Upload folder using huggingface_hub

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:39118f7368d7b390b9e48c4e9a09b46e6fcdee5e50be90f61bfd90db2c637fa8
+oid sha256:793c2c9d0135d56ef7de4863551a40cc9d5c7b029605658aa0537f5f355f5cc2
 size 25172088

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6b60f982069b7054a7c3e41818e87b3ee2a8d276881f8d520b1f0c0d70d36309
+oid sha256:e2802faa1f7ee79dc302ab9df00e418e656ecd332ea2c97989901116bd2f2085
 size 50372538

rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e5994634e90ce1b45d012a45568063be05fea876e791cd66b48a4efc924164b2
+oid sha256:feded292b067a53a2aeb0e2a23dd8fd5fd080c27efc2767fe4b430da7e5f7d6f
 size 14244

trainer_state.json

@@ -12,2184 +12,2184 @@
       "epoch": 0.0032,
       "grad_norm": 10.0,
       "learning_rate": 2.132196162046908e-06,
-      "loss": 22.4733,
+      "loss": 25.0099,
       "step": 10
     },
     {
       "epoch": 0.0064,
-      "grad_norm": 9.999999046325684,
+      "grad_norm": 10.0,
       "learning_rate": 4.264392324093816e-06,
-      "loss": 18.8042,
+      "loss": 21.14,
       "step": 20
     },
     {
       "epoch": 0.0096,
-      "grad_norm": 10.000001907348633,
+      "grad_norm": 10.0,
       "learning_rate": 6.396588486140726e-06,
-      "loss": 19.6935,
+      "loss": 19.5656,
       "step": 30
     },
     {
       "epoch": 0.0128,
-      "grad_norm": 9.999999046325684,
+      "grad_norm": 10.0,
       "learning_rate": 8.528784648187633e-06,
-      "loss": 19.4804,
+      "loss": 22.2556,
       "step": 40
     },
     {
       "epoch": 0.016,
-      "grad_norm": 10.000000953674316,
+      "grad_norm": 9.999999046325684,
       "learning_rate": 1.0660980810234541e-05,
-      "loss": 17.4569,
+      "loss": 19.8975,
       "step": 50
     },
     {
       "epoch": 0.0192,
       "grad_norm": 10.0,
       "learning_rate": 1.2793176972281452e-05,
-      "loss": 18.3143,
+      "loss": 19.9186,
       "step": 60
     },
     {
       "epoch": 0.0224,
       "grad_norm": 10.0,
       "learning_rate": 1.4925373134328357e-05,
-      "loss": 16.6743,
+      "loss": 19.9392,
       "step": 70
     },
     {
       "epoch": 0.0256,
-      "grad_norm": 9.999999046325684,
+      "grad_norm": 10.0,
       "learning_rate": 1.7057569296375266e-05,
-      "loss": 14.7713,
+      "loss": 18.3684,
       "step": 80
     },
     {
       "epoch": 0.0288,
-      "grad_norm": 9.999999046325684,
+      "grad_norm": 9.999998092651367,
       "learning_rate": 1.9189765458422178e-05,
-      "loss": 14.883,
+      "loss": 16.6577,
       "step": 90
     },
     {
       "epoch": 0.032,
       "grad_norm": 9.999999046325684,
       "learning_rate": 2.1321961620469083e-05,
-      "loss": 14.1393,
+      "loss": 13.9967,
       "step": 100
     },
     {
       "epoch": 0.0352,
-      "grad_norm": 9.999999046325684,
+      "grad_norm": 10.0,
       "learning_rate": 2.345415778251599e-05,
-      "loss": 11.5208,
+      "loss": 15.5673,
       "step": 110
     },
     {
       "epoch": 0.0384,
-      "grad_norm": 10.0,
+      "grad_norm": 9.999999046325684,
       "learning_rate": 2.5586353944562904e-05,
-      "loss": 12.865,
+      "loss": 12.1718,
       "step": 120
     },
     {
       "epoch": 0.0416,
       "grad_norm": 10.0,
       "learning_rate": 2.771855010660981e-05,
-      "loss": 10.3148,
+      "loss": 11.1047,
       "step": 130
     },
     {
       "epoch": 0.0448,
-      "grad_norm": 9.999998092651367,
+      "grad_norm": 10.0,
       "learning_rate": 2.9850746268656714e-05,
-      "loss": 9.7127,
+      "loss": 10.4134,
       "step": 140
     },
     {
       "epoch": 0.048,
-      "grad_norm": 9.999998092651367,
+      "grad_norm": 10.0,
       "learning_rate": 3.1982942430703626e-05,
-      "loss": 8.3006,
+      "loss": 10.6204,
       "step": 150
     },
     {
       "epoch": 0.0512,
       "grad_norm": 9.999999046325684,
       "learning_rate": 3.411513859275053e-05,
-      "loss": 7.292,
+      "loss": 7.4005,
       "step": 160
     },
     {
       "epoch": 0.0544,
-      "grad_norm": 9.999999046325684,
+      "grad_norm": 9.999998092651367,
       "learning_rate": 3.624733475479744e-05,
-      "loss": 6.0268,
+      "loss": 6.7938,
       "step": 170
     },
     {
       "epoch": 0.0576,
-      "grad_norm": 9.999998092651367,
+      "grad_norm": 9.999999046325684,
       "learning_rate": 3.8379530916844355e-05,
-      "loss": 5.7297,
+      "loss": 5.8487,
       "step": 180
     },
     {
       "epoch": 0.0608,
-      "grad_norm": 10.0,
+      "grad_norm": 9.999999046325684,
       "learning_rate": 4.051172707889126e-05,
-      "loss": 4.6862,
+      "loss": 5.6155,
       "step": 190
     },
     {
       "epoch": 0.064,
-      "grad_norm": 9.999999046325684,
+      "grad_norm": 9.999998092651367,
       "learning_rate": 4.2643923240938166e-05,
-      "loss": 4.0643,
+      "loss": 4.978,
       "step": 200
     },
     {
       "epoch": 0.0672,
-      "grad_norm": 9.999998092651367,
+      "grad_norm": 9.999999046325684,
       "learning_rate": 4.477611940298508e-05,
-      "loss": 3.3502,
+      "loss": 3.7961,
       "step": 210
     },
     {
       "epoch": 0.0704,
-      "grad_norm": 9.999998092651367,
+      "grad_norm": 9.999999046325684,
       "learning_rate": 4.690831556503198e-05,
-      "loss": 2.6549,
+      "loss": 3.2914,
       "step": 220
     },
     {
       "epoch": 0.0736,
-      "grad_norm": 8.796483039855957,
+      "grad_norm": 8.79616641998291,
       "learning_rate": 4.904051172707889e-05,
-      "loss": 2.883,
+      "loss": 3.3423,
       "step": 230
     },
     {
       "epoch": 0.0768,
-      "grad_norm": 8.624096870422363,
+      "grad_norm": 8.134244918823242,
       "learning_rate": 5.117270788912581e-05,
-      "loss": 2.3625,
+      "loss": 2.5389,
       "step": 240
     },
     {
       "epoch": 0.08,
-      "grad_norm": 4.12373161315918,
+      "grad_norm": 8.448352813720703,
       "learning_rate": 5.330490405117271e-05,
-      "loss": 2.0516,
+      "loss": 2.2754,
       "step": 250
     },
     {
       "epoch": 0.0832,
-      "grad_norm": 6.344674110412598,
+      "grad_norm": 5.989101409912109,
       "learning_rate": 5.543710021321962e-05,
-      "loss": 1.8814,
+      "loss": 2.0337,
       "step": 260
     },
     {
       "epoch": 0.0864,
-      "grad_norm": 5.940646171569824,
+      "grad_norm": 5.194303512573242,
       "learning_rate": 5.756929637526652e-05,
-      "loss": 1.8997,
+      "loss": 1.8128,
       "step": 270
     },
     {
       "epoch": 0.0896,
-      "grad_norm": 4.136787414550781,
+      "grad_norm": 4.9008893966674805,
       "learning_rate": 5.970149253731343e-05,
-      "loss": 1.6641,
+      "loss": 1.8,
       "step": 280
     },
     {
       "epoch": 0.0928,
-      "grad_norm": 3.336697578430176,
+      "grad_norm": 5.970029354095459,
       "learning_rate": 6.183368869936035e-05,
-      "loss": 1.5741,
+      "loss": 1.6032,
       "step": 290
     },
     {
       "epoch": 0.096,
-      "grad_norm": 4.003772735595703,
+      "grad_norm": 5.243438720703125,
       "learning_rate": 6.396588486140725e-05,
-      "loss": 1.4759,
+      "loss": 1.7633,
       "step": 300
     },
     {
       "epoch": 0.0992,
-      "grad_norm": 5.183838367462158,
+      "grad_norm": 4.298349857330322,
       "learning_rate": 6.609808102345416e-05,
-      "loss": 1.5434,
+      "loss": 1.5381,
       "step": 310
     },
     {
       "epoch": 0.1024,
-      "grad_norm": 4.6075849533081055,
+      "grad_norm": 4.325716495513916,
       "learning_rate": 6.823027718550106e-05,
-      "loss": 1.5185,
+      "loss": 1.4,
       "step": 320
     },
     {
       "epoch": 0.1056,
-      "grad_norm": 4.707767486572266,
+      "grad_norm": 4.040591716766357,
       "learning_rate": 7.036247334754798e-05,
-      "loss": 1.3706,
+      "loss": 1.485,
       "step": 330
     },
     {
       "epoch": 0.1088,
-      "grad_norm": 3.7312469482421875,
+      "grad_norm": 4.4948015213012695,
       "learning_rate": 7.249466950959489e-05,
-      "loss": 1.3455,
+      "loss": 1.3199,
       "step": 340
     },
     {
       "epoch": 0.112,
-      "grad_norm": 3.605818033218384,
+      "grad_norm": 4.134742736816406,
       "learning_rate": 7.46268656716418e-05,
-      "loss": 1.2145,
+      "loss": 1.3535,
       "step": 350
     },
     {
       "epoch": 0.1152,
-      "grad_norm": 4.285920143127441,
+      "grad_norm": 3.762752056121826,
       "learning_rate": 7.675906183368871e-05,
-      "loss": 1.303,
+      "loss": 1.2331,
       "step": 360
     },
     {
       "epoch": 0.1184,
-      "grad_norm": 3.215698480606079,
+      "grad_norm": 4.193824291229248,
       "learning_rate": 7.889125799573562e-05,
-      "loss": 1.168,
+      "loss": 1.2674,
       "step": 370
     },
     {
       "epoch": 0.1216,
-      "grad_norm": 4.043213844299316,
+      "grad_norm": 3.6042652130126953,
       "learning_rate": 8.102345415778252e-05,
-      "loss": 1.0962,
+      "loss": 1.2238,
       "step": 380
     },
     {
       "epoch": 0.1248,
-      "grad_norm": 4.1487555503845215,
+      "grad_norm": 3.8218977451324463,
       "learning_rate": 8.315565031982943e-05,
-      "loss": 1.1853,
+      "loss": 1.18,
       "step": 390
     },
     {
       "epoch": 0.128,
-      "grad_norm": 3.942502498626709,
+      "grad_norm": 4.003774166107178,
       "learning_rate": 8.528784648187633e-05,
-      "loss": 1.0418,
+      "loss": 1.2161,
       "step": 400
     },
     {
       "epoch": 0.1312,
-      "grad_norm": 4.188830852508545,
+      "grad_norm": 3.954782247543335,
       "learning_rate": 8.742004264392325e-05,
-      "loss": 1.0992,
+      "loss": 1.1235,
       "step": 410
     },
     {
       "epoch": 0.1344,
-      "grad_norm": 4.130038738250732,
+      "grad_norm": 4.31278657913208,
       "learning_rate": 8.955223880597016e-05,
-      "loss": 0.976,
+      "loss": 1.0786,
       "step": 420
     },
     {
       "epoch": 0.1376,
-      "grad_norm": 3.643944501876831,
+      "grad_norm": 3.7707276344299316,
       "learning_rate": 9.168443496801706e-05,
-      "loss": 1.0592,
+      "loss": 1.0585,
       "step": 430
     },
     {
       "epoch": 0.1408,
-      "grad_norm": 3.760075092315674,
+      "grad_norm": 5.498402118682861,
       "learning_rate": 9.381663113006397e-05,
-      "loss": 1.0115,
+      "loss": 1.0594,
       "step": 440
     },
     {
       "epoch": 0.144,
-      "grad_norm": 2.4692914485931396,
+      "grad_norm": 4.374643802642822,
       "learning_rate": 9.594882729211087e-05,
-      "loss": 1.0096,
+      "loss": 1.0838,
       "step": 450
     },
     {
       "epoch": 0.1472,
-      "grad_norm": 3.1593716144561768,
+      "grad_norm": 4.22064733505249,
       "learning_rate": 9.808102345415778e-05,
-      "loss": 0.9181,
+      "loss": 1.0391,
       "step": 460
     },
     {
       "epoch": 0.1504,
-      "grad_norm": 3.6118807792663574,
+      "grad_norm": 3.9745569229125977,
       "learning_rate": 9.998877161464182e-05,
-      "loss": 0.921,
+      "loss": 0.8823,
       "step": 470
     },
     {
       "epoch": 0.1536,
-      "grad_norm": 3.3329732418060303,
+      "grad_norm": 3.797243595123291,
       "learning_rate": 9.987648776105997e-05,
-      "loss": 0.8714,
+      "loss": 0.889,
       "step": 480
     },
     {
       "epoch": 0.1568,
-      "grad_norm": 2.9491333961486816,
+      "grad_norm": 3.7148423194885254,
       "learning_rate": 9.97642039074781e-05,
-      "loss": 0.8553,
+      "loss": 0.8886,
       "step": 490
     },
     {
       "epoch": 0.16,
-      "grad_norm": 3.0345425605773926,
+      "grad_norm": 3.3785812854766846,
       "learning_rate": 9.965192005389625e-05,
-      "loss": 0.9232,
+      "loss": 0.9073,
       "step": 500
     },
     {
       "epoch": 0.1632,
-      "grad_norm": 3.3396542072296143,
+      "grad_norm": 3.5189311504364014,
       "learning_rate": 9.95396362003144e-05,
-      "loss": 0.8112,
+      "loss": 0.8472,
       "step": 510
     },
     {
       "epoch": 0.1664,
-      "grad_norm": 3.1262471675872803,
+      "grad_norm": 4.278319835662842,
       "learning_rate": 9.942735234673256e-05,
-      "loss": 0.8252,
+      "loss": 0.8266,
       "step": 520
     },
     {
       "epoch": 0.1696,
-      "grad_norm": 2.431586742401123,
+      "grad_norm": 2.666747570037842,
       "learning_rate": 9.931506849315069e-05,
-      "loss": 0.7863,
+      "loss": 0.7793,
       "step": 530
     },
     {
       "epoch": 0.1728,
-      "grad_norm": 3.4569954872131348,
+      "grad_norm": 6.504436016082764,
       "learning_rate": 9.920278463956883e-05,
-      "loss": 0.7916,
+      "loss": 0.8237,
       "step": 540
     },
     {
       "epoch": 0.176,
-      "grad_norm": 3.2024927139282227,
+      "grad_norm": 3.135028839111328,
       "learning_rate": 9.909050078598698e-05,
-      "loss": 0.8355,
+      "loss": 0.771,
       "step": 550
     },
     {
       "epoch": 0.1792,
-      "grad_norm": 3.4040327072143555,
+      "grad_norm": 3.2984988689422607,
       "learning_rate": 9.897821693240512e-05,
-      "loss": 0.8548,
+      "loss": 0.7429,
       "step": 560
     },
     {
       "epoch": 0.1824,
-      "grad_norm": 2.5302040576934814,
+      "grad_norm": 4.183766841888428,
       "learning_rate": 9.886593307882327e-05,
-      "loss": 0.7237,
+      "loss": 0.788,
       "step": 570
     },
     {
       "epoch": 0.1856,
-      "grad_norm": 3.6245014667510986,
+      "grad_norm": 3.8883821964263916,
       "learning_rate": 9.875364922524142e-05,
-      "loss": 0.7395,
+      "loss": 0.8378,
       "step": 580
     },
     {
       "epoch": 0.1888,
-      "grad_norm": 2.6966214179992676,
+      "grad_norm": 3.9806365966796875,
       "learning_rate": 9.864136537165956e-05,
-      "loss": 0.7004,
+      "loss": 0.7939,
       "step": 590
     },
     {
       "epoch": 0.192,
-      "grad_norm": 3.207789421081543,
+      "grad_norm": 3.4956815242767334,
       "learning_rate": 9.852908151807771e-05,
-      "loss": 0.7419,
+      "loss": 0.7603,
       "step": 600
     },
     {
       "epoch": 0.1952,
-      "grad_norm": 3.4613256454467773,
+      "grad_norm": 5.5648627281188965,
       "learning_rate": 9.841679766449586e-05,
-      "loss": 0.7264,
+      "loss": 0.7235,
       "step": 610
     },
     {
       "epoch": 0.1984,
-      "grad_norm": 3.311279058456421,
+      "grad_norm": 3.4344027042388916,
       "learning_rate": 9.8304513810914e-05,
-      "loss": 0.6852,
+      "loss": 0.6992,
       "step": 620
     },
     {
       "epoch": 0.2016,
-      "grad_norm": 2.7546231746673584,
+      "grad_norm": 3.3964436054229736,
       "learning_rate": 9.819222995733213e-05,
-      "loss": 0.6922,
+      "loss": 0.6704,
       "step": 630
     },
     {
       "epoch": 0.2048,
-      "grad_norm": 3.5927953720092773,
+      "grad_norm": 4.435855388641357,
       "learning_rate": 9.807994610375028e-05,
-      "loss": 0.7391,
+      "loss": 0.6606,
       "step": 640
     },
     {
       "epoch": 0.208,
-      "grad_norm": 2.975539207458496,
+      "grad_norm": 3.452620029449463,
       "learning_rate": 9.796766225016843e-05,
-      "loss": 0.699,
+      "loss": 0.6262,
       "step": 650
     },
     {
       "epoch": 0.2112,
-      "grad_norm": 3.653235673904419,
+      "grad_norm": 3.708383798599243,
       "learning_rate": 9.785537839658657e-05,
-      "loss": 0.6669,
+      "loss": 0.6535,
       "step": 660
     },
     {
       "epoch": 0.2144,
-      "grad_norm": 3.6133604049682617,
+      "grad_norm": 4.044270038604736,
       "learning_rate": 9.774309454300472e-05,
-      "loss": 0.7107,
+      "loss": 0.5877,
       "step": 670
     },
     {
       "epoch": 0.2176,
-      "grad_norm": 2.855388641357422,
+      "grad_norm": 3.3694796562194824,
       "learning_rate": 9.763081068942287e-05,
-      "loss": 0.6938,
+      "loss": 0.5658,
       "step": 680
     },
     {
       "epoch": 0.2208,
-      "grad_norm": 2.790356159210205,
+      "grad_norm": 3.8326478004455566,
       "learning_rate": 9.751852683584101e-05,
-      "loss": 0.6208,
+      "loss": 0.5767,
       "step": 690
     },
     {
       "epoch": 0.224,
-      "grad_norm": 3.2559432983398438,
+      "grad_norm": 3.358949899673462,
       "learning_rate": 9.740624298225916e-05,
-      "loss": 0.6327,
+      "loss": 0.5449,
       "step": 700
     },
     {
       "epoch": 0.2272,
-      "grad_norm": 2.8997764587402344,
+      "grad_norm": 4.013234615325928,
       "learning_rate": 9.729395912867731e-05,
-      "loss": 0.6176,
+      "loss": 0.5099,
       "step": 710
     },
     {
       "epoch": 0.2304,
-      "grad_norm": 3.3574609756469727,
+      "grad_norm": 3.565805435180664,
       "learning_rate": 9.718167527509545e-05,
-      "loss": 0.601,
+      "loss": 0.5736,
       "step": 720
     },
     {
       "epoch": 0.2336,
-      "grad_norm": 2.94797945022583,
+      "grad_norm": 3.975719690322876,
       "learning_rate": 9.706939142151358e-05,
-      "loss": 0.5904,
+      "loss": 0.5294,
       "step": 730
     },
     {
       "epoch": 0.2368,
-      "grad_norm": 2.8964426517486572,
+      "grad_norm": 6.230079650878906,
       "learning_rate": 9.695710756793174e-05,
-      "loss": 0.5889,
+      "loss": 0.5179,
       "step": 740
     },
     {
       "epoch": 0.24,
-      "grad_norm": 2.505166530609131,
+      "grad_norm": 3.283184051513672,
       "learning_rate": 9.684482371434989e-05,
-      "loss": 0.5491,
+      "loss": 0.496,
       "step": 750
     },
     {
       "epoch": 0.2432,
-      "grad_norm": 2.8895745277404785,
+      "grad_norm": 3.611090660095215,
       "learning_rate": 9.673253986076802e-05,
-      "loss": 0.5751,
+      "loss": 0.5017,
       "step": 760
     },
     {
       "epoch": 0.2464,
-      "grad_norm": 2.348055601119995,
+      "grad_norm": 6.290208339691162,
       "learning_rate": 9.662025600718617e-05,
-      "loss": 0.5702,
+      "loss": 0.4952,
       "step": 770
     },
     {
       "epoch": 0.2496,
-      "grad_norm": 3.0331923961639404,
+      "grad_norm": 4.165639400482178,
       "learning_rate": 9.650797215360432e-05,
-      "loss": 0.5299,
+      "loss": 0.4312,
       "step": 780
     },
     {
       "epoch": 0.2528,
-      "grad_norm": 2.881728172302246,
+      "grad_norm": 3.7825958728790283,
       "learning_rate": 9.639568830002246e-05,
-      "loss": 0.4772,
+      "loss": 0.412,
       "step": 790
     },
     {
       "epoch": 0.256,
-      "grad_norm": 2.9465715885162354,
+      "grad_norm": 6.383756637573242,
       "learning_rate": 9.628340444644061e-05,
-      "loss": 0.5137,
+      "loss": 0.4097,
       "step": 800
     },
     {
       "epoch": 0.2592,
-      "grad_norm": 2.664348602294922,
+      "grad_norm": 2.7888543605804443,
       "learning_rate": 9.617112059285875e-05,
-      "loss": 0.4398,
+      "loss": 0.3707,
       "step": 810
     },
     {
       "epoch": 0.2624,
-      "grad_norm": 2.931985378265381,
+      "grad_norm": 1.9508731365203857,
       "learning_rate": 9.605883673927689e-05,
-      "loss": 0.4969,
+      "loss": 0.3584,
       "step": 820
     },
     {
       "epoch": 0.2656,
-      "grad_norm": 2.953338861465454,
+      "grad_norm": 3.938863515853882,
       "learning_rate": 9.594655288569504e-05,
-      "loss": 0.4829,
+      "loss": 0.3721,
       "step": 830
     },
     {
       "epoch": 0.2688,
-      "grad_norm": 3.0215139389038086,
+      "grad_norm": 2.89081072807312,
       "learning_rate": 9.583426903211319e-05,
-      "loss": 0.463,
+      "loss": 0.3412,
       "step": 840
     },
     {
       "epoch": 0.272,
-      "grad_norm": 3.6885106563568115,
+      "grad_norm": 2.9330146312713623,
       "learning_rate": 9.572198517853134e-05,
-      "loss": 0.4601,
+      "loss": 0.2764,
       "step": 850
     },
     {
       "epoch": 0.2752,
-      "grad_norm": 3.8273849487304688,
+      "grad_norm": 2.5014054775238037,
       "learning_rate": 9.560970132494948e-05,
-      "loss": 0.5616,
+      "loss": 0.2725,
       "step": 860
     },
     {
       "epoch": 0.2784,
-      "grad_norm": 3.266014337539673,
+      "grad_norm": 2.38750958442688,
       "learning_rate": 9.549741747136763e-05,
-      "loss": 0.4581,
+      "loss": 0.2933,
       "step": 870
     },
     {
       "epoch": 0.2816,
-      "grad_norm": 2.304288387298584,
+      "grad_norm": 3.248847246170044,
       "learning_rate": 9.538513361778578e-05,
-      "loss": 0.4192,
+      "loss": 0.309,
       "step": 880
     },
     {
       "epoch": 0.2848,
-      "grad_norm": 2.7526509761810303,
+      "grad_norm": 4.140264987945557,
       "learning_rate": 9.527284976420391e-05,
-      "loss": 0.3809,
+      "loss": 0.2656,
       "step": 890
     },
     {
       "epoch": 0.288,
-      "grad_norm": 3.033849000930786,
+      "grad_norm": 2.2702248096466064,
       "learning_rate": 9.516056591062205e-05,
-      "loss": 0.3865,
+      "loss": 0.2928,
       "step": 900
     },
     {
       "epoch": 0.2912,
-      "grad_norm": 2.9246816635131836,
+      "grad_norm": 5.701258659362793,
       "learning_rate": 9.50482820570402e-05,
-      "loss": 0.3565,
+      "loss": 0.267,
       "step": 910
     },
     {
       "epoch": 0.2944,
-      "grad_norm": 2.111208438873291,
+      "grad_norm": 2.0174479484558105,
       "learning_rate": 9.493599820345834e-05,
-      "loss": 0.3507,
+      "loss": 0.2388,
       "step": 920
     },
     {
       "epoch": 0.2976,
-      "grad_norm": 3.193631172180176,
+      "grad_norm": 2.4099040031433105,
       "learning_rate": 9.482371434987649e-05,
-      "loss": 0.3136,
+      "loss": 0.2303,
       "step": 930
     },
     {
       "epoch": 0.3008,
-      "grad_norm": 2.647897481918335,
+      "grad_norm": 2.7431418895721436,
       "learning_rate": 9.471143049629464e-05,
-      "loss": 0.3034,
+      "loss": 0.2389,
       "step": 940
     },
     {
       "epoch": 0.304,
-      "grad_norm": 4.482762813568115,
+      "grad_norm": 1.7483993768692017,
       "learning_rate": 9.459914664271278e-05,
-      "loss": 0.3168,
+      "loss": 0.2555,
       "step": 950
     },
     {
       "epoch": 0.3072,
-      "grad_norm": 2.408997058868408,
+      "grad_norm": 2.9412496089935303,
       "learning_rate": 9.448686278913093e-05,
-      "loss": 0.2838,
+      "loss": 0.2189,
       "step": 960
     },
     {
       "epoch": 0.3104,
-      "grad_norm": 2.701946496963501,
+      "grad_norm": 2.591343402862549,
       "learning_rate": 9.437457893554908e-05,
-      "loss": 0.2945,
+      "loss": 0.1972,
       "step": 970
     },
     {
       "epoch": 0.3136,
-      "grad_norm": 2.5696310997009277,
+      "grad_norm": 2.46079421043396,
       "learning_rate": 9.426229508196722e-05,
-      "loss": 0.2483,
+      "loss": 0.1959,
       "step": 980
     },
     {
       "epoch": 0.3168,
-      "grad_norm": 1.9443378448486328,
+      "grad_norm": 3.11035418510437,
       "learning_rate": 9.415001122838537e-05,
-      "loss": 0.2379,
+      "loss": 0.1952,
       "step": 990
     },
     {
       "epoch": 0.32,
-      "grad_norm": 2.137860059738159,
+      "grad_norm": 1.9760693311691284,
       "learning_rate": 9.40377273748035e-05,
-      "loss": 0.2396,
+      "loss": 0.1732,
       "step": 1000
     },
     {
       "epoch": 0.3232,
-      "grad_norm": 2.6493215560913086,
+      "grad_norm": 1.6762940883636475,
       "learning_rate": 9.392544352122165e-05,
-      "loss": 0.2294,
+      "loss": 0.1591,
       "step": 1010
     },
     {
       "epoch": 0.3264,
-      "grad_norm": 3.381498336791992,
+      "grad_norm": 2.380981922149658,
       "learning_rate": 9.381315966763979e-05,
-      "loss": 0.2162,
+      "loss": 0.1799,
       "step": 1020
     },
     {
       "epoch": 0.3296,
-      "grad_norm": 2.907684564590454,
+      "grad_norm": 3.2577524185180664,
       "learning_rate": 9.370087581405794e-05,
-      "loss": 0.194,
+      "loss": 0.1994,
       "step": 1030
     },
     {
       "epoch": 0.3328,
-      "grad_norm": 2.3227946758270264,
+      "grad_norm": 1.5915032625198364,
       "learning_rate": 9.358859196047609e-05,
-      "loss": 0.2071,
+      "loss": 0.1466,
       "step": 1040
     },
     {
       "epoch": 0.336,
-      "grad_norm": 2.3173024654388428,
+      "grad_norm": 1.7364068031311035,
       "learning_rate": 9.347630810689423e-05,
-      "loss": 0.2018,
+      "loss": 0.164,
       "step": 1050
     },
     {
       "epoch": 0.3392,
-      "grad_norm": 2.9500386714935303,
+      "grad_norm": 2.3017282485961914,
       "learning_rate": 9.336402425331238e-05,
-      "loss": 0.1987,
+      "loss": 0.1564,
       "step": 1060
     },
     {
       "epoch": 0.3424,
-      "grad_norm": 5.206624984741211,
+      "grad_norm": 1.5479563474655151,
       "learning_rate": 9.325174039973053e-05,
-      "loss": 0.2036,
+      "loss": 0.1413,
       "step": 1070
     },
     {
       "epoch": 0.3456,
-      "grad_norm": 1.9075260162353516,
+      "grad_norm": 2.007779836654663,
       "learning_rate": 9.313945654614867e-05,
-      "loss": 0.1785,
+      "loss": 0.143,
       "step": 1080
     },
     {
       "epoch": 0.3488,
-      "grad_norm": 1.7478675842285156,
+      "grad_norm": 1.2506276369094849,
       "learning_rate": 9.30271726925668e-05,
-      "loss": 0.1956,
+      "loss": 0.1297,
       "step": 1090
     },
     {
       "epoch": 0.352,
-      "grad_norm": 2.1234798431396484,
+      "grad_norm": 1.4856570959091187,
       "learning_rate": 9.291488883898496e-05,
-      "loss": 0.175,
+      "loss": 0.1355,
       "step": 1100
     },
     {
       "epoch": 0.3552,
-      "grad_norm": 1.828234076499939,
+      "grad_norm": 2.262418270111084,
       "learning_rate": 9.280260498540311e-05,
-      "loss": 0.1591,
+      "loss": 0.1374,
       "step": 1110
     },
     {
       "epoch": 0.3584,
-      "grad_norm": 1.8341890573501587,
+      "grad_norm": 1.535651683807373,
       "learning_rate": 9.269032113182124e-05,
-      "loss": 0.1649,
+      "loss": 0.1331,
       "step": 1120
     },
     {
       "epoch": 0.3616,
-      "grad_norm": 1.3575913906097412,
+      "grad_norm": 1.3784823417663574,
       "learning_rate": 9.25780372782394e-05,
-      "loss": 0.1694,
+      "loss": 0.1495,
       "step": 1130
     },
     {
       "epoch": 0.3648,
-      "grad_norm": 2.1410560607910156,
+      "grad_norm": 1.548283338546753,
       "learning_rate": 9.246575342465755e-05,
-      "loss": 0.1659,
+      "loss": 0.1194,
       "step": 1140
     },
     {
       "epoch": 0.368,
-      "grad_norm": 1.7794946432113647,
+      "grad_norm": 2.1643035411834717,
       "learning_rate": 9.235346957107568e-05,
-      "loss": 0.1479,
+      "loss": 0.1285,
       "step": 1150
     },
     {
       "epoch": 0.3712,
-      "grad_norm": 2.1806530952453613,
+      "grad_norm": 1.8928327560424805,
       "learning_rate": 9.224118571749383e-05,
-      "loss": 0.151,
+      "loss": 0.1189,
       "step": 1160
     },
     {
       "epoch": 0.3744,
-      "grad_norm": 1.5767009258270264,
+      "grad_norm": 1.3566118478775024,
       "learning_rate": 9.212890186391197e-05,
-      "loss": 0.1541,
+      "loss": 0.1165,
       "step": 1170
     },
     {
       "epoch": 0.3776,
-      "grad_norm": 1.453840970993042,
+      "grad_norm": 1.5916873216629028,
       "learning_rate": 9.201661801033011e-05,
-      "loss": 0.1368,
+      "loss": 0.1129,
       "step": 1180
     },
     {
       "epoch": 0.3808,
-      "grad_norm": 1.8884810209274292,
+      "grad_norm": 2.1428229808807373,
       "learning_rate": 9.190433415674826e-05,
-      "loss": 0.1394,
+      "loss": 0.1176,
       "step": 1190
     },
     {
       "epoch": 0.384,
-      "grad_norm": 1.6665995121002197,
+      "grad_norm": 1.2154109477996826,
       "learning_rate": 9.179205030316641e-05,
-      "loss": 0.1356,
+      "loss": 0.1045,
       "step": 1200
     },
     {
       "epoch": 0.3872,
-      "grad_norm": 1.6958074569702148,
+      "grad_norm": 1.4908652305603027,
       "learning_rate": 9.167976644958456e-05,
-      "loss": 0.1468,
+      "loss": 0.1061,
       "step": 1210
     },
     {
       "epoch": 0.3904,
-      "grad_norm": 1.609466552734375,
+      "grad_norm": 1.2576334476470947,
       "learning_rate": 9.15674825960027e-05,
-      "loss": 0.1396,
+      "loss": 0.0973,
       "step": 1220
     },
     {
       "epoch": 0.3936,
-      "grad_norm": 2.045232057571411,
+      "grad_norm": 2.442065954208374,
       "learning_rate": 9.145519874242085e-05,
-      "loss": 0.1234,
+      "loss": 0.1057,
       "step": 1230
     },
     {
       "epoch": 0.3968,
-      "grad_norm": 1.8025046586990356,
+      "grad_norm": 1.8123060464859009,
       "learning_rate": 9.1342914888839e-05,
-      "loss": 0.1335,
+      "loss": 0.1037,
       "step": 1240
     },
     {
       "epoch": 0.4,
-      "grad_norm": 1.4416459798812866,
+      "grad_norm": 3.8450021743774414,
       "learning_rate": 9.123063103525713e-05,
-      "loss": 0.1325,
+      "loss": 0.1042,
       "step": 1250
     },
     {
       "epoch": 0.4032,
-      "grad_norm": 2.590508460998535,
+      "grad_norm": 1.4633779525756836,
       "learning_rate": 9.111834718167527e-05,
-      "loss": 0.1258,
+      "loss": 0.1081,
       "step": 1260
     },
     {
       "epoch": 0.4064,
-      "grad_norm": 1.7362828254699707,
+      "grad_norm": 2.0343596935272217,
       "learning_rate": 9.100606332809342e-05,
-      "loss": 0.1337,
+      "loss": 0.0845,
       "step": 1270
     },
     {
       "epoch": 0.4096,
-      "grad_norm": 1.726511836051941,
+      "grad_norm": 2.1787729263305664,
       "learning_rate": 9.089377947451156e-05,
-      "loss": 0.1274,
+      "loss": 0.1067,
       "step": 1280
     },
     {
       "epoch": 0.4128,
-      "grad_norm": 1.3185865879058838,
+      "grad_norm": 1.5552949905395508,
       "learning_rate": 9.078149562092971e-05,
-      "loss": 0.1244,
+      "loss": 0.107,
       "step": 1290
     },
     {
       "epoch": 0.416,
-      "grad_norm": 1.6214089393615723,
+      "grad_norm": 2.656562566757202,
       "learning_rate": 9.066921176734786e-05,
-      "loss": 0.1002,
+      "loss": 0.0911,
       "step": 1300
     },
     {
       "epoch": 0.4192,
-      "grad_norm": 1.206648826599121,
+      "grad_norm": 1.4937834739685059,
       "learning_rate": 9.055692791376601e-05,
-      "loss": 0.1169,
+      "loss": 0.0992,
       "step": 1310
     },
     {
       "epoch": 0.4224,
-      "grad_norm": 1.020280122756958,
+      "grad_norm": 1.6240930557250977,
       "learning_rate": 9.044464406018415e-05,
-      "loss": 0.1036,
+      "loss": 0.0914,
       "step": 1320
     },
     {
       "epoch": 0.4256,
-      "grad_norm": 1.86745285987854,
+      "grad_norm": 1.9437700510025024,
       "learning_rate": 9.03323602066023e-05,
-      "loss": 0.1163,
+      "loss": 0.0985,
       "step": 1330
     },
     {
       "epoch": 0.4288,
-      "grad_norm": 1.2367403507232666,
+      "grad_norm": 1.0026746988296509,
       "learning_rate": 9.022007635302045e-05,
-      "loss": 0.1166,
+      "loss": 0.0917,
       "step": 1340
     },
     {
       "epoch": 0.432,
-      "grad_norm": 1.3708258867263794,
+      "grad_norm": 0.9967594742774963,
       "learning_rate": 9.010779249943859e-05,
-      "loss": 0.0956,
+      "loss": 0.0793,
       "step": 1350
     },
     {
       "epoch": 0.4352,
-      "grad_norm": 1.9553757905960083,
+      "grad_norm": 0.9784688353538513,
       "learning_rate": 8.999550864585672e-05,
-      "loss": 0.1139,
+      "loss": 0.0826,
       "step": 1360
     },
     {
       "epoch": 0.4384,
-      "grad_norm": 2.1358702182769775,
+      "grad_norm": 1.4356553554534912,
       "learning_rate": 8.988322479227488e-05,
-      "loss": 0.1094,
+      "loss": 0.0936,
       "step": 1370
     },
     {
       "epoch": 0.4416,
-      "grad_norm": 3.4062862396240234,
+      "grad_norm": 1.7505671977996826,
       "learning_rate": 8.977094093869301e-05,
-      "loss": 0.1017,
+      "loss": 0.0775,
       "step": 1380
     },
     {
       "epoch": 0.4448,
-      "grad_norm": 1.4105799198150635,
+      "grad_norm": 1.1895841360092163,
       "learning_rate": 8.965865708511116e-05,
-      "loss": 0.1008,
+      "loss": 0.087,
       "step": 1390
     },
     {
       "epoch": 0.448,
-      "grad_norm": 1.713500738143921,
+      "grad_norm": 1.9759968519210815,
       "learning_rate": 8.954637323152931e-05,
-      "loss": 0.1078,
+      "loss": 0.0777,
       "step": 1400
     },
     {
       "epoch": 0.4512,
-      "grad_norm": 1.128848671913147,
+      "grad_norm": 1.276041865348816,
       "learning_rate": 8.943408937794746e-05,
-      "loss": 0.093,
+      "loss": 0.0937,
       "step": 1410
     },
     {
       "epoch": 0.4544,
-      "grad_norm": 1.3671025037765503,
+      "grad_norm": 1.0501399040222168,
       "learning_rate": 8.93218055243656e-05,
-      "loss": 0.1093,
+      "loss": 0.0821,
       "step": 1420
     },
     {
       "epoch": 0.4576,
-      "grad_norm": 1.8151606321334839,
+      "grad_norm": 0.9872506856918335,
       "learning_rate": 8.920952167078375e-05,
-      "loss": 0.1088,
+      "loss": 0.076,
       "step": 1430
     },
     {
       "epoch": 0.4608,
-      "grad_norm": 1.1207199096679688,
+      "grad_norm": 1.2520620822906494,
       "learning_rate": 8.909723781720189e-05,
-      "loss": 0.1051,
+      "loss": 0.0719,
       "step": 1440
     },
     {
       "epoch": 0.464,
-      "grad_norm": 1.0649629831314087,
+      "grad_norm": 1.0118720531463623,
       "learning_rate": 8.898495396362003e-05,
-      "loss": 0.1,
+      "loss": 0.0712,
       "step": 1450
     },
     {
       "epoch": 0.4672,
-      "grad_norm": 1.4365625381469727,
+      "grad_norm": 1.3171415328979492,
       "learning_rate": 8.887267011003818e-05,
-      "loss": 0.0861,
+      "loss": 0.0701,
       "step": 1460
     },
     {
       "epoch": 0.4704,
-      "grad_norm": 1.1462955474853516,
+      "grad_norm": 1.0808926820755005,
       "learning_rate": 8.876038625645633e-05,
-      "loss": 0.0866,
+      "loss": 0.0754,
       "step": 1470
     },
     {
       "epoch": 0.4736,
-      "grad_norm": 1.3217401504516602,
+      "grad_norm": 2.9019877910614014,
       "learning_rate": 8.864810240287447e-05,
-      "loss": 0.0979,
+      "loss": 0.0767,
       "step": 1480
     },
     {
       "epoch": 0.4768,
-      "grad_norm": 1.3830331563949585,
+      "grad_norm": 1.3460652828216553,
       "learning_rate": 8.853581854929262e-05,
-      "loss": 0.0792,
+      "loss": 0.0782,
       "step": 1490
     },
     {
       "epoch": 0.48,
-      "grad_norm": 1.777252435684204,
+      "grad_norm": 1.6336652040481567,
       "learning_rate": 8.842353469571077e-05,
-      "loss": 0.0727,
+      "loss": 0.0723,
       "step": 1500
     },
     {
       "epoch": 0.4832,
-      "grad_norm": 1.220021367073059,
+      "grad_norm": 0.9963213205337524,
       "learning_rate": 8.83112508421289e-05,
-      "loss": 0.0836,
+      "loss": 0.068,
       "step": 1510
     },
     {
       "epoch": 0.4864,
-      "grad_norm": 2.3963851928710938,
+      "grad_norm": 1.618598222732544,
       "learning_rate": 8.819896698854705e-05,
-      "loss": 0.0851,
+      "loss": 0.0773,
       "step": 1520
     },
     {
       "epoch": 0.4896,
-      "grad_norm": 1.1223018169403076,
+      "grad_norm": 0.9730167984962463,
       "learning_rate": 8.808668313496519e-05,
-      "loss": 0.082,
+      "loss": 0.0633,
       "step": 1530
     },
     {
       "epoch": 0.4928,
-      "grad_norm": 1.148827314376831,
+      "grad_norm": 1.4831032752990723,
       "learning_rate": 8.797439928138334e-05,
-      "loss": 0.0758,
+      "loss": 0.0641,
       "step": 1540
     },
     {
       "epoch": 0.496,
-      "grad_norm": 0.8612151145935059,
+      "grad_norm": 0.9782041311264038,
       "learning_rate": 8.786211542780148e-05,
-      "loss": 0.0783,
+      "loss": 0.0758,
       "step": 1550
     },
     {
       "epoch": 0.4992,
-      "grad_norm": 1.1042686700820923,
+      "grad_norm": 1.8618321418762207,
       "learning_rate": 8.774983157421963e-05,
-      "loss": 0.0846,
+      "loss": 0.0657,
       "step": 1560
     },
     {
       "epoch": 0.5024,
-      "grad_norm": 1.3059678077697754,
+      "grad_norm": 1.637278437614441,
       "learning_rate": 8.763754772063778e-05,
-      "loss": 0.0842,
+      "loss": 0.0615,
       "step": 1570
     },
     {
       "epoch": 0.5056,
-      "grad_norm": 1.976445198059082,
+      "grad_norm": 0.9550865292549133,
       "learning_rate": 8.752526386705592e-05,
-      "loss": 0.0811,
+      "loss": 0.0614,
       "step": 1580
     },
     {
       "epoch": 0.5088,
-      "grad_norm": 1.7577661275863647,
+      "grad_norm": 2.0610525608062744,
       "learning_rate": 8.741298001347407e-05,
-      "loss": 0.0848,
+      "loss": 0.0657,
       "step": 1590
     },
     {
       "epoch": 0.512,
-      "grad_norm": 1.2085758447647095,
+      "grad_norm": 0.7099101543426514,
       "learning_rate": 8.730069615989222e-05,
-      "loss": 0.077,
+      "loss": 0.0631,
       "step": 1600
     },
     {
       "epoch": 0.5152,
-      "grad_norm": 1.1045840978622437,
+      "grad_norm": 1.1444408893585205,
       "learning_rate": 8.718841230631036e-05,
-      "loss": 0.0794,
+      "loss": 0.0601,
       "step": 1610
     },
     {
       "epoch": 0.5184,
-      "grad_norm": 2.760986328125,
+      "grad_norm": 2.0309245586395264,
       "learning_rate": 8.70761284527285e-05,
-      "loss": 0.0887,
+      "loss": 0.0564,
       "step": 1620
     },
     {
       "epoch": 0.5216,
-      "grad_norm": 1.1649103164672852,
+      "grad_norm": 1.3110053539276123,
       "learning_rate": 8.696384459914664e-05,
-      "loss": 0.0806,
+      "loss": 0.0637,
       "step": 1630
     },
     {
       "epoch": 0.5248,
-      "grad_norm": 2.1718943119049072,
+      "grad_norm": 1.2481273412704468,
       "learning_rate": 8.68515607455648e-05,
-      "loss": 0.0774,
+      "loss": 0.0562,
       "step": 1640
     },
     {
       "epoch": 0.528,
-      "grad_norm": 5.1032586097717285,
+      "grad_norm": 1.6358749866485596,
       "learning_rate": 8.673927689198293e-05,
-      "loss": 0.0823,
+      "loss": 0.0612,
       "step": 1650
     },
     {
       "epoch": 0.5312,
-      "grad_norm": 1.3188016414642334,
+      "grad_norm": 1.46366548538208,
       "learning_rate": 8.662699303840108e-05,
-      "loss": 0.0723,
+      "loss": 0.0603,
       "step": 1660
     },
     {
       "epoch": 0.5344,
-      "grad_norm": 1.1091735363006592,
+      "grad_norm": 0.9687389731407166,
       "learning_rate": 8.651470918481923e-05,
-      "loss": 0.0801,
+      "loss": 0.055,
       "step": 1670
     },
     {
       "epoch": 0.5376,
-      "grad_norm": 1.944667935371399,
+      "grad_norm": 0.8391016125679016,
       "learning_rate": 8.640242533123737e-05,
-      "loss": 0.0656,
+      "loss": 0.0545,
       "step": 1680
     },
     {
       "epoch": 0.5408,
-      "grad_norm": 1.409705400466919,
+      "grad_norm": 1.6513214111328125,
       "learning_rate": 8.629014147765552e-05,
-      "loss": 0.0783,
+      "loss": 0.0581,
       "step": 1690
     },
     {
       "epoch": 0.544,
-      "grad_norm": 1.0202471017837524,
+      "grad_norm": 0.5965524315834045,
       "learning_rate": 8.617785762407367e-05,
-      "loss": 0.0698,
+      "loss": 0.0556,
       "step": 1700
     },
     {
       "epoch": 0.5472,
-      "grad_norm": 0.8339371681213379,
+      "grad_norm": 0.8696075677871704,
       "learning_rate": 8.606557377049181e-05,
-      "loss": 0.0646,
+      "loss": 0.0608,
       "step": 1710
     },
     {
       "epoch": 0.5504,
-      "grad_norm": 1.0416721105575562,
+      "grad_norm": 1.1388493776321411,
       "learning_rate": 8.595328991690995e-05,
-      "loss": 0.0631,
+      "loss": 0.0526,
       "step": 1720
     },
     {
       "epoch": 0.5536,
-      "grad_norm": 0.812588632106781,
+      "grad_norm": 0.9238974452018738,
       "learning_rate": 8.58410060633281e-05,
-      "loss": 0.0718,
+      "loss": 0.0545,
       "step": 1730
     },
     {
       "epoch": 0.5568,
-      "grad_norm": 1.2019861936569214,
+      "grad_norm": 1.38905930519104,
       "learning_rate": 8.572872220974623e-05,
-      "loss": 0.0766,
+      "loss": 0.0585,
       "step": 1740
     },
     {
       "epoch": 0.56,
-      "grad_norm": 0.8724514842033386,
+      "grad_norm": 1.6556098461151123,
       "learning_rate": 8.561643835616438e-05,
-      "loss": 0.066,
+      "loss": 0.054,
       "step": 1750
     },
     {
       "epoch": 0.5632,
-      "grad_norm": 0.9269486665725708,
+      "grad_norm": 0.9753937721252441,
       "learning_rate": 8.550415450258253e-05,
-      "loss": 0.0711,
+      "loss": 0.0489,
       "step": 1760
     },
     {
       "epoch": 0.5664,
-      "grad_norm": 1.1435014009475708,
+      "grad_norm": 1.2688230276107788,
       "learning_rate": 8.539187064900069e-05,
-      "loss": 0.0649,
+      "loss": 0.0496,
       "step": 1770
     },
     {
       "epoch": 0.5696,
-      "grad_norm": 0.9017223119735718,
+      "grad_norm": 1.6871953010559082,
       "learning_rate": 8.527958679541882e-05,
-      "loss": 0.0743,
+      "loss": 0.052,
       "step": 1780
     },
     {
       "epoch": 0.5728,
-      "grad_norm": 1.5114269256591797,
+      "grad_norm": 1.1844048500061035,
       "learning_rate": 8.516730294183697e-05,
-      "loss": 0.0711,
+      "loss": 0.0544,
       "step": 1790
     },
     {
       "epoch": 0.576,
-      "grad_norm": 1.1478126049041748,
+      "grad_norm": 0.7625532746315002,
       "learning_rate": 8.505501908825511e-05,
-      "loss": 0.0749,
+      "loss": 0.0501,
       "step": 1800
     },
     {
       "epoch": 0.5792,
-      "grad_norm": 1.3925341367721558,
+      "grad_norm": 0.9352998733520508,
       "learning_rate": 8.494273523467325e-05,
-      "loss": 0.0623,
+      "loss": 0.0517,
       "step": 1810
     },
     {
       "epoch": 0.5824,
-      "grad_norm": 0.7855392098426819,
+      "grad_norm": 2.186729669570923,
       "learning_rate": 8.48304513810914e-05,
-      "loss": 0.0568,
+      "loss": 0.0565,
       "step": 1820
     },
     {
       "epoch": 0.5856,
-      "grad_norm": 0.7520506381988525,
+      "grad_norm": 0.6474363803863525,
       "learning_rate": 8.471816752750955e-05,
-      "loss": 0.0654,
+      "loss": 0.0512,
       "step": 1830
     },
     {
       "epoch": 0.5888,
-      "grad_norm": 0.8448010087013245,
+      "grad_norm": 1.0736746788024902,
       "learning_rate": 8.460588367392769e-05,
-      "loss": 0.0624,
+      "loss": 0.0484,
       "step": 1840
     },
     {
       "epoch": 0.592,
-      "grad_norm": 0.7805534601211548,
+      "grad_norm": 1.6603658199310303,
       "learning_rate": 8.449359982034584e-05,
-      "loss": 0.0676,
+      "loss": 0.055,
       "step": 1850
     },
     {
       "epoch": 0.5952,
-      "grad_norm": 1.1754975318908691,
+      "grad_norm": 0.8258435726165771,
       "learning_rate": 8.438131596676399e-05,
-      "loss": 0.0609,
+      "loss": 0.0521,
       "step": 1860
     },
     {
       "epoch": 0.5984,
-      "grad_norm": 0.7776190638542175,
+      "grad_norm": 0.8024755716323853,
       "learning_rate": 8.426903211318214e-05,
-      "loss": 0.0601,
+      "loss": 0.0456,
       "step": 1870
     },
     {
       "epoch": 0.6016,
-      "grad_norm": 1.610683560371399,
+      "grad_norm": 1.215813159942627,
       "learning_rate": 8.415674825960028e-05,
-      "loss": 0.0618,
+      "loss": 0.0465,
       "step": 1880
     },
     {
       "epoch": 0.6048,
-      "grad_norm": 0.8926658630371094,
+      "grad_norm": 0.6433802247047424,
       "learning_rate": 8.404446440601843e-05,
-      "loss": 0.0582,
+      "loss": 0.0511,
       "step": 1890
     },
     {
       "epoch": 0.608,
-      "grad_norm": 1.2540363073349,
+      "grad_norm": 0.5610775947570801,
       "learning_rate": 8.393218055243656e-05,
-      "loss": 0.0623,
+      "loss": 0.048,
       "step": 1900
     },
     {
       "epoch": 0.6112,
-      "grad_norm": 1.463120937347412,
+      "grad_norm": 1.1887115240097046,
       "learning_rate": 8.38198966988547e-05,
-      "loss": 0.0595,
+      "loss": 0.047,
       "step": 1910
     },
     {
       "epoch": 0.6144,
-      "grad_norm": 1.3101223707199097,
+      "grad_norm": 0.7299818396568298,
       "learning_rate": 8.370761284527285e-05,
-      "loss": 0.0653,
+      "loss": 0.0498,
       "step": 1920
     },
     {
       "epoch": 0.6176,
-      "grad_norm": 1.1716495752334595,
+      "grad_norm": 1.1920089721679688,
       "learning_rate": 8.3595328991691e-05,
-      "loss": 0.0518,
+      "loss": 0.0458,
       "step": 1930
     },
     {
       "epoch": 0.6208,
-      "grad_norm": 1.539556860923767,
+      "grad_norm": 0.6778120994567871,
       "learning_rate": 8.348304513810914e-05,
-      "loss": 0.0496,
+      "loss": 0.0455,
       "step": 1940
     },
     {
       "epoch": 0.624,
-      "grad_norm": 0.8535395264625549,
+      "grad_norm": 1.1930702924728394,
       "learning_rate": 8.337076128452729e-05,
-      "loss": 0.0581,
+      "loss": 0.1967,
       "step": 1950
     },
     {
       "epoch": 0.6272,
-      "grad_norm": 0.9112345576286316,
+      "grad_norm": 1.2795714139938354,
       "learning_rate": 8.325847743094544e-05,
-      "loss": 0.0568,
+      "loss": 0.9457,
       "step": 1960
     },
     {
       "epoch": 0.6304,
-      "grad_norm": 0.9361368417739868,
+      "grad_norm": 1.2569034099578857,
       "learning_rate": 8.314619357736358e-05,
-      "loss": 0.0507,
+      "loss": 0.0661,
       "step": 1970
     },
     {
       "epoch": 0.6336,
-      "grad_norm": 1.4482288360595703,
+      "grad_norm": 1.6149742603302002,
       "learning_rate": 8.303390972378173e-05,
-      "loss": 0.0521,
+      "loss": 0.0981,
       "step": 1980
     },
     {
       "epoch": 0.6368,
-      "grad_norm": 0.9236279129981995,
+      "grad_norm": 10.0,
       "learning_rate": 8.292162587019986e-05,
-      "loss": 0.0512,
+      "loss": 0.0525,
       "step": 1990
     },
     {
       "epoch": 0.64,
-      "grad_norm": 1.238692045211792,
+      "grad_norm": 2.119809150695801,
       "learning_rate": 8.280934201661802e-05,
-      "loss": 0.0538,
+      "loss": 0.0533,
       "step": 2000
     },
     {
       "epoch": 0.6432,
-      "grad_norm": 0.9093800187110901,
+      "grad_norm": 1.103276252746582,
       "learning_rate": 8.269705816303615e-05,
-      "loss": 0.054,
+      "loss": 0.0471,
       "step": 2010
     },
     {
       "epoch": 0.6464,
-      "grad_norm": 0.7657427787780762,
+      "grad_norm": 0.8100212216377258,
       "learning_rate": 8.25847743094543e-05,
-      "loss": 0.049,
+      "loss": 0.0463,
       "step": 2020
     },
     {
       "epoch": 0.6496,
-      "grad_norm": 1.5526565313339233,
+      "grad_norm": 0.9989579319953918,
       "learning_rate": 8.247249045587245e-05,
-      "loss": 0.0524,
+      "loss": 0.044,
       "step": 2030
     },
     {
       "epoch": 0.6528,
-      "grad_norm": 0.7204543352127075,
+      "grad_norm": 0.8777104616165161,
       "learning_rate": 8.236020660229059e-05,
-      "loss": 0.0555,
+      "loss": 0.0477,
       "step": 2040
     },
     {
       "epoch": 0.656,
-      "grad_norm": 0.8594505190849304,
+      "grad_norm": 0.8403288125991821,
       "learning_rate": 8.224792274870874e-05,
-      "loss": 0.0529,
+      "loss": 0.0427,
       "step": 2050
     },
     {
       "epoch": 0.6592,
-      "grad_norm": 1.0530890226364136,
+      "grad_norm": 0.920377254486084,
       "learning_rate": 8.213563889512689e-05,
-      "loss": 0.0621,
+      "loss": 0.0414,
       "step": 2060
     },
     {
       "epoch": 0.6624,
-      "grad_norm": 0.92835533618927,
+      "grad_norm": 0.8990156650543213,
       "learning_rate": 8.202335504154503e-05,
-      "loss": 0.0434,
+      "loss": 0.0441,
       "step": 2070
     },
     {
       "epoch": 0.6656,
-      "grad_norm": 1.0584304332733154,
+      "grad_norm": 0.6968216896057129,
       "learning_rate": 8.191107118796317e-05,
-      "loss": 0.0515,
+      "loss": 0.0408,
       "step": 2080
     },
     {
       "epoch": 0.6688,
-      "grad_norm": 0.7033383250236511,
+      "grad_norm": 0.8257766366004944,
       "learning_rate": 8.179878733438132e-05,
-      "loss": 0.0451,
+      "loss": 0.0435,
       "step": 2090
     },
     {
       "epoch": 0.672,
-      "grad_norm": 1.09385085105896,
+      "grad_norm": 1.689205288887024,
       "learning_rate": 8.168650348079947e-05,
-      "loss": 0.0486,
+      "loss": 0.0402,
       "step": 2100
     },
     {
       "epoch": 0.6752,
-      "grad_norm": 1.0709513425827026,
+      "grad_norm": 0.766890287399292,
       "learning_rate": 8.15742196272176e-05,
-      "loss": 0.0571,
+      "loss": 0.0393,
       "step": 2110
     },
     {
       "epoch": 0.6784,
-      "grad_norm": 0.7316601276397705,
+      "grad_norm": 0.8452049493789673,
       "learning_rate": 8.146193577363576e-05,
-      "loss": 0.0496,
+      "loss": 0.037,
       "step": 2120
     },
     {
       "epoch": 0.6816,
-      "grad_norm": 0.9458874464035034,
+      "grad_norm": 0.6207525134086609,
       "learning_rate": 8.13496519200539e-05,
-      "loss": 0.054,
+      "loss": 0.0406,
       "step": 2130
     },
     {
       "epoch": 0.6848,
-      "grad_norm": 2.1238739490509033,
+      "grad_norm": 1.4125391244888306,
       "learning_rate": 8.123736806647204e-05,
-      "loss": 0.0535,
+      "loss": 0.041,
       "step": 2140
     },
     {
       "epoch": 0.688,
-      "grad_norm": 1.011391520500183,
+      "grad_norm": 1.1123186349868774,
       "learning_rate": 8.11250842128902e-05,
-      "loss": 0.0527,
+      "loss": 0.0331,
       "step": 2150
     },
     {
       "epoch": 0.6912,
-      "grad_norm": 0.8783167004585266,
+      "grad_norm": 0.769235372543335,
       "learning_rate": 8.101280035930835e-05,
-      "loss": 0.0424,
+      "loss": 0.0432,
       "step": 2160
     },
     {
       "epoch": 0.6944,
-      "grad_norm": 0.9206530451774597,
+      "grad_norm": 0.7295809984207153,
       "learning_rate": 8.090051650572648e-05,
-      "loss": 0.0588,
+      "loss": 0.0395,
       "step": 2170
     },
     {
       "epoch": 0.6976,
-      "grad_norm": 0.5304967164993286,
+      "grad_norm": 1.110095739364624,
       "learning_rate": 8.078823265214462e-05,
-      "loss": 0.0574,
+      "loss": 0.036,
       "step": 2180
     },
     {
       "epoch": 0.7008,
-      "grad_norm": 0.5870644450187683,
+      "grad_norm": 0.5310615301132202,
       "learning_rate": 8.067594879856277e-05,
-      "loss": 0.0512,
+      "loss": 0.0384,
       "step": 2190
     },
     {
       "epoch": 0.704,
-      "grad_norm": 0.7831016182899475,
+      "grad_norm": 0.8980706930160522,
       "learning_rate": 8.056366494498092e-05,
-      "loss": 0.0485,
+      "loss": 0.0331,
       "step": 2200
     },
     {
       "epoch": 0.7072,
-      "grad_norm": 2.5291478633880615,
+      "grad_norm": 0.9225451946258545,
       "learning_rate": 8.045138109139906e-05,
-      "loss": 0.058,
+      "loss": 0.0385,
       "step": 2210
     },
     {
       "epoch": 0.7104,
-      "grad_norm": 0.705797553062439,
+      "grad_norm": 0.6329707503318787,
       "learning_rate": 8.033909723781721e-05,
-      "loss": 0.0534,
+      "loss": 0.038,
       "step": 2220
     },
     {
       "epoch": 0.7136,
-      "grad_norm": 0.7620034217834473,
+      "grad_norm": 0.588439404964447,
       "learning_rate": 8.022681338423536e-05,
-      "loss": 0.0466,
+      "loss": 0.0354,
       "step": 2230
     },
     {
       "epoch": 0.7168,
-      "grad_norm": 2.0571022033691406,
+      "grad_norm": 0.7347375750541687,
       "learning_rate": 8.01145295306535e-05,
-      "loss": 0.0469,
+      "loss": 0.0379,
       "step": 2240
     },
     {
       "epoch": 0.72,
-      "grad_norm": 1.094312310218811,
+      "grad_norm": 1.577707290649414,
       "learning_rate": 8.000224567707165e-05,
-      "loss": 0.0482,
+      "loss": 0.0391,
       "step": 2250
     },
     {
       "epoch": 0.7232,
-      "grad_norm": 0.857584536075592,
+      "grad_norm": 1.1854794025421143,
       "learning_rate": 7.988996182348978e-05,
-      "loss": 0.044,
+      "loss": 0.0354,
       "step": 2260
     },
     {
       "epoch": 0.7264,
-      "grad_norm": 0.7637543082237244,
+      "grad_norm": 0.8479386568069458,
       "learning_rate": 7.977767796990792e-05,
-      "loss": 0.0493,
+      "loss": 0.0399,
       "step": 2270
     },
     {
       "epoch": 0.7296,
-      "grad_norm": 0.6703894138336182,
+      "grad_norm": 0.870980441570282,
       "learning_rate": 7.966539411632607e-05,
-      "loss": 0.0462,
+      "loss": 0.0396,
       "step": 2280
     },
     {
       "epoch": 0.7328,
-      "grad_norm": 0.7242295145988464,
+      "grad_norm": 1.1692736148834229,
       "learning_rate": 7.955311026274422e-05,
-      "loss": 0.044,
+      "loss": 0.0403,
       "step": 2290
     },
     {
       "epoch": 0.736,
-      "grad_norm": 1.0875325202941895,
+      "grad_norm": 1.4128285646438599,
       "learning_rate": 7.944082640916236e-05,
-      "loss": 0.0458,
+      "loss": 0.0408,
       "step": 2300
     },
     {
       "epoch": 0.7392,
-      "grad_norm": 0.8322548270225525,
+      "grad_norm": 0.9887726306915283,
       "learning_rate": 7.932854255558051e-05,
-      "loss": 0.0471,
+      "loss": 0.0382,
       "step": 2310
     },
     {
       "epoch": 0.7424,
-      "grad_norm": 1.6921989917755127,
+      "grad_norm": 0.6805205345153809,
       "learning_rate": 7.921625870199866e-05,
-      "loss": 0.0481,
+      "loss": 0.0344,
       "step": 2320
     },
     {
       "epoch": 0.7456,
-      "grad_norm": 0.9339900612831116,
+      "grad_norm": 0.6173204779624939,
       "learning_rate": 7.910397484841681e-05,
-      "loss": 0.0438,
+      "loss": 0.0339,
       "step": 2330
     },
     {
       "epoch": 0.7488,
-      "grad_norm": 0.9007784724235535,
+      "grad_norm": 1.1889104843139648,
       "learning_rate": 7.899169099483495e-05,
-      "loss": 0.051,
+      "loss": 0.0362,
       "step": 2340
     },
     {
       "epoch": 0.752,
-      "grad_norm": 0.7366623282432556,
+      "grad_norm": 0.812631368637085,
       "learning_rate": 7.887940714125309e-05,
-      "loss": 0.0488,
+      "loss": 0.0373,
       "step": 2350
     },
     {
       "epoch": 0.7552,
-      "grad_norm": 1.0581986904144287,
+      "grad_norm": 0.7912474274635315,
       "learning_rate": 7.876712328767124e-05,
-      "loss": 0.0465,
+      "loss": 0.0347,
       "step": 2360
     },
     {
       "epoch": 0.7584,
-      "grad_norm": 0.8398572206497192,
+      "grad_norm": 0.8621821403503418,
       "learning_rate": 7.865483943408937e-05,
-      "loss": 0.0444,
+      "loss": 0.039,
       "step": 2370
     },
     {
       "epoch": 0.7616,
-      "grad_norm": 0.829765796661377,
+      "grad_norm": 1.7311482429504395,
       "learning_rate": 7.854255558050752e-05,
-      "loss": 0.0468,
+      "loss": 0.0334,
       "step": 2380
     },
     {
       "epoch": 0.7648,
-      "grad_norm": 0.8922726511955261,
+      "grad_norm": 1.260198712348938,
       "learning_rate": 7.843027172692568e-05,
-      "loss": 0.0429,
+      "loss": 0.0364,
       "step": 2390
     },
     {
       "epoch": 0.768,
-      "grad_norm": 0.9981004595756531,
+      "grad_norm": 0.5560032725334167,
       "learning_rate": 7.831798787334381e-05,
-      "loss": 0.0425,
+      "loss": 0.0331,
       "step": 2400
     },
     {
       "epoch": 0.7712,
-      "grad_norm": 1.1781072616577148,
+      "grad_norm": 0.7563883066177368,
       "learning_rate": 7.820570401976196e-05,
-      "loss": 0.0467,
+      "loss": 0.0335,
       "step": 2410
     },
     {
       "epoch": 0.7744,
-      "grad_norm": 1.295114517211914,
+      "grad_norm": 0.991214394569397,
       "learning_rate": 7.809342016618011e-05,
-      "loss": 0.0429,
+      "loss": 0.0348,
       "step": 2420
     },
     {
       "epoch": 0.7776,
-      "grad_norm": 1.2923245429992676,
+      "grad_norm": 0.5576212406158447,
       "learning_rate": 7.798113631259825e-05,
-      "loss": 0.0411,
+      "loss": 0.0344,
       "step": 2430
     },
     {
       "epoch": 0.7808,
-      "grad_norm": 0.9972735643386841,
+      "grad_norm": 0.9191322326660156,
       "learning_rate": 7.78688524590164e-05,
-      "loss": 0.0392,
+      "loss": 0.0376,
       "step": 2440
     },
     {
       "epoch": 0.784,
-      "grad_norm": 0.7741293907165527,
+      "grad_norm": 4.326545715332031,
       "learning_rate": 7.775656860543454e-05,
-      "loss": 0.0432,
+      "loss": 0.0822,
       "step": 2450
     },
     {
       "epoch": 0.7872,
-      "grad_norm": 0.5855127573013306,
+      "grad_norm": 0.7509451508522034,
       "learning_rate": 7.764428475185269e-05,
-      "loss": 0.0468,
+      "loss": 0.0359,
       "step": 2460
     },
     {
       "epoch": 0.7904,
-      "grad_norm": 0.6745654940605164,
+      "grad_norm": 1.5423306226730347,
       "learning_rate": 7.753200089827083e-05,
-      "loss": 0.0394,
+      "loss": 0.0349,
       "step": 2470
     },
     {
       "epoch": 0.7936,
-      "grad_norm": 1.2831262350082397,
+      "grad_norm": 0.6744070053100586,
       "learning_rate": 7.741971704468898e-05,
-      "loss": 0.037,
+      "loss": 0.0322,
       "step": 2480
     },
     {
       "epoch": 0.7968,
-      "grad_norm": 0.6621804237365723,
+      "grad_norm": 0.46497827768325806,
       "learning_rate": 7.730743319110713e-05,
-      "loss": 0.0413,
+      "loss": 0.0299,
       "step": 2490
     },
     {
       "epoch": 0.8,
-      "grad_norm": 0.5354174375534058,
+      "grad_norm": 0.7538796067237854,
       "learning_rate": 7.719514933752526e-05,
-      "loss": 0.0375,
+      "loss": 0.03,
       "step": 2500
     },
     {
       "epoch": 0.8032,
-      "grad_norm": 0.9441866278648376,
+      "grad_norm": 1.8027441501617432,
       "learning_rate": 7.708286548394342e-05,
-      "loss": 0.0364,
+      "loss": 0.0325,
       "step": 2510
     },
     {
       "epoch": 0.8064,
-      "grad_norm": 0.6830460429191589,
+      "grad_norm": 0.7862191796302795,
       "learning_rate": 7.697058163036157e-05,
-      "loss": 0.042,
+      "loss": 0.0346,
       "step": 2520
     },
     {
       "epoch": 0.8096,
-      "grad_norm": 0.6004045605659485,
+      "grad_norm": 0.9970144033432007,
       "learning_rate": 7.68582977767797e-05,
-      "loss": 0.04,
+      "loss": 0.0349,
       "step": 2530
     },
     {
       "epoch": 0.8128,
-      "grad_norm": 0.6852745413780212,
+      "grad_norm": 0.6299024820327759,
       "learning_rate": 7.674601392319784e-05,
-      "loss": 0.0404,
+      "loss": 0.0305,
       "step": 2540
     },
     {
       "epoch": 0.816,
-      "grad_norm": 1.0080032348632812,
+      "grad_norm": 0.883230447769165,
       "learning_rate": 7.663373006961599e-05,
-      "loss": 0.043,
+      "loss": 0.0384,
       "step": 2550
     },
     {
       "epoch": 0.8192,
-      "grad_norm": 0.7791699767112732,
+      "grad_norm": 1.0779157876968384,
       "learning_rate": 7.652144621603414e-05,
-      "loss": 0.0411,
+      "loss": 0.0377,
       "step": 2560
     },
     {
       "epoch": 0.8224,
-      "grad_norm": 1.192091464996338,
+      "grad_norm": 0.8372220396995544,
       "learning_rate": 7.640916236245228e-05,
-      "loss": 0.0413,
+      "loss": 0.0299,
       "step": 2570
     },
     {
       "epoch": 0.8256,
-      "grad_norm": 0.744788646697998,
+      "grad_norm": 0.5274995565414429,
       "learning_rate": 7.629687850887043e-05,
-      "loss": 0.0401,
+      "loss": 0.0338,
       "step": 2580
     },
     {
       "epoch": 0.8288,
-      "grad_norm": 0.6732345223426819,
+      "grad_norm": 0.8475728034973145,
       "learning_rate": 7.618459465528858e-05,
-      "loss": 0.0426,
+      "loss": 0.0314,
       "step": 2590
     },
     {
       "epoch": 0.832,
-      "grad_norm": 1.0868829488754272,
+      "grad_norm": 1.0869860649108887,
       "learning_rate": 7.607231080170672e-05,
-      "loss": 0.0368,
+      "loss": 0.0288,
       "step": 2600
     },
     {
       "epoch": 0.8352,
-      "grad_norm": 1.4248939752578735,
+      "grad_norm": 0.8045733571052551,
       "learning_rate": 7.596002694812487e-05,
-      "loss": 0.0369,
+      "loss": 0.031,
       "step": 2610
     },
     {
       "epoch": 0.8384,
-      "grad_norm": 0.7218601107597351,
+      "grad_norm": 0.815394937992096,
       "learning_rate": 7.5847743094543e-05,
-      "loss": 0.0415,
+      "loss": 0.028,
       "step": 2620
     },
     {
       "epoch": 0.8416,
-      "grad_norm": 0.803717851638794,
+      "grad_norm": 0.7129034996032715,
       "learning_rate": 7.573545924096114e-05,
-      "loss": 0.0352,
+      "loss": 0.031,
       "step": 2630
     },
     {
       "epoch": 0.8448,
-      "grad_norm": 0.821607768535614,
+      "grad_norm": 0.6846269965171814,
       "learning_rate": 7.562317538737929e-05,
-      "loss": 0.039,
+      "loss": 0.036,
       "step": 2640
     },
     {
       "epoch": 0.848,
-      "grad_norm": 1.1404283046722412,
+      "grad_norm": 2.122080087661743,
       "learning_rate": 7.551089153379744e-05,
-      "loss": 0.0369,
+      "loss": 0.0301,
       "step": 2650
     },
     {
       "epoch": 0.8512,
-      "grad_norm": 1.2288737297058105,
+      "grad_norm": 1.3065335750579834,
       "learning_rate": 7.53986076802156e-05,
-      "loss": 0.0417,
+      "loss": 0.0314,
       "step": 2660
     },
     {
       "epoch": 0.8544,
-      "grad_norm": 1.0263468027114868,
+      "grad_norm": 1.0742933750152588,
       "learning_rate": 7.528632382663373e-05,
-      "loss": 0.0423,
+      "loss": 0.0318,
       "step": 2670
     },
     {
       "epoch": 0.8576,
-      "grad_norm": 0.8517736196517944,
+      "grad_norm": 0.6569668650627136,
       "learning_rate": 7.517403997305188e-05,
-      "loss": 0.0364,
+      "loss": 0.0281,
       "step": 2680
     },
     {
       "epoch": 0.8608,
-      "grad_norm": 0.8727993369102478,
+      "grad_norm": 0.6425995826721191,
       "learning_rate": 7.506175611947003e-05,
-      "loss": 0.0382,
+      "loss": 0.0282,
       "step": 2690
     },
     {
       "epoch": 0.864,
-      "grad_norm": 0.7277560234069824,
+      "grad_norm": 0.6735262870788574,
       "learning_rate": 7.494947226588817e-05,
-      "loss": 0.0368,
+      "loss": 0.0303,
       "step": 2700
     },
     {
       "epoch": 0.8672,
-      "grad_norm": 0.854989230632782,
+      "grad_norm": 0.7769756317138672,
       "learning_rate": 7.483718841230631e-05,
-      "loss": 0.0431,
+      "loss": 0.0283,
       "step": 2710
     },
     {
       "epoch": 0.8704,
-      "grad_norm": 0.47089987993240356,
+      "grad_norm": 0.7856665849685669,
       "learning_rate": 7.472490455872446e-05,
-      "loss": 0.0372,
+      "loss": 0.0278,
       "step": 2720
     },
     {
       "epoch": 0.8736,
-      "grad_norm": 0.716643750667572,
+      "grad_norm": 0.9197099208831787,
       "learning_rate": 7.46126207051426e-05,
-      "loss": 0.0348,
+      "loss": 0.0317,
       "step": 2730
     },
     {
       "epoch": 0.8768,
-      "grad_norm": 0.8277871012687683,
+      "grad_norm": 0.4260849952697754,
       "learning_rate": 7.450033685156075e-05,
-      "loss": 0.0369,
+      "loss": 0.0287,
       "step": 2740
     },
     {
       "epoch": 0.88,
-      "grad_norm": 0.9618933796882629,
+      "grad_norm": 0.8799973130226135,
       "learning_rate": 7.43880529979789e-05,
-      "loss": 0.0377,
+      "loss": 0.0294,
       "step": 2750
     },
     {
       "epoch": 0.8832,
-      "grad_norm": 0.6898852586746216,
+      "grad_norm": 1.2114665508270264,
       "learning_rate": 7.427576914439703e-05,
-      "loss": 0.0543,
+      "loss": 0.0331,
       "step": 2760
     },
     {
       "epoch": 0.8864,
-      "grad_norm": 1.4362825155258179,
+      "grad_norm": 0.6824879050254822,
       "learning_rate": 7.416348529081518e-05,
-      "loss": 0.0397,
+      "loss": 0.0296,
       "step": 2770
     },
     {
       "epoch": 0.8896,
-      "grad_norm": 1.1972767114639282,
+      "grad_norm": 1.274061918258667,
       "learning_rate": 7.405120143723333e-05,
-      "loss": 0.0324,
+      "loss": 0.0289,
       "step": 2780
     },
     {
       "epoch": 0.8928,
-      "grad_norm": 0.5438815355300903,
+      "grad_norm": 0.6117509603500366,
       "learning_rate": 7.393891758365149e-05,
-      "loss": 0.0397,
+      "loss": 0.0255,
       "step": 2790
     },
     {
       "epoch": 0.896,
-      "grad_norm": 0.513469398021698,
+      "grad_norm": 0.5068930983543396,
       "learning_rate": 7.382663373006962e-05,
-      "loss": 0.0338,
+      "loss": 0.0281,
       "step": 2800
     },
     {
       "epoch": 0.8992,
-      "grad_norm": 0.5743911266326904,
+      "grad_norm": 0.5966737270355225,
       "learning_rate": 7.371434987648776e-05,
-      "loss": 0.0313,
+      "loss": 0.026,
       "step": 2810
     },
     {
       "epoch": 0.9024,
-      "grad_norm": 1.011957049369812,
+      "grad_norm": 0.7945902347564697,
       "learning_rate": 7.360206602290591e-05,
-      "loss": 0.0374,
+      "loss": 0.0271,
       "step": 2820
     },
     {
       "epoch": 0.9056,
-      "grad_norm": 0.6926620602607727,
+      "grad_norm": 0.7903382182121277,
       "learning_rate": 7.348978216932405e-05,
-      "loss": 0.0392,
+      "loss": 0.0261,
       "step": 2830
     },
     {
       "epoch": 0.9088,
-      "grad_norm": 0.6338510513305664,
+      "grad_norm": 1.1210882663726807,
       "learning_rate": 7.33774983157422e-05,
-      "loss": 0.0359,
+      "loss": 0.0319,
       "step": 2840
     },
     {
       "epoch": 0.912,
-      "grad_norm": 0.7649824023246765,
+      "grad_norm": 0.7618020176887512,
       "learning_rate": 7.326521446216035e-05,
-      "loss": 0.0353,
+      "loss": 0.0319,
       "step": 2850
     },
     {
       "epoch": 0.9152,
-      "grad_norm": 0.8123289346694946,
+      "grad_norm": 0.5775326490402222,
       "learning_rate": 7.315293060857849e-05,
-      "loss": 0.0322,
+      "loss": 0.0293,
       "step": 2860
     },
     {
       "epoch": 0.9184,
-      "grad_norm": 0.8033359050750732,
+      "grad_norm": 0.532517671585083,
       "learning_rate": 7.304064675499664e-05,
-      "loss": 0.0331,
+      "loss": 0.0272,
       "step": 2870
     },
     {
       "epoch": 0.9216,
-      "grad_norm": 0.8859496116638184,
+      "grad_norm": 1.226197361946106,
       "learning_rate": 7.292836290141479e-05,
-      "loss": 0.0352,
+      "loss": 0.0278,
       "step": 2880
     },
     {
       "epoch": 0.9248,
-      "grad_norm": 0.7962930202484131,
+      "grad_norm": 0.5860388875007629,
       "learning_rate": 7.281607904783292e-05,
-      "loss": 0.0373,
+      "loss": 0.0274,
       "step": 2890
     },
     {
       "epoch": 0.928,
-      "grad_norm": 0.746497392654419,
+      "grad_norm": 0.4597097337245941,
       "learning_rate": 7.270379519425106e-05,
-      "loss": 0.0426,
+      "loss": 0.0259,
       "step": 2900
     },
     {
       "epoch": 0.9312,
-      "grad_norm": 0.8344641327857971,
+      "grad_norm": 0.654680073261261,
       "learning_rate": 7.259151134066921e-05,
-      "loss": 0.0349,
+      "loss": 0.0258,
       "step": 2910
     },
     {
       "epoch": 0.9344,
-      "grad_norm": 0.8275250792503357,
+      "grad_norm": 0.47539767622947693,
       "learning_rate": 7.247922748708736e-05,
-      "loss": 0.0358,
+      "loss": 0.023,
       "step": 2920
     },
     {
       "epoch": 0.9376,
-      "grad_norm": 0.5994471907615662,
+      "grad_norm": 0.5486841201782227,
       "learning_rate": 7.23669436335055e-05,
-      "loss": 0.0344,
+      "loss": 0.0227,
       "step": 2930
     },
     {
       "epoch": 0.9408,
-      "grad_norm": 0.6452350616455078,
+      "grad_norm": 0.9400618672370911,
       "learning_rate": 7.225465977992365e-05,
-      "loss": 0.0358,
+      "loss": 0.029,
       "step": 2940
     },
     {
       "epoch": 0.944,
-      "grad_norm": 1.0141571760177612,
+      "grad_norm": 0.8965266942977905,
       "learning_rate": 7.21423759263418e-05,
-      "loss": 0.0347,
+      "loss": 0.029,
       "step": 2950
     },
     {
       "epoch": 0.9472,
-      "grad_norm": 0.832384467124939,
+      "grad_norm": 0.6586723923683167,
       "learning_rate": 7.203009207275994e-05,
-      "loss": 0.0332,
+      "loss": 0.0241,
       "step": 2960
     },
     {
       "epoch": 0.9504,
-      "grad_norm": 0.7129203677177429,
+      "grad_norm": 0.7518870830535889,
       "learning_rate": 7.191780821917809e-05,
-      "loss": 0.0313,
+      "loss": 0.0275,
       "step": 2970
     },
     {
       "epoch": 0.9536,
-      "grad_norm": 0.7890746593475342,
+      "grad_norm": 1.0358211994171143,
       "learning_rate": 7.180552436559623e-05,
-      "loss": 0.0331,
+      "loss": 0.0249,
       "step": 2980
     },
     {
       "epoch": 0.9568,
-      "grad_norm": 1.432335615158081,
+      "grad_norm": 0.6047177314758301,
       "learning_rate": 7.169324051201438e-05,
-      "loss": 0.0353,
+      "loss": 0.029,
       "step": 2990
     },
     {
       "epoch": 0.96,
-      "grad_norm": 1.0536537170410156,
+      "grad_norm": 0.6489489078521729,
       "learning_rate": 7.158095665843251e-05,
-      "loss": 0.039,
+      "loss": 0.0259,
       "step": 3000
     },
     {
       "epoch": 0.9632,
-      "grad_norm": 0.7935389280319214,
+      "grad_norm": 0.6869999766349792,
       "learning_rate": 7.146867280485066e-05,
-      "loss": 0.0334,
+      "loss": 0.0245,
       "step": 3010
     },
     {
       "epoch": 0.9664,
-      "grad_norm": 1.4054580926895142,
+      "grad_norm": 0.5727821588516235,
       "learning_rate": 7.135638895126882e-05,
-      "loss": 0.033,
+      "loss": 0.0236,
       "step": 3020
     },
     {
       "epoch": 0.9696,
-      "grad_norm": 0.6271975040435791,
+      "grad_norm": 0.9350295066833496,
       "learning_rate": 7.124410509768695e-05,
-      "loss": 0.0327,
+      "loss": 0.0253,
       "step": 3030
     },
     {
       "epoch": 0.9728,
-      "grad_norm": 0.5951416492462158,
+      "grad_norm": 0.8133239150047302,
       "learning_rate": 7.11318212441051e-05,
-      "loss": 0.0325,
+      "loss": 0.0266,
       "step": 3040
     },
     {
       "epoch": 0.976,
-      "grad_norm": 0.6794223785400391,
+      "grad_norm": 0.39781591296195984,
       "learning_rate": 7.101953739052325e-05,
-      "loss": 0.0336,
+      "loss": 0.0241,
       "step": 3050
     },
     {
       "epoch": 0.9792,
-      "grad_norm": 1.084647536277771,
+      "grad_norm": 0.8405945897102356,
       "learning_rate": 7.090725353694139e-05,
-      "loss": 0.035,
+      "loss": 0.0253,
       "step": 3060
     },
     {
       "epoch": 0.9824,
-      "grad_norm": 0.40548598766326904,
+      "grad_norm": 0.6194741129875183,
       "learning_rate": 7.079496968335954e-05,
-      "loss": 0.0277,
+      "loss": 0.0253,
       "step": 3070
     },
     {
       "epoch": 0.9856,
-      "grad_norm": 0.6343255043029785,
+      "grad_norm": 0.5068514943122864,
       "learning_rate": 7.068268582977768e-05,
-      "loss": 0.0282,
+      "loss": 0.0226,
       "step": 3080
     },
     {
       "epoch": 0.9888,
-      "grad_norm": 0.53138667345047,
+      "grad_norm": 0.8302213549613953,
       "learning_rate": 7.057040197619582e-05,
-      "loss": 0.032,
+      "loss": 0.0237,
       "step": 3090
     },
     {
       "epoch": 0.992,
-      "grad_norm": 0.7178220748901367,
+      "grad_norm": 0.9723827838897705,
       "learning_rate": 7.045811812261397e-05,
-      "loss": 0.0323,
+      "loss": 0.0256,
       "step": 3100
     },
     {
       "epoch": 0.9952,
-      "grad_norm": 0.5384820103645325,
+      "grad_norm": 0.9212355017662048,
       "learning_rate": 7.034583426903212e-05,
-      "loss": 0.0319,
+      "loss": 0.0243,
       "step": 3110
     },
     {
       "epoch": 0.9984,
-      "grad_norm": 1.4491897821426392,
+      "grad_norm": 0.3177473247051239,
       "learning_rate": 7.023355041545027e-05,
-      "loss": 0.0345,
+      "loss": 0.0201,
       "step": 3120
     }
   ],

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:58dc4832b9ecbedb58e177e8210247010a0ec93903efa66a51b87b3bc91d64e4
+oid sha256:f1a838d849888b7d1ac8e4ee59146d9c162c6903d6b779d4f8666af6807ac008
 size 5304