diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6314 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.03797345732829604,
+  "eval_steps": 1,
+  "global_step": 392,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 9.687106461300009e-05,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.9688,
+      "step": 1
+    },
+    {
+      "epoch": 9.687106461300009e-05,
+      "eval_accuracy": 4.851833655867371e-05,
+      "eval_loss": 10.96875,
+      "eval_runtime": 239.4378,
+      "eval_samples_per_second": 141.026,
+      "eval_steps_per_second": 2.205,
+      "step": 1
+    },
+    {
+      "epoch": 0.00019374212922600018,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.9609,
+      "step": 2
+    },
+    {
+      "epoch": 0.00019374212922600018,
+      "eval_accuracy": 4.851833655867371e-05,
+      "eval_loss": 10.96875,
+      "eval_runtime": 239.1345,
+      "eval_samples_per_second": 141.205,
+      "eval_steps_per_second": 2.208,
+      "step": 2
+    },
+    {
+      "epoch": 0.0002906131938390003,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.9609,
+      "step": 3
+    },
+    {
+      "epoch": 0.0002906131938390003,
+      "eval_accuracy": 4.851833655867371e-05,
+      "eval_loss": 10.96875,
+      "eval_runtime": 238.5289,
+      "eval_samples_per_second": 141.564,
+      "eval_steps_per_second": 2.214,
+      "step": 3
+    },
+    {
+      "epoch": 0.00038748425845200037,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.9609,
+      "step": 4
+    },
+    {
+      "epoch": 0.00038748425845200037,
+      "eval_accuracy": 4.851833655867371e-05,
+      "eval_loss": 10.96875,
+      "eval_runtime": 238.6311,
+      "eval_samples_per_second": 141.503,
+      "eval_steps_per_second": 2.213,
+      "step": 4
+    },
+    {
+      "epoch": 0.0004843553230650005,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.9609,
+      "step": 5
+    },
+    {
+      "epoch": 0.0004843553230650005,
+      "eval_accuracy": 4.851833655867371e-05,
+      "eval_loss": 10.96875,
+      "eval_runtime": 239.4512,
+      "eval_samples_per_second": 141.018,
+      "eval_steps_per_second": 2.205,
+      "step": 5
+    },
+    {
+      "epoch": 0.0005812263876780006,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.9688,
+      "step": 6
+    },
+    {
+      "epoch": 0.0005812263876780006,
+      "eval_accuracy": 4.851833655867371e-05,
+      "eval_loss": 10.96875,
+      "eval_runtime": 239.7687,
+      "eval_samples_per_second": 140.832,
+      "eval_steps_per_second": 2.202,
+      "step": 6
+    },
+    {
+      "epoch": 0.0006780974522910007,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.9609,
+      "step": 7
+    },
+    {
+      "epoch": 0.0006780974522910007,
+      "eval_accuracy": 4.851833655867371e-05,
+      "eval_loss": 10.96875,
+      "eval_runtime": 239.881,
+      "eval_samples_per_second": 140.766,
+      "eval_steps_per_second": 2.201,
+      "step": 7
+    },
+    {
+      "epoch": 0.0007749685169040007,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.9609,
+      "step": 8
+    },
+    {
+      "epoch": 0.0007749685169040007,
+      "eval_accuracy": 4.851833655867371e-05,
+      "eval_loss": 10.96875,
+      "eval_runtime": 238.2807,
+      "eval_samples_per_second": 141.711,
+      "eval_steps_per_second": 2.216,
+      "step": 8
+    },
+    {
+      "epoch": 0.0008718395815170009,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.9688,
+      "step": 9
+    },
+    {
+      "epoch": 0.0008718395815170009,
+      "eval_accuracy": 4.851833655867371e-05,
+      "eval_loss": 10.96875,
+      "eval_runtime": 238.8621,
+      "eval_samples_per_second": 141.366,
+      "eval_steps_per_second": 2.21,
+      "step": 9
+    },
+    {
+      "epoch": 0.000968710646130001,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.9531,
+      "step": 10
+    },
+    {
+      "epoch": 0.000968710646130001,
+      "eval_accuracy": 4.851833655867371e-05,
+      "eval_loss": 10.96875,
+      "eval_runtime": 238.2707,
+      "eval_samples_per_second": 141.717,
+      "eval_steps_per_second": 2.216,
+      "step": 10
+    },
+    {
+      "epoch": 0.001065581710743001,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.9688,
+      "step": 11
+    },
+    {
+      "epoch": 0.001065581710743001,
+      "eval_accuracy": 4.851833655867371e-05,
+      "eval_loss": 10.96875,
+      "eval_runtime": 238.7639,
+      "eval_samples_per_second": 141.424,
+      "eval_steps_per_second": 2.211,
+      "step": 11
+    },
+    {
+      "epoch": 0.0011624527753560012,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.9688,
+      "step": 12
+    },
+    {
+      "epoch": 0.0011624527753560012,
+      "eval_accuracy": 4.851833655867371e-05,
+      "eval_loss": 10.96875,
+      "eval_runtime": 236.7358,
+      "eval_samples_per_second": 142.636,
+      "eval_steps_per_second": 2.23,
+      "step": 12
+    },
+    {
+      "epoch": 0.0012593238399690012,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.9531,
+      "step": 13
+    },
+    {
+      "epoch": 0.0012593238399690012,
+      "eval_accuracy": 4.851833655867371e-05,
+      "eval_loss": 10.96875,
+      "eval_runtime": 238.6214,
+      "eval_samples_per_second": 141.509,
+      "eval_steps_per_second": 2.213,
+      "step": 13
+    },
+    {
+      "epoch": 0.0013561949045820013,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.9609,
+      "step": 14
+    },
+    {
+      "epoch": 0.0013561949045820013,
+      "eval_accuracy": 4.851833655867371e-05,
+      "eval_loss": 10.96875,
+      "eval_runtime": 238.9769,
+      "eval_samples_per_second": 141.298,
+      "eval_steps_per_second": 2.209,
+      "step": 14
+    },
+    {
+      "epoch": 0.0014530659691950015,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.9688,
+      "step": 15
+    },
+    {
+      "epoch": 0.0014530659691950015,
+      "eval_accuracy": 4.851833655867371e-05,
+      "eval_loss": 10.96875,
+      "eval_runtime": 238.6023,
+      "eval_samples_per_second": 141.52,
+      "eval_steps_per_second": 2.213,
+      "step": 15
+    },
+    {
+      "epoch": 0.0015499370338080015,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.9766,
+      "step": 16
+    },
+    {
+      "epoch": 0.0015499370338080015,
+      "eval_accuracy": 4.851833655867371e-05,
+      "eval_loss": 10.96875,
+      "eval_runtime": 238.5522,
+      "eval_samples_per_second": 141.55,
+      "eval_steps_per_second": 2.213,
+      "step": 16
+    },
+    {
+      "epoch": 0.0016468080984210016,
+      "grad_norm": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 10.9688,
+      "step": 17
+    },
+    {
+      "epoch": 0.0016468080984210016,
+      "eval_accuracy": 4.851833655867371e-05,
+      "eval_loss": 10.96875,
+      "eval_runtime": 239.2771,
+      "eval_samples_per_second": 141.121,
+      "eval_steps_per_second": 2.207,
+      "step": 17
+    },
+    {
+      "epoch": 0.0017436791630340018,
+      "grad_norm": 3.9638593196868896,
+      "learning_rate": 9.999990312893539e-06,
+      "loss": 10.9609,
+      "step": 18
+    },
+    {
+      "epoch": 0.0017436791630340018,
+      "eval_accuracy": 0.0007322042282688151,
+      "eval_loss": 10.8828125,
+      "eval_runtime": 238.4478,
+      "eval_samples_per_second": 141.612,
+      "eval_steps_per_second": 2.214,
+      "step": 18
+    },
+    {
+      "epoch": 0.0018405502276470018,
+      "grad_norm": 3.9969630241394043,
+      "learning_rate": 9.999980625787079e-06,
+      "loss": 10.8906,
+      "step": 19
+    },
+    {
+      "epoch": 0.0018405502276470018,
+      "eval_accuracy": 0.00505525749297823,
+      "eval_loss": 10.8046875,
+      "eval_runtime": 237.763,
+      "eval_samples_per_second": 142.02,
+      "eval_steps_per_second": 2.221,
+      "step": 19
+    },
+    {
+      "epoch": 0.001937421292260002,
+      "grad_norm": 3.8709299564361572,
+      "learning_rate": 9.999970938680617e-06,
+      "loss": 10.8359,
+      "step": 20
+    },
+    {
+      "epoch": 0.001937421292260002,
+      "eval_accuracy": 0.011155251410816828,
+      "eval_loss": 10.71875,
+      "eval_runtime": 238.5172,
+      "eval_samples_per_second": 141.57,
+      "eval_steps_per_second": 2.214,
+      "step": 20
+    },
+    {
+      "epoch": 0.002034292356873002,
+      "grad_norm": 3.787830352783203,
+      "learning_rate": 9.999961251574155e-06,
+      "loss": 10.75,
+      "step": 21
+    },
+    {
+      "epoch": 0.002034292356873002,
+      "eval_accuracy": 0.017461824594575886,
+      "eval_loss": 10.6484375,
+      "eval_runtime": 238.2061,
+      "eval_samples_per_second": 141.755,
+      "eval_steps_per_second": 2.217,
+      "step": 21
+    },
+    {
+      "epoch": 0.002131163421486002,
+      "grad_norm": 3.8295083045959473,
+      "learning_rate": 9.999951564467694e-06,
+      "loss": 10.6719,
+      "step": 22
+    },
+    {
+      "epoch": 0.002131163421486002,
+      "eval_accuracy": 0.02801033625841584,
+      "eval_loss": 10.578125,
+      "eval_runtime": 236.835,
+      "eval_samples_per_second": 142.576,
+      "eval_steps_per_second": 2.229,
+      "step": 22
+    },
+    {
+      "epoch": 0.002228034486099002,
+      "grad_norm": 3.6312763690948486,
+      "learning_rate": 9.999941877361234e-06,
+      "loss": 10.6172,
+      "step": 23
+    },
+    {
+      "epoch": 0.002228034486099002,
+      "eval_accuracy": 0.03920626085709957,
+      "eval_loss": 10.5,
+      "eval_runtime": 237.4625,
+      "eval_samples_per_second": 142.199,
+      "eval_steps_per_second": 2.224,
+      "step": 23
+    },
+    {
+      "epoch": 0.0023249055507120024,
+      "grad_norm": 3.621922492980957,
+      "learning_rate": 9.999932190254772e-06,
+      "loss": 10.5391,
+      "step": 24
+    },
+    {
+      "epoch": 0.0023249055507120024,
+      "eval_accuracy": 0.044748322853401586,
+      "eval_loss": 10.4375,
+      "eval_runtime": 237.2944,
+      "eval_samples_per_second": 142.3,
+      "eval_steps_per_second": 2.225,
+      "step": 24
+    },
+    {
+      "epoch": 0.0024217766153250024,
+      "grad_norm": 3.4077281951904297,
+      "learning_rate": 9.99992250314831e-06,
+      "loss": 10.5078,
+      "step": 25
+    },
+    {
+      "epoch": 0.0024217766153250024,
+      "eval_accuracy": 0.04779985410339344,
+      "eval_loss": 10.3828125,
+      "eval_runtime": 237.7331,
+      "eval_samples_per_second": 142.037,
+      "eval_steps_per_second": 2.221,
+      "step": 25
+    },
+    {
+      "epoch": 0.0025186476799380023,
+      "grad_norm": 3.273934841156006,
+      "learning_rate": 9.99991281604185e-06,
+      "loss": 10.4609,
+      "step": 26
+    },
+    {
+      "epoch": 0.0025186476799380023,
+      "eval_accuracy": 0.0499438956072986,
+      "eval_loss": 10.3125,
+      "eval_runtime": 236.731,
+      "eval_samples_per_second": 142.639,
+      "eval_steps_per_second": 2.23,
+      "step": 26
+    },
+    {
+      "epoch": 0.0026155187445510027,
+      "grad_norm": 3.3044192790985107,
+      "learning_rate": 9.999903128935388e-06,
+      "loss": 10.3906,
+      "step": 27
+    },
+    {
+      "epoch": 0.0026155187445510027,
+      "eval_accuracy": 0.05109582976502101,
+      "eval_loss": 10.265625,
+      "eval_runtime": 235.9963,
+      "eval_samples_per_second": 143.083,
+      "eval_steps_per_second": 2.237,
+      "step": 27
+    },
+    {
+      "epoch": 0.0027123898091640027,
+      "grad_norm": 3.2339956760406494,
+      "learning_rate": 9.999893441828926e-06,
+      "loss": 10.3281,
+      "step": 28
+    },
+    {
+      "epoch": 0.0027123898091640027,
+      "eval_accuracy": 0.05213538433890046,
+      "eval_loss": 10.2109375,
+      "eval_runtime": 235.9613,
+      "eval_samples_per_second": 143.104,
+      "eval_steps_per_second": 2.238,
+      "step": 28
+    },
+    {
+      "epoch": 0.0028092608737770026,
+      "grad_norm": 3.157088279724121,
+      "learning_rate": 9.999883754722464e-06,
+      "loss": 10.2656,
+      "step": 29
+    },
+    {
+      "epoch": 0.0028092608737770026,
+      "eval_accuracy": 0.053102914079034114,
+      "eval_loss": 10.1640625,
+      "eval_runtime": 236.7001,
+      "eval_samples_per_second": 142.657,
+      "eval_steps_per_second": 2.231,
+      "step": 29
+    },
+    {
+      "epoch": 0.002906131938390003,
+      "grad_norm": 2.9515769481658936,
+      "learning_rate": 9.999874067616004e-06,
+      "loss": 10.25,
+      "step": 30
+    },
+    {
+      "epoch": 0.002906131938390003,
+      "eval_accuracy": 0.05368669735769892,
+      "eval_loss": 10.1171875,
+      "eval_runtime": 237.6075,
+      "eval_samples_per_second": 142.113,
+      "eval_steps_per_second": 2.222,
+      "step": 30
+    },
+    {
+      "epoch": 0.003003003003003003,
+      "grad_norm": 2.8982396125793457,
+      "learning_rate": 9.999864380509543e-06,
+      "loss": 10.2031,
+      "step": 31
+    },
+    {
+      "epoch": 0.003003003003003003,
+      "eval_accuracy": 0.05444738729191865,
+      "eval_loss": 10.0703125,
+      "eval_runtime": 236.3017,
+      "eval_samples_per_second": 142.898,
+      "eval_steps_per_second": 2.234,
+      "step": 31
+    },
+    {
+      "epoch": 0.003099874067616003,
+      "grad_norm": 2.802349090576172,
+      "learning_rate": 9.999854693403081e-06,
+      "loss": 10.1641,
+      "step": 32
+    },
+    {
+      "epoch": 0.003099874067616003,
+      "eval_accuracy": 0.05520547182620385,
+      "eval_loss": 10.03125,
+      "eval_runtime": 236.5083,
+      "eval_samples_per_second": 142.773,
+      "eval_steps_per_second": 2.232,
+      "step": 32
+    },
+    {
+      "epoch": 0.0031967451322290033,
+      "grad_norm": 2.7052364349365234,
+      "learning_rate": 9.99984500629662e-06,
+      "loss": 10.125,
+      "step": 33
+    },
+    {
+      "epoch": 0.0031967451322290033,
+      "eval_accuracy": 0.05577098835643875,
+      "eval_loss": 9.9921875,
+      "eval_runtime": 236.8657,
+      "eval_samples_per_second": 142.558,
+      "eval_steps_per_second": 2.229,
+      "step": 33
+    },
+    {
+      "epoch": 0.0032936161968420033,
+      "grad_norm": 2.640915632247925,
+      "learning_rate": 9.99983531919016e-06,
+      "loss": 10.0859,
+      "step": 34
+    },
+    {
+      "epoch": 0.0032936161968420033,
+      "eval_accuracy": 0.0562093903187565,
+      "eval_loss": 9.9609375,
+      "eval_runtime": 236.6838,
+      "eval_samples_per_second": 142.667,
+      "eval_steps_per_second": 2.231,
+      "step": 34
+    },
+    {
+      "epoch": 0.0033904872614550032,
+      "grad_norm": 2.5900304317474365,
+      "learning_rate": 9.999825632083697e-06,
+      "loss": 10.0391,
+      "step": 35
+    },
+    {
+      "epoch": 0.0033904872614550032,
+      "eval_accuracy": 0.05660072138892365,
+      "eval_loss": 9.921875,
+      "eval_runtime": 236.5335,
+      "eval_samples_per_second": 142.758,
+      "eval_steps_per_second": 2.232,
+      "step": 35
+    },
+    {
+      "epoch": 0.0034873583260680036,
+      "grad_norm": 2.493401288986206,
+      "learning_rate": 9.999815944977235e-06,
+      "loss": 10.0156,
+      "step": 36
+    },
+    {
+      "epoch": 0.0034873583260680036,
+      "eval_accuracy": 0.0568301702764917,
+      "eval_loss": 9.890625,
+      "eval_runtime": 237.1506,
+      "eval_samples_per_second": 142.386,
+      "eval_steps_per_second": 2.226,
+      "step": 36
+    },
+    {
+      "epoch": 0.0035842293906810036,
+      "grad_norm": 2.489300012588501,
+      "learning_rate": 9.999806257870775e-06,
+      "loss": 9.9609,
+      "step": 37
+    },
+    {
+      "epoch": 0.0035842293906810036,
+      "eval_accuracy": 0.056685860069006624,
+      "eval_loss": 9.859375,
+      "eval_runtime": 236.5729,
+      "eval_samples_per_second": 142.734,
+      "eval_steps_per_second": 2.232,
+      "step": 37
+    },
+    {
+      "epoch": 0.0036811004552940035,
+      "grad_norm": 2.457721471786499,
+      "learning_rate": 9.999796570764313e-06,
+      "loss": 9.9141,
+      "step": 38
+    },
+    {
+      "epoch": 0.0036811004552940035,
+      "eval_accuracy": 0.05657735963617732,
+      "eval_loss": 9.8359375,
+      "eval_runtime": 236.757,
+      "eval_samples_per_second": 142.623,
+      "eval_steps_per_second": 2.23,
+      "step": 38
+    },
+    {
+      "epoch": 0.003777971519907004,
+      "grad_norm": 2.4300434589385986,
+      "learning_rate": 9.999786883657853e-06,
+      "loss": 9.875,
+      "step": 39
+    },
+    {
+      "epoch": 0.003777971519907004,
+      "eval_accuracy": 0.056777425402261446,
+      "eval_loss": 9.8046875,
+      "eval_runtime": 236.7279,
+      "eval_samples_per_second": 142.641,
+      "eval_steps_per_second": 2.23,
+      "step": 39
+    },
+    {
+      "epoch": 0.003874842584520004,
+      "grad_norm": 2.4308087825775146,
+      "learning_rate": 9.99977719655139e-06,
+      "loss": 9.8672,
+      "step": 40
+    },
+    {
+      "epoch": 0.003874842584520004,
+      "eval_accuracy": 0.05689365518822987,
+      "eval_loss": 9.78125,
+      "eval_runtime": 236.8414,
+      "eval_samples_per_second": 142.572,
+      "eval_steps_per_second": 2.229,
+      "step": 40
+    },
+    {
+      "epoch": 0.003971713649133004,
+      "grad_norm": 2.248959541320801,
+      "learning_rate": 9.99976750944493e-06,
+      "loss": 9.8438,
+      "step": 41
+    },
+    {
+      "epoch": 0.003971713649133004,
+      "eval_accuracy": 0.05678616796648622,
+      "eval_loss": 9.7578125,
+      "eval_runtime": 237.8773,
+      "eval_samples_per_second": 141.951,
+      "eval_steps_per_second": 2.22,
+      "step": 41
+    },
+    {
+      "epoch": 0.004068584713746004,
+      "grad_norm": 2.3188998699188232,
+      "learning_rate": 9.999757822338468e-06,
+      "loss": 9.7969,
+      "step": 42
+    },
+    {
+      "epoch": 0.004068584713746004,
+      "eval_accuracy": 0.05653222831953354,
+      "eval_loss": 9.734375,
+      "eval_runtime": 237.6567,
+      "eval_samples_per_second": 142.083,
+      "eval_steps_per_second": 2.222,
+      "step": 42
+    },
+    {
+      "epoch": 0.004165455778359004,
+      "grad_norm": 2.1342954635620117,
+      "learning_rate": 9.999748135232007e-06,
+      "loss": 9.8203,
+      "step": 43
+    },
+    {
+      "epoch": 0.004165455778359004,
+      "eval_accuracy": 0.056402595198346346,
+      "eval_loss": 9.7109375,
+      "eval_runtime": 238.2373,
+      "eval_samples_per_second": 141.737,
+      "eval_steps_per_second": 2.216,
+      "step": 43
+    },
+    {
+      "epoch": 0.004262326842972004,
+      "grad_norm": 2.0969340801239014,
+      "learning_rate": 9.999738448125546e-06,
+      "loss": 9.7891,
+      "step": 44
+    },
+    {
+      "epoch": 0.004262326842972004,
+      "eval_accuracy": 0.05635836129723557,
+      "eval_loss": 9.6875,
+      "eval_runtime": 237.2632,
+      "eval_samples_per_second": 142.319,
+      "eval_steps_per_second": 2.225,
+      "step": 44
+    },
+    {
+      "epoch": 0.004359197907585004,
+      "grad_norm": 2.200792074203491,
+      "learning_rate": 9.999728761019084e-06,
+      "loss": 9.7031,
+      "step": 45
+    },
+    {
+      "epoch": 0.004359197907585004,
+      "eval_accuracy": 0.05659519215128481,
+      "eval_loss": 9.671875,
+      "eval_runtime": 238.714,
+      "eval_samples_per_second": 141.454,
+      "eval_steps_per_second": 2.212,
+      "step": 45
+    },
+    {
+      "epoch": 0.004456068972198004,
+      "grad_norm": 2.0903360843658447,
+      "learning_rate": 9.999719073912622e-06,
+      "loss": 9.7344,
+      "step": 46
+    },
+    {
+      "epoch": 0.004456068972198004,
+      "eval_accuracy": 0.056926830614062944,
+      "eval_loss": 9.6484375,
+      "eval_runtime": 238.1183,
+      "eval_samples_per_second": 141.808,
+      "eval_steps_per_second": 2.217,
+      "step": 46
+    },
+    {
+      "epoch": 0.004552940036811005,
+      "grad_norm": 2.027561664581299,
+      "learning_rate": 9.999709386806162e-06,
+      "loss": 9.7266,
+      "step": 47
+    },
+    {
+      "epoch": 0.004552940036811005,
+      "eval_accuracy": 0.05733822326372602,
+      "eval_loss": 9.6328125,
+      "eval_runtime": 238.4469,
+      "eval_samples_per_second": 141.612,
+      "eval_steps_per_second": 2.214,
+      "step": 47
+    },
+    {
+      "epoch": 0.004649811101424005,
+      "grad_norm": 1.9840859174728394,
+      "learning_rate": 9.9996996996997e-06,
+      "loss": 9.7031,
+      "step": 48
+    },
+    {
+      "epoch": 0.004649811101424005,
+      "eval_accuracy": 0.057925625153411014,
+      "eval_loss": 9.6171875,
+      "eval_runtime": 239.2866,
+      "eval_samples_per_second": 141.115,
+      "eval_steps_per_second": 2.207,
+      "step": 48
+    },
+    {
+      "epoch": 0.004746682166037005,
+      "grad_norm": 1.8907567262649536,
+      "learning_rate": 9.99969001259324e-06,
+      "loss": 9.7109,
+      "step": 49
+    },
+    {
+      "epoch": 0.004746682166037005,
+      "eval_accuracy": 0.05849756833681777,
+      "eval_loss": 9.6015625,
+      "eval_runtime": 237.5376,
+      "eval_samples_per_second": 142.154,
+      "eval_steps_per_second": 2.223,
+      "step": 49
+    },
+    {
+      "epoch": 0.004843553230650005,
+      "grad_norm": 1.9789716005325317,
+      "learning_rate": 9.999680325486778e-06,
+      "loss": 9.6406,
+      "step": 50
+    },
+    {
+      "epoch": 0.004843553230650005,
+      "eval_accuracy": 0.059110966328071786,
+      "eval_loss": 9.578125,
+      "eval_runtime": 238.4868,
+      "eval_samples_per_second": 141.589,
+      "eval_steps_per_second": 2.214,
+      "step": 50
+    },
+    {
+      "epoch": 0.004940424295263005,
+      "grad_norm": 1.8518751859664917,
+      "learning_rate": 9.999670638380316e-06,
+      "loss": 9.6797,
+      "step": 51
+    },
+    {
+      "epoch": 0.004940424295263005,
+      "eval_accuracy": 0.059745178569913925,
+      "eval_loss": 9.5625,
+      "eval_runtime": 237.9086,
+      "eval_samples_per_second": 141.933,
+      "eval_steps_per_second": 2.219,
+      "step": 51
+    },
+    {
+      "epoch": 0.005037295359876005,
+      "grad_norm": 1.9054638147354126,
+      "learning_rate": 9.999660951273856e-06,
+      "loss": 9.6328,
+      "step": 52
+    },
+    {
+      "epoch": 0.005037295359876005,
+      "eval_accuracy": 0.06045813178755534,
+      "eval_loss": 9.546875,
+      "eval_runtime": 238.5328,
+      "eval_samples_per_second": 141.561,
+      "eval_steps_per_second": 2.214,
+      "step": 52
+    },
+    {
+      "epoch": 0.0051341664244890055,
+      "grad_norm": 1.9086819887161255,
+      "learning_rate": 9.999651264167394e-06,
+      "loss": 9.6172,
+      "step": 53
+    },
+    {
+      "epoch": 0.0051341664244890055,
+      "eval_accuracy": 0.06115973704103745,
+      "eval_loss": 9.53125,
+      "eval_runtime": 239.0323,
+      "eval_samples_per_second": 141.265,
+      "eval_steps_per_second": 2.209,
+      "step": 53
+    },
+    {
+      "epoch": 0.0052310374891020054,
+      "grad_norm": 1.8585758209228516,
+      "learning_rate": 9.999641577060932e-06,
+      "loss": 9.6172,
+      "step": 54
+    },
+    {
+      "epoch": 0.0052310374891020054,
+      "eval_accuracy": 0.06154695736908567,
+      "eval_loss": 9.5234375,
+      "eval_runtime": 237.7379,
+      "eval_samples_per_second": 142.035,
+      "eval_steps_per_second": 2.221,
+      "step": 54
+    },
+    {
+      "epoch": 0.005327908553715005,
+      "grad_norm": 1.8926104307174683,
+      "learning_rate": 9.999631889954471e-06,
+      "loss": 9.5703,
+      "step": 55
+    },
+    {
+      "epoch": 0.005327908553715005,
+      "eval_accuracy": 0.06174404139968916,
+      "eval_loss": 9.5078125,
+      "eval_runtime": 238.0821,
+      "eval_samples_per_second": 141.829,
+      "eval_steps_per_second": 2.218,
+      "step": 55
+    },
+    {
+      "epoch": 0.005424779618328005,
+      "grad_norm": 1.8241984844207764,
+      "learning_rate": 9.999622202848011e-06,
+      "loss": 9.5781,
+      "step": 56
+    },
+    {
+      "epoch": 0.005424779618328005,
+      "eval_accuracy": 0.061843017648313334,
+      "eval_loss": 9.4921875,
+      "eval_runtime": 237.2046,
+      "eval_samples_per_second": 142.354,
+      "eval_steps_per_second": 2.226,
+      "step": 56
+    },
+    {
+      "epoch": 0.005521650682941005,
+      "grad_norm": 1.7987343072891235,
+      "learning_rate": 9.999612515741549e-06,
+      "loss": 9.5938,
+      "step": 57
+    },
+    {
+      "epoch": 0.005521650682941005,
+      "eval_accuracy": 0.0619944782311743,
+      "eval_loss": 9.4765625,
+      "eval_runtime": 238.3243,
+      "eval_samples_per_second": 141.685,
+      "eval_steps_per_second": 2.215,
+      "step": 57
+    },
+    {
+      "epoch": 0.005618521747554005,
+      "grad_norm": 1.8513528108596802,
+      "learning_rate": 9.999602828635087e-06,
+      "loss": 9.5391,
+      "step": 58
+    },
+    {
+      "epoch": 0.005618521747554005,
+      "eval_accuracy": 0.06205043643199048,
+      "eval_loss": 9.46875,
+      "eval_runtime": 236.9136,
+      "eval_samples_per_second": 142.529,
+      "eval_steps_per_second": 2.229,
+      "step": 58
+    },
+    {
+      "epoch": 0.005715392812167006,
+      "grad_norm": 1.901055932044983,
+      "learning_rate": 9.999593141528627e-06,
+      "loss": 9.4922,
+      "step": 59
+    },
+    {
+      "epoch": 0.005715392812167006,
+      "eval_accuracy": 0.06204551512100302,
+      "eval_loss": 9.453125,
+      "eval_runtime": 240.0178,
+      "eval_samples_per_second": 140.685,
+      "eval_steps_per_second": 2.2,
+      "step": 59
+    },
+    {
+      "epoch": 0.005812263876780006,
+      "grad_norm": 1.8602513074874878,
+      "learning_rate": 9.999583454422165e-06,
+      "loss": 9.4688,
+      "step": 60
+    },
+    {
+      "epoch": 0.005812263876780006,
+      "eval_accuracy": 0.062001570708773865,
+      "eval_loss": 9.4375,
+      "eval_runtime": 239.415,
+      "eval_samples_per_second": 141.04,
+      "eval_steps_per_second": 2.205,
+      "step": 60
+    },
+    {
+      "epoch": 0.005909134941393006,
+      "grad_norm": 1.8280858993530273,
+      "learning_rate": 9.999573767315703e-06,
+      "loss": 9.4922,
+      "step": 61
+    },
+    {
+      "epoch": 0.005909134941393006,
+      "eval_accuracy": 0.06202125595272369,
+      "eval_loss": 9.4296875,
+      "eval_runtime": 240.2145,
+      "eval_samples_per_second": 140.57,
+      "eval_steps_per_second": 2.198,
+      "step": 61
+    },
+    {
+      "epoch": 0.006006006006006006,
+      "grad_norm": 1.8284739255905151,
+      "learning_rate": 9.999564080209243e-06,
+      "loss": 9.4609,
+      "step": 62
+    },
+    {
+      "epoch": 0.006006006006006006,
+      "eval_accuracy": 0.06204618094543074,
+      "eval_loss": 9.4140625,
+      "eval_runtime": 240.548,
+      "eval_samples_per_second": 140.375,
+      "eval_steps_per_second": 2.195,
+      "step": 62
+    },
+    {
+      "epoch": 0.006102877070619006,
+      "grad_norm": 1.8668478727340698,
+      "learning_rate": 9.99955439310278e-06,
+      "loss": 9.4297,
+      "step": 63
+    },
+    {
+      "epoch": 0.006102877070619006,
+      "eval_accuracy": 0.061995375646707304,
+      "eval_loss": 9.40625,
+      "eval_runtime": 238.875,
+      "eval_samples_per_second": 141.358,
+      "eval_steps_per_second": 2.21,
+      "step": 63
+    },
+    {
+      "epoch": 0.006199748135232006,
+      "grad_norm": 1.7616581916809082,
+      "learning_rate": 9.99954470599632e-06,
+      "loss": 9.4844,
+      "step": 64
+    },
+    {
+      "epoch": 0.006199748135232006,
+      "eval_accuracy": 0.062048844243141596,
+      "eval_loss": 9.390625,
+      "eval_runtime": 238.455,
+      "eval_samples_per_second": 141.607,
+      "eval_steps_per_second": 2.214,
+      "step": 64
+    },
+    {
+      "epoch": 0.006296619199845007,
+      "grad_norm": 1.7911943197250366,
+      "learning_rate": 9.999535018889858e-06,
+      "loss": 9.4531,
+      "step": 65
+    },
+    {
+      "epoch": 0.006296619199845007,
+      "eval_accuracy": 0.06223356709850013,
+      "eval_loss": 9.3828125,
+      "eval_runtime": 236.8696,
+      "eval_samples_per_second": 142.555,
+      "eval_steps_per_second": 2.229,
+      "step": 65
+    },
+    {
+      "epoch": 0.006393490264458007,
+      "grad_norm": 1.7897166013717651,
+      "learning_rate": 9.999525331783398e-06,
+      "loss": 9.4375,
+      "step": 66
+    },
+    {
+      "epoch": 0.006393490264458007,
+      "eval_accuracy": 0.06252389549787181,
+      "eval_loss": 9.3671875,
+      "eval_runtime": 238.3072,
+      "eval_samples_per_second": 141.695,
+      "eval_steps_per_second": 2.216,
+      "step": 66
+    },
+    {
+      "epoch": 0.006490361329071007,
+      "grad_norm": 1.7542698383331299,
+      "learning_rate": 9.999515644676936e-06,
+      "loss": 9.4375,
+      "step": 67
+    },
+    {
+      "epoch": 0.006490361329071007,
+      "eval_accuracy": 0.06279815726431386,
+      "eval_loss": 9.359375,
+      "eval_runtime": 238.8868,
+      "eval_samples_per_second": 141.351,
+      "eval_steps_per_second": 2.21,
+      "step": 67
+    },
+    {
+      "epoch": 0.0065872323936840066,
+      "grad_norm": 1.7762055397033691,
+      "learning_rate": 9.999505957570474e-06,
+      "loss": 9.3984,
+      "step": 68
+    },
+    {
+      "epoch": 0.0065872323936840066,
+      "eval_accuracy": 0.06302155583425616,
+      "eval_loss": 9.34375,
+      "eval_runtime": 238.2147,
+      "eval_samples_per_second": 141.75,
+      "eval_steps_per_second": 2.216,
+      "step": 68
+    },
+    {
+      "epoch": 0.0066841034582970065,
+      "grad_norm": 1.7549471855163574,
+      "learning_rate": 9.999496270464012e-06,
+      "loss": 9.4062,
+      "step": 69
+    },
+    {
+      "epoch": 0.0066841034582970065,
+      "eval_accuracy": 0.0631751296859529,
+      "eval_loss": 9.3359375,
+      "eval_runtime": 238.9606,
+      "eval_samples_per_second": 141.308,
+      "eval_steps_per_second": 2.21,
+      "step": 69
+    },
+    {
+      "epoch": 0.0067809745229100065,
+      "grad_norm": 1.7370997667312622,
+      "learning_rate": 9.999486583357552e-06,
+      "loss": 9.3984,
+      "step": 70
+    },
+    {
+      "epoch": 0.0067809745229100065,
+      "eval_accuracy": 0.06327500335011008,
+      "eval_loss": 9.3203125,
+      "eval_runtime": 238.5419,
+      "eval_samples_per_second": 141.556,
+      "eval_steps_per_second": 2.213,
+      "step": 70
+    },
+    {
+      "epoch": 0.006877845587523007,
+      "grad_norm": 1.6883249282836914,
+      "learning_rate": 9.99947689625109e-06,
+      "loss": 9.4375,
+      "step": 71
+    },
+    {
+      "epoch": 0.006877845587523007,
+      "eval_accuracy": 0.06334149894621705,
+      "eval_loss": 9.3125,
+      "eval_runtime": 238.5418,
+      "eval_samples_per_second": 141.556,
+      "eval_steps_per_second": 2.213,
+      "step": 71
+    },
+    {
+      "epoch": 0.006974716652136007,
+      "grad_norm": 1.7424453496932983,
+      "learning_rate": 9.99946720914463e-06,
+      "loss": 9.3828,
+      "step": 72
+    },
+    {
+      "epoch": 0.006974716652136007,
+      "eval_accuracy": 0.0633868039561898,
+      "eval_loss": 9.3046875,
+      "eval_runtime": 237.7375,
+      "eval_samples_per_second": 142.035,
+      "eval_steps_per_second": 2.221,
+      "step": 72
+    },
+    {
+      "epoch": 0.007071587716749007,
+      "grad_norm": 1.744436264038086,
+      "learning_rate": 9.999457522038168e-06,
+      "loss": 9.3594,
+      "step": 73
+    },
+    {
+      "epoch": 0.007071587716749007,
+      "eval_accuracy": 0.06341806875540422,
+      "eval_loss": 9.2890625,
+      "eval_runtime": 238.0776,
+      "eval_samples_per_second": 141.832,
+      "eval_steps_per_second": 2.218,
+      "step": 73
+    },
+    {
+      "epoch": 0.007168458781362007,
+      "grad_norm": 1.7263447046279907,
+      "learning_rate": 9.999447834931707e-06,
+      "loss": 9.3438,
+      "step": 74
+    },
+    {
+      "epoch": 0.007168458781362007,
+      "eval_accuracy": 0.06344229897479539,
+      "eval_loss": 9.28125,
+      "eval_runtime": 239.9491,
+      "eval_samples_per_second": 140.726,
+      "eval_steps_per_second": 2.2,
+      "step": 74
+    },
+    {
+      "epoch": 0.007265329845975007,
+      "grad_norm": 1.6880813837051392,
+      "learning_rate": 9.999438147825245e-06,
+      "loss": 9.3672,
+      "step": 75
+    },
+    {
+      "epoch": 0.007265329845975007,
+      "eval_accuracy": 0.06342368483970755,
+      "eval_loss": 9.2734375,
+      "eval_runtime": 237.9155,
+      "eval_samples_per_second": 141.929,
+      "eval_steps_per_second": 2.219,
+      "step": 75
+    },
+    {
+      "epoch": 0.007362200910588007,
+      "grad_norm": 1.7534254789352417,
+      "learning_rate": 9.999428460718784e-06,
+      "loss": 9.3125,
+      "step": 76
+    },
+    {
+      "epoch": 0.007362200910588007,
+      "eval_accuracy": 0.06337450067872116,
+      "eval_loss": 9.2578125,
+      "eval_runtime": 239.6783,
+      "eval_samples_per_second": 140.885,
+      "eval_steps_per_second": 2.203,
+      "step": 76
+    },
+    {
+      "epoch": 0.007459071975201007,
+      "grad_norm": 1.7223541736602783,
+      "learning_rate": 9.999418773612323e-06,
+      "loss": 9.3047,
+      "step": 77
+    },
+    {
+      "epoch": 0.007459071975201007,
+      "eval_accuracy": 0.06325945779716735,
+      "eval_loss": 9.25,
+      "eval_runtime": 238.6926,
+      "eval_samples_per_second": 141.466,
+      "eval_steps_per_second": 2.212,
+      "step": 77
+    },
+    {
+      "epoch": 0.007555943039814008,
+      "grad_norm": 1.7189525365829468,
+      "learning_rate": 9.999409086505861e-06,
+      "loss": 9.2969,
+      "step": 78
+    },
+    {
+      "epoch": 0.007555943039814008,
+      "eval_accuracy": 0.0631598736218918,
+      "eval_loss": 9.2421875,
+      "eval_runtime": 237.6208,
+      "eval_samples_per_second": 142.105,
+      "eval_steps_per_second": 2.222,
+      "step": 78
+    },
+    {
+      "epoch": 0.007652814104427008,
+      "grad_norm": 1.7101792097091675,
+      "learning_rate": 9.9993993993994e-06,
+      "loss": 9.2891,
+      "step": 79
+    },
+    {
+      "epoch": 0.007652814104427008,
+      "eval_accuracy": 0.06305487600453004,
+      "eval_loss": 9.2265625,
+      "eval_runtime": 239.1168,
+      "eval_samples_per_second": 141.216,
+      "eval_steps_per_second": 2.208,
+      "step": 79
+    },
+    {
+      "epoch": 0.007749685169040008,
+      "grad_norm": 1.7495548725128174,
+      "learning_rate": 9.999389712292939e-06,
+      "loss": 9.2812,
+      "step": 80
+    },
+    {
+      "epoch": 0.007749685169040008,
+      "eval_accuracy": 0.06309847303010126,
+      "eval_loss": 9.21875,
+      "eval_runtime": 238.7884,
+      "eval_samples_per_second": 141.41,
+      "eval_steps_per_second": 2.211,
+      "step": 80
+    },
+    {
+      "epoch": 0.007846556233653008,
+      "grad_norm": 1.6896053552627563,
+      "learning_rate": 9.999380025186479e-06,
+      "loss": 9.2656,
+      "step": 81
+    },
+    {
+      "epoch": 0.007846556233653008,
+      "eval_accuracy": 0.0631880698389611,
+      "eval_loss": 9.2109375,
+      "eval_runtime": 240.0662,
+      "eval_samples_per_second": 140.657,
+      "eval_steps_per_second": 2.199,
+      "step": 81
+    },
+    {
+      "epoch": 0.007943427298266009,
+      "grad_norm": 1.7166608572006226,
+      "learning_rate": 9.999370338080017e-06,
+      "loss": 9.2422,
+      "step": 82
+    },
+    {
+      "epoch": 0.007943427298266009,
+      "eval_accuracy": 0.06330250479386351,
+      "eval_loss": 9.203125,
+      "eval_runtime": 238.0584,
+      "eval_samples_per_second": 141.843,
+      "eval_steps_per_second": 2.218,
+      "step": 82
+    },
+    {
+      "epoch": 0.008040298362879008,
+      "grad_norm": 1.6938972473144531,
+      "learning_rate": 9.999360650973555e-06,
+      "loss": 9.2656,
+      "step": 83
+    },
+    {
+      "epoch": 0.008040298362879008,
+      "eval_accuracy": 0.06352989831037209,
+      "eval_loss": 9.1875,
+      "eval_runtime": 240.419,
+      "eval_samples_per_second": 140.451,
+      "eval_steps_per_second": 2.196,
+      "step": 83
+    },
+    {
+      "epoch": 0.008137169427492008,
+      "grad_norm": 1.6488450765609741,
+      "learning_rate": 9.999350963867095e-06,
+      "loss": 9.25,
+      "step": 84
+    },
+    {
+      "epoch": 0.008137169427492008,
+      "eval_accuracy": 0.06372730077874536,
+      "eval_loss": 9.1796875,
+      "eval_runtime": 239.4979,
+      "eval_samples_per_second": 140.991,
+      "eval_steps_per_second": 2.205,
+      "step": 84
+    },
+    {
+      "epoch": 0.008234040492105008,
+      "grad_norm": 1.695817470550537,
+      "learning_rate": 9.999341276760633e-06,
+      "loss": 9.2344,
+      "step": 85
+    },
+    {
+      "epoch": 0.008234040492105008,
+      "eval_accuracy": 0.06391222627632101,
+      "eval_loss": 9.171875,
+      "eval_runtime": 237.1194,
+      "eval_samples_per_second": 142.405,
+      "eval_steps_per_second": 2.227,
+      "step": 85
+    },
+    {
+      "epoch": 0.008330911556718008,
+      "grad_norm": 1.6468489170074463,
+      "learning_rate": 9.99933158965417e-06,
+      "loss": 9.2266,
+      "step": 86
+    },
+    {
+      "epoch": 0.008330911556718008,
+      "eval_accuracy": 0.0640310325133358,
+      "eval_loss": 9.15625,
+      "eval_runtime": 238.4317,
+      "eval_samples_per_second": 141.621,
+      "eval_steps_per_second": 2.214,
+      "step": 86
+    },
+    {
+      "epoch": 0.00842778262133101,
+      "grad_norm": 1.5998897552490234,
+      "learning_rate": 9.99932190254771e-06,
+      "loss": 9.25,
+      "step": 87
+    },
+    {
+      "epoch": 0.00842778262133101,
+      "eval_accuracy": 0.06412291628436041,
+      "eval_loss": 9.1484375,
+      "eval_runtime": 239.3988,
+      "eval_samples_per_second": 141.049,
+      "eval_steps_per_second": 2.206,
+      "step": 87
+    },
+    {
+      "epoch": 0.008524653685944008,
+      "grad_norm": 1.74087655544281,
+      "learning_rate": 9.999312215441248e-06,
+      "loss": 9.1406,
+      "step": 88
+    },
+    {
+      "epoch": 0.008524653685944008,
+      "eval_accuracy": 0.06414216729498781,
+      "eval_loss": 9.140625,
+      "eval_runtime": 239.1559,
+      "eval_samples_per_second": 141.192,
+      "eval_steps_per_second": 2.208,
+      "step": 88
+    },
+    {
+      "epoch": 0.008621524750557009,
+      "grad_norm": 1.723071813583374,
+      "learning_rate": 9.999302528334788e-06,
+      "loss": 9.1562,
+      "step": 89
+    },
+    {
+      "epoch": 0.008621524750557009,
+      "eval_accuracy": 0.06415910239456228,
+      "eval_loss": 9.1328125,
+      "eval_runtime": 239.2993,
+      "eval_samples_per_second": 141.108,
+      "eval_steps_per_second": 2.206,
+      "step": 89
+    },
+    {
+      "epoch": 0.008718395815170008,
+      "grad_norm": 1.605566143989563,
+      "learning_rate": 9.999292841228326e-06,
+      "loss": 9.2031,
+      "step": 90
+    },
+    {
+      "epoch": 0.008718395815170008,
+      "eval_accuracy": 0.0641378828595399,
+      "eval_loss": 9.1171875,
+      "eval_runtime": 239.3127,
+      "eval_samples_per_second": 141.1,
+      "eval_steps_per_second": 2.206,
+      "step": 90
+    },
+    {
+      "epoch": 0.008815266879783009,
+      "grad_norm": 1.699105143547058,
+      "learning_rate": 9.999283154121864e-06,
+      "loss": 9.1406,
+      "step": 91
+    },
+    {
+      "epoch": 0.008815266879783009,
+      "eval_accuracy": 0.06416199728337843,
+      "eval_loss": 9.109375,
+      "eval_runtime": 239.2239,
+      "eval_samples_per_second": 141.152,
+      "eval_steps_per_second": 2.207,
+      "step": 91
+    },
+    {
+      "epoch": 0.008912137944396008,
+      "grad_norm": 1.6730599403381348,
+      "learning_rate": 9.999273467015404e-06,
+      "loss": 9.1406,
+      "step": 92
+    },
+    {
+      "epoch": 0.008912137944396008,
+      "eval_accuracy": 0.06426725544073365,
+      "eval_loss": 9.1015625,
+      "eval_runtime": 239.2839,
+      "eval_samples_per_second": 141.117,
+      "eval_steps_per_second": 2.207,
+      "step": 92
+    },
+    {
+      "epoch": 0.009009009009009009,
+      "grad_norm": 1.6537282466888428,
+      "learning_rate": 9.999263779908942e-06,
+      "loss": 9.1406,
+      "step": 93
+    },
+    {
+      "epoch": 0.009009009009009009,
+      "eval_accuracy": 0.06438559849553786,
+      "eval_loss": 9.09375,
+      "eval_runtime": 237.9124,
+      "eval_samples_per_second": 141.93,
+      "eval_steps_per_second": 2.219,
+      "step": 93
+    },
+    {
+      "epoch": 0.00910588007362201,
+      "grad_norm": 1.648465871810913,
+      "learning_rate": 9.99925409280248e-06,
+      "loss": 9.1328,
+      "step": 94
+    },
+    {
+      "epoch": 0.00910588007362201,
+      "eval_accuracy": 0.06439923342186193,
+      "eval_loss": 9.078125,
+      "eval_runtime": 239.3262,
+      "eval_samples_per_second": 141.092,
+      "eval_steps_per_second": 2.206,
+      "step": 94
+    },
+    {
+      "epoch": 0.009202751138235009,
+      "grad_norm": 1.6511099338531494,
+      "learning_rate": 9.99924440569602e-06,
+      "loss": 9.125,
+      "step": 95
+    },
+    {
+      "epoch": 0.009202751138235009,
+      "eval_accuracy": 0.06449600955498583,
+      "eval_loss": 9.0703125,
+      "eval_runtime": 238.7857,
+      "eval_samples_per_second": 141.411,
+      "eval_steps_per_second": 2.211,
+      "step": 95
+    },
+    {
+      "epoch": 0.00929962220284801,
+      "grad_norm": 1.6301484107971191,
+      "learning_rate": 9.999234718589558e-06,
+      "loss": 9.1016,
+      "step": 96
+    },
+    {
+      "epoch": 0.00929962220284801,
+      "eval_accuracy": 0.0646362669181283,
+      "eval_loss": 9.0625,
+      "eval_runtime": 238.0729,
+      "eval_samples_per_second": 141.835,
+      "eval_steps_per_second": 2.218,
+      "step": 96
+    },
+    {
+      "epoch": 0.009396493267461009,
+      "grad_norm": 1.5811512470245361,
+      "learning_rate": 9.999225031483096e-06,
+      "loss": 9.125,
+      "step": 97
+    },
+    {
+      "epoch": 0.009396493267461009,
+      "eval_accuracy": 0.06484397519068705,
+      "eval_loss": 9.0546875,
+      "eval_runtime": 238.6104,
+      "eval_samples_per_second": 141.515,
+      "eval_steps_per_second": 2.213,
+      "step": 97
+    },
+    {
+      "epoch": 0.00949336433207401,
+      "grad_norm": 1.6737486124038696,
+      "learning_rate": 9.999215344376635e-06,
+      "loss": 9.0625,
+      "step": 98
+    },
+    {
+      "epoch": 0.00949336433207401,
+      "eval_accuracy": 0.06518476150212422,
+      "eval_loss": 9.0390625,
+      "eval_runtime": 238.844,
+      "eval_samples_per_second": 141.377,
+      "eval_steps_per_second": 2.211,
+      "step": 98
+    },
+    {
+      "epoch": 0.00959023539668701,
+      "grad_norm": 1.6142433881759644,
+      "learning_rate": 9.999205657270175e-06,
+      "loss": 9.0859,
+      "step": 99
+    },
+    {
+      "epoch": 0.00959023539668701,
+      "eval_accuracy": 0.06548032965025313,
+      "eval_loss": 9.03125,
+      "eval_runtime": 238.064,
+      "eval_samples_per_second": 141.84,
+      "eval_steps_per_second": 2.218,
+      "step": 99
+    },
+    {
+      "epoch": 0.00968710646130001,
+      "grad_norm": 1.640156865119934,
+      "learning_rate": 9.999195970163713e-06,
+      "loss": 9.0547,
+      "step": 100
+    },
+    {
+      "epoch": 0.00968710646130001,
+      "eval_accuracy": 0.06569275659158222,
+      "eval_loss": 9.0234375,
+      "eval_runtime": 238.6131,
+      "eval_samples_per_second": 141.514,
+      "eval_steps_per_second": 2.213,
+      "step": 100
+    },
+    {
+      "epoch": 0.00978397752591301,
+      "grad_norm": 1.6266826391220093,
+      "learning_rate": 9.999186283057251e-06,
+      "loss": 9.0547,
+      "step": 101
+    },
+    {
+      "epoch": 0.00978397752591301,
+      "eval_accuracy": 0.06578165862712619,
+      "eval_loss": 9.015625,
+      "eval_runtime": 239.9035,
+      "eval_samples_per_second": 140.752,
+      "eval_steps_per_second": 2.201,
+      "step": 101
+    },
+    {
+      "epoch": 0.00988084859052601,
+      "grad_norm": 1.5783159732818604,
+      "learning_rate": 9.999176595950791e-06,
+      "loss": 9.0625,
+      "step": 102
+    },
+    {
+      "epoch": 0.00988084859052601,
+      "eval_accuracy": 0.0659039097818322,
+      "eval_loss": 9.0078125,
+      "eval_runtime": 239.5936,
+      "eval_samples_per_second": 140.934,
+      "eval_steps_per_second": 2.204,
+      "step": 102
+    },
+    {
+      "epoch": 0.00997771965513901,
+      "grad_norm": 1.6107426881790161,
+      "learning_rate": 9.999166908844329e-06,
+      "loss": 9.0547,
+      "step": 103
+    },
+    {
+      "epoch": 0.00997771965513901,
+      "eval_accuracy": 0.0660996332146921,
+      "eval_loss": 8.9921875,
+      "eval_runtime": 240.4185,
+      "eval_samples_per_second": 140.451,
+      "eval_steps_per_second": 2.196,
+      "step": 103
+    },
+    {
+      "epoch": 0.01007459071975201,
+      "grad_norm": 1.6132723093032837,
+      "learning_rate": 9.999157221737867e-06,
+      "loss": 9.0156,
+      "step": 104
+    },
+    {
+      "epoch": 0.01007459071975201,
+      "eval_accuracy": 0.06618155856818915,
+      "eval_loss": 8.984375,
+      "eval_runtime": 238.4229,
+      "eval_samples_per_second": 141.627,
+      "eval_steps_per_second": 2.215,
+      "step": 104
+    },
+    {
+      "epoch": 0.01017146178436501,
+      "grad_norm": 1.594551682472229,
+      "learning_rate": 9.999147534631407e-06,
+      "loss": 9.0391,
+      "step": 105
+    },
+    {
+      "epoch": 0.01017146178436501,
+      "eval_accuracy": 0.06635901525261914,
+      "eval_loss": 8.9765625,
+      "eval_runtime": 239.2842,
+      "eval_samples_per_second": 141.117,
+      "eval_steps_per_second": 2.207,
+      "step": 105
+    },
+    {
+      "epoch": 0.010268332848978011,
+      "grad_norm": 1.6009005308151245,
+      "learning_rate": 9.999137847524946e-06,
+      "loss": 9.0234,
+      "step": 106
+    },
+    {
+      "epoch": 0.010268332848978011,
+      "eval_accuracy": 0.06637285282116034,
+      "eval_loss": 8.96875,
+      "eval_runtime": 239.6129,
+      "eval_samples_per_second": 140.923,
+      "eval_steps_per_second": 2.204,
+      "step": 106
+    },
+    {
+      "epoch": 0.01036520391359101,
+      "grad_norm": 1.5228526592254639,
+      "learning_rate": 9.999128160418484e-06,
+      "loss": 9.0234,
+      "step": 107
+    },
+    {
+      "epoch": 0.01036520391359101,
+      "eval_accuracy": 0.06640226489153242,
+      "eval_loss": 8.9609375,
+      "eval_runtime": 240.652,
+      "eval_samples_per_second": 140.315,
+      "eval_steps_per_second": 2.194,
+      "step": 107
+    },
+    {
+      "epoch": 0.010462074978204011,
+      "grad_norm": 1.6139883995056152,
+      "learning_rate": 9.999118473312022e-06,
+      "loss": 8.9766,
+      "step": 108
+    },
+    {
+      "epoch": 0.010462074978204011,
+      "eval_accuracy": 0.06641254174682976,
+      "eval_loss": 8.9453125,
+      "eval_runtime": 241.0995,
+      "eval_samples_per_second": 140.054,
+      "eval_steps_per_second": 2.19,
+      "step": 108
+    },
+    {
+      "epoch": 0.01055894604281701,
+      "grad_norm": 1.5845446586608887,
+      "learning_rate": 9.999108786205562e-06,
+      "loss": 8.9922,
+      "step": 109
+    },
+    {
+      "epoch": 0.01055894604281701,
+      "eval_accuracy": 0.06653754304591111,
+      "eval_loss": 8.9375,
+      "eval_runtime": 241.1972,
+      "eval_samples_per_second": 139.997,
+      "eval_steps_per_second": 2.189,
+      "step": 109
+    },
+    {
+      "epoch": 0.01065581710743001,
+      "grad_norm": 1.6032410860061646,
+      "learning_rate": 9.9990990990991e-06,
+      "loss": 8.9453,
+      "step": 110
+    },
+    {
+      "epoch": 0.01065581710743001,
+      "eval_accuracy": 0.06648474027390454,
+      "eval_loss": 8.9296875,
+      "eval_runtime": 240.7345,
+      "eval_samples_per_second": 140.267,
+      "eval_steps_per_second": 2.193,
+      "step": 110
+    },
+    {
+      "epoch": 0.010752688172043012,
+      "grad_norm": 1.5953596830368042,
+      "learning_rate": 9.999089411992638e-06,
+      "loss": 8.9609,
+      "step": 111
+    },
+    {
+      "epoch": 0.010752688172043012,
+      "eval_accuracy": 0.06636981318790339,
+      "eval_loss": 8.921875,
+      "eval_runtime": 241.4759,
+      "eval_samples_per_second": 139.836,
+      "eval_steps_per_second": 2.187,
+      "step": 111
+    },
+    {
+      "epoch": 0.01084955923665601,
+      "grad_norm": 1.5376003980636597,
+      "learning_rate": 9.999079724886176e-06,
+      "loss": 8.9766,
+      "step": 112
+    },
+    {
+      "epoch": 0.01084955923665601,
+      "eval_accuracy": 0.06635956528149421,
+      "eval_loss": 8.9140625,
+      "eval_runtime": 238.6698,
+      "eval_samples_per_second": 141.48,
+      "eval_steps_per_second": 2.212,
+      "step": 112
+    },
+    {
+      "epoch": 0.010946430301269012,
+      "grad_norm": 1.5216182470321655,
+      "learning_rate": 9.999070037779716e-06,
+      "loss": 8.9844,
+      "step": 113
+    },
+    {
+      "epoch": 0.010946430301269012,
+      "eval_accuracy": 0.06662239223711247,
+      "eval_loss": 8.8984375,
+      "eval_runtime": 240.1305,
+      "eval_samples_per_second": 140.619,
+      "eval_steps_per_second": 2.199,
+      "step": 113
+    },
+    {
+      "epoch": 0.01104330136588201,
+      "grad_norm": 1.575067400932312,
+      "learning_rate": 9.999060350673254e-06,
+      "loss": 8.9453,
+      "step": 114
+    },
+    {
+      "epoch": 0.01104330136588201,
+      "eval_accuracy": 0.06693289801153271,
+      "eval_loss": 8.890625,
+      "eval_runtime": 239.4069,
+      "eval_samples_per_second": 141.044,
+      "eval_steps_per_second": 2.205,
+      "step": 114
+    },
+    {
+      "epoch": 0.011140172430495011,
+      "grad_norm": 1.509313941001892,
+      "learning_rate": 9.999050663566794e-06,
+      "loss": 8.9688,
+      "step": 115
+    },
+    {
+      "epoch": 0.011140172430495011,
+      "eval_accuracy": 0.06726662079425849,
+      "eval_loss": 8.8828125,
+      "eval_runtime": 242.3159,
+      "eval_samples_per_second": 139.351,
+      "eval_steps_per_second": 2.179,
+      "step": 115
+    },
+    {
+      "epoch": 0.01123704349510801,
+      "grad_norm": 1.48069167137146,
+      "learning_rate": 9.999040976460332e-06,
+      "loss": 8.9766,
+      "step": 116
+    },
+    {
+      "epoch": 0.01123704349510801,
+      "eval_accuracy": 0.06765149626236562,
+      "eval_loss": 8.875,
+      "eval_runtime": 241.5301,
+      "eval_samples_per_second": 139.805,
+      "eval_steps_per_second": 2.186,
+      "step": 116
+    },
+    {
+      "epoch": 0.011333914559721011,
+      "grad_norm": 1.5143989324569702,
+      "learning_rate": 9.999031289353872e-06,
+      "loss": 8.9297,
+      "step": 117
+    },
+    {
+      "epoch": 0.011333914559721011,
+      "eval_accuracy": 0.0682274054434505,
+      "eval_loss": 8.8671875,
+      "eval_runtime": 241.8791,
+      "eval_samples_per_second": 139.603,
+      "eval_steps_per_second": 2.183,
+      "step": 117
+    },
+    {
+      "epoch": 0.011430785624334012,
+      "grad_norm": 1.5148530006408691,
+      "learning_rate": 9.99902160224741e-06,
+      "loss": 8.9297,
+      "step": 118
+    },
+    {
+      "epoch": 0.011430785624334012,
+      "eval_accuracy": 0.06885003813002805,
+      "eval_loss": 8.859375,
+      "eval_runtime": 240.8979,
+      "eval_samples_per_second": 140.171,
+      "eval_steps_per_second": 2.192,
+      "step": 118
+    },
+    {
+      "epoch": 0.011527656688947011,
+      "grad_norm": 1.5727568864822388,
+      "learning_rate": 9.999011915140948e-06,
+      "loss": 8.8672,
+      "step": 119
+    },
+    {
+      "epoch": 0.011527656688947011,
+      "eval_accuracy": 0.06944707999947082,
+      "eval_loss": 8.8515625,
+      "eval_runtime": 239.842,
+      "eval_samples_per_second": 140.789,
+      "eval_steps_per_second": 2.201,
+      "step": 119
+    },
+    {
+      "epoch": 0.011624527753560012,
+      "grad_norm": 1.5209897756576538,
+      "learning_rate": 9.999002228034487e-06,
+      "loss": 8.8906,
+      "step": 120
+    },
+    {
+      "epoch": 0.011624527753560012,
+      "eval_accuracy": 0.0699609517132256,
+      "eval_loss": 8.8359375,
+      "eval_runtime": 239.7155,
+      "eval_samples_per_second": 140.863,
+      "eval_steps_per_second": 2.203,
+      "step": 120
+    },
+    {
+      "epoch": 0.011721398818173011,
+      "grad_norm": 1.4772562980651855,
+      "learning_rate": 9.998992540928025e-06,
+      "loss": 8.8984,
+      "step": 121
+    },
+    {
+      "epoch": 0.011721398818173011,
+      "eval_accuracy": 0.07026170171233542,
+      "eval_loss": 8.828125,
+      "eval_runtime": 240.3458,
+      "eval_samples_per_second": 140.493,
+      "eval_steps_per_second": 2.197,
+      "step": 121
+    },
+    {
+      "epoch": 0.011818269882786012,
+      "grad_norm": 1.4959176778793335,
+      "learning_rate": 9.998982853821563e-06,
+      "loss": 8.8984,
+      "step": 122
+    },
+    {
+      "epoch": 0.011818269882786012,
+      "eval_accuracy": 0.07044769831877305,
+      "eval_loss": 8.8203125,
+      "eval_runtime": 242.3847,
+      "eval_samples_per_second": 139.312,
+      "eval_steps_per_second": 2.178,
+      "step": 122
+    },
+    {
+      "epoch": 0.011915140947399011,
+      "grad_norm": 1.5008248090744019,
+      "learning_rate": 9.998973166715103e-06,
+      "loss": 8.8828,
+      "step": 123
+    },
+    {
+      "epoch": 0.011915140947399011,
+      "eval_accuracy": 0.07059994052161438,
+      "eval_loss": 8.8125,
+      "eval_runtime": 242.5169,
+      "eval_samples_per_second": 139.236,
+      "eval_steps_per_second": 2.177,
+      "step": 123
+    },
+    {
+      "epoch": 0.012012012012012012,
+      "grad_norm": 1.5148494243621826,
+      "learning_rate": 9.998963479608643e-06,
+      "loss": 8.8594,
+      "step": 124
+    },
+    {
+      "epoch": 0.012012012012012012,
+      "eval_accuracy": 0.07070221694348897,
+      "eval_loss": 8.8046875,
+      "eval_runtime": 241.6035,
+      "eval_samples_per_second": 139.762,
+      "eval_steps_per_second": 2.185,
+      "step": 124
+    },
+    {
+      "epoch": 0.012108883076625013,
+      "grad_norm": 1.513382077217102,
+      "learning_rate": 9.99895379250218e-06,
+      "loss": 8.8281,
+      "step": 125
+    },
+    {
+      "epoch": 0.012108883076625013,
+      "eval_accuracy": 0.07083503444237392,
+      "eval_loss": 8.796875,
+      "eval_runtime": 241.6868,
+      "eval_samples_per_second": 139.714,
+      "eval_steps_per_second": 2.185,
+      "step": 125
+    },
+    {
+      "epoch": 0.012205754141238012,
+      "grad_norm": 1.5034332275390625,
+      "learning_rate": 9.998944105395719e-06,
+      "loss": 8.8359,
+      "step": 126
+    },
+    {
+      "epoch": 0.012205754141238012,
+      "eval_accuracy": 0.07095690926153383,
+      "eval_loss": 8.78125,
+      "eval_runtime": 240.0043,
+      "eval_samples_per_second": 140.693,
+      "eval_steps_per_second": 2.2,
+      "step": 126
+    },
+    {
+      "epoch": 0.012302625205851013,
+      "grad_norm": 1.4775885343551636,
+      "learning_rate": 9.998934418289259e-06,
+      "loss": 8.8359,
+      "step": 127
+    },
+    {
+      "epoch": 0.012302625205851013,
+      "eval_accuracy": 0.07105356959910508,
+      "eval_loss": 8.7734375,
+      "eval_runtime": 238.0369,
+      "eval_samples_per_second": 141.856,
+      "eval_steps_per_second": 2.218,
+      "step": 127
+    },
+    {
+      "epoch": 0.012399496270464012,
+      "grad_norm": 1.493371605873108,
+      "learning_rate": 9.998924731182797e-06,
+      "loss": 8.8281,
+      "step": 128
+    },
+    {
+      "epoch": 0.012399496270464012,
+      "eval_accuracy": 0.07101825195554806,
+      "eval_loss": 8.765625,
+      "eval_runtime": 240.7256,
+      "eval_samples_per_second": 140.272,
+      "eval_steps_per_second": 2.193,
+      "step": 128
+    },
+    {
+      "epoch": 0.012496367335077013,
+      "grad_norm": 1.4377527236938477,
+      "learning_rate": 9.998915044076335e-06,
+      "loss": 8.8438,
+      "step": 129
+    },
+    {
+      "epoch": 0.012496367335077013,
+      "eval_accuracy": 0.07066889677321507,
+      "eval_loss": 8.7578125,
+      "eval_runtime": 241.9868,
+      "eval_samples_per_second": 139.541,
+      "eval_steps_per_second": 2.182,
+      "step": 129
+    },
+    {
+      "epoch": 0.012593238399690013,
+      "grad_norm": 1.4930979013442993,
+      "learning_rate": 9.998905356969873e-06,
+      "loss": 8.7578,
+      "step": 130
+    },
+    {
+      "epoch": 0.012593238399690013,
+      "eval_accuracy": 0.07018139749657541,
+      "eval_loss": 8.75,
+      "eval_runtime": 241.4545,
+      "eval_samples_per_second": 139.848,
+      "eval_steps_per_second": 2.187,
+      "step": 130
+    },
+    {
+      "epoch": 0.012690109464303012,
+      "grad_norm": 1.4573019742965698,
+      "learning_rate": 9.998895669863412e-06,
+      "loss": 8.7812,
+      "step": 131
+    },
+    {
+      "epoch": 0.012690109464303012,
+      "eval_accuracy": 0.06976317290930623,
+      "eval_loss": 8.7421875,
+      "eval_runtime": 241.0814,
+      "eval_samples_per_second": 140.065,
+      "eval_steps_per_second": 2.19,
+      "step": 131
+    },
+    {
+      "epoch": 0.012786980528916013,
+      "grad_norm": 1.4660903215408325,
+      "learning_rate": 9.998885982756952e-06,
+      "loss": 8.7734,
+      "step": 132
+    },
+    {
+      "epoch": 0.012786980528916013,
+      "eval_accuracy": 0.06973908743435586,
+      "eval_loss": 8.734375,
+      "eval_runtime": 240.7814,
+      "eval_samples_per_second": 140.239,
+      "eval_steps_per_second": 2.193,
+      "step": 132
+    },
+    {
+      "epoch": 0.012883851593529012,
+      "grad_norm": 1.4869788885116577,
+      "learning_rate": 9.99887629565049e-06,
+      "loss": 8.7812,
+      "step": 133
+    },
+    {
+      "epoch": 0.012883851593529012,
+      "eval_accuracy": 0.07012303653804183,
+      "eval_loss": 8.7265625,
+      "eval_runtime": 239.4345,
+      "eval_samples_per_second": 141.028,
+      "eval_steps_per_second": 2.205,
+      "step": 133
+    },
+    {
+      "epoch": 0.012980722658142013,
+      "grad_norm": 1.41987943649292,
+      "learning_rate": 9.998866608544028e-06,
+      "loss": 8.7891,
+      "step": 134
+    },
+    {
+      "epoch": 0.012980722658142013,
+      "eval_accuracy": 0.07073081844499253,
+      "eval_loss": 8.71875,
+      "eval_runtime": 238.6401,
+      "eval_samples_per_second": 141.498,
+      "eval_steps_per_second": 2.213,
+      "step": 134
+    },
+    {
+      "epoch": 0.013077593722755012,
+      "grad_norm": 1.4336059093475342,
+      "learning_rate": 9.998856921437568e-06,
+      "loss": 8.7656,
+      "step": 135
+    },
+    {
+      "epoch": 0.013077593722755012,
+      "eval_accuracy": 0.0712553722984789,
+      "eval_loss": 8.703125,
+      "eval_runtime": 239.3711,
+      "eval_samples_per_second": 141.065,
+      "eval_steps_per_second": 2.206,
+      "step": 135
+    },
+    {
+      "epoch": 0.013174464787368013,
+      "grad_norm": 1.4126527309417725,
+      "learning_rate": 9.998847234331106e-06,
+      "loss": 8.7891,
+      "step": 136
+    },
+    {
+      "epoch": 0.013174464787368013,
+      "eval_accuracy": 0.07193187886592499,
+      "eval_loss": 8.6953125,
+      "eval_runtime": 241.9039,
+      "eval_samples_per_second": 139.588,
+      "eval_steps_per_second": 2.183,
+      "step": 136
+    },
+    {
+      "epoch": 0.013271335851981014,
+      "grad_norm": 1.4684745073318481,
+      "learning_rate": 9.998837547224644e-06,
+      "loss": 8.7188,
+      "step": 137
+    },
+    {
+      "epoch": 0.013271335851981014,
+      "eval_accuracy": 0.0726333393749663,
+      "eval_loss": 8.6875,
+      "eval_runtime": 241.6523,
+      "eval_samples_per_second": 139.734,
+      "eval_steps_per_second": 2.185,
+      "step": 137
+    },
+    {
+      "epoch": 0.013368206916594013,
+      "grad_norm": 1.434983730316162,
+      "learning_rate": 9.998827860118184e-06,
+      "loss": 8.7266,
+      "step": 138
+    },
+    {
+      "epoch": 0.013368206916594013,
+      "eval_accuracy": 0.07333210763740858,
+      "eval_loss": 8.6796875,
+      "eval_runtime": 241.7758,
+      "eval_samples_per_second": 139.662,
+      "eval_steps_per_second": 2.184,
+      "step": 138
+    },
+    {
+      "epoch": 0.013465077981207014,
+      "grad_norm": 1.4157376289367676,
+      "learning_rate": 9.998818173011722e-06,
+      "loss": 8.75,
+      "step": 139
+    },
+    {
+      "epoch": 0.013465077981207014,
+      "eval_accuracy": 0.07374755313141426,
+      "eval_loss": 8.671875,
+      "eval_runtime": 239.433,
+      "eval_samples_per_second": 141.029,
+      "eval_steps_per_second": 2.205,
+      "step": 139
+    },
+    {
+      "epoch": 0.013561949045820013,
+      "grad_norm": 1.4349616765975952,
+      "learning_rate": 9.998808485905261e-06,
+      "loss": 8.7188,
+      "step": 140
+    },
+    {
+      "epoch": 0.013561949045820013,
+      "eval_accuracy": 0.07402439134890268,
+      "eval_loss": 8.6640625,
+      "eval_runtime": 240.1811,
+      "eval_samples_per_second": 140.59,
+      "eval_steps_per_second": 2.198,
+      "step": 140
+    },
+    {
+      "epoch": 0.013658820110433014,
+      "grad_norm": 1.3990036249160767,
+      "learning_rate": 9.9987987987988e-06,
+      "loss": 8.7344,
+      "step": 141
+    },
+    {
+      "epoch": 0.013658820110433014,
+      "eval_accuracy": 0.07422755464602009,
+      "eval_loss": 8.65625,
+      "eval_runtime": 239.6139,
+      "eval_samples_per_second": 140.923,
+      "eval_steps_per_second": 2.204,
+      "step": 141
+    },
+    {
+      "epoch": 0.013755691175046015,
+      "grad_norm": 1.4333348274230957,
+      "learning_rate": 9.99878911169234e-06,
+      "loss": 8.6641,
+      "step": 142
+    },
+    {
+      "epoch": 0.013755691175046015,
+      "eval_accuracy": 0.0741873156914756,
+      "eval_loss": 8.6484375,
+      "eval_runtime": 240.5249,
+      "eval_samples_per_second": 140.389,
+      "eval_steps_per_second": 2.195,
+      "step": 142
+    },
+    {
+      "epoch": 0.013852562239659014,
+      "grad_norm": 1.395568609237671,
+      "learning_rate": 9.998779424585877e-06,
+      "loss": 8.7031,
+      "step": 143
+    },
+    {
+      "epoch": 0.013852562239659014,
+      "eval_accuracy": 0.07413063376845538,
+      "eval_loss": 8.640625,
+      "eval_runtime": 241.0655,
+      "eval_samples_per_second": 140.074,
+      "eval_steps_per_second": 2.19,
+      "step": 143
+    },
+    {
+      "epoch": 0.013949433304272014,
+      "grad_norm": 1.4352083206176758,
+      "learning_rate": 9.998769737479415e-06,
+      "loss": 8.6797,
+      "step": 144
+    },
+    {
+      "epoch": 0.013949433304272014,
+      "eval_accuracy": 0.07412177540867797,
+      "eval_loss": 8.6328125,
+      "eval_runtime": 241.4457,
+      "eval_samples_per_second": 139.853,
+      "eval_steps_per_second": 2.187,
+      "step": 144
+    },
+    {
+      "epoch": 0.014046304368885014,
+      "grad_norm": 1.3976655006408691,
+      "learning_rate": 9.998760050372955e-06,
+      "loss": 8.6797,
+      "step": 145
+    },
+    {
+      "epoch": 0.014046304368885014,
+      "eval_accuracy": 0.07388228125691788,
+      "eval_loss": 8.6171875,
+      "eval_runtime": 242.6627,
+      "eval_samples_per_second": 139.152,
+      "eval_steps_per_second": 2.176,
+      "step": 145
+    },
+    {
+      "epoch": 0.014143175433498014,
+      "grad_norm": 1.4085290431976318,
+      "learning_rate": 9.998750363266493e-06,
+      "loss": 8.6719,
+      "step": 146
+    },
+    {
+      "epoch": 0.014143175433498014,
+      "eval_accuracy": 0.07361215918148292,
+      "eval_loss": 8.609375,
+      "eval_runtime": 240.5995,
+      "eval_samples_per_second": 140.345,
+      "eval_steps_per_second": 2.195,
+      "step": 146
+    },
+    {
+      "epoch": 0.014240046498111014,
+      "grad_norm": 1.4239732027053833,
+      "learning_rate": 9.998740676160031e-06,
+      "loss": 8.6641,
+      "step": 147
+    },
+    {
+      "epoch": 0.014240046498111014,
+      "eval_accuracy": 0.07357530724685334,
+      "eval_loss": 8.6015625,
+      "eval_runtime": 240.087,
+      "eval_samples_per_second": 140.645,
+      "eval_steps_per_second": 2.199,
+      "step": 147
+    },
+    {
+      "epoch": 0.014336917562724014,
+      "grad_norm": 1.4121946096420288,
+      "learning_rate": 9.99873098905357e-06,
+      "loss": 8.6484,
+      "step": 148
+    },
+    {
+      "epoch": 0.014336917562724014,
+      "eval_accuracy": 0.07371012221902144,
+      "eval_loss": 8.59375,
+      "eval_runtime": 240.6261,
+      "eval_samples_per_second": 140.33,
+      "eval_steps_per_second": 2.194,
+      "step": 148
+    },
+    {
+      "epoch": 0.014433788627337015,
+      "grad_norm": 1.4172465801239014,
+      "learning_rate": 9.99872130194711e-06,
+      "loss": 8.6172,
+      "step": 149
+    },
+    {
+      "epoch": 0.014433788627337015,
+      "eval_accuracy": 0.07406176436351918,
+      "eval_loss": 8.5859375,
+      "eval_runtime": 238.2968,
+      "eval_samples_per_second": 141.701,
+      "eval_steps_per_second": 2.216,
+      "step": 149
+    },
+    {
+      "epoch": 0.014530659691950014,
+      "grad_norm": 1.366936445236206,
+      "learning_rate": 9.998711614840649e-06,
+      "loss": 8.6719,
+      "step": 150
+    },
+    {
+      "epoch": 0.014530659691950014,
+      "eval_accuracy": 0.0746068140298239,
+      "eval_loss": 8.578125,
+      "eval_runtime": 241.6018,
+      "eval_samples_per_second": 139.763,
+      "eval_steps_per_second": 2.185,
+      "step": 150
+    },
+    {
+      "epoch": 0.014627530756563015,
+      "grad_norm": 1.3992184400558472,
+      "learning_rate": 9.998701927734187e-06,
+      "loss": 8.6406,
+      "step": 151
+    },
+    {
+      "epoch": 0.014627530756563015,
+      "eval_accuracy": 0.07503204424802817,
+      "eval_loss": 8.5703125,
+      "eval_runtime": 242.6844,
+      "eval_samples_per_second": 139.14,
+      "eval_steps_per_second": 2.176,
+      "step": 151
+    },
+    {
+      "epoch": 0.014724401821176014,
+      "grad_norm": 1.3632800579071045,
+      "learning_rate": 9.998692240627725e-06,
+      "loss": 8.6172,
+      "step": 152
+    },
+    {
+      "epoch": 0.014724401821176014,
+      "eval_accuracy": 0.07536941459066229,
+      "eval_loss": 8.5625,
+      "eval_runtime": 242.1383,
+      "eval_samples_per_second": 139.453,
+      "eval_steps_per_second": 2.181,
+      "step": 152
+    },
+    {
+      "epoch": 0.014821272885789015,
+      "grad_norm": 1.3528015613555908,
+      "learning_rate": 9.998682553521264e-06,
+      "loss": 8.6094,
+      "step": 153
+    },
+    {
+      "epoch": 0.014821272885789015,
+      "eval_accuracy": 0.07561990931992374,
+      "eval_loss": 8.5546875,
+      "eval_runtime": 241.5544,
+      "eval_samples_per_second": 139.79,
+      "eval_steps_per_second": 2.186,
+      "step": 153
+    },
+    {
+      "epoch": 0.014918143950402014,
+      "grad_norm": 1.3605914115905762,
+      "learning_rate": 9.998672866414802e-06,
+      "loss": 8.6016,
+      "step": 154
+    },
+    {
+      "epoch": 0.014918143950402014,
+      "eval_accuracy": 0.07561860661995648,
+      "eval_loss": 8.546875,
+      "eval_runtime": 241.8113,
+      "eval_samples_per_second": 139.642,
+      "eval_steps_per_second": 2.184,
+      "step": 154
+    },
+    {
+      "epoch": 0.015015015015015015,
+      "grad_norm": 1.3871914148330688,
+      "learning_rate": 9.99866317930834e-06,
+      "loss": 8.5625,
+      "step": 155
+    },
+    {
+      "epoch": 0.015015015015015015,
+      "eval_accuracy": 0.07551598281142397,
+      "eval_loss": 8.5390625,
+      "eval_runtime": 239.4463,
+      "eval_samples_per_second": 141.021,
+      "eval_steps_per_second": 2.205,
+      "step": 155
+    },
+    {
+      "epoch": 0.015111886079628016,
+      "grad_norm": 1.3802464008331299,
+      "learning_rate": 9.99865349220188e-06,
+      "loss": 8.5312,
+      "step": 156
+    },
+    {
+      "epoch": 0.015111886079628016,
+      "eval_accuracy": 0.07555937719477805,
+      "eval_loss": 8.53125,
+      "eval_runtime": 240.4355,
+      "eval_samples_per_second": 140.441,
+      "eval_steps_per_second": 2.196,
+      "step": 156
+    },
+    {
+      "epoch": 0.015208757144241015,
+      "grad_norm": 1.372429609298706,
+      "learning_rate": 9.99864380509542e-06,
+      "loss": 8.5703,
+      "step": 157
+    },
+    {
+      "epoch": 0.015208757144241015,
+      "eval_accuracy": 0.07562842029304323,
+      "eval_loss": 8.5234375,
+      "eval_runtime": 241.532,
+      "eval_samples_per_second": 139.803,
+      "eval_steps_per_second": 2.186,
+      "step": 157
+    },
+    {
+      "epoch": 0.015305628208854016,
+      "grad_norm": 1.2991220951080322,
+      "learning_rate": 9.998634117988958e-06,
+      "loss": 8.6172,
+      "step": 158
+    },
+    {
+      "epoch": 0.015305628208854016,
+      "eval_accuracy": 0.07567942823398378,
+      "eval_loss": 8.515625,
+      "eval_runtime": 240.874,
+      "eval_samples_per_second": 140.185,
+      "eval_steps_per_second": 2.192,
+      "step": 158
+    },
+    {
+      "epoch": 0.015402499273467015,
+      "grad_norm": 1.3360435962677002,
+      "learning_rate": 9.998624430882496e-06,
+      "loss": 8.5781,
+      "step": 159
+    },
+    {
+      "epoch": 0.015402499273467015,
+      "eval_accuracy": 0.07568432059608309,
+      "eval_loss": 8.5078125,
+      "eval_runtime": 242.8721,
+      "eval_samples_per_second": 139.032,
+      "eval_steps_per_second": 2.174,
+      "step": 159
+    },
+    {
+      "epoch": 0.015499370338080016,
+      "grad_norm": 1.323406457901001,
+      "learning_rate": 9.998614743776036e-06,
+      "loss": 8.6016,
+      "step": 160
+    },
+    {
+      "epoch": 0.015499370338080016,
+      "eval_accuracy": 0.0758624431049408,
+      "eval_loss": 8.5,
+      "eval_runtime": 241.6758,
+      "eval_samples_per_second": 139.72,
+      "eval_steps_per_second": 2.185,
+      "step": 160
+    },
+    {
+      "epoch": 0.015596241402693016,
+      "grad_norm": 1.318803310394287,
+      "learning_rate": 9.998605056669574e-06,
+      "loss": 8.5547,
+      "step": 161
+    },
+    {
+      "epoch": 0.015596241402693016,
+      "eval_accuracy": 0.076235681120007,
+      "eval_loss": 8.4921875,
+      "eval_runtime": 241.1344,
+      "eval_samples_per_second": 140.034,
+      "eval_steps_per_second": 2.19,
+      "step": 161
+    },
+    {
+      "epoch": 0.015693112467306015,
+      "grad_norm": 1.3205766677856445,
+      "learning_rate": 9.998595369563112e-06,
+      "loss": 8.5547,
+      "step": 162
+    },
+    {
+      "epoch": 0.015693112467306015,
+      "eval_accuracy": 0.07658955232889318,
+      "eval_loss": 8.484375,
+      "eval_runtime": 241.2454,
+      "eval_samples_per_second": 139.97,
+      "eval_steps_per_second": 2.189,
+      "step": 162
+    },
+    {
+      "epoch": 0.015789983531919016,
+      "grad_norm": 1.3163981437683105,
+      "learning_rate": 9.998585682456651e-06,
+      "loss": 8.5312,
+      "step": 163
+    },
+    {
+      "epoch": 0.015789983531919016,
+      "eval_accuracy": 0.07672393306773886,
+      "eval_loss": 8.4765625,
+      "eval_runtime": 239.2463,
+      "eval_samples_per_second": 141.139,
+      "eval_steps_per_second": 2.207,
+      "step": 163
+    },
+    {
+      "epoch": 0.015886854596532017,
+      "grad_norm": 1.330255389213562,
+      "learning_rate": 9.99857599535019e-06,
+      "loss": 8.5,
+      "step": 164
+    },
+    {
+      "epoch": 0.015886854596532017,
+      "eval_accuracy": 0.07668372306208254,
+      "eval_loss": 8.46875,
+      "eval_runtime": 242.9928,
+      "eval_samples_per_second": 138.963,
+      "eval_steps_per_second": 2.173,
+      "step": 164
+    },
+    {
+      "epoch": 0.015983725661145014,
+      "grad_norm": 1.3156908750534058,
+      "learning_rate": 9.998566308243727e-06,
+      "loss": 8.5312,
+      "step": 165
+    },
+    {
+      "epoch": 0.015983725661145014,
+      "eval_accuracy": 0.07662301724360787,
+      "eval_loss": 8.4609375,
+      "eval_runtime": 242.8743,
+      "eval_samples_per_second": 139.031,
+      "eval_steps_per_second": 2.174,
+      "step": 165
+    },
+    {
+      "epoch": 0.016080596725758015,
+      "grad_norm": 1.3243170976638794,
+      "learning_rate": 9.998556621137267e-06,
+      "loss": 8.5312,
+      "step": 166
+    },
+    {
+      "epoch": 0.016080596725758015,
+      "eval_accuracy": 0.07658451522235309,
+      "eval_loss": 8.453125,
+      "eval_runtime": 242.3105,
+      "eval_samples_per_second": 139.354,
+      "eval_steps_per_second": 2.179,
+      "step": 166
+    },
+    {
+      "epoch": 0.016177467790371016,
+      "grad_norm": 1.343295931816101,
+      "learning_rate": 9.998546934030807e-06,
+      "loss": 8.4531,
+      "step": 167
+    },
+    {
+      "epoch": 0.016177467790371016,
+      "eval_accuracy": 0.07665937704713872,
+      "eval_loss": 8.4453125,
+      "eval_runtime": 242.1425,
+      "eval_samples_per_second": 139.451,
+      "eval_steps_per_second": 2.181,
+      "step": 167
+    },
+    {
+      "epoch": 0.016274338854984017,
+      "grad_norm": 1.3232625722885132,
+      "learning_rate": 9.998537246924345e-06,
+      "loss": 8.4766,
+      "step": 168
+    },
+    {
+      "epoch": 0.016274338854984017,
+      "eval_accuracy": 0.07678096237741702,
+      "eval_loss": 8.4375,
+      "eval_runtime": 241.0831,
+      "eval_samples_per_second": 140.064,
+      "eval_steps_per_second": 2.19,
+      "step": 168
+    },
+    {
+      "epoch": 0.016371209919597018,
+      "grad_norm": 1.3026939630508423,
+      "learning_rate": 9.998527559817883e-06,
+      "loss": 8.4766,
+      "step": 169
+    },
+    {
+      "epoch": 0.016371209919597018,
+      "eval_accuracy": 0.07695766639075481,
+      "eval_loss": 8.4296875,
+      "eval_runtime": 241.0848,
+      "eval_samples_per_second": 140.063,
+      "eval_steps_per_second": 2.19,
+      "step": 169
+    },
+    {
+      "epoch": 0.016468080984210015,
+      "grad_norm": 1.3087149858474731,
+      "learning_rate": 9.998517872711423e-06,
+      "loss": 8.4688,
+      "step": 170
+    },
+    {
+      "epoch": 0.016468080984210015,
+      "eval_accuracy": 0.0772238514173998,
+      "eval_loss": 8.421875,
+      "eval_runtime": 239.3699,
+      "eval_samples_per_second": 141.066,
+      "eval_steps_per_second": 2.206,
+      "step": 170
+    },
+    {
+      "epoch": 0.016564952048823016,
+      "grad_norm": 1.2863601446151733,
+      "learning_rate": 9.99850818560496e-06,
+      "loss": 8.4922,
+      "step": 171
+    },
+    {
+      "epoch": 0.016564952048823016,
+      "eval_accuracy": 0.077486128344143,
+      "eval_loss": 8.4140625,
+      "eval_runtime": 240.0556,
+      "eval_samples_per_second": 140.663,
+      "eval_steps_per_second": 2.199,
+      "step": 171
+    },
+    {
+      "epoch": 0.016661823113436017,
+      "grad_norm": 1.267603874206543,
+      "learning_rate": 9.998498498498499e-06,
+      "loss": 8.4375,
+      "step": 172
+    },
+    {
+      "epoch": 0.016661823113436017,
+      "eval_accuracy": 0.07766749312847479,
+      "eval_loss": 8.4140625,
+      "eval_runtime": 240.7221,
+      "eval_samples_per_second": 140.274,
+      "eval_steps_per_second": 2.193,
+      "step": 172
+    },
+    {
+      "epoch": 0.016758694178049018,
+      "grad_norm": 1.27737295627594,
+      "learning_rate": 9.998488811392037e-06,
+      "loss": 8.4609,
+      "step": 173
+    },
+    {
+      "epoch": 0.016758694178049018,
+      "eval_accuracy": 0.07765209231997287,
+      "eval_loss": 8.40625,
+      "eval_runtime": 241.2721,
+      "eval_samples_per_second": 139.954,
+      "eval_steps_per_second": 2.188,
+      "step": 173
+    },
+    {
+      "epoch": 0.01685556524266202,
+      "grad_norm": 1.3177382946014404,
+      "learning_rate": 9.998479124285576e-06,
+      "loss": 8.4141,
+      "step": 174
+    },
+    {
+      "epoch": 0.01685556524266202,
+      "eval_accuracy": 0.07766830369734332,
+      "eval_loss": 8.3984375,
+      "eval_runtime": 241.2269,
+      "eval_samples_per_second": 139.98,
+      "eval_steps_per_second": 2.189,
+      "step": 174
+    },
+    {
+      "epoch": 0.016952436307275016,
+      "grad_norm": 1.2761859893798828,
+      "learning_rate": 9.998469437179116e-06,
+      "loss": 8.4531,
+      "step": 175
+    },
+    {
+      "epoch": 0.016952436307275016,
+      "eval_accuracy": 0.0777974157385436,
+      "eval_loss": 8.390625,
+      "eval_runtime": 242.9014,
+      "eval_samples_per_second": 139.015,
+      "eval_steps_per_second": 2.174,
+      "step": 175
+    },
+    {
+      "epoch": 0.017049307371888017,
+      "grad_norm": 1.25885808467865,
+      "learning_rate": 9.998459750072654e-06,
+      "loss": 8.3984,
+      "step": 176
+    },
+    {
+      "epoch": 0.017049307371888017,
+      "eval_accuracy": 0.07782648042225775,
+      "eval_loss": 8.3828125,
+      "eval_runtime": 240.1918,
+      "eval_samples_per_second": 140.584,
+      "eval_steps_per_second": 2.198,
+      "step": 176
+    },
+    {
+      "epoch": 0.017146178436501017,
+      "grad_norm": 1.258347988128662,
+      "learning_rate": 9.998450062966192e-06,
+      "loss": 8.4141,
+      "step": 177
+    },
+    {
+      "epoch": 0.017146178436501017,
+      "eval_accuracy": 0.07790733466689281,
+      "eval_loss": 8.375,
+      "eval_runtime": 239.9645,
+      "eval_samples_per_second": 140.717,
+      "eval_steps_per_second": 2.2,
+      "step": 177
+    },
+    {
+      "epoch": 0.017243049501114018,
+      "grad_norm": 1.2390292882919312,
+      "learning_rate": 9.998440375859732e-06,
+      "loss": 8.4453,
+      "step": 178
+    },
+    {
+      "epoch": 0.017243049501114018,
+      "eval_accuracy": 0.07808157802473688,
+      "eval_loss": 8.3671875,
+      "eval_runtime": 238.5786,
+      "eval_samples_per_second": 141.534,
+      "eval_steps_per_second": 2.213,
+      "step": 178
+    },
+    {
+      "epoch": 0.01733992056572702,
+      "grad_norm": 1.2472869157791138,
+      "learning_rate": 9.99843068875327e-06,
+      "loss": 8.4219,
+      "step": 179
+    },
+    {
+      "epoch": 0.01733992056572702,
+      "eval_accuracy": 0.07831968262986522,
+      "eval_loss": 8.359375,
+      "eval_runtime": 240.78,
+      "eval_samples_per_second": 140.24,
+      "eval_steps_per_second": 2.193,
+      "step": 179
+    },
+    {
+      "epoch": 0.017436791630340016,
+      "grad_norm": 1.2041652202606201,
+      "learning_rate": 9.998421001646808e-06,
+      "loss": 8.4219,
+      "step": 180
+    },
+    {
+      "epoch": 0.017436791630340016,
+      "eval_accuracy": 0.07850064212976275,
+      "eval_loss": 8.3515625,
+      "eval_runtime": 243.4958,
+      "eval_samples_per_second": 138.676,
+      "eval_steps_per_second": 2.168,
+      "step": 180
+    },
+    {
+      "epoch": 0.017533662694953017,
+      "grad_norm": 1.1937860250473022,
+      "learning_rate": 9.998411314540348e-06,
+      "loss": 8.4062,
+      "step": 181
+    },
+    {
+      "epoch": 0.017533662694953017,
+      "eval_accuracy": 0.07853622031331324,
+      "eval_loss": 8.34375,
+      "eval_runtime": 242.9108,
+      "eval_samples_per_second": 139.01,
+      "eval_steps_per_second": 2.174,
+      "step": 181
+    },
+    {
+      "epoch": 0.017630533759566018,
+      "grad_norm": 1.2295143604278564,
+      "learning_rate": 9.998401627433887e-06,
+      "loss": 8.3984,
+      "step": 182
+    },
+    {
+      "epoch": 0.017630533759566018,
+      "eval_accuracy": 0.07868744930506891,
+      "eval_loss": 8.3359375,
+      "eval_runtime": 242.8021,
+      "eval_samples_per_second": 139.072,
+      "eval_steps_per_second": 2.175,
+      "step": 182
+    },
+    {
+      "epoch": 0.01772740482417902,
+      "grad_norm": 1.229699730873108,
+      "learning_rate": 9.998391940327426e-06,
+      "loss": 8.3828,
+      "step": 183
+    },
+    {
+      "epoch": 0.01772740482417902,
+      "eval_accuracy": 0.0789600320359976,
+      "eval_loss": 8.328125,
+      "eval_runtime": 242.7645,
+      "eval_samples_per_second": 139.094,
+      "eval_steps_per_second": 2.175,
+      "step": 183
+    },
+    {
+      "epoch": 0.017824275888792016,
+      "grad_norm": 1.2546368837356567,
+      "learning_rate": 9.998382253220964e-06,
+      "loss": 8.375,
+      "step": 184
+    },
+    {
+      "epoch": 0.017824275888792016,
+      "eval_accuracy": 0.07920493962984389,
+      "eval_loss": 8.3203125,
+      "eval_runtime": 239.9494,
+      "eval_samples_per_second": 140.726,
+      "eval_steps_per_second": 2.2,
+      "step": 184
+    },
+    {
+      "epoch": 0.017921146953405017,
+      "grad_norm": 1.2304656505584717,
+      "learning_rate": 9.998372566114503e-06,
+      "loss": 8.3594,
+      "step": 185
+    },
+    {
+      "epoch": 0.017921146953405017,
+      "eval_accuracy": 0.07946394533222483,
+      "eval_loss": 8.3125,
+      "eval_runtime": 239.6503,
+      "eval_samples_per_second": 140.901,
+      "eval_steps_per_second": 2.203,
+      "step": 185
+    },
+    {
+      "epoch": 0.018018018018018018,
+      "grad_norm": 1.1984411478042603,
+      "learning_rate": 9.998362879008041e-06,
+      "loss": 8.375,
+      "step": 186
+    },
+    {
+      "epoch": 0.018018018018018018,
+      "eval_accuracy": 0.07966461902496035,
+      "eval_loss": 8.3125,
+      "eval_runtime": 240.948,
+      "eval_samples_per_second": 140.142,
+      "eval_steps_per_second": 2.191,
+      "step": 186
+    },
+    {
+      "epoch": 0.01811488908263102,
+      "grad_norm": 1.2182241678237915,
+      "learning_rate": 9.99835319190158e-06,
+      "loss": 8.3125,
+      "step": 187
+    },
+    {
+      "epoch": 0.01811488908263102,
+      "eval_accuracy": 0.07963677019454897,
+      "eval_loss": 8.3046875,
+      "eval_runtime": 241.9831,
+      "eval_samples_per_second": 139.543,
+      "eval_steps_per_second": 2.182,
+      "step": 187
+    },
+    {
+      "epoch": 0.01821176014724402,
+      "grad_norm": 1.2160634994506836,
+      "learning_rate": 9.998343504795119e-06,
+      "loss": 8.3438,
+      "step": 188
+    },
+    {
+      "epoch": 0.01821176014724402,
+      "eval_accuracy": 0.07958732549356913,
+      "eval_loss": 8.296875,
+      "eval_runtime": 243.0735,
+      "eval_samples_per_second": 138.917,
+      "eval_steps_per_second": 2.172,
+      "step": 188
+    },
+    {
+      "epoch": 0.018308631211857017,
+      "grad_norm": 1.2408748865127563,
+      "learning_rate": 9.998333817688657e-06,
+      "loss": 8.3281,
+      "step": 189
+    },
+    {
+      "epoch": 0.018308631211857017,
+      "eval_accuracy": 0.0795488524212025,
+      "eval_loss": 8.2890625,
+      "eval_runtime": 242.5702,
+      "eval_samples_per_second": 139.205,
+      "eval_steps_per_second": 2.177,
+      "step": 189
+    },
+    {
+      "epoch": 0.018405502276470018,
+      "grad_norm": 1.1811424493789673,
+      "learning_rate": 9.998324130582195e-06,
+      "loss": 8.3359,
+      "step": 190
+    },
+    {
+      "epoch": 0.018405502276470018,
+      "eval_accuracy": 0.07950351846234159,
+      "eval_loss": 8.28125,
+      "eval_runtime": 241.8027,
+      "eval_samples_per_second": 139.647,
+      "eval_steps_per_second": 2.184,
+      "step": 190
+    },
+    {
+      "epoch": 0.01850237334108302,
+      "grad_norm": 1.2242190837860107,
+      "learning_rate": 9.998314443475735e-06,
+      "loss": 8.3047,
+      "step": 191
+    },
+    {
+      "epoch": 0.01850237334108302,
+      "eval_accuracy": 0.07975195782054359,
+      "eval_loss": 8.2734375,
+      "eval_runtime": 241.7121,
+      "eval_samples_per_second": 139.699,
+      "eval_steps_per_second": 2.184,
+      "step": 191
+    },
+    {
+      "epoch": 0.01859924440569602,
+      "grad_norm": 1.2018407583236694,
+      "learning_rate": 9.998304756369273e-06,
+      "loss": 8.3359,
+      "step": 192
+    },
+    {
+      "epoch": 0.01859924440569602,
+      "eval_accuracy": 0.08004521005761958,
+      "eval_loss": 8.265625,
+      "eval_runtime": 241.0872,
+      "eval_samples_per_second": 140.061,
+      "eval_steps_per_second": 2.19,
+      "step": 192
+    },
+    {
+      "epoch": 0.01869611547030902,
+      "grad_norm": 1.1608692407608032,
+      "learning_rate": 9.998295069262813e-06,
+      "loss": 8.3047,
+      "step": 193
+    },
+    {
+      "epoch": 0.01869611547030902,
+      "eval_accuracy": 0.08031692432190342,
+      "eval_loss": 8.2578125,
+      "eval_runtime": 241.524,
+      "eval_samples_per_second": 139.808,
+      "eval_steps_per_second": 2.186,
+      "step": 193
+    },
+    {
+      "epoch": 0.018792986534922018,
+      "grad_norm": 1.1747246980667114,
+      "learning_rate": 9.99828538215635e-06,
+      "loss": 8.2969,
+      "step": 194
+    },
+    {
+      "epoch": 0.018792986534922018,
+      "eval_accuracy": 0.08048132505777257,
+      "eval_loss": 8.2578125,
+      "eval_runtime": 239.0651,
+      "eval_samples_per_second": 141.246,
+      "eval_steps_per_second": 2.209,
+      "step": 194
+    },
+    {
+      "epoch": 0.01888985759953502,
+      "grad_norm": 1.14412260055542,
+      "learning_rate": 9.998275695049889e-06,
+      "loss": 8.3203,
+      "step": 195
+    },
+    {
+      "epoch": 0.01888985759953502,
+      "eval_accuracy": 0.08070339197885944,
+      "eval_loss": 8.25,
+      "eval_runtime": 241.1404,
+      "eval_samples_per_second": 140.03,
+      "eval_steps_per_second": 2.19,
+      "step": 195
+    },
+    {
+      "epoch": 0.01898672866414802,
+      "grad_norm": 1.1623828411102295,
+      "learning_rate": 9.998266007943428e-06,
+      "loss": 8.2734,
+      "step": 196
+    },
+    {
+      "epoch": 0.01898672866414802,
+      "eval_accuracy": 0.08091521099353713,
+      "eval_loss": 8.2421875,
+      "eval_runtime": 240.6022,
+      "eval_samples_per_second": 140.344,
+      "eval_steps_per_second": 2.194,
+      "step": 196
+    },
+    {
+      "epoch": 0.01908359972876102,
+      "grad_norm": 1.158718466758728,
+      "learning_rate": 9.998256320836966e-06,
+      "loss": 8.25,
+      "step": 197
+    },
+    {
+      "epoch": 0.01908359972876102,
+      "eval_accuracy": 0.08093220399088794,
+      "eval_loss": 8.234375,
+      "eval_runtime": 239.6707,
+      "eval_samples_per_second": 140.889,
+      "eval_steps_per_second": 2.203,
+      "step": 197
+    },
+    {
+      "epoch": 0.01918047079337402,
+      "grad_norm": 1.1638875007629395,
+      "learning_rate": 9.998246633730504e-06,
+      "loss": 8.2734,
+      "step": 198
+    },
+    {
+      "epoch": 0.01918047079337402,
+      "eval_accuracy": 0.08096642157669483,
+      "eval_loss": 8.2265625,
+      "eval_runtime": 239.016,
+      "eval_samples_per_second": 141.275,
+      "eval_steps_per_second": 2.209,
+      "step": 198
+    },
+    {
+      "epoch": 0.019277341857987018,
+      "grad_norm": 1.2000994682312012,
+      "learning_rate": 9.998236946624044e-06,
+      "loss": 8.2109,
+      "step": 199
+    },
+    {
+      "epoch": 0.019277341857987018,
+      "eval_accuracy": 0.08091442937355676,
+      "eval_loss": 8.21875,
+      "eval_runtime": 238.0069,
+      "eval_samples_per_second": 141.874,
+      "eval_steps_per_second": 2.218,
+      "step": 199
+    },
+    {
+      "epoch": 0.01937421292260002,
+      "grad_norm": 1.147640347480774,
+      "learning_rate": 9.998227259517584e-06,
+      "loss": 8.25,
+      "step": 200
+    },
+    {
+      "epoch": 0.01937421292260002,
+      "eval_accuracy": 0.08094085970844822,
+      "eval_loss": 8.2109375,
+      "eval_runtime": 239.3683,
+      "eval_samples_per_second": 141.067,
+      "eval_steps_per_second": 2.206,
+      "step": 200
+    },
+    {
+      "epoch": 0.01947108398721302,
+      "grad_norm": 1.143306851387024,
+      "learning_rate": 9.998217572411122e-06,
+      "loss": 8.2734,
+      "step": 201
+    },
+    {
+      "epoch": 0.01947108398721302,
+      "eval_accuracy": 0.08100220240246243,
+      "eval_loss": 8.203125,
+      "eval_runtime": 240.6044,
+      "eval_samples_per_second": 140.342,
+      "eval_steps_per_second": 2.194,
+      "step": 201
+    },
+    {
+      "epoch": 0.01956795505182602,
+      "grad_norm": 1.1851661205291748,
+      "learning_rate": 9.99820788530466e-06,
+      "loss": 8.2188,
+      "step": 202
+    },
+    {
+      "epoch": 0.01956795505182602,
+      "eval_accuracy": 0.08122783003679317,
+      "eval_loss": 8.203125,
+      "eval_runtime": 239.3597,
+      "eval_samples_per_second": 141.072,
+      "eval_steps_per_second": 2.206,
+      "step": 202
+    },
+    {
+      "epoch": 0.019664826116439018,
+      "grad_norm": 1.1528866291046143,
+      "learning_rate": 9.9981981981982e-06,
+      "loss": 8.2578,
+      "step": 203
+    },
+    {
+      "epoch": 0.019664826116439018,
+      "eval_accuracy": 0.08159895478302359,
+      "eval_loss": 8.1953125,
+      "eval_runtime": 239.5354,
+      "eval_samples_per_second": 140.969,
+      "eval_steps_per_second": 2.204,
+      "step": 203
+    },
+    {
+      "epoch": 0.01976169718105202,
+      "grad_norm": 1.121721625328064,
+      "learning_rate": 9.998188511091738e-06,
+      "loss": 8.2344,
+      "step": 204
+    },
+    {
+      "epoch": 0.01976169718105202,
+      "eval_accuracy": 0.08193849629226983,
+      "eval_loss": 8.1875,
+      "eval_runtime": 240.2982,
+      "eval_samples_per_second": 140.521,
+      "eval_steps_per_second": 2.197,
+      "step": 204
+    },
+    {
+      "epoch": 0.01985856824566502,
+      "grad_norm": 1.120973825454712,
+      "learning_rate": 9.998178823985276e-06,
+      "loss": 8.2969,
+      "step": 205
+    },
+    {
+      "epoch": 0.01985856824566502,
+      "eval_accuracy": 0.0822807879458914,
+      "eval_loss": 8.1796875,
+      "eval_runtime": 239.9882,
+      "eval_samples_per_second": 140.703,
+      "eval_steps_per_second": 2.2,
+      "step": 205
+    },
+    {
+      "epoch": 0.01995543931027802,
+      "grad_norm": 1.1114741563796997,
+      "learning_rate": 9.998169136878815e-06,
+      "loss": 8.2812,
+      "step": 206
+    },
+    {
+      "epoch": 0.01995543931027802,
+      "eval_accuracy": 0.08245699982813046,
+      "eval_loss": 8.171875,
+      "eval_runtime": 237.9497,
+      "eval_samples_per_second": 141.908,
+      "eval_steps_per_second": 2.219,
+      "step": 206
+    },
+    {
+      "epoch": 0.02005231037489102,
+      "grad_norm": 1.1044974327087402,
+      "learning_rate": 9.998159449772353e-06,
+      "loss": 8.2578,
+      "step": 207
+    },
+    {
+      "epoch": 0.02005231037489102,
+      "eval_accuracy": 0.08243728563529247,
+      "eval_loss": 8.1640625,
+      "eval_runtime": 239.4885,
+      "eval_samples_per_second": 140.996,
+      "eval_steps_per_second": 2.205,
+      "step": 207
+    },
+    {
+      "epoch": 0.02014918143950402,
+      "grad_norm": 1.156922459602356,
+      "learning_rate": 9.998149762665893e-06,
+      "loss": 8.2031,
+      "step": 208
+    },
+    {
+      "epoch": 0.02014918143950402,
+      "eval_accuracy": 0.08235234959742663,
+      "eval_loss": 8.1640625,
+      "eval_runtime": 241.2039,
+      "eval_samples_per_second": 139.994,
+      "eval_steps_per_second": 2.189,
+      "step": 208
+    },
+    {
+      "epoch": 0.02024605250411702,
+      "grad_norm": 1.1297597885131836,
+      "learning_rate": 9.998140075559431e-06,
+      "loss": 8.1953,
+      "step": 209
+    },
+    {
+      "epoch": 0.02024605250411702,
+      "eval_accuracy": 0.08222390338065405,
+      "eval_loss": 8.15625,
+      "eval_runtime": 240.9498,
+      "eval_samples_per_second": 140.141,
+      "eval_steps_per_second": 2.191,
+      "step": 209
+    },
+    {
+      "epoch": 0.02034292356873002,
+      "grad_norm": 1.0749586820602417,
+      "learning_rate": 9.998130388452971e-06,
+      "loss": 8.2344,
+      "step": 210
+    },
+    {
+      "epoch": 0.02034292356873002,
+      "eval_accuracy": 0.08212646142310244,
+      "eval_loss": 8.1484375,
+      "eval_runtime": 240.1358,
+      "eval_samples_per_second": 140.616,
+      "eval_steps_per_second": 2.199,
+      "step": 210
+    },
+    {
+      "epoch": 0.02043979463334302,
+      "grad_norm": 1.1359708309173584,
+      "learning_rate": 9.998120701346509e-06,
+      "loss": 8.1484,
+      "step": 211
+    },
+    {
+      "epoch": 0.02043979463334302,
+      "eval_accuracy": 0.08224295174906432,
+      "eval_loss": 8.140625,
+      "eval_runtime": 239.69,
+      "eval_samples_per_second": 140.878,
+      "eval_steps_per_second": 2.203,
+      "step": 211
+    },
+    {
+      "epoch": 0.020536665697956022,
+      "grad_norm": 1.0708686113357544,
+      "learning_rate": 9.998111014240047e-06,
+      "loss": 8.2188,
+      "step": 212
+    },
+    {
+      "epoch": 0.020536665697956022,
+      "eval_accuracy": 0.08242179798012607,
+      "eval_loss": 8.1328125,
+      "eval_runtime": 239.7594,
+      "eval_samples_per_second": 140.837,
+      "eval_steps_per_second": 2.202,
+      "step": 212
+    },
+    {
+      "epoch": 0.02063353676256902,
+      "grad_norm": 1.1410462856292725,
+      "learning_rate": 9.998101327133585e-06,
+      "loss": 8.1406,
+      "step": 213
+    },
+    {
+      "epoch": 0.02063353676256902,
+      "eval_accuracy": 0.08258810934261389,
+      "eval_loss": 8.1328125,
+      "eval_runtime": 236.9832,
+      "eval_samples_per_second": 142.487,
+      "eval_steps_per_second": 2.228,
+      "step": 213
+    },
+    {
+      "epoch": 0.02073040782718202,
+      "grad_norm": 1.1414846181869507,
+      "learning_rate": 9.998091640027125e-06,
+      "loss": 8.1641,
+      "step": 214
+    },
+    {
+      "epoch": 0.02073040782718202,
+      "eval_accuracy": 0.08288260638188082,
+      "eval_loss": 8.125,
+      "eval_runtime": 238.8778,
+      "eval_samples_per_second": 141.357,
+      "eval_steps_per_second": 2.21,
+      "step": 214
+    },
+    {
+      "epoch": 0.02082727889179502,
+      "grad_norm": 1.1182080507278442,
+      "learning_rate": 9.998081952920663e-06,
+      "loss": 8.1328,
+      "step": 215
+    },
+    {
+      "epoch": 0.02082727889179502,
+      "eval_accuracy": 0.08311975567369982,
+      "eval_loss": 8.1171875,
+      "eval_runtime": 241.545,
+      "eval_samples_per_second": 139.796,
+      "eval_steps_per_second": 2.186,
+      "step": 215
+    },
+    {
+      "epoch": 0.020924149956408022,
+      "grad_norm": 1.0699214935302734,
+      "learning_rate": 9.998072265814202e-06,
+      "loss": 8.1875,
+      "step": 216
+    },
+    {
+      "epoch": 0.020924149956408022,
+      "eval_accuracy": 0.08333493275940426,
+      "eval_loss": 8.109375,
+      "eval_runtime": 240.2573,
+      "eval_samples_per_second": 140.545,
+      "eval_steps_per_second": 2.198,
+      "step": 216
+    },
+    {
+      "epoch": 0.021021021021021023,
+      "grad_norm": 1.0673527717590332,
+      "learning_rate": 9.99806257870774e-06,
+      "loss": 8.1719,
+      "step": 217
+    },
+    {
+      "epoch": 0.021021021021021023,
+      "eval_accuracy": 0.0834616709917753,
+      "eval_loss": 8.1015625,
+      "eval_runtime": 241.6637,
+      "eval_samples_per_second": 139.727,
+      "eval_steps_per_second": 2.185,
+      "step": 217
+    },
+    {
+      "epoch": 0.02111789208563402,
+      "grad_norm": 1.0935176610946655,
+      "learning_rate": 9.99805289160128e-06,
+      "loss": 8.125,
+      "step": 218
+    },
+    {
+      "epoch": 0.02111789208563402,
+      "eval_accuracy": 0.08352278209468422,
+      "eval_loss": 8.1015625,
+      "eval_runtime": 239.5275,
+      "eval_samples_per_second": 140.973,
+      "eval_steps_per_second": 2.204,
+      "step": 218
+    },
+    {
+      "epoch": 0.02121476315024702,
+      "grad_norm": 1.0975334644317627,
+      "learning_rate": 9.998043204494818e-06,
+      "loss": 8.1172,
+      "step": 219
+    },
+    {
+      "epoch": 0.02121476315024702,
+      "eval_accuracy": 0.08350500747735307,
+      "eval_loss": 8.09375,
+      "eval_runtime": 238.965,
+      "eval_samples_per_second": 141.305,
+      "eval_steps_per_second": 2.21,
+      "step": 219
+    },
+    {
+      "epoch": 0.02131163421486002,
+      "grad_norm": 1.051080346107483,
+      "learning_rate": 9.998033517388356e-06,
+      "loss": 8.1172,
+      "step": 220
+    },
+    {
+      "epoch": 0.02131163421486002,
+      "eval_accuracy": 0.0834144553551839,
+      "eval_loss": 8.0859375,
+      "eval_runtime": 238.1182,
+      "eval_samples_per_second": 141.808,
+      "eval_steps_per_second": 2.217,
+      "step": 220
+    },
+    {
+      "epoch": 0.021408505279473022,
+      "grad_norm": 1.077547550201416,
+      "learning_rate": 9.998023830281896e-06,
+      "loss": 8.1562,
+      "step": 221
+    },
+    {
+      "epoch": 0.021408505279473022,
+      "eval_accuracy": 0.08347733234027067,
+      "eval_loss": 8.078125,
+      "eval_runtime": 237.0397,
+      "eval_samples_per_second": 142.453,
+      "eval_steps_per_second": 2.227,
+      "step": 221
+    },
+    {
+      "epoch": 0.021505376344086023,
+      "grad_norm": 1.069908857345581,
+      "learning_rate": 9.998014143175434e-06,
+      "loss": 8.0781,
+      "step": 222
+    },
+    {
+      "epoch": 0.021505376344086023,
+      "eval_accuracy": 0.08379307786344815,
+      "eval_loss": 8.078125,
+      "eval_runtime": 239.359,
+      "eval_samples_per_second": 141.073,
+      "eval_steps_per_second": 2.206,
+      "step": 222
+    },
+    {
+      "epoch": 0.02160224740869902,
+      "grad_norm": 1.05584716796875,
+      "learning_rate": 9.998004456068972e-06,
+      "loss": 8.1094,
+      "step": 223
+    },
+    {
+      "epoch": 0.02160224740869902,
+      "eval_accuracy": 0.08401352364679797,
+      "eval_loss": 8.0703125,
+      "eval_runtime": 239.5111,
+      "eval_samples_per_second": 140.983,
+      "eval_steps_per_second": 2.204,
+      "step": 223
+    },
+    {
+      "epoch": 0.02169911847331202,
+      "grad_norm": 1.0520967245101929,
+      "learning_rate": 9.997994768962512e-06,
+      "loss": 8.0938,
+      "step": 224
+    },
+    {
+      "epoch": 0.02169911847331202,
+      "eval_accuracy": 0.08430202826621548,
+      "eval_loss": 8.0625,
+      "eval_runtime": 240.2811,
+      "eval_samples_per_second": 140.531,
+      "eval_steps_per_second": 2.197,
+      "step": 224
+    },
+    {
+      "epoch": 0.021795989537925022,
+      "grad_norm": 1.042554497718811,
+      "learning_rate": 9.997985081856052e-06,
+      "loss": 8.0938,
+      "step": 225
+    },
+    {
+      "epoch": 0.021795989537925022,
+      "eval_accuracy": 0.08457012391947913,
+      "eval_loss": 8.0546875,
+      "eval_runtime": 239.2715,
+      "eval_samples_per_second": 141.124,
+      "eval_steps_per_second": 2.207,
+      "step": 225
+    },
+    {
+      "epoch": 0.021892860602538023,
+      "grad_norm": 1.0677285194396973,
+      "learning_rate": 9.99797539474959e-06,
+      "loss": 8.1016,
+      "step": 226
+    },
+    {
+      "epoch": 0.021892860602538023,
+      "eval_accuracy": 0.08467908753451901,
+      "eval_loss": 8.046875,
+      "eval_runtime": 237.6539,
+      "eval_samples_per_second": 142.085,
+      "eval_steps_per_second": 2.222,
+      "step": 226
+    },
+    {
+      "epoch": 0.02198973166715102,
+      "grad_norm": 1.0176194906234741,
+      "learning_rate": 9.997965707643128e-06,
+      "loss": 8.1094,
+      "step": 227
+    },
+    {
+      "epoch": 0.02198973166715102,
+      "eval_accuracy": 0.08459319618334385,
+      "eval_loss": 8.046875,
+      "eval_runtime": 240.1907,
+      "eval_samples_per_second": 140.584,
+      "eval_steps_per_second": 2.198,
+      "step": 227
+    },
+    {
+      "epoch": 0.02208660273176402,
+      "grad_norm": 1.0204542875289917,
+      "learning_rate": 9.997956020536667e-06,
+      "loss": 8.1016,
+      "step": 228
+    },
+    {
+      "epoch": 0.02208660273176402,
+      "eval_accuracy": 0.08444465943818719,
+      "eval_loss": 8.0390625,
+      "eval_runtime": 239.7402,
+      "eval_samples_per_second": 140.848,
+      "eval_steps_per_second": 2.202,
+      "step": 228
+    },
+    {
+      "epoch": 0.022183473796377022,
+      "grad_norm": 1.0234394073486328,
+      "learning_rate": 9.997946333430205e-06,
+      "loss": 8.0859,
+      "step": 229
+    },
+    {
+      "epoch": 0.022183473796377022,
+      "eval_accuracy": 0.0843871379974103,
+      "eval_loss": 8.03125,
+      "eval_runtime": 238.5255,
+      "eval_samples_per_second": 141.566,
+      "eval_steps_per_second": 2.214,
+      "step": 229
+    },
+    {
+      "epoch": 0.022280344860990023,
+      "grad_norm": 1.0190110206604004,
+      "learning_rate": 9.997936646323743e-06,
+      "loss": 8.0859,
+      "step": 230
+    },
+    {
+      "epoch": 0.022280344860990023,
+      "eval_accuracy": 0.08446454732435414,
+      "eval_loss": 8.03125,
+      "eval_runtime": 239.3718,
+      "eval_samples_per_second": 141.065,
+      "eval_steps_per_second": 2.206,
+      "step": 230
+    },
+    {
+      "epoch": 0.022377215925603024,
+      "grad_norm": 1.0486043691635132,
+      "learning_rate": 9.997926959217283e-06,
+      "loss": 8.1094,
+      "step": 231
+    },
+    {
+      "epoch": 0.022377215925603024,
+      "eval_accuracy": 0.08493841167466973,
+      "eval_loss": 8.0234375,
+      "eval_runtime": 240.0196,
+      "eval_samples_per_second": 140.684,
+      "eval_steps_per_second": 2.2,
+      "step": 231
+    },
+    {
+      "epoch": 0.02247408699021602,
+      "grad_norm": 1.023290753364563,
+      "learning_rate": 9.997917272110821e-06,
+      "loss": 8.1016,
+      "step": 232
+    },
+    {
+      "epoch": 0.02247408699021602,
+      "eval_accuracy": 0.0852934987368587,
+      "eval_loss": 8.015625,
+      "eval_runtime": 238.7993,
+      "eval_samples_per_second": 141.403,
+      "eval_steps_per_second": 2.211,
+      "step": 232
+    },
+    {
+      "epoch": 0.022570958054829022,
+      "grad_norm": 1.0117347240447998,
+      "learning_rate": 9.997907585004361e-06,
+      "loss": 8.0859,
+      "step": 233
+    },
+    {
+      "epoch": 0.022570958054829022,
+      "eval_accuracy": 0.08559141174492868,
+      "eval_loss": 8.0078125,
+      "eval_runtime": 237.5708,
+      "eval_samples_per_second": 142.134,
+      "eval_steps_per_second": 2.222,
+      "step": 233
+    },
+    {
+      "epoch": 0.022667829119442023,
+      "grad_norm": 0.9845523238182068,
+      "learning_rate": 9.997897897897899e-06,
+      "loss": 8.0859,
+      "step": 234
+    },
+    {
+      "epoch": 0.022667829119442023,
+      "eval_accuracy": 0.08566416030087853,
+      "eval_loss": 8.0078125,
+      "eval_runtime": 238.2296,
+      "eval_samples_per_second": 141.741,
+      "eval_steps_per_second": 2.216,
+      "step": 234
+    },
+    {
+      "epoch": 0.022764700184055024,
+      "grad_norm": 0.9753006100654602,
+      "learning_rate": 9.997888210791437e-06,
+      "loss": 8.0781,
+      "step": 235
+    },
+    {
+      "epoch": 0.022764700184055024,
+      "eval_accuracy": 0.08569174859129644,
+      "eval_loss": 8.0,
+      "eval_runtime": 238.7612,
+      "eval_samples_per_second": 141.426,
+      "eval_steps_per_second": 2.211,
+      "step": 235
+    },
+    {
+      "epoch": 0.022861571248668024,
+      "grad_norm": 1.0460307598114014,
+      "learning_rate": 9.997878523684977e-06,
+      "loss": 8.0234,
+      "step": 236
+    },
+    {
+      "epoch": 0.022861571248668024,
+      "eval_accuracy": 0.08559063012494832,
+      "eval_loss": 7.9921875,
+      "eval_runtime": 241.525,
+      "eval_samples_per_second": 139.807,
+      "eval_steps_per_second": 2.186,
+      "step": 236
+    },
+    {
+      "epoch": 0.02295844231328102,
+      "grad_norm": 1.0023905038833618,
+      "learning_rate": 9.997868836578515e-06,
+      "loss": 8.0391,
+      "step": 237
+    },
+    {
+      "epoch": 0.02295844231328102,
+      "eval_accuracy": 0.08547078172795972,
+      "eval_loss": 7.98828125,
+      "eval_runtime": 240.1103,
+      "eval_samples_per_second": 140.631,
+      "eval_steps_per_second": 2.199,
+      "step": 237
+    },
+    {
+      "epoch": 0.023055313377894023,
+      "grad_norm": 1.004289984703064,
+      "learning_rate": 9.997859149472053e-06,
+      "loss": 8.0078,
+      "step": 238
+    },
+    {
+      "epoch": 0.023055313377894023,
+      "eval_accuracy": 0.08553345607082936,
+      "eval_loss": 7.984375,
+      "eval_runtime": 240.4054,
+      "eval_samples_per_second": 140.459,
+      "eval_steps_per_second": 2.196,
+      "step": 238
+    },
+    {
+      "epoch": 0.023152184442507023,
+      "grad_norm": 0.9898872971534729,
+      "learning_rate": 9.997849462365592e-06,
+      "loss": 8.0078,
+      "step": 239
+    },
+    {
+      "epoch": 0.023152184442507023,
+      "eval_accuracy": 0.08570740993979181,
+      "eval_loss": 7.9765625,
+      "eval_runtime": 240.3894,
+      "eval_samples_per_second": 140.468,
+      "eval_steps_per_second": 2.196,
+      "step": 239
+    },
+    {
+      "epoch": 0.023249055507120024,
+      "grad_norm": 1.0223126411437988,
+      "learning_rate": 9.99783977525913e-06,
+      "loss": 7.9883,
+      "step": 240
+    },
+    {
+      "epoch": 0.023249055507120024,
+      "eval_accuracy": 0.08616876837042163,
+      "eval_loss": 7.97265625,
+      "eval_runtime": 237.9077,
+      "eval_samples_per_second": 141.933,
+      "eval_steps_per_second": 2.219,
+      "step": 240
+    },
+    {
+      "epoch": 0.023345926571733025,
+      "grad_norm": 0.9876830577850342,
+      "learning_rate": 9.997830088152668e-06,
+      "loss": 7.9805,
+      "step": 241
+    },
+    {
+      "epoch": 0.023345926571733025,
+      "eval_accuracy": 0.08653523234565806,
+      "eval_loss": 7.96484375,
+      "eval_runtime": 236.1897,
+      "eval_samples_per_second": 142.966,
+      "eval_steps_per_second": 2.235,
+      "step": 241
+    },
+    {
+      "epoch": 0.023442797636346022,
+      "grad_norm": 0.9565869569778442,
+      "learning_rate": 9.997820401046208e-06,
+      "loss": 8.0234,
+      "step": 242
+    },
+    {
+      "epoch": 0.023442797636346022,
+      "eval_accuracy": 0.08681360485421905,
+      "eval_loss": 7.9609375,
+      "eval_runtime": 239.1608,
+      "eval_samples_per_second": 141.19,
+      "eval_steps_per_second": 2.208,
+      "step": 242
+    },
+    {
+      "epoch": 0.023539668700959023,
+      "grad_norm": 0.9919081926345825,
+      "learning_rate": 9.997810713939748e-06,
+      "loss": 7.9961,
+      "step": 243
+    },
+    {
+      "epoch": 0.023539668700959023,
+      "eval_accuracy": 0.0869957512585312,
+      "eval_loss": 7.95703125,
+      "eval_runtime": 238.8915,
+      "eval_samples_per_second": 141.349,
+      "eval_steps_per_second": 2.21,
+      "step": 243
+    },
+    {
+      "epoch": 0.023636539765572024,
+      "grad_norm": 0.9676252007484436,
+      "learning_rate": 9.997801026833286e-06,
+      "loss": 8.0156,
+      "step": 244
+    },
+    {
+      "epoch": 0.023636539765572024,
+      "eval_accuracy": 0.08696109943940189,
+      "eval_loss": 7.94921875,
+      "eval_runtime": 239.0437,
+      "eval_samples_per_second": 141.259,
+      "eval_steps_per_second": 2.209,
+      "step": 244
+    },
+    {
+      "epoch": 0.023733410830185025,
+      "grad_norm": 0.9586087465286255,
+      "learning_rate": 9.997791339726824e-06,
+      "loss": 7.9766,
+      "step": 245
+    },
+    {
+      "epoch": 0.023733410830185025,
+      "eval_accuracy": 0.08687515019045039,
+      "eval_loss": 7.9453125,
+      "eval_runtime": 241.5167,
+      "eval_samples_per_second": 139.812,
+      "eval_steps_per_second": 2.186,
+      "step": 245
+    },
+    {
+      "epoch": 0.023830281894798022,
+      "grad_norm": 0.9658289551734924,
+      "learning_rate": 9.997781652620364e-06,
+      "loss": 7.9297,
+      "step": 246
+    },
+    {
+      "epoch": 0.023830281894798022,
+      "eval_accuracy": 0.08664124317410547,
+      "eval_loss": 7.94140625,
+      "eval_runtime": 239.9096,
+      "eval_samples_per_second": 140.749,
+      "eval_steps_per_second": 2.201,
+      "step": 246
+    },
+    {
+      "epoch": 0.023927152959411023,
+      "grad_norm": 0.9587231874465942,
+      "learning_rate": 9.997771965513902e-06,
+      "loss": 7.9336,
+      "step": 247
+    },
+    {
+      "epoch": 0.023927152959411023,
+      "eval_accuracy": 0.08649270642894882,
+      "eval_loss": 7.9375,
+      "eval_runtime": 238.655,
+      "eval_samples_per_second": 141.489,
+      "eval_steps_per_second": 2.212,
+      "step": 247
+    },
+    {
+      "epoch": 0.024024024024024024,
+      "grad_norm": 0.9744223356246948,
+      "learning_rate": 9.99776227840744e-06,
+      "loss": 7.9219,
+      "step": 248
+    },
+    {
+      "epoch": 0.024024024024024024,
+      "eval_accuracy": 0.08659249324644151,
+      "eval_loss": 7.9296875,
+      "eval_runtime": 238.0999,
+      "eval_samples_per_second": 141.819,
+      "eval_steps_per_second": 2.218,
+      "step": 248
+    },
+    {
+      "epoch": 0.024120895088637025,
+      "grad_norm": 0.9676837921142578,
+      "learning_rate": 9.99775259130098e-06,
+      "loss": 7.957,
+      "step": 249
+    },
+    {
+      "epoch": 0.024120895088637025,
+      "eval_accuracy": 0.08693261373345097,
+      "eval_loss": 7.92578125,
+      "eval_runtime": 239.9677,
+      "eval_samples_per_second": 140.715,
+      "eval_steps_per_second": 2.2,
+      "step": 249
+    },
+    {
+      "epoch": 0.024217766153250025,
+      "grad_norm": 0.9536520838737488,
+      "learning_rate": 9.99774290419452e-06,
+      "loss": 7.9453,
+      "step": 250
+    },
+    {
+      "epoch": 0.024217766153250025,
+      "eval_accuracy": 0.08742873977876275,
+      "eval_loss": 7.91796875,
+      "eval_runtime": 240.6934,
+      "eval_samples_per_second": 140.291,
+      "eval_steps_per_second": 2.194,
+      "step": 250
+    },
+    {
+      "epoch": 0.024314637217863023,
+      "grad_norm": 0.9051578044891357,
+      "learning_rate": 9.997733217088057e-06,
+      "loss": 7.9805,
+      "step": 251
+    },
+    {
+      "epoch": 0.024314637217863023,
+      "eval_accuracy": 0.08791403893990214,
+      "eval_loss": 7.9140625,
+      "eval_runtime": 241.0758,
+      "eval_samples_per_second": 140.068,
+      "eval_steps_per_second": 2.19,
+      "step": 251
+    },
+    {
+      "epoch": 0.024411508282476024,
+      "grad_norm": 0.9311773180961609,
+      "learning_rate": 9.997723529981595e-06,
+      "loss": 7.9531,
+      "step": 252
+    },
+    {
+      "epoch": 0.024411508282476024,
+      "eval_accuracy": 0.08829436943256792,
+      "eval_loss": 7.91015625,
+      "eval_runtime": 239.1867,
+      "eval_samples_per_second": 141.174,
+      "eval_steps_per_second": 2.207,
+      "step": 252
+    },
+    {
+      "epoch": 0.024508379347089024,
+      "grad_norm": 1.0077519416809082,
+      "learning_rate": 9.997713842875133e-06,
+      "loss": 7.9102,
+      "step": 253
+    },
+    {
+      "epoch": 0.024508379347089024,
+      "eval_accuracy": 0.08852494732677427,
+      "eval_loss": 7.90625,
+      "eval_runtime": 240.0697,
+      "eval_samples_per_second": 140.655,
+      "eval_steps_per_second": 2.199,
+      "step": 253
+    },
+    {
+      "epoch": 0.024605250411702025,
+      "grad_norm": 0.8876093029975891,
+      "learning_rate": 9.997704155768673e-06,
+      "loss": 7.9844,
+      "step": 254
+    },
+    {
+      "epoch": 0.024605250411702025,
+      "eval_accuracy": 0.08859361408949334,
+      "eval_loss": 7.8984375,
+      "eval_runtime": 239.2375,
+      "eval_samples_per_second": 141.144,
+      "eval_steps_per_second": 2.207,
+      "step": 254
+    },
+    {
+      "epoch": 0.024702121476315026,
+      "grad_norm": 0.9206761717796326,
+      "learning_rate": 9.997694468662211e-06,
+      "loss": 7.9414,
+      "step": 255
+    },
+    {
+      "epoch": 0.024702121476315026,
+      "eval_accuracy": 0.08849816960522489,
+      "eval_loss": 7.89453125,
+      "eval_runtime": 240.2162,
+      "eval_samples_per_second": 140.569,
+      "eval_steps_per_second": 2.198,
+      "step": 255
+    },
+    {
+      "epoch": 0.024798992540928023,
+      "grad_norm": 0.8806933164596558,
+      "learning_rate": 9.997684781555749e-06,
+      "loss": 7.9453,
+      "step": 256
+    },
+    {
+      "epoch": 0.024798992540928023,
+      "eval_accuracy": 0.088308004358892,
+      "eval_loss": 7.890625,
+      "eval_runtime": 241.0125,
+      "eval_samples_per_second": 140.105,
+      "eval_steps_per_second": 2.191,
+      "step": 256
+    },
+    {
+      "epoch": 0.024895863605541024,
+      "grad_norm": 0.9106225371360779,
+      "learning_rate": 9.997675094449289e-06,
+      "loss": 7.9219,
+      "step": 257
+    },
+    {
+      "epoch": 0.024895863605541024,
+      "eval_accuracy": 0.08826322042890615,
+      "eval_loss": 7.88671875,
+      "eval_runtime": 242.9806,
+      "eval_samples_per_second": 138.97,
+      "eval_steps_per_second": 2.173,
+      "step": 257
+    },
+    {
+      "epoch": 0.024992734670154025,
+      "grad_norm": 0.9404253959655762,
+      "learning_rate": 9.997665407342827e-06,
+      "loss": 7.9141,
+      "step": 258
+    },
+    {
+      "epoch": 0.024992734670154025,
+      "eval_accuracy": 0.08845868332177259,
+      "eval_loss": 7.8828125,
+      "eval_runtime": 239.9439,
+      "eval_samples_per_second": 140.729,
+      "eval_steps_per_second": 2.201,
+      "step": 258
+    },
+    {
+      "epoch": 0.025089605734767026,
+      "grad_norm": 0.943131685256958,
+      "learning_rate": 9.997655720236367e-06,
+      "loss": 7.9258,
+      "step": 259
+    },
+    {
+      "epoch": 0.025089605734767026,
+      "eval_accuracy": 0.08890710159939423,
+      "eval_loss": 7.875,
+      "eval_runtime": 240.2599,
+      "eval_samples_per_second": 140.544,
+      "eval_steps_per_second": 2.198,
+      "step": 259
+    },
+    {
+      "epoch": 0.025186476799380027,
+      "grad_norm": 0.8273116946220398,
+      "learning_rate": 9.997646033129905e-06,
+      "loss": 7.957,
+      "step": 260
+    },
+    {
+      "epoch": 0.025186476799380027,
+      "eval_accuracy": 0.08928497143656629,
+      "eval_loss": 7.87109375,
+      "eval_runtime": 239.6344,
+      "eval_samples_per_second": 140.91,
+      "eval_steps_per_second": 2.203,
+      "step": 260
+    },
+    {
+      "epoch": 0.025283347863993024,
+      "grad_norm": 0.8834772109985352,
+      "learning_rate": 9.997636346023444e-06,
+      "loss": 7.8984,
+      "step": 261
+    },
+    {
+      "epoch": 0.025283347863993024,
+      "eval_accuracy": 0.08963823471880107,
+      "eval_loss": 7.8671875,
+      "eval_runtime": 240.9633,
+      "eval_samples_per_second": 140.133,
+      "eval_steps_per_second": 2.191,
+      "step": 261
+    },
+    {
+      "epoch": 0.025380218928606025,
+      "grad_norm": 0.8725751638412476,
+      "learning_rate": 9.997626658916982e-06,
+      "loss": 7.8945,
+      "step": 262
+    },
+    {
+      "epoch": 0.025380218928606025,
+      "eval_accuracy": 0.08984756412909688,
+      "eval_loss": 7.86328125,
+      "eval_runtime": 239.3388,
+      "eval_samples_per_second": 141.085,
+      "eval_steps_per_second": 2.206,
+      "step": 262
+    },
+    {
+      "epoch": 0.025477089993219026,
+      "grad_norm": 0.8974217176437378,
+      "learning_rate": 9.99761697181052e-06,
+      "loss": 7.9141,
+      "step": 263
+    },
+    {
+      "epoch": 0.025477089993219026,
+      "eval_accuracy": 0.08987159170627092,
+      "eval_loss": 7.859375,
+      "eval_runtime": 239.8452,
+      "eval_samples_per_second": 140.787,
+      "eval_steps_per_second": 2.201,
+      "step": 263
+    },
+    {
+      "epoch": 0.025573961057832027,
+      "grad_norm": 0.8629518747329712,
+      "learning_rate": 9.99760728470406e-06,
+      "loss": 7.9453,
+      "step": 264
+    },
+    {
+      "epoch": 0.025573961057832027,
+      "eval_accuracy": 0.0898933323212802,
+      "eval_loss": 7.85546875,
+      "eval_runtime": 242.7767,
+      "eval_samples_per_second": 139.087,
+      "eval_steps_per_second": 2.175,
+      "step": 264
+    },
+    {
+      "epoch": 0.025670832122445024,
+      "grad_norm": 0.8912859559059143,
+      "learning_rate": 9.997597597597598e-06,
+      "loss": 7.8672,
+      "step": 265
+    },
+    {
+      "epoch": 0.025670832122445024,
+      "eval_accuracy": 0.08997965790577779,
+      "eval_loss": 7.84765625,
+      "eval_runtime": 242.1493,
+      "eval_samples_per_second": 139.447,
+      "eval_steps_per_second": 2.18,
+      "step": 265
+    },
+    {
+      "epoch": 0.025767703187058025,
+      "grad_norm": 0.8678880333900452,
+      "learning_rate": 9.997587910491136e-06,
+      "loss": 7.9375,
+      "step": 266
+    },
+    {
+      "epoch": 0.025767703187058025,
+      "eval_accuracy": 0.09024167429252752,
+      "eval_loss": 7.84375,
+      "eval_runtime": 241.9128,
+      "eval_samples_per_second": 139.583,
+      "eval_steps_per_second": 2.183,
+      "step": 266
+    },
+    {
+      "epoch": 0.025864574251671026,
+      "grad_norm": 0.8691763281822205,
+      "learning_rate": 9.997578223384676e-06,
+      "loss": 7.9219,
+      "step": 267
+    },
+    {
+      "epoch": 0.025864574251671026,
+      "eval_accuracy": 0.09048151583094556,
+      "eval_loss": 7.83984375,
+      "eval_runtime": 240.5035,
+      "eval_samples_per_second": 140.401,
+      "eval_steps_per_second": 2.195,
+      "step": 267
+    },
+    {
+      "epoch": 0.025961445316284026,
+      "grad_norm": 0.8926984071731567,
+      "learning_rate": 9.997568536278216e-06,
+      "loss": 7.8555,
+      "step": 268
+    },
+    {
+      "epoch": 0.025961445316284026,
+      "eval_accuracy": 0.09069718504774874,
+      "eval_loss": 7.8359375,
+      "eval_runtime": 239.1407,
+      "eval_samples_per_second": 141.201,
+      "eval_steps_per_second": 2.208,
+      "step": 268
+    },
+    {
+      "epoch": 0.026058316380897027,
+      "grad_norm": 0.8526946306228638,
+      "learning_rate": 9.997558849171754e-06,
+      "loss": 7.8984,
+      "step": 269
+    },
+    {
+      "epoch": 0.026058316380897027,
+      "eval_accuracy": 0.0907848133322136,
+      "eval_loss": 7.83203125,
+      "eval_runtime": 238.5966,
+      "eval_samples_per_second": 141.523,
+      "eval_steps_per_second": 2.213,
+      "step": 269
+    },
+    {
+      "epoch": 0.026155187445510025,
+      "grad_norm": 0.8387218117713928,
+      "learning_rate": 9.997549162065292e-06,
+      "loss": 7.8906,
+      "step": 270
+    },
+    {
+      "epoch": 0.026155187445510025,
+      "eval_accuracy": 0.09087632076769209,
+      "eval_loss": 7.828125,
+      "eval_runtime": 240.2401,
+      "eval_samples_per_second": 140.555,
+      "eval_steps_per_second": 2.198,
+      "step": 270
+    },
+    {
+      "epoch": 0.026252058510123025,
+      "grad_norm": 0.8393184542655945,
+      "learning_rate": 9.997539474958831e-06,
+      "loss": 7.8711,
+      "step": 271
+    },
+    {
+      "epoch": 0.026252058510123025,
+      "eval_accuracy": 0.09095222475245154,
+      "eval_loss": 7.82421875,
+      "eval_runtime": 239.8682,
+      "eval_samples_per_second": 140.773,
+      "eval_steps_per_second": 2.201,
+      "step": 271
+    },
+    {
+      "epoch": 0.026348929574736026,
+      "grad_norm": 0.8263446092605591,
+      "learning_rate": 9.99752978785237e-06,
+      "loss": 7.8633,
+      "step": 272
+    },
+    {
+      "epoch": 0.026348929574736026,
+      "eval_accuracy": 0.09090057993597143,
+      "eval_loss": 7.8203125,
+      "eval_runtime": 240.3911,
+      "eval_samples_per_second": 140.467,
+      "eval_steps_per_second": 2.196,
+      "step": 272
+    },
+    {
+      "epoch": 0.026445800639349027,
+      "grad_norm": 0.8394054770469666,
+      "learning_rate": 9.997520100745907e-06,
+      "loss": 7.8633,
+      "step": 273
+    },
+    {
+      "epoch": 0.026445800639349027,
+      "eval_accuracy": 0.09090938039797253,
+      "eval_loss": 7.81640625,
+      "eval_runtime": 238.99,
+      "eval_samples_per_second": 141.29,
+      "eval_steps_per_second": 2.209,
+      "step": 273
+    },
+    {
+      "epoch": 0.026542671703962028,
+      "grad_norm": 0.8645299077033997,
+      "learning_rate": 9.997510413639445e-06,
+      "loss": 7.8789,
+      "step": 274
+    },
+    {
+      "epoch": 0.026542671703962028,
+      "eval_accuracy": 0.09088737924296979,
+      "eval_loss": 7.8125,
+      "eval_runtime": 239.144,
+      "eval_samples_per_second": 141.199,
+      "eval_steps_per_second": 2.208,
+      "step": 274
+    },
+    {
+      "epoch": 0.026639542768575025,
+      "grad_norm": 0.832476019859314,
+      "learning_rate": 9.997500726532985e-06,
+      "loss": 7.8438,
+      "step": 275
+    },
+    {
+      "epoch": 0.026639542768575025,
+      "eval_accuracy": 0.09096122785666977,
+      "eval_loss": 7.80859375,
+      "eval_runtime": 238.0741,
+      "eval_samples_per_second": 141.834,
+      "eval_steps_per_second": 2.218,
+      "step": 275
+    },
+    {
+      "epoch": 0.026736413833188026,
+      "grad_norm": 0.8168690800666809,
+      "learning_rate": 9.997491039426525e-06,
+      "loss": 7.8789,
+      "step": 276
+    },
+    {
+      "epoch": 0.026736413833188026,
+      "eval_accuracy": 0.09111972301935398,
+      "eval_loss": 7.8046875,
+      "eval_runtime": 237.8823,
+      "eval_samples_per_second": 141.948,
+      "eval_steps_per_second": 2.22,
+      "step": 276
+    },
+    {
+      "epoch": 0.026833284897801027,
+      "grad_norm": 0.8217095732688904,
+      "learning_rate": 9.997481352320063e-06,
+      "loss": 7.8516,
+      "step": 277
+    },
+    {
+      "epoch": 0.026833284897801027,
+      "eval_accuracy": 0.09120419587500923,
+      "eval_loss": 7.80078125,
+      "eval_runtime": 238.7067,
+      "eval_samples_per_second": 141.458,
+      "eval_steps_per_second": 2.212,
+      "step": 277
+    },
+    {
+      "epoch": 0.026930155962414028,
+      "grad_norm": 0.8041301965713501,
+      "learning_rate": 9.997471665213601e-06,
+      "loss": 7.8711,
+      "step": 278
+    },
+    {
+      "epoch": 0.026930155962414028,
+      "eval_accuracy": 0.09132303106091219,
+      "eval_loss": 7.796875,
+      "eval_runtime": 239.1861,
+      "eval_samples_per_second": 141.175,
+      "eval_steps_per_second": 2.207,
+      "step": 278
+    },
+    {
+      "epoch": 0.02702702702702703,
+      "grad_norm": 0.8525278568267822,
+      "learning_rate": 9.99746197810714e-06,
+      "loss": 7.8008,
+      "step": 279
+    },
+    {
+      "epoch": 0.02702702702702703,
+      "eval_accuracy": 0.0915536379040067,
+      "eval_loss": 7.79296875,
+      "eval_runtime": 239.3931,
+      "eval_samples_per_second": 141.053,
+      "eval_steps_per_second": 2.206,
+      "step": 279
+    },
+    {
+      "epoch": 0.027123898091640026,
+      "grad_norm": 0.8100795745849609,
+      "learning_rate": 9.997452291000679e-06,
+      "loss": 7.8477,
+      "step": 280
+    },
+    {
+      "epoch": 0.027123898091640026,
+      "eval_accuracy": 0.09175946449883497,
+      "eval_loss": 7.7890625,
+      "eval_runtime": 239.2146,
+      "eval_samples_per_second": 141.158,
+      "eval_steps_per_second": 2.207,
+      "step": 280
+    },
+    {
+      "epoch": 0.027220769156253027,
+      "grad_norm": 0.8111329674720764,
+      "learning_rate": 9.997442603894217e-06,
+      "loss": 7.8086,
+      "step": 281
+    },
+    {
+      "epoch": 0.027220769156253027,
+      "eval_accuracy": 0.09185091403653714,
+      "eval_loss": 7.78515625,
+      "eval_runtime": 238.4234,
+      "eval_samples_per_second": 141.626,
+      "eval_steps_per_second": 2.215,
+      "step": 281
+    },
+    {
+      "epoch": 0.027317640220866028,
+      "grad_norm": 0.7890406847000122,
+      "learning_rate": 9.997432916787756e-06,
+      "loss": 7.8398,
+      "step": 282
+    },
+    {
+      "epoch": 0.027317640220866028,
+      "eval_accuracy": 0.09197820229778327,
+      "eval_loss": 7.78125,
+      "eval_runtime": 239.0276,
+      "eval_samples_per_second": 141.268,
+      "eval_steps_per_second": 2.209,
+      "step": 282
+    },
+    {
+      "epoch": 0.02741451128547903,
+      "grad_norm": 0.8093599677085876,
+      "learning_rate": 9.997423229681295e-06,
+      "loss": 7.8008,
+      "step": 283
+    },
+    {
+      "epoch": 0.02741451128547903,
+      "eval_accuracy": 0.09215345886671297,
+      "eval_loss": 7.77734375,
+      "eval_runtime": 236.4329,
+      "eval_samples_per_second": 142.819,
+      "eval_steps_per_second": 2.233,
+      "step": 283
+    },
+    {
+      "epoch": 0.02751138235009203,
+      "grad_norm": 0.7843554615974426,
+      "learning_rate": 9.997413542574834e-06,
+      "loss": 7.8281,
+      "step": 284
+    },
+    {
+      "epoch": 0.02751138235009203,
+      "eval_accuracy": 0.09223813436458536,
+      "eval_loss": 7.7734375,
+      "eval_runtime": 240.4286,
+      "eval_samples_per_second": 140.445,
+      "eval_steps_per_second": 2.196,
+      "step": 284
+    },
+    {
+      "epoch": 0.027608253414705027,
+      "grad_norm": 0.8685352206230164,
+      "learning_rate": 9.997403855468372e-06,
+      "loss": 7.7852,
+      "step": 285
+    },
+    {
+      "epoch": 0.027608253414705027,
+      "eval_accuracy": 0.09260729058642081,
+      "eval_loss": 7.76953125,
+      "eval_runtime": 240.1204,
+      "eval_samples_per_second": 140.625,
+      "eval_steps_per_second": 2.199,
+      "step": 285
+    },
+    {
+      "epoch": 0.027705124479318027,
+      "grad_norm": 0.810741662979126,
+      "learning_rate": 9.997394168361912e-06,
+      "loss": 7.793,
+      "step": 286
+    },
+    {
+      "epoch": 0.027705124479318027,
+      "eval_accuracy": 0.09294211342689672,
+      "eval_loss": 7.765625,
+      "eval_runtime": 240.2366,
+      "eval_samples_per_second": 140.557,
+      "eval_steps_per_second": 2.198,
+      "step": 286
+    },
+    {
+      "epoch": 0.027801995543931028,
+      "grad_norm": 0.7687424421310425,
+      "learning_rate": 9.99738448125545e-06,
+      "loss": 7.8086,
+      "step": 287
+    },
+    {
+      "epoch": 0.027801995543931028,
+      "eval_accuracy": 0.09305313241299606,
+      "eval_loss": 7.76171875,
+      "eval_runtime": 239.5713,
+      "eval_samples_per_second": 140.948,
+      "eval_steps_per_second": 2.204,
+      "step": 287
+    },
+    {
+      "epoch": 0.02789886660854403,
+      "grad_norm": 0.7878952026367188,
+      "learning_rate": 9.997374794148988e-06,
+      "loss": 7.7812,
+      "step": 288
+    },
+    {
+      "epoch": 0.02789886660854403,
+      "eval_accuracy": 0.09312049647574788,
+      "eval_loss": 7.7578125,
+      "eval_runtime": 240.4397,
+      "eval_samples_per_second": 140.439,
+      "eval_steps_per_second": 2.196,
+      "step": 288
+    },
+    {
+      "epoch": 0.027995737673157026,
+      "grad_norm": 0.7822087407112122,
+      "learning_rate": 9.997365107042528e-06,
+      "loss": 7.793,
+      "step": 289
+    },
+    {
+      "epoch": 0.027995737673157026,
+      "eval_accuracy": 0.09311163811597047,
+      "eval_loss": 7.75390625,
+      "eval_runtime": 239.2769,
+      "eval_samples_per_second": 141.121,
+      "eval_steps_per_second": 2.207,
+      "step": 289
+    },
+    {
+      "epoch": 0.028092608737770027,
+      "grad_norm": 0.8028809428215027,
+      "learning_rate": 9.997355419936066e-06,
+      "loss": 7.7539,
+      "step": 290
+    },
+    {
+      "epoch": 0.028092608737770027,
+      "eval_accuracy": 0.09305567991515429,
+      "eval_loss": 7.75,
+      "eval_runtime": 236.6986,
+      "eval_samples_per_second": 142.658,
+      "eval_steps_per_second": 2.231,
+      "step": 290
+    },
+    {
+      "epoch": 0.028189479802383028,
+      "grad_norm": 0.8122669458389282,
+      "learning_rate": 9.997345732829604e-06,
+      "loss": 7.75,
+      "step": 291
+    },
+    {
+      "epoch": 0.028189479802383028,
+      "eval_accuracy": 0.09302276502931466,
+      "eval_loss": 7.74609375,
+      "eval_runtime": 239.075,
+      "eval_samples_per_second": 141.24,
+      "eval_steps_per_second": 2.209,
+      "step": 291
+    },
+    {
+      "epoch": 0.02828635086699603,
+      "grad_norm": 0.7843779921531677,
+      "learning_rate": 9.997336045723144e-06,
+      "loss": 7.8164,
+      "step": 292
+    },
+    {
+      "epoch": 0.02828635086699603,
+      "eval_accuracy": 0.09302994435357871,
+      "eval_loss": 7.7421875,
+      "eval_runtime": 239.8333,
+      "eval_samples_per_second": 140.794,
+      "eval_steps_per_second": 2.202,
+      "step": 292
+    },
+    {
+      "epoch": 0.02838322193160903,
+      "grad_norm": 0.8030862808227539,
+      "learning_rate": 9.997326358616683e-06,
+      "loss": 7.7539,
+      "step": 293
+    },
+    {
+      "epoch": 0.02838322193160903,
+      "eval_accuracy": 0.09310758527162785,
+      "eval_loss": 7.7421875,
+      "eval_runtime": 238.9644,
+      "eval_samples_per_second": 141.306,
+      "eval_steps_per_second": 2.21,
+      "step": 293
+    },
+    {
+      "epoch": 0.028480092996222027,
+      "grad_norm": 0.7855885624885559,
+      "learning_rate": 9.997316671510221e-06,
+      "loss": 7.8086,
+      "step": 294
+    },
+    {
+      "epoch": 0.028480092996222027,
+      "eval_accuracy": 0.09324769789032951,
+      "eval_loss": 7.73828125,
+      "eval_runtime": 238.9575,
+      "eval_samples_per_second": 141.31,
+      "eval_steps_per_second": 2.21,
+      "step": 294
+    },
+    {
+      "epoch": 0.028576964060835028,
+      "grad_norm": 0.7736480832099915,
+      "learning_rate": 9.99730698440376e-06,
+      "loss": 7.793,
+      "step": 295
+    },
+    {
+      "epoch": 0.028576964060835028,
+      "eval_accuracy": 0.09355316655820965,
+      "eval_loss": 7.734375,
+      "eval_runtime": 239.2168,
+      "eval_samples_per_second": 141.156,
+      "eval_steps_per_second": 2.207,
+      "step": 295
+    },
+    {
+      "epoch": 0.02867383512544803,
+      "grad_norm": 0.7640628814697266,
+      "learning_rate": 9.997297297297297e-06,
+      "loss": 7.7695,
+      "step": 296
+    },
+    {
+      "epoch": 0.02867383512544803,
+      "eval_accuracy": 0.0937199121540199,
+      "eval_loss": 7.73046875,
+      "eval_runtime": 238.9578,
+      "eval_samples_per_second": 141.309,
+      "eval_steps_per_second": 2.21,
+      "step": 296
+    },
+    {
+      "epoch": 0.02877070619006103,
+      "grad_norm": 0.787857174873352,
+      "learning_rate": 9.997287610190837e-06,
+      "loss": 7.75,
+      "step": 297
+    },
+    {
+      "epoch": 0.02877070619006103,
+      "eval_accuracy": 0.09381628300270953,
+      "eval_loss": 7.7265625,
+      "eval_runtime": 235.9881,
+      "eval_samples_per_second": 143.088,
+      "eval_steps_per_second": 2.237,
+      "step": 297
+    },
+    {
+      "epoch": 0.02886757725467403,
+      "grad_norm": 0.7313410639762878,
+      "learning_rate": 9.997277923084375e-06,
+      "loss": 7.7891,
+      "step": 298
+    },
+    {
+      "epoch": 0.02886757725467403,
+      "eval_accuracy": 0.09376585403953219,
+      "eval_loss": 7.72265625,
+      "eval_runtime": 238.897,
+      "eval_samples_per_second": 141.345,
+      "eval_steps_per_second": 2.21,
+      "step": 298
+    },
+    {
+      "epoch": 0.028964448319287028,
+      "grad_norm": 0.7264304757118225,
+      "learning_rate": 9.997268235977913e-06,
+      "loss": 7.7773,
+      "step": 299
+    },
+    {
+      "epoch": 0.028964448319287028,
+      "eval_accuracy": 0.09357499401988342,
+      "eval_loss": 7.71875,
+      "eval_runtime": 239.7653,
+      "eval_samples_per_second": 140.834,
+      "eval_steps_per_second": 2.202,
+      "step": 299
+    },
+    {
+      "epoch": 0.02906131938390003,
+      "grad_norm": 0.8194934725761414,
+      "learning_rate": 9.997258548871453e-06,
+      "loss": 7.7227,
+      "step": 300
+    },
+    {
+      "epoch": 0.02906131938390003,
+      "eval_accuracy": 0.09352896528770664,
+      "eval_loss": 7.71484375,
+      "eval_runtime": 241.2504,
+      "eval_samples_per_second": 139.967,
+      "eval_steps_per_second": 2.189,
+      "step": 300
+    },
+    {
+      "epoch": 0.02915819044851303,
+      "grad_norm": 0.7791144251823425,
+      "learning_rate": 9.997248861764993e-06,
+      "loss": 7.7109,
+      "step": 301
+    },
+    {
+      "epoch": 0.02915819044851303,
+      "eval_accuracy": 0.09365309812014315,
+      "eval_loss": 7.71484375,
+      "eval_runtime": 240.0076,
+      "eval_samples_per_second": 140.691,
+      "eval_steps_per_second": 2.2,
+      "step": 301
+    },
+    {
+      "epoch": 0.02925506151312603,
+      "grad_norm": 0.8002768754959106,
+      "learning_rate": 9.99723917465853e-06,
+      "loss": 7.7148,
+      "step": 302
+    },
+    {
+      "epoch": 0.02925506151312603,
+      "eval_accuracy": 0.09385024004852296,
+      "eval_loss": 7.7109375,
+      "eval_runtime": 238.1093,
+      "eval_samples_per_second": 141.813,
+      "eval_steps_per_second": 2.217,
+      "step": 302
+    },
+    {
+      "epoch": 0.02935193257773903,
+      "grad_norm": 0.7393519282341003,
+      "learning_rate": 9.997229487552069e-06,
+      "loss": 7.7812,
+      "step": 303
+    },
+    {
+      "epoch": 0.02935193257773903,
+      "eval_accuracy": 0.09395888522579307,
+      "eval_loss": 7.70703125,
+      "eval_runtime": 237.8539,
+      "eval_samples_per_second": 141.965,
+      "eval_steps_per_second": 2.22,
+      "step": 303
+    },
+    {
+      "epoch": 0.02944880364235203,
+      "grad_norm": 0.7291700839996338,
+      "learning_rate": 9.997219800445608e-06,
+      "loss": 7.7109,
+      "step": 304
+    },
+    {
+      "epoch": 0.02944880364235203,
+      "eval_accuracy": 0.0940500742235018,
+      "eval_loss": 7.703125,
+      "eval_runtime": 237.0915,
+      "eval_samples_per_second": 142.422,
+      "eval_steps_per_second": 2.227,
+      "step": 304
+    },
+    {
+      "epoch": 0.02954567470696503,
+      "grad_norm": 0.7372825145721436,
+      "learning_rate": 9.997210113339146e-06,
+      "loss": 7.7539,
+      "step": 305
+    },
+    {
+      "epoch": 0.02954567470696503,
+      "eval_accuracy": 0.09420005841306653,
+      "eval_loss": 7.69921875,
+      "eval_runtime": 238.3146,
+      "eval_samples_per_second": 141.691,
+      "eval_steps_per_second": 2.216,
+      "step": 305
+    },
+    {
+      "epoch": 0.02964254577157803,
+      "grad_norm": 0.7321212887763977,
+      "learning_rate": 9.997200426232684e-06,
+      "loss": 7.7734,
+      "step": 306
+    },
+    {
+      "epoch": 0.02964254577157803,
+      "eval_accuracy": 0.09429781880838792,
+      "eval_loss": 7.69921875,
+      "eval_runtime": 241.1284,
+      "eval_samples_per_second": 140.037,
+      "eval_steps_per_second": 2.19,
+      "step": 306
+    },
+    {
+      "epoch": 0.02973941683619103,
+      "grad_norm": 0.733534574508667,
+      "learning_rate": 9.997190739126224e-06,
+      "loss": 7.6914,
+      "step": 307
+    },
+    {
+      "epoch": 0.02973941683619103,
+      "eval_accuracy": 0.09434127108951833,
+      "eval_loss": 7.6953125,
+      "eval_runtime": 238.2743,
+      "eval_samples_per_second": 141.715,
+      "eval_steps_per_second": 2.216,
+      "step": 307
+    },
+    {
+      "epoch": 0.029836287900804028,
+      "grad_norm": 0.7814671993255615,
+      "learning_rate": 9.997181052019762e-06,
+      "loss": 7.6445,
+      "step": 308
+    },
+    {
+      "epoch": 0.029836287900804028,
+      "eval_accuracy": 0.09439731613699899,
+      "eval_loss": 7.69140625,
+      "eval_runtime": 236.2657,
+      "eval_samples_per_second": 142.92,
+      "eval_steps_per_second": 2.235,
+      "step": 308
+    },
+    {
+      "epoch": 0.02993315896541703,
+      "grad_norm": 0.7276116609573364,
+      "learning_rate": 9.997171364913302e-06,
+      "loss": 7.6953,
+      "step": 309
+    },
+    {
+      "epoch": 0.02993315896541703,
+      "eval_accuracy": 0.09446468019975081,
+      "eval_loss": 7.6875,
+      "eval_runtime": 235.6433,
+      "eval_samples_per_second": 143.297,
+      "eval_steps_per_second": 2.241,
+      "step": 309
+    },
+    {
+      "epoch": 0.03003003003003003,
+      "grad_norm": 0.7116464376449585,
+      "learning_rate": 9.99716167780684e-06,
+      "loss": 7.75,
+      "step": 310
+    },
+    {
+      "epoch": 0.03003003003003003,
+      "eval_accuracy": 0.09461590919150648,
+      "eval_loss": 7.68359375,
+      "eval_runtime": 237.3533,
+      "eval_samples_per_second": 142.265,
+      "eval_steps_per_second": 2.225,
+      "step": 310
+    },
+    {
+      "epoch": 0.03012690109464303,
+      "grad_norm": 0.7537593245506287,
+      "learning_rate": 9.99715199070038e-06,
+      "loss": 7.7539,
+      "step": 311
+    },
+    {
+      "epoch": 0.03012690109464303,
+      "eval_accuracy": 0.09491029043522077,
+      "eval_loss": 7.68359375,
+      "eval_runtime": 236.2372,
+      "eval_samples_per_second": 142.937,
+      "eval_steps_per_second": 2.235,
+      "step": 311
+    },
+    {
+      "epoch": 0.03022377215925603,
+      "grad_norm": 0.7783864140510559,
+      "learning_rate": 9.997142303593918e-06,
+      "loss": 7.6953,
+      "step": 312
+    },
+    {
+      "epoch": 0.03022377215925603,
+      "eval_accuracy": 0.0951206620054904,
+      "eval_loss": 7.6796875,
+      "eval_runtime": 235.999,
+      "eval_samples_per_second": 143.081,
+      "eval_steps_per_second": 2.237,
+      "step": 312
+    },
+    {
+      "epoch": 0.03032064322386903,
+      "grad_norm": 0.7092785835266113,
+      "learning_rate": 9.997132616487456e-06,
+      "loss": 7.7188,
+      "step": 313
+    },
+    {
+      "epoch": 0.03032064322386903,
+      "eval_accuracy": 0.09514008670944675,
+      "eval_loss": 7.67578125,
+      "eval_runtime": 236.3642,
+      "eval_samples_per_second": 142.86,
+      "eval_steps_per_second": 2.234,
+      "step": 313
+    },
+    {
+      "epoch": 0.03041751428848203,
+      "grad_norm": 0.7478684782981873,
+      "learning_rate": 9.997122929380995e-06,
+      "loss": 7.6914,
+      "step": 314
+    },
+    {
+      "epoch": 0.03041751428848203,
+      "eval_accuracy": 0.09526714337958758,
+      "eval_loss": 7.671875,
+      "eval_runtime": 235.6142,
+      "eval_samples_per_second": 143.315,
+      "eval_steps_per_second": 2.241,
+      "step": 314
+    },
+    {
+      "epoch": 0.03051438535309503,
+      "grad_norm": 0.7141132354736328,
+      "learning_rate": 9.997113242274533e-06,
+      "loss": 7.7344,
+      "step": 315
+    },
+    {
+      "epoch": 0.03051438535309503,
+      "eval_accuracy": 0.09541362475368477,
+      "eval_loss": 7.671875,
+      "eval_runtime": 237.3782,
+      "eval_samples_per_second": 142.25,
+      "eval_steps_per_second": 2.224,
+      "step": 315
+    },
+    {
+      "epoch": 0.03061125641770803,
+      "grad_norm": 0.677810549736023,
+      "learning_rate": 9.997103555168072e-06,
+      "loss": 7.7383,
+      "step": 316
+    },
+    {
+      "epoch": 0.03061125641770803,
+      "eval_accuracy": 0.0953166749272319,
+      "eval_loss": 7.66796875,
+      "eval_runtime": 237.8956,
+      "eval_samples_per_second": 141.94,
+      "eval_steps_per_second": 2.219,
+      "step": 316
+    },
+    {
+      "epoch": 0.030708127482321032,
+      "grad_norm": 0.6948665976524353,
+      "learning_rate": 9.99709386806161e-06,
+      "loss": 7.6875,
+      "step": 317
+    },
+    {
+      "epoch": 0.030708127482321032,
+      "eval_accuracy": 0.09496401957164852,
+      "eval_loss": 7.6640625,
+      "eval_runtime": 236.596,
+      "eval_samples_per_second": 142.72,
+      "eval_steps_per_second": 2.232,
+      "step": 317
+    },
+    {
+      "epoch": 0.03080499854693403,
+      "grad_norm": 0.7147016525268555,
+      "learning_rate": 9.99708418095515e-06,
+      "loss": 7.6914,
+      "step": 318
+    },
+    {
+      "epoch": 0.03080499854693403,
+      "eval_accuracy": 0.09470200318489878,
+      "eval_loss": 7.66015625,
+      "eval_runtime": 235.4121,
+      "eval_samples_per_second": 143.438,
+      "eval_steps_per_second": 2.243,
+      "step": 318
+    },
+    {
+      "epoch": 0.03090186961154703,
+      "grad_norm": 0.7114992141723633,
+      "learning_rate": 9.997074493848689e-06,
+      "loss": 7.6758,
+      "step": 319
+    },
+    {
+      "epoch": 0.03090186961154703,
+      "eval_accuracy": 0.09453548918019383,
+      "eval_loss": 7.66015625,
+      "eval_runtime": 235.7073,
+      "eval_samples_per_second": 143.258,
+      "eval_steps_per_second": 2.24,
+      "step": 319
+    },
+    {
+      "epoch": 0.03099874067616003,
+      "grad_norm": 0.7744684219360352,
+      "learning_rate": 9.997064806742227e-06,
+      "loss": 7.6836,
+      "step": 320
+    },
+    {
+      "epoch": 0.03099874067616003,
+      "eval_accuracy": 0.0947446159482725,
+      "eval_loss": 7.65625,
+      "eval_runtime": 236.6536,
+      "eval_samples_per_second": 142.685,
+      "eval_steps_per_second": 2.231,
+      "step": 320
+    },
+    {
+      "epoch": 0.031095611740773032,
+      "grad_norm": 0.6939721703529358,
+      "learning_rate": 9.997055119635765e-06,
+      "loss": 7.6914,
+      "step": 321
+    },
+    {
+      "epoch": 0.031095611740773032,
+      "eval_accuracy": 0.09504519225405336,
+      "eval_loss": 7.65234375,
+      "eval_runtime": 236.3373,
+      "eval_samples_per_second": 142.876,
+      "eval_steps_per_second": 2.234,
+      "step": 321
+    },
+    {
+      "epoch": 0.031192482805386033,
+      "grad_norm": 0.7224457859992981,
+      "learning_rate": 9.997045432529305e-06,
+      "loss": 7.6719,
+      "step": 322
+    },
+    {
+      "epoch": 0.031192482805386033,
+      "eval_accuracy": 0.0953886418632014,
+      "eval_loss": 7.65234375,
+      "eval_runtime": 236.1013,
+      "eval_samples_per_second": 143.019,
+      "eval_steps_per_second": 2.236,
+      "step": 322
+    },
+    {
+      "epoch": 0.03128935386999903,
+      "grad_norm": 0.6798779368400574,
+      "learning_rate": 9.997035745422843e-06,
+      "loss": 7.6914,
+      "step": 323
+    },
+    {
+      "epoch": 0.03128935386999903,
+      "eval_accuracy": 0.09575362944514158,
+      "eval_loss": 7.6484375,
+      "eval_runtime": 236.3544,
+      "eval_samples_per_second": 142.866,
+      "eval_steps_per_second": 2.234,
+      "step": 323
+    },
+    {
+      "epoch": 0.03138622493461203,
+      "grad_norm": 0.7235939502716064,
+      "learning_rate": 9.99702605831638e-06,
+      "loss": 7.6094,
+      "step": 324
+    },
+    {
+      "epoch": 0.03138622493461203,
+      "eval_accuracy": 0.09607695957701737,
+      "eval_loss": 7.64453125,
+      "eval_runtime": 236.9857,
+      "eval_samples_per_second": 142.485,
+      "eval_steps_per_second": 2.228,
+      "step": 324
+    },
+    {
+      "epoch": 0.03148309599922503,
+      "grad_norm": 0.6780532002449036,
+      "learning_rate": 9.99701637120992e-06,
+      "loss": 7.7148,
+      "step": 325
+    },
+    {
+      "epoch": 0.03148309599922503,
+      "eval_accuracy": 0.09617480681900324,
+      "eval_loss": 7.640625,
+      "eval_runtime": 236.9965,
+      "eval_samples_per_second": 142.479,
+      "eval_steps_per_second": 2.228,
+      "step": 325
+    },
+    {
+      "epoch": 0.03157996706383803,
+      "grad_norm": 0.7239165306091309,
+      "learning_rate": 9.99700668410346e-06,
+      "loss": 7.6641,
+      "step": 326
+    },
+    {
+      "epoch": 0.03157996706383803,
+      "eval_accuracy": 0.09612828595572771,
+      "eval_loss": 7.640625,
+      "eval_runtime": 236.7231,
+      "eval_samples_per_second": 142.643,
+      "eval_steps_per_second": 2.23,
+      "step": 326
+    },
+    {
+      "epoch": 0.03167683812845103,
+      "grad_norm": 0.6952512860298157,
+      "learning_rate": 9.996996996996998e-06,
+      "loss": 7.6602,
+      "step": 327
+    },
+    {
+      "epoch": 0.03167683812845103,
+      "eval_accuracy": 0.09610260829192846,
+      "eval_loss": 7.63671875,
+      "eval_runtime": 237.3447,
+      "eval_samples_per_second": 142.27,
+      "eval_steps_per_second": 2.225,
+      "step": 327
+    },
+    {
+      "epoch": 0.031773709193064034,
+      "grad_norm": 0.6919338703155518,
+      "learning_rate": 9.996987309890536e-06,
+      "loss": 7.7031,
+      "step": 328
+    },
+    {
+      "epoch": 0.031773709193064034,
+      "eval_accuracy": 0.0962992001914332,
+      "eval_loss": 7.6328125,
+      "eval_runtime": 236.9539,
+      "eval_samples_per_second": 142.505,
+      "eval_steps_per_second": 2.228,
+      "step": 328
+    },
+    {
+      "epoch": 0.031870580257677035,
+      "grad_norm": 0.6624383926391602,
+      "learning_rate": 9.996977622784076e-06,
+      "loss": 7.6953,
+      "step": 329
+    },
+    {
+      "epoch": 0.031870580257677035,
+      "eval_accuracy": 0.09658234926654084,
+      "eval_loss": 7.6328125,
+      "eval_runtime": 237.5409,
+      "eval_samples_per_second": 142.152,
+      "eval_steps_per_second": 2.223,
+      "step": 329
+    },
+    {
+      "epoch": 0.03196745132229003,
+      "grad_norm": 0.6538208723068237,
+      "learning_rate": 9.996967935677614e-06,
+      "loss": 7.6445,
+      "step": 330
+    },
+    {
+      "epoch": 0.03196745132229003,
+      "eval_accuracy": 0.09676490095528725,
+      "eval_loss": 7.62890625,
+      "eval_runtime": 236.9649,
+      "eval_samples_per_second": 142.498,
+      "eval_steps_per_second": 2.228,
+      "step": 330
+    },
+    {
+      "epoch": 0.03206432238690303,
+      "grad_norm": 0.6456217765808105,
+      "learning_rate": 9.996958248571152e-06,
+      "loss": 7.6445,
+      "step": 331
+    },
+    {
+      "epoch": 0.03206432238690303,
+      "eval_accuracy": 0.09686503515943788,
+      "eval_loss": 7.625,
+      "eval_runtime": 236.8552,
+      "eval_samples_per_second": 142.564,
+      "eval_steps_per_second": 2.229,
+      "step": 331
+    },
+    {
+      "epoch": 0.03216119345151603,
+      "grad_norm": 0.6584651470184326,
+      "learning_rate": 9.996948561464692e-06,
+      "loss": 7.6445,
+      "step": 332
+    },
+    {
+      "epoch": 0.03216119345151603,
+      "eval_accuracy": 0.09685333980862064,
+      "eval_loss": 7.625,
+      "eval_runtime": 235.6971,
+      "eval_samples_per_second": 143.264,
+      "eval_steps_per_second": 2.24,
+      "step": 332
+    },
+    {
+      "epoch": 0.03225806451612903,
+      "grad_norm": 0.6445140242576599,
+      "learning_rate": 9.99693887435823e-06,
+      "loss": 7.668,
+      "step": 333
+    },
+    {
+      "epoch": 0.03225806451612903,
+      "eval_accuracy": 0.09675618733995064,
+      "eval_loss": 7.62109375,
+      "eval_runtime": 236.584,
+      "eval_samples_per_second": 142.727,
+      "eval_steps_per_second": 2.232,
+      "step": 333
+    },
+    {
+      "epoch": 0.03235493558074203,
+      "grad_norm": 0.6661810874938965,
+      "learning_rate": 9.996929187251768e-06,
+      "loss": 7.6523,
+      "step": 334
+    },
+    {
+      "epoch": 0.03235493558074203,
+      "eval_accuracy": 0.09667113550653215,
+      "eval_loss": 7.6171875,
+      "eval_runtime": 237.5698,
+      "eval_samples_per_second": 142.135,
+      "eval_steps_per_second": 2.223,
+      "step": 334
+    },
+    {
+      "epoch": 0.03245180664535503,
+      "grad_norm": 0.6913946270942688,
+      "learning_rate": 9.996919500145308e-06,
+      "loss": 7.6602,
+      "step": 335
+    },
+    {
+      "epoch": 0.03245180664535503,
+      "eval_accuracy": 0.0968231461182682,
+      "eval_loss": 7.6171875,
+      "eval_runtime": 236.2729,
+      "eval_samples_per_second": 142.915,
+      "eval_steps_per_second": 2.235,
+      "step": 335
+    },
+    {
+      "epoch": 0.032548677709968034,
+      "grad_norm": 0.6751811504364014,
+      "learning_rate": 9.996909813038846e-06,
+      "loss": 7.6328,
+      "step": 336
+    },
+    {
+      "epoch": 0.032548677709968034,
+      "eval_accuracy": 0.09719745524219639,
+      "eval_loss": 7.61328125,
+      "eval_runtime": 236.6177,
+      "eval_samples_per_second": 142.707,
+      "eval_steps_per_second": 2.231,
+      "step": 336
+    },
+    {
+      "epoch": 0.032645548774581035,
+      "grad_norm": 0.6774271130561829,
+      "learning_rate": 9.996900125932385e-06,
+      "loss": 7.6523,
+      "step": 337
+    },
+    {
+      "epoch": 0.032645548774581035,
+      "eval_accuracy": 0.09763750729113935,
+      "eval_loss": 7.609375,
+      "eval_runtime": 236.1388,
+      "eval_samples_per_second": 142.996,
+      "eval_steps_per_second": 2.236,
+      "step": 337
+    },
+    {
+      "epoch": 0.032742419839194035,
+      "grad_norm": 0.6921724677085876,
+      "learning_rate": 9.996890438825923e-06,
+      "loss": 7.6133,
+      "step": 338
+    },
+    {
+      "epoch": 0.032742419839194035,
+      "eval_accuracy": 0.09806679035368622,
+      "eval_loss": 7.609375,
+      "eval_runtime": 237.6892,
+      "eval_samples_per_second": 142.064,
+      "eval_steps_per_second": 2.221,
+      "step": 338
+    },
+    {
+      "epoch": 0.032839290903807036,
+      "grad_norm": 0.6743577122688293,
+      "learning_rate": 9.996880751719461e-06,
+      "loss": 7.6367,
+      "step": 339
+    },
+    {
+      "epoch": 0.032839290903807036,
+      "eval_accuracy": 0.09835668451973549,
+      "eval_loss": 7.60546875,
+      "eval_runtime": 236.3158,
+      "eval_samples_per_second": 142.889,
+      "eval_steps_per_second": 2.234,
+      "step": 339
+    },
+    {
+      "epoch": 0.03293616196842003,
+      "grad_norm": 0.7336851358413696,
+      "learning_rate": 9.996871064613001e-06,
+      "loss": 7.6641,
+      "step": 340
+    },
+    {
+      "epoch": 0.03293616196842003,
+      "eval_accuracy": 0.09847195899239458,
+      "eval_loss": 7.6015625,
+      "eval_runtime": 235.9339,
+      "eval_samples_per_second": 143.121,
+      "eval_steps_per_second": 2.238,
+      "step": 340
+    },
+    {
+      "epoch": 0.03303303303303303,
+      "grad_norm": 0.6606389880180359,
+      "learning_rate": 9.99686137750654e-06,
+      "loss": 7.6367,
+      "step": 341
+    },
+    {
+      "epoch": 0.03303303303303303,
+      "eval_accuracy": 0.09851552706907764,
+      "eval_loss": 7.6015625,
+      "eval_runtime": 236.0582,
+      "eval_samples_per_second": 143.045,
+      "eval_steps_per_second": 2.237,
+      "step": 341
+    },
+    {
+      "epoch": 0.03312990409764603,
+      "grad_norm": 0.6327222585678101,
+      "learning_rate": 9.996851690400077e-06,
+      "loss": 7.6133,
+      "step": 342
+    },
+    {
+      "epoch": 0.03312990409764603,
+      "eval_accuracy": 0.09849986572058227,
+      "eval_loss": 7.59765625,
+      "eval_runtime": 239.9306,
+      "eval_samples_per_second": 140.737,
+      "eval_steps_per_second": 2.201,
+      "step": 342
+    },
+    {
+      "epoch": 0.03322677516225903,
+      "grad_norm": 0.6791363954544067,
+      "learning_rate": 9.996842003293617e-06,
+      "loss": 7.6016,
+      "step": 343
+    },
+    {
+      "epoch": 0.03322677516225903,
+      "eval_accuracy": 0.09839912358978024,
+      "eval_loss": 7.59765625,
+      "eval_runtime": 238.699,
+      "eval_samples_per_second": 141.463,
+      "eval_steps_per_second": 2.212,
+      "step": 343
+    },
+    {
+      "epoch": 0.033323646226872033,
+      "grad_norm": 0.6186023354530334,
+      "learning_rate": 9.996832316187157e-06,
+      "loss": 7.668,
+      "step": 344
+    },
+    {
+      "epoch": 0.033323646226872033,
+      "eval_accuracy": 0.09836293747957837,
+      "eval_loss": 7.59375,
+      "eval_runtime": 239.8775,
+      "eval_samples_per_second": 140.768,
+      "eval_steps_per_second": 2.201,
+      "step": 344
+    },
+    {
+      "epoch": 0.033420517291485034,
+      "grad_norm": 0.6474988460540771,
+      "learning_rate": 9.996822629080695e-06,
+      "loss": 7.6172,
+      "step": 345
+    },
+    {
+      "epoch": 0.033420517291485034,
+      "eval_accuracy": 0.09842074840923688,
+      "eval_loss": 7.58984375,
+      "eval_runtime": 237.6838,
+      "eval_samples_per_second": 142.067,
+      "eval_steps_per_second": 2.221,
+      "step": 345
+    },
+    {
+      "epoch": 0.033517388356098035,
+      "grad_norm": 0.6356053352355957,
+      "learning_rate": 9.996812941974233e-06,
+      "loss": 7.6016,
+      "step": 346
+    },
+    {
+      "epoch": 0.033517388356098035,
+      "eval_accuracy": 0.09849783929841095,
+      "eval_loss": 7.58984375,
+      "eval_runtime": 236.5606,
+      "eval_samples_per_second": 142.741,
+      "eval_steps_per_second": 2.232,
+      "step": 346
+    },
+    {
+      "epoch": 0.033614259420711036,
+      "grad_norm": 0.6287916898727417,
+      "learning_rate": 9.996803254867772e-06,
+      "loss": 7.6328,
+      "step": 347
+    },
+    {
+      "epoch": 0.033614259420711036,
+      "eval_accuracy": 0.0984772855878163,
+      "eval_loss": 7.5859375,
+      "eval_runtime": 235.9208,
+      "eval_samples_per_second": 143.129,
+      "eval_steps_per_second": 2.238,
+      "step": 347
+    },
+    {
+      "epoch": 0.03371113048532404,
+      "grad_norm": 0.602911114692688,
+      "learning_rate": 9.99679356776131e-06,
+      "loss": 7.668,
+      "step": 348
+    },
+    {
+      "epoch": 0.03371113048532404,
+      "eval_accuracy": 0.09856062943683325,
+      "eval_loss": 7.58203125,
+      "eval_runtime": 235.2916,
+      "eval_samples_per_second": 143.511,
+      "eval_steps_per_second": 2.244,
+      "step": 348
+    },
+    {
+      "epoch": 0.03380800154993703,
+      "grad_norm": 0.6358947157859802,
+      "learning_rate": 9.996783880654848e-06,
+      "loss": 7.6719,
+      "step": 349
+    },
+    {
+      "epoch": 0.03380800154993703,
+      "eval_accuracy": 0.09871625865958947,
+      "eval_loss": 7.58203125,
+      "eval_runtime": 237.9978,
+      "eval_samples_per_second": 141.879,
+      "eval_steps_per_second": 2.219,
+      "step": 349
+    },
+    {
+      "epoch": 0.03390487261455003,
+      "grad_norm": 0.6221932768821716,
+      "learning_rate": 9.996774193548388e-06,
+      "loss": 7.6602,
+      "step": 350
+    },
+    {
+      "epoch": 0.03390487261455003,
+      "eval_accuracy": 0.09892819346981982,
+      "eval_loss": 7.578125,
+      "eval_runtime": 235.489,
+      "eval_samples_per_second": 143.391,
+      "eval_steps_per_second": 2.242,
+      "step": 350
+    },
+    {
+      "epoch": 0.03400174367916303,
+      "grad_norm": 0.6026197075843811,
+      "learning_rate": 9.996764506441926e-06,
+      "loss": 7.6641,
+      "step": 351
+    },
+    {
+      "epoch": 0.03400174367916303,
+      "eval_accuracy": 0.09917735655022585,
+      "eval_loss": 7.57421875,
+      "eval_runtime": 236.3887,
+      "eval_samples_per_second": 142.845,
+      "eval_steps_per_second": 2.234,
+      "step": 351
+    },
+    {
+      "epoch": 0.03409861474377603,
+      "grad_norm": 0.6385203003883362,
+      "learning_rate": 9.996754819335466e-06,
+      "loss": 7.6445,
+      "step": 352
+    },
+    {
+      "epoch": 0.03409861474377603,
+      "eval_accuracy": 0.09936850605875623,
+      "eval_loss": 7.57421875,
+      "eval_runtime": 236.1623,
+      "eval_samples_per_second": 142.982,
+      "eval_steps_per_second": 2.236,
+      "step": 352
+    },
+    {
+      "epoch": 0.034195485808389034,
+      "grad_norm": 0.6434333324432373,
+      "learning_rate": 9.996745132229004e-06,
+      "loss": 7.5781,
+      "step": 353
+    },
+    {
+      "epoch": 0.034195485808389034,
+      "eval_accuracy": 0.09951446635286651,
+      "eval_loss": 7.5703125,
+      "eval_runtime": 236.8614,
+      "eval_samples_per_second": 142.56,
+      "eval_steps_per_second": 2.229,
+      "step": 353
+    },
+    {
+      "epoch": 0.034292356873002035,
+      "grad_norm": 0.588945209980011,
+      "learning_rate": 9.996735445122544e-06,
+      "loss": 7.6523,
+      "step": 354
+    },
+    {
+      "epoch": 0.034292356873002035,
+      "eval_accuracy": 0.09960855023939139,
+      "eval_loss": 7.5703125,
+      "eval_runtime": 238.835,
+      "eval_samples_per_second": 141.382,
+      "eval_steps_per_second": 2.211,
+      "step": 354
+    },
+    {
+      "epoch": 0.034389227937615036,
+      "grad_norm": 0.6065830588340759,
+      "learning_rate": 9.996725758016082e-06,
+      "loss": 7.6562,
+      "step": 355
+    },
+    {
+      "epoch": 0.034389227937615036,
+      "eval_accuracy": 0.09964855760283058,
+      "eval_loss": 7.56640625,
+      "eval_runtime": 237.4072,
+      "eval_samples_per_second": 142.232,
+      "eval_steps_per_second": 2.224,
+      "step": 355
+    },
+    {
+      "epoch": 0.034486099002228036,
+      "grad_norm": 0.6374432444572449,
+      "learning_rate": 9.99671607090962e-06,
+      "loss": 7.5977,
+      "step": 356
+    },
+    {
+      "epoch": 0.034486099002228036,
+      "eval_accuracy": 0.09976296360884482,
+      "eval_loss": 7.56640625,
+      "eval_runtime": 237.2058,
+      "eval_samples_per_second": 142.353,
+      "eval_steps_per_second": 2.226,
+      "step": 356
+    },
+    {
+      "epoch": 0.03458297006684104,
+      "grad_norm": 0.6186094284057617,
+      "learning_rate": 9.996706383803158e-06,
+      "loss": 7.5977,
+      "step": 357
+    },
+    {
+      "epoch": 0.03458297006684104,
+      "eval_accuracy": 0.09979237567921691,
+      "eval_loss": 7.5625,
+      "eval_runtime": 237.4024,
+      "eval_samples_per_second": 142.235,
+      "eval_steps_per_second": 2.224,
+      "step": 357
+    },
+    {
+      "epoch": 0.03467984113145404,
+      "grad_norm": 0.6533841490745544,
+      "learning_rate": 9.996696696696698e-06,
+      "loss": 7.5508,
+      "step": 358
+    },
+    {
+      "epoch": 0.03467984113145404,
+      "eval_accuracy": 0.09973910972499975,
+      "eval_loss": 7.5625,
+      "eval_runtime": 236.1905,
+      "eval_samples_per_second": 142.965,
+      "eval_steps_per_second": 2.235,
+      "step": 358
+    },
+    {
+      "epoch": 0.03477671219606703,
+      "grad_norm": 0.6210293769836426,
+      "learning_rate": 9.996687009590236e-06,
+      "loss": 7.6172,
+      "step": 359
+    },
+    {
+      "epoch": 0.03477671219606703,
+      "eval_accuracy": 0.09971864286106957,
+      "eval_loss": 7.55859375,
+      "eval_runtime": 237.087,
+      "eval_samples_per_second": 142.425,
+      "eval_steps_per_second": 2.227,
+      "step": 359
+    },
+    {
+      "epoch": 0.03487358326068003,
+      "grad_norm": 0.6240441799163818,
+      "learning_rate": 9.996677322483775e-06,
+      "loss": 7.5469,
+      "step": 360
+    },
+    {
+      "epoch": 0.03487358326068003,
+      "eval_accuracy": 0.09969887077045526,
+      "eval_loss": 7.5546875,
+      "eval_runtime": 236.921,
+      "eval_samples_per_second": 142.524,
+      "eval_steps_per_second": 2.229,
+      "step": 360
+    },
+    {
+      "epoch": 0.034970454325293034,
+      "grad_norm": 0.593199610710144,
+      "learning_rate": 9.996667635377313e-06,
+      "loss": 7.6172,
+      "step": 361
+    },
+    {
+      "epoch": 0.034970454325293034,
+      "eval_accuracy": 0.09969548375054037,
+      "eval_loss": 7.5546875,
+      "eval_runtime": 237.3099,
+      "eval_samples_per_second": 142.291,
+      "eval_steps_per_second": 2.225,
+      "step": 361
+    },
+    {
+      "epoch": 0.035067325389906034,
+      "grad_norm": 0.6026429533958435,
+      "learning_rate": 9.996657948270853e-06,
+      "loss": 7.625,
+      "step": 362
+    },
+    {
+      "epoch": 0.035067325389906034,
+      "eval_accuracy": 0.0997558421823571,
+      "eval_loss": 7.55078125,
+      "eval_runtime": 236.3278,
+      "eval_samples_per_second": 142.882,
+      "eval_steps_per_second": 2.234,
+      "step": 362
+    },
+    {
+      "epoch": 0.035164196454519035,
+      "grad_norm": 0.5891857743263245,
+      "learning_rate": 9.996648261164391e-06,
+      "loss": 7.6289,
+      "step": 363
+    },
+    {
+      "epoch": 0.035164196454519035,
+      "eval_accuracy": 0.09990139719203311,
+      "eval_loss": 7.55078125,
+      "eval_runtime": 236.9824,
+      "eval_samples_per_second": 142.487,
+      "eval_steps_per_second": 2.228,
+      "step": 363
+    },
+    {
+      "epoch": 0.035261067519132036,
+      "grad_norm": 0.628919780254364,
+      "learning_rate": 9.996638574057929e-06,
+      "loss": 7.5234,
+      "step": 364
+    },
+    {
+      "epoch": 0.035261067519132036,
+      "eval_accuracy": 0.10019818119346481,
+      "eval_loss": 7.546875,
+      "eval_runtime": 238.3945,
+      "eval_samples_per_second": 141.643,
+      "eval_steps_per_second": 2.215,
+      "step": 364
+    },
+    {
+      "epoch": 0.03535793858374504,
+      "grad_norm": 0.6015390157699585,
+      "learning_rate": 9.996628886951469e-06,
+      "loss": 7.5703,
+      "step": 365
+    },
+    {
+      "epoch": 0.03535793858374504,
+      "eval_accuracy": 0.10060615787432482,
+      "eval_loss": 7.54296875,
+      "eval_runtime": 235.2415,
+      "eval_samples_per_second": 143.542,
+      "eval_steps_per_second": 2.245,
+      "step": 365
+    },
+    {
+      "epoch": 0.03545480964835804,
+      "grad_norm": 0.5987712740898132,
+      "learning_rate": 9.996619199845007e-06,
+      "loss": 7.5859,
+      "step": 366
+    },
+    {
+      "epoch": 0.03545480964835804,
+      "eval_accuracy": 0.1009971705067222,
+      "eval_loss": 7.54296875,
+      "eval_runtime": 237.4998,
+      "eval_samples_per_second": 142.177,
+      "eval_steps_per_second": 2.223,
+      "step": 366
+    },
+    {
+      "epoch": 0.03555168071297104,
+      "grad_norm": 0.6239945888519287,
+      "learning_rate": 9.996609512738545e-06,
+      "loss": 7.5469,
+      "step": 367
+    },
+    {
+      "epoch": 0.03555168071297104,
+      "eval_accuracy": 0.10135602092437215,
+      "eval_loss": 7.5390625,
+      "eval_runtime": 238.6262,
+      "eval_samples_per_second": 141.506,
+      "eval_steps_per_second": 2.213,
+      "step": 367
+    },
+    {
+      "epoch": 0.03564855177758403,
+      "grad_norm": 0.5930135250091553,
+      "learning_rate": 9.996599825632085e-06,
+      "loss": 7.5508,
+      "step": 368
+    },
+    {
+      "epoch": 0.03564855177758403,
+      "eval_accuracy": 0.10160049428489602,
+      "eval_loss": 7.5390625,
+      "eval_runtime": 241.151,
+      "eval_samples_per_second": 140.024,
+      "eval_steps_per_second": 2.189,
+      "step": 368
+    },
+    {
+      "epoch": 0.03574542284219703,
+      "grad_norm": 0.5912957191467285,
+      "learning_rate": 9.996590138525624e-06,
+      "loss": 7.6172,
+      "step": 369
+    },
+    {
+      "epoch": 0.03574542284219703,
+      "eval_accuracy": 0.10172242700183226,
+      "eval_loss": 7.53515625,
+      "eval_runtime": 236.1926,
+      "eval_samples_per_second": 142.964,
+      "eval_steps_per_second": 2.235,
+      "step": 369
+    },
+    {
+      "epoch": 0.035842293906810034,
+      "grad_norm": 0.5632056593894958,
+      "learning_rate": 9.996580451419162e-06,
+      "loss": 7.6172,
+      "step": 370
+    },
+    {
+      "epoch": 0.035842293906810034,
+      "eval_accuracy": 0.1017233823151416,
+      "eval_loss": 7.53515625,
+      "eval_runtime": 236.4678,
+      "eval_samples_per_second": 142.797,
+      "eval_steps_per_second": 2.233,
+      "step": 370
+    },
+    {
+      "epoch": 0.035939164971423035,
+      "grad_norm": 0.5804527401924133,
+      "learning_rate": 9.9965707643127e-06,
+      "loss": 7.5352,
+      "step": 371
+    },
+    {
+      "epoch": 0.035939164971423035,
+      "eval_accuracy": 0.10176124746085684,
+      "eval_loss": 7.53125,
+      "eval_runtime": 236.525,
+      "eval_samples_per_second": 142.763,
+      "eval_steps_per_second": 2.232,
+      "step": 371
+    },
+    {
+      "epoch": 0.036036036036036036,
+      "grad_norm": 0.5636327862739563,
+      "learning_rate": 9.99656107720624e-06,
+      "loss": 7.5859,
+      "step": 372
+    },
+    {
+      "epoch": 0.036036036036036036,
+      "eval_accuracy": 0.10177870364041822,
+      "eval_loss": 7.53125,
+      "eval_runtime": 235.9325,
+      "eval_samples_per_second": 143.121,
+      "eval_steps_per_second": 2.238,
+      "step": 372
+    },
+    {
+      "epoch": 0.03613290710064904,
+      "grad_norm": 0.5637556910514832,
+      "learning_rate": 9.996551390099778e-06,
+      "loss": 7.5586,
+      "step": 373
+    },
+    {
+      "epoch": 0.03613290710064904,
+      "eval_accuracy": 0.10174599139679573,
+      "eval_loss": 7.52734375,
+      "eval_runtime": 237.5048,
+      "eval_samples_per_second": 142.174,
+      "eval_steps_per_second": 2.223,
+      "step": 373
+    },
+    {
+      "epoch": 0.03622977816526204,
+      "grad_norm": 0.5569087862968445,
+      "learning_rate": 9.996541702993316e-06,
+      "loss": 7.6406,
+      "step": 374
+    },
+    {
+      "epoch": 0.03622977816526204,
+      "eval_accuracy": 0.10170641826667895,
+      "eval_loss": 7.52734375,
+      "eval_runtime": 238.2389,
+      "eval_samples_per_second": 141.736,
+      "eval_steps_per_second": 2.216,
+      "step": 374
+    },
+    {
+      "epoch": 0.03632664922987504,
+      "grad_norm": 0.5960782766342163,
+      "learning_rate": 9.996532015886856e-06,
+      "loss": 7.5273,
+      "step": 375
+    },
+    {
+      "epoch": 0.03632664922987504,
+      "eval_accuracy": 0.10178799623351806,
+      "eval_loss": 7.5234375,
+      "eval_runtime": 236.2956,
+      "eval_samples_per_second": 142.902,
+      "eval_steps_per_second": 2.234,
+      "step": 375
+    },
+    {
+      "epoch": 0.03642352029448804,
+      "grad_norm": 0.5876879096031189,
+      "learning_rate": 9.996522328780394e-06,
+      "loss": 7.5312,
+      "step": 376
+    },
+    {
+      "epoch": 0.03642352029448804,
+      "eval_accuracy": 0.10199868624155746,
+      "eval_loss": 7.51953125,
+      "eval_runtime": 236.7211,
+      "eval_samples_per_second": 142.645,
+      "eval_steps_per_second": 2.23,
+      "step": 376
+    },
+    {
+      "epoch": 0.03652039135910104,
+      "grad_norm": 0.5734599828720093,
+      "learning_rate": 9.996512641673934e-06,
+      "loss": 7.5898,
+      "step": 377
+    },
+    {
+      "epoch": 0.03652039135910104,
+      "eval_accuracy": 0.10233654871529031,
+      "eval_loss": 7.51953125,
+      "eval_runtime": 237.8605,
+      "eval_samples_per_second": 141.961,
+      "eval_steps_per_second": 2.22,
+      "step": 377
+    },
+    {
+      "epoch": 0.036617262423714034,
+      "grad_norm": 0.6157567501068115,
+      "learning_rate": 9.996502954567472e-06,
+      "loss": 7.5898,
+      "step": 378
+    },
+    {
+      "epoch": 0.036617262423714034,
+      "eval_accuracy": 0.10266948987803573,
+      "eval_loss": 7.515625,
+      "eval_runtime": 236.7976,
+      "eval_samples_per_second": 142.599,
+      "eval_steps_per_second": 2.23,
+      "step": 378
+    },
+    {
+      "epoch": 0.036714133488327035,
+      "grad_norm": 0.5708773136138916,
+      "learning_rate": 9.99649326746101e-06,
+      "loss": 7.543,
+      "step": 379
+    },
+    {
+      "epoch": 0.036714133488327035,
+      "eval_accuracy": 0.10289378586351103,
+      "eval_loss": 7.515625,
+      "eval_runtime": 236.0204,
+      "eval_samples_per_second": 143.068,
+      "eval_steps_per_second": 2.237,
+      "step": 379
+    },
+    {
+      "epoch": 0.036811004552940035,
+      "grad_norm": 0.5848063230514526,
+      "learning_rate": 9.99648358035455e-06,
+      "loss": 7.5156,
+      "step": 380
+    },
+    {
+      "epoch": 0.036811004552940035,
+      "eval_accuracy": 0.10301302633384825,
+      "eval_loss": 7.51171875,
+      "eval_runtime": 237.3865,
+      "eval_samples_per_second": 142.245,
+      "eval_steps_per_second": 2.224,
+      "step": 380
+    },
+    {
+      "epoch": 0.036907875617553036,
+      "grad_norm": 0.5812904238700867,
+      "learning_rate": 9.996473893248087e-06,
+      "loss": 7.5664,
+      "step": 381
+    },
+    {
+      "epoch": 0.036907875617553036,
+      "eval_accuracy": 0.1031038968937872,
+      "eval_loss": 7.51171875,
+      "eval_runtime": 236.029,
+      "eval_samples_per_second": 143.063,
+      "eval_steps_per_second": 2.237,
+      "step": 381
+    },
+    {
+      "epoch": 0.03700474668216604,
+      "grad_norm": 0.5508126020431519,
+      "learning_rate": 9.996464206141625e-06,
+      "loss": 7.5625,
+      "step": 382
+    },
+    {
+      "epoch": 0.03700474668216604,
+      "eval_accuracy": 0.10313258524195525,
+      "eval_loss": 7.5078125,
+      "eval_runtime": 239.8272,
+      "eval_samples_per_second": 140.797,
+      "eval_steps_per_second": 2.202,
+      "step": 382
+    },
+    {
+      "epoch": 0.03710161774677904,
+      "grad_norm": 0.5484282374382019,
+      "learning_rate": 9.996454519035165e-06,
+      "loss": 7.5312,
+      "step": 383
+    },
+    {
+      "epoch": 0.03710161774677904,
+      "eval_accuracy": 0.10315090988816147,
+      "eval_loss": 7.5078125,
+      "eval_runtime": 237.933,
+      "eval_samples_per_second": 141.918,
+      "eval_steps_per_second": 2.219,
+      "step": 383
+    },
+    {
+      "epoch": 0.03719848881139204,
+      "grad_norm": 0.5489575266838074,
+      "learning_rate": 9.996444831928703e-06,
+      "loss": 7.625,
+      "step": 384
+    },
+    {
+      "epoch": 0.03719848881139204,
+      "eval_accuracy": 0.10320996562001093,
+      "eval_loss": 7.5078125,
+      "eval_runtime": 238.4765,
+      "eval_samples_per_second": 141.595,
+      "eval_steps_per_second": 2.214,
+      "step": 384
+    },
+    {
+      "epoch": 0.03729535987600504,
+      "grad_norm": 0.5515839457511902,
+      "learning_rate": 9.996435144822241e-06,
+      "loss": 7.5898,
+      "step": 385
+    },
+    {
+      "epoch": 0.03729535987600504,
+      "eval_accuracy": 0.1033970622841987,
+      "eval_loss": 7.50390625,
+      "eval_runtime": 237.4396,
+      "eval_samples_per_second": 142.213,
+      "eval_steps_per_second": 2.224,
+      "step": 385
+    },
+    {
+      "epoch": 0.03739223094061804,
+      "grad_norm": 0.5489226579666138,
+      "learning_rate": 9.996425457715781e-06,
+      "loss": 7.5625,
+      "step": 386
+    },
+    {
+      "epoch": 0.03739223094061804,
+      "eval_accuracy": 0.10353740649400565,
+      "eval_loss": 7.5,
+      "eval_runtime": 235.8228,
+      "eval_samples_per_second": 143.188,
+      "eval_steps_per_second": 2.239,
+      "step": 386
+    },
+    {
+      "epoch": 0.037489102005231034,
+      "grad_norm": 0.5816370248794556,
+      "learning_rate": 9.99641577060932e-06,
+      "loss": 7.5664,
+      "step": 387
+    },
+    {
+      "epoch": 0.037489102005231034,
+      "eval_accuracy": 0.10371405260956713,
+      "eval_loss": 7.5,
+      "eval_runtime": 236.7279,
+      "eval_samples_per_second": 142.641,
+      "eval_steps_per_second": 2.23,
+      "step": 387
+    },
+    {
+      "epoch": 0.037585973069844035,
+      "grad_norm": 0.6061241030693054,
+      "learning_rate": 9.996406083502859e-06,
+      "loss": 7.4609,
+      "step": 388
+    },
+    {
+      "epoch": 0.037585973069844035,
+      "eval_accuracy": 0.10385063346391309,
+      "eval_loss": 7.49609375,
+      "eval_runtime": 236.382,
+      "eval_samples_per_second": 142.849,
+      "eval_steps_per_second": 2.234,
+      "step": 388
+    },
+    {
+      "epoch": 0.037682844134457036,
+      "grad_norm": 0.5538076758384705,
+      "learning_rate": 9.996396396396397e-06,
+      "loss": 7.5469,
+      "step": 389
+    },
+    {
+      "epoch": 0.037682844134457036,
+      "eval_accuracy": 0.10396234722332831,
+      "eval_loss": 7.49609375,
+      "eval_runtime": 236.2463,
+      "eval_samples_per_second": 142.931,
+      "eval_steps_per_second": 2.235,
+      "step": 389
+    },
+    {
+      "epoch": 0.03777971519907004,
+      "grad_norm": 0.5397905707359314,
+      "learning_rate": 9.996386709289936e-06,
+      "loss": 7.5742,
+      "step": 390
+    },
+    {
+      "epoch": 0.03777971519907004,
+      "eval_accuracy": 0.10398232195615975,
+      "eval_loss": 7.4921875,
+      "eval_runtime": 236.8546,
+      "eval_samples_per_second": 142.564,
+      "eval_steps_per_second": 2.229,
+      "step": 390
+    },
+    {
+      "epoch": 0.03787658626368304,
+      "grad_norm": 0.5839679837226868,
+      "learning_rate": 9.996377022183475e-06,
+      "loss": 7.4375,
+      "step": 391
+    },
+    {
+      "epoch": 0.03787658626368304,
+      "eval_accuracy": 0.10396709484098679,
+      "eval_loss": 7.4921875,
+      "eval_runtime": 236.6285,
+      "eval_samples_per_second": 142.7,
+      "eval_steps_per_second": 2.231,
+      "step": 391
+    },
+    {
+      "epoch": 0.03797345732829604,
+      "grad_norm": 0.5725728869438171,
+      "learning_rate": 9.996367335077013e-06,
+      "loss": 7.4961,
+      "step": 392
+    },
+    {
+      "epoch": 0.03797345732829604,
+      "eval_accuracy": 0.10393614847954215,
+      "eval_loss": 7.48828125,
+      "eval_runtime": 237.6574,
+      "eval_samples_per_second": 142.083,
+      "eval_steps_per_second": 2.222,
+      "step": 392
+    },
+    {
+      "epoch": 0.03797345732829604,
+      "step": 392,
+      "total_flos": 2185295466332160.0,
+      "train_loss": 8.552704480229592,
+      "train_runtime": 93913.2173,
+      "train_samples_per_second": 703.461,
+      "train_steps_per_second": 10.992
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1032300,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 100,
+  "save_steps": 1000000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2185295466332160.0,
+  "train_batch_size": 64,
+  "trial_name": null,
+  "trial_params": null
+}