_UR50D-finetuned-receptor_pred_cluspro_propedia_pepnn/checkpoint-282331",
+  "epoch": 9.999893741742849,
+  "eval_steps": 500,
+  "global_step": 352910,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.614975690841675,
+      "learning_rate": 1.997166416366779e-05,
+      "loss": 7.4545,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.9494409561157227,
+      "learning_rate": 1.9943328327335584e-05,
+      "loss": 6.7082,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.9211537837982178,
+      "learning_rate": 1.9914992491003373e-05,
+      "loss": 6.2699,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.27163553237915,
+      "learning_rate": 1.9886656654671162e-05,
+      "loss": 6.0185,
+      "step": 2000
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.343883991241455,
+      "learning_rate": 1.9858320818338955e-05,
+      "loss": 5.8031,
+      "step": 2500
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.585379123687744,
+      "learning_rate": 1.9829984982006744e-05,
+      "loss": 5.6227,
+      "step": 3000
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.197877883911133,
+      "learning_rate": 1.9801649145674537e-05,
+      "loss": 5.4341,
+      "step": 3500
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 5.071269512176514,
+      "learning_rate": 1.977331330934233e-05,
+      "loss": 5.282,
+      "step": 4000
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.63652229309082,
+      "learning_rate": 1.974497747301012e-05,
+      "loss": 5.0468,
+      "step": 4500
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.155987739562988,
+      "learning_rate": 1.9716641636677908e-05,
+      "loss": 4.8222,
+      "step": 5000
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.847218036651611,
+      "learning_rate": 1.96883058003457e-05,
+      "loss": 4.7802,
+      "step": 5500
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 4.746351718902588,
+      "learning_rate": 1.965996996401349e-05,
+      "loss": 4.5951,
+      "step": 6000
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.8343939781188965,
+      "learning_rate": 1.9631634127681282e-05,
+      "loss": 4.4475,
+      "step": 6500
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.076068878173828,
+      "learning_rate": 1.960329829134907e-05,
+      "loss": 4.266,
+      "step": 7000
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.7831315994262695,
+      "learning_rate": 1.957496245501686e-05,
+      "loss": 4.1708,
+      "step": 7500
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 4.758873462677002,
+      "learning_rate": 1.9546626618684653e-05,
+      "loss": 4.0085,
+      "step": 8000
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.676722049713135,
+      "learning_rate": 1.9518290782352443e-05,
+      "loss": 3.7978,
+      "step": 8500
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 6.958544731140137,
+      "learning_rate": 1.9489954946020232e-05,
+      "loss": 3.7165,
+      "step": 9000
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 6.333941459655762,
+      "learning_rate": 1.9461619109688024e-05,
+      "loss": 3.629,
+      "step": 9500
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 6.761513710021973,
+      "learning_rate": 1.9433283273355814e-05,
+      "loss": 3.4656,
+      "step": 10000
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 5.946700096130371,
+      "learning_rate": 1.9404947437023603e-05,
+      "loss": 3.3146,
+      "step": 10500
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 6.658429145812988,
+      "learning_rate": 1.9376611600691396e-05,
+      "loss": 3.2532,
+      "step": 11000
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 5.190903663635254,
+      "learning_rate": 1.9348275764359185e-05,
+      "loss": 3.1859,
+      "step": 11500
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 4.14591646194458,
+      "learning_rate": 1.9319939928026977e-05,
+      "loss": 2.9643,
+      "step": 12000
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 4.31158447265625,
+      "learning_rate": 1.929160409169477e-05,
+      "loss": 2.9481,
+      "step": 12500
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.342416763305664,
+      "learning_rate": 1.926326825536256e-05,
+      "loss": 2.8739,
+      "step": 13000
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 4.94139289855957,
+      "learning_rate": 1.923493241903035e-05,
+      "loss": 2.7749,
+      "step": 13500
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 3.6980950832366943,
+      "learning_rate": 1.920659658269814e-05,
+      "loss": 2.5832,
+      "step": 14000
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 4.225959777832031,
+      "learning_rate": 1.917826074636593e-05,
+      "loss": 2.6088,
+      "step": 14500
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 4.793159484863281,
+      "learning_rate": 1.9149924910033723e-05,
+      "loss": 2.5069,
+      "step": 15000
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.690148830413818,
+      "learning_rate": 1.9121589073701512e-05,
+      "loss": 2.4667,
+      "step": 15500
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.850300312042236,
+      "learning_rate": 1.90932532373693e-05,
+      "loss": 2.3974,
+      "step": 16000
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.971298933029175,
+      "learning_rate": 1.9064917401037094e-05,
+      "loss": 2.3625,
+      "step": 16500
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.971684217453003,
+      "learning_rate": 1.9036581564704883e-05,
+      "loss": 2.3342,
+      "step": 17000
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 4.61134147644043,
+      "learning_rate": 1.9008245728372673e-05,
+      "loss": 2.2263,
+      "step": 17500
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.9258928298950195,
+      "learning_rate": 1.8979909892040465e-05,
+      "loss": 2.2668,
+      "step": 18000
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 3.6153666973114014,
+      "learning_rate": 1.8951574055708255e-05,
+      "loss": 2.1847,
+      "step": 18500
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 3.767505645751953,
+      "learning_rate": 1.8923238219376044e-05,
+      "loss": 2.1653,
+      "step": 19000
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.658140659332275,
+      "learning_rate": 1.8894902383043836e-05,
+      "loss": 2.1216,
+      "step": 19500
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 5.809523582458496,
+      "learning_rate": 1.886656654671163e-05,
+      "loss": 2.1138,
+      "step": 20000
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.878647327423096,
+      "learning_rate": 1.883823071037942e-05,
+      "loss": 2.0043,
+      "step": 20500
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 4.507823467254639,
+      "learning_rate": 1.880989487404721e-05,
+      "loss": 1.9965,
+      "step": 21000
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 4.389089107513428,
+      "learning_rate": 1.8781559037715e-05,
+      "loss": 1.9863,
+      "step": 21500
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.876653671264648,
+      "learning_rate": 1.875322320138279e-05,
+      "loss": 1.9836,
+      "step": 22000
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.195200443267822,
+      "learning_rate": 1.8724887365050582e-05,
+      "loss": 1.9457,
+      "step": 22500
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.636641502380371,
+      "learning_rate": 1.869655152871837e-05,
+      "loss": 1.9271,
+      "step": 23000
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.557649850845337,
+      "learning_rate": 1.8668215692386164e-05,
+      "loss": 1.8922,
+      "step": 23500
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.440142631530762,
+      "learning_rate": 1.8639879856053953e-05,
+      "loss": 1.8782,
+      "step": 24000
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.779470920562744,
+      "learning_rate": 1.8611544019721743e-05,
+      "loss": 1.8265,
+      "step": 24500
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.1893739700317383,
+      "learning_rate": 1.8583208183389535e-05,
+      "loss": 1.8315,
+      "step": 25000
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.712510585784912,
+      "learning_rate": 1.8554872347057324e-05,
+      "loss": 1.8203,
+      "step": 25500
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.149935722351074,
+      "learning_rate": 1.8526536510725114e-05,
+      "loss": 1.8559,
+      "step": 26000
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 6.496318817138672,
+      "learning_rate": 1.8498200674392906e-05,
+      "loss": 1.7193,
+      "step": 26500
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 9.580072402954102,
+      "learning_rate": 1.8469864838060696e-05,
+      "loss": 1.7393,
+      "step": 27000
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.071699619293213,
+      "learning_rate": 1.8441529001728488e-05,
+      "loss": 1.7212,
+      "step": 27500
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 6.9725470542907715,
+      "learning_rate": 1.841319316539628e-05,
+      "loss": 1.7477,
+      "step": 28000
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 3.311365842819214,
+      "learning_rate": 1.838485732906407e-05,
+      "loss": 1.6702,
+      "step": 28500
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 3.8417956829071045,
+      "learning_rate": 1.835652149273186e-05,
+      "loss": 1.6546,
+      "step": 29000
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.032675266265869,
+      "learning_rate": 1.8328185656399652e-05,
+      "loss": 1.6497,
+      "step": 29500
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 7.1385064125061035,
+      "learning_rate": 1.829984982006744e-05,
+      "loss": 1.6649,
+      "step": 30000
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.662952423095703,
+      "learning_rate": 1.827151398373523e-05,
+      "loss": 1.6297,
+      "step": 30500
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 6.526814937591553,
+      "learning_rate": 1.8243178147403023e-05,
+      "loss": 1.5955,
+      "step": 31000
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 7.363564491271973,
+      "learning_rate": 1.8214842311070812e-05,
+      "loss": 1.6026,
+      "step": 31500
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.3690953254699707,
+      "learning_rate": 1.8186506474738605e-05,
+      "loss": 1.5352,
+      "step": 32000
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 7.161606311798096,
+      "learning_rate": 1.8158170638406394e-05,
+      "loss": 1.6161,
+      "step": 32500
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.376026153564453,
+      "learning_rate": 1.8129834802074183e-05,
+      "loss": 1.6121,
+      "step": 33000
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 3.4227709770202637,
+      "learning_rate": 1.8101498965741976e-05,
+      "loss": 1.4918,
+      "step": 33500
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 6.050368785858154,
+      "learning_rate": 1.8073163129409765e-05,
+      "loss": 1.4924,
+      "step": 34000
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 8.424703598022461,
+      "learning_rate": 1.8044827293077555e-05,
+      "loss": 1.5637,
+      "step": 34500
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.532356262207031,
+      "learning_rate": 1.8016491456745347e-05,
+      "loss": 1.4816,
+      "step": 35000
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.6981436813974987,
+      "eval_loss": 1.491289734840393,
+      "eval_runtime": 2747.372,
+      "eval_samples_per_second": 34.255,
+      "eval_steps_per_second": 34.255,
+      "step": 35291
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 3.4316861629486084,
+      "learning_rate": 1.7988155620413136e-05,
+      "loss": 1.5285,
+      "step": 35500
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 7.24040412902832,
+      "learning_rate": 1.795981978408093e-05,
+      "loss": 1.4974,
+      "step": 36000
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 3.9627156257629395,
+      "learning_rate": 1.7931483947748722e-05,
+      "loss": 1.5143,
+      "step": 36500
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 5.244486331939697,
+      "learning_rate": 1.790314811141651e-05,
+      "loss": 1.4756,
+      "step": 37000
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 6.913219928741455,
+      "learning_rate": 1.78748122750843e-05,
+      "loss": 1.4885,
+      "step": 37500
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 7.830247402191162,
+      "learning_rate": 1.7846476438752093e-05,
+      "loss": 1.4262,
+      "step": 38000
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 4.569477081298828,
+      "learning_rate": 1.7818140602419882e-05,
+      "loss": 1.4444,
+      "step": 38500
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 7.476200103759766,
+      "learning_rate": 1.778980476608767e-05,
+      "loss": 1.4168,
+      "step": 39000
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 5.149771213531494,
+      "learning_rate": 1.7761468929755464e-05,
+      "loss": 1.44,
+      "step": 39500
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 6.547873020172119,
+      "learning_rate": 1.7733133093423253e-05,
+      "loss": 1.3884,
+      "step": 40000
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 5.382103443145752,
+      "learning_rate": 1.7704797257091046e-05,
+      "loss": 1.3833,
+      "step": 40500
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 5.518494606018066,
+      "learning_rate": 1.7676461420758835e-05,
+      "loss": 1.4183,
+      "step": 41000
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 6.328868389129639,
+      "learning_rate": 1.7648125584426624e-05,
+      "loss": 1.4387,
+      "step": 41500
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 3.008655548095703,
+      "learning_rate": 1.7619789748094417e-05,
+      "loss": 1.3613,
+      "step": 42000
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 6.35288667678833,
+      "learning_rate": 1.7591453911762206e-05,
+      "loss": 1.3565,
+      "step": 42500
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 6.100168704986572,
+      "learning_rate": 1.7563118075429995e-05,
+      "loss": 1.3506,
+      "step": 43000
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 3.2157440185546875,
+      "learning_rate": 1.7534782239097788e-05,
+      "loss": 1.3395,
+      "step": 43500
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 2.379615306854248,
+      "learning_rate": 1.750644640276558e-05,
+      "loss": 1.3396,
+      "step": 44000
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 2.793736457824707,
+      "learning_rate": 1.747811056643337e-05,
+      "loss": 1.3319,
+      "step": 44500
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 6.41762113571167,
+      "learning_rate": 1.7449774730101163e-05,
+      "loss": 1.301,
+      "step": 45000
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 6.420482635498047,
+      "learning_rate": 1.7421438893768952e-05,
+      "loss": 1.3574,
+      "step": 45500
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 2.2033274173736572,
+      "learning_rate": 1.739310305743674e-05,
+      "loss": 1.316,
+      "step": 46000
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 5.3276801109313965,
+      "learning_rate": 1.7364767221104534e-05,
+      "loss": 1.3369,
+      "step": 46500
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 4.429421901702881,
+      "learning_rate": 1.7336431384772323e-05,
+      "loss": 1.3232,
+      "step": 47000
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 7.140500068664551,
+      "learning_rate": 1.7308095548440112e-05,
+      "loss": 1.3108,
+      "step": 47500
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 4.972752571105957,
+      "learning_rate": 1.7279759712107905e-05,
+      "loss": 1.2425,
+      "step": 48000
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 4.659143447875977,
+      "learning_rate": 1.7251423875775694e-05,
+      "loss": 1.2828,
+      "step": 48500
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 3.439525842666626,
+      "learning_rate": 1.7223088039443487e-05,
+      "loss": 1.2559,
+      "step": 49000
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 5.8029656410217285,
+      "learning_rate": 1.7194752203111276e-05,
+      "loss": 1.2982,
+      "step": 49500
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 4.9915771484375,
+      "learning_rate": 1.7166416366779065e-05,
+      "loss": 1.2627,
+      "step": 50000
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 5.220378875732422,
+      "learning_rate": 1.7138080530446858e-05,
+      "loss": 1.32,
+      "step": 50500
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 6.145462989807129,
+      "learning_rate": 1.7109744694114647e-05,
+      "loss": 1.2215,
+      "step": 51000
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 3.8394155502319336,
+      "learning_rate": 1.7081408857782436e-05,
+      "loss": 1.2521,
+      "step": 51500
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 4.981057643890381,
+      "learning_rate": 1.705307302145023e-05,
+      "loss": 1.2893,
+      "step": 52000
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.4461212158203125,
+      "learning_rate": 1.702473718511802e-05,
+      "loss": 1.3046,
+      "step": 52500
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.8721215724945068,
+      "learning_rate": 1.699640134878581e-05,
+      "loss": 1.2423,
+      "step": 53000
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 5.1794843673706055,
+      "learning_rate": 1.6968065512453604e-05,
+      "loss": 1.2604,
+      "step": 53500
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 4.042840480804443,
+      "learning_rate": 1.6939729676121393e-05,
+      "loss": 1.2357,
+      "step": 54000
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 5.862013816833496,
+      "learning_rate": 1.6911393839789182e-05,
+      "loss": 1.2256,
+      "step": 54500
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 3.3453152179718018,
+      "learning_rate": 1.6883058003456975e-05,
+      "loss": 1.1843,
+      "step": 55000
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 8.414984703063965,
+      "learning_rate": 1.6854722167124764e-05,
+      "loss": 1.2016,
+      "step": 55500
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 2.847531795501709,
+      "learning_rate": 1.6826386330792553e-05,
+      "loss": 1.1997,
+      "step": 56000
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 3.7932987213134766,
+      "learning_rate": 1.6798050494460346e-05,
+      "loss": 1.2276,
+      "step": 56500
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 4.8044562339782715,
+      "learning_rate": 1.6769714658128135e-05,
+      "loss": 1.1733,
+      "step": 57000
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 5.216464996337891,
+      "learning_rate": 1.6741378821795928e-05,
+      "loss": 1.1884,
+      "step": 57500
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 4.36829137802124,
+      "learning_rate": 1.6713042985463717e-05,
+      "loss": 1.1851,
+      "step": 58000
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 4.850276947021484,
+      "learning_rate": 1.6684707149131506e-05,
+      "loss": 1.1902,
+      "step": 58500
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 5.7681989669799805,
+      "learning_rate": 1.66563713127993e-05,
+      "loss": 1.1787,
+      "step": 59000
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 2.202500581741333,
+      "learning_rate": 1.662803547646709e-05,
+      "loss": 1.1648,
+      "step": 59500
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 3.1508541107177734,
+      "learning_rate": 1.659969964013488e-05,
+      "loss": 1.1421,
+      "step": 60000
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 7.434388637542725,
+      "learning_rate": 1.657136380380267e-05,
+      "loss": 1.174,
+      "step": 60500
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.1375887393951416,
+      "learning_rate": 1.6543027967470463e-05,
+      "loss": 1.1941,
+      "step": 61000
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 5.083194732666016,
+      "learning_rate": 1.6514692131138252e-05,
+      "loss": 1.1457,
+      "step": 61500
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 5.154632568359375,
+      "learning_rate": 1.6486356294806044e-05,
+      "loss": 1.2178,
+      "step": 62000
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 3.088094711303711,
+      "learning_rate": 1.6458020458473834e-05,
+      "loss": 1.1661,
+      "step": 62500
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 5.301969528198242,
+      "learning_rate": 1.6429684622141623e-05,
+      "loss": 1.119,
+      "step": 63000
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 4.429632186889648,
+      "learning_rate": 1.6401348785809416e-05,
+      "loss": 1.1533,
+      "step": 63500
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 4.935004234313965,
+      "learning_rate": 1.6373012949477205e-05,
+      "loss": 1.1492,
+      "step": 64000
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 4.975069999694824,
+      "learning_rate": 1.6344677113144994e-05,
+      "loss": 1.1214,
+      "step": 64500
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 2.3240180015563965,
+      "learning_rate": 1.6316341276812787e-05,
+      "loss": 1.2013,
+      "step": 65000
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 3.9896786212921143,
+      "learning_rate": 1.6288005440480576e-05,
+      "loss": 1.1176,
+      "step": 65500
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 5.830460071563721,
+      "learning_rate": 1.625966960414837e-05,
+      "loss": 1.117,
+      "step": 66000
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 5.111226558685303,
+      "learning_rate": 1.6231333767816158e-05,
+      "loss": 1.1321,
+      "step": 66500
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 7.154306411743164,
+      "learning_rate": 1.6202997931483947e-05,
+      "loss": 1.1416,
+      "step": 67000
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 2.1327767372131348,
+      "learning_rate": 1.617466209515174e-05,
+      "loss": 1.1103,
+      "step": 67500
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 4.014294624328613,
+      "learning_rate": 1.6146326258819532e-05,
+      "loss": 1.0615,
+      "step": 68000
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.03513578325510025,
+      "learning_rate": 1.611799042248732e-05,
+      "loss": 1.1205,
+      "step": 68500
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 2.805485725402832,
+      "learning_rate": 1.608965458615511e-05,
+      "loss": 1.136,
+      "step": 69000
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 3.611574172973633,
+      "learning_rate": 1.6061318749822904e-05,
+      "loss": 1.1159,
+      "step": 69500
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 5.143590450286865,
+      "learning_rate": 1.6032982913490693e-05,
+      "loss": 1.111,
+      "step": 70000
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 3.690364360809326,
+      "learning_rate": 1.6004647077158485e-05,
+      "loss": 1.1491,
+      "step": 70500
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.7476809299656788,
+      "eval_loss": 1.085469126701355,
+      "eval_runtime": 2745.5383,
+      "eval_samples_per_second": 34.278,
+      "eval_steps_per_second": 34.278,
+      "step": 70582
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 4.023772239685059,
+      "learning_rate": 1.5976311240826275e-05,
+      "loss": 1.055,
+      "step": 71000
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 3.253685474395752,
+      "learning_rate": 1.5947975404494064e-05,
+      "loss": 1.0495,
+      "step": 71500
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 4.396245956420898,
+      "learning_rate": 1.5919639568161857e-05,
+      "loss": 1.0747,
+      "step": 72000
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 8.438695907592773,
+      "learning_rate": 1.5891303731829646e-05,
+      "loss": 1.0484,
+      "step": 72500
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 4.8430914878845215,
+      "learning_rate": 1.5862967895497435e-05,
+      "loss": 1.1275,
+      "step": 73000
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 3.183419704437256,
+      "learning_rate": 1.5834632059165228e-05,
+      "loss": 1.0809,
+      "step": 73500
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 2.4836373329162598,
+      "learning_rate": 1.5806296222833017e-05,
+      "loss": 1.1175,
+      "step": 74000
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 5.866794109344482,
+      "learning_rate": 1.577796038650081e-05,
+      "loss": 1.0826,
+      "step": 74500
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 7.779801368713379,
+      "learning_rate": 1.57496245501686e-05,
+      "loss": 1.0051,
+      "step": 75000
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 7.138259410858154,
+      "learning_rate": 1.572128871383639e-05,
+      "loss": 1.0388,
+      "step": 75500
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 5.617405414581299,
+      "learning_rate": 1.569295287750418e-05,
+      "loss": 1.0332,
+      "step": 76000
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 3.4417684078216553,
+      "learning_rate": 1.5664617041171973e-05,
+      "loss": 1.1104,
+      "step": 76500
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 3.4162983894348145,
+      "learning_rate": 1.5636281204839763e-05,
+      "loss": 1.0992,
+      "step": 77000
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 8.070796966552734,
+      "learning_rate": 1.5607945368507552e-05,
+      "loss": 1.0486,
+      "step": 77500
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 4.65944242477417,
+      "learning_rate": 1.5579609532175344e-05,
+      "loss": 1.0532,
+      "step": 78000
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 7.3272385597229,
+      "learning_rate": 1.5551273695843134e-05,
+      "loss": 1.0081,
+      "step": 78500
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 2.9811854362487793,
+      "learning_rate": 1.5522937859510926e-05,
+      "loss": 1.0228,
+      "step": 79000
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 2.284489154815674,
+      "learning_rate": 1.5494602023178716e-05,
+      "loss": 1.0208,
+      "step": 79500
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 6.056519985198975,
+      "learning_rate": 1.5466266186846505e-05,
+      "loss": 1.032,
+      "step": 80000
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.048289719969034195,
+      "learning_rate": 1.5437930350514297e-05,
+      "loss": 1.0771,
+      "step": 80500
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 4.455281734466553,
+      "learning_rate": 1.5409594514182087e-05,
+      "loss": 1.0366,
+      "step": 81000
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 5.394782066345215,
+      "learning_rate": 1.5381258677849876e-05,
+      "loss": 1.0811,
+      "step": 81500
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 4.059425354003906,
+      "learning_rate": 1.535292284151767e-05,
+      "loss": 1.0072,
+      "step": 82000
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 5.580400466918945,
+      "learning_rate": 1.5324587005185458e-05,
+      "loss": 1.053,
+      "step": 82500
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 2.141561508178711,
+      "learning_rate": 1.529625116885325e-05,
+      "loss": 1.0425,
+      "step": 83000
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 2.186253547668457,
+      "learning_rate": 1.5267915332521043e-05,
+      "loss": 1.0205,
+      "step": 83500
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 3.237588405609131,
+      "learning_rate": 1.5239579496188832e-05,
+      "loss": 0.9956,
+      "step": 84000
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 4.83936071395874,
+      "learning_rate": 1.5211243659856623e-05,
+      "loss": 0.9885,
+      "step": 84500
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 5.182650089263916,
+      "learning_rate": 1.5182907823524413e-05,
+      "loss": 1.0748,
+      "step": 85000
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.2307519912719727,
+      "learning_rate": 1.5154571987192203e-05,
+      "loss": 0.9792,
+      "step": 85500
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 4.785423278808594,
+      "learning_rate": 1.5126236150859994e-05,
+      "loss": 1.0311,
+      "step": 86000
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 6.569101333618164,
+      "learning_rate": 1.5097900314527785e-05,
+      "loss": 0.9967,
+      "step": 86500
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 2.734511613845825,
+      "learning_rate": 1.5069564478195575e-05,
+      "loss": 1.0641,
+      "step": 87000
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 3.6517486572265625,
+      "learning_rate": 1.5041228641863366e-05,
+      "loss": 0.9817,
+      "step": 87500
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 2.1969916820526123,
+      "learning_rate": 1.5012892805531156e-05,
+      "loss": 0.9867,
+      "step": 88000
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 3.1304757595062256,
+      "learning_rate": 1.4984556969198947e-05,
+      "loss": 1.0322,
+      "step": 88500
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 3.666090965270996,
+      "learning_rate": 1.4956221132866737e-05,
+      "loss": 1.0294,
+      "step": 89000
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 6.77361536026001,
+      "learning_rate": 1.4927885296534528e-05,
+      "loss": 0.9871,
+      "step": 89500
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 5.582375526428223,
+      "learning_rate": 1.4899549460202319e-05,
+      "loss": 0.9809,
+      "step": 90000
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 5.1192827224731445,
+      "learning_rate": 1.487121362387011e-05,
+      "loss": 0.9964,
+      "step": 90500
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 3.1294760704040527,
+      "learning_rate": 1.4842877787537899e-05,
+      "loss": 1.0106,
+      "step": 91000
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 2.579064130783081,
+      "learning_rate": 1.4814541951205691e-05,
+      "loss": 0.9774,
+      "step": 91500
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 5.767348766326904,
+      "learning_rate": 1.4786206114873482e-05,
+      "loss": 0.9883,
+      "step": 92000
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 3.4669153690338135,
+      "learning_rate": 1.4757870278541273e-05,
+      "loss": 1.0007,
+      "step": 92500
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 3.2495174407958984,
+      "learning_rate": 1.4729534442209064e-05,
+      "loss": 1.0275,
+      "step": 93000
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 4.541181564331055,
+      "learning_rate": 1.4701198605876853e-05,
+      "loss": 1.0074,
+      "step": 93500
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 4.682742118835449,
+      "learning_rate": 1.4672862769544644e-05,
+      "loss": 1.0366,
+      "step": 94000
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 3.1074297428131104,
+      "learning_rate": 1.4644526933212435e-05,
+      "loss": 1.0267,
+      "step": 94500
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 2.548527240753174,
+      "learning_rate": 1.4616191096880226e-05,
+      "loss": 0.9915,
+      "step": 95000
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 2.168649673461914,
+      "learning_rate": 1.4587855260548016e-05,
+      "loss": 1.0054,
+      "step": 95500
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 5.411528587341309,
+      "learning_rate": 1.4559519424215806e-05,
+      "loss": 1.0045,
+      "step": 96000
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 3.786655902862549,
+      "learning_rate": 1.4531183587883597e-05,
+      "loss": 0.9932,
+      "step": 96500
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 3.0948727130889893,
+      "learning_rate": 1.4502847751551388e-05,
+      "loss": 0.9707,
+      "step": 97000
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 2.989771842956543,
+      "learning_rate": 1.4474511915219178e-05,
+      "loss": 0.9774,
+      "step": 97500
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 3.2359373569488525,
+      "learning_rate": 1.4446176078886969e-05,
+      "loss": 0.9306,
+      "step": 98000
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 8.07835865020752,
+      "learning_rate": 1.441784024255476e-05,
+      "loss": 0.9524,
+      "step": 98500
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 3.9400320053100586,
+      "learning_rate": 1.438950440622255e-05,
+      "loss": 0.9501,
+      "step": 99000
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 4.814745903015137,
+      "learning_rate": 1.4361168569890343e-05,
+      "loss": 0.9887,
+      "step": 99500
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 2.6948797702789307,
+      "learning_rate": 1.4332832733558132e-05,
+      "loss": 0.9708,
+      "step": 100000
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 2.974473476409912,
+      "learning_rate": 1.4304496897225923e-05,
+      "loss": 0.9709,
+      "step": 100500
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 7.195821762084961,
+      "learning_rate": 1.4276161060893714e-05,
+      "loss": 0.8956,
+      "step": 101000
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 4.091536045074463,
+      "learning_rate": 1.4247825224561505e-05,
+      "loss": 0.9442,
+      "step": 101500
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 2.1364340782165527,
+      "learning_rate": 1.4219489388229294e-05,
+      "loss": 1.0104,
+      "step": 102000
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 2.952352523803711,
+      "learning_rate": 1.4191153551897085e-05,
+      "loss": 0.918,
+      "step": 102500
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 2.175086736679077,
+      "learning_rate": 1.4162817715564876e-05,
+      "loss": 0.9514,
+      "step": 103000
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 3.7156364917755127,
+      "learning_rate": 1.4134481879232667e-05,
+      "loss": 0.9795,
+      "step": 103500
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 3.0967416763305664,
+      "learning_rate": 1.4106146042900456e-05,
+      "loss": 0.9774,
+      "step": 104000
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 3.4768569469451904,
+      "learning_rate": 1.4077810206568247e-05,
+      "loss": 0.9217,
+      "step": 104500
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 8.810585975646973,
+      "learning_rate": 1.4049474370236038e-05,
+      "loss": 0.9517,
+      "step": 105000
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 4.477392673492432,
+      "learning_rate": 1.402113853390383e-05,
+      "loss": 0.9538,
+      "step": 105500
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.7630351393567171,
+      "eval_loss": 0.9388064742088318,
+      "eval_runtime": 2748.5992,
+      "eval_samples_per_second": 34.24,
+      "eval_steps_per_second": 34.24,
+      "step": 105874
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 3.1107003688812256,
+      "learning_rate": 1.3992802697571619e-05,
+      "loss": 0.9243,
+      "step": 106000
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 4.796999931335449,
+      "learning_rate": 1.396446686123941e-05,
+      "loss": 0.9163,
+      "step": 106500
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 3.348123073577881,
+      "learning_rate": 1.3936131024907202e-05,
+      "loss": 0.9476,
+      "step": 107000
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 3.8326098918914795,
+      "learning_rate": 1.3907795188574993e-05,
+      "loss": 0.9742,
+      "step": 107500
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 3.05477237701416,
+      "learning_rate": 1.3879459352242784e-05,
+      "loss": 0.9613,
+      "step": 108000
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 2.1962602138519287,
+      "learning_rate": 1.3851123515910573e-05,
+      "loss": 0.9059,
+      "step": 108500
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 4.850438117980957,
+      "learning_rate": 1.3822787679578364e-05,
+      "loss": 0.9147,
+      "step": 109000
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 4.167685508728027,
+      "learning_rate": 1.3794451843246155e-05,
+      "loss": 0.8815,
+      "step": 109500
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 3.5904150009155273,
+      "learning_rate": 1.3766116006913946e-05,
+      "loss": 0.9336,
+      "step": 110000
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 3.102593421936035,
+      "learning_rate": 1.3737780170581735e-05,
+      "loss": 0.9543,
+      "step": 110500
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 3.3601255416870117,
+      "learning_rate": 1.3709444334249526e-05,
+      "loss": 0.8912,
+      "step": 111000
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 3.960665702819824,
+      "learning_rate": 1.3681108497917317e-05,
+      "loss": 0.9416,
+      "step": 111500
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 2.2088279724121094,
+      "learning_rate": 1.3652772661585108e-05,
+      "loss": 0.9303,
+      "step": 112000
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 3.3079452514648438,
+      "learning_rate": 1.3624436825252897e-05,
+      "loss": 0.9343,
+      "step": 112500
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 3.0884592533111572,
+      "learning_rate": 1.3596100988920688e-05,
+      "loss": 0.8972,
+      "step": 113000
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 3.246960401535034,
+      "learning_rate": 1.356776515258848e-05,
+      "loss": 0.9042,
+      "step": 113500
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.40205252170562744,
+      "learning_rate": 1.353942931625627e-05,
+      "loss": 0.9483,
+      "step": 114000
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 3.8221993446350098,
+      "learning_rate": 1.351109347992406e-05,
+      "loss": 0.9081,
+      "step": 114500
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 2.204099655151367,
+      "learning_rate": 1.3482757643591852e-05,
+      "loss": 0.9174,
+      "step": 115000
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 5.331653118133545,
+      "learning_rate": 1.3454421807259643e-05,
+      "loss": 0.9708,
+      "step": 115500
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 0.37105703353881836,
+      "learning_rate": 1.3426085970927434e-05,
+      "loss": 0.9209,
+      "step": 116000
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 4.132419109344482,
+      "learning_rate": 1.3397750134595225e-05,
+      "loss": 0.9223,
+      "step": 116500
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 3.2373697757720947,
+      "learning_rate": 1.3369414298263014e-05,
+      "loss": 0.9449,
+      "step": 117000
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 2.200536012649536,
+      "learning_rate": 1.3341078461930805e-05,
+      "loss": 0.924,
+      "step": 117500
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 4.251534938812256,
+      "learning_rate": 1.3312742625598596e-05,
+      "loss": 0.9245,
+      "step": 118000
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 4.964018821716309,
+      "learning_rate": 1.3284406789266387e-05,
+      "loss": 0.8797,
+      "step": 118500
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 8.006040573120117,
+      "learning_rate": 1.3256070952934176e-05,
+      "loss": 0.9154,
+      "step": 119000
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 3.682969331741333,
+      "learning_rate": 1.3227735116601967e-05,
+      "loss": 0.9182,
+      "step": 119500
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 5.011803150177002,
+      "learning_rate": 1.3199399280269758e-05,
+      "loss": 0.9166,
+      "step": 120000
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 3.1646368503570557,
+      "learning_rate": 1.3171063443937549e-05,
+      "loss": 0.9276,
+      "step": 120500
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 3.1225647926330566,
+      "learning_rate": 1.3142727607605338e-05,
+      "loss": 0.9428,
+      "step": 121000
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 3.139483690261841,
+      "learning_rate": 1.311439177127313e-05,
+      "loss": 0.9391,
+      "step": 121500
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 4.643157482147217,
+      "learning_rate": 1.308605593494092e-05,
+      "loss": 0.9136,
+      "step": 122000
+    },
+    {
+      "epoch": 3.47,
+      "grad_norm": 4.578587532043457,
+      "learning_rate": 1.3057720098608711e-05,
+      "loss": 0.9092,
+      "step": 122500
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 2.2071352005004883,
+      "learning_rate": 1.3029384262276504e-05,
+      "loss": 0.915,
+      "step": 123000
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 2.4921517372131348,
+      "learning_rate": 1.3001048425944293e-05,
+      "loss": 0.8868,
+      "step": 123500
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 4.749142646789551,
+      "learning_rate": 1.2972712589612084e-05,
+      "loss": 0.9282,
+      "step": 124000
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 2.181434154510498,
+      "learning_rate": 1.2944376753279875e-05,
+      "loss": 0.9086,
+      "step": 124500
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 4.025723457336426,
+      "learning_rate": 1.2916040916947666e-05,
+      "loss": 0.8491,
+      "step": 125000
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 3.551264524459839,
+      "learning_rate": 1.2887705080615455e-05,
+      "loss": 0.9228,
+      "step": 125500
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 2.884403705596924,
+      "learning_rate": 1.2859369244283246e-05,
+      "loss": 0.9004,
+      "step": 126000
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 2.999499559402466,
+      "learning_rate": 1.2831033407951037e-05,
+      "loss": 0.8974,
+      "step": 126500
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.017392095178365707,
+      "learning_rate": 1.2802697571618828e-05,
+      "loss": 0.9185,
+      "step": 127000
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 3.908944845199585,
+      "learning_rate": 1.2774361735286617e-05,
+      "loss": 0.9096,
+      "step": 127500
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 4.624220848083496,
+      "learning_rate": 1.2746025898954408e-05,
+      "loss": 0.9109,
+      "step": 128000
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 3.6561858654022217,
+      "learning_rate": 1.2717690062622199e-05,
+      "loss": 0.9003,
+      "step": 128500
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.10583093762397766,
+      "learning_rate": 1.268935422628999e-05,
+      "loss": 0.9076,
+      "step": 129000
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 3.653130292892456,
+      "learning_rate": 1.266101838995778e-05,
+      "loss": 0.9232,
+      "step": 129500
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 3.191444158554077,
+      "learning_rate": 1.263268255362557e-05,
+      "loss": 0.9331,
+      "step": 130000
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 5.10831880569458,
+      "learning_rate": 1.2604346717293361e-05,
+      "loss": 0.9038,
+      "step": 130500
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 3.668541193008423,
+      "learning_rate": 1.2576010880961154e-05,
+      "loss": 0.8835,
+      "step": 131000
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 4.350325584411621,
+      "learning_rate": 1.2547675044628945e-05,
+      "loss": 0.8649,
+      "step": 131500
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 5.0062079429626465,
+      "learning_rate": 1.2519339208296734e-05,
+      "loss": 0.9066,
+      "step": 132000
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 3.533888816833496,
+      "learning_rate": 1.2491003371964525e-05,
+      "loss": 0.893,
+      "step": 132500
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 2.9693057537078857,
+      "learning_rate": 1.2462667535632316e-05,
+      "loss": 0.8672,
+      "step": 133000
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 3.295539140701294,
+      "learning_rate": 1.2434331699300107e-05,
+      "loss": 0.8736,
+      "step": 133500
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 2.2124862670898438,
+      "learning_rate": 1.2405995862967896e-05,
+      "loss": 0.8584,
+      "step": 134000
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 3.2220728397369385,
+      "learning_rate": 1.2377660026635687e-05,
+      "loss": 0.8644,
+      "step": 134500
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 4.965516567230225,
+      "learning_rate": 1.2349324190303478e-05,
+      "loss": 0.8814,
+      "step": 135000
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 4.129695415496826,
+      "learning_rate": 1.2320988353971269e-05,
+      "loss": 0.86,
+      "step": 135500
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 2.1921725273132324,
+      "learning_rate": 1.2292652517639058e-05,
+      "loss": 0.8896,
+      "step": 136000
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.049632515758275986,
+      "learning_rate": 1.2264316681306849e-05,
+      "loss": 0.8845,
+      "step": 136500
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 3.807011604309082,
+      "learning_rate": 1.223598084497464e-05,
+      "loss": 0.9152,
+      "step": 137000
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 5.094349384307861,
+      "learning_rate": 1.2207645008642431e-05,
+      "loss": 0.8885,
+      "step": 137500
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 6.3345537185668945,
+      "learning_rate": 1.217930917231022e-05,
+      "loss": 0.8714,
+      "step": 138000
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 4.590448379516602,
+      "learning_rate": 1.2150973335978011e-05,
+      "loss": 0.857,
+      "step": 138500
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 3.5549352169036865,
+      "learning_rate": 1.2122637499645804e-05,
+      "loss": 0.9091,
+      "step": 139000
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 3.162019968032837,
+      "learning_rate": 1.2094301663313595e-05,
+      "loss": 0.8527,
+      "step": 139500
+    },
+    {
+      "epoch": 3.97,
+      "grad_norm": 7.230122089385986,
+      "learning_rate": 1.2065965826981386e-05,
+      "loss": 0.8646,
+      "step": 140000
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.9568886756896973,
+      "learning_rate": 1.2037629990649175e-05,
+      "loss": 0.8996,
+      "step": 140500
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 6.046046257019043,
+      "learning_rate": 1.2009294154316966e-05,
+      "loss": 0.902,
+      "step": 141000
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.770122514902615,
+      "eval_loss": 0.8713550567626953,
+      "eval_runtime": 2750.3895,
+      "eval_samples_per_second": 34.217,
+      "eval_steps_per_second": 34.217,
+      "step": 141165
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 4.080121040344238,
+      "learning_rate": 1.1980958317984757e-05,
+      "loss": 0.8701,
+      "step": 141500
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 3.5020222663879395,
+      "learning_rate": 1.1952622481652548e-05,
+      "loss": 0.9201,
+      "step": 142000
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 5.760317802429199,
+      "learning_rate": 1.1924286645320337e-05,
+      "loss": 0.8844,
+      "step": 142500
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 3.131807565689087,
+      "learning_rate": 1.1895950808988128e-05,
+      "loss": 0.8403,
+      "step": 143000
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 2.6976137161254883,
+      "learning_rate": 1.1867614972655919e-05,
+      "loss": 0.837,
+      "step": 143500
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 4.363113880157471,
+      "learning_rate": 1.183927913632371e-05,
+      "loss": 0.8528,
+      "step": 144000
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 3.6650633811950684,
+      "learning_rate": 1.1810943299991499e-05,
+      "loss": 0.866,
+      "step": 144500
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 2.1963698863983154,
+      "learning_rate": 1.178260746365929e-05,
+      "loss": 0.8799,
+      "step": 145000
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 6.11692476272583,
+      "learning_rate": 1.1754271627327081e-05,
+      "loss": 0.862,
+      "step": 145500
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 3.8074452877044678,
+      "learning_rate": 1.1725935790994872e-05,
+      "loss": 0.8481,
+      "step": 146000
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 3.0709991455078125,
+      "learning_rate": 1.1697599954662661e-05,
+      "loss": 0.8673,
+      "step": 146500
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 4.780150413513184,
+      "learning_rate": 1.1669264118330454e-05,
+      "loss": 0.8413,
+      "step": 147000
+    },
+    {
+      "epoch": 4.18,
+      "grad_norm": 3.2353837490081787,
+      "learning_rate": 1.1640928281998245e-05,
+      "loss": 0.8593,
+      "step": 147500
+    },
+    {
+      "epoch": 4.19,
+      "grad_norm": 2.6519854068756104,
+      "learning_rate": 1.1612592445666036e-05,
+      "loss": 0.8726,
+      "step": 148000
+    },
+    {
+      "epoch": 4.21,
+      "grad_norm": 6.630087852478027,
+      "learning_rate": 1.1584256609333827e-05,
+      "loss": 0.8422,
+      "step": 148500
+    },
+    {
+      "epoch": 4.22,
+      "grad_norm": 4.064183712005615,
+      "learning_rate": 1.1555920773001616e-05,
+      "loss": 0.8463,
+      "step": 149000
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 2.2233190536499023,
+      "learning_rate": 1.1527584936669407e-05,
+      "loss": 0.8529,
+      "step": 149500
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.32811805605888367,
+      "learning_rate": 1.1499249100337198e-05,
+      "loss": 0.8887,
+      "step": 150000
+    },
+    {
+      "epoch": 4.26,
+      "grad_norm": 3.161423683166504,
+      "learning_rate": 1.1470913264004989e-05,
+      "loss": 0.8435,
+      "step": 150500
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 0.14471185207366943,
+      "learning_rate": 1.1442577427672778e-05,
+      "loss": 0.8758,
+      "step": 151000
+    },
+    {
+      "epoch": 4.29,
+      "grad_norm": 2.2590670585632324,
+      "learning_rate": 1.1414241591340569e-05,
+      "loss": 0.8808,
+      "step": 151500
+    },
+    {
+      "epoch": 4.31,
+      "grad_norm": 8.431534767150879,
+      "learning_rate": 1.138590575500836e-05,
+      "loss": 0.8463,
+      "step": 152000
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 3.280068874359131,
+      "learning_rate": 1.135756991867615e-05,
+      "loss": 0.8644,
+      "step": 152500
+    },
+    {
+      "epoch": 4.34,
+      "grad_norm": 3.7241225242614746,
+      "learning_rate": 1.132923408234394e-05,
+      "loss": 0.8531,
+      "step": 153000
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 3.3345284461975098,
+      "learning_rate": 1.130089824601173e-05,
+      "loss": 0.8671,
+      "step": 153500
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 0.11300093680620193,
+      "learning_rate": 1.1272562409679522e-05,
+      "loss": 0.8489,
+      "step": 154000
+    },
+    {
+      "epoch": 4.38,
+      "grad_norm": 3.3431150913238525,
+      "learning_rate": 1.1244226573347313e-05,
+      "loss": 0.9055,
+      "step": 154500
+    },
+    {
+      "epoch": 4.39,
+      "grad_norm": 2.2208619117736816,
+      "learning_rate": 1.1215890737015105e-05,
+      "loss": 0.8338,
+      "step": 155000
+    },
+    {
+      "epoch": 4.41,
+      "grad_norm": 3.340219020843506,
+      "learning_rate": 1.1187554900682895e-05,
+      "loss": 0.8709,
+      "step": 155500
+    },
+    {
+      "epoch": 4.42,
+      "grad_norm": 0.9095796942710876,
+      "learning_rate": 1.1159219064350686e-05,
+      "loss": 0.8795,
+      "step": 156000
+    },
+    {
+      "epoch": 4.43,
+      "grad_norm": 6.427603721618652,
+      "learning_rate": 1.1130883228018476e-05,
+      "loss": 0.8372,
+      "step": 156500
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 3.5089049339294434,
+      "learning_rate": 1.1102547391686267e-05,
+      "loss": 0.9082,
+      "step": 157000
+    },
+    {
+      "epoch": 4.46,
+      "grad_norm": 0.014750667847692966,
+      "learning_rate": 1.1074211555354057e-05,
+      "loss": 0.8351,
+      "step": 157500
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 5.520388126373291,
+      "learning_rate": 1.1045875719021848e-05,
+      "loss": 0.8342,
+      "step": 158000
+    },
+    {
+      "epoch": 4.49,
+      "grad_norm": 4.782327651977539,
+      "learning_rate": 1.1017539882689639e-05,
+      "loss": 0.8209,
+      "step": 158500
+    },
+    {
+      "epoch": 4.51,
+      "grad_norm": 2.2101173400878906,
+      "learning_rate": 1.098920404635743e-05,
+      "loss": 0.7976,
+      "step": 159000
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 5.672389030456543,
+      "learning_rate": 1.0960868210025219e-05,
+      "loss": 0.8703,
+      "step": 159500
+    },
+    {
+      "epoch": 4.53,
+      "grad_norm": 3.9403274059295654,
+      "learning_rate": 1.093253237369301e-05,
+      "loss": 0.8647,
+      "step": 160000
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 4.439927577972412,
+      "learning_rate": 1.09041965373608e-05,
+      "loss": 0.8725,
+      "step": 160500
+    },
+    {
+      "epoch": 4.56,
+      "grad_norm": 3.5873897075653076,
+      "learning_rate": 1.0875860701028592e-05,
+      "loss": 0.8628,
+      "step": 161000
+    },
+    {
+      "epoch": 4.58,
+      "grad_norm": 2.261143922805786,
+      "learning_rate": 1.084752486469638e-05,
+      "loss": 0.8657,
+      "step": 161500
+    },
+    {
+      "epoch": 4.59,
+      "grad_norm": 0.026868639513850212,
+      "learning_rate": 1.0819189028364172e-05,
+      "loss": 0.8614,
+      "step": 162000
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 3.7286293506622314,
+      "learning_rate": 1.0790853192031964e-05,
+      "loss": 0.8624,
+      "step": 162500
+    },
+    {
+      "epoch": 4.62,
+      "grad_norm": 2.272576093673706,
+      "learning_rate": 1.0762517355699755e-05,
+      "loss": 0.8414,
+      "step": 163000
+    },
+    {
+      "epoch": 4.63,
+      "grad_norm": 3.6233222484588623,
+      "learning_rate": 1.0734181519367546e-05,
+      "loss": 0.8429,
+      "step": 163500
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 2.8637378215789795,
+      "learning_rate": 1.0705845683035336e-05,
+      "loss": 0.8615,
+      "step": 164000
+    },
+    {
+      "epoch": 4.66,
+      "grad_norm": 3.5243945121765137,
+      "learning_rate": 1.0677509846703126e-05,
+      "loss": 0.8324,
+      "step": 164500
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 3.835604667663574,
+      "learning_rate": 1.0649174010370917e-05,
+      "loss": 0.8001,
+      "step": 165000
+    },
+    {
+      "epoch": 4.69,
+      "grad_norm": 4.268076419830322,
+      "learning_rate": 1.0620838174038708e-05,
+      "loss": 0.8573,
+      "step": 165500
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 5.4616312980651855,
+      "learning_rate": 1.0592502337706498e-05,
+      "loss": 0.8347,
+      "step": 166000
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 4.885875701904297,
+      "learning_rate": 1.0564166501374289e-05,
+      "loss": 0.8819,
+      "step": 166500
+    },
+    {
+      "epoch": 4.73,
+      "grad_norm": 3.8010919094085693,
+      "learning_rate": 1.053583066504208e-05,
+      "loss": 0.8585,
+      "step": 167000
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 2.4292032718658447,
+      "learning_rate": 1.050749482870987e-05,
+      "loss": 0.8869,
+      "step": 167500
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 3.1070659160614014,
+      "learning_rate": 1.047915899237766e-05,
+      "loss": 0.8033,
+      "step": 168000
+    },
+    {
+      "epoch": 4.77,
+      "grad_norm": 3.7243080139160156,
+      "learning_rate": 1.045082315604545e-05,
+      "loss": 0.8518,
+      "step": 168500
+    },
+    {
+      "epoch": 4.79,
+      "grad_norm": 3.1299855709075928,
+      "learning_rate": 1.0422487319713242e-05,
+      "loss": 0.8439,
+      "step": 169000
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.01023823581635952,
+      "learning_rate": 1.0394151483381032e-05,
+      "loss": 0.8251,
+      "step": 169500
+    },
+    {
+      "epoch": 4.82,
+      "grad_norm": 5.466165542602539,
+      "learning_rate": 1.0365815647048822e-05,
+      "loss": 0.8065,
+      "step": 170000
+    },
+    {
+      "epoch": 4.83,
+      "grad_norm": 3.4839234352111816,
+      "learning_rate": 1.0337479810716614e-05,
+      "loss": 0.8351,
+      "step": 170500
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 2.2356865406036377,
+      "learning_rate": 1.0309143974384405e-05,
+      "loss": 0.8775,
+      "step": 171000
+    },
+    {
+      "epoch": 4.86,
+      "grad_norm": 4.999177932739258,
+      "learning_rate": 1.0280808138052196e-05,
+      "loss": 0.8641,
+      "step": 171500
+    },
+    {
+      "epoch": 4.87,
+      "grad_norm": 3.0965590476989746,
+      "learning_rate": 1.0252472301719987e-05,
+      "loss": 0.8145,
+      "step": 172000
+    },
+    {
+      "epoch": 4.89,
+      "grad_norm": 3.6084883213043213,
+      "learning_rate": 1.0224136465387776e-05,
+      "loss": 0.8015,
+      "step": 172500
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 0.03004506602883339,
+      "learning_rate": 1.0195800629055567e-05,
+      "loss": 0.8532,
+      "step": 173000
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 7.562097072601318,
+      "learning_rate": 1.0167464792723358e-05,
+      "loss": 0.8199,
+      "step": 173500
+    },
+    {
+      "epoch": 4.93,
+      "grad_norm": 5.141594886779785,
+      "learning_rate": 1.013912895639115e-05,
+      "loss": 0.8355,
+      "step": 174000
+    },
+    {
+      "epoch": 4.94,
+      "grad_norm": 3.5604329109191895,
+      "learning_rate": 1.0110793120058939e-05,
+      "loss": 0.8121,
+      "step": 174500
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 2.2066256999969482,
+      "learning_rate": 1.008245728372673e-05,
+      "loss": 0.8497,
+      "step": 175000
+    },
+    {
+      "epoch": 4.97,
+      "grad_norm": 4.226958274841309,
+      "learning_rate": 1.005412144739452e-05,
+      "loss": 0.7996,
+      "step": 175500
+    },
+    {
+      "epoch": 4.99,
+      "grad_norm": 3.455479383468628,
+      "learning_rate": 1.0025785611062311e-05,
+      "loss": 0.8377,
+      "step": 176000
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.7734058717896951,
+      "eval_loss": 0.8303263783454895,
+      "eval_runtime": 2743.1767,
+      "eval_samples_per_second": 34.307,
+      "eval_steps_per_second": 34.307,
+      "step": 176456
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.007345318328589201,
+      "learning_rate": 9.997449774730102e-06,
+      "loss": 0.8228,
+      "step": 176500
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 3.3040854930877686,
+      "learning_rate": 9.969113938397893e-06,
+      "loss": 0.8059,
+      "step": 177000
+    },
+    {
+      "epoch": 5.03,
+      "grad_norm": 0.11892475187778473,
+      "learning_rate": 9.940778102065684e-06,
+      "loss": 0.8316,
+      "step": 177500
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 3.803138017654419,
+      "learning_rate": 9.912442265733473e-06,
+      "loss": 0.8172,
+      "step": 178000
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 10.063855171203613,
+      "learning_rate": 9.884106429401264e-06,
+      "loss": 0.7939,
+      "step": 178500
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 2.2473671436309814,
+      "learning_rate": 9.855770593069055e-06,
+      "loss": 0.8246,
+      "step": 179000
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 3.1692700386047363,
+      "learning_rate": 9.827434756736846e-06,
+      "loss": 0.816,
+      "step": 179500
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 2.5137248039245605,
+      "learning_rate": 9.799098920404635e-06,
+      "loss": 0.7863,
+      "step": 180000
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 2.4693000316619873,
+      "learning_rate": 9.770763084072428e-06,
+      "loss": 0.8297,
+      "step": 180500
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 3.0473134517669678,
+      "learning_rate": 9.742427247740217e-06,
+      "loss": 0.8124,
+      "step": 181000
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 3.729397773742676,
+      "learning_rate": 9.714091411408008e-06,
+      "loss": 0.8146,
+      "step": 181500
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 1.7608509063720703,
+      "learning_rate": 9.6857555750758e-06,
+      "loss": 0.8478,
+      "step": 182000
+    },
+    {
+      "epoch": 5.17,
+      "grad_norm": 2.2976291179656982,
+      "learning_rate": 9.65741973874359e-06,
+      "loss": 0.8438,
+      "step": 182500
+    },
+    {
+      "epoch": 5.19,
+      "grad_norm": 3.66800856590271,
+      "learning_rate": 9.62908390241138e-06,
+      "loss": 0.8117,
+      "step": 183000
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 2.1089248657226562,
+      "learning_rate": 9.60074806607917e-06,
+      "loss": 0.7711,
+      "step": 183500
+    },
+    {
+      "epoch": 5.21,
+      "grad_norm": 2.214057207107544,
+      "learning_rate": 9.572412229746961e-06,
+      "loss": 0.8131,
+      "step": 184000
+    },
+    {
+      "epoch": 5.23,
+      "grad_norm": 2.953176736831665,
+      "learning_rate": 9.544076393414752e-06,
+      "loss": 0.8358,
+      "step": 184500
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 4.549135684967041,
+      "learning_rate": 9.515740557082543e-06,
+      "loss": 0.7986,
+      "step": 185000
+    },
+    {
+      "epoch": 5.26,
+      "grad_norm": 5.39935827255249,
+      "learning_rate": 9.487404720750334e-06,
+      "loss": 0.8279,
+      "step": 185500
+    },
+    {
+      "epoch": 5.27,
+      "grad_norm": 2.80328631401062,
+      "learning_rate": 9.459068884418125e-06,
+      "loss": 0.8442,
+      "step": 186000
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 3.071382522583008,
+      "learning_rate": 9.430733048085914e-06,
+      "loss": 0.8557,
+      "step": 186500
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 3.065718412399292,
+      "learning_rate": 9.402397211753705e-06,
+      "loss": 0.8119,
+      "step": 187000
+    },
+    {
+      "epoch": 5.31,
+      "grad_norm": 3.371381998062134,
+      "learning_rate": 9.374061375421496e-06,
+      "loss": 0.7994,
+      "step": 187500
+    },
+    {
+      "epoch": 5.33,
+      "grad_norm": 4.33438777923584,
+      "learning_rate": 9.345725539089287e-06,
+      "loss": 0.804,
+      "step": 188000
+    },
+    {
+      "epoch": 5.34,
+      "grad_norm": 5.77909517288208,
+      "learning_rate": 9.317389702757078e-06,
+      "loss": 0.8357,
+      "step": 188500
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 2.6496472358703613,
+      "learning_rate": 9.289053866424869e-06,
+      "loss": 0.8209,
+      "step": 189000
+    },
+    {
+      "epoch": 5.37,
+      "grad_norm": 4.3701348304748535,
+      "learning_rate": 9.260718030092658e-06,
+      "loss": 0.8317,
+      "step": 189500
+    },
+    {
+      "epoch": 5.38,
+      "grad_norm": 3.137705087661743,
+      "learning_rate": 9.23238219376045e-06,
+      "loss": 0.7959,
+      "step": 190000
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 2.3616700172424316,
+      "learning_rate": 9.20404635742824e-06,
+      "loss": 0.7894,
+      "step": 190500
+    },
+    {
+      "epoch": 5.41,
+      "grad_norm": 2.2935869693756104,
+      "learning_rate": 9.175710521096031e-06,
+      "loss": 0.8328,
+      "step": 191000
+    },
+    {
+      "epoch": 5.43,
+      "grad_norm": 4.851805686950684,
+      "learning_rate": 9.14737468476382e-06,
+      "loss": 0.8681,
+      "step": 191500
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 9.607928276062012,
+      "learning_rate": 9.119038848431611e-06,
+      "loss": 0.7672,
+      "step": 192000
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 3.8751378059387207,
+      "learning_rate": 9.090703012099404e-06,
+      "loss": 0.8453,
+      "step": 192500
+    },
+    {
+      "epoch": 5.47,
+      "grad_norm": 4.63936710357666,
+      "learning_rate": 9.062367175767193e-06,
+      "loss": 0.8093,
+      "step": 193000
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 3.8076059818267822,
+      "learning_rate": 9.034031339434984e-06,
+      "loss": 0.7815,
+      "step": 193500
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 3.0797765254974365,
+      "learning_rate": 9.005695503102775e-06,
+      "loss": 0.805,
+      "step": 194000
+    },
+    {
+      "epoch": 5.51,
+      "grad_norm": 4.025183200836182,
+      "learning_rate": 8.977359666770566e-06,
+      "loss": 0.8292,
+      "step": 194500
+    },
+    {
+      "epoch": 5.53,
+      "grad_norm": 3.398597002029419,
+      "learning_rate": 8.949023830438355e-06,
+      "loss": 0.8072,
+      "step": 195000
+    },
+    {
+      "epoch": 5.54,
+      "grad_norm": 3.7402262687683105,
+      "learning_rate": 8.920687994106146e-06,
+      "loss": 0.8492,
+      "step": 195500
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 4.7148919105529785,
+      "learning_rate": 8.892352157773937e-06,
+      "loss": 0.8189,
+      "step": 196000
+    },
+    {
+      "epoch": 5.57,
+      "grad_norm": 2.135432004928589,
+      "learning_rate": 8.864016321441728e-06,
+      "loss": 0.8146,
+      "step": 196500
+    },
+    {
+      "epoch": 5.58,
+      "grad_norm": 3.6518137454986572,
+      "learning_rate": 8.835680485109519e-06,
+      "loss": 0.8169,
+      "step": 197000
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 0.9229252338409424,
+      "learning_rate": 8.80734464877731e-06,
+      "loss": 0.8226,
+      "step": 197500
+    },
+    {
+      "epoch": 5.61,
+      "grad_norm": 0.1361069679260254,
+      "learning_rate": 8.7790088124451e-06,
+      "loss": 0.8053,
+      "step": 198000
+    },
+    {
+      "epoch": 5.62,
+      "grad_norm": 0.059051357209682465,
+      "learning_rate": 8.75067297611289e-06,
+      "loss": 0.8288,
+      "step": 198500
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 4.741939544677734,
+      "learning_rate": 8.722337139780681e-06,
+      "loss": 0.8408,
+      "step": 199000
+    },
+    {
+      "epoch": 5.65,
+      "grad_norm": 3.188978433609009,
+      "learning_rate": 8.694001303448472e-06,
+      "loss": 0.8116,
+      "step": 199500
+    },
+    {
+      "epoch": 5.67,
+      "grad_norm": 3.483867883682251,
+      "learning_rate": 8.665665467116261e-06,
+      "loss": 0.8168,
+      "step": 200000
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 3.1518492698669434,
+      "learning_rate": 8.637329630784054e-06,
+      "loss": 0.832,
+      "step": 200500
+    },
+    {
+      "epoch": 5.7,
+      "grad_norm": 5.365024089813232,
+      "learning_rate": 8.608993794451845e-06,
+      "loss": 0.8259,
+      "step": 201000
+    },
+    {
+      "epoch": 5.71,
+      "grad_norm": 3.5916318893432617,
+      "learning_rate": 8.580657958119634e-06,
+      "loss": 0.8245,
+      "step": 201500
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 3.5970184803009033,
+      "learning_rate": 8.552322121787425e-06,
+      "loss": 0.8316,
+      "step": 202000
+    },
+    {
+      "epoch": 5.74,
+      "grad_norm": 4.369210243225098,
+      "learning_rate": 8.523986285455216e-06,
+      "loss": 0.8032,
+      "step": 202500
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 2.211902618408203,
+      "learning_rate": 8.495650449123007e-06,
+      "loss": 0.8216,
+      "step": 203000
+    },
+    {
+      "epoch": 5.77,
+      "grad_norm": 0.11311174184083939,
+      "learning_rate": 8.467314612790796e-06,
+      "loss": 0.8386,
+      "step": 203500
+    },
+    {
+      "epoch": 5.78,
+      "grad_norm": 2.7481818199157715,
+      "learning_rate": 8.438978776458587e-06,
+      "loss": 0.8298,
+      "step": 204000
+    },
+    {
+      "epoch": 5.79,
+      "grad_norm": 2.241297960281372,
+      "learning_rate": 8.41064294012638e-06,
+      "loss": 0.7622,
+      "step": 204500
+    },
+    {
+      "epoch": 5.81,
+      "grad_norm": 3.433582067489624,
+      "learning_rate": 8.382307103794169e-06,
+      "loss": 0.7888,
+      "step": 205000
+    },
+    {
+      "epoch": 5.82,
+      "grad_norm": 3.3529977798461914,
+      "learning_rate": 8.35397126746196e-06,
+      "loss": 0.8105,
+      "step": 205500
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 2.6061885356903076,
+      "learning_rate": 8.325635431129751e-06,
+      "loss": 0.8184,
+      "step": 206000
+    },
+    {
+      "epoch": 5.85,
+      "grad_norm": 2.5410923957824707,
+      "learning_rate": 8.297299594797542e-06,
+      "loss": 0.8144,
+      "step": 206500
+    },
+    {
+      "epoch": 5.87,
+      "grad_norm": 4.426595211029053,
+      "learning_rate": 8.268963758465331e-06,
+      "loss": 0.8693,
+      "step": 207000
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 4.819573879241943,
+      "learning_rate": 8.240627922133122e-06,
+      "loss": 0.8314,
+      "step": 207500
+    },
+    {
+      "epoch": 5.89,
+      "grad_norm": 2.823551893234253,
+      "learning_rate": 8.212292085800913e-06,
+      "loss": 0.789,
+      "step": 208000
+    },
+    {
+      "epoch": 5.91,
+      "grad_norm": 3.179856777191162,
+      "learning_rate": 8.183956249468704e-06,
+      "loss": 0.8096,
+      "step": 208500
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 3.1199803352355957,
+      "learning_rate": 8.155620413136495e-06,
+      "loss": 0.8197,
+      "step": 209000
+    },
+    {
+      "epoch": 5.94,
+      "grad_norm": 0.029315292835235596,
+      "learning_rate": 8.127284576804286e-06,
+      "loss": 0.7781,
+      "step": 209500
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 4.753857135772705,
+      "learning_rate": 8.098948740472075e-06,
+      "loss": 0.7788,
+      "step": 210000
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 2.8551526069641113,
+      "learning_rate": 8.070612904139866e-06,
+      "loss": 0.8241,
+      "step": 210500
+    },
+    {
+      "epoch": 5.98,
+      "grad_norm": 2.6687989234924316,
+      "learning_rate": 8.042277067807657e-06,
+      "loss": 0.805,
+      "step": 211000
+    },
+    {
+      "epoch": 5.99,
+      "grad_norm": 3.9277312755584717,
+      "learning_rate": 8.013941231475448e-06,
+      "loss": 0.8204,
+      "step": 211500
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.7752760038677732,
+      "eval_loss": 0.8054434061050415,
+      "eval_runtime": 2742.5432,
+      "eval_samples_per_second": 34.315,
+      "eval_steps_per_second": 34.315,
+      "step": 211748
+    },
+    {
+      "epoch": 6.01,
+      "grad_norm": 2.4980616569519043,
+      "learning_rate": 7.985605395143239e-06,
+      "loss": 0.8555,
+      "step": 212000
+    },
+    {
+      "epoch": 6.02,
+      "grad_norm": 1.9361926317214966,
+      "learning_rate": 7.95726955881103e-06,
+      "loss": 0.7816,
+      "step": 212500
+    },
+    {
+      "epoch": 6.04,
+      "grad_norm": 2.223845958709717,
+      "learning_rate": 7.92893372247882e-06,
+      "loss": 0.7757,
+      "step": 213000
+    },
+    {
+      "epoch": 6.05,
+      "grad_norm": 2.364459991455078,
+      "learning_rate": 7.90059788614661e-06,
+      "loss": 0.8488,
+      "step": 213500
+    },
+    {
+      "epoch": 6.06,
+      "grad_norm": 4.380248069763184,
+      "learning_rate": 7.872262049814401e-06,
+      "loss": 0.7997,
+      "step": 214000
+    },
+    {
+      "epoch": 6.08,
+      "grad_norm": 4.055036544799805,
+      "learning_rate": 7.843926213482192e-06,
+      "loss": 0.7863,
+      "step": 214500
+    },
+    {
+      "epoch": 6.09,
+      "grad_norm": 0.13833500444889069,
+      "learning_rate": 7.815590377149983e-06,
+      "loss": 0.8032,
+      "step": 215000
+    },
+    {
+      "epoch": 6.11,
+      "grad_norm": 3.1694560050964355,
+      "learning_rate": 7.787254540817772e-06,
+      "loss": 0.793,
+      "step": 215500
+    },
+    {
+      "epoch": 6.12,
+      "grad_norm": 3.2722058296203613,
+      "learning_rate": 7.758918704485565e-06,
+      "loss": 0.8196,
+      "step": 216000
+    },
+    {
+      "epoch": 6.13,
+      "grad_norm": 3.0875654220581055,
+      "learning_rate": 7.730582868153354e-06,
+      "loss": 0.7951,
+      "step": 216500
+    },
+    {
+      "epoch": 6.15,
+      "grad_norm": 1.9850670099258423,
+      "learning_rate": 7.702247031821145e-06,
+      "loss": 0.804,
+      "step": 217000
+    },
+    {
+      "epoch": 6.16,
+      "grad_norm": 3.0951926708221436,
+      "learning_rate": 7.673911195488936e-06,
+      "loss": 0.8405,
+      "step": 217500
+    },
+    {
+      "epoch": 6.18,
+      "grad_norm": 3.2018356323242188,
+      "learning_rate": 7.645575359156727e-06,
+      "loss": 0.8358,
+      "step": 218000
+    },
+    {
+      "epoch": 6.19,
+      "grad_norm": 2.13598895072937,
+      "learning_rate": 7.617239522824517e-06,
+      "loss": 0.7789,
+      "step": 218500
+    },
+    {
+      "epoch": 6.21,
+      "grad_norm": 3.089243173599243,
+      "learning_rate": 7.588903686492307e-06,
+      "loss": 0.7905,
+      "step": 219000
+    },
+    {
+      "epoch": 6.22,
+      "grad_norm": 5.413113594055176,
+      "learning_rate": 7.560567850160098e-06,
+      "loss": 0.834,
+      "step": 219500
+    },
+    {
+      "epoch": 6.23,
+      "grad_norm": 1.8769762516021729,
+      "learning_rate": 7.53223201382789e-06,
+      "loss": 0.7967,
+      "step": 220000
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 0.30497410893440247,
+      "learning_rate": 7.50389617749568e-06,
+      "loss": 0.8116,
+      "step": 220500
+    },
+    {
+      "epoch": 6.26,
+      "grad_norm": 3.9341723918914795,
+      "learning_rate": 7.475560341163471e-06,
+      "loss": 0.8227,
+      "step": 221000
+    },
+    {
+      "epoch": 6.28,
+      "grad_norm": 4.053406238555908,
+      "learning_rate": 7.447224504831261e-06,
+      "loss": 0.7984,
+      "step": 221500
+    },
+    {
+      "epoch": 6.29,
+      "grad_norm": 4.562232494354248,
+      "learning_rate": 7.418888668499052e-06,
+      "loss": 0.7988,
+      "step": 222000
+    },
+    {
+      "epoch": 6.3,
+      "grad_norm": 4.173889636993408,
+      "learning_rate": 7.390552832166842e-06,
+      "loss": 0.8079,
+      "step": 222500
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 2.6685173511505127,
+      "learning_rate": 7.362216995834633e-06,
+      "loss": 0.7419,
+      "step": 223000
+    },
+    {
+      "epoch": 6.33,
+      "grad_norm": 3.2578206062316895,
+      "learning_rate": 7.333881159502423e-06,
+      "loss": 0.783,
+      "step": 223500
+    },
+    {
+      "epoch": 6.35,
+      "grad_norm": 3.737518548965454,
+      "learning_rate": 7.305545323170215e-06,
+      "loss": 0.7971,
+      "step": 224000
+    },
+    {
+      "epoch": 6.36,
+      "grad_norm": 3.875011920928955,
+      "learning_rate": 7.277209486838005e-06,
+      "loss": 0.7866,
+      "step": 224500
+    },
+    {
+      "epoch": 6.38,
+      "grad_norm": 2.1486947536468506,
+      "learning_rate": 7.248873650505796e-06,
+      "loss": 0.7771,
+      "step": 225000
+    },
+    {
+      "epoch": 6.39,
+      "grad_norm": 3.4426918029785156,
+      "learning_rate": 7.220537814173586e-06,
+      "loss": 0.7717,
+      "step": 225500
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 3.809903621673584,
+      "learning_rate": 7.192201977841377e-06,
+      "loss": 0.8186,
+      "step": 226000
+    },
+    {
+      "epoch": 6.42,
+      "grad_norm": 0.17552083730697632,
+      "learning_rate": 7.163866141509167e-06,
+      "loss": 0.7982,
+      "step": 226500
+    },
+    {
+      "epoch": 6.43,
+      "grad_norm": 3.331289768218994,
+      "learning_rate": 7.135530305176958e-06,
+      "loss": 0.7963,
+      "step": 227000
+    },
+    {
+      "epoch": 6.45,
+      "grad_norm": 3.937941074371338,
+      "learning_rate": 7.107194468844748e-06,
+      "loss": 0.8209,
+      "step": 227500
+    },
+    {
+      "epoch": 6.46,
+      "grad_norm": 12.834056854248047,
+      "learning_rate": 7.07885863251254e-06,
+      "loss": 0.8353,
+      "step": 228000
+    },
+    {
+      "epoch": 6.47,
+      "grad_norm": 3.121537923812866,
+      "learning_rate": 7.0505227961803305e-06,
+      "loss": 0.7912,
+      "step": 228500
+    },
+    {
+      "epoch": 6.49,
+      "grad_norm": 3.6076176166534424,
+      "learning_rate": 7.022186959848121e-06,
+      "loss": 0.7881,
+      "step": 229000
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 2.1992852687835693,
+      "learning_rate": 6.993851123515912e-06,
+      "loss": 0.7787,
+      "step": 229500
+    },
+    {
+      "epoch": 6.52,
+      "grad_norm": 2.540170192718506,
+      "learning_rate": 6.965515287183702e-06,
+      "loss": 0.7942,
+      "step": 230000
+    },
+    {
+      "epoch": 6.53,
+      "grad_norm": 4.526458740234375,
+      "learning_rate": 6.937179450851493e-06,
+      "loss": 0.8014,
+      "step": 230500
+    },
+    {
+      "epoch": 6.55,
+      "grad_norm": 3.644343376159668,
+      "learning_rate": 6.908843614519283e-06,
+      "loss": 0.8333,
+      "step": 231000
+    },
+    {
+      "epoch": 6.56,
+      "grad_norm": 3.894827127456665,
+      "learning_rate": 6.880507778187074e-06,
+      "loss": 0.801,
+      "step": 231500
+    },
+    {
+      "epoch": 6.57,
+      "grad_norm": 3.7568604946136475,
+      "learning_rate": 6.852171941854865e-06,
+      "loss": 0.7923,
+      "step": 232000
+    },
+    {
+      "epoch": 6.59,
+      "grad_norm": 4.713413238525391,
+      "learning_rate": 6.8238361055226555e-06,
+      "loss": 0.777,
+      "step": 232500
+    },
+    {
+      "epoch": 6.6,
+      "grad_norm": 2.4898388385772705,
+      "learning_rate": 6.795500269190446e-06,
+      "loss": 0.8041,
+      "step": 233000
+    },
+    {
+      "epoch": 6.62,
+      "grad_norm": 2.9944307804107666,
+      "learning_rate": 6.7671644328582366e-06,
+      "loss": 0.8332,
+      "step": 233500
+    },
+    {
+      "epoch": 6.63,
+      "grad_norm": 4.278744697570801,
+      "learning_rate": 6.738828596526027e-06,
+      "loss": 0.7571,
+      "step": 234000
+    },
+    {
+      "epoch": 6.64,
+      "grad_norm": 2.043044328689575,
+      "learning_rate": 6.710492760193818e-06,
+      "loss": 0.7547,
+      "step": 234500
+    },
+    {
+      "epoch": 6.66,
+      "grad_norm": 3.0850257873535156,
+      "learning_rate": 6.682156923861608e-06,
+      "loss": 0.7262,
+      "step": 235000
+    },
+    {
+      "epoch": 6.67,
+      "grad_norm": 3.963240385055542,
+      "learning_rate": 6.653821087529399e-06,
+      "loss": 0.7709,
+      "step": 235500
+    },
+    {
+      "epoch": 6.69,
+      "grad_norm": 2.9758195877075195,
+      "learning_rate": 6.6254852511971904e-06,
+      "loss": 0.8393,
+      "step": 236000
+    },
+    {
+      "epoch": 6.7,
+      "grad_norm": 3.1119496822357178,
+      "learning_rate": 6.5971494148649805e-06,
+      "loss": 0.8186,
+      "step": 236500
+    },
+    {
+      "epoch": 6.72,
+      "grad_norm": 0.15055912733078003,
+      "learning_rate": 6.5688135785327715e-06,
+      "loss": 0.7837,
+      "step": 237000
+    },
+    {
+      "epoch": 6.73,
+      "grad_norm": 4.089719295501709,
+      "learning_rate": 6.5404777422005616e-06,
+      "loss": 0.8045,
+      "step": 237500
+    },
+    {
+      "epoch": 6.74,
+      "grad_norm": 4.55492639541626,
+      "learning_rate": 6.5121419058683525e-06,
+      "loss": 0.7813,
+      "step": 238000
+    },
+    {
+      "epoch": 6.76,
+      "grad_norm": 4.595581531524658,
+      "learning_rate": 6.483806069536143e-06,
+      "loss": 0.8184,
+      "step": 238500
+    },
+    {
+      "epoch": 6.77,
+      "grad_norm": 1.8500324487686157,
+      "learning_rate": 6.4554702332039335e-06,
+      "loss": 0.8151,
+      "step": 239000
+    },
+    {
+      "epoch": 6.79,
+      "grad_norm": 4.76149320602417,
+      "learning_rate": 6.427134396871724e-06,
+      "loss": 0.7687,
+      "step": 239500
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 4.253459930419922,
+      "learning_rate": 6.398798560539515e-06,
+      "loss": 0.7684,
+      "step": 240000
+    },
+    {
+      "epoch": 6.81,
+      "grad_norm": 4.767841815948486,
+      "learning_rate": 6.3704627242073055e-06,
+      "loss": 0.777,
+      "step": 240500
+    },
+    {
+      "epoch": 6.83,
+      "grad_norm": 13.834861755371094,
+      "learning_rate": 6.3421268878750965e-06,
+      "loss": 0.8122,
+      "step": 241000
+    },
+    {
+      "epoch": 6.84,
+      "grad_norm": 2.1560964584350586,
+      "learning_rate": 6.3137910515428866e-06,
+      "loss": 0.8281,
+      "step": 241500
+    },
+    {
+      "epoch": 6.86,
+      "grad_norm": 3.589592218399048,
+      "learning_rate": 6.2854552152106775e-06,
+      "loss": 0.7957,
+      "step": 242000
+    },
+    {
+      "epoch": 6.87,
+      "grad_norm": 2.0722033977508545,
+      "learning_rate": 6.257119378878468e-06,
+      "loss": 0.7793,
+      "step": 242500
+    },
+    {
+      "epoch": 6.89,
+      "grad_norm": 3.5615785121917725,
+      "learning_rate": 6.2287835425462585e-06,
+      "loss": 0.7737,
+      "step": 243000
+    },
+    {
+      "epoch": 6.9,
+      "grad_norm": 3.982910394668579,
+      "learning_rate": 6.200447706214049e-06,
+      "loss": 0.749,
+      "step": 243500
+    },
+    {
+      "epoch": 6.91,
+      "grad_norm": 4.356442451477051,
+      "learning_rate": 6.17211186988184e-06,
+      "loss": 0.7996,
+      "step": 244000
+    },
+    {
+      "epoch": 6.93,
+      "grad_norm": 0.04942583665251732,
+      "learning_rate": 6.143776033549631e-06,
+      "loss": 0.7455,
+      "step": 244500
+    },
+    {
+      "epoch": 6.94,
+      "grad_norm": 1.8689194917678833,
+      "learning_rate": 6.1154401972174214e-06,
+      "loss": 0.7809,
+      "step": 245000
+    },
+    {
+      "epoch": 6.96,
+      "grad_norm": 2.919508934020996,
+      "learning_rate": 6.087104360885212e-06,
+      "loss": 0.7589,
+      "step": 245500
+    },
+    {
+      "epoch": 6.97,
+      "grad_norm": 3.5863306522369385,
+      "learning_rate": 6.0587685245530025e-06,
+      "loss": 0.7854,
+      "step": 246000
+    },
+    {
+      "epoch": 6.98,
+      "grad_norm": 1.6574618816375732,
+      "learning_rate": 6.030432688220793e-06,
+      "loss": 0.7658,
+      "step": 246500
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 4.045894145965576,
+      "learning_rate": 6.0020968518885835e-06,
+      "loss": 0.7482,
+      "step": 247000
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.7764873394183465,
+      "eval_loss": 0.7901994585990906,
+      "eval_runtime": 2745.9709,
+      "eval_samples_per_second": 34.272,
+      "eval_steps_per_second": 34.272,
+      "step": 247039
+    },
+    {
+      "epoch": 7.01,
+      "grad_norm": 3.3407113552093506,
+      "learning_rate": 5.9737610155563745e-06,
+      "loss": 0.7677,
+      "step": 247500
+    },
+    {
+      "epoch": 7.03,
+      "grad_norm": 4.4228692054748535,
+      "learning_rate": 5.945425179224165e-06,
+      "loss": 0.7874,
+      "step": 248000
+    },
+    {
+      "epoch": 7.04,
+      "grad_norm": 3.2873871326446533,
+      "learning_rate": 5.917089342891956e-06,
+      "loss": 0.7588,
+      "step": 248500
+    },
+    {
+      "epoch": 7.06,
+      "grad_norm": 2.2169015407562256,
+      "learning_rate": 5.8887535065597464e-06,
+      "loss": 0.7579,
+      "step": 249000
+    },
+    {
+      "epoch": 7.07,
+      "grad_norm": 1.9542386531829834,
+      "learning_rate": 5.860417670227537e-06,
+      "loss": 0.7618,
+      "step": 249500
+    },
+    {
+      "epoch": 7.08,
+      "grad_norm": 3.451892852783203,
+      "learning_rate": 5.8320818338953275e-06,
+      "loss": 0.7734,
+      "step": 250000
+    },
+    {
+      "epoch": 7.1,
+      "grad_norm": 4.9330034255981445,
+      "learning_rate": 5.803745997563118e-06,
+      "loss": 0.8035,
+      "step": 250500
+    },
+    {
+      "epoch": 7.11,
+      "grad_norm": 3.184312105178833,
+      "learning_rate": 5.7754101612309085e-06,
+      "loss": 0.7663,
+      "step": 251000
+    },
+    {
+      "epoch": 7.13,
+      "grad_norm": 3.387808084487915,
+      "learning_rate": 5.7470743248986994e-06,
+      "loss": 0.77,
+      "step": 251500
+    },
+    {
+      "epoch": 7.14,
+      "grad_norm": 2.910405158996582,
+      "learning_rate": 5.718738488566491e-06,
+      "loss": 0.8006,
+      "step": 252000
+    },
+    {
+      "epoch": 7.15,
+      "grad_norm": 3.2209646701812744,
+      "learning_rate": 5.690402652234281e-06,
+      "loss": 0.7645,
+      "step": 252500
+    },
+    {
+      "epoch": 7.17,
+      "grad_norm": 2.874213218688965,
+      "learning_rate": 5.662066815902072e-06,
+      "loss": 0.7371,
+      "step": 253000
+    },
+    {
+      "epoch": 7.18,
+      "grad_norm": 4.348395824432373,
+      "learning_rate": 5.633730979569862e-06,
+      "loss": 0.7177,
+      "step": 253500
+    },
+    {
+      "epoch": 7.2,
+      "grad_norm": 2.4898056983947754,
+      "learning_rate": 5.605395143237653e-06,
+      "loss": 0.7727,
+      "step": 254000
+    },
+    {
+      "epoch": 7.21,
+      "grad_norm": 4.091867923736572,
+      "learning_rate": 5.577059306905443e-06,
+      "loss": 0.8085,
+      "step": 254500
+    },
+    {
+      "epoch": 7.23,
+      "grad_norm": 4.121910095214844,
+      "learning_rate": 5.548723470573234e-06,
+      "loss": 0.7768,
+      "step": 255000
+    },
+    {
+      "epoch": 7.24,
+      "grad_norm": 2.1994853019714355,
+      "learning_rate": 5.5203876342410244e-06,
+      "loss": 0.7847,
+      "step": 255500
+    },
+    {
+      "epoch": 7.25,
+      "grad_norm": 3.1094110012054443,
+      "learning_rate": 5.492051797908816e-06,
+      "loss": 0.7496,
+      "step": 256000
+    },
+    {
+      "epoch": 7.27,
+      "grad_norm": 4.486825942993164,
+      "learning_rate": 5.463715961576606e-06,
+      "loss": 0.7434,
+      "step": 256500
+    },
+    {
+      "epoch": 7.28,
+      "grad_norm": 4.349071502685547,
+      "learning_rate": 5.435380125244397e-06,
+      "loss": 0.7563,
+      "step": 257000
+    },
+    {
+      "epoch": 7.3,
+      "grad_norm": 2.1635329723358154,
+      "learning_rate": 5.407044288912187e-06,
+      "loss": 0.8022,
+      "step": 257500
+    },
+    {
+      "epoch": 7.31,
+      "grad_norm": 3.0839200019836426,
+      "learning_rate": 5.378708452579978e-06,
+      "loss": 0.7507,
+      "step": 258000
+    },
+    {
+      "epoch": 7.32,
+      "grad_norm": 3.0056662559509277,
+      "learning_rate": 5.350372616247768e-06,
+      "loss": 0.7807,
+      "step": 258500
+    },
+    {
+      "epoch": 7.34,
+      "grad_norm": 3.784560441970825,
+      "learning_rate": 5.322036779915559e-06,
+      "loss": 0.7763,
+      "step": 259000
+    },
+    {
+      "epoch": 7.35,
+      "grad_norm": 4.388574123382568,
+      "learning_rate": 5.293700943583351e-06,
+      "loss": 0.8446,
+      "step": 259500
+    },
+    {
+      "epoch": 7.37,
+      "grad_norm": 3.1023905277252197,
+      "learning_rate": 5.265365107251141e-06,
+      "loss": 0.7713,
+      "step": 260000
+    },
+    {
+      "epoch": 7.38,
+      "grad_norm": 4.7277045249938965,
+      "learning_rate": 5.237029270918932e-06,
+      "loss": 0.7975,
+      "step": 260500
+    },
+    {
+      "epoch": 7.4,
+      "grad_norm": 5.080424785614014,
+      "learning_rate": 5.208693434586722e-06,
+      "loss": 0.7836,
+      "step": 261000
+    },
+    {
+      "epoch": 7.41,
+      "grad_norm": 3.0420889854431152,
+      "learning_rate": 5.180357598254513e-06,
+      "loss": 0.8012,
+      "step": 261500
+    },
+    {
+      "epoch": 7.42,
+      "grad_norm": 2.417198657989502,
+      "learning_rate": 5.152021761922303e-06,
+      "loss": 0.8363,
+      "step": 262000
+    },
+    {
+      "epoch": 7.44,
+      "grad_norm": 3.7752904891967773,
+      "learning_rate": 5.123685925590094e-06,
+      "loss": 0.7595,
+      "step": 262500
+    },
+    {
+      "epoch": 7.45,
+      "grad_norm": 2.991297721862793,
+      "learning_rate": 5.095350089257884e-06,
+      "loss": 0.7564,
+      "step": 263000
+    },
+    {
+      "epoch": 7.47,
+      "grad_norm": 2.7910547256469727,
+      "learning_rate": 5.067014252925676e-06,
+      "loss": 0.7458,
+      "step": 263500
+    },
+    {
+      "epoch": 7.48,
+      "grad_norm": 3.118992567062378,
+      "learning_rate": 5.038678416593466e-06,
+      "loss": 0.8259,
+      "step": 264000
+    },
+    {
+      "epoch": 7.49,
+      "grad_norm": 2.306201696395874,
+      "learning_rate": 5.010342580261257e-06,
+      "loss": 0.7895,
+      "step": 264500
+    },
+    {
+      "epoch": 7.51,
+      "grad_norm": 2.723785638809204,
+      "learning_rate": 4.982006743929047e-06,
+      "loss": 0.8224,
+      "step": 265000
+    },
+    {
+      "epoch": 7.52,
+      "grad_norm": 5.79103422164917,
+      "learning_rate": 4.953670907596838e-06,
+      "loss": 0.7698,
+      "step": 265500
+    },
+    {
+      "epoch": 7.54,
+      "grad_norm": 2.896071195602417,
+      "learning_rate": 4.925335071264628e-06,
+      "loss": 0.8117,
+      "step": 266000
+    },
+    {
+      "epoch": 7.55,
+      "grad_norm": 4.713535308837891,
+      "learning_rate": 4.89699923493242e-06,
+      "loss": 0.8028,
+      "step": 266500
+    },
+    {
+      "epoch": 7.57,
+      "grad_norm": 3.329005241394043,
+      "learning_rate": 4.86866339860021e-06,
+      "loss": 0.762,
+      "step": 267000
+    },
+    {
+      "epoch": 7.58,
+      "grad_norm": 2.388073444366455,
+      "learning_rate": 4.840327562268001e-06,
+      "loss": 0.7489,
+      "step": 267500
+    },
+    {
+      "epoch": 7.59,
+      "grad_norm": 2.9465584754943848,
+      "learning_rate": 4.811991725935791e-06,
+      "loss": 0.7755,
+      "step": 268000
+    },
+    {
+      "epoch": 7.61,
+      "grad_norm": 4.375637531280518,
+      "learning_rate": 4.783655889603582e-06,
+      "loss": 0.7994,
+      "step": 268500
+    },
+    {
+      "epoch": 7.62,
+      "grad_norm": 3.1931121349334717,
+      "learning_rate": 4.755320053271373e-06,
+      "loss": 0.8111,
+      "step": 269000
+    },
+    {
+      "epoch": 7.64,
+      "grad_norm": 4.697356700897217,
+      "learning_rate": 4.726984216939163e-06,
+      "loss": 0.7568,
+      "step": 269500
+    },
+    {
+      "epoch": 7.65,
+      "grad_norm": 3.1443934440612793,
+      "learning_rate": 4.698648380606954e-06,
+      "loss": 0.7723,
+      "step": 270000
+    },
+    {
+      "epoch": 7.66,
+      "grad_norm": 4.355726718902588,
+      "learning_rate": 4.670312544274745e-06,
+      "loss": 0.7451,
+      "step": 270500
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 3.7708301544189453,
+      "learning_rate": 4.641976707942535e-06,
+      "loss": 0.7662,
+      "step": 271000
+    },
+    {
+      "epoch": 7.69,
+      "grad_norm": 3.0741500854492188,
+      "learning_rate": 4.613640871610326e-06,
+      "loss": 0.7939,
+      "step": 271500
+    },
+    {
+      "epoch": 7.71,
+      "grad_norm": 4.382043361663818,
+      "learning_rate": 4.585305035278116e-06,
+      "loss": 0.7949,
+      "step": 272000
+    },
+    {
+      "epoch": 7.72,
+      "grad_norm": 0.0032790284603834152,
+      "learning_rate": 4.556969198945907e-06,
+      "loss": 0.776,
+      "step": 272500
+    },
+    {
+      "epoch": 7.74,
+      "grad_norm": 4.499131679534912,
+      "learning_rate": 4.528633362613698e-06,
+      "loss": 0.7683,
+      "step": 273000
+    },
+    {
+      "epoch": 7.75,
+      "grad_norm": 2.200047016143799,
+      "learning_rate": 4.500297526281488e-06,
+      "loss": 0.8006,
+      "step": 273500
+    },
+    {
+      "epoch": 7.76,
+      "grad_norm": 3.781383752822876,
+      "learning_rate": 4.471961689949279e-06,
+      "loss": 0.7454,
+      "step": 274000
+    },
+    {
+      "epoch": 7.78,
+      "grad_norm": 3.239480972290039,
+      "learning_rate": 4.44362585361707e-06,
+      "loss": 0.7665,
+      "step": 274500
+    },
+    {
+      "epoch": 7.79,
+      "grad_norm": 0.20325414836406708,
+      "learning_rate": 4.415290017284861e-06,
+      "loss": 0.8246,
+      "step": 275000
+    },
+    {
+      "epoch": 7.81,
+      "grad_norm": 3.527724504470825,
+      "learning_rate": 4.386954180952651e-06,
+      "loss": 0.7981,
+      "step": 275500
+    },
+    {
+      "epoch": 7.82,
+      "grad_norm": 4.077188968658447,
+      "learning_rate": 4.358618344620442e-06,
+      "loss": 0.7873,
+      "step": 276000
+    },
+    {
+      "epoch": 7.83,
+      "grad_norm": 2.584451675415039,
+      "learning_rate": 4.330282508288233e-06,
+      "loss": 0.8097,
+      "step": 276500
+    },
+    {
+      "epoch": 7.85,
+      "grad_norm": 3.91054368019104,
+      "learning_rate": 4.301946671956023e-06,
+      "loss": 0.781,
+      "step": 277000
+    },
+    {
+      "epoch": 7.86,
+      "grad_norm": 1.5081390142440796,
+      "learning_rate": 4.273610835623814e-06,
+      "loss": 0.7504,
+      "step": 277500
+    },
+    {
+      "epoch": 7.88,
+      "grad_norm": 2.7629621028900146,
+      "learning_rate": 4.245274999291604e-06,
+      "loss": 0.7698,
+      "step": 278000
+    },
+    {
+      "epoch": 7.89,
+      "grad_norm": 0.04077434912323952,
+      "learning_rate": 4.216939162959395e-06,
+      "loss": 0.794,
+      "step": 278500
+    },
+    {
+      "epoch": 7.91,
+      "grad_norm": 3.088022232055664,
+      "learning_rate": 4.188603326627186e-06,
+      "loss": 0.7957,
+      "step": 279000
+    },
+    {
+      "epoch": 7.92,
+      "grad_norm": 0.009913645684719086,
+      "learning_rate": 4.160267490294976e-06,
+      "loss": 0.7477,
+      "step": 279500
+    },
+    {
+      "epoch": 7.93,
+      "grad_norm": 3.2786879539489746,
+      "learning_rate": 4.131931653962767e-06,
+      "loss": 0.7746,
+      "step": 280000
+    },
+    {
+      "epoch": 7.95,
+      "grad_norm": 4.847500801086426,
+      "learning_rate": 4.103595817630558e-06,
+      "loss": 0.7769,
+      "step": 280500
+    },
+    {
+      "epoch": 7.96,
+      "grad_norm": 3.0839521884918213,
+      "learning_rate": 4.075259981298348e-06,
+      "loss": 0.7969,
+      "step": 281000
+    },
+    {
+      "epoch": 7.98,
+      "grad_norm": 5.6063337326049805,
+      "learning_rate": 4.046924144966139e-06,
+      "loss": 0.8007,
+      "step": 281500
+    },
+    {
+      "epoch": 7.99,
+      "grad_norm": 2.7218992710113525,
+      "learning_rate": 4.018588308633929e-06,
+      "loss": 0.777,
+      "step": 282000
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.7773267737033928,
+      "eval_loss": 0.7822893857955933,
+      "eval_runtime": 2744.096,
+      "eval_samples_per_second": 34.296,
+      "eval_steps_per_second": 34.296,
+      "step": 282331
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 3.1049580574035645,
+      "learning_rate": 3.990252472301721e-06,
+      "loss": 0.792,
+      "step": 282500
+    },
+    {
+      "epoch": 8.02,
+      "grad_norm": 2.2240233421325684,
+      "learning_rate": 3.961916635969511e-06,
+      "loss": 0.7679,
+      "step": 283000
+    },
+    {
+      "epoch": 8.03,
+      "grad_norm": 3.7481648921966553,
+      "learning_rate": 3.933580799637302e-06,
+      "loss": 0.783,
+      "step": 283500
+    },
+    {
+      "epoch": 8.05,
+      "grad_norm": 2.1633012294769287,
+      "learning_rate": 3.905244963305092e-06,
+      "loss": 0.7946,
+      "step": 284000
+    },
+    {
+      "epoch": 8.06,
+      "grad_norm": 2.2068793773651123,
+      "learning_rate": 3.876909126972883e-06,
+      "loss": 0.739,
+      "step": 284500
+    },
+    {
+      "epoch": 8.08,
+      "grad_norm": 2.721085548400879,
+      "learning_rate": 3.848573290640674e-06,
+      "loss": 0.7119,
+      "step": 285000
+    },
+    {
+      "epoch": 8.09,
+      "grad_norm": 4.786925792694092,
+      "learning_rate": 3.820237454308464e-06,
+      "loss": 0.7804,
+      "step": 285500
+    },
+    {
+      "epoch": 8.1,
+      "grad_norm": 2.967390537261963,
+      "learning_rate": 3.7919016179762545e-06,
+      "loss": 0.8006,
+      "step": 286000
+    },
+    {
+      "epoch": 8.12,
+      "grad_norm": 3.6855547428131104,
+      "learning_rate": 3.763565781644046e-06,
+      "loss": 0.7571,
+      "step": 286500
+    },
+    {
+      "epoch": 8.13,
+      "grad_norm": 3.163161516189575,
+      "learning_rate": 3.7352299453118364e-06,
+      "loss": 0.7706,
+      "step": 287000
+    },
+    {
+      "epoch": 8.15,
+      "grad_norm": 2.217069387435913,
+      "learning_rate": 3.706894108979627e-06,
+      "loss": 0.7855,
+      "step": 287500
+    },
+    {
+      "epoch": 8.16,
+      "grad_norm": 3.2176852226257324,
+      "learning_rate": 3.6785582726474174e-06,
+      "loss": 0.7804,
+      "step": 288000
+    },
+    {
+      "epoch": 8.17,
+      "grad_norm": 4.42471981048584,
+      "learning_rate": 3.6502224363152084e-06,
+      "loss": 0.7368,
+      "step": 288500
+    },
+    {
+      "epoch": 8.19,
+      "grad_norm": 3.7135393619537354,
+      "learning_rate": 3.621886599982999e-06,
+      "loss": 0.7546,
+      "step": 289000
+    },
+    {
+      "epoch": 8.2,
+      "grad_norm": 1.9810535907745361,
+      "learning_rate": 3.5935507636507894e-06,
+      "loss": 0.7748,
+      "step": 289500
+    },
+    {
+      "epoch": 8.22,
+      "grad_norm": 2.208085060119629,
+      "learning_rate": 3.56521492731858e-06,
+      "loss": 0.7382,
+      "step": 290000
+    },
+    {
+      "epoch": 8.23,
+      "grad_norm": 4.069818019866943,
+      "learning_rate": 3.536879090986371e-06,
+      "loss": 0.7717,
+      "step": 290500
+    },
+    {
+      "epoch": 8.25,
+      "grad_norm": 3.083171844482422,
+      "learning_rate": 3.5085432546541614e-06,
+      "loss": 0.7637,
+      "step": 291000
+    },
+    {
+      "epoch": 8.26,
+      "grad_norm": 1.8702301979064941,
+      "learning_rate": 3.480207418321952e-06,
+      "loss": 0.7632,
+      "step": 291500
+    },
+    {
+      "epoch": 8.27,
+      "grad_norm": 2.7875142097473145,
+      "learning_rate": 3.4518715819897424e-06,
+      "loss": 0.7447,
+      "step": 292000
+    },
+    {
+      "epoch": 8.29,
+      "grad_norm": 2.318441152572632,
+      "learning_rate": 3.4235357456575333e-06,
+      "loss": 0.7445,
+      "step": 292500
+    },
+    {
+      "epoch": 8.3,
+      "grad_norm": 3.098846912384033,
+      "learning_rate": 3.395199909325324e-06,
+      "loss": 0.7527,
+      "step": 293000
+    },
+    {
+      "epoch": 8.32,
+      "grad_norm": 4.272964954376221,
+      "learning_rate": 3.3668640729931144e-06,
+      "loss": 0.7604,
+      "step": 293500
+    },
+    {
+      "epoch": 8.33,
+      "grad_norm": 2.8859071731567383,
+      "learning_rate": 3.338528236660905e-06,
+      "loss": 0.7715,
+      "step": 294000
+    },
+    {
+      "epoch": 8.34,
+      "grad_norm": 2.219268560409546,
+      "learning_rate": 3.3101924003286963e-06,
+      "loss": 0.7369,
+      "step": 294500
+    },
+    {
+      "epoch": 8.36,
+      "grad_norm": 3.0765175819396973,
+      "learning_rate": 3.2818565639964868e-06,
+      "loss": 0.7948,
+      "step": 295000
+    },
+    {
+      "epoch": 8.37,
+      "grad_norm": 3.4554049968719482,
+      "learning_rate": 3.2535207276642773e-06,
+      "loss": 0.7596,
+      "step": 295500
+    },
+    {
+      "epoch": 8.39,
+      "grad_norm": 3.8051280975341797,
+      "learning_rate": 3.225184891332068e-06,
+      "loss": 0.792,
+      "step": 296000
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 2.201718330383301,
+      "learning_rate": 3.1968490549998588e-06,
+      "loss": 0.7968,
+      "step": 296500
+    },
+    {
+      "epoch": 8.42,
+      "grad_norm": 3.527601718902588,
+      "learning_rate": 3.1685132186676493e-06,
+      "loss": 0.7582,
+      "step": 297000
+    },
+    {
+      "epoch": 8.43,
+      "grad_norm": 3.275254726409912,
+      "learning_rate": 3.14017738233544e-06,
+      "loss": 0.8061,
+      "step": 297500
+    },
+    {
+      "epoch": 8.44,
+      "grad_norm": 2.419063091278076,
+      "learning_rate": 3.1118415460032303e-06,
+      "loss": 0.7874,
+      "step": 298000
+    },
+    {
+      "epoch": 8.46,
+      "grad_norm": 2.1207940578460693,
+      "learning_rate": 3.0835057096710213e-06,
+      "loss": 0.7635,
+      "step": 298500
+    },
+    {
+      "epoch": 8.47,
+      "grad_norm": 3.7095024585723877,
+      "learning_rate": 3.0551698733388118e-06,
+      "loss": 0.7789,
+      "step": 299000
+    },
+    {
+      "epoch": 8.49,
+      "grad_norm": 3.1537423133850098,
+      "learning_rate": 3.0268340370066023e-06,
+      "loss": 0.7847,
+      "step": 299500
+    },
+    {
+      "epoch": 8.5,
+      "grad_norm": 2.2112996578216553,
+      "learning_rate": 2.998498200674393e-06,
+      "loss": 0.7716,
+      "step": 300000
+    },
+    {
+      "epoch": 8.51,
+      "grad_norm": 2.808809280395508,
+      "learning_rate": 2.970162364342184e-06,
+      "loss": 0.7903,
+      "step": 300500
+    },
+    {
+      "epoch": 8.53,
+      "grad_norm": 2.72741961479187,
+      "learning_rate": 2.9418265280099747e-06,
+      "loss": 0.7859,
+      "step": 301000
+    },
+    {
+      "epoch": 8.54,
+      "grad_norm": 2.9039785861968994,
+      "learning_rate": 2.9134906916777648e-06,
+      "loss": 0.7756,
+      "step": 301500
+    },
+    {
+      "epoch": 8.56,
+      "grad_norm": 0.06406772136688232,
+      "learning_rate": 2.8851548553455553e-06,
+      "loss": 0.7672,
+      "step": 302000
+    },
+    {
+      "epoch": 8.57,
+      "grad_norm": 0.002435769187286496,
+      "learning_rate": 2.8568190190133467e-06,
+      "loss": 0.7799,
+      "step": 302500
+    },
+    {
+      "epoch": 8.59,
+      "grad_norm": 2.7473349571228027,
+      "learning_rate": 2.828483182681137e-06,
+      "loss": 0.7607,
+      "step": 303000
+    },
+    {
+      "epoch": 8.6,
+      "grad_norm": 4.246364116668701,
+      "learning_rate": 2.8001473463489277e-06,
+      "loss": 0.7616,
+      "step": 303500
+    },
+    {
+      "epoch": 8.61,
+      "grad_norm": 4.9080891609191895,
+      "learning_rate": 2.7718115100167182e-06,
+      "loss": 0.7813,
+      "step": 304000
+    },
+    {
+      "epoch": 8.63,
+      "grad_norm": 2.181150197982788,
+      "learning_rate": 2.743475673684509e-06,
+      "loss": 0.7604,
+      "step": 304500
+    },
+    {
+      "epoch": 8.64,
+      "grad_norm": 4.190260410308838,
+      "learning_rate": 2.7151398373522997e-06,
+      "loss": 0.7934,
+      "step": 305000
+    },
+    {
+      "epoch": 8.66,
+      "grad_norm": 3.007934331893921,
+      "learning_rate": 2.68680400102009e-06,
+      "loss": 0.7615,
+      "step": 305500
+    },
+    {
+      "epoch": 8.67,
+      "grad_norm": 3.8845651149749756,
+      "learning_rate": 2.6584681646878807e-06,
+      "loss": 0.7446,
+      "step": 306000
+    },
+    {
+      "epoch": 8.68,
+      "grad_norm": 3.1428420543670654,
+      "learning_rate": 2.6301323283556717e-06,
+      "loss": 0.7861,
+      "step": 306500
+    },
+    {
+      "epoch": 8.7,
+      "grad_norm": 2.3537344932556152,
+      "learning_rate": 2.601796492023462e-06,
+      "loss": 0.7959,
+      "step": 307000
+    },
+    {
+      "epoch": 8.71,
+      "grad_norm": 2.065415143966675,
+      "learning_rate": 2.5734606556912527e-06,
+      "loss": 0.7744,
+      "step": 307500
+    },
+    {
+      "epoch": 8.73,
+      "grad_norm": 2.2324647903442383,
+      "learning_rate": 2.545124819359043e-06,
+      "loss": 0.7655,
+      "step": 308000
+    },
+    {
+      "epoch": 8.74,
+      "grad_norm": 3.919121742248535,
+      "learning_rate": 2.5167889830268346e-06,
+      "loss": 0.7631,
+      "step": 308500
+    },
+    {
+      "epoch": 8.76,
+      "grad_norm": 1.9113390445709229,
+      "learning_rate": 2.488453146694625e-06,
+      "loss": 0.7907,
+      "step": 309000
+    },
+    {
+      "epoch": 8.77,
+      "grad_norm": 3.220702648162842,
+      "learning_rate": 2.4601173103624156e-06,
+      "loss": 0.7797,
+      "step": 309500
+    },
+    {
+      "epoch": 8.78,
+      "grad_norm": 3.0107169151306152,
+      "learning_rate": 2.431781474030206e-06,
+      "loss": 0.7633,
+      "step": 310000
+    },
+    {
+      "epoch": 8.8,
+      "grad_norm": 3.342789649963379,
+      "learning_rate": 2.403445637697997e-06,
+      "loss": 0.7681,
+      "step": 310500
+    },
+    {
+      "epoch": 8.81,
+      "grad_norm": 4.381540298461914,
+      "learning_rate": 2.3751098013657876e-06,
+      "loss": 0.7389,
+      "step": 311000
+    },
+    {
+      "epoch": 8.83,
+      "grad_norm": 2.1664342880249023,
+      "learning_rate": 2.346773965033578e-06,
+      "loss": 0.7606,
+      "step": 311500
+    },
+    {
+      "epoch": 8.84,
+      "grad_norm": 1.6208741664886475,
+      "learning_rate": 2.3184381287013686e-06,
+      "loss": 0.7427,
+      "step": 312000
+    },
+    {
+      "epoch": 8.85,
+      "grad_norm": 3.916785955429077,
+      "learning_rate": 2.2901022923691596e-06,
+      "loss": 0.7548,
+      "step": 312500
+    },
+    {
+      "epoch": 8.87,
+      "grad_norm": 4.270482063293457,
+      "learning_rate": 2.26176645603695e-06,
+      "loss": 0.7702,
+      "step": 313000
+    },
+    {
+      "epoch": 8.88,
+      "grad_norm": 2.844374418258667,
+      "learning_rate": 2.233430619704741e-06,
+      "loss": 0.7754,
+      "step": 313500
+    },
+    {
+      "epoch": 8.9,
+      "grad_norm": 5.132666110992432,
+      "learning_rate": 2.2050947833725315e-06,
+      "loss": 0.7667,
+      "step": 314000
+    },
+    {
+      "epoch": 8.91,
+      "grad_norm": 3.2758193016052246,
+      "learning_rate": 2.176758947040322e-06,
+      "loss": 0.792,
+      "step": 314500
+    },
+    {
+      "epoch": 8.93,
+      "grad_norm": 0.15504547953605652,
+      "learning_rate": 2.1484231107081126e-06,
+      "loss": 0.7549,
+      "step": 315000
+    },
+    {
+      "epoch": 8.94,
+      "grad_norm": 4.870330333709717,
+      "learning_rate": 2.1200872743759035e-06,
+      "loss": 0.7631,
+      "step": 315500
+    },
+    {
+      "epoch": 8.95,
+      "grad_norm": 7.014795303344727,
+      "learning_rate": 2.091751438043694e-06,
+      "loss": 0.7784,
+      "step": 316000
+    },
+    {
+      "epoch": 8.97,
+      "grad_norm": 9.451394081115723,
+      "learning_rate": 2.063415601711485e-06,
+      "loss": 0.7642,
+      "step": 316500
+    },
+    {
+      "epoch": 8.98,
+      "grad_norm": 0.6156451106071472,
+      "learning_rate": 2.0350797653792755e-06,
+      "loss": 0.7692,
+      "step": 317000
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 4.711402893066406,
+      "learning_rate": 2.006743929047066e-06,
+      "loss": 0.8045,
+      "step": 317500
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.7766467256750008,
+      "eval_loss": 0.7740858197212219,
+      "eval_runtime": 2742.9816,
+      "eval_samples_per_second": 34.31,
+      "eval_steps_per_second": 34.31,
+      "step": 317622
+    },
+    {
+      "epoch": 9.01,
+      "grad_norm": 3.5747580528259277,
+      "learning_rate": 1.9784080927148565e-06,
+      "loss": 0.75,
+      "step": 318000
+    },
+    {
+      "epoch": 9.02,
+      "grad_norm": 2.482762575149536,
+      "learning_rate": 1.9500722563826475e-06,
+      "loss": 0.7759,
+      "step": 318500
+    },
+    {
+      "epoch": 9.04,
+      "grad_norm": 3.532433032989502,
+      "learning_rate": 1.921736420050438e-06,
+      "loss": 0.79,
+      "step": 319000
+    },
+    {
+      "epoch": 9.05,
+      "grad_norm": 1.5587611198425293,
+      "learning_rate": 1.8934005837182287e-06,
+      "loss": 0.7424,
+      "step": 319500
+    },
+    {
+      "epoch": 9.07,
+      "grad_norm": 3.8560965061187744,
+      "learning_rate": 1.8650647473860192e-06,
+      "loss": 0.7464,
+      "step": 320000
+    },
+    {
+      "epoch": 9.08,
+      "grad_norm": 3.5473949909210205,
+      "learning_rate": 1.83672891105381e-06,
+      "loss": 0.7857,
+      "step": 320500
+    },
+    {
+      "epoch": 9.1,
+      "grad_norm": 2.0293056964874268,
+      "learning_rate": 1.8083930747216005e-06,
+      "loss": 0.7538,
+      "step": 321000
+    },
+    {
+      "epoch": 9.11,
+      "grad_norm": 5.718314170837402,
+      "learning_rate": 1.7800572383893912e-06,
+      "loss": 0.7467,
+      "step": 321500
+    },
+    {
+      "epoch": 9.12,
+      "grad_norm": 3.7162270545959473,
+      "learning_rate": 1.7517214020571817e-06,
+      "loss": 0.7627,
+      "step": 322000
+    },
+    {
+      "epoch": 9.14,
+      "grad_norm": 3.278439998626709,
+      "learning_rate": 1.7233855657249727e-06,
+      "loss": 0.7898,
+      "step": 322500
+    },
+    {
+      "epoch": 9.15,
+      "grad_norm": 3.3073856830596924,
+      "learning_rate": 1.6950497293927632e-06,
+      "loss": 0.7535,
+      "step": 323000
+    },
+    {
+      "epoch": 9.17,
+      "grad_norm": 0.3235242962837219,
+      "learning_rate": 1.666713893060554e-06,
+      "loss": 0.7559,
+      "step": 323500
+    },
+    {
+      "epoch": 9.18,
+      "grad_norm": 3.050361394882202,
+      "learning_rate": 1.6383780567283444e-06,
+      "loss": 0.7876,
+      "step": 324000
+    },
+    {
+      "epoch": 9.19,
+      "grad_norm": 2.169379472732544,
+      "learning_rate": 1.6100422203961352e-06,
+      "loss": 0.7864,
+      "step": 324500
+    },
+    {
+      "epoch": 9.21,
+      "grad_norm": 2.979144811630249,
+      "learning_rate": 1.5817063840639257e-06,
+      "loss": 0.7855,
+      "step": 325000
+    },
+    {
+      "epoch": 9.22,
+      "grad_norm": 2.187973976135254,
+      "learning_rate": 1.5533705477317166e-06,
+      "loss": 0.7744,
+      "step": 325500
+    },
+    {
+      "epoch": 9.24,
+      "grad_norm": 2.242929220199585,
+      "learning_rate": 1.525034711399507e-06,
+      "loss": 0.7929,
+      "step": 326000
+    },
+    {
+      "epoch": 9.25,
+      "grad_norm": 3.240936756134033,
+      "learning_rate": 1.4966988750672979e-06,
+      "loss": 0.7504,
+      "step": 326500
+    },
+    {
+      "epoch": 9.27,
+      "grad_norm": 5.772107124328613,
+      "learning_rate": 1.4683630387350884e-06,
+      "loss": 0.7653,
+      "step": 327000
+    },
+    {
+      "epoch": 9.28,
+      "grad_norm": 3.1591999530792236,
+      "learning_rate": 1.4400272024028791e-06,
+      "loss": 0.7102,
+      "step": 327500
+    },
+    {
+      "epoch": 9.29,
+      "grad_norm": 2.9955251216888428,
+      "learning_rate": 1.4116913660706696e-06,
+      "loss": 0.7689,
+      "step": 328000
+    },
+    {
+      "epoch": 9.31,
+      "grad_norm": 3.1016738414764404,
+      "learning_rate": 1.3833555297384604e-06,
+      "loss": 0.7729,
+      "step": 328500
+    },
+    {
+      "epoch": 9.32,
+      "grad_norm": 4.220294952392578,
+      "learning_rate": 1.3550196934062509e-06,
+      "loss": 0.7901,
+      "step": 329000
+    },
+    {
+      "epoch": 9.34,
+      "grad_norm": 2.13430118560791,
+      "learning_rate": 1.3266838570740418e-06,
+      "loss": 0.764,
+      "step": 329500
+    },
+    {
+      "epoch": 9.35,
+      "grad_norm": 3.9360506534576416,
+      "learning_rate": 1.2983480207418323e-06,
+      "loss": 0.7443,
+      "step": 330000
+    },
+    {
+      "epoch": 9.36,
+      "grad_norm": 2.117910623550415,
+      "learning_rate": 1.270012184409623e-06,
+      "loss": 0.7349,
+      "step": 330500
+    },
+    {
+      "epoch": 9.38,
+      "grad_norm": 4.577652931213379,
+      "learning_rate": 1.2416763480774136e-06,
+      "loss": 0.7891,
+      "step": 331000
+    },
+    {
+      "epoch": 9.39,
+      "grad_norm": 4.693737506866455,
+      "learning_rate": 1.2133405117452041e-06,
+      "loss": 0.7812,
+      "step": 331500
+    },
+    {
+      "epoch": 9.41,
+      "grad_norm": 3.70219349861145,
+      "learning_rate": 1.1850046754129948e-06,
+      "loss": 0.77,
+      "step": 332000
+    },
+    {
+      "epoch": 9.42,
+      "grad_norm": 3.6610214710235596,
+      "learning_rate": 1.1566688390807856e-06,
+      "loss": 0.7,
+      "step": 332500
+    },
+    {
+      "epoch": 9.44,
+      "grad_norm": 2.40402889251709,
+      "learning_rate": 1.128333002748576e-06,
+      "loss": 0.7968,
+      "step": 333000
+    },
+    {
+      "epoch": 9.45,
+      "grad_norm": 3.7406198978424072,
+      "learning_rate": 1.0999971664163668e-06,
+      "loss": 0.7786,
+      "step": 333500
+    },
+    {
+      "epoch": 9.46,
+      "grad_norm": 4.944614410400391,
+      "learning_rate": 1.0716613300841576e-06,
+      "loss": 0.7554,
+      "step": 334000
+    },
+    {
+      "epoch": 9.48,
+      "grad_norm": 0.21894210577011108,
+      "learning_rate": 1.043325493751948e-06,
+      "loss": 0.7585,
+      "step": 334500
+    },
+    {
+      "epoch": 9.49,
+      "grad_norm": 4.879526138305664,
+      "learning_rate": 1.0149896574197388e-06,
+      "loss": 0.767,
+      "step": 335000
+    },
+    {
+      "epoch": 9.51,
+      "grad_norm": 3.0120060443878174,
+      "learning_rate": 9.866538210875293e-07,
+      "loss": 0.7692,
+      "step": 335500
+    },
+    {
+      "epoch": 9.52,
+      "grad_norm": 2.9848334789276123,
+      "learning_rate": 9.5831798475532e-07,
+      "loss": 0.8005,
+      "step": 336000
+    },
+    {
+      "epoch": 9.53,
+      "grad_norm": 4.2996110916137695,
+      "learning_rate": 9.299821484231109e-07,
+      "loss": 0.7887,
+      "step": 336500
+    },
+    {
+      "epoch": 9.55,
+      "grad_norm": 3.8129289150238037,
+      "learning_rate": 9.016463120909015e-07,
+      "loss": 0.7608,
+      "step": 337000
+    },
+    {
+      "epoch": 9.56,
+      "grad_norm": 3.8413565158843994,
+      "learning_rate": 8.733104757586921e-07,
+      "loss": 0.7477,
+      "step": 337500
+    },
+    {
+      "epoch": 9.58,
+      "grad_norm": 2.2064590454101562,
+      "learning_rate": 8.449746394264829e-07,
+      "loss": 0.7587,
+      "step": 338000
+    },
+    {
+      "epoch": 9.59,
+      "grad_norm": 5.598100662231445,
+      "learning_rate": 8.166388030942735e-07,
+      "loss": 0.782,
+      "step": 338500
+    },
+    {
+      "epoch": 9.61,
+      "grad_norm": 3.09802508354187,
+      "learning_rate": 7.883029667620641e-07,
+      "loss": 0.7799,
+      "step": 339000
+    },
+    {
+      "epoch": 9.62,
+      "grad_norm": 2.217261791229248,
+      "learning_rate": 7.599671304298547e-07,
+      "loss": 0.744,
+      "step": 339500
+    },
+    {
+      "epoch": 9.63,
+      "grad_norm": 2.2989118099212646,
+      "learning_rate": 7.316312940976455e-07,
+      "loss": 0.7724,
+      "step": 340000
+    },
+    {
+      "epoch": 9.65,
+      "grad_norm": 4.708602428436279,
+      "learning_rate": 7.032954577654361e-07,
+      "loss": 0.7564,
+      "step": 340500
+    },
+    {
+      "epoch": 9.66,
+      "grad_norm": 2.177694797515869,
+      "learning_rate": 6.749596214332267e-07,
+      "loss": 0.7935,
+      "step": 341000
+    },
+    {
+      "epoch": 9.68,
+      "grad_norm": 3.017601728439331,
+      "learning_rate": 6.466237851010173e-07,
+      "loss": 0.7764,
+      "step": 341500
+    },
+    {
+      "epoch": 9.69,
+      "grad_norm": 4.6326494216918945,
+      "learning_rate": 6.18287948768808e-07,
+      "loss": 0.7858,
+      "step": 342000
+    },
+    {
+      "epoch": 9.7,
+      "grad_norm": 3.898557424545288,
+      "learning_rate": 5.899521124365986e-07,
+      "loss": 0.7539,
+      "step": 342500
+    },
+    {
+      "epoch": 9.72,
+      "grad_norm": 3.546487808227539,
+      "learning_rate": 5.616162761043892e-07,
+      "loss": 0.7912,
+      "step": 343000
+    },
+    {
+      "epoch": 9.73,
+      "grad_norm": 3.868692398071289,
+      "learning_rate": 5.332804397721799e-07,
+      "loss": 0.7661,
+      "step": 343500
+    },
+    {
+      "epoch": 9.75,
+      "grad_norm": 4.169196605682373,
+      "learning_rate": 5.049446034399706e-07,
+      "loss": 0.7642,
+      "step": 344000
+    },
+    {
+      "epoch": 9.76,
+      "grad_norm": 0.08999128639698029,
+      "learning_rate": 4.766087671077612e-07,
+      "loss": 0.7434,
+      "step": 344500
+    },
+    {
+      "epoch": 9.78,
+      "grad_norm": 3.686347007751465,
+      "learning_rate": 4.4827293077555185e-07,
+      "loss": 0.7124,
+      "step": 345000
+    },
+    {
+      "epoch": 9.79,
+      "grad_norm": 4.009047031402588,
+      "learning_rate": 4.1993709444334253e-07,
+      "loss": 0.75,
+      "step": 345500
+    },
+    {
+      "epoch": 9.8,
+      "grad_norm": 3.0990748405456543,
+      "learning_rate": 3.9160125811113315e-07,
+      "loss": 0.7539,
+      "step": 346000
+    },
+    {
+      "epoch": 9.82,
+      "grad_norm": 2.179380178451538,
+      "learning_rate": 3.6326542177892383e-07,
+      "loss": 0.7654,
+      "step": 346500
+    },
+    {
+      "epoch": 9.83,
+      "grad_norm": 4.667028903961182,
+      "learning_rate": 3.3492958544671445e-07,
+      "loss": 0.7645,
+      "step": 347000
+    },
+    {
+      "epoch": 9.85,
+      "grad_norm": 0.0834590345621109,
+      "learning_rate": 3.0659374911450513e-07,
+      "loss": 0.74,
+      "step": 347500
+    },
+    {
+      "epoch": 9.86,
+      "grad_norm": 1.7832657098770142,
+      "learning_rate": 2.782579127822958e-07,
+      "loss": 0.7738,
+      "step": 348000
+    },
+    {
+      "epoch": 9.87,
+      "grad_norm": 3.8339791297912598,
+      "learning_rate": 2.4992207645008643e-07,
+      "loss": 0.7494,
+      "step": 348500
+    },
+    {
+      "epoch": 9.89,
+      "grad_norm": 4.224211692810059,
+      "learning_rate": 2.215862401178771e-07,
+      "loss": 0.7532,
+      "step": 349000
+    },
+    {
+      "epoch": 9.9,
+      "grad_norm": 4.752499580383301,
+      "learning_rate": 1.9325040378566776e-07,
+      "loss": 0.7239,
+      "step": 349500
+    },
+    {
+      "epoch": 9.92,
+      "grad_norm": 2.172379732131958,
+      "learning_rate": 1.649145674534584e-07,
+      "loss": 0.7724,
+      "step": 350000
+    },
+    {
+      "epoch": 9.93,
+      "grad_norm": 2.954164743423462,
+      "learning_rate": 1.3657873112124906e-07,
+      "loss": 0.7716,
+      "step": 350500
+    },
+    {
+      "epoch": 9.95,
+      "grad_norm": 3.167682647705078,
+      "learning_rate": 1.082428947890397e-07,
+      "loss": 0.7789,
+      "step": 351000
+    },
+    {
+      "epoch": 9.96,
+      "grad_norm": 3.5074291229248047,
+      "learning_rate": 7.990705845683036e-08,
+      "loss": 0.7563,
+      "step": 351500
+    },
+    {
+      "epoch": 9.97,
+      "grad_norm": 2.857052803039551,
+      "learning_rate": 5.1571222124621016e-08,
+      "loss": 0.7174,
+      "step": 352000
+    },
+    {
+      "epoch": 9.99,
+      "grad_norm": 4.333151340484619,
+      "learning_rate": 2.3235385792411667e-08,
+      "loss": 0.7338,
+      "step": 352500
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.7761048124023759,
+      "eval_loss": 0.7718502879142761,
+      "eval_runtime": 2734.4878,
+      "eval_samples_per_second": 34.416,
+      "eval_steps_per_second": 34.416,
+      "step": 352910
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 352910,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "total_flos": 1.4322010384662528e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}