diff --git "a/checkpoints/checkpoint-28325/trainer_state.json" "b/checkpoints/checkpoint-28325/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoints/checkpoint-28325/trainer_state.json"
@@ -0,0 +1,4219 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.999558732680258,
+  "eval_steps": 1000,
+  "global_step": 28325,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.008825346394845997,
+      "grad_norm": 4.154237270355225,
+      "learning_rate": 4.9915269196822594e-05,
+      "loss": 1.761,
+      "step": 50
+    },
+    {
+      "epoch": 0.017650692789691995,
+      "grad_norm": 5.022342681884766,
+      "learning_rate": 4.98270079435128e-05,
+      "loss": 1.1951,
+      "step": 100
+    },
+    {
+      "epoch": 0.026476039184537992,
+      "grad_norm": 4.0111613273620605,
+      "learning_rate": 4.9738746690203006e-05,
+      "loss": 1.1727,
+      "step": 150
+    },
+    {
+      "epoch": 0.03530138557938399,
+      "grad_norm": 4.008307933807373,
+      "learning_rate": 4.9650485436893205e-05,
+      "loss": 1.0397,
+      "step": 200
+    },
+    {
+      "epoch": 0.04412673197422999,
+      "grad_norm": 3.561506748199463,
+      "learning_rate": 4.956222418358341e-05,
+      "loss": 1.1023,
+      "step": 250
+    },
+    {
+      "epoch": 0.052952078369075985,
+      "grad_norm": 3.71460223197937,
+      "learning_rate": 4.947396293027361e-05,
+      "loss": 1.0662,
+      "step": 300
+    },
+    {
+      "epoch": 0.061777424763921986,
+      "grad_norm": 3.7053279876708984,
+      "learning_rate": 4.9385701676963815e-05,
+      "loss": 1.0294,
+      "step": 350
+    },
+    {
+      "epoch": 0.07060277115876798,
+      "grad_norm": 4.443565845489502,
+      "learning_rate": 4.929744042365402e-05,
+      "loss": 1.0758,
+      "step": 400
+    },
+    {
+      "epoch": 0.07942811755361398,
+      "grad_norm": 3.3921597003936768,
+      "learning_rate": 4.920917917034422e-05,
+      "loss": 0.979,
+      "step": 450
+    },
+    {
+      "epoch": 0.08825346394845998,
+      "grad_norm": 3.324091672897339,
+      "learning_rate": 4.9120917917034426e-05,
+      "loss": 1.0505,
+      "step": 500
+    },
+    {
+      "epoch": 0.09707881034330597,
+      "grad_norm": 3.984813690185547,
+      "learning_rate": 4.903265666372463e-05,
+      "loss": 1.0452,
+      "step": 550
+    },
+    {
+      "epoch": 0.10590415673815197,
+      "grad_norm": 4.010895729064941,
+      "learning_rate": 4.894439541041483e-05,
+      "loss": 1.0423,
+      "step": 600
+    },
+    {
+      "epoch": 0.11472950313299797,
+      "grad_norm": 4.561792373657227,
+      "learning_rate": 4.885613415710503e-05,
+      "loss": 1.042,
+      "step": 650
+    },
+    {
+      "epoch": 0.12355484952784397,
+      "grad_norm": 3.540585517883301,
+      "learning_rate": 4.8767872903795235e-05,
+      "loss": 1.0458,
+      "step": 700
+    },
+    {
+      "epoch": 0.13238019592268996,
+      "grad_norm": 3.182912588119507,
+      "learning_rate": 4.867961165048544e-05,
+      "loss": 1.0393,
+      "step": 750
+    },
+    {
+      "epoch": 0.14120554231753596,
+      "grad_norm": 3.2969844341278076,
+      "learning_rate": 4.8591350397175647e-05,
+      "loss": 1.0733,
+      "step": 800
+    },
+    {
+      "epoch": 0.15003088871238196,
+      "grad_norm": 3.535918712615967,
+      "learning_rate": 4.8503089143865845e-05,
+      "loss": 0.9854,
+      "step": 850
+    },
+    {
+      "epoch": 0.15885623510722796,
+      "grad_norm": 4.074019432067871,
+      "learning_rate": 4.8414827890556044e-05,
+      "loss": 1.0172,
+      "step": 900
+    },
+    {
+      "epoch": 0.16768158150207396,
+      "grad_norm": 2.4251832962036133,
+      "learning_rate": 4.832656663724625e-05,
+      "loss": 0.9895,
+      "step": 950
+    },
+    {
+      "epoch": 0.17650692789691996,
+      "grad_norm": 2.9533393383026123,
+      "learning_rate": 4.8238305383936456e-05,
+      "loss": 1.0214,
+      "step": 1000
+    },
+    {
+      "epoch": 0.17650692789691996,
+      "eval_loss": 0.9503015279769897,
+      "eval_runtime": 181.6386,
+      "eval_samples_per_second": 155.953,
+      "eval_steps_per_second": 15.597,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18533227429176596,
+      "grad_norm": 3.5368824005126953,
+      "learning_rate": 4.815004413062666e-05,
+      "loss": 0.9633,
+      "step": 1050
+    },
+    {
+      "epoch": 0.19415762068661194,
+      "grad_norm": 3.8759989738464355,
+      "learning_rate": 4.806178287731686e-05,
+      "loss": 0.9748,
+      "step": 1100
+    },
+    {
+      "epoch": 0.20298296708145794,
+      "grad_norm": 3.02571177482605,
+      "learning_rate": 4.797352162400706e-05,
+      "loss": 0.9824,
+      "step": 1150
+    },
+    {
+      "epoch": 0.21180831347630394,
+      "grad_norm": 4.157599449157715,
+      "learning_rate": 4.7885260370697265e-05,
+      "loss": 0.9452,
+      "step": 1200
+    },
+    {
+      "epoch": 0.22063365987114994,
+      "grad_norm": 3.4217748641967773,
+      "learning_rate": 4.779699911738747e-05,
+      "loss": 0.9668,
+      "step": 1250
+    },
+    {
+      "epoch": 0.22945900626599594,
+      "grad_norm": 3.7693021297454834,
+      "learning_rate": 4.770873786407768e-05,
+      "loss": 0.9758,
+      "step": 1300
+    },
+    {
+      "epoch": 0.23828435266084194,
+      "grad_norm": 4.903484344482422,
+      "learning_rate": 4.7620476610767876e-05,
+      "loss": 0.9817,
+      "step": 1350
+    },
+    {
+      "epoch": 0.24710969905568794,
+      "grad_norm": 3.7475759983062744,
+      "learning_rate": 4.7532215357458075e-05,
+      "loss": 0.968,
+      "step": 1400
+    },
+    {
+      "epoch": 0.2559350454505339,
+      "grad_norm": 2.6870646476745605,
+      "learning_rate": 4.744395410414828e-05,
+      "loss": 0.9143,
+      "step": 1450
+    },
+    {
+      "epoch": 0.2647603918453799,
+      "grad_norm": 3.0536086559295654,
+      "learning_rate": 4.7355692850838486e-05,
+      "loss": 0.9469,
+      "step": 1500
+    },
+    {
+      "epoch": 0.2735857382402259,
+      "grad_norm": 3.0142056941986084,
+      "learning_rate": 4.726743159752869e-05,
+      "loss": 0.9072,
+      "step": 1550
+    },
+    {
+      "epoch": 0.2824110846350719,
+      "grad_norm": 2.2967958450317383,
+      "learning_rate": 4.717917034421889e-05,
+      "loss": 0.9582,
+      "step": 1600
+    },
+    {
+      "epoch": 0.2912364310299179,
+      "grad_norm": 3.881225824356079,
+      "learning_rate": 4.709090909090909e-05,
+      "loss": 0.9517,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3000617774247639,
+      "grad_norm": 3.5175845623016357,
+      "learning_rate": 4.7002647837599295e-05,
+      "loss": 0.9889,
+      "step": 1700
+    },
+    {
+      "epoch": 0.3088871238196099,
+      "grad_norm": 2.7161202430725098,
+      "learning_rate": 4.69143865842895e-05,
+      "loss": 0.8666,
+      "step": 1750
+    },
+    {
+      "epoch": 0.3177124702144559,
+      "grad_norm": 3.3932855129241943,
+      "learning_rate": 4.68261253309797e-05,
+      "loss": 0.908,
+      "step": 1800
+    },
+    {
+      "epoch": 0.3265378166093019,
+      "grad_norm": 2.1355509757995605,
+      "learning_rate": 4.6737864077669906e-05,
+      "loss": 0.9444,
+      "step": 1850
+    },
+    {
+      "epoch": 0.3353631630041479,
+      "grad_norm": 3.1375510692596436,
+      "learning_rate": 4.6649602824360105e-05,
+      "loss": 0.8773,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3441885093989939,
+      "grad_norm": 3.356266975402832,
+      "learning_rate": 4.656134157105031e-05,
+      "loss": 0.9471,
+      "step": 1950
+    },
+    {
+      "epoch": 0.3530138557938399,
+      "grad_norm": 3.24814510345459,
+      "learning_rate": 4.6473080317740516e-05,
+      "loss": 0.9159,
+      "step": 2000
+    },
+    {
+      "epoch": 0.3530138557938399,
+      "eval_loss": 0.8810886144638062,
+      "eval_runtime": 181.6697,
+      "eval_samples_per_second": 155.926,
+      "eval_steps_per_second": 15.594,
+      "step": 2000
+    },
+    {
+      "epoch": 0.3618392021886859,
+      "grad_norm": 2.7671000957489014,
+      "learning_rate": 4.6384819064430715e-05,
+      "loss": 0.8957,
+      "step": 2050
+    },
+    {
+      "epoch": 0.3706645485835319,
+      "grad_norm": 3.4630813598632812,
+      "learning_rate": 4.629655781112092e-05,
+      "loss": 0.9142,
+      "step": 2100
+    },
+    {
+      "epoch": 0.37948989497837793,
+      "grad_norm": 3.2977519035339355,
+      "learning_rate": 4.620829655781112e-05,
+      "loss": 0.8814,
+      "step": 2150
+    },
+    {
+      "epoch": 0.3883152413732239,
+      "grad_norm": 2.398122787475586,
+      "learning_rate": 4.6120035304501326e-05,
+      "loss": 0.8989,
+      "step": 2200
+    },
+    {
+      "epoch": 0.3971405877680699,
+      "grad_norm": 2.642117500305176,
+      "learning_rate": 4.603177405119153e-05,
+      "loss": 0.9322,
+      "step": 2250
+    },
+    {
+      "epoch": 0.4059659341629159,
+      "grad_norm": 3.492455005645752,
+      "learning_rate": 4.594351279788173e-05,
+      "loss": 0.9167,
+      "step": 2300
+    },
+    {
+      "epoch": 0.4147912805577619,
+      "grad_norm": 3.0943338871002197,
+      "learning_rate": 4.5855251544571936e-05,
+      "loss": 0.9265,
+      "step": 2350
+    },
+    {
+      "epoch": 0.4236166269526079,
+      "grad_norm": 3.072920322418213,
+      "learning_rate": 4.576699029126214e-05,
+      "loss": 0.9321,
+      "step": 2400
+    },
+    {
+      "epoch": 0.4324419733474539,
+      "grad_norm": 2.8082592487335205,
+      "learning_rate": 4.567872903795234e-05,
+      "loss": 0.8699,
+      "step": 2450
+    },
+    {
+      "epoch": 0.4412673197422999,
+      "grad_norm": 3.1018335819244385,
+      "learning_rate": 4.5590467784642546e-05,
+      "loss": 0.8869,
+      "step": 2500
+    },
+    {
+      "epoch": 0.4500926661371459,
+      "grad_norm": 3.1571199893951416,
+      "learning_rate": 4.5502206531332745e-05,
+      "loss": 1.0116,
+      "step": 2550
+    },
+    {
+      "epoch": 0.4589180125319919,
+      "grad_norm": 3.358121871948242,
+      "learning_rate": 4.541394527802295e-05,
+      "loss": 0.9374,
+      "step": 2600
+    },
+    {
+      "epoch": 0.4677433589268379,
+      "grad_norm": 2.648608684539795,
+      "learning_rate": 4.532568402471316e-05,
+      "loss": 0.8901,
+      "step": 2650
+    },
+    {
+      "epoch": 0.4765687053216839,
+      "grad_norm": 3.6951816082000732,
+      "learning_rate": 4.5237422771403356e-05,
+      "loss": 0.8871,
+      "step": 2700
+    },
+    {
+      "epoch": 0.4853940517165299,
+      "grad_norm": 2.734888792037964,
+      "learning_rate": 4.5149161518093555e-05,
+      "loss": 0.8735,
+      "step": 2750
+    },
+    {
+      "epoch": 0.4942193981113759,
+      "grad_norm": 2.99570369720459,
+      "learning_rate": 4.506090026478376e-05,
+      "loss": 0.8863,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5030447445062218,
+      "grad_norm": 3.6677873134613037,
+      "learning_rate": 4.4972639011473966e-05,
+      "loss": 0.9086,
+      "step": 2850
+    },
+    {
+      "epoch": 0.5118700909010678,
+      "grad_norm": 2.688584804534912,
+      "learning_rate": 4.488437775816417e-05,
+      "loss": 0.9028,
+      "step": 2900
+    },
+    {
+      "epoch": 0.5206954372959138,
+      "grad_norm": 3.2178659439086914,
+      "learning_rate": 4.479611650485437e-05,
+      "loss": 0.9089,
+      "step": 2950
+    },
+    {
+      "epoch": 0.5295207836907598,
+      "grad_norm": 2.8180041313171387,
+      "learning_rate": 4.470785525154457e-05,
+      "loss": 0.8956,
+      "step": 3000
+    },
+    {
+      "epoch": 0.5295207836907598,
+      "eval_loss": 0.8461005687713623,
+      "eval_runtime": 181.6855,
+      "eval_samples_per_second": 155.912,
+      "eval_steps_per_second": 15.593,
+      "step": 3000
+    },
+    {
+      "epoch": 0.5383461300856058,
+      "grad_norm": 3.1606943607330322,
+      "learning_rate": 4.4619593998234776e-05,
+      "loss": 0.8886,
+      "step": 3050
+    },
+    {
+      "epoch": 0.5471714764804518,
+      "grad_norm": 2.718230724334717,
+      "learning_rate": 4.453133274492498e-05,
+      "loss": 0.8704,
+      "step": 3100
+    },
+    {
+      "epoch": 0.5559968228752978,
+      "grad_norm": 3.1943600177764893,
+      "learning_rate": 4.444307149161519e-05,
+      "loss": 0.9335,
+      "step": 3150
+    },
+    {
+      "epoch": 0.5648221692701438,
+      "grad_norm": 3.1927852630615234,
+      "learning_rate": 4.4354810238305386e-05,
+      "loss": 0.89,
+      "step": 3200
+    },
+    {
+      "epoch": 0.5736475156649898,
+      "grad_norm": 3.269458293914795,
+      "learning_rate": 4.4266548984995585e-05,
+      "loss": 0.8806,
+      "step": 3250
+    },
+    {
+      "epoch": 0.5824728620598358,
+      "grad_norm": 3.280489921569824,
+      "learning_rate": 4.417828773168579e-05,
+      "loss": 0.8952,
+      "step": 3300
+    },
+    {
+      "epoch": 0.5912982084546818,
+      "grad_norm": 3.4834930896759033,
+      "learning_rate": 4.4090026478375996e-05,
+      "loss": 0.8698,
+      "step": 3350
+    },
+    {
+      "epoch": 0.6001235548495278,
+      "grad_norm": 2.7703258991241455,
+      "learning_rate": 4.40017652250662e-05,
+      "loss": 0.8516,
+      "step": 3400
+    },
+    {
+      "epoch": 0.6089489012443738,
+      "grad_norm": 3.2723336219787598,
+      "learning_rate": 4.39135039717564e-05,
+      "loss": 0.8564,
+      "step": 3450
+    },
+    {
+      "epoch": 0.6177742476392198,
+      "grad_norm": 2.9629292488098145,
+      "learning_rate": 4.38252427184466e-05,
+      "loss": 0.8509,
+      "step": 3500
+    },
+    {
+      "epoch": 0.6265995940340658,
+      "grad_norm": 2.5405166149139404,
+      "learning_rate": 4.3736981465136806e-05,
+      "loss": 0.875,
+      "step": 3550
+    },
+    {
+      "epoch": 0.6354249404289118,
+      "grad_norm": 2.6754391193389893,
+      "learning_rate": 4.364872021182701e-05,
+      "loss": 0.8813,
+      "step": 3600
+    },
+    {
+      "epoch": 0.6442502868237578,
+      "grad_norm": 3.0264227390289307,
+      "learning_rate": 4.356045895851722e-05,
+      "loss": 0.8518,
+      "step": 3650
+    },
+    {
+      "epoch": 0.6530756332186038,
+      "grad_norm": 2.8763012886047363,
+      "learning_rate": 4.3472197705207416e-05,
+      "loss": 0.8272,
+      "step": 3700
+    },
+    {
+      "epoch": 0.6619009796134498,
+      "grad_norm": 2.9555325508117676,
+      "learning_rate": 4.3383936451897615e-05,
+      "loss": 0.9122,
+      "step": 3750
+    },
+    {
+      "epoch": 0.6707263260082958,
+      "grad_norm": 3.2128372192382812,
+      "learning_rate": 4.329567519858782e-05,
+      "loss": 0.825,
+      "step": 3800
+    },
+    {
+      "epoch": 0.6795516724031418,
+      "grad_norm": 2.1403987407684326,
+      "learning_rate": 4.320741394527803e-05,
+      "loss": 0.8808,
+      "step": 3850
+    },
+    {
+      "epoch": 0.6883770187979878,
+      "grad_norm": 4.011631965637207,
+      "learning_rate": 4.3119152691968226e-05,
+      "loss": 0.8071,
+      "step": 3900
+    },
+    {
+      "epoch": 0.6972023651928339,
+      "grad_norm": 2.8009603023529053,
+      "learning_rate": 4.303089143865843e-05,
+      "loss": 0.8236,
+      "step": 3950
+    },
+    {
+      "epoch": 0.7060277115876799,
+      "grad_norm": 2.3197402954101562,
+      "learning_rate": 4.294263018534863e-05,
+      "loss": 0.9126,
+      "step": 4000
+    },
+    {
+      "epoch": 0.7060277115876799,
+      "eval_loss": 0.8137471675872803,
+      "eval_runtime": 181.6839,
+      "eval_samples_per_second": 155.914,
+      "eval_steps_per_second": 15.593,
+      "step": 4000
+    },
+    {
+      "epoch": 0.7148530579825259,
+      "grad_norm": 2.557133913040161,
+      "learning_rate": 4.2854368932038836e-05,
+      "loss": 0.8425,
+      "step": 4050
+    },
+    {
+      "epoch": 0.7236784043773719,
+      "grad_norm": 3.1375222206115723,
+      "learning_rate": 4.276610767872904e-05,
+      "loss": 0.8095,
+      "step": 4100
+    },
+    {
+      "epoch": 0.7325037507722179,
+      "grad_norm": 3.2000410556793213,
+      "learning_rate": 4.267784642541924e-05,
+      "loss": 0.8318,
+      "step": 4150
+    },
+    {
+      "epoch": 0.7413290971670639,
+      "grad_norm": 2.4913811683654785,
+      "learning_rate": 4.2589585172109446e-05,
+      "loss": 0.8729,
+      "step": 4200
+    },
+    {
+      "epoch": 0.7501544435619099,
+      "grad_norm": 2.605590581893921,
+      "learning_rate": 4.250132391879965e-05,
+      "loss": 0.7657,
+      "step": 4250
+    },
+    {
+      "epoch": 0.7589797899567559,
+      "grad_norm": 2.879939317703247,
+      "learning_rate": 4.241306266548985e-05,
+      "loss": 0.8319,
+      "step": 4300
+    },
+    {
+      "epoch": 0.7678051363516017,
+      "grad_norm": 3.4146721363067627,
+      "learning_rate": 4.232480141218006e-05,
+      "loss": 0.8513,
+      "step": 4350
+    },
+    {
+      "epoch": 0.7766304827464477,
+      "grad_norm": 2.533024787902832,
+      "learning_rate": 4.2236540158870256e-05,
+      "loss": 0.8228,
+      "step": 4400
+    },
+    {
+      "epoch": 0.7854558291412937,
+      "grad_norm": 3.1727654933929443,
+      "learning_rate": 4.214827890556046e-05,
+      "loss": 0.8579,
+      "step": 4450
+    },
+    {
+      "epoch": 0.7942811755361397,
+      "grad_norm": 3.0712013244628906,
+      "learning_rate": 4.206001765225067e-05,
+      "loss": 0.8504,
+      "step": 4500
+    },
+    {
+      "epoch": 0.8031065219309858,
+      "grad_norm": 4.400243282318115,
+      "learning_rate": 4.1971756398940866e-05,
+      "loss": 0.8606,
+      "step": 4550
+    },
+    {
+      "epoch": 0.8119318683258318,
+      "grad_norm": 3.040689468383789,
+      "learning_rate": 4.188349514563107e-05,
+      "loss": 0.822,
+      "step": 4600
+    },
+    {
+      "epoch": 0.8207572147206778,
+      "grad_norm": 2.518383026123047,
+      "learning_rate": 4.179523389232127e-05,
+      "loss": 0.8013,
+      "step": 4650
+    },
+    {
+      "epoch": 0.8295825611155238,
+      "grad_norm": 3.1982264518737793,
+      "learning_rate": 4.170697263901148e-05,
+      "loss": 0.8374,
+      "step": 4700
+    },
+    {
+      "epoch": 0.8384079075103698,
+      "grad_norm": 3.7081596851348877,
+      "learning_rate": 4.161871138570168e-05,
+      "loss": 0.8281,
+      "step": 4750
+    },
+    {
+      "epoch": 0.8472332539052158,
+      "grad_norm": 2.829268217086792,
+      "learning_rate": 4.153045013239188e-05,
+      "loss": 0.8736,
+      "step": 4800
+    },
+    {
+      "epoch": 0.8560586003000618,
+      "grad_norm": 2.8219809532165527,
+      "learning_rate": 4.144218887908208e-05,
+      "loss": 0.8324,
+      "step": 4850
+    },
+    {
+      "epoch": 0.8648839466949078,
+      "grad_norm": 3.261457681655884,
+      "learning_rate": 4.1353927625772286e-05,
+      "loss": 0.7835,
+      "step": 4900
+    },
+    {
+      "epoch": 0.8737092930897538,
+      "grad_norm": 2.952995777130127,
+      "learning_rate": 4.126566637246249e-05,
+      "loss": 0.8674,
+      "step": 4950
+    },
+    {
+      "epoch": 0.8825346394845998,
+      "grad_norm": 2.9086506366729736,
+      "learning_rate": 4.11774051191527e-05,
+      "loss": 0.8263,
+      "step": 5000
+    },
+    {
+      "epoch": 0.8825346394845998,
+      "eval_loss": 0.7903227806091309,
+      "eval_runtime": 181.7285,
+      "eval_samples_per_second": 155.875,
+      "eval_steps_per_second": 15.589,
+      "step": 5000
+    },
+    {
+      "epoch": 0.8913599858794458,
+      "grad_norm": 2.967153549194336,
+      "learning_rate": 4.1089143865842896e-05,
+      "loss": 0.8394,
+      "step": 5050
+    },
+    {
+      "epoch": 0.9001853322742918,
+      "grad_norm": 3.2790939807891846,
+      "learning_rate": 4.1000882612533095e-05,
+      "loss": 0.8037,
+      "step": 5100
+    },
+    {
+      "epoch": 0.9090106786691378,
+      "grad_norm": 2.5177745819091797,
+      "learning_rate": 4.09126213592233e-05,
+      "loss": 0.8217,
+      "step": 5150
+    },
+    {
+      "epoch": 0.9178360250639838,
+      "grad_norm": 3.300760507583618,
+      "learning_rate": 4.082436010591351e-05,
+      "loss": 0.8177,
+      "step": 5200
+    },
+    {
+      "epoch": 0.9266613714588298,
+      "grad_norm": 2.8145620822906494,
+      "learning_rate": 4.073609885260371e-05,
+      "loss": 0.8196,
+      "step": 5250
+    },
+    {
+      "epoch": 0.9354867178536758,
+      "grad_norm": 2.681286096572876,
+      "learning_rate": 4.064783759929391e-05,
+      "loss": 0.8417,
+      "step": 5300
+    },
+    {
+      "epoch": 0.9443120642485218,
+      "grad_norm": 2.404695510864258,
+      "learning_rate": 4.055957634598411e-05,
+      "loss": 0.841,
+      "step": 5350
+    },
+    {
+      "epoch": 0.9531374106433678,
+      "grad_norm": 2.606860876083374,
+      "learning_rate": 4.0471315092674316e-05,
+      "loss": 0.8194,
+      "step": 5400
+    },
+    {
+      "epoch": 0.9619627570382138,
+      "grad_norm": 2.41960072517395,
+      "learning_rate": 4.038305383936452e-05,
+      "loss": 0.8009,
+      "step": 5450
+    },
+    {
+      "epoch": 0.9707881034330598,
+      "grad_norm": 3.663374185562134,
+      "learning_rate": 4.029479258605473e-05,
+      "loss": 0.8017,
+      "step": 5500
+    },
+    {
+      "epoch": 0.9796134498279058,
+      "grad_norm": 2.751960277557373,
+      "learning_rate": 4.0206531332744927e-05,
+      "loss": 0.7816,
+      "step": 5550
+    },
+    {
+      "epoch": 0.9884387962227518,
+      "grad_norm": 2.9253251552581787,
+      "learning_rate": 4.0118270079435126e-05,
+      "loss": 0.8022,
+      "step": 5600
+    },
+    {
+      "epoch": 0.9972641426175978,
+      "grad_norm": 3.4623188972473145,
+      "learning_rate": 4.003000882612533e-05,
+      "loss": 0.7701,
+      "step": 5650
+    },
+    {
+      "epoch": 1.0060894890124437,
+      "grad_norm": 2.0556461811065674,
+      "learning_rate": 3.994174757281554e-05,
+      "loss": 0.7143,
+      "step": 5700
+    },
+    {
+      "epoch": 1.0149148354072897,
+      "grad_norm": 2.6380767822265625,
+      "learning_rate": 3.985348631950574e-05,
+      "loss": 0.6442,
+      "step": 5750
+    },
+    {
+      "epoch": 1.0237401818021357,
+      "grad_norm": 3.469467878341675,
+      "learning_rate": 3.976522506619594e-05,
+      "loss": 0.6443,
+      "step": 5800
+    },
+    {
+      "epoch": 1.0325655281969817,
+      "grad_norm": 3.0494396686553955,
+      "learning_rate": 3.967696381288615e-05,
+      "loss": 0.6686,
+      "step": 5850
+    },
+    {
+      "epoch": 1.0413908745918277,
+      "grad_norm": 3.4059927463531494,
+      "learning_rate": 3.9588702559576346e-05,
+      "loss": 0.6985,
+      "step": 5900
+    },
+    {
+      "epoch": 1.0502162209866737,
+      "grad_norm": 2.977858781814575,
+      "learning_rate": 3.950044130626655e-05,
+      "loss": 0.6614,
+      "step": 5950
+    },
+    {
+      "epoch": 1.0590415673815197,
+      "grad_norm": 2.572244644165039,
+      "learning_rate": 3.941218005295675e-05,
+      "loss": 0.6959,
+      "step": 6000
+    },
+    {
+      "epoch": 1.0590415673815197,
+      "eval_loss": 0.7684289813041687,
+      "eval_runtime": 181.6989,
+      "eval_samples_per_second": 155.901,
+      "eval_steps_per_second": 15.592,
+      "step": 6000
+    },
+    {
+      "epoch": 1.0678669137763657,
+      "grad_norm": 2.360053777694702,
+      "learning_rate": 3.932568402471315e-05,
+      "loss": 0.6979,
+      "step": 6050
+    },
+    {
+      "epoch": 1.0766922601712117,
+      "grad_norm": 2.46508526802063,
+      "learning_rate": 3.923742277140335e-05,
+      "loss": 0.6179,
+      "step": 6100
+    },
+    {
+      "epoch": 1.0855176065660577,
+      "grad_norm": 2.3281443119049072,
+      "learning_rate": 3.914916151809356e-05,
+      "loss": 0.6245,
+      "step": 6150
+    },
+    {
+      "epoch": 1.0943429529609037,
+      "grad_norm": 2.59844708442688,
+      "learning_rate": 3.9060900264783764e-05,
+      "loss": 0.6938,
+      "step": 6200
+    },
+    {
+      "epoch": 1.1031682993557497,
+      "grad_norm": 3.2816011905670166,
+      "learning_rate": 3.897263901147396e-05,
+      "loss": 0.6129,
+      "step": 6250
+    },
+    {
+      "epoch": 1.1119936457505957,
+      "grad_norm": 2.310105800628662,
+      "learning_rate": 3.888437775816417e-05,
+      "loss": 0.6435,
+      "step": 6300
+    },
+    {
+      "epoch": 1.1208189921454417,
+      "grad_norm": 2.9647128582000732,
+      "learning_rate": 3.879611650485437e-05,
+      "loss": 0.6694,
+      "step": 6350
+    },
+    {
+      "epoch": 1.1296443385402877,
+      "grad_norm": 2.9377100467681885,
+      "learning_rate": 3.870785525154457e-05,
+      "loss": 0.7068,
+      "step": 6400
+    },
+    {
+      "epoch": 1.1384696849351337,
+      "grad_norm": 2.559523105621338,
+      "learning_rate": 3.861959399823478e-05,
+      "loss": 0.6432,
+      "step": 6450
+    },
+    {
+      "epoch": 1.1472950313299797,
+      "grad_norm": 2.5071003437042236,
+      "learning_rate": 3.853133274492498e-05,
+      "loss": 0.6542,
+      "step": 6500
+    },
+    {
+      "epoch": 1.1561203777248257,
+      "grad_norm": 2.6376101970672607,
+      "learning_rate": 3.8443071491615184e-05,
+      "loss": 0.6715,
+      "step": 6550
+    },
+    {
+      "epoch": 1.1649457241196717,
+      "grad_norm": 2.169046401977539,
+      "learning_rate": 3.835481023830538e-05,
+      "loss": 0.6367,
+      "step": 6600
+    },
+    {
+      "epoch": 1.1737710705145177,
+      "grad_norm": 2.852400779724121,
+      "learning_rate": 3.826654898499559e-05,
+      "loss": 0.6711,
+      "step": 6650
+    },
+    {
+      "epoch": 1.1825964169093637,
+      "grad_norm": 2.8903868198394775,
+      "learning_rate": 3.8178287731685794e-05,
+      "loss": 0.6603,
+      "step": 6700
+    },
+    {
+      "epoch": 1.1914217633042097,
+      "grad_norm": 3.0550076961517334,
+      "learning_rate": 3.809002647837599e-05,
+      "loss": 0.7009,
+      "step": 6750
+    },
+    {
+      "epoch": 1.2002471096990557,
+      "grad_norm": 3.017362356185913,
+      "learning_rate": 3.80017652250662e-05,
+      "loss": 0.651,
+      "step": 6800
+    },
+    {
+      "epoch": 1.2090724560939017,
+      "grad_norm": 3.1694257259368896,
+      "learning_rate": 3.7913503971756404e-05,
+      "loss": 0.6674,
+      "step": 6850
+    },
+    {
+      "epoch": 1.2178978024887477,
+      "grad_norm": 3.285419464111328,
+      "learning_rate": 3.78270079435128e-05,
+      "loss": 0.6425,
+      "step": 6900
+    },
+    {
+      "epoch": 1.2267231488835937,
+      "grad_norm": 3.4212896823883057,
+      "learning_rate": 3.7738746690203006e-05,
+      "loss": 0.7068,
+      "step": 6950
+    },
+    {
+      "epoch": 1.2355484952784397,
+      "grad_norm": 2.7931535243988037,
+      "learning_rate": 3.7650485436893205e-05,
+      "loss": 0.6685,
+      "step": 7000
+    },
+    {
+      "epoch": 1.2355484952784397,
+      "eval_loss": 0.7610744833946228,
+      "eval_runtime": 181.743,
+      "eval_samples_per_second": 155.863,
+      "eval_steps_per_second": 15.588,
+      "step": 7000
+    },
+    {
+      "epoch": 1.2443738416732857,
+      "grad_norm": 3.0050013065338135,
+      "learning_rate": 3.756222418358341e-05,
+      "loss": 0.63,
+      "step": 7050
+    },
+    {
+      "epoch": 1.2531991880681317,
+      "grad_norm": 3.8987817764282227,
+      "learning_rate": 3.747396293027361e-05,
+      "loss": 0.6285,
+      "step": 7100
+    },
+    {
+      "epoch": 1.2620245344629777,
+      "grad_norm": 2.7524607181549072,
+      "learning_rate": 3.7385701676963815e-05,
+      "loss": 0.649,
+      "step": 7150
+    },
+    {
+      "epoch": 1.2708498808578237,
+      "grad_norm": 3.0095231533050537,
+      "learning_rate": 3.729744042365402e-05,
+      "loss": 0.6298,
+      "step": 7200
+    },
+    {
+      "epoch": 1.2796752272526697,
+      "grad_norm": 2.8414998054504395,
+      "learning_rate": 3.720917917034422e-05,
+      "loss": 0.5854,
+      "step": 7250
+    },
+    {
+      "epoch": 1.2885005736475157,
+      "grad_norm": 2.4056193828582764,
+      "learning_rate": 3.7120917917034425e-05,
+      "loss": 0.6555,
+      "step": 7300
+    },
+    {
+      "epoch": 1.2973259200423617,
+      "grad_norm": 2.7647316455841064,
+      "learning_rate": 3.7032656663724624e-05,
+      "loss": 0.6433,
+      "step": 7350
+    },
+    {
+      "epoch": 1.3061512664372077,
+      "grad_norm": 3.025959014892578,
+      "learning_rate": 3.694439541041483e-05,
+      "loss": 0.6319,
+      "step": 7400
+    },
+    {
+      "epoch": 1.3149766128320537,
+      "grad_norm": 2.893221378326416,
+      "learning_rate": 3.685613415710503e-05,
+      "loss": 0.6474,
+      "step": 7450
+    },
+    {
+      "epoch": 1.3238019592268997,
+      "grad_norm": 2.621736764907837,
+      "learning_rate": 3.6767872903795235e-05,
+      "loss": 0.6302,
+      "step": 7500
+    },
+    {
+      "epoch": 1.3326273056217457,
+      "grad_norm": 3.4629733562469482,
+      "learning_rate": 3.667961165048544e-05,
+      "loss": 0.6259,
+      "step": 7550
+    },
+    {
+      "epoch": 1.3414526520165917,
+      "grad_norm": 2.6440625190734863,
+      "learning_rate": 3.6591350397175646e-05,
+      "loss": 0.7059,
+      "step": 7600
+    },
+    {
+      "epoch": 1.3502779984114377,
+      "grad_norm": 3.098646879196167,
+      "learning_rate": 3.6503089143865845e-05,
+      "loss": 0.6563,
+      "step": 7650
+    },
+    {
+      "epoch": 1.3591033448062837,
+      "grad_norm": 3.1059629917144775,
+      "learning_rate": 3.6414827890556044e-05,
+      "loss": 0.7017,
+      "step": 7700
+    },
+    {
+      "epoch": 1.3679286912011297,
+      "grad_norm": 3.305082082748413,
+      "learning_rate": 3.632656663724625e-05,
+      "loss": 0.6417,
+      "step": 7750
+    },
+    {
+      "epoch": 1.3767540375959757,
+      "grad_norm": 2.580493688583374,
+      "learning_rate": 3.6238305383936456e-05,
+      "loss": 0.6768,
+      "step": 7800
+    },
+    {
+      "epoch": 1.3855793839908217,
+      "grad_norm": 2.5949904918670654,
+      "learning_rate": 3.615004413062666e-05,
+      "loss": 0.6778,
+      "step": 7850
+    },
+    {
+      "epoch": 1.3944047303856677,
+      "grad_norm": 3.4138495922088623,
+      "learning_rate": 3.606354810238306e-05,
+      "loss": 0.644,
+      "step": 7900
+    },
+    {
+      "epoch": 1.4032300767805137,
+      "grad_norm": 2.841855764389038,
+      "learning_rate": 3.5975286849073256e-05,
+      "loss": 0.6587,
+      "step": 7950
+    },
+    {
+      "epoch": 1.4120554231753597,
+      "grad_norm": 3.4159514904022217,
+      "learning_rate": 3.588702559576346e-05,
+      "loss": 0.6483,
+      "step": 8000
+    },
+    {
+      "epoch": 1.4120554231753597,
+      "eval_loss": 0.7551769018173218,
+      "eval_runtime": 181.7484,
+      "eval_samples_per_second": 155.858,
+      "eval_steps_per_second": 15.587,
+      "step": 8000
+    },
+    {
+      "epoch": 1.4208807695702057,
+      "grad_norm": 3.678314447402954,
+      "learning_rate": 3.579876434245367e-05,
+      "loss": 0.6326,
+      "step": 8050
+    },
+    {
+      "epoch": 1.4297061159650517,
+      "grad_norm": 2.811630964279175,
+      "learning_rate": 3.5710503089143866e-05,
+      "loss": 0.6684,
+      "step": 8100
+    },
+    {
+      "epoch": 1.4385314623598977,
+      "grad_norm": 2.802046775817871,
+      "learning_rate": 3.562224183583407e-05,
+      "loss": 0.6581,
+      "step": 8150
+    },
+    {
+      "epoch": 1.4473568087547437,
+      "grad_norm": 3.1183340549468994,
+      "learning_rate": 3.553398058252427e-05,
+      "loss": 0.6444,
+      "step": 8200
+    },
+    {
+      "epoch": 1.4561821551495897,
+      "grad_norm": 2.275667428970337,
+      "learning_rate": 3.544571932921448e-05,
+      "loss": 0.6042,
+      "step": 8250
+    },
+    {
+      "epoch": 1.4650075015444357,
+      "grad_norm": 3.0248420238494873,
+      "learning_rate": 3.535745807590468e-05,
+      "loss": 0.6638,
+      "step": 8300
+    },
+    {
+      "epoch": 1.4738328479392817,
+      "grad_norm": 3.109173059463501,
+      "learning_rate": 3.526919682259488e-05,
+      "loss": 0.6636,
+      "step": 8350
+    },
+    {
+      "epoch": 1.4826581943341277,
+      "grad_norm": 3.1378836631774902,
+      "learning_rate": 3.518093556928509e-05,
+      "loss": 0.6723,
+      "step": 8400
+    },
+    {
+      "epoch": 1.4914835407289737,
+      "grad_norm": 2.956021547317505,
+      "learning_rate": 3.5092674315975286e-05,
+      "loss": 0.6449,
+      "step": 8450
+    },
+    {
+      "epoch": 1.5003088871238197,
+      "grad_norm": 2.559032440185547,
+      "learning_rate": 3.500441306266549e-05,
+      "loss": 0.6343,
+      "step": 8500
+    },
+    {
+      "epoch": 1.5091342335186657,
+      "grad_norm": 2.106903553009033,
+      "learning_rate": 3.49161518093557e-05,
+      "loss": 0.6494,
+      "step": 8550
+    },
+    {
+      "epoch": 1.5179595799135117,
+      "grad_norm": 4.132986068725586,
+      "learning_rate": 3.48278905560459e-05,
+      "loss": 0.6556,
+      "step": 8600
+    },
+    {
+      "epoch": 1.5267849263083577,
+      "grad_norm": 3.011594533920288,
+      "learning_rate": 3.4739629302736095e-05,
+      "loss": 0.6318,
+      "step": 8650
+    },
+    {
+      "epoch": 1.5356102727032037,
+      "grad_norm": 3.080071210861206,
+      "learning_rate": 3.46513680494263e-05,
+      "loss": 0.6635,
+      "step": 8700
+    },
+    {
+      "epoch": 1.5444356190980497,
+      "grad_norm": 2.3603274822235107,
+      "learning_rate": 3.456310679611651e-05,
+      "loss": 0.6224,
+      "step": 8750
+    },
+    {
+      "epoch": 1.5532609654928957,
+      "grad_norm": 3.5411500930786133,
+      "learning_rate": 3.447484554280671e-05,
+      "loss": 0.6349,
+      "step": 8800
+    },
+    {
+      "epoch": 1.5620863118877417,
+      "grad_norm": 3.3874282836914062,
+      "learning_rate": 3.438658428949691e-05,
+      "loss": 0.6537,
+      "step": 8850
+    },
+    {
+      "epoch": 1.5709116582825877,
+      "grad_norm": 3.0514631271362305,
+      "learning_rate": 3.429832303618711e-05,
+      "loss": 0.6496,
+      "step": 8900
+    },
+    {
+      "epoch": 1.5797370046774337,
+      "grad_norm": 3.084991931915283,
+      "learning_rate": 3.4210061782877316e-05,
+      "loss": 0.6647,
+      "step": 8950
+    },
+    {
+      "epoch": 1.5885623510722797,
+      "grad_norm": 3.14204478263855,
+      "learning_rate": 3.412180052956752e-05,
+      "loss": 0.6551,
+      "step": 9000
+    },
+    {
+      "epoch": 1.5885623510722797,
+      "eval_loss": 0.7328709363937378,
+      "eval_runtime": 181.7304,
+      "eval_samples_per_second": 155.874,
+      "eval_steps_per_second": 15.589,
+      "step": 9000
+    },
+    {
+      "epoch": 1.5973876974671257,
+      "grad_norm": 2.8025941848754883,
+      "learning_rate": 3.403353927625773e-05,
+      "loss": 0.6385,
+      "step": 9050
+    },
+    {
+      "epoch": 1.6062130438619717,
+      "grad_norm": 2.485394239425659,
+      "learning_rate": 3.3945278022947927e-05,
+      "loss": 0.5855,
+      "step": 9100
+    },
+    {
+      "epoch": 1.6150383902568177,
+      "grad_norm": 3.1365981101989746,
+      "learning_rate": 3.3857016769638126e-05,
+      "loss": 0.6335,
+      "step": 9150
+    },
+    {
+      "epoch": 1.6238637366516637,
+      "grad_norm": 2.991084098815918,
+      "learning_rate": 3.376875551632833e-05,
+      "loss": 0.649,
+      "step": 9200
+    },
+    {
+      "epoch": 1.6326890830465097,
+      "grad_norm": 3.4233882427215576,
+      "learning_rate": 3.368049426301854e-05,
+      "loss": 0.6562,
+      "step": 9250
+    },
+    {
+      "epoch": 1.6415144294413557,
+      "grad_norm": 2.715102195739746,
+      "learning_rate": 3.359223300970874e-05,
+      "loss": 0.6356,
+      "step": 9300
+    },
+    {
+      "epoch": 1.6503397758362017,
+      "grad_norm": 3.2025671005249023,
+      "learning_rate": 3.350397175639894e-05,
+      "loss": 0.6451,
+      "step": 9350
+    },
+    {
+      "epoch": 1.6591651222310477,
+      "grad_norm": 2.856401205062866,
+      "learning_rate": 3.341571050308915e-05,
+      "loss": 0.6527,
+      "step": 9400
+    },
+    {
+      "epoch": 1.6679904686258937,
+      "grad_norm": 2.430091142654419,
+      "learning_rate": 3.3327449249779346e-05,
+      "loss": 0.6901,
+      "step": 9450
+    },
+    {
+      "epoch": 1.6768158150207397,
+      "grad_norm": 3.5278499126434326,
+      "learning_rate": 3.323918799646955e-05,
+      "loss": 0.6499,
+      "step": 9500
+    },
+    {
+      "epoch": 1.6856411614155855,
+      "grad_norm": 2.8567793369293213,
+      "learning_rate": 3.315092674315976e-05,
+      "loss": 0.6201,
+      "step": 9550
+    },
+    {
+      "epoch": 1.6944665078104315,
+      "grad_norm": 2.610034465789795,
+      "learning_rate": 3.306266548984996e-05,
+      "loss": 0.6407,
+      "step": 9600
+    },
+    {
+      "epoch": 1.7032918542052775,
+      "grad_norm": 3.1110494136810303,
+      "learning_rate": 3.297440423654016e-05,
+      "loss": 0.6229,
+      "step": 9650
+    },
+    {
+      "epoch": 1.7121172006001235,
+      "grad_norm": 3.0331647396087646,
+      "learning_rate": 3.288614298323036e-05,
+      "loss": 0.6331,
+      "step": 9700
+    },
+    {
+      "epoch": 1.7209425469949695,
+      "grad_norm": 2.306661605834961,
+      "learning_rate": 3.279788172992057e-05,
+      "loss": 0.6221,
+      "step": 9750
+    },
+    {
+      "epoch": 1.7297678933898155,
+      "grad_norm": 3.5243642330169678,
+      "learning_rate": 3.2709620476610766e-05,
+      "loss": 0.652,
+      "step": 9800
+    },
+    {
+      "epoch": 1.7385932397846615,
+      "grad_norm": 3.3176615238189697,
+      "learning_rate": 3.262135922330097e-05,
+      "loss": 0.6614,
+      "step": 9850
+    },
+    {
+      "epoch": 1.7474185861795075,
+      "grad_norm": 2.874948501586914,
+      "learning_rate": 3.253309796999118e-05,
+      "loss": 0.6597,
+      "step": 9900
+    },
+    {
+      "epoch": 1.7562439325743535,
+      "grad_norm": 3.0557405948638916,
+      "learning_rate": 3.2444836716681377e-05,
+      "loss": 0.62,
+      "step": 9950
+    },
+    {
+      "epoch": 1.7650692789691995,
+      "grad_norm": 3.2125186920166016,
+      "learning_rate": 3.235657546337158e-05,
+      "loss": 0.6577,
+      "step": 10000
+    },
+    {
+      "epoch": 1.7650692789691995,
+      "eval_loss": 0.7176974415779114,
+      "eval_runtime": 181.8356,
+      "eval_samples_per_second": 155.784,
+      "eval_steps_per_second": 15.58,
+      "step": 10000
+    },
+    {
+      "epoch": 1.7738946253640455,
+      "grad_norm": 3.4087321758270264,
+      "learning_rate": 3.226831421006178e-05,
+      "loss": 0.6412,
+      "step": 10050
+    },
+    {
+      "epoch": 1.7827199717588915,
+      "grad_norm": 3.135465383529663,
+      "learning_rate": 3.218005295675199e-05,
+      "loss": 0.6436,
+      "step": 10100
+    },
+    {
+      "epoch": 1.7915453181537375,
+      "grad_norm": 3.334916830062866,
+      "learning_rate": 3.209179170344219e-05,
+      "loss": 0.6381,
+      "step": 10150
+    },
+    {
+      "epoch": 1.8003706645485835,
+      "grad_norm": 2.579524278640747,
+      "learning_rate": 3.200353045013239e-05,
+      "loss": 0.6083,
+      "step": 10200
+    },
+    {
+      "epoch": 1.8091960109434295,
+      "grad_norm": 3.2345893383026123,
+      "learning_rate": 3.19152691968226e-05,
+      "loss": 0.6353,
+      "step": 10250
+    },
+    {
+      "epoch": 1.8180213573382755,
+      "grad_norm": 3.3575096130371094,
+      "learning_rate": 3.1827007943512796e-05,
+      "loss": 0.627,
+      "step": 10300
+    },
+    {
+      "epoch": 1.8268467037331215,
+      "grad_norm": 2.8366053104400635,
+      "learning_rate": 3.1738746690203e-05,
+      "loss": 0.6247,
+      "step": 10350
+    },
+    {
+      "epoch": 1.8356720501279675,
+      "grad_norm": 3.510075330734253,
+      "learning_rate": 3.165048543689321e-05,
+      "loss": 0.6429,
+      "step": 10400
+    },
+    {
+      "epoch": 1.8444973965228135,
+      "grad_norm": 2.613879919052124,
+      "learning_rate": 3.1562224183583414e-05,
+      "loss": 0.6243,
+      "step": 10450
+    },
+    {
+      "epoch": 1.8533227429176595,
+      "grad_norm": 3.2765021324157715,
+      "learning_rate": 3.147396293027361e-05,
+      "loss": 0.6741,
+      "step": 10500
+    },
+    {
+      "epoch": 1.8621480893125055,
+      "grad_norm": 3.1736323833465576,
+      "learning_rate": 3.138570167696381e-05,
+      "loss": 0.6523,
+      "step": 10550
+    },
+    {
+      "epoch": 1.8709734357073515,
+      "grad_norm": 2.467745065689087,
+      "learning_rate": 3.129744042365402e-05,
+      "loss": 0.6149,
+      "step": 10600
+    },
+    {
+      "epoch": 1.8797987821021975,
+      "grad_norm": 3.506396532058716,
+      "learning_rate": 3.120917917034422e-05,
+      "loss": 0.625,
+      "step": 10650
+    },
+    {
+      "epoch": 1.8886241284970435,
+      "grad_norm": 3.8226113319396973,
+      "learning_rate": 3.112091791703443e-05,
+      "loss": 0.6314,
+      "step": 10700
+    },
+    {
+      "epoch": 1.8974494748918895,
+      "grad_norm": 3.097027540206909,
+      "learning_rate": 3.103265666372462e-05,
+      "loss": 0.6384,
+      "step": 10750
+    },
+    {
+      "epoch": 1.9062748212867355,
+      "grad_norm": 2.8758792877197266,
+      "learning_rate": 3.0944395410414827e-05,
+      "loss": 0.6459,
+      "step": 10800
+    },
+    {
+      "epoch": 1.9151001676815815,
+      "grad_norm": 3.3849267959594727,
+      "learning_rate": 3.085613415710503e-05,
+      "loss": 0.6229,
+      "step": 10850
+    },
+    {
+      "epoch": 1.9239255140764275,
+      "grad_norm": 3.4073870182037354,
+      "learning_rate": 3.076787290379524e-05,
+      "loss": 0.6299,
+      "step": 10900
+    },
+    {
+      "epoch": 1.9327508604712735,
+      "grad_norm": 2.961178779602051,
+      "learning_rate": 3.067961165048544e-05,
+      "loss": 0.6157,
+      "step": 10950
+    },
+    {
+      "epoch": 1.9415762068661195,
+      "grad_norm": 3.150435447692871,
+      "learning_rate": 3.0591350397175636e-05,
+      "loss": 0.6094,
+      "step": 11000
+    },
+    {
+      "epoch": 1.9415762068661195,
+      "eval_loss": 0.7086746692657471,
+      "eval_runtime": 181.7779,
+      "eval_samples_per_second": 155.833,
+      "eval_steps_per_second": 15.585,
+      "step": 11000
+    },
+    {
+      "epoch": 1.9504015532609655,
+      "grad_norm": 2.8941028118133545,
+      "learning_rate": 3.050308914386584e-05,
+      "loss": 0.6292,
+      "step": 11050
+    },
+    {
+      "epoch": 1.9592268996558115,
+      "grad_norm": 3.9323644638061523,
+      "learning_rate": 3.0414827890556047e-05,
+      "loss": 0.6188,
+      "step": 11100
+    },
+    {
+      "epoch": 1.9680522460506575,
+      "grad_norm": 3.0189285278320312,
+      "learning_rate": 3.032656663724625e-05,
+      "loss": 0.6135,
+      "step": 11150
+    },
+    {
+      "epoch": 1.9768775924455035,
+      "grad_norm": 4.591734886169434,
+      "learning_rate": 3.0238305383936455e-05,
+      "loss": 0.6226,
+      "step": 11200
+    },
+    {
+      "epoch": 1.9857029388403493,
+      "grad_norm": 2.5225682258605957,
+      "learning_rate": 3.0150044130626658e-05,
+      "loss": 0.6162,
+      "step": 11250
+    },
+    {
+      "epoch": 1.9945282852351953,
+      "grad_norm": 2.399808406829834,
+      "learning_rate": 3.0061782877316857e-05,
+      "loss": 0.6034,
+      "step": 11300
+    },
+    {
+      "epoch": 2.0033536316300413,
+      "grad_norm": 2.693629503250122,
+      "learning_rate": 2.9973521624007062e-05,
+      "loss": 0.5803,
+      "step": 11350
+    },
+    {
+      "epoch": 2.0121789780248873,
+      "grad_norm": 2.8983335494995117,
+      "learning_rate": 2.9885260370697265e-05,
+      "loss": 0.496,
+      "step": 11400
+    },
+    {
+      "epoch": 2.0210043244197333,
+      "grad_norm": 3.2255337238311768,
+      "learning_rate": 2.979699911738747e-05,
+      "loss": 0.4722,
+      "step": 11450
+    },
+    {
+      "epoch": 2.0298296708145793,
+      "grad_norm": 2.8371284008026123,
+      "learning_rate": 2.9708737864077673e-05,
+      "loss": 0.4893,
+      "step": 11500
+    },
+    {
+      "epoch": 2.0386550172094253,
+      "grad_norm": 2.7969467639923096,
+      "learning_rate": 2.9620476610767872e-05,
+      "loss": 0.4663,
+      "step": 11550
+    },
+    {
+      "epoch": 2.0474803636042713,
+      "grad_norm": 2.523054361343384,
+      "learning_rate": 2.9532215357458078e-05,
+      "loss": 0.4998,
+      "step": 11600
+    },
+    {
+      "epoch": 2.0563057099991173,
+      "grad_norm": 3.1918458938598633,
+      "learning_rate": 2.944395410414828e-05,
+      "loss": 0.4968,
+      "step": 11650
+    },
+    {
+      "epoch": 2.0651310563939633,
+      "grad_norm": 3.167325735092163,
+      "learning_rate": 2.9355692850838486e-05,
+      "loss": 0.4688,
+      "step": 11700
+    },
+    {
+      "epoch": 2.0739564027888093,
+      "grad_norm": 3.2759768962860107,
+      "learning_rate": 2.9267431597528688e-05,
+      "loss": 0.5058,
+      "step": 11750
+    },
+    {
+      "epoch": 2.0827817491836553,
+      "grad_norm": 2.962247371673584,
+      "learning_rate": 2.9179170344218887e-05,
+      "loss": 0.4622,
+      "step": 11800
+    },
+    {
+      "epoch": 2.0916070955785013,
+      "grad_norm": 3.7349026203155518,
+      "learning_rate": 2.909090909090909e-05,
+      "loss": 0.4978,
+      "step": 11850
+    },
+    {
+      "epoch": 2.1004324419733473,
+      "grad_norm": 2.856795072555542,
+      "learning_rate": 2.9002647837599295e-05,
+      "loss": 0.4611,
+      "step": 11900
+    },
+    {
+      "epoch": 2.1092577883681933,
+      "grad_norm": 2.8157849311828613,
+      "learning_rate": 2.8914386584289497e-05,
+      "loss": 0.4769,
+      "step": 11950
+    },
+    {
+      "epoch": 2.1180831347630393,
+      "grad_norm": 3.712787389755249,
+      "learning_rate": 2.8826125330979703e-05,
+      "loss": 0.5045,
+      "step": 12000
+    },
+    {
+      "epoch": 2.1180831347630393,
+      "eval_loss": 0.7179752588272095,
+      "eval_runtime": 181.8101,
+      "eval_samples_per_second": 155.805,
+      "eval_steps_per_second": 15.582,
+      "step": 12000
+    },
+    {
+      "epoch": 2.1269084811578853,
+      "grad_norm": 2.407057285308838,
+      "learning_rate": 2.8737864077669902e-05,
+      "loss": 0.4973,
+      "step": 12050
+    },
+    {
+      "epoch": 2.1357338275527313,
+      "grad_norm": 2.9012491703033447,
+      "learning_rate": 2.8649602824360104e-05,
+      "loss": 0.4973,
+      "step": 12100
+    },
+    {
+      "epoch": 2.1445591739475773,
+      "grad_norm": 2.8666646480560303,
+      "learning_rate": 2.856134157105031e-05,
+      "loss": 0.4711,
+      "step": 12150
+    },
+    {
+      "epoch": 2.1533845203424233,
+      "grad_norm": 2.6431496143341064,
+      "learning_rate": 2.8473080317740512e-05,
+      "loss": 0.4555,
+      "step": 12200
+    },
+    {
+      "epoch": 2.1622098667372693,
+      "grad_norm": 2.309817314147949,
+      "learning_rate": 2.8384819064430718e-05,
+      "loss": 0.4654,
+      "step": 12250
+    },
+    {
+      "epoch": 2.1710352131321153,
+      "grad_norm": 2.9013850688934326,
+      "learning_rate": 2.829655781112092e-05,
+      "loss": 0.4767,
+      "step": 12300
+    },
+    {
+      "epoch": 2.1798605595269613,
+      "grad_norm": 3.384150266647339,
+      "learning_rate": 2.820829655781112e-05,
+      "loss": 0.5275,
+      "step": 12350
+    },
+    {
+      "epoch": 2.1886859059218073,
+      "grad_norm": 3.1525537967681885,
+      "learning_rate": 2.8120035304501325e-05,
+      "loss": 0.4777,
+      "step": 12400
+    },
+    {
+      "epoch": 2.1975112523166533,
+      "grad_norm": 2.6473567485809326,
+      "learning_rate": 2.8031774051191528e-05,
+      "loss": 0.4789,
+      "step": 12450
+    },
+    {
+      "epoch": 2.2063365987114993,
+      "grad_norm": 2.334127187728882,
+      "learning_rate": 2.7943512797881733e-05,
+      "loss": 0.4749,
+      "step": 12500
+    },
+    {
+      "epoch": 2.2151619451063453,
+      "grad_norm": 2.6121950149536133,
+      "learning_rate": 2.7855251544571936e-05,
+      "loss": 0.4807,
+      "step": 12550
+    },
+    {
+      "epoch": 2.2239872915011913,
+      "grad_norm": 2.5503621101379395,
+      "learning_rate": 2.7766990291262135e-05,
+      "loss": 0.4568,
+      "step": 12600
+    },
+    {
+      "epoch": 2.2328126378960373,
+      "grad_norm": 3.1507601737976074,
+      "learning_rate": 2.767872903795234e-05,
+      "loss": 0.5129,
+      "step": 12650
+    },
+    {
+      "epoch": 2.2416379842908833,
+      "grad_norm": 4.06856107711792,
+      "learning_rate": 2.7590467784642543e-05,
+      "loss": 0.4652,
+      "step": 12700
+    },
+    {
+      "epoch": 2.2504633306857293,
+      "grad_norm": 3.199040412902832,
+      "learning_rate": 2.750220653133275e-05,
+      "loss": 0.4995,
+      "step": 12750
+    },
+    {
+      "epoch": 2.2592886770805753,
+      "grad_norm": 3.08388352394104,
+      "learning_rate": 2.741394527802295e-05,
+      "loss": 0.5213,
+      "step": 12800
+    },
+    {
+      "epoch": 2.2681140234754213,
+      "grad_norm": 2.8300905227661133,
+      "learning_rate": 2.732568402471315e-05,
+      "loss": 0.5109,
+      "step": 12850
+    },
+    {
+      "epoch": 2.2769393698702673,
+      "grad_norm": 3.372019052505493,
+      "learning_rate": 2.7237422771403352e-05,
+      "loss": 0.482,
+      "step": 12900
+    },
+    {
+      "epoch": 2.2857647162651133,
+      "grad_norm": 3.004849433898926,
+      "learning_rate": 2.7149161518093558e-05,
+      "loss": 0.4918,
+      "step": 12950
+    },
+    {
+      "epoch": 2.2945900626599594,
+      "grad_norm": 3.0937423706054688,
+      "learning_rate": 2.706090026478376e-05,
+      "loss": 0.5041,
+      "step": 13000
+    },
+    {
+      "epoch": 2.2945900626599594,
+      "eval_loss": 0.7087730169296265,
+      "eval_runtime": 181.7582,
+      "eval_samples_per_second": 155.85,
+      "eval_steps_per_second": 15.587,
+      "step": 13000
+    },
+    {
+      "epoch": 2.3034154090548054,
+      "grad_norm": 3.027329444885254,
+      "learning_rate": 2.6972639011473966e-05,
+      "loss": 0.4969,
+      "step": 13050
+    },
+    {
+      "epoch": 2.3122407554496514,
+      "grad_norm": 2.688856840133667,
+      "learning_rate": 2.6884377758164168e-05,
+      "loss": 0.5064,
+      "step": 13100
+    },
+    {
+      "epoch": 2.3210661018444974,
+      "grad_norm": 3.5008068084716797,
+      "learning_rate": 2.6796116504854367e-05,
+      "loss": 0.4931,
+      "step": 13150
+    },
+    {
+      "epoch": 2.3298914482393434,
+      "grad_norm": 2.946478843688965,
+      "learning_rate": 2.6707855251544573e-05,
+      "loss": 0.5058,
+      "step": 13200
+    },
+    {
+      "epoch": 2.3387167946341894,
+      "grad_norm": 3.2251195907592773,
+      "learning_rate": 2.6619593998234775e-05,
+      "loss": 0.4723,
+      "step": 13250
+    },
+    {
+      "epoch": 2.3475421410290354,
+      "grad_norm": 2.8397462368011475,
+      "learning_rate": 2.653133274492498e-05,
+      "loss": 0.4753,
+      "step": 13300
+    },
+    {
+      "epoch": 2.3563674874238814,
+      "grad_norm": 2.383877992630005,
+      "learning_rate": 2.6443071491615183e-05,
+      "loss": 0.4403,
+      "step": 13350
+    },
+    {
+      "epoch": 2.3651928338187274,
+      "grad_norm": 2.9594268798828125,
+      "learning_rate": 2.6354810238305382e-05,
+      "loss": 0.4968,
+      "step": 13400
+    },
+    {
+      "epoch": 2.3740181802135734,
+      "grad_norm": 3.557107925415039,
+      "learning_rate": 2.6266548984995588e-05,
+      "loss": 0.4739,
+      "step": 13450
+    },
+    {
+      "epoch": 2.3828435266084194,
+      "grad_norm": 3.5984597206115723,
+      "learning_rate": 2.617828773168579e-05,
+      "loss": 0.5027,
+      "step": 13500
+    },
+    {
+      "epoch": 2.3916688730032654,
+      "grad_norm": 2.6769216060638428,
+      "learning_rate": 2.6090026478375996e-05,
+      "loss": 0.4781,
+      "step": 13550
+    },
+    {
+      "epoch": 2.4004942193981114,
+      "grad_norm": 4.396824836730957,
+      "learning_rate": 2.60017652250662e-05,
+      "loss": 0.5125,
+      "step": 13600
+    },
+    {
+      "epoch": 2.4093195657929574,
+      "grad_norm": 3.351771831512451,
+      "learning_rate": 2.5915269196822594e-05,
+      "loss": 0.5022,
+      "step": 13650
+    },
+    {
+      "epoch": 2.4181449121878034,
+      "grad_norm": 3.4032695293426514,
+      "learning_rate": 2.58270079435128e-05,
+      "loss": 0.4641,
+      "step": 13700
+    },
+    {
+      "epoch": 2.4269702585826494,
+      "grad_norm": 2.782778263092041,
+      "learning_rate": 2.5738746690203002e-05,
+      "loss": 0.4862,
+      "step": 13750
+    },
+    {
+      "epoch": 2.4357956049774954,
+      "grad_norm": 2.8130016326904297,
+      "learning_rate": 2.5650485436893208e-05,
+      "loss": 0.4982,
+      "step": 13800
+    },
+    {
+      "epoch": 2.4446209513723414,
+      "grad_norm": 3.3339552879333496,
+      "learning_rate": 2.556222418358341e-05,
+      "loss": 0.4999,
+      "step": 13850
+    },
+    {
+      "epoch": 2.4534462977671874,
+      "grad_norm": 2.952395439147949,
+      "learning_rate": 2.547396293027361e-05,
+      "loss": 0.5009,
+      "step": 13900
+    },
+    {
+      "epoch": 2.4622716441620334,
+      "grad_norm": 2.979508399963379,
+      "learning_rate": 2.5385701676963815e-05,
+      "loss": 0.4805,
+      "step": 13950
+    },
+    {
+      "epoch": 2.4710969905568794,
+      "grad_norm": 2.709047555923462,
+      "learning_rate": 2.5297440423654017e-05,
+      "loss": 0.5152,
+      "step": 14000
+    },
+    {
+      "epoch": 2.4710969905568794,
+      "eval_loss": 0.7000770568847656,
+      "eval_runtime": 181.7422,
+      "eval_samples_per_second": 155.864,
+      "eval_steps_per_second": 15.588,
+      "step": 14000
+    },
+    {
+      "epoch": 2.4799223369517254,
+      "grad_norm": 2.9618706703186035,
+      "learning_rate": 2.5209179170344223e-05,
+      "loss": 0.4763,
+      "step": 14050
+    },
+    {
+      "epoch": 2.4887476833465714,
+      "grad_norm": 2.6275711059570312,
+      "learning_rate": 2.5120917917034425e-05,
+      "loss": 0.5057,
+      "step": 14100
+    },
+    {
+      "epoch": 2.4975730297414174,
+      "grad_norm": 2.0146267414093018,
+      "learning_rate": 2.5032656663724624e-05,
+      "loss": 0.4706,
+      "step": 14150
+    },
+    {
+      "epoch": 2.5063983761362634,
+      "grad_norm": 3.7594754695892334,
+      "learning_rate": 2.4944395410414826e-05,
+      "loss": 0.5005,
+      "step": 14200
+    },
+    {
+      "epoch": 2.5152237225311094,
+      "grad_norm": 2.0959393978118896,
+      "learning_rate": 2.4856134157105032e-05,
+      "loss": 0.513,
+      "step": 14250
+    },
+    {
+      "epoch": 2.5240490689259554,
+      "grad_norm": 2.7866508960723877,
+      "learning_rate": 2.4767872903795238e-05,
+      "loss": 0.5094,
+      "step": 14300
+    },
+    {
+      "epoch": 2.5328744153208014,
+      "grad_norm": 3.7597134113311768,
+      "learning_rate": 2.4679611650485437e-05,
+      "loss": 0.4843,
+      "step": 14350
+    },
+    {
+      "epoch": 2.5416997617156474,
+      "grad_norm": 2.447920083999634,
+      "learning_rate": 2.4591350397175643e-05,
+      "loss": 0.4716,
+      "step": 14400
+    },
+    {
+      "epoch": 2.5505251081104934,
+      "grad_norm": 3.173151731491089,
+      "learning_rate": 2.450308914386584e-05,
+      "loss": 0.4899,
+      "step": 14450
+    },
+    {
+      "epoch": 2.5593504545053394,
+      "grad_norm": 2.6911866664886475,
+      "learning_rate": 2.4414827890556047e-05,
+      "loss": 0.4829,
+      "step": 14500
+    },
+    {
+      "epoch": 2.5681758009001854,
+      "grad_norm": 2.1910979747772217,
+      "learning_rate": 2.432656663724625e-05,
+      "loss": 0.5169,
+      "step": 14550
+    },
+    {
+      "epoch": 2.5770011472950314,
+      "grad_norm": 3.5892837047576904,
+      "learning_rate": 2.4238305383936452e-05,
+      "loss": 0.5169,
+      "step": 14600
+    },
+    {
+      "epoch": 2.5858264936898774,
+      "grad_norm": 4.015511989593506,
+      "learning_rate": 2.4150044130626658e-05,
+      "loss": 0.4537,
+      "step": 14650
+    },
+    {
+      "epoch": 2.5946518400847234,
+      "grad_norm": 3.547804832458496,
+      "learning_rate": 2.406178287731686e-05,
+      "loss": 0.5112,
+      "step": 14700
+    },
+    {
+      "epoch": 2.6034771864795694,
+      "grad_norm": 3.3924198150634766,
+      "learning_rate": 2.3973521624007062e-05,
+      "loss": 0.4676,
+      "step": 14750
+    },
+    {
+      "epoch": 2.6123025328744154,
+      "grad_norm": 2.915863037109375,
+      "learning_rate": 2.3885260370697265e-05,
+      "loss": 0.486,
+      "step": 14800
+    },
+    {
+      "epoch": 2.6211278792692614,
+      "grad_norm": 2.734058380126953,
+      "learning_rate": 2.3796999117387467e-05,
+      "loss": 0.4795,
+      "step": 14850
+    },
+    {
+      "epoch": 2.6299532256641074,
+      "grad_norm": 2.50942325592041,
+      "learning_rate": 2.370873786407767e-05,
+      "loss": 0.4739,
+      "step": 14900
+    },
+    {
+      "epoch": 2.6387785720589534,
+      "grad_norm": 3.3483402729034424,
+      "learning_rate": 2.3620476610767875e-05,
+      "loss": 0.5242,
+      "step": 14950
+    },
+    {
+      "epoch": 2.6476039184537994,
+      "grad_norm": 2.779780149459839,
+      "learning_rate": 2.3532215357458077e-05,
+      "loss": 0.465,
+      "step": 15000
+    },
+    {
+      "epoch": 2.6476039184537994,
+      "eval_loss": 0.7009094953536987,
+      "eval_runtime": 181.7609,
+      "eval_samples_per_second": 155.848,
+      "eval_steps_per_second": 15.586,
+      "step": 15000
+    },
+    {
+      "epoch": 2.6564292648486454,
+      "grad_norm": 3.3387420177459717,
+      "learning_rate": 2.344395410414828e-05,
+      "loss": 0.5071,
+      "step": 15050
+    },
+    {
+      "epoch": 2.6652546112434914,
+      "grad_norm": 3.358072519302368,
+      "learning_rate": 2.3355692850838485e-05,
+      "loss": 0.4635,
+      "step": 15100
+    },
+    {
+      "epoch": 2.6740799576383374,
+      "grad_norm": 3.029219627380371,
+      "learning_rate": 2.3267431597528684e-05,
+      "loss": 0.4748,
+      "step": 15150
+    },
+    {
+      "epoch": 2.6829053040331834,
+      "grad_norm": 2.3864011764526367,
+      "learning_rate": 2.317917034421889e-05,
+      "loss": 0.4895,
+      "step": 15200
+    },
+    {
+      "epoch": 2.6917306504280294,
+      "grad_norm": 2.5701119899749756,
+      "learning_rate": 2.309090909090909e-05,
+      "loss": 0.4693,
+      "step": 15250
+    },
+    {
+      "epoch": 2.7005559968228754,
+      "grad_norm": 2.640472173690796,
+      "learning_rate": 2.3002647837599295e-05,
+      "loss": 0.4556,
+      "step": 15300
+    },
+    {
+      "epoch": 2.7093813432177214,
+      "grad_norm": 2.6650307178497314,
+      "learning_rate": 2.29143865842895e-05,
+      "loss": 0.4795,
+      "step": 15350
+    },
+    {
+      "epoch": 2.7182066896125674,
+      "grad_norm": 2.955892562866211,
+      "learning_rate": 2.28261253309797e-05,
+      "loss": 0.458,
+      "step": 15400
+    },
+    {
+      "epoch": 2.7270320360074134,
+      "grad_norm": 3.361410617828369,
+      "learning_rate": 2.2737864077669905e-05,
+      "loss": 0.5199,
+      "step": 15450
+    },
+    {
+      "epoch": 2.7358573824022594,
+      "grad_norm": 3.844050645828247,
+      "learning_rate": 2.2649602824360108e-05,
+      "loss": 0.4946,
+      "step": 15500
+    },
+    {
+      "epoch": 2.7446827287971054,
+      "grad_norm": 3.419942617416382,
+      "learning_rate": 2.256134157105031e-05,
+      "loss": 0.5105,
+      "step": 15550
+    },
+    {
+      "epoch": 2.7535080751919514,
+      "grad_norm": 2.58404278755188,
+      "learning_rate": 2.2473080317740512e-05,
+      "loss": 0.5022,
+      "step": 15600
+    },
+    {
+      "epoch": 2.7623334215867974,
+      "grad_norm": 2.5541560649871826,
+      "learning_rate": 2.2384819064430715e-05,
+      "loss": 0.5038,
+      "step": 15650
+    },
+    {
+      "epoch": 2.7711587679816434,
+      "grad_norm": 2.7854466438293457,
+      "learning_rate": 2.229655781112092e-05,
+      "loss": 0.501,
+      "step": 15700
+    },
+    {
+      "epoch": 2.7799841143764894,
+      "grad_norm": 2.999455213546753,
+      "learning_rate": 2.2208296557811123e-05,
+      "loss": 0.4941,
+      "step": 15750
+    },
+    {
+      "epoch": 2.7888094607713354,
+      "grad_norm": 3.169633388519287,
+      "learning_rate": 2.2120035304501325e-05,
+      "loss": 0.5011,
+      "step": 15800
+    },
+    {
+      "epoch": 2.7976348071661814,
+      "grad_norm": 2.6727561950683594,
+      "learning_rate": 2.2031774051191527e-05,
+      "loss": 0.4743,
+      "step": 15850
+    },
+    {
+      "epoch": 2.8064601535610274,
+      "grad_norm": 2.7246663570404053,
+      "learning_rate": 2.1943512797881733e-05,
+      "loss": 0.4549,
+      "step": 15900
+    },
+    {
+      "epoch": 2.8152854999558734,
+      "grad_norm": 3.7977099418640137,
+      "learning_rate": 2.1855251544571932e-05,
+      "loss": 0.4873,
+      "step": 15950
+    },
+    {
+      "epoch": 2.8241108463507194,
+      "grad_norm": 3.3139097690582275,
+      "learning_rate": 2.1766990291262138e-05,
+      "loss": 0.4491,
+      "step": 16000
+    },
+    {
+      "epoch": 2.8241108463507194,
+      "eval_loss": 0.6963682174682617,
+      "eval_runtime": 181.7512,
+      "eval_samples_per_second": 155.856,
+      "eval_steps_per_second": 15.587,
+      "step": 16000
+    },
+    {
+      "epoch": 2.8329361927455654,
+      "grad_norm": 3.014883279800415,
+      "learning_rate": 2.167872903795234e-05,
+      "loss": 0.503,
+      "step": 16050
+    },
+    {
+      "epoch": 2.8417615391404114,
+      "grad_norm": 3.094257116317749,
+      "learning_rate": 2.1590467784642542e-05,
+      "loss": 0.466,
+      "step": 16100
+    },
+    {
+      "epoch": 2.8505868855352574,
+      "grad_norm": 3.347677707672119,
+      "learning_rate": 2.1502206531332748e-05,
+      "loss": 0.4772,
+      "step": 16150
+    },
+    {
+      "epoch": 2.8594122319301034,
+      "grad_norm": 3.1895172595977783,
+      "learning_rate": 2.1413945278022947e-05,
+      "loss": 0.4722,
+      "step": 16200
+    },
+    {
+      "epoch": 2.8682375783249494,
+      "grad_norm": 3.3283205032348633,
+      "learning_rate": 2.1325684024713153e-05,
+      "loss": 0.4596,
+      "step": 16250
+    },
+    {
+      "epoch": 2.8770629247197954,
+      "grad_norm": 2.760470390319824,
+      "learning_rate": 2.1237422771403355e-05,
+      "loss": 0.4908,
+      "step": 16300
+    },
+    {
+      "epoch": 2.8858882711146414,
+      "grad_norm": 2.9019477367401123,
+      "learning_rate": 2.1149161518093558e-05,
+      "loss": 0.4889,
+      "step": 16350
+    },
+    {
+      "epoch": 2.8947136175094874,
+      "grad_norm": 2.706892251968384,
+      "learning_rate": 2.1060900264783763e-05,
+      "loss": 0.4896,
+      "step": 16400
+    },
+    {
+      "epoch": 2.9035389639043334,
+      "grad_norm": 2.5221893787384033,
+      "learning_rate": 2.0972639011473962e-05,
+      "loss": 0.4779,
+      "step": 16450
+    },
+    {
+      "epoch": 2.9123643102991794,
+      "grad_norm": 2.687797784805298,
+      "learning_rate": 2.0884377758164168e-05,
+      "loss": 0.4829,
+      "step": 16500
+    },
+    {
+      "epoch": 2.9211896566940254,
+      "grad_norm": 3.208223581314087,
+      "learning_rate": 2.079611650485437e-05,
+      "loss": 0.5011,
+      "step": 16550
+    },
+    {
+      "epoch": 2.9300150030888714,
+      "grad_norm": 3.5684010982513428,
+      "learning_rate": 2.0707855251544573e-05,
+      "loss": 0.5349,
+      "step": 16600
+    },
+    {
+      "epoch": 2.9388403494837174,
+      "grad_norm": 2.910231351852417,
+      "learning_rate": 2.0619593998234775e-05,
+      "loss": 0.4876,
+      "step": 16650
+    },
+    {
+      "epoch": 2.9476656958785634,
+      "grad_norm": 2.2015023231506348,
+      "learning_rate": 2.0531332744924977e-05,
+      "loss": 0.4947,
+      "step": 16700
+    },
+    {
+      "epoch": 2.9564910422734094,
+      "grad_norm": 3.053490400314331,
+      "learning_rate": 2.0443071491615183e-05,
+      "loss": 0.532,
+      "step": 16750
+    },
+    {
+      "epoch": 2.9653163886682554,
+      "grad_norm": 2.6305556297302246,
+      "learning_rate": 2.0354810238305385e-05,
+      "loss": 0.4773,
+      "step": 16800
+    },
+    {
+      "epoch": 2.9741417350631014,
+      "grad_norm": 1.92122220993042,
+      "learning_rate": 2.0266548984995588e-05,
+      "loss": 0.4753,
+      "step": 16850
+    },
+    {
+      "epoch": 2.9829670814579474,
+      "grad_norm": 2.6048147678375244,
+      "learning_rate": 2.017828773168579e-05,
+      "loss": 0.5042,
+      "step": 16900
+    },
+    {
+      "epoch": 2.9917924278527934,
+      "grad_norm": 2.4308252334594727,
+      "learning_rate": 2.0090026478375996e-05,
+      "loss": 0.4722,
+      "step": 16950
+    },
+    {
+      "epoch": 3.0006177742476394,
+      "grad_norm": 2.56939435005188,
+      "learning_rate": 2.0003530450132395e-05,
+      "loss": 0.4925,
+      "step": 17000
+    },
+    {
+      "epoch": 3.0006177742476394,
+      "eval_loss": 0.6830546855926514,
+      "eval_runtime": 181.7333,
+      "eval_samples_per_second": 155.871,
+      "eval_steps_per_second": 15.589,
+      "step": 17000
+    },
+    {
+      "epoch": 3.0094431206424854,
+      "grad_norm": 2.973703622817993,
+      "learning_rate": 1.9915269196822594e-05,
+      "loss": 0.3581,
+      "step": 17050
+    },
+    {
+      "epoch": 3.0182684670373314,
+      "grad_norm": 5.493050575256348,
+      "learning_rate": 1.98270079435128e-05,
+      "loss": 0.3559,
+      "step": 17100
+    },
+    {
+      "epoch": 3.0270938134321774,
+      "grad_norm": 2.2883782386779785,
+      "learning_rate": 1.9738746690203002e-05,
+      "loss": 0.3607,
+      "step": 17150
+    },
+    {
+      "epoch": 3.0359191598270234,
+      "grad_norm": 2.020956039428711,
+      "learning_rate": 1.9650485436893204e-05,
+      "loss": 0.3722,
+      "step": 17200
+    },
+    {
+      "epoch": 3.0447445062218694,
+      "grad_norm": 3.439941883087158,
+      "learning_rate": 1.9562224183583406e-05,
+      "loss": 0.3458,
+      "step": 17250
+    },
+    {
+      "epoch": 3.0535698526167154,
+      "grad_norm": 2.0934534072875977,
+      "learning_rate": 1.9473962930273612e-05,
+      "loss": 0.3543,
+      "step": 17300
+    },
+    {
+      "epoch": 3.062395199011561,
+      "grad_norm": 2.9070968627929688,
+      "learning_rate": 1.9385701676963815e-05,
+      "loss": 0.3893,
+      "step": 17350
+    },
+    {
+      "epoch": 3.071220545406407,
+      "grad_norm": 2.8557307720184326,
+      "learning_rate": 1.9297440423654017e-05,
+      "loss": 0.3579,
+      "step": 17400
+    },
+    {
+      "epoch": 3.080045891801253,
+      "grad_norm": 3.181415557861328,
+      "learning_rate": 1.920917917034422e-05,
+      "loss": 0.3933,
+      "step": 17450
+    },
+    {
+      "epoch": 3.088871238196099,
+      "grad_norm": 3.302639961242676,
+      "learning_rate": 1.912091791703442e-05,
+      "loss": 0.4011,
+      "step": 17500
+    },
+    {
+      "epoch": 3.097696584590945,
+      "grad_norm": 2.84566330909729,
+      "learning_rate": 1.9032656663724627e-05,
+      "loss": 0.3722,
+      "step": 17550
+    },
+    {
+      "epoch": 3.106521930985791,
+      "grad_norm": 2.536458730697632,
+      "learning_rate": 1.894439541041483e-05,
+      "loss": 0.3794,
+      "step": 17600
+    },
+    {
+      "epoch": 3.115347277380637,
+      "grad_norm": 2.6080989837646484,
+      "learning_rate": 1.8856134157105032e-05,
+      "loss": 0.3696,
+      "step": 17650
+    },
+    {
+      "epoch": 3.124172623775483,
+      "grad_norm": 3.233208417892456,
+      "learning_rate": 1.8767872903795238e-05,
+      "loss": 0.3693,
+      "step": 17700
+    },
+    {
+      "epoch": 3.132997970170329,
+      "grad_norm": 2.3864200115203857,
+      "learning_rate": 1.8679611650485437e-05,
+      "loss": 0.3647,
+      "step": 17750
+    },
+    {
+      "epoch": 3.141823316565175,
+      "grad_norm": 3.92567777633667,
+      "learning_rate": 1.8591350397175642e-05,
+      "loss": 0.3672,
+      "step": 17800
+    },
+    {
+      "epoch": 3.150648662960021,
+      "grad_norm": 2.5370068550109863,
+      "learning_rate": 1.850308914386584e-05,
+      "loss": 0.3576,
+      "step": 17850
+    },
+    {
+      "epoch": 3.159474009354867,
+      "grad_norm": 2.7078917026519775,
+      "learning_rate": 1.8414827890556047e-05,
+      "loss": 0.3667,
+      "step": 17900
+    },
+    {
+      "epoch": 3.168299355749713,
+      "grad_norm": 3.0950546264648438,
+      "learning_rate": 1.832656663724625e-05,
+      "loss": 0.3988,
+      "step": 17950
+    },
+    {
+      "epoch": 3.177124702144559,
+      "grad_norm": 2.078380823135376,
+      "learning_rate": 1.8238305383936452e-05,
+      "loss": 0.3736,
+      "step": 18000
+    },
+    {
+      "epoch": 3.177124702144559,
+      "eval_loss": 0.6996409296989441,
+      "eval_runtime": 181.8922,
+      "eval_samples_per_second": 155.735,
+      "eval_steps_per_second": 15.575,
+      "step": 18000
+    },
+    {
+      "epoch": 3.185950048539405,
+      "grad_norm": 2.491473913192749,
+      "learning_rate": 1.8150044130626657e-05,
+      "loss": 0.3493,
+      "step": 18050
+    },
+    {
+      "epoch": 3.194775394934251,
+      "grad_norm": 3.2102303504943848,
+      "learning_rate": 1.806178287731686e-05,
+      "loss": 0.3671,
+      "step": 18100
+    },
+    {
+      "epoch": 3.203600741329097,
+      "grad_norm": 2.8939054012298584,
+      "learning_rate": 1.7973521624007062e-05,
+      "loss": 0.3849,
+      "step": 18150
+    },
+    {
+      "epoch": 3.212426087723943,
+      "grad_norm": 2.372587203979492,
+      "learning_rate": 1.7885260370697265e-05,
+      "loss": 0.3651,
+      "step": 18200
+    },
+    {
+      "epoch": 3.221251434118789,
+      "grad_norm": 3.5877439975738525,
+      "learning_rate": 1.7796999117387467e-05,
+      "loss": 0.3736,
+      "step": 18250
+    },
+    {
+      "epoch": 3.230076780513635,
+      "grad_norm": 3.3625779151916504,
+      "learning_rate": 1.770873786407767e-05,
+      "loss": 0.3653,
+      "step": 18300
+    },
+    {
+      "epoch": 3.238902126908481,
+      "grad_norm": 3.4853782653808594,
+      "learning_rate": 1.7620476610767875e-05,
+      "loss": 0.3615,
+      "step": 18350
+    },
+    {
+      "epoch": 3.247727473303327,
+      "grad_norm": 4.208776950836182,
+      "learning_rate": 1.7532215357458077e-05,
+      "loss": 0.3907,
+      "step": 18400
+    },
+    {
+      "epoch": 3.256552819698173,
+      "grad_norm": 3.6717019081115723,
+      "learning_rate": 1.744395410414828e-05,
+      "loss": 0.3682,
+      "step": 18450
+    },
+    {
+      "epoch": 3.265378166093019,
+      "grad_norm": 2.0552759170532227,
+      "learning_rate": 1.7355692850838482e-05,
+      "loss": 0.3786,
+      "step": 18500
+    },
+    {
+      "epoch": 3.274203512487865,
+      "grad_norm": 2.2906410694122314,
+      "learning_rate": 1.7267431597528684e-05,
+      "loss": 0.3941,
+      "step": 18550
+    },
+    {
+      "epoch": 3.283028858882711,
+      "grad_norm": 2.546051263809204,
+      "learning_rate": 1.717917034421889e-05,
+      "loss": 0.374,
+      "step": 18600
+    },
+    {
+      "epoch": 3.291854205277557,
+      "grad_norm": 3.243657350540161,
+      "learning_rate": 1.7090909090909092e-05,
+      "loss": 0.4133,
+      "step": 18650
+    },
+    {
+      "epoch": 3.300679551672403,
+      "grad_norm": 3.139357328414917,
+      "learning_rate": 1.7002647837599295e-05,
+      "loss": 0.3807,
+      "step": 18700
+    },
+    {
+      "epoch": 3.309504898067249,
+      "grad_norm": 4.064445972442627,
+      "learning_rate": 1.69143865842895e-05,
+      "loss": 0.386,
+      "step": 18750
+    },
+    {
+      "epoch": 3.318330244462095,
+      "grad_norm": 2.3872506618499756,
+      "learning_rate": 1.68261253309797e-05,
+      "loss": 0.3618,
+      "step": 18800
+    },
+    {
+      "epoch": 3.327155590856941,
+      "grad_norm": 2.3966455459594727,
+      "learning_rate": 1.6737864077669905e-05,
+      "loss": 0.3862,
+      "step": 18850
+    },
+    {
+      "epoch": 3.335980937251787,
+      "grad_norm": 3.9382779598236084,
+      "learning_rate": 1.6649602824360104e-05,
+      "loss": 0.344,
+      "step": 18900
+    },
+    {
+      "epoch": 3.344806283646633,
+      "grad_norm": 3.825925588607788,
+      "learning_rate": 1.656134157105031e-05,
+      "loss": 0.3657,
+      "step": 18950
+    },
+    {
+      "epoch": 3.353631630041479,
+      "grad_norm": 2.955763101577759,
+      "learning_rate": 1.6473080317740512e-05,
+      "loss": 0.4116,
+      "step": 19000
+    },
+    {
+      "epoch": 3.353631630041479,
+      "eval_loss": 0.7010481953620911,
+      "eval_runtime": 181.7645,
+      "eval_samples_per_second": 155.845,
+      "eval_steps_per_second": 15.586,
+      "step": 19000
+    },
+    {
+      "epoch": 3.362456976436325,
+      "grad_norm": 3.2211873531341553,
+      "learning_rate": 1.6384819064430714e-05,
+      "loss": 0.3662,
+      "step": 19050
+    },
+    {
+      "epoch": 3.371282322831171,
+      "grad_norm": 3.256361722946167,
+      "learning_rate": 1.629655781112092e-05,
+      "loss": 0.3738,
+      "step": 19100
+    },
+    {
+      "epoch": 3.380107669226017,
+      "grad_norm": 2.363861560821533,
+      "learning_rate": 1.6208296557811123e-05,
+      "loss": 0.4157,
+      "step": 19150
+    },
+    {
+      "epoch": 3.388933015620863,
+      "grad_norm": 3.10766339302063,
+      "learning_rate": 1.6120035304501325e-05,
+      "loss": 0.3781,
+      "step": 19200
+    },
+    {
+      "epoch": 3.397758362015709,
+      "grad_norm": 2.534269332885742,
+      "learning_rate": 1.6031774051191527e-05,
+      "loss": 0.3727,
+      "step": 19250
+    },
+    {
+      "epoch": 3.406583708410555,
+      "grad_norm": 2.8136239051818848,
+      "learning_rate": 1.594351279788173e-05,
+      "loss": 0.3843,
+      "step": 19300
+    },
+    {
+      "epoch": 3.415409054805401,
+      "grad_norm": 2.956463575363159,
+      "learning_rate": 1.5855251544571932e-05,
+      "loss": 0.3829,
+      "step": 19350
+    },
+    {
+      "epoch": 3.424234401200247,
+      "grad_norm": 3.122279644012451,
+      "learning_rate": 1.5766990291262138e-05,
+      "loss": 0.3771,
+      "step": 19400
+    },
+    {
+      "epoch": 3.433059747595093,
+      "grad_norm": 3.2590718269348145,
+      "learning_rate": 1.567872903795234e-05,
+      "loss": 0.3508,
+      "step": 19450
+    },
+    {
+      "epoch": 3.441885093989939,
+      "grad_norm": 2.1195740699768066,
+      "learning_rate": 1.5590467784642542e-05,
+      "loss": 0.3806,
+      "step": 19500
+    },
+    {
+      "epoch": 3.450710440384785,
+      "grad_norm": 2.7716526985168457,
+      "learning_rate": 1.5502206531332748e-05,
+      "loss": 0.372,
+      "step": 19550
+    },
+    {
+      "epoch": 3.459535786779631,
+      "grad_norm": 2.7619106769561768,
+      "learning_rate": 1.5413945278022947e-05,
+      "loss": 0.3641,
+      "step": 19600
+    },
+    {
+      "epoch": 3.468361133174477,
+      "grad_norm": 3.310124635696411,
+      "learning_rate": 1.5325684024713153e-05,
+      "loss": 0.3525,
+      "step": 19650
+    },
+    {
+      "epoch": 3.477186479569323,
+      "grad_norm": 2.739682674407959,
+      "learning_rate": 1.523918799646955e-05,
+      "loss": 0.3682,
+      "step": 19700
+    },
+    {
+      "epoch": 3.486011825964169,
+      "grad_norm": 3.138272523880005,
+      "learning_rate": 1.5150926743159754e-05,
+      "loss": 0.3577,
+      "step": 19750
+    },
+    {
+      "epoch": 3.494837172359015,
+      "grad_norm": 2.602055549621582,
+      "learning_rate": 1.5062665489849956e-05,
+      "loss": 0.3866,
+      "step": 19800
+    },
+    {
+      "epoch": 3.503662518753861,
+      "grad_norm": 3.8529045581817627,
+      "learning_rate": 1.497440423654016e-05,
+      "loss": 0.369,
+      "step": 19850
+    },
+    {
+      "epoch": 3.512487865148707,
+      "grad_norm": 2.2044355869293213,
+      "learning_rate": 1.4886142983230364e-05,
+      "loss": 0.3472,
+      "step": 19900
+    },
+    {
+      "epoch": 3.521313211543553,
+      "grad_norm": 4.661579132080078,
+      "learning_rate": 1.4797881729920565e-05,
+      "loss": 0.3698,
+      "step": 19950
+    },
+    {
+      "epoch": 3.530138557938399,
+      "grad_norm": 2.468601703643799,
+      "learning_rate": 1.4709620476610769e-05,
+      "loss": 0.3905,
+      "step": 20000
+    },
+    {
+      "epoch": 3.530138557938399,
+      "eval_loss": 0.6975388526916504,
+      "eval_runtime": 181.7923,
+      "eval_samples_per_second": 155.821,
+      "eval_steps_per_second": 15.584,
+      "step": 20000
+    },
+    {
+      "epoch": 3.538963904333245,
+      "grad_norm": 3.216001272201538,
+      "learning_rate": 1.462135922330097e-05,
+      "loss": 0.3641,
+      "step": 20050
+    },
+    {
+      "epoch": 3.547789250728091,
+      "grad_norm": 2.9274120330810547,
+      "learning_rate": 1.4533097969991174e-05,
+      "loss": 0.3772,
+      "step": 20100
+    },
+    {
+      "epoch": 3.556614597122937,
+      "grad_norm": 3.954235076904297,
+      "learning_rate": 1.4444836716681378e-05,
+      "loss": 0.3594,
+      "step": 20150
+    },
+    {
+      "epoch": 3.565439943517783,
+      "grad_norm": 2.441312551498413,
+      "learning_rate": 1.435657546337158e-05,
+      "loss": 0.3602,
+      "step": 20200
+    },
+    {
+      "epoch": 3.574265289912629,
+      "grad_norm": 2.377568006515503,
+      "learning_rate": 1.4268314210061784e-05,
+      "loss": 0.37,
+      "step": 20250
+    },
+    {
+      "epoch": 3.583090636307475,
+      "grad_norm": 2.7893033027648926,
+      "learning_rate": 1.4180052956751985e-05,
+      "loss": 0.3747,
+      "step": 20300
+    },
+    {
+      "epoch": 3.591915982702321,
+      "grad_norm": 5.03574800491333,
+      "learning_rate": 1.4091791703442189e-05,
+      "loss": 0.3832,
+      "step": 20350
+    },
+    {
+      "epoch": 3.600741329097167,
+      "grad_norm": 3.8684732913970947,
+      "learning_rate": 1.4003530450132393e-05,
+      "loss": 0.3635,
+      "step": 20400
+    },
+    {
+      "epoch": 3.609566675492013,
+      "grad_norm": 2.8690998554229736,
+      "learning_rate": 1.3915269196822595e-05,
+      "loss": 0.3532,
+      "step": 20450
+    },
+    {
+      "epoch": 3.618392021886859,
+      "grad_norm": 3.5299575328826904,
+      "learning_rate": 1.38270079435128e-05,
+      "loss": 0.3592,
+      "step": 20500
+    },
+    {
+      "epoch": 3.627217368281705,
+      "grad_norm": 2.8447256088256836,
+      "learning_rate": 1.3738746690203003e-05,
+      "loss": 0.3809,
+      "step": 20550
+    },
+    {
+      "epoch": 3.636042714676551,
+      "grad_norm": 2.9593961238861084,
+      "learning_rate": 1.3650485436893204e-05,
+      "loss": 0.3842,
+      "step": 20600
+    },
+    {
+      "epoch": 3.644868061071397,
+      "grad_norm": 2.8917315006256104,
+      "learning_rate": 1.3562224183583408e-05,
+      "loss": 0.3778,
+      "step": 20650
+    },
+    {
+      "epoch": 3.653693407466243,
+      "grad_norm": 2.705604314804077,
+      "learning_rate": 1.3473962930273609e-05,
+      "loss": 0.3856,
+      "step": 20700
+    },
+    {
+      "epoch": 3.662518753861089,
+      "grad_norm": 3.337663173675537,
+      "learning_rate": 1.3385701676963813e-05,
+      "loss": 0.3554,
+      "step": 20750
+    },
+    {
+      "epoch": 3.671344100255935,
+      "grad_norm": 2.7694709300994873,
+      "learning_rate": 1.3297440423654017e-05,
+      "loss": 0.3576,
+      "step": 20800
+    },
+    {
+      "epoch": 3.680169446650781,
+      "grad_norm": 3.0217745304107666,
+      "learning_rate": 1.3209179170344219e-05,
+      "loss": 0.3568,
+      "step": 20850
+    },
+    {
+      "epoch": 3.688994793045627,
+      "grad_norm": 3.2156639099121094,
+      "learning_rate": 1.3120917917034423e-05,
+      "loss": 0.3733,
+      "step": 20900
+    },
+    {
+      "epoch": 3.697820139440473,
+      "grad_norm": 3.0890026092529297,
+      "learning_rate": 1.3032656663724627e-05,
+      "loss": 0.3755,
+      "step": 20950
+    },
+    {
+      "epoch": 3.706645485835319,
+      "grad_norm": 2.598747730255127,
+      "learning_rate": 1.2944395410414828e-05,
+      "loss": 0.3959,
+      "step": 21000
+    },
+    {
+      "epoch": 3.706645485835319,
+      "eval_loss": 0.6940839290618896,
+      "eval_runtime": 181.7434,
+      "eval_samples_per_second": 155.863,
+      "eval_steps_per_second": 15.588,
+      "step": 21000
+    },
+    {
+      "epoch": 3.715470832230165,
+      "grad_norm": 3.0001120567321777,
+      "learning_rate": 1.2856134157105032e-05,
+      "loss": 0.3494,
+      "step": 21050
+    },
+    {
+      "epoch": 3.724296178625011,
+      "grad_norm": 3.8093442916870117,
+      "learning_rate": 1.2767872903795232e-05,
+      "loss": 0.4024,
+      "step": 21100
+    },
+    {
+      "epoch": 3.733121525019857,
+      "grad_norm": 3.1789517402648926,
+      "learning_rate": 1.2679611650485437e-05,
+      "loss": 0.3872,
+      "step": 21150
+    },
+    {
+      "epoch": 3.741946871414703,
+      "grad_norm": 3.2620937824249268,
+      "learning_rate": 1.259135039717564e-05,
+      "loss": 0.3594,
+      "step": 21200
+    },
+    {
+      "epoch": 3.750772217809549,
+      "grad_norm": 4.006773471832275,
+      "learning_rate": 1.2503089143865843e-05,
+      "loss": 0.3807,
+      "step": 21250
+    },
+    {
+      "epoch": 3.759597564204395,
+      "grad_norm": 3.351083517074585,
+      "learning_rate": 1.2414827890556047e-05,
+      "loss": 0.3591,
+      "step": 21300
+    },
+    {
+      "epoch": 3.768422910599241,
+      "grad_norm": 3.535079002380371,
+      "learning_rate": 1.232656663724625e-05,
+      "loss": 0.3901,
+      "step": 21350
+    },
+    {
+      "epoch": 3.777248256994087,
+      "grad_norm": 3.1112380027770996,
+      "learning_rate": 1.2238305383936452e-05,
+      "loss": 0.3548,
+      "step": 21400
+    },
+    {
+      "epoch": 3.786073603388933,
+      "grad_norm": 3.1333882808685303,
+      "learning_rate": 1.2150044130626656e-05,
+      "loss": 0.3696,
+      "step": 21450
+    },
+    {
+      "epoch": 3.794898949783779,
+      "grad_norm": 3.432013511657715,
+      "learning_rate": 1.2061782877316858e-05,
+      "loss": 0.3492,
+      "step": 21500
+    },
+    {
+      "epoch": 3.803724296178625,
+      "grad_norm": 3.25502347946167,
+      "learning_rate": 1.1973521624007062e-05,
+      "loss": 0.381,
+      "step": 21550
+    },
+    {
+      "epoch": 3.812549642573471,
+      "grad_norm": 3.45226788520813,
+      "learning_rate": 1.1885260370697264e-05,
+      "loss": 0.3853,
+      "step": 21600
+    },
+    {
+      "epoch": 3.821374988968317,
+      "grad_norm": 3.1285228729248047,
+      "learning_rate": 1.1796999117387468e-05,
+      "loss": 0.3897,
+      "step": 21650
+    },
+    {
+      "epoch": 3.830200335363163,
+      "grad_norm": 2.432709217071533,
+      "learning_rate": 1.170873786407767e-05,
+      "loss": 0.3672,
+      "step": 21700
+    },
+    {
+      "epoch": 3.839025681758009,
+      "grad_norm": 2.7926766872406006,
+      "learning_rate": 1.1620476610767873e-05,
+      "loss": 0.365,
+      "step": 21750
+    },
+    {
+      "epoch": 3.847851028152855,
+      "grad_norm": 3.418755531311035,
+      "learning_rate": 1.1532215357458075e-05,
+      "loss": 0.3828,
+      "step": 21800
+    },
+    {
+      "epoch": 3.856676374547701,
+      "grad_norm": 3.320993185043335,
+      "learning_rate": 1.144395410414828e-05,
+      "loss": 0.3527,
+      "step": 21850
+    },
+    {
+      "epoch": 3.865501720942547,
+      "grad_norm": 3.38420033454895,
+      "learning_rate": 1.1355692850838483e-05,
+      "loss": 0.3768,
+      "step": 21900
+    },
+    {
+      "epoch": 3.874327067337393,
+      "grad_norm": 2.989563226699829,
+      "learning_rate": 1.1267431597528686e-05,
+      "loss": 0.3536,
+      "step": 21950
+    },
+    {
+      "epoch": 3.883152413732239,
+      "grad_norm": 2.897488832473755,
+      "learning_rate": 1.1179170344218888e-05,
+      "loss": 0.3853,
+      "step": 22000
+    },
+    {
+      "epoch": 3.883152413732239,
+      "eval_loss": 0.689664900302887,
+      "eval_runtime": 181.7323,
+      "eval_samples_per_second": 155.872,
+      "eval_steps_per_second": 15.589,
+      "step": 22000
+    },
+    {
+      "epoch": 3.891977760127085,
+      "grad_norm": 3.514223575592041,
+      "learning_rate": 1.1090909090909092e-05,
+      "loss": 0.3693,
+      "step": 22050
+    },
+    {
+      "epoch": 3.900803106521931,
+      "grad_norm": 3.4816884994506836,
+      "learning_rate": 1.1002647837599295e-05,
+      "loss": 0.3776,
+      "step": 22100
+    },
+    {
+      "epoch": 3.909628452916777,
+      "grad_norm": 3.4098060131073,
+      "learning_rate": 1.0914386584289497e-05,
+      "loss": 0.3825,
+      "step": 22150
+    },
+    {
+      "epoch": 3.918453799311623,
+      "grad_norm": 2.6594343185424805,
+      "learning_rate": 1.08261253309797e-05,
+      "loss": 0.3863,
+      "step": 22200
+    },
+    {
+      "epoch": 3.927279145706469,
+      "grad_norm": 2.3485662937164307,
+      "learning_rate": 1.07396293027361e-05,
+      "loss": 0.3735,
+      "step": 22250
+    },
+    {
+      "epoch": 3.936104492101315,
+      "grad_norm": 2.5646393299102783,
+      "learning_rate": 1.0651368049426302e-05,
+      "loss": 0.3884,
+      "step": 22300
+    },
+    {
+      "epoch": 3.944929838496161,
+      "grad_norm": 2.7431557178497314,
+      "learning_rate": 1.0563106796116505e-05,
+      "loss": 0.3416,
+      "step": 22350
+    },
+    {
+      "epoch": 3.953755184891007,
+      "grad_norm": 2.6949362754821777,
+      "learning_rate": 1.0474845542806707e-05,
+      "loss": 0.3718,
+      "step": 22400
+    },
+    {
+      "epoch": 3.962580531285853,
+      "grad_norm": 2.811133861541748,
+      "learning_rate": 1.0386584289496913e-05,
+      "loss": 0.3376,
+      "step": 22450
+    },
+    {
+      "epoch": 3.971405877680699,
+      "grad_norm": 2.7501027584075928,
+      "learning_rate": 1.0298323036187115e-05,
+      "loss": 0.3899,
+      "step": 22500
+    },
+    {
+      "epoch": 3.980231224075545,
+      "grad_norm": 3.2334673404693604,
+      "learning_rate": 1.0210061782877317e-05,
+      "loss": 0.3666,
+      "step": 22550
+    },
+    {
+      "epoch": 3.989056570470391,
+      "grad_norm": 2.5137345790863037,
+      "learning_rate": 1.012180052956752e-05,
+      "loss": 0.4036,
+      "step": 22600
+    },
+    {
+      "epoch": 3.997881916865237,
+      "grad_norm": 3.6843361854553223,
+      "learning_rate": 1.0033539276257724e-05,
+      "loss": 0.378,
+      "step": 22650
+    },
+    {
+      "epoch": 4.006707263260083,
+      "grad_norm": 3.4136714935302734,
+      "learning_rate": 9.945278022947926e-06,
+      "loss": 0.3131,
+      "step": 22700
+    },
+    {
+      "epoch": 4.015532609654929,
+      "grad_norm": 2.0005850791931152,
+      "learning_rate": 9.857016769638128e-06,
+      "loss": 0.2745,
+      "step": 22750
+    },
+    {
+      "epoch": 4.024357956049775,
+      "grad_norm": 3.076453447341919,
+      "learning_rate": 9.768755516328332e-06,
+      "loss": 0.2694,
+      "step": 22800
+    },
+    {
+      "epoch": 4.033183302444621,
+      "grad_norm": 2.3047375679016113,
+      "learning_rate": 9.680494263018536e-06,
+      "loss": 0.2986,
+      "step": 22850
+    },
+    {
+      "epoch": 4.042008648839467,
+      "grad_norm": 3.2981176376342773,
+      "learning_rate": 9.592233009708739e-06,
+      "loss": 0.3034,
+      "step": 22900
+    },
+    {
+      "epoch": 4.050833995234313,
+      "grad_norm": 1.8951621055603027,
+      "learning_rate": 9.503971756398941e-06,
+      "loss": 0.2673,
+      "step": 22950
+    },
+    {
+      "epoch": 4.059659341629159,
+      "grad_norm": 2.2991392612457275,
+      "learning_rate": 9.415710503089143e-06,
+      "loss": 0.2809,
+      "step": 23000
+    },
+    {
+      "epoch": 4.059659341629159,
+      "eval_loss": 0.7109093070030212,
+      "eval_runtime": 181.7083,
+      "eval_samples_per_second": 155.893,
+      "eval_steps_per_second": 15.591,
+      "step": 23000
+    },
+    {
+      "epoch": 4.068484688024005,
+      "grad_norm": 3.2459051609039307,
+      "learning_rate": 9.327449249779347e-06,
+      "loss": 0.2973,
+      "step": 23050
+    },
+    {
+      "epoch": 4.077310034418851,
+      "grad_norm": 2.494835615158081,
+      "learning_rate": 9.23918799646955e-06,
+      "loss": 0.2891,
+      "step": 23100
+    },
+    {
+      "epoch": 4.086135380813697,
+      "grad_norm": 3.9098784923553467,
+      "learning_rate": 9.150926743159754e-06,
+      "loss": 0.2718,
+      "step": 23150
+    },
+    {
+      "epoch": 4.094960727208543,
+      "grad_norm": 3.0526154041290283,
+      "learning_rate": 9.062665489849956e-06,
+      "loss": 0.3181,
+      "step": 23200
+    },
+    {
+      "epoch": 4.103786073603389,
+      "grad_norm": 2.8916709423065186,
+      "learning_rate": 8.97440423654016e-06,
+      "loss": 0.304,
+      "step": 23250
+    },
+    {
+      "epoch": 4.112611419998235,
+      "grad_norm": 2.576984405517578,
+      "learning_rate": 8.886142983230363e-06,
+      "loss": 0.2993,
+      "step": 23300
+    },
+    {
+      "epoch": 4.121436766393081,
+      "grad_norm": 2.776900053024292,
+      "learning_rate": 8.797881729920565e-06,
+      "loss": 0.2662,
+      "step": 23350
+    },
+    {
+      "epoch": 4.130262112787927,
+      "grad_norm": 2.9947192668914795,
+      "learning_rate": 8.709620476610767e-06,
+      "loss": 0.2965,
+      "step": 23400
+    },
+    {
+      "epoch": 4.139087459182773,
+      "grad_norm": 3.5091171264648438,
+      "learning_rate": 8.621359223300971e-06,
+      "loss": 0.311,
+      "step": 23450
+    },
+    {
+      "epoch": 4.147912805577619,
+      "grad_norm": 3.0500705242156982,
+      "learning_rate": 8.533097969991175e-06,
+      "loss": 0.3022,
+      "step": 23500
+    },
+    {
+      "epoch": 4.156738151972465,
+      "grad_norm": 3.1017823219299316,
+      "learning_rate": 8.444836716681378e-06,
+      "loss": 0.2615,
+      "step": 23550
+    },
+    {
+      "epoch": 4.165563498367311,
+      "grad_norm": 3.331235408782959,
+      "learning_rate": 8.35657546337158e-06,
+      "loss": 0.3127,
+      "step": 23600
+    },
+    {
+      "epoch": 4.174388844762157,
+      "grad_norm": 2.713762044906616,
+      "learning_rate": 8.268314210061784e-06,
+      "loss": 0.3146,
+      "step": 23650
+    },
+    {
+      "epoch": 4.183214191157003,
+      "grad_norm": 2.274113178253174,
+      "learning_rate": 8.180052956751986e-06,
+      "loss": 0.2899,
+      "step": 23700
+    },
+    {
+      "epoch": 4.192039537551849,
+      "grad_norm": 2.678398370742798,
+      "learning_rate": 8.091791703442189e-06,
+      "loss": 0.2949,
+      "step": 23750
+    },
+    {
+      "epoch": 4.200864883946695,
+      "grad_norm": 2.3312747478485107,
+      "learning_rate": 8.003530450132391e-06,
+      "loss": 0.3019,
+      "step": 23800
+    },
+    {
+      "epoch": 4.209690230341541,
+      "grad_norm": 2.080692768096924,
+      "learning_rate": 7.915269196822595e-06,
+      "loss": 0.281,
+      "step": 23850
+    },
+    {
+      "epoch": 4.218515576736387,
+      "grad_norm": 2.982898235321045,
+      "learning_rate": 7.827007943512799e-06,
+      "loss": 0.3062,
+      "step": 23900
+    },
+    {
+      "epoch": 4.227340923131233,
+      "grad_norm": 2.4919323921203613,
+      "learning_rate": 7.738746690203001e-06,
+      "loss": 0.2851,
+      "step": 23950
+    },
+    {
+      "epoch": 4.236166269526079,
+      "grad_norm": 2.330500841140747,
+      "learning_rate": 7.650485436893204e-06,
+      "loss": 0.3078,
+      "step": 24000
+    },
+    {
+      "epoch": 4.236166269526079,
+      "eval_loss": 0.7070273756980896,
+      "eval_runtime": 181.7435,
+      "eval_samples_per_second": 155.863,
+      "eval_steps_per_second": 15.588,
+      "step": 24000
+    },
+    {
+      "epoch": 4.244991615920925,
+      "grad_norm": 3.2452151775360107,
+      "learning_rate": 7.562224183583408e-06,
+      "loss": 0.3133,
+      "step": 24050
+    },
+    {
+      "epoch": 4.253816962315771,
+      "grad_norm": 3.1312410831451416,
+      "learning_rate": 7.47396293027361e-06,
+      "loss": 0.2953,
+      "step": 24100
+    },
+    {
+      "epoch": 4.262642308710617,
+      "grad_norm": 3.128032684326172,
+      "learning_rate": 7.385701676963813e-06,
+      "loss": 0.2757,
+      "step": 24150
+    },
+    {
+      "epoch": 4.271467655105463,
+      "grad_norm": 3.9495296478271484,
+      "learning_rate": 7.297440423654016e-06,
+      "loss": 0.2845,
+      "step": 24200
+    },
+    {
+      "epoch": 4.280293001500309,
+      "grad_norm": 3.0301661491394043,
+      "learning_rate": 7.20917917034422e-06,
+      "loss": 0.3086,
+      "step": 24250
+    },
+    {
+      "epoch": 4.289118347895155,
+      "grad_norm": 3.9362733364105225,
+      "learning_rate": 7.120917917034423e-06,
+      "loss": 0.2845,
+      "step": 24300
+    },
+    {
+      "epoch": 4.297943694290001,
+      "grad_norm": 2.506880760192871,
+      "learning_rate": 7.032656663724625e-06,
+      "loss": 0.3013,
+      "step": 24350
+    },
+    {
+      "epoch": 4.306769040684847,
+      "grad_norm": 2.264787435531616,
+      "learning_rate": 6.944395410414828e-06,
+      "loss": 0.2966,
+      "step": 24400
+    },
+    {
+      "epoch": 4.315594387079693,
+      "grad_norm": 2.2223780155181885,
+      "learning_rate": 6.856134157105031e-06,
+      "loss": 0.2934,
+      "step": 24450
+    },
+    {
+      "epoch": 4.324419733474539,
+      "grad_norm": 3.4193592071533203,
+      "learning_rate": 6.767872903795235e-06,
+      "loss": 0.2961,
+      "step": 24500
+    },
+    {
+      "epoch": 4.333245079869385,
+      "grad_norm": 3.005798101425171,
+      "learning_rate": 6.679611650485437e-06,
+      "loss": 0.2712,
+      "step": 24550
+    },
+    {
+      "epoch": 4.342070426264231,
+      "grad_norm": 3.0497212409973145,
+      "learning_rate": 6.5913503971756395e-06,
+      "loss": 0.2939,
+      "step": 24600
+    },
+    {
+      "epoch": 4.350895772659077,
+      "grad_norm": 2.77939772605896,
+      "learning_rate": 6.503089143865843e-06,
+      "loss": 0.2936,
+      "step": 24650
+    },
+    {
+      "epoch": 4.359721119053923,
+      "grad_norm": 2.7358789443969727,
+      "learning_rate": 6.414827890556047e-06,
+      "loss": 0.3102,
+      "step": 24700
+    },
+    {
+      "epoch": 4.368546465448769,
+      "grad_norm": 3.289128303527832,
+      "learning_rate": 6.326566637246249e-06,
+      "loss": 0.2994,
+      "step": 24750
+    },
+    {
+      "epoch": 4.377371811843615,
+      "grad_norm": 2.8301990032196045,
+      "learning_rate": 6.238305383936452e-06,
+      "loss": 0.2993,
+      "step": 24800
+    },
+    {
+      "epoch": 4.386197158238461,
+      "grad_norm": 3.143252372741699,
+      "learning_rate": 6.1500441306266555e-06,
+      "loss": 0.2728,
+      "step": 24850
+    },
+    {
+      "epoch": 4.395022504633307,
+      "grad_norm": 2.6168856620788574,
+      "learning_rate": 6.061782877316858e-06,
+      "loss": 0.2935,
+      "step": 24900
+    },
+    {
+      "epoch": 4.403847851028153,
+      "grad_norm": 5.8992228507995605,
+      "learning_rate": 5.973521624007061e-06,
+      "loss": 0.3128,
+      "step": 24950
+    },
+    {
+      "epoch": 4.412673197422999,
+      "grad_norm": 3.9669909477233887,
+      "learning_rate": 5.88702559576346e-06,
+      "loss": 0.3087,
+      "step": 25000
+    },
+    {
+      "epoch": 4.412673197422999,
+      "eval_loss": 0.7075334191322327,
+      "eval_runtime": 181.7426,
+      "eval_samples_per_second": 155.863,
+      "eval_steps_per_second": 15.588,
+      "step": 25000
+    },
+    {
+      "epoch": 4.421498543817845,
+      "grad_norm": 1.8581490516662598,
+      "learning_rate": 5.798764342453663e-06,
+      "loss": 0.2811,
+      "step": 25050
+    },
+    {
+      "epoch": 4.430323890212691,
+      "grad_norm": 2.18139386177063,
+      "learning_rate": 5.710503089143866e-06,
+      "loss": 0.296,
+      "step": 25100
+    },
+    {
+      "epoch": 4.439149236607537,
+      "grad_norm": 2.703723907470703,
+      "learning_rate": 5.622241835834069e-06,
+      "loss": 0.2826,
+      "step": 25150
+    },
+    {
+      "epoch": 4.447974583002383,
+      "grad_norm": 3.0339534282684326,
+      "learning_rate": 5.533980582524273e-06,
+      "loss": 0.2828,
+      "step": 25200
+    },
+    {
+      "epoch": 4.456799929397229,
+      "grad_norm": 3.030467987060547,
+      "learning_rate": 5.445719329214475e-06,
+      "loss": 0.2899,
+      "step": 25250
+    },
+    {
+      "epoch": 4.465625275792075,
+      "grad_norm": 2.0632028579711914,
+      "learning_rate": 5.357458075904678e-06,
+      "loss": 0.3062,
+      "step": 25300
+    },
+    {
+      "epoch": 4.474450622186921,
+      "grad_norm": 2.691420793533325,
+      "learning_rate": 5.2691968225948806e-06,
+      "loss": 0.2933,
+      "step": 25350
+    },
+    {
+      "epoch": 4.483275968581767,
+      "grad_norm": 2.518090009689331,
+      "learning_rate": 5.180935569285085e-06,
+      "loss": 0.3056,
+      "step": 25400
+    },
+    {
+      "epoch": 4.492101314976613,
+      "grad_norm": 3.4329051971435547,
+      "learning_rate": 5.092674315975287e-06,
+      "loss": 0.2922,
+      "step": 25450
+    },
+    {
+      "epoch": 4.500926661371459,
+      "grad_norm": 1.8490076065063477,
+      "learning_rate": 5.00441306266549e-06,
+      "loss": 0.3004,
+      "step": 25500
+    },
+    {
+      "epoch": 4.509752007766305,
+      "grad_norm": 2.656369209289551,
+      "learning_rate": 4.916151809355693e-06,
+      "loss": 0.2917,
+      "step": 25550
+    },
+    {
+      "epoch": 4.518577354161151,
+      "grad_norm": 3.1583127975463867,
+      "learning_rate": 4.8278905560458965e-06,
+      "loss": 0.3037,
+      "step": 25600
+    },
+    {
+      "epoch": 4.527402700555997,
+      "grad_norm": 2.609161138534546,
+      "learning_rate": 4.739629302736099e-06,
+      "loss": 0.3102,
+      "step": 25650
+    },
+    {
+      "epoch": 4.536228046950843,
+      "grad_norm": 3.1805193424224854,
+      "learning_rate": 4.651368049426302e-06,
+      "loss": 0.2879,
+      "step": 25700
+    },
+    {
+      "epoch": 4.545053393345689,
+      "grad_norm": 2.557835578918457,
+      "learning_rate": 4.563106796116505e-06,
+      "loss": 0.2958,
+      "step": 25750
+    },
+    {
+      "epoch": 4.553878739740535,
+      "grad_norm": 2.510469675064087,
+      "learning_rate": 4.474845542806708e-06,
+      "loss": 0.2947,
+      "step": 25800
+    },
+    {
+      "epoch": 4.562704086135381,
+      "grad_norm": 3.4820563793182373,
+      "learning_rate": 4.386584289496911e-06,
+      "loss": 0.2927,
+      "step": 25850
+    },
+    {
+      "epoch": 4.571529432530227,
+      "grad_norm": 3.2949700355529785,
+      "learning_rate": 4.298323036187114e-06,
+      "loss": 0.2697,
+      "step": 25900
+    },
+    {
+      "epoch": 4.580354778925073,
+      "grad_norm": 2.9834752082824707,
+      "learning_rate": 4.210061782877317e-06,
+      "loss": 0.3102,
+      "step": 25950
+    },
+    {
+      "epoch": 4.589180125319919,
+      "grad_norm": 3.7537124156951904,
+      "learning_rate": 4.12180052956752e-06,
+      "loss": 0.2956,
+      "step": 26000
+    },
+    {
+      "epoch": 4.589180125319919,
+      "eval_loss": 0.7058725953102112,
+      "eval_runtime": 181.7357,
+      "eval_samples_per_second": 155.869,
+      "eval_steps_per_second": 15.589,
+      "step": 26000
+    },
+    {
+      "epoch": 4.598005471714765,
+      "grad_norm": 2.9796440601348877,
+      "learning_rate": 4.033539276257723e-06,
+      "loss": 0.2979,
+      "step": 26050
+    },
+    {
+      "epoch": 4.606830818109611,
+      "grad_norm": 3.2112295627593994,
+      "learning_rate": 3.945278022947927e-06,
+      "loss": 0.2921,
+      "step": 26100
+    },
+    {
+      "epoch": 4.615656164504457,
+      "grad_norm": 2.114686965942383,
+      "learning_rate": 3.857016769638129e-06,
+      "loss": 0.2757,
+      "step": 26150
+    },
+    {
+      "epoch": 4.624481510899303,
+      "grad_norm": 2.44917368888855,
+      "learning_rate": 3.7687555163283322e-06,
+      "loss": 0.2593,
+      "step": 26200
+    },
+    {
+      "epoch": 4.633306857294149,
+      "grad_norm": 2.6650853157043457,
+      "learning_rate": 3.680494263018535e-06,
+      "loss": 0.3195,
+      "step": 26250
+    },
+    {
+      "epoch": 4.642132203688995,
+      "grad_norm": 2.660585403442383,
+      "learning_rate": 3.5922330097087378e-06,
+      "loss": 0.3108,
+      "step": 26300
+    },
+    {
+      "epoch": 4.650957550083841,
+      "grad_norm": 3.4605438709259033,
+      "learning_rate": 3.503971756398941e-06,
+      "loss": 0.2904,
+      "step": 26350
+    },
+    {
+      "epoch": 4.659782896478687,
+      "grad_norm": 2.301084041595459,
+      "learning_rate": 3.4157105030891437e-06,
+      "loss": 0.3159,
+      "step": 26400
+    },
+    {
+      "epoch": 4.668608242873533,
+      "grad_norm": 3.1047492027282715,
+      "learning_rate": 3.327449249779347e-06,
+      "loss": 0.3224,
+      "step": 26450
+    },
+    {
+      "epoch": 4.677433589268379,
+      "grad_norm": 2.4581010341644287,
+      "learning_rate": 3.2391879964695497e-06,
+      "loss": 0.284,
+      "step": 26500
+    },
+    {
+      "epoch": 4.686258935663225,
+      "grad_norm": 3.301814317703247,
+      "learning_rate": 3.150926743159753e-06,
+      "loss": 0.31,
+      "step": 26550
+    },
+    {
+      "epoch": 4.695084282058071,
+      "grad_norm": 2.921429395675659,
+      "learning_rate": 3.062665489849956e-06,
+      "loss": 0.2928,
+      "step": 26600
+    },
+    {
+      "epoch": 4.703909628452917,
+      "grad_norm": 2.5449726581573486,
+      "learning_rate": 2.9744042365401592e-06,
+      "loss": 0.298,
+      "step": 26650
+    },
+    {
+      "epoch": 4.712734974847763,
+      "grad_norm": 3.2756364345550537,
+      "learning_rate": 2.886142983230362e-06,
+      "loss": 0.3066,
+      "step": 26700
+    },
+    {
+      "epoch": 4.721560321242609,
+      "grad_norm": 2.0160672664642334,
+      "learning_rate": 2.797881729920565e-06,
+      "loss": 0.2864,
+      "step": 26750
+    },
+    {
+      "epoch": 4.730385667637455,
+      "grad_norm": 3.0313892364501953,
+      "learning_rate": 2.709620476610768e-06,
+      "loss": 0.2866,
+      "step": 26800
+    },
+    {
+      "epoch": 4.739211014032301,
+      "grad_norm": 3.7765920162200928,
+      "learning_rate": 2.621359223300971e-06,
+      "loss": 0.2816,
+      "step": 26850
+    },
+    {
+      "epoch": 4.748036360427147,
+      "grad_norm": 2.8294875621795654,
+      "learning_rate": 2.533097969991174e-06,
+      "loss": 0.2752,
+      "step": 26900
+    },
+    {
+      "epoch": 4.756861706821993,
+      "grad_norm": 2.9688727855682373,
+      "learning_rate": 2.444836716681377e-06,
+      "loss": 0.2961,
+      "step": 26950
+    },
+    {
+      "epoch": 4.765687053216839,
+      "grad_norm": 4.593882083892822,
+      "learning_rate": 2.3565754633715803e-06,
+      "loss": 0.2898,
+      "step": 27000
+    },
+    {
+      "epoch": 4.765687053216839,
+      "eval_loss": 0.705726683139801,
+      "eval_runtime": 181.7545,
+      "eval_samples_per_second": 155.853,
+      "eval_steps_per_second": 15.587,
+      "step": 27000
+    },
+    {
+      "epoch": 4.774512399611685,
+      "grad_norm": 3.7863380908966064,
+      "learning_rate": 2.268314210061783e-06,
+      "loss": 0.3242,
+      "step": 27050
+    },
+    {
+      "epoch": 4.783337746006531,
+      "grad_norm": 2.438889980316162,
+      "learning_rate": 2.1800529567519862e-06,
+      "loss": 0.2942,
+      "step": 27100
+    },
+    {
+      "epoch": 4.792163092401377,
+      "grad_norm": 2.7048282623291016,
+      "learning_rate": 2.091791703442189e-06,
+      "loss": 0.287,
+      "step": 27150
+    },
+    {
+      "epoch": 4.800988438796223,
+      "grad_norm": 3.619891881942749,
+      "learning_rate": 2.003530450132392e-06,
+      "loss": 0.2992,
+      "step": 27200
+    },
+    {
+      "epoch": 4.809813785191069,
+      "grad_norm": 1.8524075746536255,
+      "learning_rate": 1.915269196822595e-06,
+      "loss": 0.2748,
+      "step": 27250
+    },
+    {
+      "epoch": 4.818639131585915,
+      "grad_norm": 3.449615955352783,
+      "learning_rate": 1.8270079435127981e-06,
+      "loss": 0.3181,
+      "step": 27300
+    },
+    {
+      "epoch": 4.827464477980761,
+      "grad_norm": 3.258453130722046,
+      "learning_rate": 1.7387466902030011e-06,
+      "loss": 0.2837,
+      "step": 27350
+    },
+    {
+      "epoch": 4.836289824375607,
+      "grad_norm": 3.91930890083313,
+      "learning_rate": 1.650485436893204e-06,
+      "loss": 0.2842,
+      "step": 27400
+    },
+    {
+      "epoch": 4.845115170770453,
+      "grad_norm": 3.170217275619507,
+      "learning_rate": 1.5622241835834069e-06,
+      "loss": 0.2763,
+      "step": 27450
+    },
+    {
+      "epoch": 4.853940517165299,
+      "grad_norm": 3.373209238052368,
+      "learning_rate": 1.47396293027361e-06,
+      "loss": 0.2703,
+      "step": 27500
+    },
+    {
+      "epoch": 4.862765863560145,
+      "grad_norm": 3.7620341777801514,
+      "learning_rate": 1.385701676963813e-06,
+      "loss": 0.2945,
+      "step": 27550
+    },
+    {
+      "epoch": 4.871591209954991,
+      "grad_norm": 3.2243876457214355,
+      "learning_rate": 1.297440423654016e-06,
+      "loss": 0.2975,
+      "step": 27600
+    },
+    {
+      "epoch": 4.880416556349837,
+      "grad_norm": 3.1117992401123047,
+      "learning_rate": 1.209179170344219e-06,
+      "loss": 0.2922,
+      "step": 27650
+    },
+    {
+      "epoch": 4.889241902744683,
+      "grad_norm": 3.212538957595825,
+      "learning_rate": 1.120917917034422e-06,
+      "loss": 0.2898,
+      "step": 27700
+    },
+    {
+      "epoch": 4.898067249139529,
+      "grad_norm": 2.3424935340881348,
+      "learning_rate": 1.032656663724625e-06,
+      "loss": 0.2852,
+      "step": 27750
+    },
+    {
+      "epoch": 4.906892595534375,
+      "grad_norm": 2.8400442600250244,
+      "learning_rate": 9.44395410414828e-07,
+      "loss": 0.2948,
+      "step": 27800
+    },
+    {
+      "epoch": 4.915717941929221,
+      "grad_norm": 3.1694393157958984,
+      "learning_rate": 8.56134157105031e-07,
+      "loss": 0.2583,
+      "step": 27850
+    },
+    {
+      "epoch": 4.924543288324067,
+      "grad_norm": 2.80452299118042,
+      "learning_rate": 7.678729037952339e-07,
+      "loss": 0.2816,
+      "step": 27900
+    },
+    {
+      "epoch": 4.933368634718913,
+      "grad_norm": 2.7664706707000732,
+      "learning_rate": 6.813768755516329e-07,
+      "loss": 0.2806,
+      "step": 27950
+    },
+    {
+      "epoch": 4.942193981113759,
+      "grad_norm": 2.9960453510284424,
+      "learning_rate": 5.931156222418359e-07,
+      "loss": 0.2858,
+      "step": 28000
+    },
+    {
+      "epoch": 4.942193981113759,
+      "eval_loss": 0.7043320536613464,
+      "eval_runtime": 181.7383,
+      "eval_samples_per_second": 155.867,
+      "eval_steps_per_second": 15.588,
+      "step": 28000
+    },
+    {
+      "epoch": 4.951019327508605,
+      "grad_norm": 3.455805540084839,
+      "learning_rate": 5.048543689320388e-07,
+      "loss": 0.3116,
+      "step": 28050
+    },
+    {
+      "epoch": 4.959844673903451,
+      "grad_norm": 3.7969307899475098,
+      "learning_rate": 4.165931156222418e-07,
+      "loss": 0.2845,
+      "step": 28100
+    },
+    {
+      "epoch": 4.968670020298297,
+      "grad_norm": 3.1473402976989746,
+      "learning_rate": 3.2833186231244484e-07,
+      "loss": 0.2933,
+      "step": 28150
+    },
+    {
+      "epoch": 4.977495366693143,
+      "grad_norm": 3.197662115097046,
+      "learning_rate": 2.4007060900264787e-07,
+      "loss": 0.2918,
+      "step": 28200
+    },
+    {
+      "epoch": 4.986320713087989,
+      "grad_norm": 2.3752260208129883,
+      "learning_rate": 1.5180935569285085e-07,
+      "loss": 0.2969,
+      "step": 28250
+    },
+    {
+      "epoch": 4.995146059482835,
+      "grad_norm": 3.3417346477508545,
+      "learning_rate": 6.354810238305384e-08,
+      "loss": 0.2904,
+      "step": 28300
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 28325,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.3071621043067904e+18,
+  "train_batch_size": 10,
+  "trial_name": null,
+  "trial_params": null
+}