add newer models

with_input/1b_52k/best_model/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1f205dca0a65bf622caa6d439f4ff02151cac0a5c4ba8a1d3f4d6aad08d17898
+oid sha256:3180dc69e81027d5923f12d04cfe14e43dd8e5a9054cbf3e96394a6367507167
 size 67201357

with_input/1b_52k/best_model/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1b5e6e20b1087f6644e6895c74078975f305a563890b6467d0cc6bcf54f2475a
+oid sha256:68246debaa80f39fb978b925e1b301396a62e3cad3a43aace0d78955cccb9f49
 size 134433093

with_input/1b_52k/best_model/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d56b42bf9cdca4b0cc0311bae98becd374f3c91f842129455b13c79b24023f56
+oid sha256:7abed5479a9ca4c9ea8664980b3819aa15f0253b2cd93603bf403f8cc838445b
 size 14575

with_input/1b_52k/best_model/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:abfed23e6d6d347e99d70b8c08d200a84418499ce9da89b5fe35ff4e4e832571
+oid sha256:6b2d3080c8e2e4ab287873ffeaf2b06cd1df976c27dba320c4b0ac6a28dd37d5
 size 627

with_input/1b_52k/best_model/trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.4508701264858246,
+  "best_metric": 0.44935256242752075,
   "best_model_checkpoint": "checkpoints/instrucode/with_input/1b_52k/checkpoint-4600",
   "epoch": 2.958199356913183,
   "eval_steps": 200,
@@ -11,2945 +11,2945 @@
     {
       "epoch": 0.01,
       "learning_rate": 2.6999999999999996e-05,
-      "loss": 1.2405,
+      "loss": 1.27,
       "step": 10
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.6999999999999996e-05,
-      "loss": 1.2308,
+      "loss": 1.2423,
       "step": 20
     },
     {
       "epoch": 0.02,
       "learning_rate": 8.699999999999999e-05,
-      "loss": 1.0636,
+      "loss": 1.0757,
       "step": 30
     },
     {
       "epoch": 0.03,
       "learning_rate": 0.000117,
-      "loss": 0.8083,
+      "loss": 0.8238,
       "step": 40
     },
     {
       "epoch": 0.03,
       "learning_rate": 0.000147,
-      "loss": 0.7358,
+      "loss": 0.7379,
       "step": 50
     },
     {
       "epoch": 0.04,
       "learning_rate": 0.00017699999999999997,
-      "loss": 0.6593,
+      "loss": 0.6487,
       "step": 60
     },
     {
       "epoch": 0.05,
       "learning_rate": 0.00020699999999999996,
-      "loss": 0.557,
+      "loss": 0.5623,
       "step": 70
     },
     {
       "epoch": 0.05,
       "learning_rate": 0.000237,
-      "loss": 0.5436,
+      "loss": 0.5591,
       "step": 80
     },
     {
       "epoch": 0.06,
       "learning_rate": 0.000267,
-      "loss": 0.5564,
+      "loss": 0.5358,
       "step": 90
     },
     {
       "epoch": 0.06,
       "learning_rate": 0.00029699999999999996,
-      "loss": 0.5154,
+      "loss": 0.5251,
       "step": 100
     },
     {
       "epoch": 0.07,
       "learning_rate": 0.00029964820846905535,
-      "loss": 0.5257,
+      "loss": 0.5371,
       "step": 110
     },
     {
       "epoch": 0.08,
       "learning_rate": 0.000299257328990228,
-      "loss": 0.5174,
+      "loss": 0.5192,
       "step": 120
     },
     {
       "epoch": 0.08,
       "learning_rate": 0.00029886644951140066,
-      "loss": 0.5303,
+      "loss": 0.4993,
       "step": 130
     },
     {
       "epoch": 0.09,
       "learning_rate": 0.00029847557003257326,
-      "loss": 0.5081,
+      "loss": 0.4993,
       "step": 140
     },
     {
       "epoch": 0.1,
       "learning_rate": 0.0002980846905537459,
-      "loss": 0.5031,
+      "loss": 0.5304,
       "step": 150
     },
     {
       "epoch": 0.1,
       "learning_rate": 0.00029769381107491857,
-      "loss": 0.4832,
+      "loss": 0.5094,
       "step": 160
     },
     {
       "epoch": 0.11,
       "learning_rate": 0.00029730293159609117,
-      "loss": 0.4949,
+      "loss": 0.492,
       "step": 170
     },
     {
       "epoch": 0.12,
       "learning_rate": 0.0002969120521172638,
-      "loss": 0.5122,
+      "loss": 0.5123,
       "step": 180
     },
     {
       "epoch": 0.12,
       "learning_rate": 0.0002965211726384364,
-      "loss": 0.5149,
+      "loss": 0.5018,
       "step": 190
     },
     {
       "epoch": 0.13,
       "learning_rate": 0.0002961302931596091,
-      "loss": 0.4987,
+      "loss": 0.4899,
       "step": 200
     },
     {
       "epoch": 0.13,
-      "eval_loss": 0.49822479486465454,
-      "eval_runtime": 389.6569,
-      "eval_samples_per_second": 5.133,
-      "eval_steps_per_second": 0.642,
+      "eval_loss": 0.49921754002571106,
+      "eval_runtime": 220.0251,
+      "eval_samples_per_second": 9.09,
+      "eval_steps_per_second": 1.136,
       "step": 200
     },
     {
       "epoch": 0.14,
       "learning_rate": 0.00029573941368078174,
-      "loss": 0.5059,
+      "loss": 0.5064,
       "step": 210
     },
     {
       "epoch": 0.14,
       "learning_rate": 0.0002953485342019544,
-      "loss": 0.5117,
+      "loss": 0.4656,
       "step": 220
     },
     {
       "epoch": 0.15,
       "learning_rate": 0.00029495765472312705,
-      "loss": 0.4844,
+      "loss": 0.4961,
       "step": 230
     },
     {
       "epoch": 0.15,
       "learning_rate": 0.00029456677524429965,
-      "loss": 0.4967,
+      "loss": 0.4909,
       "step": 240
     },
     {
       "epoch": 0.16,
       "learning_rate": 0.0002941758957654723,
-      "loss": 0.4805,
+      "loss": 0.484,
       "step": 250
     },
     {
       "epoch": 0.17,
       "learning_rate": 0.0002937850162866449,
-      "loss": 0.4713,
+      "loss": 0.5019,
       "step": 260
     },
     {
       "epoch": 0.17,
       "learning_rate": 0.00029339413680781756,
-      "loss": 0.4857,
+      "loss": 0.4887,
       "step": 270
     },
     {
       "epoch": 0.18,
       "learning_rate": 0.0002930032573289902,
-      "loss": 0.4935,
+      "loss": 0.4761,
       "step": 280
     },
     {
       "epoch": 0.19,
       "learning_rate": 0.0002926123778501628,
-      "loss": 0.4827,
+      "loss": 0.4874,
       "step": 290
     },
     {
       "epoch": 0.19,
       "learning_rate": 0.00029222149837133547,
-      "loss": 0.4954,
+      "loss": 0.479,
       "step": 300
     },
     {
       "epoch": 0.2,
       "learning_rate": 0.0002918306188925081,
-      "loss": 0.4902,
+      "loss": 0.5078,
       "step": 310
     },
     {
       "epoch": 0.21,
       "learning_rate": 0.0002914397394136808,
-      "loss": 0.4707,
+      "loss": 0.4875,
       "step": 320
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.0002910488599348534,
-      "loss": 0.4909,
+      "learning_rate": 0.00029108794788273616,
+      "loss": 0.4858,
       "step": 330
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.00029065798045602604,
-      "loss": 0.4763,
+      "learning_rate": 0.0002906970684039088,
+      "loss": 0.4633,
       "step": 340
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.00029026710097719864,
-      "loss": 0.4896,
+      "learning_rate": 0.0002903061889250814,
+      "loss": 0.4804,
       "step": 350
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.0002898762214983713,
-      "loss": 0.4778,
+      "learning_rate": 0.00028991530944625407,
+      "loss": 0.4981,
       "step": 360
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.00028948534201954395,
-      "loss": 0.483,
+      "learning_rate": 0.00028952442996742667,
+      "loss": 0.4784,
       "step": 370
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.0002890944625407166,
-      "loss": 0.4811,
+      "learning_rate": 0.0002891335504885993,
+      "loss": 0.4923,
       "step": 380
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.0002887035830618892,
-      "loss": 0.4908,
+      "learning_rate": 0.000288742671009772,
+      "loss": 0.4892,
       "step": 390
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.00028831270358306186,
-      "loss": 0.478,
+      "learning_rate": 0.0002883517915309446,
+      "loss": 0.4902,
       "step": 400
     },
     {
       "epoch": 0.26,
-      "eval_loss": 0.48327910900115967,
-      "eval_runtime": 403.39,
-      "eval_samples_per_second": 4.958,
-      "eval_steps_per_second": 0.62,
+      "eval_loss": 0.48341748118400574,
+      "eval_runtime": 220.2998,
+      "eval_samples_per_second": 9.079,
+      "eval_steps_per_second": 1.135,
       "step": 400
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0002879218241042345,
-      "loss": 0.4708,
+      "learning_rate": 0.00028796091205211723,
+      "loss": 0.4725,
       "step": 410
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.0002875309446254071,
-      "loss": 0.4814,
+      "learning_rate": 0.0002875700325732899,
+      "loss": 0.4873,
       "step": 420
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.0002871400651465798,
-      "loss": 0.4852,
+      "learning_rate": 0.00028717915309446255,
+      "loss": 0.483,
       "step": 430
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.00028674918566775243,
-      "loss": 0.4682,
+      "learning_rate": 0.00028678827361563515,
+      "loss": 0.4587,
       "step": 440
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.00028635830618892503,
-      "loss": 0.4594,
+      "learning_rate": 0.0002863973941368078,
+      "loss": 0.4735,
       "step": 450
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0002859674267100977,
-      "loss": 0.4759,
+      "learning_rate": 0.0002860065146579804,
+      "loss": 0.5109,
       "step": 460
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.00028557654723127034,
-      "loss": 0.4779,
+      "learning_rate": 0.00028561563517915306,
+      "loss": 0.478,
       "step": 470
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.000285185667752443,
-      "loss": 0.4739,
+      "learning_rate": 0.0002852247557003257,
+      "loss": 0.4645,
       "step": 480
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.00028479478827361565,
-      "loss": 0.4698,
+      "learning_rate": 0.00028483387622149837,
+      "loss": 0.4932,
       "step": 490
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.00028440390879478825,
-      "loss": 0.4819,
+      "learning_rate": 0.00028444299674267097,
+      "loss": 0.4663,
       "step": 500
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0002840130293159609,
-      "loss": 0.4768,
+      "learning_rate": 0.0002840521172638436,
+      "loss": 0.478,
       "step": 510
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0002836221498371335,
-      "loss": 0.4705,
+      "learning_rate": 0.0002836612377850163,
+      "loss": 0.4567,
       "step": 520
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.00028323127035830616,
-      "loss": 0.4755,
+      "learning_rate": 0.0002832703583061889,
+      "loss": 0.4811,
       "step": 530
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0002828403908794788,
-      "loss": 0.4686,
+      "learning_rate": 0.00028287947882736154,
+      "loss": 0.4678,
       "step": 540
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0002824495114006514,
-      "loss": 0.4745,
+      "learning_rate": 0.00028248859934853414,
+      "loss": 0.4714,
       "step": 550
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0002820586319218241,
-      "loss": 0.4889,
+      "learning_rate": 0.0002820977198697068,
+      "loss": 0.4714,
       "step": 560
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.00028166775244299673,
-      "loss": 0.4696,
+      "learning_rate": 0.00028170684039087945,
+      "loss": 0.4583,
       "step": 570
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0002812768729641694,
-      "loss": 0.4896,
+      "learning_rate": 0.0002813159609120521,
+      "loss": 0.4493,
       "step": 580
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.000280885993485342,
-      "loss": 0.4772,
+      "learning_rate": 0.00028092508143322476,
+      "loss": 0.4551,
       "step": 590
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.00028049511400651464,
-      "loss": 0.4824,
+      "learning_rate": 0.00028053420195439736,
+      "loss": 0.5053,
       "step": 600
     },
     {
       "epoch": 0.39,
-      "eval_loss": 0.47541937232017517,
-      "eval_runtime": 392.9458,
-      "eval_samples_per_second": 5.09,
-      "eval_steps_per_second": 0.636,
+      "eval_loss": 0.4742714762687683,
+      "eval_runtime": 220.0995,
+      "eval_samples_per_second": 9.087,
+      "eval_steps_per_second": 1.136,
       "step": 600
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.00028010423452768724,
-      "loss": 0.481,
+      "learning_rate": 0.00028014332247557,
+      "loss": 0.4556,
       "step": 610
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0002797133550488599,
-      "loss": 0.4695,
+      "learning_rate": 0.00027975244299674267,
+      "loss": 0.4819,
       "step": 620
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.00027932247557003255,
-      "loss": 0.4725,
+      "learning_rate": 0.00027936156351791527,
+      "loss": 0.4729,
       "step": 630
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0002789315960912052,
-      "loss": 0.4688,
+      "learning_rate": 0.0002789706840390879,
+      "loss": 0.4798,
       "step": 640
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0002785407166123778,
-      "loss": 0.4782,
+      "learning_rate": 0.0002785798045602606,
+      "loss": 0.4804,
       "step": 650
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.00027814983713355046,
-      "loss": 0.4775,
+      "learning_rate": 0.0002781889250814332,
+      "loss": 0.4661,
       "step": 660
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0002777589576547231,
-      "loss": 0.4702,
+      "learning_rate": 0.00027779804560260584,
+      "loss": 0.465,
       "step": 670
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0002773680781758957,
-      "loss": 0.4606,
+      "learning_rate": 0.0002774071661237785,
+      "loss": 0.4698,
       "step": 680
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0002769771986970684,
-      "loss": 0.4625,
+      "learning_rate": 0.00027701628664495115,
+      "loss": 0.4824,
       "step": 690
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.000276586319218241,
-      "loss": 0.4768,
+      "learning_rate": 0.00027662540716612375,
+      "loss": 0.4575,
       "step": 700
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.00027619543973941363,
-      "loss": 0.4695,
+      "learning_rate": 0.0002762345276872964,
+      "loss": 0.4747,
       "step": 710
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0002758045602605863,
-      "loss": 0.472,
+      "learning_rate": 0.000275843648208469,
+      "loss": 0.461,
       "step": 720
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.00027541368078175894,
-      "loss": 0.4786,
+      "learning_rate": 0.00027545276872964166,
+      "loss": 0.4588,
       "step": 730
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0002750228013029316,
-      "loss": 0.4832,
+      "learning_rate": 0.0002750618892508143,
+      "loss": 0.464,
       "step": 740
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0002746319218241042,
-      "loss": 0.4788,
+      "learning_rate": 0.00027467100977198697,
+      "loss": 0.4829,
       "step": 750
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.00027424104234527685,
-      "loss": 0.465,
+      "learning_rate": 0.00027428013029315957,
+      "loss": 0.4703,
       "step": 760
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0002738501628664495,
-      "loss": 0.4868,
+      "learning_rate": 0.00027388925081433223,
+      "loss": 0.4697,
       "step": 770
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0002734592833876221,
-      "loss": 0.4659,
+      "learning_rate": 0.0002734983713355049,
+      "loss": 0.4774,
       "step": 780
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.00027306840390879477,
-      "loss": 0.4667,
+      "learning_rate": 0.0002731074918566775,
+      "loss": 0.467,
       "step": 790
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.00027267752442996737,
-      "loss": 0.4643,
+      "learning_rate": 0.00027271661237785014,
+      "loss": 0.4924,
       "step": 800
     },
     {
       "epoch": 0.51,
-      "eval_loss": 0.4702647626399994,
-      "eval_runtime": 395.8141,
-      "eval_samples_per_second": 5.053,
-      "eval_steps_per_second": 0.632,
+      "eval_loss": 0.4694923460483551,
+      "eval_runtime": 220.2001,
+      "eval_samples_per_second": 9.083,
+      "eval_steps_per_second": 1.135,
       "step": 800
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.00027228664495114,
-      "loss": 0.4593,
+      "learning_rate": 0.00027232573289902274,
+      "loss": 0.4788,
       "step": 810
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0002718957654723127,
-      "loss": 0.4602,
+      "learning_rate": 0.0002719348534201954,
+      "loss": 0.4635,
       "step": 820
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.00027150488599348533,
-      "loss": 0.4663,
+      "learning_rate": 0.00027154397394136805,
+      "loss": 0.4816,
       "step": 830
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.000271114006514658,
-      "loss": 0.4812,
+      "learning_rate": 0.0002711530944625407,
+      "loss": 0.4582,
       "step": 840
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0002707231270358306,
-      "loss": 0.4682,
+      "learning_rate": 0.00027076221498371336,
+      "loss": 0.4541,
       "step": 850
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.00027033224755700324,
-      "loss": 0.4506,
+      "learning_rate": 0.00027037133550488596,
+      "loss": 0.4748,
       "step": 860
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.00026994136807817584,
-      "loss": 0.4786,
+      "learning_rate": 0.0002699804560260586,
+      "loss": 0.4762,
       "step": 870
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0002695504885993485,
-      "loss": 0.4516,
+      "learning_rate": 0.0002695895765472312,
+      "loss": 0.4678,
       "step": 880
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.00026915960912052116,
-      "loss": 0.4591,
+      "learning_rate": 0.0002691986970684039,
+      "loss": 0.4581,
       "step": 890
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0002687687296416938,
-      "loss": 0.4581,
+      "learning_rate": 0.00026880781758957653,
+      "loss": 0.4439,
       "step": 900
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0002683778501628664,
-      "loss": 0.4509,
+      "learning_rate": 0.00026841693811074913,
+      "loss": 0.4881,
       "step": 910
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.00026798697068403907,
-      "loss": 0.475,
+      "learning_rate": 0.0002680260586319218,
+      "loss": 0.4636,
       "step": 920
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0002675960912052117,
-      "loss": 0.4828,
+      "learning_rate": 0.00026763517915309444,
+      "loss": 0.4649,
       "step": 930
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0002672052117263843,
-      "loss": 0.4663,
+      "learning_rate": 0.0002672442996742671,
+      "loss": 0.4769,
       "step": 940
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.000266814332247557,
-      "loss": 0.4767,
+      "learning_rate": 0.00026685342019543975,
+      "loss": 0.4568,
       "step": 950
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0002664234527687296,
-      "loss": 0.4633,
+      "learning_rate": 0.00026646254071661235,
+      "loss": 0.4673,
       "step": 960
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.00026603257328990223,
-      "loss": 0.4877,
+      "learning_rate": 0.000266071661237785,
+      "loss": 0.4485,
       "step": 970
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0002656416938110749,
-      "loss": 0.4529,
+      "learning_rate": 0.0002656807817589576,
+      "loss": 0.471,
       "step": 980
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.00026525081433224755,
-      "loss": 0.4742,
+      "learning_rate": 0.00026528990228013026,
+      "loss": 0.4631,
       "step": 990
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0002648599348534202,
-      "loss": 0.468,
+      "learning_rate": 0.0002648990228013029,
+      "loss": 0.4724,
       "step": 1000
     },
     {
       "epoch": 0.64,
-      "eval_loss": 0.46662116050720215,
-      "eval_runtime": 395.1813,
-      "eval_samples_per_second": 5.061,
-      "eval_steps_per_second": 0.633,
+      "eval_loss": 0.46559515595436096,
+      "eval_runtime": 220.2191,
+      "eval_samples_per_second": 9.082,
+      "eval_steps_per_second": 1.135,
       "step": 1000
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0002644690553745928,
-      "loss": 0.4739,
+      "learning_rate": 0.0002645081433224756,
+      "loss": 0.465,
       "step": 1010
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.00026407817589576546,
-      "loss": 0.4653,
+      "learning_rate": 0.0002641172638436482,
+      "loss": 0.4504,
       "step": 1020
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0002636872964169381,
-      "loss": 0.448,
+      "learning_rate": 0.00026372638436482083,
+      "loss": 0.4583,
       "step": 1030
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0002632964169381107,
-      "loss": 0.4752,
+      "learning_rate": 0.0002633355048859935,
+      "loss": 0.471,
       "step": 1040
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.00026290553745928337,
-      "loss": 0.4625,
+      "learning_rate": 0.0002629446254071661,
+      "loss": 0.4779,
       "step": 1050
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.00026251465798045597,
-      "loss": 0.4717,
+      "learning_rate": 0.00026255374592833874,
+      "loss": 0.4543,
       "step": 1060
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0002621237785016286,
-      "loss": 0.4715,
+      "learning_rate": 0.00026216286644951134,
+      "loss": 0.4629,
       "step": 1070
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0002617328990228013,
-      "loss": 0.4846,
+      "learning_rate": 0.000261771986970684,
+      "loss": 0.45,
       "step": 1080
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.00026134201954397393,
-      "loss": 0.4464,
+      "learning_rate": 0.00026138110749185665,
+      "loss": 0.4493,
       "step": 1090
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0002609511400651466,
-      "loss": 0.465,
+      "learning_rate": 0.0002609902280130293,
+      "loss": 0.4695,
       "step": 1100
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0002605602605863192,
-      "loss": 0.4637,
+      "learning_rate": 0.00026059934853420196,
+      "loss": 0.4538,
       "step": 1110
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.00026016938110749185,
-      "loss": 0.4535,
+      "learning_rate": 0.00026020846905537456,
+      "loss": 0.4672,
       "step": 1120
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.00025977850162866445,
-      "loss": 0.4752,
+      "learning_rate": 0.0002598175895765472,
+      "loss": 0.4802,
       "step": 1130
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0002593876221498371,
-      "loss": 0.4563,
+      "learning_rate": 0.0002594267100977198,
+      "loss": 0.4656,
       "step": 1140
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.00025899674267100976,
-      "loss": 0.4621,
+      "learning_rate": 0.0002590358306188925,
+      "loss": 0.4466,
       "step": 1150
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.00025860586319218236,
-      "loss": 0.4634,
+      "learning_rate": 0.00025864495114006513,
+      "loss": 0.475,
       "step": 1160
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.000258214983713355,
-      "loss": 0.4593,
+      "learning_rate": 0.00025825407166123773,
+      "loss": 0.4857,
       "step": 1170
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.00025782410423452767,
-      "loss": 0.464,
+      "learning_rate": 0.0002578631921824104,
+      "loss": 0.4479,
       "step": 1180
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0002574332247557003,
-      "loss": 0.4588,
+      "learning_rate": 0.00025747231270358304,
+      "loss": 0.4488,
       "step": 1190
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0002570423452768729,
-      "loss": 0.4605,
+      "learning_rate": 0.0002570814332247557,
+      "loss": 0.4646,
       "step": 1200
     },
     {
       "epoch": 0.77,
-      "eval_loss": 0.46422210335731506,
-      "eval_runtime": 395.5692,
-      "eval_samples_per_second": 5.056,
-      "eval_steps_per_second": 0.632,
+      "eval_loss": 0.46300560235977173,
+      "eval_runtime": 220.2387,
+      "eval_samples_per_second": 9.081,
+      "eval_steps_per_second": 1.135,
       "step": 1200
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0002566514657980456,
-      "loss": 0.4725,
+      "learning_rate": 0.00025669055374592835,
+      "loss": 0.4722,
       "step": 1210
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0002562605863192182,
-      "loss": 0.4552,
+      "learning_rate": 0.00025629967426710095,
+      "loss": 0.4661,
       "step": 1220
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.00025586970684039084,
-      "loss": 0.4766,
+      "learning_rate": 0.0002559087947882736,
+      "loss": 0.4489,
       "step": 1230
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0002554788273615635,
-      "loss": 0.4615,
+      "learning_rate": 0.0002555179153094462,
+      "loss": 0.4769,
       "step": 1240
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.00025508794788273615,
-      "loss": 0.447,
+      "learning_rate": 0.00025512703583061887,
+      "loss": 0.4707,
       "step": 1250
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0002546970684039088,
-      "loss": 0.4557,
+      "learning_rate": 0.0002547361563517915,
+      "loss": 0.4448,
       "step": 1260
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0002543061889250814,
-      "loss": 0.4758,
+      "learning_rate": 0.0002543452768729641,
+      "loss": 0.4647,
       "step": 1270
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00025391530944625406,
-      "loss": 0.4569,
+      "learning_rate": 0.0002539543973941368,
+      "loss": 0.4694,
       "step": 1280
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0002535244299674267,
-      "loss": 0.4772,
+      "learning_rate": 0.00025356351791530943,
+      "loss": 0.475,
       "step": 1290
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0002531335504885993,
-      "loss": 0.4646,
+      "learning_rate": 0.0002531726384364821,
+      "loss": 0.4589,
       "step": 1300
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00025274267100977197,
-      "loss": 0.4583,
+      "learning_rate": 0.0002527817589576547,
+      "loss": 0.483,
       "step": 1310
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.00025235179153094457,
-      "loss": 0.4427,
+      "learning_rate": 0.00025239087947882734,
+      "loss": 0.449,
       "step": 1320
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00025196091205211723,
-      "loss": 0.454,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 0.4533,
       "step": 1330
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0002515700325732899,
-      "loss": 0.4564,
+      "learning_rate": 0.0002516091205211726,
+      "loss": 0.4429,
       "step": 1340
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.00025117915309446254,
-      "loss": 0.4752,
+      "learning_rate": 0.00025121824104234526,
+      "loss": 0.4623,
       "step": 1350
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0002507882736156352,
-      "loss": 0.454,
+      "learning_rate": 0.0002508273615635179,
+      "loss": 0.4767,
       "step": 1360
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0002503973941368078,
-      "loss": 0.4518,
+      "learning_rate": 0.00025043648208469057,
+      "loss": 0.4803,
       "step": 1370
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00025000651465798045,
-      "loss": 0.4723,
+      "learning_rate": 0.00025004560260586317,
+      "loss": 0.4499,
       "step": 1380
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00024961563517915305,
-      "loss": 0.448,
+      "learning_rate": 0.0002496547231270358,
+      "loss": 0.444,
       "step": 1390
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0002492247557003257,
-      "loss": 0.47,
+      "learning_rate": 0.0002492638436482084,
+      "loss": 0.477,
       "step": 1400
     },
     {
       "epoch": 0.9,
-      "eval_loss": 0.46156543493270874,
-      "eval_runtime": 401.374,
-      "eval_samples_per_second": 4.983,
-      "eval_steps_per_second": 0.623,
+      "eval_loss": 0.46069058775901794,
+      "eval_runtime": 220.249,
+      "eval_samples_per_second": 9.081,
+      "eval_steps_per_second": 1.135,
       "step": 1400
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00024883387622149836,
-      "loss": 0.4864,
+      "learning_rate": 0.0002488729641693811,
+      "loss": 0.4628,
       "step": 1410
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00024844299674267096,
-      "loss": 0.4461,
+      "learning_rate": 0.00024848208469055373,
+      "loss": 0.4528,
       "step": 1420
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0002480521172638436,
-      "loss": 0.4608,
+      "learning_rate": 0.00024809120521172634,
+      "loss": 0.4669,
       "step": 1430
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00024766123778501627,
-      "loss": 0.4562,
+      "learning_rate": 0.000247700325732899,
+      "loss": 0.4607,
       "step": 1440
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00024727035830618893,
-      "loss": 0.4753,
+      "learning_rate": 0.00024730944625407165,
+      "loss": 0.4703,
       "step": 1450
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00024687947882736153,
-      "loss": 0.4485,
+      "learning_rate": 0.0002469185667752443,
+      "loss": 0.4613,
       "step": 1460
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0002464885993485342,
-      "loss": 0.4752,
+      "learning_rate": 0.00024652768729641696,
+      "loss": 0.4593,
       "step": 1470
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0002460977198697068,
-      "loss": 0.457,
+      "learning_rate": 0.00024613680781758956,
+      "loss": 0.4515,
       "step": 1480
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00024570684039087944,
-      "loss": 0.4668,
+      "learning_rate": 0.0002457459283387622,
+      "loss": 0.4493,
       "step": 1490
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0002453159609120521,
-      "loss": 0.4655,
+      "learning_rate": 0.0002453550488599348,
+      "loss": 0.4572,
       "step": 1500
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.00024492508143322475,
-      "loss": 0.4448,
+      "learning_rate": 0.00024496416938110747,
+      "loss": 0.4396,
       "step": 1510
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.00024453420195439735,
-      "loss": 0.4525,
+      "learning_rate": 0.0002445732899022801,
+      "loss": 0.4572,
       "step": 1520
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.00024414332247557,
-      "loss": 0.4414,
+      "learning_rate": 0.0002441824104234527,
+      "loss": 0.4652,
       "step": 1530
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00024375244299674263,
-      "loss": 0.4635,
+      "learning_rate": 0.00024379153094462538,
+      "loss": 0.4449,
       "step": 1540
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0002433615635179153,
-      "loss": 0.4837,
+      "learning_rate": 0.00024340065146579804,
+      "loss": 0.4574,
       "step": 1550
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.00024297068403908795,
-      "loss": 0.4482,
+      "learning_rate": 0.00024300977198697066,
+      "loss": 0.4336,
       "step": 1560
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.00024257980456026055,
-      "loss": 0.443,
+      "learning_rate": 0.00024261889250814332,
+      "loss": 0.4368,
       "step": 1570
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0002421889250814332,
-      "loss": 0.4539,
+      "learning_rate": 0.00024222801302931592,
+      "loss": 0.45,
       "step": 1580
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00024179804560260583,
-      "loss": 0.4554,
+      "learning_rate": 0.00024183713355048858,
+      "loss": 0.4272,
       "step": 1590
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.00024140716612377849,
-      "loss": 0.4459,
+      "learning_rate": 0.0002414462540716612,
+      "loss": 0.4475,
       "step": 1600
     },
     {
       "epoch": 1.03,
-      "eval_loss": 0.45973867177963257,
-      "eval_runtime": 402.8623,
-      "eval_samples_per_second": 4.964,
-      "eval_steps_per_second": 0.621,
+      "eval_loss": 0.4588911533355713,
+      "eval_runtime": 220.3526,
+      "eval_samples_per_second": 9.076,
+      "eval_steps_per_second": 1.135,
       "step": 1600
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.00024101628664495114,
-      "loss": 0.4464,
+      "learning_rate": 0.00024105537459283386,
+      "loss": 0.4423,
       "step": 1610
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.00024062540716612377,
-      "loss": 0.4333,
+      "learning_rate": 0.00024066449511400651,
+      "loss": 0.442,
       "step": 1620
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0002402345276872964,
-      "loss": 0.4484,
+      "learning_rate": 0.00024027361563517911,
+      "loss": 0.4412,
       "step": 1630
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00023984364820846902,
-      "loss": 0.4449,
+      "learning_rate": 0.00023988273615635177,
+      "loss": 0.4403,
       "step": 1640
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.00023945276872964168,
-      "loss": 0.4403,
+      "learning_rate": 0.0002394918566775244,
+      "loss": 0.4397,
       "step": 1650
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0002390618892508143,
-      "loss": 0.4382,
+      "learning_rate": 0.00023910097719869705,
+      "loss": 0.44,
       "step": 1660
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.00023867100977198696,
-      "loss": 0.4467,
+      "learning_rate": 0.0002387100977198697,
+      "loss": 0.4295,
       "step": 1670
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.00023828013029315956,
-      "loss": 0.4335,
+      "learning_rate": 0.0002383192182410423,
+      "loss": 0.4305,
       "step": 1680
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00023788925081433222,
-      "loss": 0.4477,
+      "learning_rate": 0.00023792833876221497,
+      "loss": 0.433,
       "step": 1690
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00023749837133550487,
-      "loss": 0.4468,
+      "learning_rate": 0.0002375374592833876,
+      "loss": 0.4514,
       "step": 1700
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0002371074918566775,
-      "loss": 0.4219,
+      "learning_rate": 0.00023714657980456025,
+      "loss": 0.4696,
       "step": 1710
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.00023671661237785016,
-      "loss": 0.4577,
+      "learning_rate": 0.00023675570032573288,
+      "loss": 0.4497,
       "step": 1720
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.00023632573289902276,
-      "loss": 0.4597,
+      "learning_rate": 0.0002363648208469055,
+      "loss": 0.433,
       "step": 1730
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00023593485342019541,
-      "loss": 0.4396,
+      "learning_rate": 0.00023597394136807813,
+      "loss": 0.4441,
       "step": 1740
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.00023554397394136807,
-      "loss": 0.4559,
+      "learning_rate": 0.0002355830618892508,
+      "loss": 0.4283,
       "step": 1750
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0002351530944625407,
-      "loss": 0.4504,
+      "learning_rate": 0.00023519218241042344,
+      "loss": 0.46,
       "step": 1760
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00023476221498371335,
-      "loss": 0.4387,
+      "learning_rate": 0.00023480130293159607,
+      "loss": 0.4386,
       "step": 1770
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00023437133550488595,
-      "loss": 0.4391,
+      "learning_rate": 0.00023441042345276873,
+      "loss": 0.4548,
       "step": 1780
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0002339804560260586,
-      "loss": 0.4502,
+      "learning_rate": 0.00023401954397394133,
+      "loss": 0.4546,
       "step": 1790
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.00023358957654723124,
-      "loss": 0.4358,
+      "learning_rate": 0.00023362866449511398,
+      "loss": 0.4482,
       "step": 1800
     },
     {
       "epoch": 1.16,
-      "eval_loss": 0.4591727554798126,
-      "eval_runtime": 407.7185,
-      "eval_samples_per_second": 4.905,
-      "eval_steps_per_second": 0.613,
+      "eval_loss": 0.4585164189338684,
+      "eval_runtime": 220.4297,
+      "eval_samples_per_second": 9.073,
+      "eval_steps_per_second": 1.134,
       "step": 1800
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0002331986970684039,
-      "loss": 0.4466,
+      "learning_rate": 0.00023323778501628664,
+      "loss": 0.4383,
       "step": 1810
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.00023280781758957655,
-      "loss": 0.4487,
+      "learning_rate": 0.00023284690553745927,
+      "loss": 0.4464,
       "step": 1820
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.00023241693811074915,
-      "loss": 0.429,
+      "learning_rate": 0.00023245602605863192,
+      "loss": 0.4354,
       "step": 1830
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0002320260586319218,
-      "loss": 0.4414,
+      "learning_rate": 0.00023206514657980452,
+      "loss": 0.4403,
       "step": 1840
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.00023163517915309443,
-      "loss": 0.4586,
+      "learning_rate": 0.00023167426710097718,
+      "loss": 0.4367,
       "step": 1850
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0002312442996742671,
-      "loss": 0.4363,
+      "learning_rate": 0.0002312833876221498,
+      "loss": 0.4485,
       "step": 1860
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.00023085342019543974,
-      "loss": 0.44,
+      "learning_rate": 0.00023089250814332246,
+      "loss": 0.4553,
       "step": 1870
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.00023046254071661234,
-      "loss": 0.4623,
+      "learning_rate": 0.00023050162866449512,
+      "loss": 0.4262,
       "step": 1880
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.000230071661237785,
-      "loss": 0.4352,
+      "learning_rate": 0.00023011074918566772,
+      "loss": 0.4394,
       "step": 1890
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00022968078175895763,
-      "loss": 0.434,
+      "learning_rate": 0.00022971986970684037,
+      "loss": 0.4441,
       "step": 1900
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.00022928990228013028,
-      "loss": 0.4411,
+      "learning_rate": 0.000229328990228013,
+      "loss": 0.4405,
       "step": 1910
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0002288990228013029,
-      "loss": 0.454,
+      "learning_rate": 0.00022893811074918566,
+      "loss": 0.434,
       "step": 1920
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.00022850814332247554,
-      "loss": 0.4238,
+      "learning_rate": 0.0002285472312703583,
+      "loss": 0.4326,
       "step": 1930
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00022811726384364817,
-      "loss": 0.4485,
+      "learning_rate": 0.0002281563517915309,
+      "loss": 0.4401,
       "step": 1940
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00022772638436482082,
-      "loss": 0.4397,
+      "learning_rate": 0.00022776547231270357,
+      "loss": 0.4394,
       "step": 1950
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00022733550488599348,
-      "loss": 0.43,
+      "learning_rate": 0.0002273745928338762,
+      "loss": 0.4404,
       "step": 1960
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0002269446254071661,
-      "loss": 0.4491,
+      "learning_rate": 0.00022698371335504885,
+      "loss": 0.4437,
       "step": 1970
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.00022655374592833873,
-      "loss": 0.4432,
+      "learning_rate": 0.00022659283387622148,
+      "loss": 0.4598,
       "step": 1980
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.00022616286644951136,
-      "loss": 0.4213,
+      "learning_rate": 0.0002262019543973941,
+      "loss": 0.4308,
       "step": 1990
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00022577198697068402,
-      "loss": 0.4368,
+      "learning_rate": 0.00022581107491856674,
+      "loss": 0.4498,
       "step": 2000
     },
     {
       "epoch": 1.29,
-      "eval_loss": 0.45750632882118225,
-      "eval_runtime": 399.3044,
-      "eval_samples_per_second": 5.009,
-      "eval_steps_per_second": 0.626,
+      "eval_loss": 0.456809401512146,
+      "eval_runtime": 220.2549,
+      "eval_samples_per_second": 9.08,
+      "eval_steps_per_second": 1.135,
       "step": 2000
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00022538110749185667,
-      "loss": 0.4461,
+      "learning_rate": 0.0002254201954397394,
+      "loss": 0.4555,
       "step": 2010
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0002249902280130293,
-      "loss": 0.4253,
+      "learning_rate": 0.00022502931596091205,
+      "loss": 0.4397,
       "step": 2020
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.00022459934853420196,
-      "loss": 0.4429,
+      "learning_rate": 0.00022463843648208467,
+      "loss": 0.4459,
       "step": 2030
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.00022420846905537456,
-      "loss": 0.4436,
+      "learning_rate": 0.0002242475570032573,
+      "loss": 0.4373,
       "step": 2040
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0002238175895765472,
-      "loss": 0.4418,
+      "learning_rate": 0.00022385667752442993,
+      "loss": 0.4346,
       "step": 2050
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00022342671009771984,
-      "loss": 0.4156,
+      "learning_rate": 0.00022346579804560259,
+      "loss": 0.4224,
       "step": 2060
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0002230358306188925,
-      "loss": 0.439,
+      "learning_rate": 0.00022307491856677524,
+      "loss": 0.4539,
       "step": 2070
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.00022264495114006515,
-      "loss": 0.451,
+      "learning_rate": 0.00022268403908794787,
+      "loss": 0.4431,
       "step": 2080
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.00022225407166123775,
-      "loss": 0.4468,
+      "learning_rate": 0.0002222931596091205,
+      "loss": 0.4436,
       "step": 2090
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0002218631921824104,
-      "loss": 0.4423,
+      "learning_rate": 0.00022190228013029313,
+      "loss": 0.4453,
       "step": 2100
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00022147231270358304,
-      "loss": 0.4267,
+      "learning_rate": 0.00022151140065146578,
+      "loss": 0.4268,
       "step": 2110
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0002210814332247557,
-      "loss": 0.4329,
+      "learning_rate": 0.0002211205211726384,
+      "loss": 0.4121,
       "step": 2120
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.00022069055374592835,
-      "loss": 0.4417,
+      "learning_rate": 0.00022072964169381106,
+      "loss": 0.4614,
       "step": 2130
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00022029967426710095,
-      "loss": 0.4544,
+      "learning_rate": 0.00022033876221498372,
+      "loss": 0.452,
       "step": 2140
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0002199087947882736,
-      "loss": 0.4294,
+      "learning_rate": 0.00021994788273615632,
+      "loss": 0.4472,
       "step": 2150
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00021951791530944623,
-      "loss": 0.4484,
+      "learning_rate": 0.00021955700325732898,
+      "loss": 0.4318,
       "step": 2160
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.00021912703583061889,
-      "loss": 0.4521,
+      "learning_rate": 0.0002191661237785016,
+      "loss": 0.4476,
       "step": 2170
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.00021873615635179151,
-      "loss": 0.4362,
+      "learning_rate": 0.00021877524429967426,
+      "loss": 0.4378,
       "step": 2180
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.00021834527687296414,
-      "loss": 0.4342,
+      "learning_rate": 0.00021838436482084691,
+      "loss": 0.425,
       "step": 2190
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.00021795439739413677,
-      "loss": 0.4394,
+      "learning_rate": 0.00021799348534201952,
+      "loss": 0.4392,
       "step": 2200
     },
     {
       "epoch": 1.41,
-      "eval_loss": 0.4563552439212799,
-      "eval_runtime": 401.35,
-      "eval_samples_per_second": 4.983,
-      "eval_steps_per_second": 0.623,
+      "eval_loss": 0.45531219244003296,
+      "eval_runtime": 220.3519,
+      "eval_samples_per_second": 9.076,
+      "eval_steps_per_second": 1.135,
       "step": 2200
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00021756351791530943,
-      "loss": 0.444,
+      "learning_rate": 0.00021760260586319217,
+      "loss": 0.4422,
       "step": 2210
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.00021717263843648208,
-      "loss": 0.4434,
+      "learning_rate": 0.0002172117263843648,
+      "loss": 0.4403,
       "step": 2220
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0002167817589576547,
-      "loss": 0.4451,
+      "learning_rate": 0.00021682084690553745,
+      "loss": 0.45,
       "step": 2230
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.00021639087947882734,
-      "loss": 0.4347,
+      "learning_rate": 0.00021642996742671008,
+      "loss": 0.4282,
       "step": 2240
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00021599999999999996,
-      "loss": 0.4347,
+      "learning_rate": 0.0002160390879478827,
+      "loss": 0.4319,
       "step": 2250
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00021560912052117262,
-      "loss": 0.4532,
+      "learning_rate": 0.00021564820846905534,
+      "loss": 0.4438,
       "step": 2260
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.00021521824104234528,
-      "loss": 0.4578,
+      "learning_rate": 0.000215257328990228,
+      "loss": 0.4509,
       "step": 2270
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0002148273615635179,
-      "loss": 0.4479,
+      "learning_rate": 0.00021486644951140065,
+      "loss": 0.4454,
       "step": 2280
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.00021443648208469053,
-      "loss": 0.4606,
+      "learning_rate": 0.00021447557003257328,
+      "loss": 0.4176,
       "step": 2290
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.00021404560260586316,
-      "loss": 0.431,
+      "learning_rate": 0.0002140846905537459,
+      "loss": 0.4554,
       "step": 2300
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00021365472312703581,
-      "loss": 0.4584,
+      "learning_rate": 0.00021369381107491853,
+      "loss": 0.4388,
       "step": 2310
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00021326384364820844,
-      "loss": 0.4459,
+      "learning_rate": 0.0002133029315960912,
+      "loss": 0.456,
       "step": 2320
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0002128729641693811,
-      "loss": 0.4469,
+      "learning_rate": 0.00021291205211726384,
+      "loss": 0.4257,
       "step": 2330
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0002124820846905537,
-      "loss": 0.4335,
+      "learning_rate": 0.00021252117263843647,
+      "loss": 0.4428,
       "step": 2340
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.00021209120521172635,
-      "loss": 0.4467,
+      "learning_rate": 0.0002121302931596091,
+      "loss": 0.4266,
       "step": 2350
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.000211700325732899,
-      "loss": 0.4325,
+      "learning_rate": 0.00021173941368078173,
+      "loss": 0.4591,
       "step": 2360
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00021130944625407164,
-      "loss": 0.447,
+      "learning_rate": 0.00021134853420195438,
+      "loss": 0.453,
       "step": 2370
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0002109185667752443,
-      "loss": 0.4314,
+      "learning_rate": 0.000210957654723127,
+      "loss": 0.4414,
       "step": 2380
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.00021052768729641695,
-      "loss": 0.4451,
+      "learning_rate": 0.00021056677524429967,
+      "loss": 0.4619,
       "step": 2390
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.00021013680781758955,
-      "loss": 0.4548,
+      "learning_rate": 0.00021017589576547227,
+      "loss": 0.4285,
       "step": 2400
     },
     {
       "epoch": 1.54,
-      "eval_loss": 0.4556201696395874,
-      "eval_runtime": 401.3952,
-      "eval_samples_per_second": 4.983,
-      "eval_steps_per_second": 0.623,
+      "eval_loss": 0.45426830649375916,
+      "eval_runtime": 220.2316,
+      "eval_samples_per_second": 9.081,
+      "eval_steps_per_second": 1.135,
       "step": 2400
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0002097459283387622,
-      "loss": 0.4303,
+      "learning_rate": 0.00020978501628664492,
+      "loss": 0.4476,
       "step": 2410
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00020935504885993483,
-      "loss": 0.4248,
+      "learning_rate": 0.00020939413680781758,
+      "loss": 0.4361,
       "step": 2420
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0002089641693811075,
-      "loss": 0.4205,
+      "learning_rate": 0.0002090032573289902,
+      "loss": 0.452,
       "step": 2430
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.00020857328990228012,
-      "loss": 0.4434,
+      "learning_rate": 0.00020861237785016286,
+      "loss": 0.4358,
       "step": 2440
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.00020818241042345274,
-      "loss": 0.4544,
+      "learning_rate": 0.00020822149837133546,
+      "loss": 0.447,
       "step": 2450
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.00020779153094462537,
-      "loss": 0.4467,
+      "learning_rate": 0.00020783061889250812,
+      "loss": 0.4592,
       "step": 2460
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00020740065146579803,
-      "loss": 0.4448,
+      "learning_rate": 0.00020743973941368077,
+      "loss": 0.4529,
       "step": 2470
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00020700977198697068,
-      "loss": 0.4384,
+      "learning_rate": 0.0002070488599348534,
+      "loss": 0.4545,
       "step": 2480
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0002066188925081433,
-      "loss": 0.4623,
+      "learning_rate": 0.00020665798045602606,
+      "loss": 0.4303,
       "step": 2490
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.00020622801302931594,
-      "loss": 0.4413,
+      "learning_rate": 0.00020626710097719866,
+      "loss": 0.4489,
       "step": 2500
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.00020583713355048857,
-      "loss": 0.4557,
+      "learning_rate": 0.0002058762214983713,
+      "loss": 0.4306,
       "step": 2510
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00020544625407166122,
-      "loss": 0.4339,
+      "learning_rate": 0.00020548534201954394,
+      "loss": 0.4441,
       "step": 2520
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00020505537459283388,
-      "loss": 0.4239,
+      "learning_rate": 0.0002050944625407166,
+      "loss": 0.4132,
       "step": 2530
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0002046644951140065,
-      "loss": 0.4323,
+      "learning_rate": 0.00020470358306188925,
+      "loss": 0.428,
       "step": 2540
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.00020427361563517913,
-      "loss": 0.4479,
+      "learning_rate": 0.00020431270358306188,
+      "loss": 0.4242,
       "step": 2550
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00020388273615635176,
-      "loss": 0.4448,
+      "learning_rate": 0.0002039218241042345,
+      "loss": 0.4372,
       "step": 2560
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00020349185667752442,
-      "loss": 0.4348,
+      "learning_rate": 0.00020353094462540714,
+      "loss": 0.4339,
       "step": 2570
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.00020310097719869705,
-      "loss": 0.4344,
+      "learning_rate": 0.0002031400651465798,
+      "loss": 0.434,
       "step": 2580
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0002027100977198697,
-      "loss": 0.4401,
+      "learning_rate": 0.00020274918566775245,
+      "loss": 0.4282,
       "step": 2590
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0002023192182410423,
-      "loss": 0.4366,
+      "learning_rate": 0.00020235830618892507,
+      "loss": 0.4352,
       "step": 2600
     },
     {
       "epoch": 1.67,
-      "eval_loss": 0.45435893535614014,
-      "eval_runtime": 403.3395,
-      "eval_samples_per_second": 4.959,
-      "eval_steps_per_second": 0.62,
+      "eval_loss": 0.4528008699417114,
+      "eval_runtime": 220.0694,
+      "eval_samples_per_second": 9.088,
+      "eval_steps_per_second": 1.136,
       "step": 2600
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00020192833876221496,
-      "loss": 0.4367,
+      "learning_rate": 0.0002019674267100977,
+      "loss": 0.4478,
       "step": 2610
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.0002015374592833876,
-      "loss": 0.4538,
+      "learning_rate": 0.00020157654723127033,
+      "loss": 0.4317,
       "step": 2620
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00020114657980456024,
-      "loss": 0.4388,
+      "learning_rate": 0.00020118566775244299,
+      "loss": 0.4447,
       "step": 2630
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.0002007557003257329,
-      "loss": 0.4264,
+      "learning_rate": 0.00020079478827361561,
+      "loss": 0.4454,
       "step": 2640
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.0002003648208469055,
-      "loss": 0.4315,
+      "learning_rate": 0.00020040390879478827,
+      "loss": 0.4422,
       "step": 2650
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00019997394136807815,
-      "loss": 0.4304,
+      "learning_rate": 0.00020001302931596087,
+      "loss": 0.4504,
       "step": 2660
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.0001995830618892508,
-      "loss": 0.4332,
+      "learning_rate": 0.00019962214983713353,
+      "loss": 0.4322,
       "step": 2670
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00019919218241042344,
-      "loss": 0.4431,
+      "learning_rate": 0.00019923127035830618,
+      "loss": 0.4387,
       "step": 2680
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.0001988013029315961,
-      "loss": 0.4386,
+      "learning_rate": 0.0001988403908794788,
+      "loss": 0.4398,
       "step": 2690
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0001984104234527687,
-      "loss": 0.4556,
+      "learning_rate": 0.00019844951140065146,
+      "loss": 0.4384,
       "step": 2700
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00019801954397394135,
-      "loss": 0.4463,
+      "learning_rate": 0.00019805863192182407,
+      "loss": 0.4412,
       "step": 2710
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00019762866449511398,
-      "loss": 0.4496,
+      "learning_rate": 0.00019766775244299672,
+      "loss": 0.4535,
       "step": 2720
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00019723778501628663,
-      "loss": 0.4313,
+      "learning_rate": 0.00019727687296416938,
+      "loss": 0.437,
       "step": 2730
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00019684690553745929,
-      "loss": 0.4378,
+      "learning_rate": 0.000196885993485342,
+      "loss": 0.4403,
       "step": 2740
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00019645602605863191,
-      "loss": 0.4352,
+      "learning_rate": 0.00019649511400651466,
+      "loss": 0.4343,
       "step": 2750
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00019606514657980454,
-      "loss": 0.4385,
+      "learning_rate": 0.00019610423452768726,
+      "loss": 0.4281,
       "step": 2760
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00019567426710097717,
-      "loss": 0.4312,
+      "learning_rate": 0.00019571335504885992,
+      "loss": 0.4364,
       "step": 2770
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00019528338762214983,
-      "loss": 0.4339,
+      "learning_rate": 0.00019532247557003254,
+      "loss": 0.4473,
       "step": 2780
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00019489250814332248,
-      "loss": 0.4609,
+      "learning_rate": 0.0001949315960912052,
+      "loss": 0.4421,
       "step": 2790
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.0001945016286644951,
-      "loss": 0.4428,
+      "learning_rate": 0.00019454071661237785,
+      "loss": 0.4427,
       "step": 2800
     },
     {
       "epoch": 1.8,
-      "eval_loss": 0.45374733209609985,
-      "eval_runtime": 397.5954,
-      "eval_samples_per_second": 5.03,
-      "eval_steps_per_second": 0.629,
+      "eval_loss": 0.4516027867794037,
+      "eval_runtime": 220.0573,
+      "eval_samples_per_second": 9.089,
+      "eval_steps_per_second": 1.136,
       "step": 2800
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00019411074918566774,
-      "loss": 0.4311,
+      "learning_rate": 0.00019414983713355046,
+      "loss": 0.449,
       "step": 2810
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00019371986970684037,
-      "loss": 0.45,
+      "learning_rate": 0.0001937589576547231,
+      "loss": 0.433,
       "step": 2820
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00019332899022801302,
-      "loss": 0.4485,
+      "learning_rate": 0.00019336807817589574,
+      "loss": 0.4368,
       "step": 2830
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00019293811074918565,
-      "loss": 0.4488,
+      "learning_rate": 0.0001929771986970684,
+      "loss": 0.4432,
       "step": 2840
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.0001925472312703583,
-      "loss": 0.4474,
+      "learning_rate": 0.00019258631921824105,
+      "loss": 0.4394,
       "step": 2850
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.0001921563517915309,
-      "loss": 0.4434,
+      "learning_rate": 0.00019219543973941365,
+      "loss": 0.4421,
       "step": 2860
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00019176547231270356,
-      "loss": 0.4237,
+      "learning_rate": 0.0001918045602605863,
+      "loss": 0.4526,
       "step": 2870
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00019137459283387622,
-      "loss": 0.4413,
+      "learning_rate": 0.00019141368078175893,
+      "loss": 0.4322,
       "step": 2880
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00019098371335504884,
-      "loss": 0.444,
+      "learning_rate": 0.0001910228013029316,
+      "loss": 0.4376,
       "step": 2890
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.0001905928338762215,
-      "loss": 0.4351,
+      "learning_rate": 0.00019063192182410422,
+      "loss": 0.4496,
       "step": 2900
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.0001902019543973941,
-      "loss": 0.4528,
+      "learning_rate": 0.00019024104234527687,
+      "loss": 0.4429,
       "step": 2910
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00018981107491856675,
-      "loss": 0.4369,
+      "learning_rate": 0.00018985016286644947,
+      "loss": 0.4269,
       "step": 2920
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.0001894201954397394,
-      "loss": 0.4209,
+      "learning_rate": 0.00018945928338762213,
+      "loss": 0.4409,
       "step": 2930
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00018902931596091204,
-      "loss": 0.4513,
+      "learning_rate": 0.00018906840390879478,
+      "loss": 0.4328,
       "step": 2940
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.0001886384364820847,
-      "loss": 0.4353,
+      "learning_rate": 0.0001886775244299674,
+      "loss": 0.4438,
       "step": 2950
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.0001882475570032573,
-      "loss": 0.438,
+      "learning_rate": 0.00018828664495114007,
+      "loss": 0.4238,
       "step": 2960
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00018785667752442995,
-      "loss": 0.4522,
+      "learning_rate": 0.00018789576547231267,
+      "loss": 0.4288,
       "step": 2970
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00018746579804560258,
-      "loss": 0.4303,
+      "learning_rate": 0.00018750488599348532,
+      "loss": 0.4356,
       "step": 2980
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00018707491856677523,
-      "loss": 0.4698,
+      "learning_rate": 0.00018711400651465798,
+      "loss": 0.453,
       "step": 2990
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.0001866840390879479,
-      "loss": 0.4386,
+      "learning_rate": 0.0001867231270358306,
+      "loss": 0.4574,
       "step": 3000
     },
     {
       "epoch": 1.93,
-      "eval_loss": 0.45254799723625183,
-      "eval_runtime": 392.9711,
-      "eval_samples_per_second": 5.089,
-      "eval_steps_per_second": 0.636,
+      "eval_loss": 0.4507245421409607,
+      "eval_runtime": 219.8338,
+      "eval_samples_per_second": 9.098,
+      "eval_steps_per_second": 1.137,
       "step": 3000
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.0001862931596091205,
-      "loss": 0.4681,
+      "learning_rate": 0.00018633224755700326,
+      "loss": 0.4401,
       "step": 3010
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00018590228013029314,
-      "loss": 0.4451,
+      "learning_rate": 0.00018594136807817586,
+      "loss": 0.4562,
       "step": 3020
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00018551140065146577,
-      "loss": 0.4464,
+      "learning_rate": 0.00018555048859934852,
+      "loss": 0.4458,
       "step": 3030
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00018512052117263843,
-      "loss": 0.4556,
+      "learning_rate": 0.00018515960912052115,
+      "loss": 0.4391,
       "step": 3040
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00018472964169381108,
-      "loss": 0.4266,
+      "learning_rate": 0.0001847687296416938,
+      "loss": 0.4328,
       "step": 3050
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00018433876221498368,
-      "loss": 0.4527,
+      "learning_rate": 0.00018437785016286646,
+      "loss": 0.4397,
       "step": 3060
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00018394788273615634,
-      "loss": 0.4418,
+      "learning_rate": 0.00018398697068403906,
+      "loss": 0.4199,
       "step": 3070
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00018355700325732897,
-      "loss": 0.4444,
+      "learning_rate": 0.0001835960912052117,
+      "loss": 0.439,
       "step": 3080
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00018316612377850162,
-      "loss": 0.4578,
+      "learning_rate": 0.00018320521172638434,
+      "loss": 0.4354,
       "step": 3090
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00018277524429967425,
-      "loss": 0.4311,
+      "learning_rate": 0.000182814332247557,
+      "loss": 0.4265,
       "step": 3100
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00018238436482084688,
-      "loss": 0.4411,
+      "learning_rate": 0.00018242345276872965,
+      "loss": 0.4338,
       "step": 3110
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.0001819934853420195,
-      "loss": 0.4216,
+      "learning_rate": 0.00018203257328990225,
+      "loss": 0.4107,
       "step": 3120
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00018160260586319216,
-      "loss": 0.4057,
+      "learning_rate": 0.0001816416938110749,
+      "loss": 0.4254,
       "step": 3130
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00018121172638436482,
-      "loss": 0.4127,
+      "learning_rate": 0.00018125081433224754,
+      "loss": 0.4312,
       "step": 3140
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00018082084690553745,
-      "loss": 0.4325,
+      "learning_rate": 0.0001808599348534202,
+      "loss": 0.4049,
       "step": 3150
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.0001804299674267101,
-      "loss": 0.4169,
+      "learning_rate": 0.00018046905537459282,
+      "loss": 0.4226,
       "step": 3160
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.0001800390879478827,
-      "loss": 0.4052,
+      "learning_rate": 0.00018007817589576545,
+      "loss": 0.4172,
       "step": 3170
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00017964820846905536,
-      "loss": 0.4059,
+      "learning_rate": 0.00017968729641693808,
+      "loss": 0.4047,
       "step": 3180
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.000179257328990228,
-      "loss": 0.4234,
+      "learning_rate": 0.00017929641693811073,
+      "loss": 0.4269,
       "step": 3190
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00017886644951140064,
-      "loss": 0.4145,
+      "learning_rate": 0.00017890553745928339,
+      "loss": 0.4027,
       "step": 3200
     },
     {
       "epoch": 2.06,
-      "eval_loss": 0.4534437358379364,
-      "eval_runtime": 397.6364,
-      "eval_samples_per_second": 5.03,
-      "eval_steps_per_second": 0.629,
+      "eval_loss": 0.45280253887176514,
+      "eval_runtime": 219.8269,
+      "eval_samples_per_second": 9.098,
+      "eval_steps_per_second": 1.137,
       "step": 3200
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.0001784755700325733,
-      "loss": 0.4233,
+      "learning_rate": 0.00017851465798045601,
+      "loss": 0.4149,
       "step": 3210
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.0001780846905537459,
-      "loss": 0.3946,
+      "learning_rate": 0.00017812377850162864,
+      "loss": 0.4317,
       "step": 3220
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00017769381107491855,
-      "loss": 0.4178,
+      "learning_rate": 0.00017773289902280127,
+      "loss": 0.4243,
       "step": 3230
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00017730293159609118,
-      "loss": 0.4335,
+      "learning_rate": 0.00017734201954397393,
+      "loss": 0.4036,
       "step": 3240
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00017691205211726384,
-      "loss": 0.4059,
+      "learning_rate": 0.00017695114006514658,
+      "loss": 0.4113,
       "step": 3250
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.0001765211726384365,
-      "loss": 0.4228,
+      "learning_rate": 0.0001765602605863192,
+      "loss": 0.4206,
       "step": 3260
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.0001761302931596091,
-      "loss": 0.4188,
+      "learning_rate": 0.00017616938110749184,
+      "loss": 0.4194,
       "step": 3270
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.00017573941368078175,
-      "loss": 0.4234,
+      "learning_rate": 0.00017577850162866447,
+      "loss": 0.4203,
       "step": 3280
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.00017534853420195438,
-      "loss": 0.411,
+      "learning_rate": 0.00017538762214983712,
+      "loss": 0.4149,
       "step": 3290
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.00017495765472312703,
-      "loss": 0.4268,
+      "learning_rate": 0.00017499674267100975,
+      "loss": 0.3972,
       "step": 3300
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.00017456677524429966,
-      "loss": 0.4105,
+      "learning_rate": 0.0001746058631921824,
+      "loss": 0.3971,
       "step": 3310
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.0001741758957654723,
-      "loss": 0.4298,
+      "learning_rate": 0.00017421498371335506,
+      "loss": 0.4309,
       "step": 3320
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.00017378501628664494,
-      "loss": 0.4178,
+      "learning_rate": 0.00017382410423452766,
+      "loss": 0.4216,
       "step": 3330
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.00017339413680781757,
-      "loss": 0.4222,
+      "learning_rate": 0.00017343322475570032,
+      "loss": 0.4291,
       "step": 3340
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.00017300325732899023,
-      "loss": 0.4199,
+      "learning_rate": 0.00017304234527687294,
+      "loss": 0.4186,
       "step": 3350
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.00017261237785016285,
+      "learning_rate": 0.0001726514657980456,
       "loss": 0.4258,
       "step": 3360
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.00017222149837133548,
-      "loss": 0.4154,
+      "learning_rate": 0.00017226058631921825,
+      "loss": 0.4114,
       "step": 3370
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.0001718306188925081,
-      "loss": 0.4105,
+      "learning_rate": 0.00017186970684039086,
+      "loss": 0.4166,
       "step": 3380
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.00017143973941368077,
-      "loss": 0.4259,
+      "learning_rate": 0.0001714788273615635,
+      "loss": 0.4157,
       "step": 3390
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.00017104885993485342,
-      "loss": 0.4048,
+      "learning_rate": 0.00017108794788273614,
+      "loss": 0.4302,
       "step": 3400
     },
     {
       "epoch": 2.19,
-      "eval_loss": 0.45390984416007996,
-      "eval_runtime": 397.2706,
-      "eval_samples_per_second": 5.034,
-      "eval_steps_per_second": 0.629,
+      "eval_loss": 0.4515763223171234,
+      "eval_runtime": 219.868,
+      "eval_samples_per_second": 9.096,
+      "eval_steps_per_second": 1.137,
       "step": 3400
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.00017065798045602605,
-      "loss": 0.4078,
+      "learning_rate": 0.0001706970684039088,
+      "loss": 0.4098,
       "step": 3410
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.00017026710097719868,
-      "loss": 0.413,
+      "learning_rate": 0.00017030618892508142,
+      "loss": 0.4326,
       "step": 3420
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.0001698762214983713,
-      "loss": 0.4175,
+      "learning_rate": 0.00016991530944625405,
+      "loss": 0.4303,
       "step": 3430
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.00016948534201954396,
-      "loss": 0.419,
+      "learning_rate": 0.00016952442996742668,
+      "loss": 0.4325,
       "step": 3440
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.0001690944625407166,
-      "loss": 0.4286,
+      "learning_rate": 0.00016913355048859933,
+      "loss": 0.4164,
       "step": 3450
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.00016870358306188924,
-      "loss": 0.4179,
+      "learning_rate": 0.000168742671009772,
+      "loss": 0.4195,
       "step": 3460
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.00016831270358306187,
-      "loss": 0.4121,
+      "learning_rate": 0.00016835179153094462,
+      "loss": 0.408,
       "step": 3470
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.0001679218241042345,
-      "loss": 0.4238,
+      "learning_rate": 0.00016796091205211725,
+      "loss": 0.4185,
       "step": 3480
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.00016753094462540716,
-      "loss": 0.4247,
+      "learning_rate": 0.00016757003257328987,
+      "loss": 0.4155,
       "step": 3490
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.00016714006514657978,
-      "loss": 0.4181,
+      "learning_rate": 0.00016717915309446253,
+      "loss": 0.4004,
       "step": 3500
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.00016674918566775244,
-      "loss": 0.4175,
+      "learning_rate": 0.00016678827361563518,
+      "loss": 0.4078,
       "step": 3510
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.0001663583061889251,
-      "loss": 0.4338,
+      "learning_rate": 0.0001663973941368078,
+      "loss": 0.411,
       "step": 3520
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.0001659674267100977,
-      "loss": 0.4131,
+      "learning_rate": 0.00016600651465798044,
+      "loss": 0.4242,
       "step": 3530
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.00016557654723127035,
-      "loss": 0.4215,
+      "learning_rate": 0.00016561563517915307,
+      "loss": 0.4092,
       "step": 3540
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.00016518566775244298,
-      "loss": 0.4366,
+      "learning_rate": 0.00016522475570032572,
+      "loss": 0.4193,
       "step": 3550
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.00016479478827361563,
-      "loss": 0.424,
+      "learning_rate": 0.00016483387622149835,
+      "loss": 0.419,
       "step": 3560
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.00016440390879478826,
-      "loss": 0.4147,
+      "learning_rate": 0.000164442996742671,
+      "loss": 0.4173,
       "step": 3570
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.0001640130293159609,
-      "loss": 0.4223,
+      "learning_rate": 0.0001640521172638436,
+      "loss": 0.414,
       "step": 3580
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.00016362214983713352,
-      "loss": 0.4132,
+      "learning_rate": 0.00016366123778501626,
+      "loss": 0.4115,
       "step": 3590
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.00016323127035830617,
-      "loss": 0.4019,
+      "learning_rate": 0.00016327035830618892,
+      "loss": 0.4227,
       "step": 3600
     },
     {
       "epoch": 2.32,
-      "eval_loss": 0.45374155044555664,
-      "eval_runtime": 403.7993,
-      "eval_samples_per_second": 4.953,
-      "eval_steps_per_second": 0.619,
+      "eval_loss": 0.45233967900276184,
+      "eval_runtime": 220.0403,
+      "eval_samples_per_second": 9.089,
+      "eval_steps_per_second": 1.136,
       "step": 3600
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.00016284039087947883,
-      "loss": 0.4101,
+      "learning_rate": 0.00016287947882736155,
+      "loss": 0.4238,
       "step": 3610
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.00016244951140065146,
-      "loss": 0.4234,
+      "learning_rate": 0.0001624885993485342,
+      "loss": 0.4066,
       "step": 3620
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.00016205863192182408,
-      "loss": 0.4177,
+      "learning_rate": 0.0001620977198697068,
+      "loss": 0.4226,
       "step": 3630
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.0001616677524429967,
-      "loss": 0.433,
+      "learning_rate": 0.00016170684039087946,
+      "loss": 0.429,
       "step": 3640
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00016127687296416937,
-      "loss": 0.4249,
+      "learning_rate": 0.0001613159609120521,
+      "loss": 0.4181,
       "step": 3650
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00016088599348534202,
-      "loss": 0.4233,
+      "learning_rate": 0.00016092508143322474,
+      "loss": 0.4276,
       "step": 3660
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.00016049511400651465,
-      "loss": 0.4229,
+      "learning_rate": 0.0001605342019543974,
+      "loss": 0.4169,
       "step": 3670
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.00016010423452768728,
-      "loss": 0.4221,
+      "learning_rate": 0.00016014332247557003,
+      "loss": 0.4223,
       "step": 3680
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.0001597133550488599,
-      "loss": 0.4166,
+      "learning_rate": 0.00015975244299674265,
+      "loss": 0.4136,
       "step": 3690
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.00015932247557003256,
-      "loss": 0.4183,
+      "learning_rate": 0.00015936156351791528,
+      "loss": 0.4137,
       "step": 3700
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.0001589315960912052,
-      "loss": 0.4279,
+      "learning_rate": 0.00015897068403908794,
+      "loss": 0.4247,
       "step": 3710
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.00015854071661237785,
-      "loss": 0.4246,
+      "learning_rate": 0.0001585798045602606,
+      "loss": 0.4052,
       "step": 3720
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.00015814983713355045,
-      "loss": 0.4225,
+      "learning_rate": 0.00015818892508143322,
+      "loss": 0.4308,
       "step": 3730
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.0001577589576547231,
-      "loss": 0.4199,
+      "learning_rate": 0.00015779804560260585,
+      "loss": 0.4115,
       "step": 3740
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.00015736807817589576,
-      "loss": 0.3944,
+      "learning_rate": 0.00015740716612377848,
+      "loss": 0.423,
       "step": 3750
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.00015697719869706839,
-      "loss": 0.4287,
+      "learning_rate": 0.00015701628664495113,
+      "loss": 0.4181,
       "step": 3760
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.00015658631921824104,
-      "loss": 0.423,
+      "learning_rate": 0.00015662540716612376,
+      "loss": 0.4069,
       "step": 3770
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.00015619543973941364,
-      "loss": 0.4218,
+      "learning_rate": 0.00015623452768729641,
+      "loss": 0.4112,
       "step": 3780
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.0001558045602605863,
-      "loss": 0.4107,
+      "learning_rate": 0.00015584364820846904,
+      "loss": 0.4221,
       "step": 3790
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.00015541368078175895,
-      "loss": 0.4177,
+      "learning_rate": 0.00015545276872964167,
+      "loss": 0.4318,
       "step": 3800
     },
     {
       "epoch": 2.44,
-      "eval_loss": 0.4529287815093994,
-      "eval_runtime": 405.0457,
-      "eval_samples_per_second": 4.938,
-      "eval_steps_per_second": 0.617,
+      "eval_loss": 0.4517545998096466,
+      "eval_runtime": 220.1454,
+      "eval_samples_per_second": 9.085,
+      "eval_steps_per_second": 1.136,
       "step": 3800
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.00015502280130293158,
-      "loss": 0.4184,
+      "learning_rate": 0.00015506188925081433,
+      "loss": 0.419,
       "step": 3810
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.00015463192182410424,
-      "loss": 0.418,
+      "learning_rate": 0.00015467100977198695,
+      "loss": 0.4357,
       "step": 3820
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.00015424104234527684,
-      "loss": 0.417,
+      "learning_rate": 0.0001542801302931596,
+      "loss": 0.4197,
       "step": 3830
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.0001538501628664495,
-      "loss": 0.4262,
+      "learning_rate": 0.0001538892508143322,
+      "loss": 0.4064,
       "step": 3840
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.00015345928338762212,
-      "loss": 0.4191,
+      "learning_rate": 0.00015349837133550487,
+      "loss": 0.421,
       "step": 3850
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.00015306840390879478,
-      "loss": 0.4301,
+      "learning_rate": 0.00015310749185667752,
+      "loss": 0.416,
       "step": 3860
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.00015267752442996743,
-      "loss": 0.4129,
+      "learning_rate": 0.00015271661237785015,
+      "loss": 0.3995,
       "step": 3870
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00015228664495114003,
-      "loss": 0.4272,
+      "learning_rate": 0.0001523257328990228,
+      "loss": 0.4222,
       "step": 3880
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.0001518957654723127,
-      "loss": 0.4196,
+      "learning_rate": 0.0001519348534201954,
+      "loss": 0.3989,
       "step": 3890
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00015150488599348532,
-      "loss": 0.4083,
+      "learning_rate": 0.00015154397394136806,
+      "loss": 0.3986,
       "step": 3900
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00015111400651465797,
-      "loss": 0.43,
+      "learning_rate": 0.0001511530944625407,
+      "loss": 0.4256,
       "step": 3910
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.00015072312703583063,
-      "loss": 0.4196,
+      "learning_rate": 0.00015076221498371334,
+      "loss": 0.4289,
       "step": 3920
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00015033224755700325,
-      "loss": 0.4373,
+      "learning_rate": 0.000150371335504886,
+      "loss": 0.4138,
       "step": 3930
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00014994136807817588,
-      "loss": 0.4228,
+      "learning_rate": 0.00014998045602605863,
+      "loss": 0.4125,
       "step": 3940
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.0001495504885993485,
-      "loss": 0.3963,
+      "learning_rate": 0.00014958957654723126,
+      "loss": 0.4236,
       "step": 3950
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.00014915960912052117,
-      "loss": 0.4132,
+      "learning_rate": 0.00014919869706840388,
+      "loss": 0.4189,
       "step": 3960
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.0001487687296416938,
-      "loss": 0.4273,
+      "learning_rate": 0.00014880781758957654,
+      "loss": 0.4155,
       "step": 3970
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.00014837785016286642,
-      "loss": 0.4153,
+      "learning_rate": 0.00014841693811074917,
+      "loss": 0.4144,
       "step": 3980
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.00014798697068403908,
-      "loss": 0.4379,
+      "learning_rate": 0.00014802605863192182,
+      "loss": 0.4295,
       "step": 3990
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.0001475960912052117,
-      "loss": 0.4195,
+      "learning_rate": 0.00014763517915309445,
+      "loss": 0.3985,
       "step": 4000
     },
     {
       "epoch": 2.57,
-      "eval_loss": 0.4522078335285187,
-      "eval_runtime": 403.1528,
-      "eval_samples_per_second": 4.961,
-      "eval_steps_per_second": 0.62,
+      "eval_loss": 0.4515686333179474,
+      "eval_runtime": 220.1634,
+      "eval_samples_per_second": 9.084,
+      "eval_steps_per_second": 1.136,
       "step": 4000
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.00014720521172638436,
-      "loss": 0.4242,
+      "learning_rate": 0.00014724429967426708,
+      "loss": 0.418,
       "step": 4010
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.000146814332247557,
-      "loss": 0.4173,
+      "learning_rate": 0.00014685342019543973,
+      "loss": 0.4267,
       "step": 4020
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.00014642345276872962,
-      "loss": 0.4152,
+      "learning_rate": 0.00014646254071661236,
+      "loss": 0.4231,
       "step": 4030
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.00014603257328990227,
-      "loss": 0.4143,
+      "learning_rate": 0.000146071661237785,
+      "loss": 0.4153,
       "step": 4040
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.0001456416938110749,
-      "loss": 0.4148,
+      "learning_rate": 0.00014568078175895765,
+      "loss": 0.4228,
       "step": 4050
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.00014525081433224756,
-      "loss": 0.4129,
+      "learning_rate": 0.00014528990228013027,
+      "loss": 0.4041,
       "step": 4060
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.00014485993485342018,
-      "loss": 0.4065,
+      "learning_rate": 0.00014489902280130293,
+      "loss": 0.4361,
       "step": 4070
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.0001444690553745928,
-      "loss": 0.4164,
+      "learning_rate": 0.00014450814332247556,
+      "loss": 0.4158,
       "step": 4080
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.00014407817589576547,
-      "loss": 0.4224,
+      "learning_rate": 0.00014411726384364819,
+      "loss": 0.4235,
       "step": 4090
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.0001436872964169381,
-      "loss": 0.4303,
+      "learning_rate": 0.00014372638436482084,
+      "loss": 0.4157,
       "step": 4100
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.00014329641693811072,
-      "loss": 0.4219,
+      "learning_rate": 0.00014333550488599347,
+      "loss": 0.4233,
       "step": 4110
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.00014290553745928338,
-      "loss": 0.4526,
+      "learning_rate": 0.00014294462540716612,
+      "loss": 0.4138,
       "step": 4120
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.000142514657980456,
-      "loss": 0.4018,
+      "learning_rate": 0.00014255374592833875,
+      "loss": 0.422,
       "step": 4130
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.00014212377850162866,
-      "loss": 0.4166,
+      "learning_rate": 0.00014216286644951138,
+      "loss": 0.4234,
       "step": 4140
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.0001417328990228013,
-      "loss": 0.4121,
+      "learning_rate": 0.00014177198697068404,
+      "loss": 0.4224,
       "step": 4150
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00014134201954397392,
-      "loss": 0.4295,
+      "learning_rate": 0.00014138110749185666,
+      "loss": 0.4256,
       "step": 4160
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00014095114006514657,
-      "loss": 0.4093,
+      "learning_rate": 0.0001409902280130293,
+      "loss": 0.4216,
       "step": 4170
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.0001405602605863192,
-      "loss": 0.423,
+      "learning_rate": 0.00014059934853420195,
+      "loss": 0.4139,
       "step": 4180
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.00014016938110749186,
-      "loss": 0.4403,
+      "learning_rate": 0.00014020846905537458,
+      "loss": 0.4112,
       "step": 4190
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00013977850162866449,
-      "loss": 0.419,
+      "learning_rate": 0.00013981758957654723,
+      "loss": 0.4203,
       "step": 4200
     },
     {
       "epoch": 2.7,
-      "eval_loss": 0.45203742384910583,
-      "eval_runtime": 217.2799,
-      "eval_samples_per_second": 9.205,
-      "eval_steps_per_second": 1.151,
+      "eval_loss": 0.45124533772468567,
+      "eval_runtime": 220.1249,
+      "eval_samples_per_second": 9.086,
+      "eval_steps_per_second": 1.136,
       "step": 4200
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.0001393876221498371,
-      "loss": 0.4119,
+      "learning_rate": 0.00013942671009771986,
+      "loss": 0.4072,
       "step": 4210
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.00013899674267100977,
-      "loss": 0.4062,
+      "learning_rate": 0.0001390358306188925,
+      "loss": 0.4134,
       "step": 4220
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.0001386058631921824,
-      "loss": 0.4231,
+      "learning_rate": 0.00013864495114006514,
+      "loss": 0.4175,
       "step": 4230
     },
     {
       "epoch": 2.73,
-      "learning_rate": 0.00013821498371335502,
-      "loss": 0.4285,
+      "learning_rate": 0.00013825407166123777,
+      "loss": 0.4215,
       "step": 4240
     },
     {
       "epoch": 2.73,
-      "learning_rate": 0.00013782410423452768,
-      "loss": 0.4026,
+      "learning_rate": 0.00013786319218241043,
+      "loss": 0.4226,
       "step": 4250
     },
     {
       "epoch": 2.74,
-      "learning_rate": 0.0001374332247557003,
-      "loss": 0.4212,
+      "learning_rate": 0.00013747231270358305,
+      "loss": 0.4003,
       "step": 4260
     },
     {
       "epoch": 2.75,
-      "learning_rate": 0.00013704234527687296,
-      "loss": 0.4314,
+      "learning_rate": 0.00013708143322475568,
+      "loss": 0.4198,
       "step": 4270
     },
     {
       "epoch": 2.75,
-      "learning_rate": 0.0001366514657980456,
-      "loss": 0.4181,
+      "learning_rate": 0.00013669055374592834,
+      "loss": 0.4225,
       "step": 4280
     },
     {
       "epoch": 2.76,
-      "learning_rate": 0.00013626058631921822,
-      "loss": 0.4077,
+      "learning_rate": 0.00013629967426710097,
+      "loss": 0.418,
       "step": 4290
     },
     {
       "epoch": 2.77,
-      "learning_rate": 0.00013586970684039087,
-      "loss": 0.4252,
+      "learning_rate": 0.0001359087947882736,
+      "loss": 0.4234,
       "step": 4300
     },
     {
       "epoch": 2.77,
-      "learning_rate": 0.0001354788273615635,
-      "loss": 0.4166,
+      "learning_rate": 0.00013551791530944622,
+      "loss": 0.421,
       "step": 4310
     },
     {
       "epoch": 2.78,
-      "learning_rate": 0.00013508794788273616,
-      "loss": 0.4277,
+      "learning_rate": 0.00013512703583061888,
+      "loss": 0.4162,
       "step": 4320
     },
     {
       "epoch": 2.78,
-      "learning_rate": 0.0001346970684039088,
-      "loss": 0.4255,
+      "learning_rate": 0.00013473615635179153,
+      "loss": 0.4103,
       "step": 4330
     },
     {
       "epoch": 2.79,
-      "learning_rate": 0.00013430618892508141,
-      "loss": 0.4257,
+      "learning_rate": 0.00013434527687296416,
+      "loss": 0.4157,
       "step": 4340
     },
     {
       "epoch": 2.8,
-      "learning_rate": 0.00013391530944625407,
-      "loss": 0.4236,
+      "learning_rate": 0.0001339543973941368,
+      "loss": 0.4144,
       "step": 4350
     },
     {
       "epoch": 2.8,
-      "learning_rate": 0.0001335244299674267,
-      "loss": 0.411,
+      "learning_rate": 0.00013356351791530944,
+      "loss": 0.4096,
       "step": 4360
     },
     {
       "epoch": 2.81,
-      "learning_rate": 0.00013313355048859933,
-      "loss": 0.4317,
+      "learning_rate": 0.00013317263843648207,
+      "loss": 0.4134,
       "step": 4370
     },
     {
       "epoch": 2.82,
-      "learning_rate": 0.00013274267100977198,
-      "loss": 0.4188,
+      "learning_rate": 0.00013278175895765473,
+      "loss": 0.42,
       "step": 4380
     },
     {
       "epoch": 2.82,
-      "learning_rate": 0.0001323517915309446,
-      "loss": 0.4295,
+      "learning_rate": 0.00013239087947882735,
+      "loss": 0.4187,
       "step": 4390
     },
     {
       "epoch": 2.83,
-      "learning_rate": 0.00013196091205211726,
-      "loss": 0.4345,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 0.4194,
       "step": 4400
     },
     {
       "epoch": 2.83,
-      "eval_loss": 0.4510750472545624,
-      "eval_runtime": 217.2642,
-      "eval_samples_per_second": 9.205,
-      "eval_steps_per_second": 1.151,
+      "eval_loss": 0.4500805735588074,
+      "eval_runtime": 220.1289,
+      "eval_samples_per_second": 9.086,
+      "eval_steps_per_second": 1.136,
       "step": 4400
     },
     {
       "epoch": 2.84,
-      "learning_rate": 0.0001315700325732899,
-      "loss": 0.4256,
+      "learning_rate": 0.00013160912052117264,
+      "loss": 0.4193,
       "step": 4410
     },
     {
       "epoch": 2.84,
-      "learning_rate": 0.00013117915309446252,
-      "loss": 0.4269,
+      "learning_rate": 0.00013121824104234527,
+      "loss": 0.4249,
       "step": 4420
     },
     {
       "epoch": 2.85,
-      "learning_rate": 0.00013078827361563518,
-      "loss": 0.431,
+      "learning_rate": 0.0001308273615635179,
+      "loss": 0.4187,
       "step": 4430
     },
     {
       "epoch": 2.86,
-      "learning_rate": 0.0001303973941368078,
-      "loss": 0.4019,
+      "learning_rate": 0.00013043648208469052,
+      "loss": 0.4215,
       "step": 4440
     },
     {
       "epoch": 2.86,
-      "learning_rate": 0.00013000651465798046,
-      "loss": 0.431,
+      "learning_rate": 0.00013004560260586318,
+      "loss": 0.4075,
       "step": 4450
     },
     {
       "epoch": 2.87,
-      "learning_rate": 0.0001296156351791531,
-      "loss": 0.4039,
+      "learning_rate": 0.00012965472312703583,
+      "loss": 0.4165,
       "step": 4460
     },
     {
       "epoch": 2.87,
-      "learning_rate": 0.00012922475570032572,
-      "loss": 0.4099,
+      "learning_rate": 0.00012926384364820846,
+      "loss": 0.4197,
       "step": 4470
     },
     {
       "epoch": 2.88,
-      "learning_rate": 0.00012883387622149837,
-      "loss": 0.3961,
+      "learning_rate": 0.0001288729641693811,
+      "loss": 0.4392,
       "step": 4480
     },
     {
       "epoch": 2.89,
-      "learning_rate": 0.000128442996742671,
-      "loss": 0.4312,
+      "learning_rate": 0.00012848208469055372,
+      "loss": 0.4064,
       "step": 4490
     },
     {
       "epoch": 2.89,
-      "learning_rate": 0.00012805211726384363,
-      "loss": 0.4248,
+      "learning_rate": 0.00012809120521172637,
+      "loss": 0.4176,
       "step": 4500
     },
     {
       "epoch": 2.9,
-      "learning_rate": 0.00012766123778501626,
-      "loss": 0.4114,
+      "learning_rate": 0.00012770032573289903,
+      "loss": 0.4116,
       "step": 4510
     },
     {
       "epoch": 2.91,
-      "learning_rate": 0.0001272703583061889,
-      "loss": 0.4213,
+      "learning_rate": 0.00012730944625407166,
+      "loss": 0.4177,
       "step": 4520
     },
     {
       "epoch": 2.91,
-      "learning_rate": 0.00012687947882736157,
-      "loss": 0.4118,
+      "learning_rate": 0.00012691856677524428,
+      "loss": 0.4109,
       "step": 4530
     },
     {
       "epoch": 2.92,
-      "learning_rate": 0.0001264885993485342,
-      "loss": 0.4281,
+      "learning_rate": 0.00012652768729641694,
+      "loss": 0.4072,
       "step": 4540
     },
     {
       "epoch": 2.93,
-      "learning_rate": 0.00012609771986970682,
-      "loss": 0.4116,
+      "learning_rate": 0.00012613680781758957,
+      "loss": 0.435,
       "step": 4550
     },
     {
       "epoch": 2.93,
-      "learning_rate": 0.00012570684039087945,
-      "loss": 0.4311,
+      "learning_rate": 0.0001257459283387622,
+      "loss": 0.401,
       "step": 4560
     },
     {
       "epoch": 2.94,
-      "learning_rate": 0.0001253159609120521,
-      "loss": 0.4187,
+      "learning_rate": 0.00012535504885993482,
+      "loss": 0.4175,
       "step": 4570
     },
     {
       "epoch": 2.95,
-      "learning_rate": 0.00012492508143322476,
-      "loss": 0.4101,
+      "learning_rate": 0.00012496416938110748,
+      "loss": 0.4245,
       "step": 4580
     },
     {
       "epoch": 2.95,
-      "learning_rate": 0.0001245342019543974,
-      "loss": 0.4029,
+      "learning_rate": 0.00012457328990228013,
+      "loss": 0.415,
       "step": 4590
     },
     {
       "epoch": 2.96,
-      "learning_rate": 0.00012414332247557002,
-      "loss": 0.4211,
+      "learning_rate": 0.00012418241042345276,
+      "loss": 0.4247,
       "step": 4600
     },
     {
       "epoch": 2.96,
-      "eval_loss": 0.4508701264858246,
-      "eval_runtime": 218.851,
-      "eval_samples_per_second": 9.139,
-      "eval_steps_per_second": 1.142,
+      "eval_loss": 0.44935256242752075,
+      "eval_runtime": 220.1783,
+      "eval_samples_per_second": 9.084,
+      "eval_steps_per_second": 1.135,
       "step": 4600
     }
   ],
@@ -2957,7 +2957,7 @@
   "max_steps": 7775,
   "num_train_epochs": 5,
   "save_steps": 200,
-  "total_flos": 2.3380063406159954e+18,
+  "total_flos": 2.3487809765824266e+18,
   "trial_name": null,
   "trial_params": null
 }

with_input/1b_52k/best_model/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a596f2d7e468432593e0605737d3cff6d7ba32b0ca82291a3251c26b06f9230b
+oid sha256:c25b69d8a4a964f3001386a5850d104ae3b22d7301634b84f8f3687fee78877b
 size 4027

with_input/decomp_code_with_intermediates/best_model/adapter_config.json

@@ -0,0 +1,21 @@
+{
+  "base_model_name_or_path": "huggyllama/llama-7b",
+  "bias": "none",
+  "enable_lora": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "lora_alpha": 16,
+  "lora_dropout": 0.05,
+  "merge_weights": false,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 16,
+  "target_modules": [
+    "q_proj",
+    "k_proj",
+    "v_proj",
+    "o_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

with_input/decomp_code_with_intermediates/best_model/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:76aa0870af36dca253bba8ee06ec331a5604fc78803f5181836c06702e84c742
+size 67201357

with_input/decomp_code_with_intermediates/best_model/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:595206e644bd8c27afa495804726feb8f34923038c2531af6f7cfb9bd0aea415
+size 134433093

with_input/decomp_code_with_intermediates/best_model/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:65c022292c1fb262c54e2784716b052f1a6d5fde0e8354f090107d64cdaffcd1
+size 14575

with_input/decomp_code_with_intermediates/best_model/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bdbb310518f28dfe4320b99ab0c420cbd81f1e17ce9a42136205bc4d89821c91
+size 627

with_input/decomp_code_with_intermediates/best_model/trainer_state.json

@@ -0,0 +1,3987 @@
+{
+  "best_metric": 0.27837860584259033,
+  "best_model_checkpoint": "checkpoints/instrucode/with_input/decomp_code_with_intermediates/checkpoint-6200",
+  "epoch": 3.987138263665595,
+  "eval_steps": 200,
+  "global_step": 6200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 1.0679,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 1.049,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 0.8986,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000117,
+      "loss": 0.6797,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000147,
+      "loss": 0.5126,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00017699999999999997,
+      "loss": 0.4582,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00020699999999999996,
+      "loss": 0.4072,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000237,
+      "loss": 0.3944,
+      "step": 80
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000267,
+      "loss": 0.3726,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029699999999999996,
+      "loss": 0.3735,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029964820846905535,
+      "loss": 0.3596,
+      "step": 110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000299257328990228,
+      "loss": 0.351,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029886644951140066,
+      "loss": 0.3575,
+      "step": 130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029847557003257326,
+      "loss": 0.3498,
+      "step": 140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002980846905537459,
+      "loss": 0.3383,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029769381107491857,
+      "loss": 0.3487,
+      "step": 160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029730293159609117,
+      "loss": 0.3397,
+      "step": 170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002969120521172638,
+      "loss": 0.3469,
+      "step": 180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002965211726384364,
+      "loss": 0.3317,
+      "step": 190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002961302931596091,
+      "loss": 0.3329,
+      "step": 200
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 0.331393301486969,
+      "eval_runtime": 441.8972,
+      "eval_samples_per_second": 4.526,
+      "eval_steps_per_second": 0.566,
+      "step": 200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029573941368078174,
+      "loss": 0.3368,
+      "step": 210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002953485342019544,
+      "loss": 0.3259,
+      "step": 220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029495765472312705,
+      "loss": 0.3223,
+      "step": 230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029456677524429965,
+      "loss": 0.3383,
+      "step": 240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002941758957654723,
+      "loss": 0.3239,
+      "step": 250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002937850162866449,
+      "loss": 0.341,
+      "step": 260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029339413680781756,
+      "loss": 0.3222,
+      "step": 270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002930032573289902,
+      "loss": 0.3235,
+      "step": 280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002926123778501628,
+      "loss": 0.3343,
+      "step": 290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029222149837133547,
+      "loss": 0.3332,
+      "step": 300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002918306188925081,
+      "loss": 0.3198,
+      "step": 310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002914397394136808,
+      "loss": 0.3269,
+      "step": 320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002910488599348534,
+      "loss": 0.3229,
+      "step": 330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029065798045602604,
+      "loss": 0.3284,
+      "step": 340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029026710097719864,
+      "loss": 0.3202,
+      "step": 350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002898762214983713,
+      "loss": 0.3214,
+      "step": 360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00028948534201954395,
+      "loss": 0.3169,
+      "step": 370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002890944625407166,
+      "loss": 0.3177,
+      "step": 380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002887035830618892,
+      "loss": 0.3254,
+      "step": 390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00028831270358306186,
+      "loss": 0.3235,
+      "step": 400
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 0.31607797741889954,
+      "eval_runtime": 442.0436,
+      "eval_samples_per_second": 4.524,
+      "eval_steps_per_second": 0.566,
+      "step": 400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002879218241042345,
+      "loss": 0.323,
+      "step": 410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002875309446254071,
+      "loss": 0.3077,
+      "step": 420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002871400651465798,
+      "loss": 0.3188,
+      "step": 430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00028674918566775243,
+      "loss": 0.3265,
+      "step": 440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00028635830618892503,
+      "loss": 0.3087,
+      "step": 450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002859674267100977,
+      "loss": 0.3102,
+      "step": 460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00028557654723127034,
+      "loss": 0.3135,
+      "step": 470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000285185667752443,
+      "loss": 0.317,
+      "step": 480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00028479478827361565,
+      "loss": 0.3141,
+      "step": 490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00028440390879478825,
+      "loss": 0.3219,
+      "step": 500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002840130293159609,
+      "loss": 0.3121,
+      "step": 510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002836221498371335,
+      "loss": 0.3111,
+      "step": 520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00028323127035830616,
+      "loss": 0.3212,
+      "step": 530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002828403908794788,
+      "loss": 0.3104,
+      "step": 540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002824495114006514,
+      "loss": 0.3152,
+      "step": 550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002820586319218241,
+      "loss": 0.3197,
+      "step": 560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028166775244299673,
+      "loss": 0.3199,
+      "step": 570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002812768729641694,
+      "loss": 0.3099,
+      "step": 580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000280885993485342,
+      "loss": 0.31,
+      "step": 590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028049511400651464,
+      "loss": 0.3109,
+      "step": 600
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 0.3083530366420746,
+      "eval_runtime": 441.735,
+      "eval_samples_per_second": 4.528,
+      "eval_steps_per_second": 0.566,
+      "step": 600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028010423452768724,
+      "loss": 0.3041,
+      "step": 610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002797133550488599,
+      "loss": 0.3056,
+      "step": 620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00027932247557003255,
+      "loss": 0.3102,
+      "step": 630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002789315960912052,
+      "loss": 0.3156,
+      "step": 640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002785407166123778,
+      "loss": 0.3121,
+      "step": 650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00027814983713355046,
+      "loss": 0.3068,
+      "step": 660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002777589576547231,
+      "loss": 0.3037,
+      "step": 670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002773680781758957,
+      "loss": 0.3103,
+      "step": 680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002769771986970684,
+      "loss": 0.308,
+      "step": 690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000276586319218241,
+      "loss": 0.3005,
+      "step": 700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00027619543973941363,
+      "loss": 0.3054,
+      "step": 710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002758045602605863,
+      "loss": 0.3087,
+      "step": 720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00027541368078175894,
+      "loss": 0.3071,
+      "step": 730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002750228013029316,
+      "loss": 0.3156,
+      "step": 740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002746319218241042,
+      "loss": 0.3072,
+      "step": 750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00027424104234527685,
+      "loss": 0.3174,
+      "step": 760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002738501628664495,
+      "loss": 0.3064,
+      "step": 770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002734592833876221,
+      "loss": 0.3049,
+      "step": 780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027306840390879477,
+      "loss": 0.3014,
+      "step": 790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00027267752442996737,
+      "loss": 0.3047,
+      "step": 800
+    },
+    {
+      "epoch": 0.51,
+      "eval_loss": 0.3024560511112213,
+      "eval_runtime": 441.654,
+      "eval_samples_per_second": 4.528,
+      "eval_steps_per_second": 0.566,
+      "step": 800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00027228664495114,
+      "loss": 0.3075,
+      "step": 810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002718957654723127,
+      "loss": 0.3149,
+      "step": 820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00027150488599348533,
+      "loss": 0.3084,
+      "step": 830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000271114006514658,
+      "loss": 0.3046,
+      "step": 840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002707231270358306,
+      "loss": 0.296,
+      "step": 850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00027033224755700324,
+      "loss": 0.3089,
+      "step": 860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00026994136807817584,
+      "loss": 0.3074,
+      "step": 870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002695504885993485,
+      "loss": 0.2972,
+      "step": 880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00026915960912052116,
+      "loss": 0.2911,
+      "step": 890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002687687296416938,
+      "loss": 0.3147,
+      "step": 900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002683778501628664,
+      "loss": 0.2985,
+      "step": 910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00026798697068403907,
+      "loss": 0.3073,
+      "step": 920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002675960912052117,
+      "loss": 0.304,
+      "step": 930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002672052117263843,
+      "loss": 0.2991,
+      "step": 940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000266814332247557,
+      "loss": 0.2949,
+      "step": 950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002664234527687296,
+      "loss": 0.2962,
+      "step": 960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00026603257328990223,
+      "loss": 0.3013,
+      "step": 970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002656416938110749,
+      "loss": 0.3004,
+      "step": 980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00026525081433224755,
+      "loss": 0.2941,
+      "step": 990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002648599348534202,
+      "loss": 0.2914,
+      "step": 1000
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 0.29872822761535645,
+      "eval_runtime": 442.1635,
+      "eval_samples_per_second": 4.523,
+      "eval_steps_per_second": 0.565,
+      "step": 1000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002644690553745928,
+      "loss": 0.2975,
+      "step": 1010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026407817589576546,
+      "loss": 0.2966,
+      "step": 1020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002636872964169381,
+      "loss": 0.2993,
+      "step": 1030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002632964169381107,
+      "loss": 0.3016,
+      "step": 1040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026290553745928337,
+      "loss": 0.303,
+      "step": 1050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026251465798045597,
+      "loss": 0.3021,
+      "step": 1060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002621237785016286,
+      "loss": 0.2993,
+      "step": 1070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002617328990228013,
+      "loss": 0.3008,
+      "step": 1080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00026134201954397393,
+      "loss": 0.2929,
+      "step": 1090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002609511400651466,
+      "loss": 0.2946,
+      "step": 1100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002605602605863192,
+      "loss": 0.3,
+      "step": 1110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00026016938110749185,
+      "loss": 0.3008,
+      "step": 1120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00025977850162866445,
+      "loss": 0.3037,
+      "step": 1130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002593876221498371,
+      "loss": 0.3102,
+      "step": 1140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00025899674267100976,
+      "loss": 0.3065,
+      "step": 1150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00025860586319218236,
+      "loss": 0.2961,
+      "step": 1160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000258214983713355,
+      "loss": 0.3014,
+      "step": 1170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00025782410423452767,
+      "loss": 0.3041,
+      "step": 1180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002574332247557003,
+      "loss": 0.2905,
+      "step": 1190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002570423452768729,
+      "loss": 0.2957,
+      "step": 1200
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 0.2957862615585327,
+      "eval_runtime": 442.3344,
+      "eval_samples_per_second": 4.521,
+      "eval_steps_per_second": 0.565,
+      "step": 1200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002566514657980456,
+      "loss": 0.2933,
+      "step": 1210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002562605863192182,
+      "loss": 0.3107,
+      "step": 1220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00025586970684039084,
+      "loss": 0.2954,
+      "step": 1230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002554788273615635,
+      "loss": 0.308,
+      "step": 1240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00025508794788273615,
+      "loss": 0.302,
+      "step": 1250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002546970684039088,
+      "loss": 0.3004,
+      "step": 1260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002543061889250814,
+      "loss": 0.3029,
+      "step": 1270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00025391530944625406,
+      "loss": 0.2893,
+      "step": 1280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002535244299674267,
+      "loss": 0.3027,
+      "step": 1290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002531335504885993,
+      "loss": 0.2948,
+      "step": 1300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00025274267100977197,
+      "loss": 0.2896,
+      "step": 1310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00025235179153094457,
+      "loss": 0.303,
+      "step": 1320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00025196091205211723,
+      "loss": 0.3001,
+      "step": 1330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002515700325732899,
+      "loss": 0.2944,
+      "step": 1340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00025117915309446254,
+      "loss": 0.3023,
+      "step": 1350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002507882736156352,
+      "loss": 0.3007,
+      "step": 1360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002503973941368078,
+      "loss": 0.3024,
+      "step": 1370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00025000651465798045,
+      "loss": 0.3012,
+      "step": 1380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00024961563517915305,
+      "loss": 0.294,
+      "step": 1390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002492247557003257,
+      "loss": 0.2987,
+      "step": 1400
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 0.2930598855018616,
+      "eval_runtime": 442.2601,
+      "eval_samples_per_second": 4.522,
+      "eval_steps_per_second": 0.565,
+      "step": 1400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00024883387622149836,
+      "loss": 0.2973,
+      "step": 1410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00024844299674267096,
+      "loss": 0.2954,
+      "step": 1420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002480521172638436,
+      "loss": 0.2907,
+      "step": 1430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00024766123778501627,
+      "loss": 0.2973,
+      "step": 1440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00024727035830618893,
+      "loss": 0.2991,
+      "step": 1450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00024687947882736153,
+      "loss": 0.2918,
+      "step": 1460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002464885993485342,
+      "loss": 0.2892,
+      "step": 1470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002460977198697068,
+      "loss": 0.2965,
+      "step": 1480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00024570684039087944,
+      "loss": 0.2965,
+      "step": 1490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002453159609120521,
+      "loss": 0.2925,
+      "step": 1500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00024492508143322475,
+      "loss": 0.292,
+      "step": 1510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00024453420195439735,
+      "loss": 0.2945,
+      "step": 1520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00024414332247557,
+      "loss": 0.2955,
+      "step": 1530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00024375244299674263,
+      "loss": 0.308,
+      "step": 1540
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002433615635179153,
+      "loss": 0.294,
+      "step": 1550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00024297068403908795,
+      "loss": 0.2937,
+      "step": 1560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00024257980456026055,
+      "loss": 0.2781,
+      "step": 1570
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002421889250814332,
+      "loss": 0.2763,
+      "step": 1580
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00024179804560260583,
+      "loss": 0.2842,
+      "step": 1590
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00024140716612377849,
+      "loss": 0.2838,
+      "step": 1600
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 0.2916465997695923,
+      "eval_runtime": 441.7075,
+      "eval_samples_per_second": 4.528,
+      "eval_steps_per_second": 0.566,
+      "step": 1600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00024101628664495114,
+      "loss": 0.2763,
+      "step": 1610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00024062540716612377,
+      "loss": 0.2812,
+      "step": 1620
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002402345276872964,
+      "loss": 0.2889,
+      "step": 1630
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00023984364820846902,
+      "loss": 0.2901,
+      "step": 1640
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00023945276872964168,
+      "loss": 0.2688,
+      "step": 1650
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002390618892508143,
+      "loss": 0.2856,
+      "step": 1660
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00023867100977198696,
+      "loss": 0.2901,
+      "step": 1670
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00023828013029315956,
+      "loss": 0.2892,
+      "step": 1680
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00023788925081433222,
+      "loss": 0.281,
+      "step": 1690
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00023749837133550487,
+      "loss": 0.284,
+      "step": 1700
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002371074918566775,
+      "loss": 0.2834,
+      "step": 1710
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00023671661237785016,
+      "loss": 0.2832,
+      "step": 1720
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00023632573289902276,
+      "loss": 0.2803,
+      "step": 1730
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00023593485342019541,
+      "loss": 0.2868,
+      "step": 1740
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00023554397394136807,
+      "loss": 0.2883,
+      "step": 1750
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002351530944625407,
+      "loss": 0.2859,
+      "step": 1760
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00023476221498371335,
+      "loss": 0.2888,
+      "step": 1770
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00023437133550488595,
+      "loss": 0.2983,
+      "step": 1780
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002339804560260586,
+      "loss": 0.2809,
+      "step": 1790
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00023358957654723124,
+      "loss": 0.2982,
+      "step": 1800
+    },
+    {
+      "epoch": 1.16,
+      "eval_loss": 0.28945282101631165,
+      "eval_runtime": 441.6135,
+      "eval_samples_per_second": 4.529,
+      "eval_steps_per_second": 0.566,
+      "step": 1800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002331986970684039,
+      "loss": 0.2848,
+      "step": 1810
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00023280781758957655,
+      "loss": 0.2821,
+      "step": 1820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00023241693811074915,
+      "loss": 0.2809,
+      "step": 1830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002320260586319218,
+      "loss": 0.2858,
+      "step": 1840
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00023163517915309443,
+      "loss": 0.2751,
+      "step": 1850
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002312442996742671,
+      "loss": 0.2779,
+      "step": 1860
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00023085342019543974,
+      "loss": 0.2797,
+      "step": 1870
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00023046254071661234,
+      "loss": 0.2888,
+      "step": 1880
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000230071661237785,
+      "loss": 0.2822,
+      "step": 1890
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00022968078175895763,
+      "loss": 0.2872,
+      "step": 1900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00022928990228013028,
+      "loss": 0.2921,
+      "step": 1910
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002288990228013029,
+      "loss": 0.2914,
+      "step": 1920
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00022850814332247554,
+      "loss": 0.2935,
+      "step": 1930
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00022811726384364817,
+      "loss": 0.2845,
+      "step": 1940
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00022772638436482082,
+      "loss": 0.3053,
+      "step": 1950
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00022733550488599348,
+      "loss": 0.3502,
+      "step": 1960
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002269446254071661,
+      "loss": 0.2942,
+      "step": 1970
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00022655374592833873,
+      "loss": 0.2895,
+      "step": 1980
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00022616286644951136,
+      "loss": 0.281,
+      "step": 1990
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00022577198697068402,
+      "loss": 0.2839,
+      "step": 2000
+    },
+    {
+      "epoch": 1.29,
+      "eval_loss": 0.28936389088630676,
+      "eval_runtime": 441.8381,
+      "eval_samples_per_second": 4.527,
+      "eval_steps_per_second": 0.566,
+      "step": 2000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00022538110749185667,
+      "loss": 0.2832,
+      "step": 2010
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002249902280130293,
+      "loss": 0.28,
+      "step": 2020
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00022459934853420196,
+      "loss": 0.2788,
+      "step": 2030
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00022420846905537456,
+      "loss": 0.298,
+      "step": 2040
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002238175895765472,
+      "loss": 0.2815,
+      "step": 2050
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00022342671009771984,
+      "loss": 0.2866,
+      "step": 2060
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002230358306188925,
+      "loss": 0.2839,
+      "step": 2070
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00022264495114006515,
+      "loss": 0.285,
+      "step": 2080
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00022225407166123775,
+      "loss": 0.2846,
+      "step": 2090
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002218631921824104,
+      "loss": 0.2903,
+      "step": 2100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00022147231270358304,
+      "loss": 0.2914,
+      "step": 2110
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002210814332247557,
+      "loss": 0.2839,
+      "step": 2120
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00022069055374592835,
+      "loss": 0.2814,
+      "step": 2130
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00022029967426710095,
+      "loss": 0.2989,
+      "step": 2140
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002199087947882736,
+      "loss": 0.286,
+      "step": 2150
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00021951791530944623,
+      "loss": 0.2875,
+      "step": 2160
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021912703583061889,
+      "loss": 0.2771,
+      "step": 2170
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021873615635179151,
+      "loss": 0.2793,
+      "step": 2180
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021834527687296414,
+      "loss": 0.2899,
+      "step": 2190
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021795439739413677,
+      "loss": 0.2871,
+      "step": 2200
+    },
+    {
+      "epoch": 1.41,
+      "eval_loss": 0.28674158453941345,
+      "eval_runtime": 441.6324,
+      "eval_samples_per_second": 4.529,
+      "eval_steps_per_second": 0.566,
+      "step": 2200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00021756351791530943,
+      "loss": 0.2856,
+      "step": 2210
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00021717263843648208,
+      "loss": 0.2838,
+      "step": 2220
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002167817589576547,
+      "loss": 0.2851,
+      "step": 2230
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00021639087947882734,
+      "loss": 0.2823,
+      "step": 2240
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 0.2872,
+      "step": 2250
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00021560912052117262,
+      "loss": 0.2782,
+      "step": 2260
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00021521824104234528,
+      "loss": 0.2722,
+      "step": 2270
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002148273615635179,
+      "loss": 0.2951,
+      "step": 2280
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00021443648208469053,
+      "loss": 0.2774,
+      "step": 2290
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00021404560260586316,
+      "loss": 0.2897,
+      "step": 2300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021365472312703581,
+      "loss": 0.2848,
+      "step": 2310
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00021326384364820844,
+      "loss": 0.2823,
+      "step": 2320
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002128729641693811,
+      "loss": 0.2795,
+      "step": 2330
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002124820846905537,
+      "loss": 0.275,
+      "step": 2340
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00021209120521172635,
+      "loss": 0.2789,
+      "step": 2350
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.000211700325732899,
+      "loss": 0.2887,
+      "step": 2360
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00021130944625407164,
+      "loss": 0.2871,
+      "step": 2370
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002109185667752443,
+      "loss": 0.2776,
+      "step": 2380
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00021052768729641695,
+      "loss": 0.2785,
+      "step": 2390
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00021013680781758955,
+      "loss": 0.2835,
+      "step": 2400
+    },
+    {
+      "epoch": 1.54,
+      "eval_loss": 0.2856335937976837,
+      "eval_runtime": 441.712,
+      "eval_samples_per_second": 4.528,
+      "eval_steps_per_second": 0.566,
+      "step": 2400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002097459283387622,
+      "loss": 0.2881,
+      "step": 2410
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00020935504885993483,
+      "loss": 0.276,
+      "step": 2420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002089641693811075,
+      "loss": 0.2883,
+      "step": 2430
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00020857328990228012,
+      "loss": 0.2801,
+      "step": 2440
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00020818241042345274,
+      "loss": 0.2815,
+      "step": 2450
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00020779153094462537,
+      "loss": 0.2925,
+      "step": 2460
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00020740065146579803,
+      "loss": 0.2694,
+      "step": 2470
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00020700977198697068,
+      "loss": 0.286,
+      "step": 2480
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002066188925081433,
+      "loss": 0.29,
+      "step": 2490
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00020622801302931594,
+      "loss": 0.285,
+      "step": 2500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00020583713355048857,
+      "loss": 0.2707,
+      "step": 2510
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00020544625407166122,
+      "loss": 0.2883,
+      "step": 2520
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00020505537459283388,
+      "loss": 0.2849,
+      "step": 2530
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002046644951140065,
+      "loss": 0.2827,
+      "step": 2540
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00020427361563517913,
+      "loss": 0.27,
+      "step": 2550
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00020388273615635176,
+      "loss": 0.2803,
+      "step": 2560
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00020349185667752442,
+      "loss": 0.2828,
+      "step": 2570
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00020310097719869705,
+      "loss": 0.2821,
+      "step": 2580
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002027100977198697,
+      "loss": 0.2732,
+      "step": 2590
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002023192182410423,
+      "loss": 0.2758,
+      "step": 2600
+    },
+    {
+      "epoch": 1.67,
+      "eval_loss": 0.28436511754989624,
+      "eval_runtime": 441.3288,
+      "eval_samples_per_second": 4.532,
+      "eval_steps_per_second": 0.566,
+      "step": 2600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00020192833876221496,
+      "loss": 0.2715,
+      "step": 2610
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002015374592833876,
+      "loss": 0.2741,
+      "step": 2620
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00020114657980456024,
+      "loss": 0.2856,
+      "step": 2630
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002007557003257329,
+      "loss": 0.2828,
+      "step": 2640
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002003648208469055,
+      "loss": 0.2695,
+      "step": 2650
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00019997394136807815,
+      "loss": 0.2914,
+      "step": 2660
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001995830618892508,
+      "loss": 0.2766,
+      "step": 2670
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00019919218241042344,
+      "loss": 0.2738,
+      "step": 2680
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001988013029315961,
+      "loss": 0.2745,
+      "step": 2690
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001984104234527687,
+      "loss": 0.2737,
+      "step": 2700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00019801954397394135,
+      "loss": 0.2734,
+      "step": 2710
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00019762866449511398,
+      "loss": 0.2658,
+      "step": 2720
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00019723778501628663,
+      "loss": 0.2856,
+      "step": 2730
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00019684690553745929,
+      "loss": 0.2762,
+      "step": 2740
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00019645602605863191,
+      "loss": 0.2755,
+      "step": 2750
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00019606514657980454,
+      "loss": 0.2702,
+      "step": 2760
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00019567426710097717,
+      "loss": 0.2836,
+      "step": 2770
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00019528338762214983,
+      "loss": 0.2703,
+      "step": 2780
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00019489250814332248,
+      "loss": 0.282,
+      "step": 2790
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001945016286644951,
+      "loss": 0.2797,
+      "step": 2800
+    },
+    {
+      "epoch": 1.8,
+      "eval_loss": 0.28348231315612793,
+      "eval_runtime": 441.491,
+      "eval_samples_per_second": 4.53,
+      "eval_steps_per_second": 0.566,
+      "step": 2800
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00019411074918566774,
+      "loss": 0.2853,
+      "step": 2810
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00019371986970684037,
+      "loss": 0.2854,
+      "step": 2820
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00019332899022801302,
+      "loss": 0.2772,
+      "step": 2830
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00019293811074918565,
+      "loss": 0.2756,
+      "step": 2840
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001925472312703583,
+      "loss": 0.2798,
+      "step": 2850
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001921563517915309,
+      "loss": 0.2765,
+      "step": 2860
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00019176547231270356,
+      "loss": 0.2835,
+      "step": 2870
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00019137459283387622,
+      "loss": 0.2756,
+      "step": 2880
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00019098371335504884,
+      "loss": 0.2777,
+      "step": 2890
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001905928338762215,
+      "loss": 0.2777,
+      "step": 2900
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001902019543973941,
+      "loss": 0.2737,
+      "step": 2910
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00018981107491856675,
+      "loss": 0.2804,
+      "step": 2920
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001894201954397394,
+      "loss": 0.2867,
+      "step": 2930
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00018902931596091204,
+      "loss": 0.2807,
+      "step": 2940
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001886384364820847,
+      "loss": 0.2782,
+      "step": 2950
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001882475570032573,
+      "loss": 0.2829,
+      "step": 2960
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00018785667752442995,
+      "loss": 0.2882,
+      "step": 2970
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00018746579804560258,
+      "loss": 0.2682,
+      "step": 2980
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00018707491856677523,
+      "loss": 0.2862,
+      "step": 2990
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001866840390879479,
+      "loss": 0.2813,
+      "step": 3000
+    },
+    {
+      "epoch": 1.93,
+      "eval_loss": 0.2826210558414459,
+      "eval_runtime": 441.4835,
+      "eval_samples_per_second": 4.53,
+      "eval_steps_per_second": 0.566,
+      "step": 3000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001862931596091205,
+      "loss": 0.2848,
+      "step": 3010
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00018590228013029314,
+      "loss": 0.2774,
+      "step": 3020
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00018551140065146577,
+      "loss": 0.2872,
+      "step": 3030
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00018512052117263843,
+      "loss": 0.2763,
+      "step": 3040
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00018472964169381108,
+      "loss": 0.2818,
+      "step": 3050
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00018433876221498368,
+      "loss": 0.2746,
+      "step": 3060
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00018394788273615634,
+      "loss": 0.2808,
+      "step": 3070
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00018355700325732897,
+      "loss": 0.2738,
+      "step": 3080
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00018316612377850162,
+      "loss": 0.2846,
+      "step": 3090
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00018277524429967425,
+      "loss": 0.2739,
+      "step": 3100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00018238436482084688,
+      "loss": 0.2845,
+      "step": 3110
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001819934853420195,
+      "loss": 0.2692,
+      "step": 3120
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00018160260586319216,
+      "loss": 0.2687,
+      "step": 3130
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00018121172638436482,
+      "loss": 0.2626,
+      "step": 3140
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00018082084690553745,
+      "loss": 0.2671,
+      "step": 3150
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001804299674267101,
+      "loss": 0.2731,
+      "step": 3160
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001800390879478827,
+      "loss": 0.2644,
+      "step": 3170
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00017964820846905536,
+      "loss": 0.2617,
+      "step": 3180
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.000179257328990228,
+      "loss": 0.2554,
+      "step": 3190
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00017886644951140064,
+      "loss": 0.264,
+      "step": 3200
+    },
+    {
+      "epoch": 2.06,
+      "eval_loss": 0.28276166319847107,
+      "eval_runtime": 441.3938,
+      "eval_samples_per_second": 4.531,
+      "eval_steps_per_second": 0.566,
+      "step": 3200
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001784755700325733,
+      "loss": 0.26,
+      "step": 3210
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001780846905537459,
+      "loss": 0.2689,
+      "step": 3220
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00017769381107491855,
+      "loss": 0.271,
+      "step": 3230
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00017730293159609118,
+      "loss": 0.2762,
+      "step": 3240
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00017691205211726384,
+      "loss": 0.2787,
+      "step": 3250
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001765211726384365,
+      "loss": 0.2698,
+      "step": 3260
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001761302931596091,
+      "loss": 0.2666,
+      "step": 3270
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00017573941368078175,
+      "loss": 0.2732,
+      "step": 3280
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00017534853420195438,
+      "loss": 0.266,
+      "step": 3290
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00017495765472312703,
+      "loss": 0.2619,
+      "step": 3300
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00017456677524429966,
+      "loss": 0.28,
+      "step": 3310
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001741758957654723,
+      "loss": 0.2725,
+      "step": 3320
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00017378501628664494,
+      "loss": 0.2668,
+      "step": 3330
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00017339413680781757,
+      "loss": 0.2721,
+      "step": 3340
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00017300325732899023,
+      "loss": 0.2514,
+      "step": 3350
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00017261237785016285,
+      "loss": 0.2623,
+      "step": 3360
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00017222149837133548,
+      "loss": 0.2569,
+      "step": 3370
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001718306188925081,
+      "loss": 0.2654,
+      "step": 3380
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00017143973941368077,
+      "loss": 0.2678,
+      "step": 3390
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00017104885993485342,
+      "loss": 0.2651,
+      "step": 3400
+    },
+    {
+      "epoch": 2.19,
+      "eval_loss": 0.28253164887428284,
+      "eval_runtime": 441.4474,
+      "eval_samples_per_second": 4.531,
+      "eval_steps_per_second": 0.566,
+      "step": 3400
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00017065798045602605,
+      "loss": 0.2587,
+      "step": 3410
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00017026710097719868,
+      "loss": 0.2835,
+      "step": 3420
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001698762214983713,
+      "loss": 0.272,
+      "step": 3430
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00016948534201954396,
+      "loss": 0.2707,
+      "step": 3440
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001690944625407166,
+      "loss": 0.2663,
+      "step": 3450
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00016870358306188924,
+      "loss": 0.2767,
+      "step": 3460
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00016831270358306187,
+      "loss": 0.2696,
+      "step": 3470
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001679218241042345,
+      "loss": 0.2735,
+      "step": 3480
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00016753094462540716,
+      "loss": 0.266,
+      "step": 3490
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00016714006514657978,
+      "loss": 0.2646,
+      "step": 3500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00016674918566775244,
+      "loss": 0.2671,
+      "step": 3510
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001663583061889251,
+      "loss": 0.2583,
+      "step": 3520
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001659674267100977,
+      "loss": 0.2754,
+      "step": 3530
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00016557654723127035,
+      "loss": 0.2599,
+      "step": 3540
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00016518566775244298,
+      "loss": 0.2618,
+      "step": 3550
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00016479478827361563,
+      "loss": 0.2669,
+      "step": 3560
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00016440390879478826,
+      "loss": 0.2656,
+      "step": 3570
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001640130293159609,
+      "loss": 0.2648,
+      "step": 3580
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00016362214983713352,
+      "loss": 0.2762,
+      "step": 3590
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016323127035830617,
+      "loss": 0.2697,
+      "step": 3600
+    },
+    {
+      "epoch": 2.32,
+      "eval_loss": 0.2820605933666229,
+      "eval_runtime": 441.4527,
+      "eval_samples_per_second": 4.53,
+      "eval_steps_per_second": 0.566,
+      "step": 3600
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016284039087947883,
+      "loss": 0.2711,
+      "step": 3610
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016244951140065146,
+      "loss": 0.2655,
+      "step": 3620
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016205863192182408,
+      "loss": 0.2705,
+      "step": 3630
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001616677524429967,
+      "loss": 0.2598,
+      "step": 3640
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016127687296416937,
+      "loss": 0.2608,
+      "step": 3650
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016088599348534202,
+      "loss": 0.264,
+      "step": 3660
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016049511400651465,
+      "loss": 0.2652,
+      "step": 3670
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016010423452768728,
+      "loss": 0.2541,
+      "step": 3680
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001597133550488599,
+      "loss": 0.2616,
+      "step": 3690
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00015932247557003256,
+      "loss": 0.2652,
+      "step": 3700
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001589315960912052,
+      "loss": 0.2587,
+      "step": 3710
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00015854071661237785,
+      "loss": 0.2629,
+      "step": 3720
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015814983713355045,
+      "loss": 0.2692,
+      "step": 3730
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001577589576547231,
+      "loss": 0.2683,
+      "step": 3740
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015736807817589576,
+      "loss": 0.2674,
+      "step": 3750
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015697719869706839,
+      "loss": 0.2591,
+      "step": 3760
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015658631921824104,
+      "loss": 0.2641,
+      "step": 3770
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015619543973941364,
+      "loss": 0.2683,
+      "step": 3780
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001558045602605863,
+      "loss": 0.2675,
+      "step": 3790
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015541368078175895,
+      "loss": 0.2593,
+      "step": 3800
+    },
+    {
+      "epoch": 2.44,
+      "eval_loss": 0.28158313035964966,
+      "eval_runtime": 441.4901,
+      "eval_samples_per_second": 4.53,
+      "eval_steps_per_second": 0.566,
+      "step": 3800
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015502280130293158,
+      "loss": 0.2751,
+      "step": 3810
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015463192182410424,
+      "loss": 0.2716,
+      "step": 3820
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015424104234527684,
+      "loss": 0.2696,
+      "step": 3830
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001538501628664495,
+      "loss": 0.2717,
+      "step": 3840
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015345928338762212,
+      "loss": 0.257,
+      "step": 3850
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015306840390879478,
+      "loss": 0.2632,
+      "step": 3860
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015267752442996743,
+      "loss": 0.2708,
+      "step": 3870
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015228664495114003,
+      "loss": 0.2769,
+      "step": 3880
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001518957654723127,
+      "loss": 0.2637,
+      "step": 3890
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015150488599348532,
+      "loss": 0.2651,
+      "step": 3900
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015111400651465797,
+      "loss": 0.2651,
+      "step": 3910
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015072312703583063,
+      "loss": 0.2712,
+      "step": 3920
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00015033224755700325,
+      "loss": 0.2677,
+      "step": 3930
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014994136807817588,
+      "loss": 0.2697,
+      "step": 3940
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001495504885993485,
+      "loss": 0.263,
+      "step": 3950
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014915960912052117,
+      "loss": 0.2662,
+      "step": 3960
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001487687296416938,
+      "loss": 0.258,
+      "step": 3970
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014837785016286642,
+      "loss": 0.2611,
+      "step": 3980
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014798697068403908,
+      "loss": 0.259,
+      "step": 3990
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001475960912052117,
+      "loss": 0.2618,
+      "step": 4000
+    },
+    {
+      "epoch": 2.57,
+      "eval_loss": 0.2809056341648102,
+      "eval_runtime": 442.2264,
+      "eval_samples_per_second": 4.523,
+      "eval_steps_per_second": 0.565,
+      "step": 4000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014720521172638436,
+      "loss": 0.2617,
+      "step": 4010
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000146814332247557,
+      "loss": 0.2612,
+      "step": 4020
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014642345276872962,
+      "loss": 0.253,
+      "step": 4030
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014603257328990227,
+      "loss": 0.2611,
+      "step": 4040
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001456416938110749,
+      "loss": 0.2574,
+      "step": 4050
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014525081433224756,
+      "loss": 0.2708,
+      "step": 4060
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014485993485342018,
+      "loss": 0.2639,
+      "step": 4070
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001444690553745928,
+      "loss": 0.2712,
+      "step": 4080
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014407817589576547,
+      "loss": 0.2644,
+      "step": 4090
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001436872964169381,
+      "loss": 0.2613,
+      "step": 4100
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00014329641693811072,
+      "loss": 0.2601,
+      "step": 4110
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014290553745928338,
+      "loss": 0.2741,
+      "step": 4120
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.000142514657980456,
+      "loss": 0.2568,
+      "step": 4130
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014212377850162866,
+      "loss": 0.2589,
+      "step": 4140
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001417328990228013,
+      "loss": 0.2781,
+      "step": 4150
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014134201954397392,
+      "loss": 0.2699,
+      "step": 4160
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014095114006514657,
+      "loss": 0.2783,
+      "step": 4170
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001405602605863192,
+      "loss": 0.2638,
+      "step": 4180
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014016938110749186,
+      "loss": 0.2602,
+      "step": 4190
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013977850162866449,
+      "loss": 0.2725,
+      "step": 4200
+    },
+    {
+      "epoch": 2.7,
+      "eval_loss": 0.28008410334587097,
+      "eval_runtime": 441.5362,
+      "eval_samples_per_second": 4.53,
+      "eval_steps_per_second": 0.566,
+      "step": 4200
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001393876221498371,
+      "loss": 0.2577,
+      "step": 4210
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013899674267100977,
+      "loss": 0.2706,
+      "step": 4220
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001386058631921824,
+      "loss": 0.2724,
+      "step": 4230
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013821498371335502,
+      "loss": 0.2638,
+      "step": 4240
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013782410423452768,
+      "loss": 0.2544,
+      "step": 4250
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001374332247557003,
+      "loss": 0.2645,
+      "step": 4260
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00013704234527687296,
+      "loss": 0.2652,
+      "step": 4270
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001366514657980456,
+      "loss": 0.2785,
+      "step": 4280
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00013626058631921822,
+      "loss": 0.2722,
+      "step": 4290
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00013586970684039087,
+      "loss": 0.27,
+      "step": 4300
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001354788273615635,
+      "loss": 0.2576,
+      "step": 4310
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00013508794788273616,
+      "loss": 0.2617,
+      "step": 4320
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001346970684039088,
+      "loss": 0.2658,
+      "step": 4330
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00013430618892508141,
+      "loss": 0.2616,
+      "step": 4340
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00013391530944625407,
+      "loss": 0.2638,
+      "step": 4350
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001335244299674267,
+      "loss": 0.2737,
+      "step": 4360
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00013313355048859933,
+      "loss": 0.278,
+      "step": 4370
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00013274267100977198,
+      "loss": 0.2753,
+      "step": 4380
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0001323517915309446,
+      "loss": 0.2553,
+      "step": 4390
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00013196091205211726,
+      "loss": 0.2618,
+      "step": 4400
+    },
+    {
+      "epoch": 2.83,
+      "eval_loss": 0.27958813309669495,
+      "eval_runtime": 441.9512,
+      "eval_samples_per_second": 4.525,
+      "eval_steps_per_second": 0.566,
+      "step": 4400
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001315700325732899,
+      "loss": 0.2588,
+      "step": 4410
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00013117915309446252,
+      "loss": 0.2558,
+      "step": 4420
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00013078827361563518,
+      "loss": 0.269,
+      "step": 4430
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001303973941368078,
+      "loss": 0.2547,
+      "step": 4440
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00013000651465798046,
+      "loss": 0.2667,
+      "step": 4450
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001296156351791531,
+      "loss": 0.2641,
+      "step": 4460
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00012922475570032572,
+      "loss": 0.2585,
+      "step": 4470
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00012883387622149837,
+      "loss": 0.2565,
+      "step": 4480
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.000128442996742671,
+      "loss": 0.2728,
+      "step": 4490
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00012805211726384363,
+      "loss": 0.2573,
+      "step": 4500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00012766123778501626,
+      "loss": 0.2633,
+      "step": 4510
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0001272703583061889,
+      "loss": 0.2582,
+      "step": 4520
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00012687947882736157,
+      "loss": 0.2625,
+      "step": 4530
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0001264885993485342,
+      "loss": 0.2672,
+      "step": 4540
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00012609771986970682,
+      "loss": 0.2667,
+      "step": 4550
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00012570684039087945,
+      "loss": 0.2624,
+      "step": 4560
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001253159609120521,
+      "loss": 0.2618,
+      "step": 4570
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00012492508143322476,
+      "loss": 0.2639,
+      "step": 4580
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0001245342019543974,
+      "loss": 0.2755,
+      "step": 4590
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00012414332247557002,
+      "loss": 0.2753,
+      "step": 4600
+    },
+    {
+      "epoch": 2.96,
+      "eval_loss": 0.27895185351371765,
+      "eval_runtime": 441.8565,
+      "eval_samples_per_second": 4.526,
+      "eval_steps_per_second": 0.566,
+      "step": 4600
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00012375244299674267,
+      "loss": 0.267,
+      "step": 4610
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001233615635179153,
+      "loss": 0.2726,
+      "step": 4620
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00012297068403908793,
+      "loss": 0.2629,
+      "step": 4630
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00012257980456026056,
+      "loss": 0.2636,
+      "step": 4640
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001221889250814332,
+      "loss": 0.2676,
+      "step": 4650
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00012179804560260585,
+      "loss": 0.2575,
+      "step": 4660
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00012140716612377848,
+      "loss": 0.2478,
+      "step": 4670
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00012101628664495114,
+      "loss": 0.2558,
+      "step": 4680
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00012062540716612377,
+      "loss": 0.2479,
+      "step": 4690
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00012023452768729641,
+      "loss": 0.2592,
+      "step": 4700
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00011984364820846905,
+      "loss": 0.2509,
+      "step": 4710
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00011945276872964168,
+      "loss": 0.2567,
+      "step": 4720
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00011906188925081432,
+      "loss": 0.2404,
+      "step": 4730
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00011867100977198695,
+      "loss": 0.2438,
+      "step": 4740
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0001182801302931596,
+      "loss": 0.2502,
+      "step": 4750
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00011788925081433224,
+      "loss": 0.2543,
+      "step": 4760
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00011749837133550487,
+      "loss": 0.252,
+      "step": 4770
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00011710749185667751,
+      "loss": 0.2583,
+      "step": 4780
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00011671661237785016,
+      "loss": 0.2575,
+      "step": 4790
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00011632573289902278,
+      "loss": 0.2559,
+      "step": 4800
+    },
+    {
+      "epoch": 3.09,
+      "eval_loss": 0.28064459562301636,
+      "eval_runtime": 442.0127,
+      "eval_samples_per_second": 4.525,
+      "eval_steps_per_second": 0.566,
+      "step": 4800
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00011593485342019544,
+      "loss": 0.2543,
+      "step": 4810
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00011554397394136807,
+      "loss": 0.2478,
+      "step": 4820
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00011515309446254071,
+      "loss": 0.2502,
+      "step": 4830
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00011476221498371335,
+      "loss": 0.255,
+      "step": 4840
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00011437133550488598,
+      "loss": 0.2542,
+      "step": 4850
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00011398045602605862,
+      "loss": 0.2562,
+      "step": 4860
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00011358957654723125,
+      "loss": 0.2463,
+      "step": 4870
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0001131986970684039,
+      "loss": 0.2592,
+      "step": 4880
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00011280781758957655,
+      "loss": 0.2601,
+      "step": 4890
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00011241693811074917,
+      "loss": 0.2548,
+      "step": 4900
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00011202605863192181,
+      "loss": 0.2569,
+      "step": 4910
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00011163517915309444,
+      "loss": 0.2543,
+      "step": 4920
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00011124429967426708,
+      "loss": 0.2532,
+      "step": 4930
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00011085342019543974,
+      "loss": 0.2493,
+      "step": 4940
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00011046254071661237,
+      "loss": 0.2533,
+      "step": 4950
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00011007166123778501,
+      "loss": 0.2632,
+      "step": 4960
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00010968078175895765,
+      "loss": 0.2457,
+      "step": 4970
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00010928990228013028,
+      "loss": 0.2538,
+      "step": 4980
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00010889902280130292,
+      "loss": 0.2536,
+      "step": 4990
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00010850814332247555,
+      "loss": 0.2555,
+      "step": 5000
+    },
+    {
+      "epoch": 3.22,
+      "eval_loss": 0.2805168330669403,
+      "eval_runtime": 442.1047,
+      "eval_samples_per_second": 4.524,
+      "eval_steps_per_second": 0.565,
+      "step": 5000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0001081172638436482,
+      "loss": 0.253,
+      "step": 5010
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00010772638436482085,
+      "loss": 0.255,
+      "step": 5020
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00010733550488599347,
+      "loss": 0.2529,
+      "step": 5030
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00010694462540716612,
+      "loss": 0.2541,
+      "step": 5040
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00010655374592833874,
+      "loss": 0.2576,
+      "step": 5050
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00010616286644951139,
+      "loss": 0.2408,
+      "step": 5060
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00010577198697068404,
+      "loss": 0.259,
+      "step": 5070
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00010538110749185667,
+      "loss": 0.2534,
+      "step": 5080
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00010499022801302931,
+      "loss": 0.2491,
+      "step": 5090
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00010459934853420194,
+      "loss": 0.2507,
+      "step": 5100
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00010420846905537458,
+      "loss": 0.2583,
+      "step": 5110
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00010381758957654722,
+      "loss": 0.2478,
+      "step": 5120
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00010342671009771985,
+      "loss": 0.2574,
+      "step": 5130
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0001030358306188925,
+      "loss": 0.2578,
+      "step": 5140
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00010264495114006515,
+      "loss": 0.2561,
+      "step": 5150
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00010225407166123778,
+      "loss": 0.257,
+      "step": 5160
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00010186319218241042,
+      "loss": 0.2458,
+      "step": 5170
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00010147231270358305,
+      "loss": 0.2484,
+      "step": 5180
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00010108143322475569,
+      "loss": 0.2503,
+      "step": 5190
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00010069055374592834,
+      "loss": 0.2661,
+      "step": 5200
+    },
+    {
+      "epoch": 3.34,
+      "eval_loss": 0.2801346182823181,
+      "eval_runtime": 442.4045,
+      "eval_samples_per_second": 4.521,
+      "eval_steps_per_second": 0.565,
+      "step": 5200
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00010029967426710097,
+      "loss": 0.2526,
+      "step": 5210
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 9.990879478827361e-05,
+      "loss": 0.25,
+      "step": 5220
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 9.951791530944624e-05,
+      "loss": 0.2536,
+      "step": 5230
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 9.912703583061888e-05,
+      "loss": 0.2444,
+      "step": 5240
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 9.873615635179152e-05,
+      "loss": 0.2491,
+      "step": 5250
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 9.834527687296415e-05,
+      "loss": 0.2494,
+      "step": 5260
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 9.795439739413681e-05,
+      "loss": 0.251,
+      "step": 5270
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.756351791530944e-05,
+      "loss": 0.2453,
+      "step": 5280
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.717263843648208e-05,
+      "loss": 0.2524,
+      "step": 5290
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.678175895765472e-05,
+      "loss": 0.2535,
+      "step": 5300
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.639087947882735e-05,
+      "loss": 0.2525,
+      "step": 5310
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 0.25,
+      "step": 5320
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 9.560912052117262e-05,
+      "loss": 0.2542,
+      "step": 5330
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 9.521824104234527e-05,
+      "loss": 0.2575,
+      "step": 5340
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 9.482736156351791e-05,
+      "loss": 0.2526,
+      "step": 5350
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.443648208469054e-05,
+      "loss": 0.2505,
+      "step": 5360
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.404560260586318e-05,
+      "loss": 0.2537,
+      "step": 5370
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.365472312703583e-05,
+      "loss": 0.2563,
+      "step": 5380
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.326384364820845e-05,
+      "loss": 0.2506,
+      "step": 5390
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.287296416938111e-05,
+      "loss": 0.2635,
+      "step": 5400
+    },
+    {
+      "epoch": 3.47,
+      "eval_loss": 0.27963170409202576,
+      "eval_runtime": 442.3292,
+      "eval_samples_per_second": 4.522,
+      "eval_steps_per_second": 0.565,
+      "step": 5400
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.248208469055374e-05,
+      "loss": 0.243,
+      "step": 5410
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.209120521172638e-05,
+      "loss": 0.2507,
+      "step": 5420
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.170032573289902e-05,
+      "loss": 0.2567,
+      "step": 5430
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.130944625407165e-05,
+      "loss": 0.2509,
+      "step": 5440
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.091856677524429e-05,
+      "loss": 0.2593,
+      "step": 5450
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 9.052768729641692e-05,
+      "loss": 0.2485,
+      "step": 5460
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 9.013680781758957e-05,
+      "loss": 0.2534,
+      "step": 5470
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.974592833876222e-05,
+      "loss": 0.2489,
+      "step": 5480
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.935504885993484e-05,
+      "loss": 0.2488,
+      "step": 5490
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.896416938110749e-05,
+      "loss": 0.2578,
+      "step": 5500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 8.857328990228011e-05,
+      "loss": 0.2408,
+      "step": 5510
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.818241042345275e-05,
+      "loss": 0.2478,
+      "step": 5520
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.779153094462541e-05,
+      "loss": 0.2476,
+      "step": 5530
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.740065146579804e-05,
+      "loss": 0.2476,
+      "step": 5540
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 8.700977198697068e-05,
+      "loss": 0.2516,
+      "step": 5550
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.661889250814332e-05,
+      "loss": 0.2595,
+      "step": 5560
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.622801302931595e-05,
+      "loss": 0.2523,
+      "step": 5570
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.583713355048859e-05,
+      "loss": 0.2594,
+      "step": 5580
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.544625407166122e-05,
+      "loss": 0.2558,
+      "step": 5590
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 8.505537459283387e-05,
+      "loss": 0.2497,
+      "step": 5600
+    },
+    {
+      "epoch": 3.6,
+      "eval_loss": 0.27932220697402954,
+      "eval_runtime": 442.4202,
+      "eval_samples_per_second": 4.521,
+      "eval_steps_per_second": 0.565,
+      "step": 5600
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.466449511400652e-05,
+      "loss": 0.2692,
+      "step": 5610
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.427361563517914e-05,
+      "loss": 0.2538,
+      "step": 5620
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.388273615635179e-05,
+      "loss": 0.2529,
+      "step": 5630
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.349185667752441e-05,
+      "loss": 0.244,
+      "step": 5640
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 8.310097719869706e-05,
+      "loss": 0.2619,
+      "step": 5650
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.271009771986971e-05,
+      "loss": 0.2532,
+      "step": 5660
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.231921824104234e-05,
+      "loss": 0.2441,
+      "step": 5670
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.192833876221498e-05,
+      "loss": 0.2607,
+      "step": 5680
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.153745928338761e-05,
+      "loss": 0.2571,
+      "step": 5690
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.114657980456025e-05,
+      "loss": 0.2599,
+      "step": 5700
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.075570032573289e-05,
+      "loss": 0.2415,
+      "step": 5710
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.036482084690552e-05,
+      "loss": 0.247,
+      "step": 5720
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 7.997394136807818e-05,
+      "loss": 0.2544,
+      "step": 5730
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 7.958306188925082e-05,
+      "loss": 0.2498,
+      "step": 5740
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 7.919218241042345e-05,
+      "loss": 0.2616,
+      "step": 5750
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 7.880130293159609e-05,
+      "loss": 0.2575,
+      "step": 5760
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 7.841042345276872e-05,
+      "loss": 0.2509,
+      "step": 5770
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 7.801954397394136e-05,
+      "loss": 0.2573,
+      "step": 5780
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 7.762866449511401e-05,
+      "loss": 0.2506,
+      "step": 5790
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.723778501628664e-05,
+      "loss": 0.2638,
+      "step": 5800
+    },
+    {
+      "epoch": 3.73,
+      "eval_loss": 0.27924951910972595,
+      "eval_runtime": 442.3315,
+      "eval_samples_per_second": 4.521,
+      "eval_steps_per_second": 0.565,
+      "step": 5800
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.684690553745928e-05,
+      "loss": 0.2502,
+      "step": 5810
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.645602605863191e-05,
+      "loss": 0.2421,
+      "step": 5820
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.606514657980455e-05,
+      "loss": 0.257,
+      "step": 5830
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.56742671009772e-05,
+      "loss": 0.2429,
+      "step": 5840
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.528338762214982e-05,
+      "loss": 0.2514,
+      "step": 5850
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.489250814332248e-05,
+      "loss": 0.2513,
+      "step": 5860
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.45016286644951e-05,
+      "loss": 0.2445,
+      "step": 5870
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.411074918566775e-05,
+      "loss": 0.2577,
+      "step": 5880
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.371986970684038e-05,
+      "loss": 0.2543,
+      "step": 5890
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.332899022801303e-05,
+      "loss": 0.2548,
+      "step": 5900
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.293811074918566e-05,
+      "loss": 0.2631,
+      "step": 5910
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.25472312703583e-05,
+      "loss": 0.2521,
+      "step": 5920
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.215635179153094e-05,
+      "loss": 0.255,
+      "step": 5930
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.176547231270357e-05,
+      "loss": 0.2528,
+      "step": 5940
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 7.137459283387621e-05,
+      "loss": 0.2469,
+      "step": 5950
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 7.098371335504885e-05,
+      "loss": 0.2568,
+      "step": 5960
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 7.05928338762215e-05,
+      "loss": 0.2428,
+      "step": 5970
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.020195439739412e-05,
+      "loss": 0.2415,
+      "step": 5980
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.981107491856678e-05,
+      "loss": 0.2537,
+      "step": 5990
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.942019543973941e-05,
+      "loss": 0.255,
+      "step": 6000
+    },
+    {
+      "epoch": 3.86,
+      "eval_loss": 0.2787570357322693,
+      "eval_runtime": 442.4383,
+      "eval_samples_per_second": 4.52,
+      "eval_steps_per_second": 0.565,
+      "step": 6000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.902931596091205e-05,
+      "loss": 0.2502,
+      "step": 6010
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.863843648208468e-05,
+      "loss": 0.2528,
+      "step": 6020
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.824755700325732e-05,
+      "loss": 0.2561,
+      "step": 6030
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.785667752442996e-05,
+      "loss": 0.2487,
+      "step": 6040
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.74657980456026e-05,
+      "loss": 0.2548,
+      "step": 6050
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.707491856677524e-05,
+      "loss": 0.2496,
+      "step": 6060
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.668403908794787e-05,
+      "loss": 0.2564,
+      "step": 6070
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.629315960912051e-05,
+      "loss": 0.2467,
+      "step": 6080
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.590228013029316e-05,
+      "loss": 0.2614,
+      "step": 6090
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.55114006514658e-05,
+      "loss": 0.2529,
+      "step": 6100
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.512052117263843e-05,
+      "loss": 0.2508,
+      "step": 6110
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.472964169381107e-05,
+      "loss": 0.2495,
+      "step": 6120
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.433876221498371e-05,
+      "loss": 0.2556,
+      "step": 6130
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.394788273615635e-05,
+      "loss": 0.2543,
+      "step": 6140
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.355700325732898e-05,
+      "loss": 0.2528,
+      "step": 6150
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.316612377850162e-05,
+      "loss": 0.2512,
+      "step": 6160
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.277524429967426e-05,
+      "loss": 0.249,
+      "step": 6170
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.23843648208469e-05,
+      "loss": 0.2591,
+      "step": 6180
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.199348534201955e-05,
+      "loss": 0.2459,
+      "step": 6190
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.160260586319217e-05,
+      "loss": 0.2633,
+      "step": 6200
+    },
+    {
+      "epoch": 3.99,
+      "eval_loss": 0.27837860584259033,
+      "eval_runtime": 442.2743,
+      "eval_samples_per_second": 4.522,
+      "eval_steps_per_second": 0.565,
+      "step": 6200
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7775,
+  "num_train_epochs": 5,
+  "save_steps": 200,
+  "total_flos": 6.311538956559385e+18,
+  "trial_name": null,
+  "trial_params": null
+}

with_input/decomp_code_with_intermediates/best_model/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:811709a41c93ebb0a7d5b7817a4a00f1ea401d394af55af88e1e3c43b8ada772
+size 4091