diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -10,5634 +10,5634 @@
   "log_history": [
     {
       "epoch": 0.0,
-      "grad_norm": 20.5,
-      "learning_rate": 2.9999999999999997e-06,
-      "loss": 1.7805,
+      "grad_norm": 20.375,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 1.7815,
       "step": 10
     },
     {
       "epoch": 0.0,
-      "grad_norm": 19.625,
-      "learning_rate": 5.999999999999999e-06,
-      "loss": 1.8292,
+      "grad_norm": 19.125,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 1.84,
       "step": 20
     },
     {
       "epoch": 0.0,
-      "grad_norm": 16.875,
-      "learning_rate": 8.999999999999999e-06,
-      "loss": 2.0677,
+      "grad_norm": 17.0,
+      "learning_rate": 9e-07,
+      "loss": 2.098,
       "step": 30
     },
     {
       "epoch": 0.0,
-      "grad_norm": 17.875,
-      "learning_rate": 1.1999999999999999e-05,
-      "loss": 1.6027,
+      "grad_norm": 20.125,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.67,
       "step": 40
     },
     {
       "epoch": 0.0,
-      "grad_norm": 21.625,
-      "learning_rate": 1.4999999999999999e-05,
-      "loss": 1.7499,
+      "grad_norm": 18.0,
+      "learning_rate": 1.5e-06,
+      "loss": 1.8452,
       "step": 50
     },
     {
       "epoch": 0.0,
-      "grad_norm": 21.125,
-      "learning_rate": 1.7999999999999997e-05,
-      "loss": 2.0222,
+      "grad_norm": 19.25,
+      "learning_rate": 1.8e-06,
+      "loss": 2.1664,
       "step": 60
     },
     {
       "epoch": 0.0,
-      "grad_norm": 22.75,
-      "learning_rate": 2.1e-05,
-      "loss": 1.4821,
+      "grad_norm": 17.25,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 1.6483,
       "step": 70
     },
     {
       "epoch": 0.0,
-      "grad_norm": 25.75,
-      "learning_rate": 2.3999999999999997e-05,
-      "loss": 1.7637,
+      "grad_norm": 18.625,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.9371,
       "step": 80
     },
     {
       "epoch": 0.0,
-      "grad_norm": 17.25,
-      "learning_rate": 2.6999999999999996e-05,
-      "loss": 1.6784,
+      "grad_norm": 16.75,
+      "learning_rate": 2.7e-06,
+      "loss": 1.8967,
       "step": 90
     },
     {
       "epoch": 0.01,
-      "grad_norm": 11.9375,
-      "learning_rate": 2.9999999999999997e-05,
-      "loss": 1.6543,
+      "grad_norm": 14.3125,
+      "learning_rate": 3e-06,
+      "loss": 1.8326,
       "step": 100
     },
     {
       "epoch": 0.01,
-      "grad_norm": 25.875,
-      "learning_rate": 3.2999999999999996e-05,
-      "loss": 1.7161,
+      "grad_norm": 15.9375,
+      "learning_rate": 3.3e-06,
+      "loss": 1.9072,
       "step": 110
     },
     {
       "epoch": 0.01,
-      "grad_norm": 38.5,
-      "learning_rate": 3.5999999999999994e-05,
-      "loss": 1.8143,
+      "grad_norm": 21.5,
+      "learning_rate": 3.6e-06,
+      "loss": 1.9562,
       "step": 120
     },
     {
       "epoch": 0.01,
-      "grad_norm": 19.375,
-      "learning_rate": 3.9e-05,
-      "loss": 1.5947,
+      "grad_norm": 21.75,
+      "learning_rate": 3.9e-06,
+      "loss": 1.6976,
       "step": 130
     },
     {
       "epoch": 0.01,
-      "grad_norm": 16.5,
-      "learning_rate": 4.2e-05,
-      "loss": 2.1477,
+      "grad_norm": 16.0,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 2.2788,
       "step": 140
     },
     {
       "epoch": 0.01,
-      "grad_norm": 18.875,
-      "learning_rate": 4.4999999999999996e-05,
-      "loss": 1.8095,
+      "grad_norm": 17.625,
+      "learning_rate": 4.5e-06,
+      "loss": 1.8362,
       "step": 150
     },
     {
       "epoch": 0.01,
-      "grad_norm": 17.875,
-      "learning_rate": 4.7999999999999994e-05,
-      "loss": 1.7693,
+      "grad_norm": 22.5,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.8372,
       "step": 160
     },
     {
       "epoch": 0.01,
-      "grad_norm": 12.9375,
-      "learning_rate": 5.1e-05,
-      "loss": 1.9624,
+      "grad_norm": 16.75,
+      "learning_rate": 5.1e-06,
+      "loss": 1.9134,
       "step": 170
     },
     {
       "epoch": 0.01,
-      "grad_norm": 11.4375,
-      "learning_rate": 5.399999999999999e-05,
-      "loss": 1.7321,
+      "grad_norm": 21.875,
+      "learning_rate": 5.4e-06,
+      "loss": 1.8973,
       "step": 180
     },
     {
       "epoch": 0.01,
-      "grad_norm": 18.125,
-      "learning_rate": 5.6999999999999996e-05,
-      "loss": 1.7955,
+      "grad_norm": 15.8125,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 1.6868,
       "step": 190
     },
     {
       "epoch": 0.01,
-      "grad_norm": 15.9375,
-      "learning_rate": 5.9999999999999995e-05,
-      "loss": 1.7011,
+      "grad_norm": 19.75,
+      "learning_rate": 6e-06,
+      "loss": 1.6079,
       "step": 200
     },
     {
       "epoch": 0.01,
-      "grad_norm": 14.0625,
-      "learning_rate": 6.299999999999999e-05,
-      "loss": 1.5647,
+      "grad_norm": 19.5,
+      "learning_rate": 6.3e-06,
+      "loss": 1.4638,
       "step": 210
     },
     {
       "epoch": 0.01,
-      "grad_norm": 12.6875,
-      "learning_rate": 6.599999999999999e-05,
-      "loss": 1.8383,
+      "grad_norm": 18.625,
+      "learning_rate": 6.6e-06,
+      "loss": 1.8714,
       "step": 220
     },
     {
       "epoch": 0.01,
-      "grad_norm": 16.125,
-      "learning_rate": 6.9e-05,
-      "loss": 1.9608,
+      "grad_norm": 24.25,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 1.9379,
       "step": 230
     },
     {
       "epoch": 0.01,
-      "grad_norm": 12.5,
-      "learning_rate": 7.199999999999999e-05,
-      "loss": 1.4068,
+      "grad_norm": 13.0,
+      "learning_rate": 7.2e-06,
+      "loss": 1.364,
       "step": 240
     },
     {
       "epoch": 0.01,
-      "grad_norm": 11.0,
-      "learning_rate": 7.5e-05,
-      "loss": 1.9903,
+      "grad_norm": 17.75,
+      "learning_rate": 7.5e-06,
+      "loss": 1.8867,
       "step": 250
     },
     {
       "epoch": 0.01,
-      "grad_norm": 10.1875,
-      "learning_rate": 7.8e-05,
-      "loss": 2.0868,
+      "grad_norm": 14.0,
+      "learning_rate": 7.8e-06,
+      "loss": 1.9215,
       "step": 260
     },
     {
       "epoch": 0.01,
-      "grad_norm": 14.8125,
-      "learning_rate": 8.1e-05,
-      "loss": 2.0174,
+      "grad_norm": 23.5,
+      "learning_rate": 8.1e-06,
+      "loss": 1.8065,
       "step": 270
     },
     {
       "epoch": 0.01,
-      "grad_norm": 10.6875,
-      "learning_rate": 8.4e-05,
-      "loss": 2.0084,
+      "grad_norm": 17.25,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.6864,
       "step": 280
     },
     {
       "epoch": 0.01,
-      "grad_norm": 10.375,
-      "learning_rate": 8.699999999999999e-05,
-      "loss": 1.9688,
+      "grad_norm": 25.25,
+      "learning_rate": 8.7e-06,
+      "loss": 1.6924,
       "step": 290
     },
     {
       "epoch": 0.01,
-      "grad_norm": 10.625,
-      "learning_rate": 8.999999999999999e-05,
-      "loss": 1.9387,
+      "grad_norm": 25.0,
+      "learning_rate": 9e-06,
+      "loss": 1.7671,
       "step": 300
     },
     {
       "epoch": 0.02,
-      "grad_norm": 10.875,
-      "learning_rate": 9.3e-05,
-      "loss": 2.1617,
+      "grad_norm": 18.375,
+      "learning_rate": 9.3e-06,
+      "loss": 1.8781,
       "step": 310
     },
     {
       "epoch": 0.02,
-      "grad_norm": 10.5625,
-      "learning_rate": 9.599999999999999e-05,
-      "loss": 1.9997,
+      "grad_norm": 20.5,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.454,
       "step": 320
     },
     {
       "epoch": 0.02,
-      "grad_norm": 8.9375,
-      "learning_rate": 9.9e-05,
-      "loss": 1.5052,
+      "grad_norm": 14.875,
+      "learning_rate": 9.9e-06,
+      "loss": 1.2016,
       "step": 330
     },
     {
       "epoch": 0.02,
-      "grad_norm": 10.4375,
-      "learning_rate": 0.000102,
-      "loss": 2.1021,
+      "grad_norm": 23.625,
+      "learning_rate": 1.02e-05,
+      "loss": 1.6703,
       "step": 340
     },
     {
       "epoch": 0.02,
-      "grad_norm": 11.3125,
-      "learning_rate": 0.00010499999999999999,
-      "loss": 1.8825,
+      "grad_norm": 26.125,
+      "learning_rate": 1.05e-05,
+      "loss": 1.3712,
       "step": 350
     },
     {
       "epoch": 0.02,
-      "grad_norm": 13.625,
-      "learning_rate": 0.00010799999999999998,
-      "loss": 1.9024,
+      "grad_norm": 18.375,
+      "learning_rate": 1.08e-05,
+      "loss": 1.3751,
       "step": 360
     },
     {
       "epoch": 0.02,
-      "grad_norm": 18.125,
-      "learning_rate": 0.00011099999999999999,
-      "loss": 2.2218,
+      "grad_norm": 17.625,
+      "learning_rate": 1.11e-05,
+      "loss": 1.7888,
       "step": 370
     },
     {
       "epoch": 0.02,
-      "grad_norm": 12.6875,
-      "learning_rate": 0.00011399999999999999,
-      "loss": 1.9759,
+      "grad_norm": 34.25,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 1.5242,
       "step": 380
     },
     {
       "epoch": 0.02,
-      "grad_norm": 10.5625,
-      "learning_rate": 0.000117,
-      "loss": 1.9719,
+      "grad_norm": 22.5,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 1.5371,
       "step": 390
     },
     {
       "epoch": 0.02,
-      "grad_norm": 7.09375,
-      "learning_rate": 0.00011999999999999999,
-      "loss": 1.7527,
+      "grad_norm": 31.375,
+      "learning_rate": 1.2e-05,
+      "loss": 1.3818,
       "step": 400
     },
     {
       "epoch": 0.02,
-      "grad_norm": 8.5,
-      "learning_rate": 0.00012299999999999998,
-      "loss": 2.3111,
+      "grad_norm": 13.375,
+      "learning_rate": 1.2299999999999999e-05,
+      "loss": 1.9301,
       "step": 410
     },
     {
       "epoch": 0.02,
-      "grad_norm": 7.9375,
-      "learning_rate": 0.00012599999999999997,
-      "loss": 2.185,
+      "grad_norm": 17.25,
+      "learning_rate": 1.26e-05,
+      "loss": 1.7813,
       "step": 420
     },
     {
       "epoch": 0.02,
-      "grad_norm": 8.375,
-      "learning_rate": 0.000129,
-      "loss": 2.0823,
+      "grad_norm": 22.125,
+      "learning_rate": 1.29e-05,
+      "loss": 1.6075,
       "step": 430
     },
     {
       "epoch": 0.02,
-      "grad_norm": 9.5,
-      "learning_rate": 0.00013199999999999998,
-      "loss": 2.155,
+      "grad_norm": 25.125,
+      "learning_rate": 1.32e-05,
+      "loss": 1.9157,
       "step": 440
     },
     {
       "epoch": 0.02,
-      "grad_norm": 8.5625,
-      "learning_rate": 0.000135,
-      "loss": 2.1027,
+      "grad_norm": 16.375,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 1.6057,
       "step": 450
     },
     {
       "epoch": 0.02,
-      "grad_norm": 7.75,
-      "learning_rate": 0.000138,
-      "loss": 2.3035,
+      "grad_norm": 19.0,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 1.6342,
       "step": 460
     },
     {
       "epoch": 0.02,
-      "grad_norm": 7.875,
-      "learning_rate": 0.00014099999999999998,
-      "loss": 1.9003,
+      "grad_norm": 13.625,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 1.4012,
       "step": 470
     },
     {
       "epoch": 0.02,
-      "grad_norm": 9.125,
-      "learning_rate": 0.00014399999999999998,
-      "loss": 2.4984,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.44e-05,
+      "loss": 1.853,
       "step": 480
     },
     {
       "epoch": 0.02,
-      "grad_norm": 8.0,
-      "learning_rate": 0.000147,
-      "loss": 2.1406,
+      "grad_norm": 21.375,
+      "learning_rate": 1.47e-05,
+      "loss": 1.6546,
       "step": 490
     },
     {
       "epoch": 0.03,
-      "grad_norm": 8.625,
-      "learning_rate": 0.00015,
-      "loss": 2.0724,
+      "grad_norm": 20.0,
+      "learning_rate": 1.5e-05,
+      "loss": 1.3404,
       "step": 500
     },
     {
       "epoch": 0.03,
-      "grad_norm": 7.8125,
-      "learning_rate": 0.00014984210526315788,
-      "loss": 2.2684,
+      "grad_norm": 18.25,
+      "learning_rate": 1.4984210526315789e-05,
+      "loss": 1.5801,
       "step": 510
     },
     {
       "epoch": 0.03,
-      "grad_norm": 7.03125,
-      "learning_rate": 0.00014968421052631578,
-      "loss": 2.0504,
+      "grad_norm": 27.5,
+      "learning_rate": 1.496842105263158e-05,
+      "loss": 1.5567,
       "step": 520
     },
     {
       "epoch": 0.03,
-      "grad_norm": 8.125,
-      "learning_rate": 0.00014952631578947368,
-      "loss": 2.3278,
+      "grad_norm": 28.25,
+      "learning_rate": 1.4952631578947368e-05,
+      "loss": 1.7417,
       "step": 530
     },
     {
       "epoch": 0.03,
-      "grad_norm": 7.5625,
-      "learning_rate": 0.00014936842105263157,
-      "loss": 2.1526,
+      "grad_norm": 27.0,
+      "learning_rate": 1.4936842105263158e-05,
+      "loss": 1.6283,
       "step": 540
     },
     {
       "epoch": 0.03,
-      "grad_norm": 7.90625,
-      "learning_rate": 0.00014921052631578947,
-      "loss": 2.3046,
+      "grad_norm": 25.5,
+      "learning_rate": 1.4921052631578947e-05,
+      "loss": 1.8534,
       "step": 550
     },
     {
       "epoch": 0.03,
-      "grad_norm": 4.625,
-      "learning_rate": 0.00014905263157894737,
-      "loss": 2.1977,
+      "grad_norm": 11.625,
+      "learning_rate": 1.4905263157894737e-05,
+      "loss": 1.668,
       "step": 560
     },
     {
       "epoch": 0.03,
-      "grad_norm": 6.25,
-      "learning_rate": 0.00014889473684210526,
-      "loss": 2.0804,
+      "grad_norm": 39.75,
+      "learning_rate": 1.4889473684210526e-05,
+      "loss": 1.4711,
       "step": 570
     },
     {
       "epoch": 0.03,
-      "grad_norm": 5.65625,
-      "learning_rate": 0.00014873684210526313,
-      "loss": 1.9854,
+      "grad_norm": 20.125,
+      "learning_rate": 1.4873684210526315e-05,
+      "loss": 1.4484,
       "step": 580
     },
     {
       "epoch": 0.03,
-      "grad_norm": 6.46875,
-      "learning_rate": 0.00014857894736842103,
-      "loss": 2.319,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.4857894736842107e-05,
+      "loss": 1.6434,
       "step": 590
     },
     {
       "epoch": 0.03,
-      "grad_norm": 7.53125,
-      "learning_rate": 0.00014842105263157893,
-      "loss": 2.5894,
+      "grad_norm": 21.375,
+      "learning_rate": 1.4842105263157895e-05,
+      "loss": 1.8766,
       "step": 600
     },
     {
       "epoch": 0.03,
-      "grad_norm": 4.625,
-      "learning_rate": 0.00014826315789473682,
-      "loss": 1.7258,
+      "grad_norm": 22.125,
+      "learning_rate": 1.4826315789473686e-05,
+      "loss": 1.329,
       "step": 610
     },
     {
       "epoch": 0.03,
-      "grad_norm": 6.3125,
-      "learning_rate": 0.00014810526315789472,
-      "loss": 2.1868,
+      "grad_norm": 22.5,
+      "learning_rate": 1.4810526315789474e-05,
+      "loss": 1.5956,
       "step": 620
     },
     {
       "epoch": 0.03,
-      "grad_norm": 5.125,
-      "learning_rate": 0.00014794736842105262,
-      "loss": 2.2646,
+      "grad_norm": 9.125,
+      "learning_rate": 1.4794736842105265e-05,
+      "loss": 1.5795,
       "step": 630
     },
     {
       "epoch": 0.03,
-      "grad_norm": 5.71875,
-      "learning_rate": 0.0001477894736842105,
-      "loss": 2.0835,
+      "grad_norm": 15.25,
+      "learning_rate": 1.4778947368421053e-05,
+      "loss": 1.7082,
       "step": 640
     },
     {
       "epoch": 0.03,
-      "grad_norm": 7.0625,
-      "learning_rate": 0.0001476315789473684,
-      "loss": 2.1409,
+      "grad_norm": 25.0,
+      "learning_rate": 1.4763157894736842e-05,
+      "loss": 1.7773,
       "step": 650
     },
     {
       "epoch": 0.03,
-      "grad_norm": 6.84375,
-      "learning_rate": 0.0001474736842105263,
-      "loss": 2.0042,
+      "grad_norm": 25.5,
+      "learning_rate": 1.4747368421052632e-05,
+      "loss": 1.3858,
       "step": 660
     },
     {
       "epoch": 0.03,
-      "grad_norm": 5.71875,
-      "learning_rate": 0.0001473157894736842,
-      "loss": 2.3385,
+      "grad_norm": 19.75,
+      "learning_rate": 1.4731578947368421e-05,
+      "loss": 1.6927,
       "step": 670
     },
     {
       "epoch": 0.03,
-      "grad_norm": 4.40625,
-      "learning_rate": 0.0001471578947368421,
-      "loss": 2.0918,
+      "grad_norm": 10.75,
+      "learning_rate": 1.4715789473684211e-05,
+      "loss": 1.5281,
       "step": 680
     },
     {
       "epoch": 0.03,
-      "grad_norm": 5.03125,
-      "learning_rate": 0.000147,
-      "loss": 1.9163,
+      "grad_norm": 30.125,
+      "learning_rate": 1.47e-05,
+      "loss": 1.3842,
       "step": 690
     },
     {
       "epoch": 0.04,
-      "grad_norm": 9.1875,
-      "learning_rate": 0.0001468421052631579,
-      "loss": 2.2878,
+      "grad_norm": 41.5,
+      "learning_rate": 1.468421052631579e-05,
+      "loss": 1.7584,
       "step": 700
     },
     {
       "epoch": 0.04,
-      "grad_norm": 4.6875,
-      "learning_rate": 0.0001466842105263158,
-      "loss": 2.0185,
+      "grad_norm": 18.875,
+      "learning_rate": 1.4668421052631579e-05,
+      "loss": 1.5485,
       "step": 710
     },
     {
       "epoch": 0.04,
-      "grad_norm": 7.03125,
-      "learning_rate": 0.00014652631578947366,
-      "loss": 2.1358,
+      "grad_norm": 35.25,
+      "learning_rate": 1.4652631578947367e-05,
+      "loss": 1.61,
       "step": 720
     },
     {
       "epoch": 0.04,
-      "grad_norm": 5.03125,
-      "learning_rate": 0.00014636842105263158,
-      "loss": 2.4189,
+      "grad_norm": 19.25,
+      "learning_rate": 1.4636842105263158e-05,
+      "loss": 1.6709,
       "step": 730
     },
     {
       "epoch": 0.04,
-      "grad_norm": 5.40625,
-      "learning_rate": 0.00014621052631578945,
-      "loss": 1.9183,
+      "grad_norm": 17.5,
+      "learning_rate": 1.4621052631578946e-05,
+      "loss": 1.4464,
       "step": 740
     },
     {
       "epoch": 0.04,
-      "grad_norm": 5.40625,
-      "learning_rate": 0.00014605263157894735,
-      "loss": 2.2969,
+      "grad_norm": 18.625,
+      "learning_rate": 1.4605263157894737e-05,
+      "loss": 1.5036,
       "step": 750
     },
     {
       "epoch": 0.04,
-      "grad_norm": 5.625,
-      "learning_rate": 0.00014589473684210524,
-      "loss": 2.2832,
+      "grad_norm": 22.25,
+      "learning_rate": 1.4589473684210527e-05,
+      "loss": 1.6983,
       "step": 760
     },
     {
       "epoch": 0.04,
-      "grad_norm": 5.625,
-      "learning_rate": 0.00014573684210526314,
-      "loss": 2.1994,
+      "grad_norm": 32.5,
+      "learning_rate": 1.4573684210526317e-05,
+      "loss": 1.4551,
       "step": 770
     },
     {
       "epoch": 0.04,
-      "grad_norm": 5.5625,
-      "learning_rate": 0.00014557894736842104,
-      "loss": 2.0053,
+      "grad_norm": 18.625,
+      "learning_rate": 1.4557894736842106e-05,
+      "loss": 1.2903,
       "step": 780
     },
     {
       "epoch": 0.04,
-      "grad_norm": 7.65625,
-      "learning_rate": 0.00014542105263157893,
-      "loss": 2.2771,
+      "grad_norm": 25.875,
+      "learning_rate": 1.4542105263157895e-05,
+      "loss": 1.5937,
       "step": 790
     },
     {
       "epoch": 0.04,
-      "grad_norm": 7.4375,
-      "learning_rate": 0.00014526315789473683,
-      "loss": 2.1748,
+      "grad_norm": 24.125,
+      "learning_rate": 1.4526315789473685e-05,
+      "loss": 1.5994,
       "step": 800
     },
     {
       "epoch": 0.04,
-      "grad_norm": 7.59375,
-      "learning_rate": 0.00014510526315789473,
-      "loss": 1.9644,
+      "grad_norm": 26.375,
+      "learning_rate": 1.4510526315789474e-05,
+      "loss": 1.4018,
       "step": 810
     },
     {
       "epoch": 0.04,
-      "grad_norm": 6.09375,
-      "learning_rate": 0.00014494736842105262,
-      "loss": 1.6369,
+      "grad_norm": 18.75,
+      "learning_rate": 1.4494736842105264e-05,
+      "loss": 1.1078,
       "step": 820
     },
     {
       "epoch": 0.04,
-      "grad_norm": 5.1875,
-      "learning_rate": 0.00014478947368421052,
-      "loss": 2.2513,
+      "grad_norm": 19.25,
+      "learning_rate": 1.4478947368421053e-05,
+      "loss": 1.5669,
       "step": 830
     },
     {
       "epoch": 0.04,
-      "grad_norm": 6.6875,
-      "learning_rate": 0.00014463157894736842,
-      "loss": 2.0433,
+      "grad_norm": 16.75,
+      "learning_rate": 1.4463157894736843e-05,
+      "loss": 1.325,
       "step": 840
     },
     {
       "epoch": 0.04,
-      "grad_norm": 6.46875,
-      "learning_rate": 0.00014447368421052631,
-      "loss": 1.8816,
+      "grad_norm": 18.5,
+      "learning_rate": 1.4447368421052632e-05,
+      "loss": 1.3645,
       "step": 850
     },
     {
       "epoch": 0.04,
-      "grad_norm": 4.75,
-      "learning_rate": 0.00014431578947368418,
-      "loss": 2.0584,
+      "grad_norm": 26.125,
+      "learning_rate": 1.443157894736842e-05,
+      "loss": 1.5002,
       "step": 860
     },
     {
       "epoch": 0.04,
-      "grad_norm": 5.03125,
-      "learning_rate": 0.0001441578947368421,
-      "loss": 1.9437,
+      "grad_norm": 22.875,
+      "learning_rate": 1.441578947368421e-05,
+      "loss": 1.3608,
       "step": 870
     },
     {
       "epoch": 0.04,
-      "grad_norm": 6.625,
-      "learning_rate": 0.00014399999999999998,
-      "loss": 2.1637,
+      "grad_norm": 20.625,
+      "learning_rate": 1.44e-05,
+      "loss": 1.5648,
       "step": 880
     },
     {
       "epoch": 0.04,
-      "grad_norm": 6.34375,
-      "learning_rate": 0.00014384210526315787,
-      "loss": 1.6931,
+      "grad_norm": 32.0,
+      "learning_rate": 1.438421052631579e-05,
+      "loss": 1.1969,
       "step": 890
     },
     {
       "epoch": 0.04,
-      "grad_norm": 6.15625,
-      "learning_rate": 0.00014368421052631577,
-      "loss": 2.1308,
+      "grad_norm": 18.25,
+      "learning_rate": 1.4368421052631578e-05,
+      "loss": 1.5523,
       "step": 900
     },
     {
       "epoch": 0.05,
-      "grad_norm": 4.75,
-      "learning_rate": 0.00014352631578947367,
-      "loss": 1.9802,
+      "grad_norm": 27.5,
+      "learning_rate": 1.4352631578947369e-05,
+      "loss": 1.5062,
       "step": 910
     },
     {
       "epoch": 0.05,
-      "grad_norm": 5.3125,
-      "learning_rate": 0.00014336842105263156,
-      "loss": 1.8052,
+      "grad_norm": 14.75,
+      "learning_rate": 1.4336842105263159e-05,
+      "loss": 1.5052,
       "step": 920
     },
     {
       "epoch": 0.05,
-      "grad_norm": 5.46875,
-      "learning_rate": 0.00014321052631578946,
-      "loss": 2.2819,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.4321052631578948e-05,
+      "loss": 1.6696,
       "step": 930
     },
     {
       "epoch": 0.05,
-      "grad_norm": 4.25,
-      "learning_rate": 0.00014305263157894736,
-      "loss": 1.9152,
+      "grad_norm": 18.0,
+      "learning_rate": 1.4305263157894738e-05,
+      "loss": 1.2593,
       "step": 940
     },
     {
       "epoch": 0.05,
-      "grad_norm": 7.03125,
-      "learning_rate": 0.00014289473684210525,
-      "loss": 1.6635,
+      "grad_norm": 22.0,
+      "learning_rate": 1.4289473684210527e-05,
+      "loss": 0.9601,
       "step": 950
     },
     {
       "epoch": 0.05,
-      "grad_norm": 5.4375,
-      "learning_rate": 0.00014273684210526315,
-      "loss": 1.8765,
+      "grad_norm": 17.5,
+      "learning_rate": 1.4273684210526317e-05,
+      "loss": 1.3436,
       "step": 960
     },
     {
       "epoch": 0.05,
-      "grad_norm": 6.21875,
-      "learning_rate": 0.00014257894736842105,
-      "loss": 2.5049,
+      "grad_norm": 29.25,
+      "learning_rate": 1.4257894736842106e-05,
+      "loss": 1.9493,
       "step": 970
     },
     {
       "epoch": 0.05,
-      "grad_norm": 4.28125,
-      "learning_rate": 0.00014242105263157894,
-      "loss": 1.8173,
+      "grad_norm": 14.375,
+      "learning_rate": 1.4242105263157896e-05,
+      "loss": 1.3317,
       "step": 980
     },
     {
       "epoch": 0.05,
-      "grad_norm": 3.609375,
-      "learning_rate": 0.00014226315789473684,
-      "loss": 1.8379,
+      "grad_norm": 12.625,
+      "learning_rate": 1.4226315789473685e-05,
+      "loss": 1.263,
       "step": 990
     },
     {
       "epoch": 0.05,
-      "grad_norm": 7.40625,
-      "learning_rate": 0.0001421052631578947,
-      "loss": 2.0426,
+      "grad_norm": 25.625,
+      "learning_rate": 1.4210526315789473e-05,
+      "loss": 1.4426,
       "step": 1000
     },
     {
       "epoch": 0.05,
-      "grad_norm": 7.25,
-      "learning_rate": 0.00014194736842105263,
-      "loss": 2.1772,
+      "grad_norm": 27.75,
+      "learning_rate": 1.4194736842105264e-05,
+      "loss": 1.738,
       "step": 1010
     },
     {
       "epoch": 0.05,
-      "grad_norm": 6.09375,
-      "learning_rate": 0.0001417894736842105,
-      "loss": 1.947,
+      "grad_norm": 20.875,
+      "learning_rate": 1.4178947368421052e-05,
+      "loss": 1.474,
       "step": 1020
     },
     {
       "epoch": 0.05,
-      "grad_norm": 5.125,
-      "learning_rate": 0.00014163157894736843,
-      "loss": 2.2714,
+      "grad_norm": 23.25,
+      "learning_rate": 1.4163157894736843e-05,
+      "loss": 1.8161,
       "step": 1030
     },
     {
       "epoch": 0.05,
-      "grad_norm": 5.125,
-      "learning_rate": 0.0001414736842105263,
-      "loss": 1.8783,
+      "grad_norm": 18.25,
+      "learning_rate": 1.4147368421052631e-05,
+      "loss": 1.2116,
       "step": 1040
     },
     {
       "epoch": 0.05,
-      "grad_norm": 6.40625,
-      "learning_rate": 0.0001413157894736842,
-      "loss": 2.409,
+      "grad_norm": 25.25,
+      "learning_rate": 1.4131578947368422e-05,
+      "loss": 1.7062,
       "step": 1050
     },
     {
       "epoch": 0.05,
-      "grad_norm": 6.96875,
-      "learning_rate": 0.0001411578947368421,
-      "loss": 2.0452,
+      "grad_norm": 21.125,
+      "learning_rate": 1.411578947368421e-05,
+      "loss": 1.5641,
       "step": 1060
     },
     {
       "epoch": 0.05,
-      "grad_norm": 4.75,
-      "learning_rate": 0.00014099999999999998,
-      "loss": 1.9155,
+      "grad_norm": 15.625,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 1.5193,
       "step": 1070
     },
     {
       "epoch": 0.05,
-      "grad_norm": 6.65625,
-      "learning_rate": 0.00014084210526315788,
-      "loss": 1.9305,
+      "grad_norm": 20.75,
+      "learning_rate": 1.408421052631579e-05,
+      "loss": 1.4775,
       "step": 1080
     },
     {
       "epoch": 0.05,
-      "grad_norm": 5.53125,
-      "learning_rate": 0.00014068421052631578,
-      "loss": 2.0101,
+      "grad_norm": 16.375,
+      "learning_rate": 1.406842105263158e-05,
+      "loss": 1.4353,
       "step": 1090
     },
     {
       "epoch": 0.06,
-      "grad_norm": 6.9375,
-      "learning_rate": 0.00014052631578947367,
-      "loss": 2.1938,
+      "grad_norm": 29.375,
+      "learning_rate": 1.405263157894737e-05,
+      "loss": 1.6741,
       "step": 1100
     },
     {
       "epoch": 0.06,
-      "grad_norm": 5.15625,
-      "learning_rate": 0.00014036842105263157,
-      "loss": 1.8036,
+      "grad_norm": 36.5,
+      "learning_rate": 1.4036842105263158e-05,
+      "loss": 1.4158,
       "step": 1110
     },
     {
       "epoch": 0.06,
-      "grad_norm": 6.65625,
-      "learning_rate": 0.00014021052631578947,
-      "loss": 1.8334,
+      "grad_norm": 20.5,
+      "learning_rate": 1.4021052631578949e-05,
+      "loss": 1.2708,
       "step": 1120
     },
     {
       "epoch": 0.06,
-      "grad_norm": 5.53125,
-      "learning_rate": 0.00014005263157894736,
-      "loss": 1.7881,
+      "grad_norm": 19.25,
+      "learning_rate": 1.4005263157894737e-05,
+      "loss": 1.1376,
       "step": 1130
     },
     {
       "epoch": 0.06,
-      "grad_norm": 7.5625,
-      "learning_rate": 0.00013989473684210523,
-      "loss": 1.9495,
+      "grad_norm": 18.5,
+      "learning_rate": 1.3989473684210526e-05,
+      "loss": 1.3171,
       "step": 1140
     },
     {
       "epoch": 0.06,
-      "grad_norm": 7.46875,
-      "learning_rate": 0.00013973684210526316,
-      "loss": 2.2284,
+      "grad_norm": 21.375,
+      "learning_rate": 1.3973684210526316e-05,
+      "loss": 1.5621,
       "step": 1150
     },
     {
       "epoch": 0.06,
-      "grad_norm": 6.28125,
-      "learning_rate": 0.00013957894736842103,
-      "loss": 2.3546,
+      "grad_norm": 19.5,
+      "learning_rate": 1.3957894736842105e-05,
+      "loss": 1.8743,
       "step": 1160
     },
     {
       "epoch": 0.06,
-      "grad_norm": 4.46875,
-      "learning_rate": 0.00013942105263157895,
-      "loss": 2.1805,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.3942105263157895e-05,
+      "loss": 1.5132,
       "step": 1170
     },
     {
       "epoch": 0.06,
-      "grad_norm": 4.78125,
-      "learning_rate": 0.00013926315789473682,
-      "loss": 2.1401,
+      "grad_norm": 22.5,
+      "learning_rate": 1.3926315789473684e-05,
+      "loss": 1.6475,
       "step": 1180
     },
     {
       "epoch": 0.06,
-      "grad_norm": 5.71875,
-      "learning_rate": 0.00013910526315789474,
-      "loss": 1.9132,
+      "grad_norm": 13.25,
+      "learning_rate": 1.3910526315789474e-05,
+      "loss": 1.4806,
       "step": 1190
     },
     {
       "epoch": 0.06,
-      "grad_norm": 3.9375,
-      "learning_rate": 0.00013894736842105261,
-      "loss": 1.9242,
+      "grad_norm": 19.0,
+      "learning_rate": 1.3894736842105263e-05,
+      "loss": 1.5761,
       "step": 1200
     },
     {
       "epoch": 0.06,
-      "grad_norm": 5.34375,
-      "learning_rate": 0.0001387894736842105,
-      "loss": 2.3808,
+      "grad_norm": 17.125,
+      "learning_rate": 1.3878947368421052e-05,
+      "loss": 1.88,
       "step": 1210
     },
     {
       "epoch": 0.06,
-      "grad_norm": 5.0,
-      "learning_rate": 0.0001386315789473684,
-      "loss": 1.8303,
+      "grad_norm": 19.0,
+      "learning_rate": 1.3863157894736842e-05,
+      "loss": 1.3399,
       "step": 1220
     },
     {
       "epoch": 0.06,
-      "grad_norm": 5.34375,
-      "learning_rate": 0.0001384736842105263,
-      "loss": 2.1689,
+      "grad_norm": 21.25,
+      "learning_rate": 1.384736842105263e-05,
+      "loss": 1.6421,
       "step": 1230
     },
     {
       "epoch": 0.06,
-      "grad_norm": 4.9375,
-      "learning_rate": 0.0001383157894736842,
-      "loss": 2.1055,
+      "grad_norm": 19.625,
+      "learning_rate": 1.3831578947368421e-05,
+      "loss": 1.7212,
       "step": 1240
     },
     {
       "epoch": 0.06,
-      "grad_norm": 4.4375,
-      "learning_rate": 0.0001381578947368421,
-      "loss": 2.1595,
+      "grad_norm": 17.75,
+      "learning_rate": 1.3815789473684211e-05,
+      "loss": 1.6537,
       "step": 1250
     },
     {
       "epoch": 0.06,
-      "grad_norm": 6.5625,
-      "learning_rate": 0.000138,
-      "loss": 2.2264,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 1.6197,
       "step": 1260
     },
     {
       "epoch": 0.06,
-      "grad_norm": 4.25,
-      "learning_rate": 0.0001378421052631579,
-      "loss": 2.0981,
+      "grad_norm": 18.875,
+      "learning_rate": 1.378421052631579e-05,
+      "loss": 1.5681,
       "step": 1270
     },
     {
       "epoch": 0.06,
-      "grad_norm": 6.90625,
-      "learning_rate": 0.00013768421052631576,
-      "loss": 2.0902,
+      "grad_norm": 40.75,
+      "learning_rate": 1.3768421052631579e-05,
+      "loss": 1.3511,
       "step": 1280
     },
     {
       "epoch": 0.06,
-      "grad_norm": 6.125,
-      "learning_rate": 0.00013752631578947368,
-      "loss": 1.9157,
+      "grad_norm": 13.0,
+      "learning_rate": 1.375263157894737e-05,
+      "loss": 1.3007,
       "step": 1290
     },
     {
       "epoch": 0.07,
-      "grad_norm": 4.5625,
-      "learning_rate": 0.00013736842105263155,
-      "loss": 1.7149,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.3736842105263158e-05,
+      "loss": 1.18,
       "step": 1300
     },
     {
       "epoch": 0.07,
-      "grad_norm": 4.6875,
-      "learning_rate": 0.00013721052631578948,
-      "loss": 2.0319,
+      "grad_norm": 19.125,
+      "learning_rate": 1.3721052631578948e-05,
+      "loss": 1.6356,
       "step": 1310
     },
     {
       "epoch": 0.07,
-      "grad_norm": 5.4375,
-      "learning_rate": 0.00013705263157894735,
-      "loss": 1.9478,
+      "grad_norm": 16.375,
+      "learning_rate": 1.3705263157894737e-05,
+      "loss": 1.4107,
       "step": 1320
     },
     {
       "epoch": 0.07,
-      "grad_norm": 6.0,
-      "learning_rate": 0.00013689473684210527,
-      "loss": 1.7866,
+      "grad_norm": 26.0,
+      "learning_rate": 1.3689473684210527e-05,
+      "loss": 1.3306,
       "step": 1330
     },
     {
       "epoch": 0.07,
-      "grad_norm": 5.1875,
-      "learning_rate": 0.00013673684210526314,
-      "loss": 2.0339,
+      "grad_norm": 22.5,
+      "learning_rate": 1.3673684210526316e-05,
+      "loss": 1.5359,
       "step": 1340
     },
     {
       "epoch": 0.07,
-      "grad_norm": 5.125,
-      "learning_rate": 0.00013657894736842104,
-      "loss": 1.7241,
+      "grad_norm": 20.0,
+      "learning_rate": 1.3657894736842106e-05,
+      "loss": 1.404,
       "step": 1350
     },
     {
       "epoch": 0.07,
-      "grad_norm": 5.25,
-      "learning_rate": 0.00013642105263157893,
-      "loss": 1.8559,
+      "grad_norm": 19.125,
+      "learning_rate": 1.3642105263157895e-05,
+      "loss": 1.2873,
       "step": 1360
     },
     {
       "epoch": 0.07,
-      "grad_norm": 5.21875,
-      "learning_rate": 0.00013626315789473683,
-      "loss": 1.847,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.3626315789473684e-05,
+      "loss": 1.3952,
       "step": 1370
     },
     {
       "epoch": 0.07,
-      "grad_norm": 7.0625,
-      "learning_rate": 0.00013610526315789473,
-      "loss": 1.8128,
+      "grad_norm": 22.0,
+      "learning_rate": 1.3610526315789474e-05,
+      "loss": 1.4608,
       "step": 1380
     },
     {
       "epoch": 0.07,
-      "grad_norm": 6.71875,
-      "learning_rate": 0.00013594736842105262,
-      "loss": 2.0117,
+      "grad_norm": 20.875,
+      "learning_rate": 1.3594736842105263e-05,
+      "loss": 1.5216,
       "step": 1390
     },
     {
       "epoch": 0.07,
-      "grad_norm": 5.1875,
-      "learning_rate": 0.00013578947368421052,
-      "loss": 2.319,
+      "grad_norm": 18.5,
+      "learning_rate": 1.3578947368421053e-05,
+      "loss": 1.4667,
       "step": 1400
     },
     {
       "epoch": 0.07,
-      "grad_norm": 10.5,
-      "learning_rate": 0.00013563157894736842,
-      "loss": 2.0572,
+      "grad_norm": 31.75,
+      "learning_rate": 1.3563157894736842e-05,
+      "loss": 1.6675,
       "step": 1410
     },
     {
       "epoch": 0.07,
-      "grad_norm": 4.6875,
-      "learning_rate": 0.0001354736842105263,
-      "loss": 1.9633,
+      "grad_norm": 22.125,
+      "learning_rate": 1.3547368421052634e-05,
+      "loss": 1.4568,
       "step": 1420
     },
     {
       "epoch": 0.07,
-      "grad_norm": 3.046875,
-      "learning_rate": 0.0001353157894736842,
-      "loss": 1.9709,
+      "grad_norm": 15.75,
+      "learning_rate": 1.3531578947368422e-05,
+      "loss": 1.4352,
       "step": 1430
     },
     {
       "epoch": 0.07,
-      "grad_norm": 5.1875,
-      "learning_rate": 0.00013515789473684208,
-      "loss": 1.9295,
+      "grad_norm": 19.75,
+      "learning_rate": 1.3515789473684211e-05,
+      "loss": 1.4362,
       "step": 1440
     },
     {
       "epoch": 0.07,
-      "grad_norm": 5.125,
-      "learning_rate": 0.000135,
-      "loss": 1.7119,
+      "grad_norm": 29.5,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 1.2622,
       "step": 1450
     },
     {
       "epoch": 0.07,
-      "grad_norm": 5.21875,
-      "learning_rate": 0.00013484210526315787,
-      "loss": 1.8842,
+      "grad_norm": 13.75,
+      "learning_rate": 1.348421052631579e-05,
+      "loss": 1.4709,
       "step": 1460
     },
     {
       "epoch": 0.07,
-      "grad_norm": 6.90625,
-      "learning_rate": 0.0001346842105263158,
-      "loss": 1.8175,
+      "grad_norm": 18.25,
+      "learning_rate": 1.346842105263158e-05,
+      "loss": 1.3638,
       "step": 1470
     },
     {
       "epoch": 0.07,
-      "grad_norm": 7.28125,
-      "learning_rate": 0.00013452631578947366,
-      "loss": 2.3633,
+      "grad_norm": 22.375,
+      "learning_rate": 1.3452631578947369e-05,
+      "loss": 1.7496,
       "step": 1480
     },
     {
       "epoch": 0.07,
-      "grad_norm": 6.875,
-      "learning_rate": 0.0001343684210526316,
-      "loss": 2.2177,
+      "grad_norm": 24.625,
+      "learning_rate": 1.343684210526316e-05,
+      "loss": 1.7741,
       "step": 1490
     },
     {
       "epoch": 0.07,
-      "grad_norm": 5.28125,
-      "learning_rate": 0.00013421052631578946,
-      "loss": 1.9366,
+      "grad_norm": 19.0,
+      "learning_rate": 1.3421052631578948e-05,
+      "loss": 1.3499,
       "step": 1500
     },
     {
       "epoch": 0.08,
-      "grad_norm": 6.09375,
-      "learning_rate": 0.00013405263157894735,
-      "loss": 1.9658,
+      "grad_norm": 31.625,
+      "learning_rate": 1.3405263157894736e-05,
+      "loss": 1.3528,
       "step": 1510
     },
     {
       "epoch": 0.08,
-      "grad_norm": 5.4375,
-      "learning_rate": 0.00013389473684210525,
-      "loss": 1.9321,
+      "grad_norm": 25.75,
+      "learning_rate": 1.3389473684210527e-05,
+      "loss": 1.4893,
       "step": 1520
     },
     {
       "epoch": 0.08,
-      "grad_norm": 4.03125,
-      "learning_rate": 0.00013373684210526315,
-      "loss": 1.8851,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.3373684210526315e-05,
+      "loss": 1.4222,
       "step": 1530
     },
     {
       "epoch": 0.08,
-      "grad_norm": 5.4375,
-      "learning_rate": 0.00013357894736842104,
-      "loss": 2.3404,
+      "grad_norm": 23.0,
+      "learning_rate": 1.3357894736842106e-05,
+      "loss": 1.7875,
       "step": 1540
     },
     {
       "epoch": 0.08,
-      "grad_norm": 6.34375,
-      "learning_rate": 0.00013342105263157894,
-      "loss": 1.9765,
+      "grad_norm": 30.375,
+      "learning_rate": 1.3342105263157894e-05,
+      "loss": 1.426,
       "step": 1550
     },
     {
       "epoch": 0.08,
-      "grad_norm": 4.40625,
-      "learning_rate": 0.00013326315789473684,
-      "loss": 2.2503,
+      "grad_norm": 14.25,
+      "learning_rate": 1.3326315789473685e-05,
+      "loss": 1.8679,
       "step": 1560
     },
     {
       "epoch": 0.08,
-      "grad_norm": 5.53125,
-      "learning_rate": 0.00013310526315789473,
-      "loss": 1.9346,
+      "grad_norm": 19.875,
+      "learning_rate": 1.3310526315789473e-05,
+      "loss": 1.4155,
       "step": 1570
     },
     {
       "epoch": 0.08,
-      "grad_norm": 6.75,
-      "learning_rate": 0.0001329473684210526,
-      "loss": 2.0895,
+      "grad_norm": 29.125,
+      "learning_rate": 1.3294736842105262e-05,
+      "loss": 1.7515,
       "step": 1580
     },
     {
       "epoch": 0.08,
-      "grad_norm": 4.4375,
-      "learning_rate": 0.00013278947368421053,
-      "loss": 1.956,
+      "grad_norm": 18.125,
+      "learning_rate": 1.3278947368421054e-05,
+      "loss": 1.4494,
       "step": 1590
     },
     {
       "epoch": 0.08,
-      "grad_norm": 6.4375,
-      "learning_rate": 0.0001326315789473684,
-      "loss": 1.8528,
+      "grad_norm": 18.75,
+      "learning_rate": 1.3263157894736843e-05,
+      "loss": 1.4159,
       "step": 1600
     },
     {
       "epoch": 0.08,
-      "grad_norm": 5.03125,
-      "learning_rate": 0.00013247368421052632,
-      "loss": 1.9809,
+      "grad_norm": 24.75,
+      "learning_rate": 1.3247368421052633e-05,
+      "loss": 1.5582,
       "step": 1610
     },
     {
       "epoch": 0.08,
-      "grad_norm": 6.03125,
-      "learning_rate": 0.0001323157894736842,
-      "loss": 1.8755,
+      "grad_norm": 24.25,
+      "learning_rate": 1.3231578947368422e-05,
+      "loss": 1.5761,
       "step": 1620
     },
     {
       "epoch": 0.08,
-      "grad_norm": 3.78125,
-      "learning_rate": 0.0001321578947368421,
-      "loss": 1.9421,
+      "grad_norm": 30.375,
+      "learning_rate": 1.3215789473684212e-05,
+      "loss": 1.5033,
       "step": 1630
     },
     {
       "epoch": 0.08,
-      "grad_norm": 5.0625,
-      "learning_rate": 0.00013199999999999998,
-      "loss": 1.8252,
+      "grad_norm": 16.75,
+      "learning_rate": 1.32e-05,
+      "loss": 1.4209,
       "step": 1640
     },
     {
       "epoch": 0.08,
-      "grad_norm": 6.1875,
-      "learning_rate": 0.00013184210526315788,
-      "loss": 2.0638,
+      "grad_norm": 22.25,
+      "learning_rate": 1.318421052631579e-05,
+      "loss": 1.5761,
       "step": 1650
     },
     {
       "epoch": 0.08,
-      "grad_norm": 5.34375,
-      "learning_rate": 0.00013168421052631578,
-      "loss": 1.9106,
+      "grad_norm": 25.0,
+      "learning_rate": 1.316842105263158e-05,
+      "loss": 1.4146,
       "step": 1660
     },
     {
       "epoch": 0.08,
-      "grad_norm": 6.40625,
-      "learning_rate": 0.00013152631578947367,
-      "loss": 1.7636,
+      "grad_norm": 20.375,
+      "learning_rate": 1.3152631578947368e-05,
+      "loss": 1.2064,
       "step": 1670
     },
     {
       "epoch": 0.08,
-      "grad_norm": 4.65625,
-      "learning_rate": 0.00013136842105263157,
-      "loss": 1.7,
+      "grad_norm": 25.75,
+      "learning_rate": 1.3136842105263159e-05,
+      "loss": 1.1254,
       "step": 1680
     },
     {
       "epoch": 0.08,
-      "grad_norm": 4.53125,
-      "learning_rate": 0.00013121052631578947,
-      "loss": 1.9783,
+      "grad_norm": 20.0,
+      "learning_rate": 1.3121052631578947e-05,
+      "loss": 1.5665,
       "step": 1690
     },
     {
       "epoch": 0.09,
-      "grad_norm": 4.75,
-      "learning_rate": 0.00013105263157894736,
-      "loss": 1.8911,
+      "grad_norm": 23.625,
+      "learning_rate": 1.3105263157894738e-05,
+      "loss": 1.5582,
       "step": 1700
     },
     {
       "epoch": 0.09,
-      "grad_norm": 5.03125,
-      "learning_rate": 0.00013089473684210526,
-      "loss": 1.6132,
+      "grad_norm": 22.875,
+      "learning_rate": 1.3089473684210526e-05,
+      "loss": 1.2198,
       "step": 1710
     },
     {
       "epoch": 0.09,
-      "grad_norm": 4.96875,
-      "learning_rate": 0.00013073684210526313,
-      "loss": 1.825,
+      "grad_norm": 15.875,
+      "learning_rate": 1.3073684210526315e-05,
+      "loss": 1.4875,
       "step": 1720
     },
     {
       "epoch": 0.09,
-      "grad_norm": 5.78125,
-      "learning_rate": 0.00013057894736842105,
-      "loss": 1.8831,
+      "grad_norm": 22.25,
+      "learning_rate": 1.3057894736842105e-05,
+      "loss": 1.3077,
       "step": 1730
     },
     {
       "epoch": 0.09,
-      "grad_norm": 3.8125,
-      "learning_rate": 0.00013042105263157892,
-      "loss": 2.0333,
+      "grad_norm": 13.125,
+      "learning_rate": 1.3042105263157894e-05,
+      "loss": 1.571,
       "step": 1740
     },
     {
       "epoch": 0.09,
-      "grad_norm": 5.4375,
-      "learning_rate": 0.00013026315789473685,
-      "loss": 1.5719,
+      "grad_norm": 16.875,
+      "learning_rate": 1.3026315789473684e-05,
+      "loss": 1.2261,
       "step": 1750
     },
     {
       "epoch": 0.09,
-      "grad_norm": 4.28125,
-      "learning_rate": 0.00013010526315789472,
-      "loss": 1.6725,
+      "grad_norm": 12.0,
+      "learning_rate": 1.3010526315789475e-05,
+      "loss": 1.1795,
       "step": 1760
     },
     {
       "epoch": 0.09,
-      "grad_norm": 6.625,
-      "learning_rate": 0.00012994736842105264,
-      "loss": 1.5617,
+      "grad_norm": 23.0,
+      "learning_rate": 1.2994736842105265e-05,
+      "loss": 1.2311,
       "step": 1770
     },
     {
       "epoch": 0.09,
-      "grad_norm": 3.296875,
-      "learning_rate": 0.0001297894736842105,
-      "loss": 1.969,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.2978947368421054e-05,
+      "loss": 1.5966,
       "step": 1780
     },
     {
       "epoch": 0.09,
-      "grad_norm": 5.46875,
-      "learning_rate": 0.0001296315789473684,
-      "loss": 1.9816,
+      "grad_norm": 18.75,
+      "learning_rate": 1.2963157894736842e-05,
+      "loss": 1.5122,
       "step": 1790
     },
     {
       "epoch": 0.09,
-      "grad_norm": 3.40625,
-      "learning_rate": 0.0001294736842105263,
-      "loss": 1.8061,
+      "grad_norm": 10.5,
+      "learning_rate": 1.2947368421052633e-05,
+      "loss": 1.2532,
       "step": 1800
     },
     {
       "epoch": 0.09,
-      "grad_norm": 2.78125,
-      "learning_rate": 0.0001293157894736842,
-      "loss": 1.796,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.2931578947368421e-05,
+      "loss": 1.3394,
       "step": 1810
     },
     {
       "epoch": 0.09,
-      "grad_norm": 4.4375,
-      "learning_rate": 0.0001291578947368421,
-      "loss": 1.9946,
+      "grad_norm": 19.375,
+      "learning_rate": 1.2915789473684212e-05,
+      "loss": 1.6067,
       "step": 1820
     },
     {
       "epoch": 0.09,
-      "grad_norm": 4.0,
-      "learning_rate": 0.000129,
-      "loss": 1.9254,
+      "grad_norm": 23.0,
+      "learning_rate": 1.29e-05,
+      "loss": 1.4977,
       "step": 1830
     },
     {
       "epoch": 0.09,
-      "grad_norm": 4.84375,
-      "learning_rate": 0.0001288421052631579,
-      "loss": 1.9615,
+      "grad_norm": 16.375,
+      "learning_rate": 1.288421052631579e-05,
+      "loss": 1.4339,
       "step": 1840
     },
     {
       "epoch": 0.09,
-      "grad_norm": 4.25,
-      "learning_rate": 0.00012868421052631578,
-      "loss": 1.9225,
+      "grad_norm": 19.625,
+      "learning_rate": 1.2868421052631579e-05,
+      "loss": 1.4789,
       "step": 1850
     },
     {
       "epoch": 0.09,
-      "grad_norm": 4.0625,
-      "learning_rate": 0.00012852631578947365,
-      "loss": 1.7678,
+      "grad_norm": 11.125,
+      "learning_rate": 1.2852631578947368e-05,
+      "loss": 1.3857,
       "step": 1860
     },
     {
       "epoch": 0.09,
-      "grad_norm": 3.921875,
-      "learning_rate": 0.00012836842105263158,
-      "loss": 2.0424,
+      "grad_norm": 9.3125,
+      "learning_rate": 1.2836842105263158e-05,
+      "loss": 1.4344,
       "step": 1870
     },
     {
       "epoch": 0.09,
-      "grad_norm": 4.125,
-      "learning_rate": 0.00012821052631578945,
-      "loss": 1.6981,
+      "grad_norm": 21.5,
+      "learning_rate": 1.2821052631578947e-05,
+      "loss": 1.416,
       "step": 1880
     },
     {
       "epoch": 0.09,
-      "grad_norm": 6.0625,
-      "learning_rate": 0.00012805263157894737,
-      "loss": 2.013,
+      "grad_norm": 23.875,
+      "learning_rate": 1.2805263157894737e-05,
+      "loss": 1.4628,
       "step": 1890
     },
     {
       "epoch": 0.1,
-      "grad_norm": 4.46875,
-      "learning_rate": 0.00012789473684210524,
-      "loss": 1.7872,
+      "grad_norm": 16.5,
+      "learning_rate": 1.2789473684210526e-05,
+      "loss": 1.3098,
       "step": 1900
     },
     {
       "epoch": 0.1,
-      "grad_norm": 4.84375,
-      "learning_rate": 0.00012773684210526316,
-      "loss": 1.7634,
+      "grad_norm": 21.375,
+      "learning_rate": 1.2773684210526316e-05,
+      "loss": 1.4423,
       "step": 1910
     },
     {
       "epoch": 0.1,
-      "grad_norm": 5.78125,
-      "learning_rate": 0.00012757894736842103,
-      "loss": 1.4218,
+      "grad_norm": 18.625,
+      "learning_rate": 1.2757894736842106e-05,
+      "loss": 1.1756,
       "step": 1920
     },
     {
       "epoch": 0.1,
-      "grad_norm": 5.125,
-      "learning_rate": 0.00012742105263157893,
-      "loss": 1.9503,
+      "grad_norm": 24.625,
+      "learning_rate": 1.2742105263157895e-05,
+      "loss": 1.3788,
       "step": 1930
     },
     {
       "epoch": 0.1,
-      "grad_norm": 4.6875,
-      "learning_rate": 0.00012726315789473683,
-      "loss": 1.6645,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.2726315789473685e-05,
+      "loss": 1.2287,
       "step": 1940
     },
     {
       "epoch": 0.1,
-      "grad_norm": 5.21875,
-      "learning_rate": 0.00012710526315789472,
-      "loss": 1.6245,
+      "grad_norm": 24.25,
+      "learning_rate": 1.2710526315789474e-05,
+      "loss": 1.1927,
       "step": 1950
     },
     {
       "epoch": 0.1,
-      "grad_norm": 4.09375,
-      "learning_rate": 0.00012694736842105262,
-      "loss": 1.934,
+      "grad_norm": 14.625,
+      "learning_rate": 1.2694736842105264e-05,
+      "loss": 1.5478,
       "step": 1960
     },
     {
       "epoch": 0.1,
-      "grad_norm": 4.28125,
-      "learning_rate": 0.00012678947368421052,
-      "loss": 1.696,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.2678947368421053e-05,
+      "loss": 1.2426,
       "step": 1970
     },
     {
       "epoch": 0.1,
-      "grad_norm": 5.59375,
-      "learning_rate": 0.0001266315789473684,
-      "loss": 1.5946,
+      "grad_norm": 30.25,
+      "learning_rate": 1.2663157894736843e-05,
+      "loss": 1.1653,
       "step": 1980
     },
     {
       "epoch": 0.1,
-      "grad_norm": 5.0625,
-      "learning_rate": 0.0001264736842105263,
-      "loss": 1.6951,
+      "grad_norm": 21.875,
+      "learning_rate": 1.2647368421052632e-05,
+      "loss": 1.3693,
       "step": 1990
     },
     {
       "epoch": 0.1,
-      "grad_norm": 3.921875,
-      "learning_rate": 0.00012631578947368418,
-      "loss": 2.1469,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 1.7071,
       "step": 2000
     },
     {
       "epoch": 0.1,
-      "eval_loss": 1.864747405052185,
-      "eval_runtime": 30.7527,
-      "eval_samples_per_second": 32.517,
-      "eval_steps_per_second": 32.517,
+      "eval_loss": 1.4587359428405762,
+      "eval_runtime": 31.2193,
+      "eval_samples_per_second": 32.032,
+      "eval_steps_per_second": 32.032,
       "step": 2000
     },
     {
       "epoch": 0.1,
-      "grad_norm": 4.3125,
-      "learning_rate": 0.0001261578947368421,
-      "loss": 1.7456,
+      "grad_norm": 12.25,
+      "learning_rate": 1.2615789473684211e-05,
+      "loss": 1.1877,
       "step": 2010
     },
     {
       "epoch": 0.1,
-      "grad_norm": 5.4375,
-      "learning_rate": 0.00012599999999999997,
-      "loss": 1.9554,
+      "grad_norm": 33.25,
+      "learning_rate": 1.26e-05,
+      "loss": 1.6576,
       "step": 2020
     },
     {
       "epoch": 0.1,
-      "grad_norm": 5.09375,
-      "learning_rate": 0.0001258421052631579,
-      "loss": 1.8303,
+      "grad_norm": 17.0,
+      "learning_rate": 1.258421052631579e-05,
+      "loss": 1.3037,
       "step": 2030
     },
     {
       "epoch": 0.1,
-      "grad_norm": 3.375,
-      "learning_rate": 0.00012568421052631577,
-      "loss": 1.7321,
+      "grad_norm": 17.625,
+      "learning_rate": 1.2568421052631579e-05,
+      "loss": 1.2634,
       "step": 2040
     },
     {
       "epoch": 0.1,
-      "grad_norm": 4.5625,
-      "learning_rate": 0.0001255263157894737,
-      "loss": 2.0467,
+      "grad_norm": 18.5,
+      "learning_rate": 1.2552631578947369e-05,
+      "loss": 1.5936,
       "step": 2050
     },
     {
       "epoch": 0.1,
-      "grad_norm": 4.625,
-      "learning_rate": 0.00012536842105263156,
-      "loss": 1.5569,
+      "grad_norm": 23.625,
+      "learning_rate": 1.2536842105263158e-05,
+      "loss": 1.1655,
       "step": 2060
     },
     {
       "epoch": 0.1,
-      "grad_norm": 4.3125,
-      "learning_rate": 0.00012521052631578946,
-      "loss": 1.6945,
+      "grad_norm": 12.625,
+      "learning_rate": 1.2521052631578946e-05,
+      "loss": 1.3446,
       "step": 2070
     },
     {
       "epoch": 0.1,
-      "grad_norm": 4.25,
-      "learning_rate": 0.00012505263157894735,
-      "loss": 1.9183,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.2505263157894737e-05,
+      "loss": 1.5295,
       "step": 2080
     },
     {
       "epoch": 0.1,
-      "grad_norm": 3.921875,
-      "learning_rate": 0.00012489473684210525,
-      "loss": 1.8488,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.2489473684210527e-05,
+      "loss": 1.5361,
       "step": 2090
     },
     {
       "epoch": 0.1,
-      "grad_norm": 4.53125,
-      "learning_rate": 0.00012473684210526315,
-      "loss": 1.7148,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.2473684210526317e-05,
+      "loss": 1.3638,
       "step": 2100
     },
     {
       "epoch": 0.11,
-      "grad_norm": 4.625,
-      "learning_rate": 0.00012457894736842104,
-      "loss": 1.9033,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.2457894736842106e-05,
+      "loss": 1.472,
       "step": 2110
     },
     {
       "epoch": 0.11,
-      "grad_norm": 4.53125,
-      "learning_rate": 0.00012442105263157894,
-      "loss": 1.4302,
+      "grad_norm": 22.0,
+      "learning_rate": 1.2442105263157896e-05,
+      "loss": 1.0827,
       "step": 2120
     },
     {
       "epoch": 0.11,
-      "grad_norm": 5.59375,
-      "learning_rate": 0.00012426315789473684,
-      "loss": 1.7065,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.2426315789473685e-05,
+      "loss": 1.2359,
       "step": 2130
     },
     {
       "epoch": 0.11,
-      "grad_norm": 4.59375,
-      "learning_rate": 0.00012410526315789473,
-      "loss": 1.9168,
+      "grad_norm": 12.75,
+      "learning_rate": 1.2410526315789474e-05,
+      "loss": 1.4345,
       "step": 2140
     },
     {
       "epoch": 0.11,
-      "grad_norm": 5.3125,
-      "learning_rate": 0.00012394736842105263,
-      "loss": 1.993,
+      "grad_norm": 19.5,
+      "learning_rate": 1.2394736842105264e-05,
+      "loss": 1.5083,
       "step": 2150
     },
     {
       "epoch": 0.11,
-      "grad_norm": 4.875,
-      "learning_rate": 0.0001237894736842105,
-      "loss": 1.7291,
+      "grad_norm": 29.125,
+      "learning_rate": 1.2378947368421053e-05,
+      "loss": 1.2703,
       "step": 2160
     },
     {
       "epoch": 0.11,
-      "grad_norm": 5.5625,
-      "learning_rate": 0.00012363157894736842,
-      "loss": 1.6801,
+      "grad_norm": 19.0,
+      "learning_rate": 1.2363157894736843e-05,
+      "loss": 1.2485,
       "step": 2170
     },
     {
       "epoch": 0.11,
-      "grad_norm": 5.9375,
-      "learning_rate": 0.0001234736842105263,
-      "loss": 1.4205,
+      "grad_norm": 30.0,
+      "learning_rate": 1.2347368421052631e-05,
+      "loss": 0.992,
       "step": 2180
     },
     {
       "epoch": 0.11,
-      "grad_norm": 5.78125,
-      "learning_rate": 0.00012331578947368421,
-      "loss": 1.8141,
+      "grad_norm": 27.625,
+      "learning_rate": 1.2331578947368422e-05,
+      "loss": 1.5069,
       "step": 2190
     },
     {
       "epoch": 0.11,
-      "grad_norm": 4.75,
-      "learning_rate": 0.00012315789473684208,
-      "loss": 1.9035,
+      "grad_norm": 23.375,
+      "learning_rate": 1.231578947368421e-05,
+      "loss": 1.5471,
       "step": 2200
     },
     {
       "epoch": 0.11,
-      "grad_norm": 4.15625,
-      "learning_rate": 0.00012299999999999998,
-      "loss": 1.846,
+      "grad_norm": 18.5,
+      "learning_rate": 1.2299999999999999e-05,
+      "loss": 1.3682,
       "step": 2210
     },
     {
       "epoch": 0.11,
-      "grad_norm": 3.921875,
-      "learning_rate": 0.00012284210526315788,
-      "loss": 1.9915,
+      "grad_norm": 17.125,
+      "learning_rate": 1.228421052631579e-05,
+      "loss": 1.5041,
       "step": 2220
     },
     {
       "epoch": 0.11,
-      "grad_norm": 4.8125,
-      "learning_rate": 0.00012268421052631577,
-      "loss": 1.9927,
+      "grad_norm": 15.875,
+      "learning_rate": 1.2268421052631578e-05,
+      "loss": 1.5487,
       "step": 2230
     },
     {
       "epoch": 0.11,
-      "grad_norm": 3.71875,
-      "learning_rate": 0.00012252631578947367,
-      "loss": 1.5576,
+      "grad_norm": 29.75,
+      "learning_rate": 1.2252631578947368e-05,
+      "loss": 1.2019,
       "step": 2240
     },
     {
       "epoch": 0.11,
-      "grad_norm": 4.09375,
-      "learning_rate": 0.00012236842105263157,
-      "loss": 1.7842,
+      "grad_norm": 12.625,
+      "learning_rate": 1.2236842105263159e-05,
+      "loss": 1.4014,
       "step": 2250
     },
     {
       "epoch": 0.11,
-      "grad_norm": 4.6875,
-      "learning_rate": 0.00012221052631578946,
-      "loss": 1.6442,
+      "grad_norm": 16.875,
+      "learning_rate": 1.2221052631578949e-05,
+      "loss": 1.3086,
       "step": 2260
     },
     {
       "epoch": 0.11,
-      "grad_norm": 4.625,
-      "learning_rate": 0.00012205263157894736,
-      "loss": 2.0708,
+      "grad_norm": 15.5,
+      "learning_rate": 1.2205263157894738e-05,
+      "loss": 1.6796,
       "step": 2270
     },
     {
       "epoch": 0.11,
-      "grad_norm": 4.3125,
-      "learning_rate": 0.00012189473684210524,
-      "loss": 1.7107,
+      "grad_norm": 18.75,
+      "learning_rate": 1.2189473684210526e-05,
+      "loss": 1.2801,
       "step": 2280
     },
     {
       "epoch": 0.11,
-      "grad_norm": 3.0,
-      "learning_rate": 0.00012173684210526315,
-      "loss": 1.6121,
+      "grad_norm": 9.75,
+      "learning_rate": 1.2173684210526317e-05,
+      "loss": 1.1822,
       "step": 2290
     },
     {
       "epoch": 0.12,
-      "grad_norm": 5.5625,
-      "learning_rate": 0.00012157894736842104,
-      "loss": 1.7322,
+      "grad_norm": 27.75,
+      "learning_rate": 1.2157894736842105e-05,
+      "loss": 1.428,
       "step": 2300
     },
     {
       "epoch": 0.12,
-      "grad_norm": 4.9375,
-      "learning_rate": 0.00012142105263157895,
-      "loss": 1.9742,
+      "grad_norm": 21.125,
+      "learning_rate": 1.2142105263157896e-05,
+      "loss": 1.4445,
       "step": 2310
     },
     {
       "epoch": 0.12,
-      "grad_norm": 5.78125,
-      "learning_rate": 0.00012126315789473683,
-      "loss": 2.2318,
+      "grad_norm": 20.5,
+      "learning_rate": 1.2126315789473684e-05,
+      "loss": 1.745,
       "step": 2320
     },
     {
       "epoch": 0.12,
-      "grad_norm": 5.1875,
-      "learning_rate": 0.00012110526315789473,
-      "loss": 2.1988,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.2110526315789475e-05,
+      "loss": 1.8124,
       "step": 2330
     },
     {
       "epoch": 0.12,
-      "grad_norm": 5.53125,
-      "learning_rate": 0.00012094736842105262,
-      "loss": 1.7186,
+      "grad_norm": 31.0,
+      "learning_rate": 1.2094736842105263e-05,
+      "loss": 1.3533,
       "step": 2340
     },
     {
       "epoch": 0.12,
-      "grad_norm": 3.375,
-      "learning_rate": 0.0001207894736842105,
-      "loss": 1.7444,
+      "grad_norm": 11.5,
+      "learning_rate": 1.2078947368421052e-05,
+      "loss": 1.4255,
       "step": 2350
     },
     {
       "epoch": 0.12,
-      "grad_norm": 4.875,
-      "learning_rate": 0.00012063157894736842,
-      "loss": 1.6583,
+      "grad_norm": 17.5,
+      "learning_rate": 1.2063157894736842e-05,
+      "loss": 1.3926,
       "step": 2360
     },
     {
       "epoch": 0.12,
-      "grad_norm": 5.0625,
-      "learning_rate": 0.0001204736842105263,
-      "loss": 1.6891,
+      "grad_norm": 13.75,
+      "learning_rate": 1.2047368421052631e-05,
+      "loss": 1.295,
       "step": 2370
     },
     {
       "epoch": 0.12,
-      "grad_norm": 4.875,
-      "learning_rate": 0.00012031578947368421,
-      "loss": 1.5208,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2031578947368421e-05,
+      "loss": 1.0575,
       "step": 2380
     },
     {
       "epoch": 0.12,
-      "grad_norm": 5.78125,
-      "learning_rate": 0.00012015789473684209,
-      "loss": 1.9399,
+      "grad_norm": 30.5,
+      "learning_rate": 1.201578947368421e-05,
+      "loss": 1.4961,
       "step": 2390
     },
     {
       "epoch": 0.12,
-      "grad_norm": 6.0,
-      "learning_rate": 0.00011999999999999999,
-      "loss": 1.9137,
+      "grad_norm": 17.5,
+      "learning_rate": 1.2e-05,
+      "loss": 1.3838,
       "step": 2400
     },
     {
       "epoch": 0.12,
-      "grad_norm": 4.71875,
-      "learning_rate": 0.00011984210526315789,
-      "loss": 1.8328,
+      "grad_norm": 16.75,
+      "learning_rate": 1.1984210526315789e-05,
+      "loss": 1.5833,
       "step": 2410
     },
     {
       "epoch": 0.12,
-      "grad_norm": 6.9375,
-      "learning_rate": 0.00011968421052631577,
-      "loss": 1.7356,
+      "grad_norm": 22.125,
+      "learning_rate": 1.196842105263158e-05,
+      "loss": 1.2771,
       "step": 2420
     },
     {
       "epoch": 0.12,
-      "grad_norm": 4.875,
-      "learning_rate": 0.00011952631578947368,
-      "loss": 2.0099,
+      "grad_norm": 19.125,
+      "learning_rate": 1.195263157894737e-05,
+      "loss": 1.6264,
       "step": 2430
     },
     {
       "epoch": 0.12,
-      "grad_norm": 4.8125,
-      "learning_rate": 0.00011936842105263156,
-      "loss": 1.8541,
+      "grad_norm": 19.5,
+      "learning_rate": 1.1936842105263158e-05,
+      "loss": 1.4304,
       "step": 2440
     },
     {
       "epoch": 0.12,
-      "grad_norm": 4.65625,
-      "learning_rate": 0.00011921052631578947,
-      "loss": 1.6947,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.1921052631578949e-05,
+      "loss": 1.1868,
       "step": 2450
     },
     {
       "epoch": 0.12,
-      "grad_norm": 5.0,
-      "learning_rate": 0.00011905263157894736,
-      "loss": 1.7433,
+      "grad_norm": 39.25,
+      "learning_rate": 1.1905263157894737e-05,
+      "loss": 1.4066,
       "step": 2460
     },
     {
       "epoch": 0.12,
-      "grad_norm": 5.71875,
-      "learning_rate": 0.00011889473684210527,
-      "loss": 1.6962,
+      "grad_norm": 24.5,
+      "learning_rate": 1.1889473684210528e-05,
+      "loss": 1.3723,
       "step": 2470
     },
     {
       "epoch": 0.12,
-      "grad_norm": 5.34375,
-      "learning_rate": 0.00011873684210526315,
-      "loss": 2.0064,
+      "grad_norm": 21.75,
+      "learning_rate": 1.1873684210526316e-05,
+      "loss": 1.507,
       "step": 2480
     },
     {
       "epoch": 0.12,
-      "grad_norm": 5.4375,
-      "learning_rate": 0.00011857894736842103,
-      "loss": 1.8146,
+      "grad_norm": 22.625,
+      "learning_rate": 1.1857894736842105e-05,
+      "loss": 1.4822,
       "step": 2490
     },
     {
       "epoch": 0.12,
-      "grad_norm": 6.25,
-      "learning_rate": 0.00011842105263157894,
-      "loss": 1.6956,
+      "grad_norm": 23.25,
+      "learning_rate": 1.1842105263157895e-05,
+      "loss": 1.3431,
       "step": 2500
     },
     {
       "epoch": 0.13,
-      "grad_norm": 4.5,
-      "learning_rate": 0.00011826315789473683,
-      "loss": 1.7244,
+      "grad_norm": 14.25,
+      "learning_rate": 1.1826315789473684e-05,
+      "loss": 1.438,
       "step": 2510
     },
     {
       "epoch": 0.13,
-      "grad_norm": 6.53125,
-      "learning_rate": 0.00011810526315789474,
-      "loss": 1.7706,
+      "grad_norm": 38.0,
+      "learning_rate": 1.1810526315789474e-05,
+      "loss": 1.3837,
       "step": 2520
     },
     {
       "epoch": 0.13,
-      "grad_norm": 6.375,
-      "learning_rate": 0.00011794736842105262,
-      "loss": 1.7227,
+      "grad_norm": 17.625,
+      "learning_rate": 1.1794736842105263e-05,
+      "loss": 1.3997,
       "step": 2530
     },
     {
       "epoch": 0.13,
-      "grad_norm": 2.640625,
-      "learning_rate": 0.00011778947368421053,
-      "loss": 1.8638,
+      "grad_norm": 13.375,
+      "learning_rate": 1.1778947368421053e-05,
+      "loss": 1.5068,
       "step": 2540
     },
     {
       "epoch": 0.13,
-      "grad_norm": 4.96875,
-      "learning_rate": 0.00011763157894736841,
-      "loss": 1.8712,
+      "grad_norm": 26.75,
+      "learning_rate": 1.1763157894736842e-05,
+      "loss": 1.5514,
       "step": 2550
     },
     {
       "epoch": 0.13,
-      "grad_norm": 4.9375,
-      "learning_rate": 0.0001174736842105263,
-      "loss": 1.8399,
+      "grad_norm": 23.375,
+      "learning_rate": 1.174736842105263e-05,
+      "loss": 1.4679,
       "step": 2560
     },
     {
       "epoch": 0.13,
-      "grad_norm": 4.9375,
-      "learning_rate": 0.0001173157894736842,
-      "loss": 1.7568,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.173157894736842e-05,
+      "loss": 1.3298,
       "step": 2570
     },
     {
       "epoch": 0.13,
-      "grad_norm": 4.1875,
-      "learning_rate": 0.00011715789473684209,
-      "loss": 1.4781,
+      "grad_norm": 16.875,
+      "learning_rate": 1.171578947368421e-05,
+      "loss": 1.2123,
       "step": 2580
     },
     {
       "epoch": 0.13,
-      "grad_norm": 5.5,
-      "learning_rate": 0.000117,
-      "loss": 1.6916,
+      "grad_norm": 21.625,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 1.3493,
       "step": 2590
     },
     {
       "epoch": 0.13,
-      "grad_norm": 6.40625,
-      "learning_rate": 0.00011684210526315788,
-      "loss": 1.8282,
+      "grad_norm": 25.75,
+      "learning_rate": 1.168421052631579e-05,
+      "loss": 1.3984,
       "step": 2600
     },
     {
       "epoch": 0.13,
-      "grad_norm": 7.46875,
-      "learning_rate": 0.00011668421052631579,
-      "loss": 1.7929,
+      "grad_norm": 20.5,
+      "learning_rate": 1.166842105263158e-05,
+      "loss": 1.435,
       "step": 2610
     },
     {
       "epoch": 0.13,
-      "grad_norm": 4.75,
-      "learning_rate": 0.00011652631578947367,
-      "loss": 1.6475,
+      "grad_norm": 23.0,
+      "learning_rate": 1.1652631578947369e-05,
+      "loss": 1.273,
       "step": 2620
     },
     {
       "epoch": 0.13,
-      "grad_norm": 3.765625,
-      "learning_rate": 0.00011636842105263156,
-      "loss": 1.6647,
+      "grad_norm": 15.75,
+      "learning_rate": 1.1636842105263158e-05,
+      "loss": 1.2547,
       "step": 2630
     },
     {
       "epoch": 0.13,
-      "grad_norm": 4.0625,
-      "learning_rate": 0.00011621052631578947,
-      "loss": 1.5357,
+      "grad_norm": 7.71875,
+      "learning_rate": 1.1621052631578948e-05,
+      "loss": 1.1227,
       "step": 2640
     },
     {
       "epoch": 0.13,
-      "grad_norm": 5.125,
-      "learning_rate": 0.00011605263157894735,
-      "loss": 1.8162,
+      "grad_norm": 29.125,
+      "learning_rate": 1.1605263157894737e-05,
+      "loss": 1.3045,
       "step": 2650
     },
     {
       "epoch": 0.13,
-      "grad_norm": 6.125,
-      "learning_rate": 0.00011589473684210526,
-      "loss": 1.7852,
+      "grad_norm": 20.25,
+      "learning_rate": 1.1589473684210527e-05,
+      "loss": 1.4858,
       "step": 2660
     },
     {
       "epoch": 0.13,
-      "grad_norm": 3.4375,
-      "learning_rate": 0.00011573684210526314,
-      "loss": 1.4552,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.1573684210526316e-05,
+      "loss": 1.1192,
       "step": 2670
     },
     {
       "epoch": 0.13,
-      "grad_norm": 5.03125,
-      "learning_rate": 0.00011557894736842105,
-      "loss": 1.7193,
+      "grad_norm": 22.875,
+      "learning_rate": 1.1557894736842106e-05,
+      "loss": 1.4049,
       "step": 2680
     },
     {
       "epoch": 0.13,
-      "grad_norm": 5.9375,
-      "learning_rate": 0.00011542105263157894,
-      "loss": 1.6664,
+      "grad_norm": 14.25,
+      "learning_rate": 1.1542105263157895e-05,
+      "loss": 1.3724,
       "step": 2690
     },
     {
       "epoch": 0.14,
-      "grad_norm": 5.0,
-      "learning_rate": 0.00011526315789473682,
-      "loss": 1.7187,
+      "grad_norm": 24.25,
+      "learning_rate": 1.1526315789473683e-05,
+      "loss": 1.3053,
       "step": 2700
     },
     {
       "epoch": 0.14,
-      "grad_norm": 4.1875,
-      "learning_rate": 0.00011510526315789473,
-      "loss": 1.7427,
+      "grad_norm": 23.0,
+      "learning_rate": 1.1510526315789474e-05,
+      "loss": 1.3986,
       "step": 2710
     },
     {
       "epoch": 0.14,
-      "grad_norm": 4.90625,
-      "learning_rate": 0.00011494736842105261,
-      "loss": 1.9004,
+      "grad_norm": 20.0,
+      "learning_rate": 1.1494736842105262e-05,
+      "loss": 1.5444,
       "step": 2720
     },
     {
       "epoch": 0.14,
-      "grad_norm": 4.15625,
-      "learning_rate": 0.00011478947368421052,
-      "loss": 2.0091,
+      "grad_norm": 16.25,
+      "learning_rate": 1.1478947368421053e-05,
+      "loss": 1.6835,
       "step": 2730
     },
     {
       "epoch": 0.14,
-      "grad_norm": 4.28125,
-      "learning_rate": 0.0001146315789473684,
-      "loss": 1.7475,
+      "grad_norm": 19.375,
+      "learning_rate": 1.1463157894736841e-05,
+      "loss": 1.2312,
       "step": 2740
     },
     {
       "epoch": 0.14,
-      "grad_norm": 5.3125,
-      "learning_rate": 0.00011447368421052632,
-      "loss": 1.4338,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.1447368421052632e-05,
+      "loss": 1.0698,
       "step": 2750
     },
     {
       "epoch": 0.14,
-      "grad_norm": 5.375,
-      "learning_rate": 0.0001143157894736842,
-      "loss": 1.7836,
+      "grad_norm": 19.875,
+      "learning_rate": 1.1431578947368422e-05,
+      "loss": 1.4871,
       "step": 2760
     },
     {
       "epoch": 0.14,
-      "grad_norm": 2.828125,
-      "learning_rate": 0.00011415789473684208,
-      "loss": 1.6579,
+      "grad_norm": 10.875,
+      "learning_rate": 1.141578947368421e-05,
+      "loss": 1.257,
       "step": 2770
     },
     {
       "epoch": 0.14,
-      "grad_norm": 4.03125,
-      "learning_rate": 0.00011399999999999999,
-      "loss": 1.9117,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 1.5711,
       "step": 2780
     },
     {
       "epoch": 0.14,
-      "grad_norm": 4.5,
-      "learning_rate": 0.00011384210526315788,
-      "loss": 1.9963,
+      "grad_norm": 22.125,
+      "learning_rate": 1.138421052631579e-05,
+      "loss": 1.592,
       "step": 2790
     },
     {
       "epoch": 0.14,
-      "grad_norm": 4.15625,
-      "learning_rate": 0.00011368421052631579,
-      "loss": 1.2637,
+      "grad_norm": 25.5,
+      "learning_rate": 1.136842105263158e-05,
+      "loss": 0.9352,
       "step": 2800
     },
     {
       "epoch": 0.14,
-      "grad_norm": 3.984375,
-      "learning_rate": 0.00011352631578947367,
-      "loss": 1.7052,
+      "grad_norm": 17.625,
+      "learning_rate": 1.1352631578947369e-05,
+      "loss": 1.3347,
       "step": 2810
     },
     {
       "epoch": 0.14,
-      "grad_norm": 5.65625,
-      "learning_rate": 0.00011336842105263158,
-      "loss": 1.3585,
+      "grad_norm": 26.75,
+      "learning_rate": 1.1336842105263159e-05,
+      "loss": 1.0488,
       "step": 2820
     },
     {
       "epoch": 0.14,
-      "grad_norm": 5.6875,
-      "learning_rate": 0.00011321052631578946,
-      "loss": 2.0309,
+      "grad_norm": 18.75,
+      "learning_rate": 1.1321052631578948e-05,
+      "loss": 1.7698,
       "step": 2830
     },
     {
       "epoch": 0.14,
-      "grad_norm": 5.3125,
-      "learning_rate": 0.00011305263157894735,
-      "loss": 1.527,
+      "grad_norm": 26.0,
+      "learning_rate": 1.1305263157894736e-05,
+      "loss": 1.2135,
       "step": 2840
     },
     {
       "epoch": 0.14,
-      "grad_norm": 5.03125,
-      "learning_rate": 0.00011289473684210526,
-      "loss": 1.7948,
+      "grad_norm": 19.125,
+      "learning_rate": 1.1289473684210527e-05,
+      "loss": 1.5619,
       "step": 2850
     },
     {
       "epoch": 0.14,
-      "grad_norm": 5.78125,
-      "learning_rate": 0.00011273684210526314,
-      "loss": 1.8825,
+      "grad_norm": 20.125,
+      "learning_rate": 1.1273684210526315e-05,
+      "loss": 1.5739,
       "step": 2860
     },
     {
       "epoch": 0.14,
-      "grad_norm": 5.15625,
-      "learning_rate": 0.00011257894736842105,
-      "loss": 1.7341,
+      "grad_norm": 51.25,
+      "learning_rate": 1.1257894736842106e-05,
+      "loss": 1.3658,
       "step": 2870
     },
     {
       "epoch": 0.14,
-      "grad_norm": 4.03125,
-      "learning_rate": 0.00011242105263157893,
-      "loss": 1.9529,
+      "grad_norm": 16.75,
+      "learning_rate": 1.1242105263157894e-05,
+      "loss": 1.6634,
       "step": 2880
     },
     {
       "epoch": 0.14,
-      "grad_norm": 8.5625,
-      "learning_rate": 0.00011226315789473684,
-      "loss": 1.6345,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1226315789473685e-05,
+      "loss": 1.1902,
       "step": 2890
     },
     {
       "epoch": 0.14,
-      "grad_norm": 6.5,
-      "learning_rate": 0.00011210526315789472,
-      "loss": 1.6036,
+      "grad_norm": 20.375,
+      "learning_rate": 1.1210526315789473e-05,
+      "loss": 1.2131,
       "step": 2900
     },
     {
       "epoch": 0.15,
-      "grad_norm": 5.125,
-      "learning_rate": 0.00011194736842105264,
-      "loss": 1.7851,
+      "grad_norm": 18.875,
+      "learning_rate": 1.1194736842105264e-05,
+      "loss": 1.4465,
       "step": 2910
     },
     {
       "epoch": 0.15,
-      "grad_norm": 4.625,
-      "learning_rate": 0.00011178947368421052,
-      "loss": 1.9272,
+      "grad_norm": 30.75,
+      "learning_rate": 1.1178947368421054e-05,
+      "loss": 1.6064,
       "step": 2920
     },
     {
       "epoch": 0.15,
-      "grad_norm": 5.03125,
-      "learning_rate": 0.0001116315789473684,
-      "loss": 1.9839,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.1163157894736842e-05,
+      "loss": 1.5864,
       "step": 2930
     },
     {
       "epoch": 0.15,
-      "grad_norm": 5.03125,
-      "learning_rate": 0.00011147368421052631,
-      "loss": 1.5799,
+      "grad_norm": 23.25,
+      "learning_rate": 1.1147368421052633e-05,
+      "loss": 1.2329,
       "step": 2940
     },
     {
       "epoch": 0.15,
-      "grad_norm": 6.34375,
-      "learning_rate": 0.0001113157894736842,
-      "loss": 1.599,
+      "grad_norm": 22.375,
+      "learning_rate": 1.1131578947368421e-05,
+      "loss": 1.3795,
       "step": 2950
     },
     {
       "epoch": 0.15,
-      "grad_norm": 7.1875,
-      "learning_rate": 0.0001111578947368421,
-      "loss": 1.8219,
+      "grad_norm": 24.0,
+      "learning_rate": 1.1115789473684212e-05,
+      "loss": 1.4095,
       "step": 2960
     },
     {
       "epoch": 0.15,
-      "grad_norm": 4.84375,
-      "learning_rate": 0.00011099999999999999,
-      "loss": 1.5486,
+      "grad_norm": 19.0,
+      "learning_rate": 1.11e-05,
+      "loss": 1.1682,
       "step": 2970
     },
     {
       "epoch": 0.15,
-      "grad_norm": 5.84375,
-      "learning_rate": 0.0001108421052631579,
-      "loss": 1.419,
+      "grad_norm": 26.0,
+      "learning_rate": 1.108421052631579e-05,
+      "loss": 1.1633,
       "step": 2980
     },
     {
       "epoch": 0.15,
-      "grad_norm": 4.03125,
-      "learning_rate": 0.00011068421052631578,
-      "loss": 1.7484,
+      "grad_norm": 18.75,
+      "learning_rate": 1.106842105263158e-05,
+      "loss": 1.4457,
       "step": 2990
     },
     {
       "epoch": 0.15,
-      "grad_norm": 6.09375,
-      "learning_rate": 0.00011052631578947366,
-      "loss": 1.7116,
+      "grad_norm": 25.75,
+      "learning_rate": 1.1052631578947368e-05,
+      "loss": 1.4336,
       "step": 3000
     },
     {
       "epoch": 0.15,
-      "grad_norm": 6.3125,
-      "learning_rate": 0.00011036842105263157,
-      "loss": 1.3924,
+      "grad_norm": 26.625,
+      "learning_rate": 1.1036842105263158e-05,
+      "loss": 1.1143,
       "step": 3010
     },
     {
       "epoch": 0.15,
-      "grad_norm": 4.125,
-      "learning_rate": 0.00011021052631578946,
-      "loss": 1.4203,
+      "grad_norm": 16.5,
+      "learning_rate": 1.1021052631578947e-05,
+      "loss": 1.2105,
       "step": 3020
     },
     {
       "epoch": 0.15,
-      "grad_norm": 5.25,
-      "learning_rate": 0.00011005263157894737,
-      "loss": 2.0427,
+      "grad_norm": 38.75,
+      "learning_rate": 1.1005263157894737e-05,
+      "loss": 1.6449,
       "step": 3030
     },
     {
       "epoch": 0.15,
-      "grad_norm": 6.09375,
-      "learning_rate": 0.00010989473684210525,
-      "loss": 1.7216,
+      "grad_norm": 23.375,
+      "learning_rate": 1.0989473684210526e-05,
+      "loss": 1.3151,
       "step": 3040
     },
     {
       "epoch": 0.15,
-      "grad_norm": 5.03125,
-      "learning_rate": 0.00010973684210526316,
-      "loss": 1.9249,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.0973684210526316e-05,
+      "loss": 1.6362,
       "step": 3050
     },
     {
       "epoch": 0.15,
-      "grad_norm": 6.15625,
-      "learning_rate": 0.00010957894736842104,
-      "loss": 1.9464,
+      "grad_norm": 24.625,
+      "learning_rate": 1.0957894736842105e-05,
+      "loss": 1.6158,
       "step": 3060
     },
     {
       "epoch": 0.15,
-      "grad_norm": 3.859375,
-      "learning_rate": 0.00010942105263157893,
-      "loss": 1.9662,
+      "grad_norm": 22.125,
+      "learning_rate": 1.0942105263157894e-05,
+      "loss": 1.6301,
       "step": 3070
     },
     {
       "epoch": 0.15,
-      "grad_norm": 4.5,
-      "learning_rate": 0.00010926315789473684,
-      "loss": 1.5929,
+      "grad_norm": 24.625,
+      "learning_rate": 1.0926315789473684e-05,
+      "loss": 1.4099,
       "step": 3080
     },
     {
       "epoch": 0.15,
-      "grad_norm": 5.28125,
-      "learning_rate": 0.00010910526315789472,
-      "loss": 1.8971,
+      "grad_norm": 28.625,
+      "learning_rate": 1.0910526315789474e-05,
+      "loss": 1.6574,
       "step": 3090
     },
     {
       "epoch": 0.15,
-      "grad_norm": 6.0625,
-      "learning_rate": 0.00010894736842105263,
-      "loss": 1.834,
+      "grad_norm": 18.75,
+      "learning_rate": 1.0894736842105265e-05,
+      "loss": 1.4458,
       "step": 3100
     },
     {
       "epoch": 0.16,
-      "grad_norm": 3.46875,
-      "learning_rate": 0.00010878947368421051,
-      "loss": 1.8009,
+      "grad_norm": 16.25,
+      "learning_rate": 1.0878947368421053e-05,
+      "loss": 1.4384,
       "step": 3110
     },
     {
       "epoch": 0.16,
-      "grad_norm": 5.09375,
-      "learning_rate": 0.00010863157894736842,
-      "loss": 1.7885,
+      "grad_norm": 17.5,
+      "learning_rate": 1.0863157894736844e-05,
+      "loss": 1.4054,
       "step": 3120
     },
     {
       "epoch": 0.16,
-      "grad_norm": 4.5,
-      "learning_rate": 0.0001084736842105263,
-      "loss": 1.3803,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.0847368421052632e-05,
+      "loss": 1.1473,
       "step": 3130
     },
     {
       "epoch": 0.16,
-      "grad_norm": 3.6875,
-      "learning_rate": 0.00010831578947368419,
-      "loss": 1.6063,
+      "grad_norm": 20.5,
+      "learning_rate": 1.0831578947368421e-05,
+      "loss": 1.2524,
       "step": 3140
     },
     {
       "epoch": 0.16,
-      "grad_norm": 4.34375,
-      "learning_rate": 0.0001081578947368421,
-      "loss": 1.6058,
+      "grad_norm": 22.375,
+      "learning_rate": 1.0815789473684211e-05,
+      "loss": 1.1765,
       "step": 3150
     },
     {
       "epoch": 0.16,
-      "grad_norm": 5.71875,
-      "learning_rate": 0.00010799999999999998,
-      "loss": 1.7769,
+      "grad_norm": 17.375,
+      "learning_rate": 1.08e-05,
+      "loss": 1.315,
       "step": 3160
     },
     {
       "epoch": 0.16,
-      "grad_norm": 6.78125,
-      "learning_rate": 0.00010784210526315789,
-      "loss": 1.8996,
+      "grad_norm": 33.75,
+      "learning_rate": 1.078421052631579e-05,
+      "loss": 1.6893,
       "step": 3170
     },
     {
       "epoch": 0.16,
-      "grad_norm": 5.375,
-      "learning_rate": 0.00010768421052631578,
-      "loss": 1.5069,
+      "grad_norm": 17.375,
+      "learning_rate": 1.0768421052631579e-05,
+      "loss": 1.3754,
       "step": 3180
     },
     {
       "epoch": 0.16,
-      "grad_norm": 4.96875,
-      "learning_rate": 0.00010752631578947369,
-      "loss": 1.4708,
+      "grad_norm": 24.125,
+      "learning_rate": 1.075263157894737e-05,
+      "loss": 1.0585,
       "step": 3190
     },
     {
       "epoch": 0.16,
-      "grad_norm": 4.59375,
-      "learning_rate": 0.00010736842105263157,
-      "loss": 1.7289,
+      "grad_norm": 26.75,
+      "learning_rate": 1.0736842105263158e-05,
+      "loss": 1.5381,
       "step": 3200
     },
     {
       "epoch": 0.16,
-      "grad_norm": 5.75,
-      "learning_rate": 0.00010721052631578945,
-      "loss": 1.509,
+      "grad_norm": 21.25,
+      "learning_rate": 1.0721052631578947e-05,
+      "loss": 1.2503,
       "step": 3210
     },
     {
       "epoch": 0.16,
-      "grad_norm": 5.375,
-      "learning_rate": 0.00010705263157894736,
-      "loss": 1.5922,
+      "grad_norm": 21.625,
+      "learning_rate": 1.0705263157894737e-05,
+      "loss": 1.4186,
       "step": 3220
     },
     {
       "epoch": 0.16,
-      "grad_norm": 5.4375,
-      "learning_rate": 0.00010689473684210525,
-      "loss": 1.8839,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.0689473684210526e-05,
+      "loss": 1.5066,
       "step": 3230
     },
     {
       "epoch": 0.16,
-      "grad_norm": 5.4375,
-      "learning_rate": 0.00010673684210526316,
-      "loss": 1.6906,
+      "grad_norm": 25.375,
+      "learning_rate": 1.0673684210526316e-05,
+      "loss": 1.2001,
       "step": 3240
     },
     {
       "epoch": 0.16,
-      "grad_norm": 3.921875,
-      "learning_rate": 0.00010657894736842104,
-      "loss": 1.6712,
+      "grad_norm": 21.5,
+      "learning_rate": 1.0657894736842106e-05,
+      "loss": 1.4222,
       "step": 3250
     },
     {
       "epoch": 0.16,
-      "grad_norm": 5.71875,
-      "learning_rate": 0.00010642105263157895,
-      "loss": 1.8901,
+      "grad_norm": 20.25,
+      "learning_rate": 1.0642105263157897e-05,
+      "loss": 1.556,
       "step": 3260
     },
     {
       "epoch": 0.16,
-      "grad_norm": 4.1875,
-      "learning_rate": 0.00010626315789473683,
-      "loss": 1.4667,
+      "grad_norm": 18.125,
+      "learning_rate": 1.0626315789473685e-05,
+      "loss": 1.1219,
       "step": 3270
     },
     {
       "epoch": 0.16,
-      "grad_norm": 6.40625,
-      "learning_rate": 0.00010610526315789471,
-      "loss": 1.8619,
+      "grad_norm": 20.125,
+      "learning_rate": 1.0610526315789474e-05,
+      "loss": 1.5552,
       "step": 3280
     },
     {
       "epoch": 0.16,
-      "grad_norm": 5.40625,
-      "learning_rate": 0.00010594736842105262,
-      "loss": 1.6954,
+      "grad_norm": 17.75,
+      "learning_rate": 1.0594736842105264e-05,
+      "loss": 1.4247,
       "step": 3290
     },
     {
       "epoch": 0.17,
-      "grad_norm": 3.703125,
-      "learning_rate": 0.00010578947368421051,
-      "loss": 1.5175,
+      "grad_norm": 17.0,
+      "learning_rate": 1.0578947368421053e-05,
+      "loss": 1.0846,
       "step": 3300
     },
     {
       "epoch": 0.17,
-      "grad_norm": 4.96875,
-      "learning_rate": 0.00010563157894736842,
-      "loss": 1.7939,
+      "grad_norm": 18.0,
+      "learning_rate": 1.0563157894736843e-05,
+      "loss": 1.4747,
       "step": 3310
     },
     {
       "epoch": 0.17,
-      "grad_norm": 4.8125,
-      "learning_rate": 0.0001054736842105263,
-      "loss": 1.6237,
+      "grad_norm": 13.5,
+      "learning_rate": 1.0547368421052632e-05,
+      "loss": 1.3696,
       "step": 3320
     },
     {
       "epoch": 0.17,
-      "grad_norm": 5.15625,
-      "learning_rate": 0.00010531578947368421,
-      "loss": 1.6149,
+      "grad_norm": 22.625,
+      "learning_rate": 1.0531578947368422e-05,
+      "loss": 1.2841,
       "step": 3330
     },
     {
       "epoch": 0.17,
-      "grad_norm": 4.5625,
-      "learning_rate": 0.0001051578947368421,
-      "loss": 1.7133,
+      "grad_norm": 19.125,
+      "learning_rate": 1.051578947368421e-05,
+      "loss": 1.4871,
       "step": 3340
     },
     {
       "epoch": 0.17,
-      "grad_norm": 4.5,
-      "learning_rate": 0.00010499999999999999,
-      "loss": 1.3957,
+      "grad_norm": 16.875,
+      "learning_rate": 1.05e-05,
+      "loss": 1.0924,
       "step": 3350
     },
     {
       "epoch": 0.17,
-      "grad_norm": 5.28125,
-      "learning_rate": 0.00010484210526315789,
-      "loss": 1.9061,
+      "grad_norm": 21.875,
+      "learning_rate": 1.048421052631579e-05,
+      "loss": 1.5268,
       "step": 3360
     },
     {
       "epoch": 0.17,
-      "grad_norm": 4.25,
-      "learning_rate": 0.00010468421052631577,
-      "loss": 1.6927,
+      "grad_norm": 23.5,
+      "learning_rate": 1.0468421052631578e-05,
+      "loss": 1.2527,
       "step": 3370
     },
     {
       "epoch": 0.17,
-      "grad_norm": 5.9375,
-      "learning_rate": 0.00010452631578947368,
-      "loss": 1.6389,
+      "grad_norm": 21.625,
+      "learning_rate": 1.0452631578947369e-05,
+      "loss": 1.0235,
       "step": 3380
     },
     {
       "epoch": 0.17,
-      "grad_norm": 4.6875,
-      "learning_rate": 0.00010436842105263156,
-      "loss": 1.6523,
+      "grad_norm": 20.5,
+      "learning_rate": 1.0436842105263157e-05,
+      "loss": 1.3699,
       "step": 3390
     },
     {
       "epoch": 0.17,
-      "grad_norm": 4.40625,
-      "learning_rate": 0.00010421052631578947,
-      "loss": 1.5598,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.0421052631578948e-05,
+      "loss": 1.2931,
       "step": 3400
     },
     {
       "epoch": 0.17,
-      "grad_norm": 4.59375,
-      "learning_rate": 0.00010405263157894736,
-      "loss": 1.2804,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.0405263157894736e-05,
+      "loss": 0.9101,
       "step": 3410
     },
     {
       "epoch": 0.17,
-      "grad_norm": 5.21875,
-      "learning_rate": 0.00010389473684210525,
-      "loss": 1.2989,
+      "grad_norm": 19.875,
+      "learning_rate": 1.0389473684210527e-05,
+      "loss": 1.0891,
       "step": 3420
     },
     {
       "epoch": 0.17,
-      "grad_norm": 6.65625,
-      "learning_rate": 0.00010373684210526315,
-      "loss": 1.4779,
+      "grad_norm": 17.5,
+      "learning_rate": 1.0373684210526317e-05,
+      "loss": 1.2518,
       "step": 3430
     },
     {
       "epoch": 0.17,
-      "grad_norm": 5.78125,
-      "learning_rate": 0.00010357894736842103,
-      "loss": 1.7402,
+      "grad_norm": 26.25,
+      "learning_rate": 1.0357894736842106e-05,
+      "loss": 1.4293,
       "step": 3440
     },
     {
       "epoch": 0.17,
-      "grad_norm": 5.09375,
-      "learning_rate": 0.00010342105263157894,
-      "loss": 1.6356,
+      "grad_norm": 13.25,
+      "learning_rate": 1.0342105263157896e-05,
+      "loss": 1.2779,
       "step": 3450
     },
     {
       "epoch": 0.17,
-      "grad_norm": 4.625,
-      "learning_rate": 0.00010326315789473683,
-      "loss": 1.9104,
+      "grad_norm": 19.75,
+      "learning_rate": 1.0326315789473685e-05,
+      "loss": 1.6036,
       "step": 3460
     },
     {
       "epoch": 0.17,
-      "grad_norm": 8.75,
-      "learning_rate": 0.00010310526315789474,
-      "loss": 1.5426,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.0310526315789475e-05,
+      "loss": 1.1933,
       "step": 3470
     },
     {
       "epoch": 0.17,
-      "grad_norm": 5.375,
-      "learning_rate": 0.00010294736842105262,
-      "loss": 1.5519,
+      "grad_norm": 24.375,
+      "learning_rate": 1.0294736842105264e-05,
+      "loss": 1.2207,
       "step": 3480
     },
     {
       "epoch": 0.17,
-      "grad_norm": 5.125,
-      "learning_rate": 0.00010278947368421052,
-      "loss": 1.3926,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.0278947368421052e-05,
+      "loss": 0.9875,
       "step": 3490
     },
     {
       "epoch": 0.17,
-      "grad_norm": 4.71875,
-      "learning_rate": 0.00010263157894736841,
-      "loss": 2.0501,
+      "grad_norm": 17.375,
+      "learning_rate": 1.0263157894736843e-05,
+      "loss": 1.6437,
       "step": 3500
     },
     {
       "epoch": 0.18,
-      "grad_norm": 5.21875,
-      "learning_rate": 0.0001024736842105263,
-      "loss": 1.8807,
+      "grad_norm": 21.125,
+      "learning_rate": 1.0247368421052631e-05,
+      "loss": 1.5689,
       "step": 3510
     },
     {
       "epoch": 0.18,
-      "grad_norm": 6.65625,
-      "learning_rate": 0.0001023157894736842,
-      "loss": 1.8071,
+      "grad_norm": 19.75,
+      "learning_rate": 1.0231578947368422e-05,
+      "loss": 1.4524,
       "step": 3520
     },
     {
       "epoch": 0.18,
-      "grad_norm": 6.0,
-      "learning_rate": 0.00010215789473684209,
-      "loss": 2.0167,
+      "grad_norm": 28.125,
+      "learning_rate": 1.021578947368421e-05,
+      "loss": 1.7354,
       "step": 3530
     },
     {
       "epoch": 0.18,
-      "grad_norm": 5.5,
-      "learning_rate": 0.000102,
-      "loss": 1.8896,
+      "grad_norm": 27.75,
+      "learning_rate": 1.02e-05,
+      "loss": 1.5075,
       "step": 3540
     },
     {
       "epoch": 0.18,
-      "grad_norm": 4.15625,
-      "learning_rate": 0.00010184210526315788,
-      "loss": 1.8863,
+      "grad_norm": 22.0,
+      "learning_rate": 1.018421052631579e-05,
+      "loss": 1.5991,
       "step": 3550
     },
     {
       "epoch": 0.18,
-      "grad_norm": 5.53125,
-      "learning_rate": 0.00010168421052631578,
-      "loss": 1.7345,
+      "grad_norm": 31.5,
+      "learning_rate": 1.0168421052631578e-05,
+      "loss": 1.3886,
       "step": 3560
     },
     {
       "epoch": 0.18,
-      "grad_norm": 5.0,
-      "learning_rate": 0.00010152631578947368,
-      "loss": 1.4917,
+      "grad_norm": 14.375,
+      "learning_rate": 1.0152631578947368e-05,
+      "loss": 1.2167,
       "step": 3570
     },
     {
       "epoch": 0.18,
-      "grad_norm": 4.8125,
-      "learning_rate": 0.00010136842105263157,
-      "loss": 1.6955,
+      "grad_norm": 15.75,
+      "learning_rate": 1.0136842105263157e-05,
+      "loss": 1.4397,
       "step": 3580
     },
     {
       "epoch": 0.18,
-      "grad_norm": 6.4375,
-      "learning_rate": 0.00010121052631578947,
-      "loss": 2.1419,
+      "grad_norm": 21.0,
+      "learning_rate": 1.0121052631578949e-05,
+      "loss": 1.8334,
       "step": 3590
     },
     {
       "epoch": 0.18,
-      "grad_norm": 4.21875,
-      "learning_rate": 0.00010105263157894735,
-      "loss": 1.7818,
+      "grad_norm": 15.75,
+      "learning_rate": 1.0105263157894738e-05,
+      "loss": 1.5306,
       "step": 3600
     },
     {
       "epoch": 0.18,
-      "grad_norm": 5.40625,
-      "learning_rate": 0.00010089473684210526,
-      "loss": 1.8861,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.0089473684210528e-05,
+      "loss": 1.5956,
       "step": 3610
     },
     {
       "epoch": 0.18,
-      "grad_norm": 5.625,
-      "learning_rate": 0.00010073684210526314,
-      "loss": 1.8296,
+      "grad_norm": 33.5,
+      "learning_rate": 1.0073684210526317e-05,
+      "loss": 1.6766,
       "step": 3620
     },
     {
       "epoch": 0.18,
-      "grad_norm": 5.5,
-      "learning_rate": 0.00010057894736842104,
-      "loss": 1.4939,
+      "grad_norm": 24.625,
+      "learning_rate": 1.0057894736842105e-05,
+      "loss": 1.2643,
       "step": 3630
     },
     {
       "epoch": 0.18,
-      "grad_norm": 3.421875,
-      "learning_rate": 0.00010042105263157894,
-      "loss": 1.6977,
+      "grad_norm": 10.75,
+      "learning_rate": 1.0042105263157896e-05,
+      "loss": 1.4199,
       "step": 3640
     },
     {
       "epoch": 0.18,
-      "grad_norm": 4.90625,
-      "learning_rate": 0.00010026315789473683,
-      "loss": 1.6357,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.0026315789473684e-05,
+      "loss": 1.3638,
       "step": 3650
     },
     {
       "epoch": 0.18,
-      "grad_norm": 5.75,
-      "learning_rate": 0.00010010526315789473,
-      "loss": 1.4076,
+      "grad_norm": 19.625,
+      "learning_rate": 1.0010526315789474e-05,
+      "loss": 1.1781,
       "step": 3660
     },
     {
       "epoch": 0.18,
-      "grad_norm": 5.4375,
-      "learning_rate": 9.994736842105261e-05,
-      "loss": 1.9151,
+      "grad_norm": 18.875,
+      "learning_rate": 9.994736842105263e-06,
+      "loss": 1.5178,
       "step": 3670
     },
     {
       "epoch": 0.18,
-      "grad_norm": 6.25,
-      "learning_rate": 9.978947368421052e-05,
-      "loss": 1.5325,
+      "grad_norm": 20.875,
+      "learning_rate": 9.978947368421053e-06,
+      "loss": 1.175,
       "step": 3680
     },
     {
       "epoch": 0.18,
-      "grad_norm": 3.453125,
-      "learning_rate": 9.963157894736841e-05,
-      "loss": 1.6342,
+      "grad_norm": 14.5625,
+      "learning_rate": 9.963157894736842e-06,
+      "loss": 1.3539,
       "step": 3690
     },
     {
       "epoch": 0.18,
-      "grad_norm": 5.9375,
-      "learning_rate": 9.94736842105263e-05,
-      "loss": 1.3291,
+      "grad_norm": 20.75,
+      "learning_rate": 9.94736842105263e-06,
+      "loss": 1.1516,
       "step": 3700
     },
     {
       "epoch": 0.19,
-      "grad_norm": 5.3125,
-      "learning_rate": 9.93157894736842e-05,
-      "loss": 1.7364,
+      "grad_norm": 21.875,
+      "learning_rate": 9.931578947368421e-06,
+      "loss": 1.3705,
       "step": 3710
     },
     {
       "epoch": 0.19,
-      "grad_norm": 4.21875,
-      "learning_rate": 9.91578947368421e-05,
-      "loss": 1.4092,
+      "grad_norm": 19.125,
+      "learning_rate": 9.91578947368421e-06,
+      "loss": 1.1967,
       "step": 3720
     },
     {
       "epoch": 0.19,
-      "grad_norm": 5.21875,
-      "learning_rate": 9.9e-05,
-      "loss": 1.515,
+      "grad_norm": 32.25,
+      "learning_rate": 9.9e-06,
+      "loss": 1.2262,
       "step": 3730
     },
     {
       "epoch": 0.19,
-      "grad_norm": 4.625,
-      "learning_rate": 9.884210526315788e-05,
-      "loss": 1.4621,
+      "grad_norm": 13.0625,
+      "learning_rate": 9.884210526315789e-06,
+      "loss": 1.0924,
       "step": 3740
     },
     {
       "epoch": 0.19,
-      "grad_norm": 6.09375,
-      "learning_rate": 9.868421052631579e-05,
-      "loss": 1.7776,
+      "grad_norm": 19.5,
+      "learning_rate": 9.868421052631579e-06,
+      "loss": 1.4826,
       "step": 3750
     },
     {
       "epoch": 0.19,
-      "grad_norm": 4.46875,
-      "learning_rate": 9.852631578947367e-05,
-      "loss": 1.8841,
+      "grad_norm": 18.5,
+      "learning_rate": 9.85263157894737e-06,
+      "loss": 1.7201,
       "step": 3760
     },
     {
       "epoch": 0.19,
-      "grad_norm": 6.53125,
-      "learning_rate": 9.836842105263157e-05,
-      "loss": 2.119,
+      "grad_norm": 22.75,
+      "learning_rate": 9.836842105263158e-06,
+      "loss": 1.8823,
       "step": 3770
     },
     {
       "epoch": 0.19,
-      "grad_norm": 4.78125,
-      "learning_rate": 9.821052631578946e-05,
-      "loss": 1.4981,
+      "grad_norm": 16.75,
+      "learning_rate": 9.821052631578948e-06,
+      "loss": 1.2102,
       "step": 3780
     },
     {
       "epoch": 0.19,
-      "grad_norm": 6.46875,
-      "learning_rate": 9.805263157894736e-05,
-      "loss": 1.3002,
+      "grad_norm": 19.0,
+      "learning_rate": 9.805263157894737e-06,
+      "loss": 0.9203,
       "step": 3790
     },
     {
       "epoch": 0.19,
-      "grad_norm": 4.9375,
-      "learning_rate": 9.789473684210526e-05,
-      "loss": 1.5721,
+      "grad_norm": 27.125,
+      "learning_rate": 9.789473684210527e-06,
+      "loss": 1.3829,
       "step": 3800
     },
     {
       "epoch": 0.19,
-      "grad_norm": 5.4375,
-      "learning_rate": 9.773684210526315e-05,
-      "loss": 1.6575,
+      "grad_norm": 16.0,
+      "learning_rate": 9.773684210526316e-06,
+      "loss": 1.4504,
       "step": 3810
     },
     {
       "epoch": 0.19,
-      "grad_norm": 5.46875,
-      "learning_rate": 9.757894736842105e-05,
-      "loss": 1.9991,
+      "grad_norm": 19.0,
+      "learning_rate": 9.757894736842106e-06,
+      "loss": 1.6228,
       "step": 3820
     },
     {
       "epoch": 0.19,
-      "grad_norm": 4.65625,
-      "learning_rate": 9.742105263157893e-05,
-      "loss": 1.4024,
+      "grad_norm": 15.0,
+      "learning_rate": 9.742105263157895e-06,
+      "loss": 1.1509,
       "step": 3830
     },
     {
       "epoch": 0.19,
-      "grad_norm": 4.0625,
-      "learning_rate": 9.726315789473683e-05,
-      "loss": 1.7038,
+      "grad_norm": 13.3125,
+      "learning_rate": 9.726315789473684e-06,
+      "loss": 1.4963,
       "step": 3840
     },
     {
       "epoch": 0.19,
-      "grad_norm": 5.9375,
-      "learning_rate": 9.710526315789473e-05,
-      "loss": 1.6067,
+      "grad_norm": 26.875,
+      "learning_rate": 9.710526315789474e-06,
+      "loss": 1.2939,
       "step": 3850
     },
     {
       "epoch": 0.19,
-      "grad_norm": 4.875,
-      "learning_rate": 9.694736842105262e-05,
-      "loss": 1.6456,
+      "grad_norm": 19.625,
+      "learning_rate": 9.694736842105263e-06,
+      "loss": 1.396,
       "step": 3860
     },
     {
       "epoch": 0.19,
-      "grad_norm": 4.34375,
-      "learning_rate": 9.678947368421052e-05,
-      "loss": 1.8511,
+      "grad_norm": 17.25,
+      "learning_rate": 9.678947368421053e-06,
+      "loss": 1.5547,
       "step": 3870
     },
     {
       "epoch": 0.19,
-      "grad_norm": 4.4375,
-      "learning_rate": 9.663157894736842e-05,
-      "loss": 1.7312,
+      "grad_norm": 23.0,
+      "learning_rate": 9.663157894736842e-06,
+      "loss": 1.3473,
       "step": 3880
     },
     {
       "epoch": 0.19,
-      "grad_norm": 4.71875,
-      "learning_rate": 9.647368421052631e-05,
-      "loss": 1.6937,
+      "grad_norm": 17.5,
+      "learning_rate": 9.647368421052632e-06,
+      "loss": 1.3847,
       "step": 3890
     },
     {
       "epoch": 0.2,
-      "grad_norm": 5.125,
-      "learning_rate": 9.63157894736842e-05,
-      "loss": 1.2875,
+      "grad_norm": 25.5,
+      "learning_rate": 9.63157894736842e-06,
+      "loss": 1.0258,
       "step": 3900
     },
     {
       "epoch": 0.2,
-      "grad_norm": 5.3125,
-      "learning_rate": 9.615789473684209e-05,
-      "loss": 1.6645,
+      "grad_norm": 20.25,
+      "learning_rate": 9.61578947368421e-06,
+      "loss": 1.4518,
       "step": 3910
     },
     {
       "epoch": 0.2,
-      "grad_norm": 4.875,
-      "learning_rate": 9.599999999999999e-05,
-      "loss": 1.8489,
+      "grad_norm": 21.5,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.5955,
       "step": 3920
     },
     {
       "epoch": 0.2,
-      "grad_norm": 4.40625,
-      "learning_rate": 9.584210526315789e-05,
-      "loss": 1.6001,
+      "grad_norm": 16.875,
+      "learning_rate": 9.58421052631579e-06,
+      "loss": 1.3555,
       "step": 3930
     },
     {
       "epoch": 0.2,
-      "grad_norm": 4.46875,
-      "learning_rate": 9.568421052631578e-05,
-      "loss": 1.546,
+      "grad_norm": 29.25,
+      "learning_rate": 9.56842105263158e-06,
+      "loss": 1.306,
       "step": 3940
     },
     {
       "epoch": 0.2,
-      "grad_norm": 4.53125,
-      "learning_rate": 9.552631578947368e-05,
-      "loss": 1.2363,
+      "grad_norm": 18.375,
+      "learning_rate": 9.552631578947369e-06,
+      "loss": 1.0536,
       "step": 3950
     },
     {
       "epoch": 0.2,
-      "grad_norm": 4.0,
-      "learning_rate": 9.536842105263158e-05,
-      "loss": 1.6655,
+      "grad_norm": 16.25,
+      "learning_rate": 9.53684210526316e-06,
+      "loss": 1.4253,
       "step": 3960
     },
     {
       "epoch": 0.2,
-      "grad_norm": 5.0,
-      "learning_rate": 9.521052631578946e-05,
-      "loss": 1.5445,
+      "grad_norm": 22.25,
+      "learning_rate": 9.521052631578948e-06,
+      "loss": 1.335,
       "step": 3970
     },
     {
       "epoch": 0.2,
-      "grad_norm": 5.625,
-      "learning_rate": 9.505263157894735e-05,
-      "loss": 1.4026,
+      "grad_norm": 27.75,
+      "learning_rate": 9.505263157894737e-06,
+      "loss": 1.2626,
       "step": 3980
     },
     {
       "epoch": 0.2,
-      "grad_norm": 5.1875,
-      "learning_rate": 9.489473684210525e-05,
-      "loss": 1.6248,
+      "grad_norm": 32.0,
+      "learning_rate": 9.489473684210527e-06,
+      "loss": 1.3404,
       "step": 3990
     },
     {
       "epoch": 0.2,
-      "grad_norm": 6.1875,
-      "learning_rate": 9.473684210526315e-05,
-      "loss": 1.9961,
+      "grad_norm": 10.25,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 1.5403,
       "step": 4000
     },
     {
       "epoch": 0.2,
-      "eval_loss": 1.6376639604568481,
-      "eval_runtime": 30.5452,
-      "eval_samples_per_second": 32.738,
-      "eval_steps_per_second": 32.738,
+      "eval_loss": 1.389373540878296,
+      "eval_runtime": 30.7104,
+      "eval_samples_per_second": 32.562,
+      "eval_steps_per_second": 32.562,
       "step": 4000
     },
     {
       "epoch": 0.2,
-      "grad_norm": 3.15625,
-      "learning_rate": 9.457894736842104e-05,
-      "loss": 1.4331,
+      "grad_norm": 23.0,
+      "learning_rate": 9.457894736842106e-06,
+      "loss": 1.2409,
       "step": 4010
     },
     {
       "epoch": 0.2,
-      "grad_norm": 3.921875,
-      "learning_rate": 9.442105263157894e-05,
-      "loss": 1.5178,
+      "grad_norm": 15.125,
+      "learning_rate": 9.442105263157894e-06,
+      "loss": 1.2763,
       "step": 4020
     },
     {
       "epoch": 0.2,
-      "grad_norm": 2.953125,
-      "learning_rate": 9.426315789473684e-05,
-      "loss": 1.7301,
+      "grad_norm": 9.625,
+      "learning_rate": 9.426315789473685e-06,
+      "loss": 1.3625,
       "step": 4030
     },
     {
       "epoch": 0.2,
-      "grad_norm": 2.75,
-      "learning_rate": 9.410526315789473e-05,
-      "loss": 1.5035,
+      "grad_norm": 10.25,
+      "learning_rate": 9.410526315789473e-06,
+      "loss": 1.2428,
       "step": 4040
     },
     {
       "epoch": 0.2,
-      "grad_norm": 4.6875,
-      "learning_rate": 9.394736842105262e-05,
-      "loss": 1.5749,
+      "grad_norm": 21.375,
+      "learning_rate": 9.394736842105262e-06,
+      "loss": 1.3786,
       "step": 4050
     },
     {
       "epoch": 0.2,
-      "grad_norm": 6.34375,
-      "learning_rate": 9.378947368421051e-05,
-      "loss": 1.7267,
+      "grad_norm": 24.5,
+      "learning_rate": 9.378947368421052e-06,
+      "loss": 1.5145,
       "step": 4060
     },
     {
       "epoch": 0.2,
-      "grad_norm": 5.25,
-      "learning_rate": 9.363157894736841e-05,
-      "loss": 1.872,
+      "grad_norm": 24.875,
+      "learning_rate": 9.363157894736841e-06,
+      "loss": 1.6081,
       "step": 4070
     },
     {
       "epoch": 0.2,
-      "grad_norm": 5.0625,
-      "learning_rate": 9.347368421052631e-05,
-      "loss": 1.991,
+      "grad_norm": 20.25,
+      "learning_rate": 9.347368421052631e-06,
+      "loss": 1.7527,
       "step": 4080
     },
     {
       "epoch": 0.2,
-      "grad_norm": 3.75,
-      "learning_rate": 9.33157894736842e-05,
-      "loss": 1.4165,
+      "grad_norm": 13.4375,
+      "learning_rate": 9.331578947368422e-06,
+      "loss": 1.2864,
       "step": 4090
     },
     {
       "epoch": 0.2,
-      "grad_norm": 4.84375,
-      "learning_rate": 9.31578947368421e-05,
-      "loss": 1.5004,
+      "grad_norm": 19.375,
+      "learning_rate": 9.315789473684212e-06,
+      "loss": 1.3088,
       "step": 4100
     },
     {
       "epoch": 0.21,
-      "grad_norm": 5.09375,
-      "learning_rate": 9.3e-05,
-      "loss": 1.4191,
+      "grad_norm": 30.375,
+      "learning_rate": 9.3e-06,
+      "loss": 1.3268,
       "step": 4110
     },
     {
       "epoch": 0.21,
-      "grad_norm": 7.0,
-      "learning_rate": 9.284210526315788e-05,
-      "loss": 1.7576,
+      "grad_norm": 16.625,
+      "learning_rate": 9.28421052631579e-06,
+      "loss": 1.4926,
       "step": 4120
     },
     {
       "epoch": 0.21,
-      "grad_norm": 4.96875,
-      "learning_rate": 9.268421052631578e-05,
-      "loss": 1.5489,
+      "grad_norm": 33.25,
+      "learning_rate": 9.26842105263158e-06,
+      "loss": 1.414,
       "step": 4130
     },
     {
       "epoch": 0.21,
-      "grad_norm": 4.6875,
-      "learning_rate": 9.252631578947367e-05,
-      "loss": 1.8809,
+      "grad_norm": 22.125,
+      "learning_rate": 9.252631578947368e-06,
+      "loss": 1.7257,
       "step": 4140
     },
     {
       "epoch": 0.21,
-      "grad_norm": 6.59375,
-      "learning_rate": 9.236842105263157e-05,
-      "loss": 1.8825,
+      "grad_norm": 25.375,
+      "learning_rate": 9.236842105263159e-06,
+      "loss": 1.5941,
       "step": 4150
     },
     {
       "epoch": 0.21,
-      "grad_norm": 4.0,
-      "learning_rate": 9.221052631578947e-05,
-      "loss": 1.5439,
+      "grad_norm": 14.375,
+      "learning_rate": 9.221052631578947e-06,
+      "loss": 1.351,
       "step": 4160
     },
     {
       "epoch": 0.21,
-      "grad_norm": 3.78125,
-      "learning_rate": 9.205263157894736e-05,
-      "loss": 1.5064,
+      "grad_norm": 26.375,
+      "learning_rate": 9.205263157894738e-06,
+      "loss": 1.2855,
       "step": 4170
     },
     {
       "epoch": 0.21,
-      "grad_norm": 4.34375,
-      "learning_rate": 9.189473684210526e-05,
-      "loss": 1.756,
+      "grad_norm": 17.0,
+      "learning_rate": 9.189473684210526e-06,
+      "loss": 1.6742,
       "step": 4180
     },
     {
       "epoch": 0.21,
-      "grad_norm": 6.0,
-      "learning_rate": 9.173684210526314e-05,
-      "loss": 1.683,
+      "grad_norm": 18.125,
+      "learning_rate": 9.173684210526315e-06,
+      "loss": 1.4892,
       "step": 4190
     },
     {
       "epoch": 0.21,
-      "grad_norm": 6.125,
-      "learning_rate": 9.157894736842104e-05,
-      "loss": 1.6729,
+      "grad_norm": 21.125,
+      "learning_rate": 9.157894736842105e-06,
+      "loss": 1.438,
       "step": 4200
     },
     {
       "epoch": 0.21,
-      "grad_norm": 7.71875,
-      "learning_rate": 9.142105263157894e-05,
-      "loss": 1.6747,
+      "grad_norm": 15.8125,
+      "learning_rate": 9.142105263157894e-06,
+      "loss": 1.5795,
       "step": 4210
     },
     {
       "epoch": 0.21,
-      "grad_norm": 6.15625,
-      "learning_rate": 9.126315789473683e-05,
-      "loss": 1.4337,
+      "grad_norm": 24.25,
+      "learning_rate": 9.126315789473684e-06,
+      "loss": 1.3318,
       "step": 4220
     },
     {
       "epoch": 0.21,
-      "grad_norm": 5.625,
-      "learning_rate": 9.110526315789473e-05,
-      "loss": 1.6269,
+      "grad_norm": 22.0,
+      "learning_rate": 9.110526315789473e-06,
+      "loss": 1.4483,
       "step": 4230
     },
     {
       "epoch": 0.21,
-      "grad_norm": 5.65625,
-      "learning_rate": 9.094736842105263e-05,
-      "loss": 1.7594,
+      "grad_norm": 21.375,
+      "learning_rate": 9.094736842105263e-06,
+      "loss": 1.547,
       "step": 4240
     },
     {
       "epoch": 0.21,
-      "grad_norm": 4.75,
-      "learning_rate": 9.078947368421052e-05,
-      "loss": 1.6758,
+      "grad_norm": 12.4375,
+      "learning_rate": 9.078947368421054e-06,
+      "loss": 1.4173,
       "step": 4250
     },
     {
       "epoch": 0.21,
-      "grad_norm": 5.875,
-      "learning_rate": 9.06315789473684e-05,
-      "loss": 1.6468,
+      "grad_norm": 21.0,
+      "learning_rate": 9.063157894736842e-06,
+      "loss": 1.3772,
       "step": 4260
     },
     {
       "epoch": 0.21,
-      "grad_norm": 5.625,
-      "learning_rate": 9.047368421052632e-05,
-      "loss": 1.9438,
+      "grad_norm": 20.75,
+      "learning_rate": 9.047368421052633e-06,
+      "loss": 1.7047,
       "step": 4270
     },
     {
       "epoch": 0.21,
-      "grad_norm": 3.890625,
-      "learning_rate": 9.03157894736842e-05,
-      "loss": 1.3541,
+      "grad_norm": 16.125,
+      "learning_rate": 9.031578947368421e-06,
+      "loss": 1.1346,
       "step": 4280
     },
     {
       "epoch": 0.21,
-      "grad_norm": 4.28125,
-      "learning_rate": 9.01578947368421e-05,
-      "loss": 1.5609,
+      "grad_norm": 13.25,
+      "learning_rate": 9.015789473684212e-06,
+      "loss": 1.2722,
       "step": 4290
     },
     {
       "epoch": 0.21,
-      "grad_norm": 7.15625,
-      "learning_rate": 8.999999999999999e-05,
-      "loss": 1.846,
+      "grad_norm": 21.25,
+      "learning_rate": 9e-06,
+      "loss": 1.611,
       "step": 4300
     },
     {
       "epoch": 0.22,
-      "grad_norm": 6.6875,
-      "learning_rate": 8.984210526315789e-05,
-      "loss": 1.6363,
+      "grad_norm": 19.125,
+      "learning_rate": 8.98421052631579e-06,
+      "loss": 1.4456,
       "step": 4310
     },
     {
       "epoch": 0.22,
-      "grad_norm": 5.875,
-      "learning_rate": 8.968421052631579e-05,
-      "loss": 1.7652,
+      "grad_norm": 15.625,
+      "learning_rate": 8.96842105263158e-06,
+      "loss": 1.5282,
       "step": 4320
     },
     {
       "epoch": 0.22,
-      "grad_norm": 5.4375,
-      "learning_rate": 8.952631578947367e-05,
-      "loss": 1.3733,
+      "grad_norm": 26.375,
+      "learning_rate": 8.952631578947368e-06,
+      "loss": 1.1709,
       "step": 4330
     },
     {
       "epoch": 0.22,
-      "grad_norm": 4.0,
-      "learning_rate": 8.936842105263158e-05,
-      "loss": 1.4821,
+      "grad_norm": 19.75,
+      "learning_rate": 8.936842105263158e-06,
+      "loss": 1.3028,
       "step": 4340
     },
     {
       "epoch": 0.22,
-      "grad_norm": 5.3125,
-      "learning_rate": 8.921052631578946e-05,
-      "loss": 1.6647,
+      "grad_norm": 13.625,
+      "learning_rate": 8.921052631578947e-06,
+      "loss": 1.3497,
       "step": 4350
     },
     {
       "epoch": 0.22,
-      "grad_norm": 5.09375,
-      "learning_rate": 8.905263157894736e-05,
-      "loss": 1.5257,
+      "grad_norm": 16.875,
+      "learning_rate": 8.905263157894737e-06,
+      "loss": 1.3032,
       "step": 4360
     },
     {
       "epoch": 0.22,
-      "grad_norm": 4.75,
-      "learning_rate": 8.889473684210525e-05,
-      "loss": 1.4503,
+      "grad_norm": 14.3125,
+      "learning_rate": 8.889473684210526e-06,
+      "loss": 1.282,
       "step": 4370
     },
     {
       "epoch": 0.22,
-      "grad_norm": 5.03125,
-      "learning_rate": 8.873684210526315e-05,
-      "loss": 1.7228,
+      "grad_norm": 22.625,
+      "learning_rate": 8.873684210526316e-06,
+      "loss": 1.404,
       "step": 4380
     },
     {
       "epoch": 0.22,
-      "grad_norm": 5.4375,
-      "learning_rate": 8.857894736842105e-05,
-      "loss": 1.4673,
+      "grad_norm": 22.125,
+      "learning_rate": 8.857894736842105e-06,
+      "loss": 1.2736,
       "step": 4390
     },
     {
       "epoch": 0.22,
-      "grad_norm": 5.96875,
-      "learning_rate": 8.842105263157893e-05,
-      "loss": 1.7307,
+      "grad_norm": 20.375,
+      "learning_rate": 8.842105263157893e-06,
+      "loss": 1.4249,
       "step": 4400
     },
     {
       "epoch": 0.22,
-      "grad_norm": 5.25,
-      "learning_rate": 8.826315789473684e-05,
-      "loss": 1.6506,
+      "grad_norm": 33.5,
+      "learning_rate": 8.826315789473684e-06,
+      "loss": 1.484,
       "step": 4410
     },
     {
       "epoch": 0.22,
-      "grad_norm": 5.125,
-      "learning_rate": 8.810526315789472e-05,
-      "loss": 1.8352,
+      "grad_norm": 27.25,
+      "learning_rate": 8.810526315789474e-06,
+      "loss": 1.5996,
       "step": 4420
     },
     {
       "epoch": 0.22,
-      "grad_norm": 4.9375,
-      "learning_rate": 8.794736842105262e-05,
-      "loss": 1.8344,
+      "grad_norm": 24.875,
+      "learning_rate": 8.794736842105264e-06,
+      "loss": 1.5918,
       "step": 4430
     },
     {
       "epoch": 0.22,
-      "grad_norm": 3.84375,
-      "learning_rate": 8.778947368421052e-05,
-      "loss": 1.5421,
+      "grad_norm": 18.25,
+      "learning_rate": 8.778947368421053e-06,
+      "loss": 1.3844,
       "step": 4440
     },
     {
       "epoch": 0.22,
-      "grad_norm": 4.875,
-      "learning_rate": 8.763157894736841e-05,
-      "loss": 1.7763,
+      "grad_norm": 20.0,
+      "learning_rate": 8.763157894736843e-06,
+      "loss": 1.3872,
       "step": 4450
     },
     {
       "epoch": 0.22,
-      "grad_norm": 4.84375,
-      "learning_rate": 8.747368421052631e-05,
-      "loss": 1.3265,
+      "grad_norm": 14.375,
+      "learning_rate": 8.747368421052632e-06,
+      "loss": 1.181,
       "step": 4460
     },
     {
       "epoch": 0.22,
-      "grad_norm": 4.5,
-      "learning_rate": 8.731578947368421e-05,
-      "loss": 1.4655,
+      "grad_norm": 26.25,
+      "learning_rate": 8.731578947368422e-06,
+      "loss": 1.3088,
       "step": 4470
     },
     {
       "epoch": 0.22,
-      "grad_norm": 4.15625,
-      "learning_rate": 8.71578947368421e-05,
-      "loss": 1.4654,
+      "grad_norm": 23.25,
+      "learning_rate": 8.715789473684211e-06,
+      "loss": 1.2746,
       "step": 4480
     },
     {
       "epoch": 0.22,
-      "grad_norm": 10.75,
-      "learning_rate": 8.699999999999999e-05,
-      "loss": 1.4957,
+      "grad_norm": 32.0,
+      "learning_rate": 8.7e-06,
+      "loss": 1.2783,
       "step": 4490
     },
     {
       "epoch": 0.23,
-      "grad_norm": 4.09375,
-      "learning_rate": 8.68421052631579e-05,
-      "loss": 1.4691,
+      "grad_norm": 7.65625,
+      "learning_rate": 8.68421052631579e-06,
+      "loss": 1.3055,
       "step": 4500
     },
     {
       "epoch": 0.23,
-      "grad_norm": 5.4375,
-      "learning_rate": 8.668421052631578e-05,
-      "loss": 1.4734,
+      "grad_norm": 16.75,
+      "learning_rate": 8.668421052631579e-06,
+      "loss": 1.2688,
       "step": 4510
     },
     {
       "epoch": 0.23,
-      "grad_norm": 5.09375,
-      "learning_rate": 8.652631578947368e-05,
-      "loss": 1.5839,
+      "grad_norm": 28.25,
+      "learning_rate": 8.652631578947369e-06,
+      "loss": 1.3703,
       "step": 4520
     },
     {
       "epoch": 0.23,
-      "grad_norm": 5.25,
-      "learning_rate": 8.636842105263157e-05,
-      "loss": 1.4329,
+      "grad_norm": 22.125,
+      "learning_rate": 8.636842105263158e-06,
+      "loss": 1.2664,
       "step": 4530
     },
     {
       "epoch": 0.23,
-      "grad_norm": 3.609375,
-      "learning_rate": 8.621052631578947e-05,
-      "loss": 1.6005,
+      "grad_norm": 18.875,
+      "learning_rate": 8.621052631578948e-06,
+      "loss": 1.3989,
       "step": 4540
     },
     {
       "epoch": 0.23,
-      "grad_norm": 6.53125,
-      "learning_rate": 8.605263157894737e-05,
-      "loss": 1.6974,
+      "grad_norm": 22.0,
+      "learning_rate": 8.605263157894737e-06,
+      "loss": 1.5021,
       "step": 4550
     },
     {
       "epoch": 0.23,
-      "grad_norm": 4.875,
-      "learning_rate": 8.589473684210525e-05,
-      "loss": 1.633,
+      "grad_norm": 18.75,
+      "learning_rate": 8.589473684210525e-06,
+      "loss": 1.4328,
       "step": 4560
     },
     {
       "epoch": 0.23,
-      "grad_norm": 4.28125,
-      "learning_rate": 8.573684210526316e-05,
-      "loss": 1.6251,
+      "grad_norm": 30.75,
+      "learning_rate": 8.573684210526316e-06,
+      "loss": 1.3705,
       "step": 4570
     },
     {
       "epoch": 0.23,
-      "grad_norm": 6.1875,
-      "learning_rate": 8.557894736842104e-05,
-      "loss": 1.3481,
+      "grad_norm": 22.625,
+      "learning_rate": 8.557894736842104e-06,
+      "loss": 1.0242,
       "step": 4580
     },
     {
       "epoch": 0.23,
-      "grad_norm": 6.96875,
-      "learning_rate": 8.542105263157894e-05,
-      "loss": 1.0885,
+      "grad_norm": 22.5,
+      "learning_rate": 8.542105263157896e-06,
+      "loss": 0.8372,
       "step": 4590
     },
     {
       "epoch": 0.23,
-      "grad_norm": 4.9375,
-      "learning_rate": 8.526315789473684e-05,
-      "loss": 1.6525,
+      "grad_norm": 17.875,
+      "learning_rate": 8.526315789473685e-06,
+      "loss": 1.4703,
       "step": 4600
     },
     {
       "epoch": 0.23,
-      "grad_norm": 5.78125,
-      "learning_rate": 8.510526315789473e-05,
-      "loss": 1.7211,
+      "grad_norm": 19.375,
+      "learning_rate": 8.510526315789475e-06,
+      "loss": 1.456,
       "step": 4610
     },
     {
       "epoch": 0.23,
-      "grad_norm": 5.4375,
-      "learning_rate": 8.494736842105263e-05,
-      "loss": 2.0992,
+      "grad_norm": 25.875,
+      "learning_rate": 8.494736842105264e-06,
+      "loss": 1.8036,
       "step": 4620
     },
     {
       "epoch": 0.23,
-      "grad_norm": 4.4375,
-      "learning_rate": 8.478947368421051e-05,
-      "loss": 1.8482,
+      "grad_norm": 17.625,
+      "learning_rate": 8.478947368421053e-06,
+      "loss": 1.7276,
       "step": 4630
     },
     {
       "epoch": 0.23,
-      "grad_norm": 4.84375,
-      "learning_rate": 8.463157894736842e-05,
-      "loss": 1.2918,
+      "grad_norm": 29.125,
+      "learning_rate": 8.463157894736843e-06,
+      "loss": 1.1681,
       "step": 4640
     },
     {
       "epoch": 0.23,
-      "grad_norm": 4.9375,
-      "learning_rate": 8.44736842105263e-05,
-      "loss": 1.7271,
+      "grad_norm": 18.875,
+      "learning_rate": 8.447368421052632e-06,
+      "loss": 1.5285,
       "step": 4650
     },
     {
       "epoch": 0.23,
-      "grad_norm": 3.53125,
-      "learning_rate": 8.43157894736842e-05,
-      "loss": 1.6719,
+      "grad_norm": 30.625,
+      "learning_rate": 8.431578947368422e-06,
+      "loss": 1.531,
       "step": 4660
     },
     {
       "epoch": 0.23,
-      "grad_norm": 4.28125,
-      "learning_rate": 8.41578947368421e-05,
-      "loss": 1.305,
+      "grad_norm": 19.375,
+      "learning_rate": 8.41578947368421e-06,
+      "loss": 1.1936,
       "step": 4670
     },
     {
       "epoch": 0.23,
-      "grad_norm": 6.4375,
-      "learning_rate": 8.4e-05,
-      "loss": 1.7411,
+      "grad_norm": 18.75,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 1.5973,
       "step": 4680
     },
     {
       "epoch": 0.23,
-      "grad_norm": 5.0,
-      "learning_rate": 8.384210526315789e-05,
-      "loss": 1.5826,
+      "grad_norm": 14.4375,
+      "learning_rate": 8.38421052631579e-06,
+      "loss": 1.5106,
       "step": 4690
     },
     {
       "epoch": 0.23,
-      "grad_norm": 4.96875,
-      "learning_rate": 8.368421052631578e-05,
-      "loss": 1.687,
+      "grad_norm": 39.0,
+      "learning_rate": 8.368421052631578e-06,
+      "loss": 1.5852,
       "step": 4700
     },
     {
       "epoch": 0.24,
-      "grad_norm": 4.59375,
-      "learning_rate": 8.352631578947369e-05,
-      "loss": 1.4675,
+      "grad_norm": 18.75,
+      "learning_rate": 8.352631578947369e-06,
+      "loss": 1.2943,
       "step": 4710
     },
     {
       "epoch": 0.24,
-      "grad_norm": 5.28125,
-      "learning_rate": 8.336842105263157e-05,
-      "loss": 1.6521,
+      "grad_norm": 19.125,
+      "learning_rate": 8.336842105263157e-06,
+      "loss": 1.3911,
       "step": 4720
     },
     {
       "epoch": 0.24,
-      "grad_norm": 5.21875,
-      "learning_rate": 8.321052631578948e-05,
-      "loss": 1.7566,
+      "grad_norm": 26.25,
+      "learning_rate": 8.321052631578947e-06,
+      "loss": 1.5374,
       "step": 4730
     },
     {
       "epoch": 0.24,
-      "grad_norm": 5.9375,
-      "learning_rate": 8.305263157894736e-05,
-      "loss": 1.6841,
+      "grad_norm": 33.75,
+      "learning_rate": 8.305263157894736e-06,
+      "loss": 1.5726,
       "step": 4740
     },
     {
       "epoch": 0.24,
-      "grad_norm": 3.578125,
-      "learning_rate": 8.289473684210526e-05,
-      "loss": 1.6672,
+      "grad_norm": 23.375,
+      "learning_rate": 8.289473684210526e-06,
+      "loss": 1.3592,
       "step": 4750
     },
     {
       "epoch": 0.24,
-      "grad_norm": 4.53125,
-      "learning_rate": 8.273684210526315e-05,
-      "loss": 1.5411,
+      "grad_norm": 20.25,
+      "learning_rate": 8.273684210526317e-06,
+      "loss": 1.378,
       "step": 4760
     },
     {
       "epoch": 0.24,
-      "grad_norm": 7.4375,
-      "learning_rate": 8.257894736842104e-05,
-      "loss": 1.721,
+      "grad_norm": 27.375,
+      "learning_rate": 8.257894736842105e-06,
+      "loss": 1.5718,
       "step": 4770
     },
     {
       "epoch": 0.24,
-      "grad_norm": 3.953125,
-      "learning_rate": 8.242105263157895e-05,
-      "loss": 1.7061,
+      "grad_norm": 25.25,
+      "learning_rate": 8.242105263157896e-06,
+      "loss": 1.5038,
       "step": 4780
     },
     {
       "epoch": 0.24,
-      "grad_norm": 3.953125,
-      "learning_rate": 8.226315789473683e-05,
-      "loss": 1.5112,
+      "grad_norm": 13.1875,
+      "learning_rate": 8.226315789473684e-06,
+      "loss": 1.4199,
       "step": 4790
     },
     {
       "epoch": 0.24,
-      "grad_norm": 4.75,
-      "learning_rate": 8.210526315789474e-05,
-      "loss": 1.3537,
+      "grad_norm": 19.25,
+      "learning_rate": 8.210526315789475e-06,
+      "loss": 1.1204,
       "step": 4800
     },
     {
       "epoch": 0.24,
-      "grad_norm": 5.125,
-      "learning_rate": 8.194736842105262e-05,
-      "loss": 1.3723,
+      "grad_norm": 14.5625,
+      "learning_rate": 8.194736842105263e-06,
+      "loss": 1.3619,
       "step": 4810
     },
     {
       "epoch": 0.24,
-      "grad_norm": 4.8125,
-      "learning_rate": 8.178947368421052e-05,
-      "loss": 1.9272,
+      "grad_norm": 21.375,
+      "learning_rate": 8.178947368421054e-06,
+      "loss": 1.9199,
       "step": 4820
     },
     {
       "epoch": 0.24,
-      "grad_norm": 13.3125,
-      "learning_rate": 8.163157894736842e-05,
-      "loss": 1.3583,
+      "grad_norm": 21.0,
+      "learning_rate": 8.163157894736842e-06,
+      "loss": 1.2522,
       "step": 4830
     },
     {
       "epoch": 0.24,
-      "grad_norm": 10.25,
-      "learning_rate": 8.14736842105263e-05,
-      "loss": 1.7233,
+      "grad_norm": 21.75,
+      "learning_rate": 8.147368421052631e-06,
+      "loss": 1.4897,
       "step": 4840
     },
     {
       "epoch": 0.24,
-      "grad_norm": 5.3125,
-      "learning_rate": 8.131578947368421e-05,
-      "loss": 1.6048,
+      "grad_norm": 22.25,
+      "learning_rate": 8.131578947368421e-06,
+      "loss": 1.4521,
       "step": 4850
     },
     {
       "epoch": 0.24,
-      "grad_norm": 5.4375,
-      "learning_rate": 8.11578947368421e-05,
-      "loss": 1.7952,
+      "grad_norm": 22.875,
+      "learning_rate": 8.11578947368421e-06,
+      "loss": 1.554,
       "step": 4860
     },
     {
       "epoch": 0.24,
-      "grad_norm": 4.6875,
-      "learning_rate": 8.1e-05,
-      "loss": 1.6838,
+      "grad_norm": 16.5,
+      "learning_rate": 8.1e-06,
+      "loss": 1.562,
       "step": 4870
     },
     {
       "epoch": 0.24,
-      "grad_norm": 3.546875,
-      "learning_rate": 8.084210526315789e-05,
-      "loss": 1.9311,
+      "grad_norm": 16.375,
+      "learning_rate": 8.084210526315789e-06,
+      "loss": 1.6152,
       "step": 4880
     },
     {
       "epoch": 0.24,
-      "grad_norm": 5.28125,
-      "learning_rate": 8.068421052631578e-05,
-      "loss": 1.5036,
+      "grad_norm": 21.0,
+      "learning_rate": 8.06842105263158e-06,
+      "loss": 1.3458,
       "step": 4890
     },
     {
       "epoch": 0.24,
-      "grad_norm": 4.90625,
-      "learning_rate": 8.052631578947368e-05,
-      "loss": 1.55,
+      "grad_norm": 24.375,
+      "learning_rate": 8.052631578947368e-06,
+      "loss": 1.2805,
       "step": 4900
     },
     {
       "epoch": 0.25,
-      "grad_norm": 5.21875,
-      "learning_rate": 8.036842105263156e-05,
-      "loss": 1.5131,
+      "grad_norm": 18.375,
+      "learning_rate": 8.036842105263157e-06,
+      "loss": 1.3182,
       "step": 4910
     },
     {
       "epoch": 0.25,
-      "grad_norm": 5.21875,
-      "learning_rate": 8.021052631578947e-05,
-      "loss": 1.5671,
+      "grad_norm": 33.5,
+      "learning_rate": 8.021052631578949e-06,
+      "loss": 1.2983,
       "step": 4920
     },
     {
       "epoch": 0.25,
-      "grad_norm": 4.71875,
-      "learning_rate": 8.005263157894736e-05,
-      "loss": 1.6518,
+      "grad_norm": 23.875,
+      "learning_rate": 8.005263157894737e-06,
+      "loss": 1.4759,
       "step": 4930
     },
     {
       "epoch": 0.25,
-      "grad_norm": 4.46875,
-      "learning_rate": 7.989473684210527e-05,
-      "loss": 1.7936,
+      "grad_norm": 18.5,
+      "learning_rate": 7.989473684210528e-06,
+      "loss": 1.5965,
       "step": 4940
     },
     {
       "epoch": 0.25,
-      "grad_norm": 9.4375,
-      "learning_rate": 7.973684210526315e-05,
-      "loss": 1.5896,
+      "grad_norm": 20.25,
+      "learning_rate": 7.973684210526316e-06,
+      "loss": 1.349,
       "step": 4950
     },
     {
       "epoch": 0.25,
-      "grad_norm": 4.40625,
-      "learning_rate": 7.957894736842106e-05,
-      "loss": 1.6038,
+      "grad_norm": 15.5,
+      "learning_rate": 7.957894736842107e-06,
+      "loss": 1.4254,
       "step": 4960
     },
     {
       "epoch": 0.25,
-      "grad_norm": 5.28125,
-      "learning_rate": 7.942105263157894e-05,
-      "loss": 1.4422,
+      "grad_norm": 12.0,
+      "learning_rate": 7.942105263157895e-06,
+      "loss": 1.2403,
       "step": 4970
     },
     {
       "epoch": 0.25,
-      "grad_norm": 3.625,
-      "learning_rate": 7.926315789473683e-05,
-      "loss": 1.4781,
+      "grad_norm": 17.375,
+      "learning_rate": 7.926315789473684e-06,
+      "loss": 1.3203,
       "step": 4980
     },
     {
       "epoch": 0.25,
-      "grad_norm": 4.90625,
-      "learning_rate": 7.910526315789474e-05,
-      "loss": 1.554,
+      "grad_norm": 17.0,
+      "learning_rate": 7.910526315789474e-06,
+      "loss": 1.2978,
       "step": 4990
     },
     {
       "epoch": 0.25,
-      "grad_norm": 4.25,
-      "learning_rate": 7.894736842105262e-05,
-      "loss": 1.4183,
+      "grad_norm": 21.375,
+      "learning_rate": 7.894736842105263e-06,
+      "loss": 1.2262,
       "step": 5000
     },
     {
       "epoch": 0.25,
-      "grad_norm": 5.3125,
-      "learning_rate": 7.878947368421053e-05,
-      "loss": 1.5773,
+      "grad_norm": 15.875,
+      "learning_rate": 7.878947368421053e-06,
+      "loss": 1.4521,
       "step": 5010
     },
     {
       "epoch": 0.25,
-      "grad_norm": 8.9375,
-      "learning_rate": 7.863157894736841e-05,
-      "loss": 1.7739,
+      "grad_norm": 13.4375,
+      "learning_rate": 7.863157894736842e-06,
+      "loss": 1.713,
       "step": 5020
     },
     {
       "epoch": 0.25,
-      "grad_norm": 4.6875,
-      "learning_rate": 7.847368421052632e-05,
-      "loss": 1.6497,
+      "grad_norm": 22.0,
+      "learning_rate": 7.847368421052632e-06,
+      "loss": 1.4544,
       "step": 5030
     },
     {
       "epoch": 0.25,
-      "grad_norm": 3.296875,
-      "learning_rate": 7.83157894736842e-05,
-      "loss": 1.6733,
+      "grad_norm": 65.5,
+      "learning_rate": 7.831578947368421e-06,
+      "loss": 1.5237,
       "step": 5040
     },
     {
       "epoch": 0.25,
-      "grad_norm": 5.09375,
-      "learning_rate": 7.815789473684209e-05,
-      "loss": 1.509,
+      "grad_norm": 16.625,
+      "learning_rate": 7.81578947368421e-06,
+      "loss": 1.3374,
       "step": 5050
     },
     {
       "epoch": 0.25,
-      "grad_norm": 4.9375,
-      "learning_rate": 7.8e-05,
-      "loss": 1.7451,
+      "grad_norm": 15.625,
+      "learning_rate": 7.8e-06,
+      "loss": 1.5768,
       "step": 5060
     },
     {
       "epoch": 0.25,
-      "grad_norm": 3.984375,
-      "learning_rate": 7.784210526315788e-05,
-      "loss": 1.1635,
+      "grad_norm": 11.0625,
+      "learning_rate": 7.784210526315789e-06,
+      "loss": 1.0055,
       "step": 5070
     },
     {
       "epoch": 0.25,
-      "grad_norm": 5.71875,
-      "learning_rate": 7.768421052631579e-05,
-      "loss": 1.3425,
+      "grad_norm": 19.125,
+      "learning_rate": 7.768421052631579e-06,
+      "loss": 1.2459,
       "step": 5080
     },
     {
       "epoch": 0.25,
-      "grad_norm": 5.71875,
-      "learning_rate": 7.752631578947367e-05,
-      "loss": 1.6612,
+      "grad_norm": 20.625,
+      "learning_rate": 7.75263157894737e-06,
+      "loss": 1.4636,
       "step": 5090
     },
     {
       "epoch": 0.26,
-      "grad_norm": 6.4375,
-      "learning_rate": 7.736842105263159e-05,
-      "loss": 1.5136,
+      "grad_norm": 21.0,
+      "learning_rate": 7.73684210526316e-06,
+      "loss": 1.3925,
       "step": 5100
     },
     {
       "epoch": 0.26,
-      "grad_norm": 4.375,
-      "learning_rate": 7.721052631578947e-05,
-      "loss": 1.4066,
+      "grad_norm": 17.125,
+      "learning_rate": 7.721052631578948e-06,
+      "loss": 1.1989,
       "step": 5110
     },
     {
       "epoch": 0.26,
-      "grad_norm": 5.15625,
-      "learning_rate": 7.705263157894735e-05,
-      "loss": 1.5721,
+      "grad_norm": 27.25,
+      "learning_rate": 7.705263157894737e-06,
+      "loss": 1.4218,
       "step": 5120
     },
     {
       "epoch": 0.26,
-      "grad_norm": 5.75,
-      "learning_rate": 7.689473684210526e-05,
-      "loss": 1.7248,
+      "grad_norm": 38.25,
+      "learning_rate": 7.689473684210527e-06,
+      "loss": 1.5287,
       "step": 5130
     },
     {
       "epoch": 0.26,
-      "grad_norm": 4.40625,
-      "learning_rate": 7.673684210526314e-05,
-      "loss": 1.5728,
+      "grad_norm": 15.3125,
+      "learning_rate": 7.673684210526316e-06,
+      "loss": 1.3331,
       "step": 5140
     },
     {
       "epoch": 0.26,
-      "grad_norm": 4.40625,
-      "learning_rate": 7.657894736842105e-05,
-      "loss": 1.6477,
+      "grad_norm": 10.0,
+      "learning_rate": 7.657894736842106e-06,
+      "loss": 1.562,
       "step": 5150
     },
     {
       "epoch": 0.26,
-      "grad_norm": 4.1875,
-      "learning_rate": 7.642105263157894e-05,
-      "loss": 1.7276,
+      "grad_norm": 24.0,
+      "learning_rate": 7.642105263157895e-06,
+      "loss": 1.5789,
       "step": 5160
     },
     {
       "epoch": 0.26,
-      "grad_norm": 6.1875,
-      "learning_rate": 7.626315789473685e-05,
-      "loss": 1.5607,
+      "grad_norm": 29.375,
+      "learning_rate": 7.626315789473685e-06,
+      "loss": 1.2715,
       "step": 5170
     },
     {
       "epoch": 0.26,
-      "grad_norm": 4.8125,
-      "learning_rate": 7.610526315789473e-05,
-      "loss": 1.7054,
+      "grad_norm": 19.0,
+      "learning_rate": 7.610526315789474e-06,
+      "loss": 1.5294,
       "step": 5180
     },
     {
       "epoch": 0.26,
-      "grad_norm": 4.5625,
-      "learning_rate": 7.594736842105261e-05,
-      "loss": 1.5041,
+      "grad_norm": 23.0,
+      "learning_rate": 7.594736842105262e-06,
+      "loss": 1.2915,
       "step": 5190
     },
     {
       "epoch": 0.26,
-      "grad_norm": 4.28125,
-      "learning_rate": 7.578947368421052e-05,
-      "loss": 1.3939,
+      "grad_norm": 19.125,
+      "learning_rate": 7.578947368421053e-06,
+      "loss": 1.2321,
       "step": 5200
     },
     {
       "epoch": 0.26,
-      "grad_norm": 5.28125,
-      "learning_rate": 7.563157894736841e-05,
-      "loss": 1.3901,
+      "grad_norm": 23.75,
+      "learning_rate": 7.563157894736842e-06,
+      "loss": 1.3039,
       "step": 5210
     },
     {
       "epoch": 0.26,
-      "grad_norm": 5.84375,
-      "learning_rate": 7.547368421052632e-05,
-      "loss": 1.3635,
+      "grad_norm": 29.375,
+      "learning_rate": 7.5473684210526326e-06,
+      "loss": 1.2118,
       "step": 5220
     },
     {
       "epoch": 0.26,
-      "grad_norm": 5.5,
-      "learning_rate": 7.53157894736842e-05,
-      "loss": 1.6156,
+      "grad_norm": 26.125,
+      "learning_rate": 7.531578947368421e-06,
+      "loss": 1.3432,
       "step": 5230
     },
     {
       "epoch": 0.26,
-      "grad_norm": 6.5625,
-      "learning_rate": 7.515789473684211e-05,
-      "loss": 1.4224,
+      "grad_norm": 26.25,
+      "learning_rate": 7.5157894736842115e-06,
+      "loss": 1.2973,
       "step": 5240
     },
     {
       "epoch": 0.26,
-      "grad_norm": 5.6875,
-      "learning_rate": 7.5e-05,
-      "loss": 1.5507,
+      "grad_norm": 17.5,
+      "learning_rate": 7.5e-06,
+      "loss": 1.3631,
       "step": 5250
     },
     {
       "epoch": 0.26,
-      "grad_norm": 6.5,
-      "learning_rate": 7.484210526315789e-05,
-      "loss": 1.5214,
+      "grad_norm": 19.25,
+      "learning_rate": 7.48421052631579e-06,
+      "loss": 1.5015,
       "step": 5260
     },
     {
       "epoch": 0.26,
-      "grad_norm": 4.21875,
-      "learning_rate": 7.468421052631579e-05,
-      "loss": 1.4005,
+      "grad_norm": 15.0,
+      "learning_rate": 7.468421052631579e-06,
+      "loss": 1.0094,
       "step": 5270
     },
     {
       "epoch": 0.26,
-      "grad_norm": 9.8125,
-      "learning_rate": 7.452631578947368e-05,
-      "loss": 1.649,
+      "grad_norm": 18.5,
+      "learning_rate": 7.452631578947369e-06,
+      "loss": 1.4658,
       "step": 5280
     },
     {
       "epoch": 0.26,
-      "grad_norm": 6.25,
-      "learning_rate": 7.436842105263157e-05,
-      "loss": 1.3222,
+      "grad_norm": 25.5,
+      "learning_rate": 7.436842105263157e-06,
+      "loss": 1.2255,
       "step": 5290
     },
     {
       "epoch": 0.27,
-      "grad_norm": 4.875,
-      "learning_rate": 7.421052631578946e-05,
-      "loss": 1.4316,
+      "grad_norm": 22.125,
+      "learning_rate": 7.421052631578948e-06,
+      "loss": 1.1919,
       "step": 5300
     },
     {
       "epoch": 0.27,
-      "grad_norm": 4.59375,
-      "learning_rate": 7.405263157894736e-05,
-      "loss": 1.4701,
+      "grad_norm": 19.75,
+      "learning_rate": 7.405263157894737e-06,
+      "loss": 1.2764,
       "step": 5310
     },
     {
       "epoch": 0.27,
-      "grad_norm": 5.96875,
-      "learning_rate": 7.389473684210526e-05,
-      "loss": 1.8238,
+      "grad_norm": 20.875,
+      "learning_rate": 7.389473684210527e-06,
+      "loss": 1.5922,
       "step": 5320
     },
     {
       "epoch": 0.27,
-      "grad_norm": 4.375,
-      "learning_rate": 7.373684210526315e-05,
-      "loss": 1.4409,
+      "grad_norm": 15.375,
+      "learning_rate": 7.373684210526316e-06,
+      "loss": 1.3,
       "step": 5330
     },
     {
       "epoch": 0.27,
-      "grad_norm": 5.5,
-      "learning_rate": 7.357894736842105e-05,
-      "loss": 1.4107,
+      "grad_norm": 23.75,
+      "learning_rate": 7.357894736842106e-06,
+      "loss": 1.3065,
       "step": 5340
     },
     {
       "epoch": 0.27,
-      "grad_norm": 6.0,
-      "learning_rate": 7.342105263157895e-05,
-      "loss": 1.5908,
+      "grad_norm": 14.25,
+      "learning_rate": 7.342105263157895e-06,
+      "loss": 1.35,
       "step": 5350
     },
     {
       "epoch": 0.27,
-      "grad_norm": 4.46875,
-      "learning_rate": 7.326315789473683e-05,
-      "loss": 1.1515,
+      "grad_norm": 15.8125,
+      "learning_rate": 7.326315789473684e-06,
+      "loss": 0.9403,
       "step": 5360
     },
     {
       "epoch": 0.27,
-      "grad_norm": 5.625,
-      "learning_rate": 7.310526315789473e-05,
-      "loss": 1.8387,
+      "grad_norm": 44.75,
+      "learning_rate": 7.310526315789473e-06,
+      "loss": 1.5757,
       "step": 5370
     },
     {
       "epoch": 0.27,
-      "grad_norm": 7.34375,
-      "learning_rate": 7.294736842105262e-05,
-      "loss": 1.7581,
+      "grad_norm": 40.25,
+      "learning_rate": 7.2947368421052636e-06,
+      "loss": 1.7024,
       "step": 5380
     },
     {
       "epoch": 0.27,
-      "grad_norm": 6.40625,
-      "learning_rate": 7.278947368421052e-05,
-      "loss": 1.8786,
+      "grad_norm": 20.625,
+      "learning_rate": 7.278947368421053e-06,
+      "loss": 1.7692,
       "step": 5390
     },
     {
       "epoch": 0.27,
-      "grad_norm": 6.09375,
-      "learning_rate": 7.263157894736842e-05,
-      "loss": 1.5486,
+      "grad_norm": 25.5,
+      "learning_rate": 7.2631578947368426e-06,
+      "loss": 1.4494,
       "step": 5400
     },
     {
       "epoch": 0.27,
-      "grad_norm": 6.375,
-      "learning_rate": 7.247368421052631e-05,
-      "loss": 1.6973,
+      "grad_norm": 22.75,
+      "learning_rate": 7.247368421052632e-06,
+      "loss": 1.3622,
       "step": 5410
     },
     {
       "epoch": 0.27,
-      "grad_norm": 2.53125,
-      "learning_rate": 7.231578947368421e-05,
-      "loss": 1.1965,
+      "grad_norm": 8.75,
+      "learning_rate": 7.2315789473684215e-06,
+      "loss": 1.1007,
       "step": 5420
     },
     {
       "epoch": 0.27,
-      "grad_norm": 6.375,
-      "learning_rate": 7.215789473684209e-05,
-      "loss": 1.4671,
+      "grad_norm": 16.375,
+      "learning_rate": 7.21578947368421e-06,
+      "loss": 1.2776,
       "step": 5430
     },
     {
       "epoch": 0.27,
-      "grad_norm": 6.03125,
-      "learning_rate": 7.199999999999999e-05,
-      "loss": 1.4568,
+      "grad_norm": 23.125,
+      "learning_rate": 7.2e-06,
+      "loss": 1.2975,
       "step": 5440
     },
     {
       "epoch": 0.27,
-      "grad_norm": 6.0625,
-      "learning_rate": 7.184210526315788e-05,
-      "loss": 1.4513,
+      "grad_norm": 23.0,
+      "learning_rate": 7.184210526315789e-06,
+      "loss": 1.3911,
       "step": 5450
     },
     {
       "epoch": 0.27,
-      "grad_norm": 4.4375,
-      "learning_rate": 7.168421052631578e-05,
-      "loss": 1.2652,
+      "grad_norm": 16.375,
+      "learning_rate": 7.1684210526315795e-06,
+      "loss": 1.0816,
       "step": 5460
     },
     {
       "epoch": 0.27,
-      "grad_norm": 5.125,
-      "learning_rate": 7.152631578947368e-05,
-      "loss": 1.3024,
+      "grad_norm": 23.125,
+      "learning_rate": 7.152631578947369e-06,
+      "loss": 1.0682,
       "step": 5470
     },
     {
       "epoch": 0.27,
-      "grad_norm": 4.3125,
-      "learning_rate": 7.136842105263157e-05,
-      "loss": 1.3969,
+      "grad_norm": 14.0625,
+      "learning_rate": 7.1368421052631585e-06,
+      "loss": 1.2556,
       "step": 5480
     },
     {
       "epoch": 0.27,
-      "grad_norm": 5.96875,
-      "learning_rate": 7.121052631578947e-05,
-      "loss": 1.5425,
+      "grad_norm": 14.75,
+      "learning_rate": 7.121052631578948e-06,
+      "loss": 1.4108,
       "step": 5490
     },
     {
       "epoch": 0.28,
-      "grad_norm": 6.78125,
-      "learning_rate": 7.105263157894735e-05,
-      "loss": 1.5634,
+      "grad_norm": 24.5,
+      "learning_rate": 7.105263157894737e-06,
+      "loss": 1.3118,
       "step": 5500
     },
     {
       "epoch": 0.28,
-      "grad_norm": 4.03125,
-      "learning_rate": 7.089473684210525e-05,
-      "loss": 1.3567,
+      "grad_norm": 8.75,
+      "learning_rate": 7.089473684210526e-06,
+      "loss": 1.2675,
       "step": 5510
     },
     {
       "epoch": 0.28,
-      "grad_norm": 5.25,
-      "learning_rate": 7.073684210526315e-05,
-      "loss": 1.5746,
+      "grad_norm": 23.0,
+      "learning_rate": 7.073684210526316e-06,
+      "loss": 1.4626,
       "step": 5520
     },
     {
       "epoch": 0.28,
-      "grad_norm": 5.875,
-      "learning_rate": 7.057894736842104e-05,
-      "loss": 1.6727,
+      "grad_norm": 17.5,
+      "learning_rate": 7.057894736842105e-06,
+      "loss": 1.6127,
       "step": 5530
     },
     {
       "epoch": 0.28,
-      "grad_norm": 5.15625,
-      "learning_rate": 7.042105263157894e-05,
-      "loss": 1.709,
+      "grad_norm": 24.875,
+      "learning_rate": 7.042105263157895e-06,
+      "loss": 1.4965,
       "step": 5540
     },
     {
       "epoch": 0.28,
-      "grad_norm": 5.4375,
-      "learning_rate": 7.026315789473684e-05,
-      "loss": 1.6377,
+      "grad_norm": 15.9375,
+      "learning_rate": 7.026315789473685e-06,
+      "loss": 1.5175,
       "step": 5550
     },
     {
       "epoch": 0.28,
-      "grad_norm": 4.75,
-      "learning_rate": 7.010526315789473e-05,
-      "loss": 1.368,
+      "grad_norm": 21.625,
+      "learning_rate": 7.010526315789474e-06,
+      "loss": 1.181,
       "step": 5560
     },
     {
       "epoch": 0.28,
-      "grad_norm": 11.0,
-      "learning_rate": 6.994736842105262e-05,
-      "loss": 1.1897,
+      "grad_norm": 21.125,
+      "learning_rate": 6.994736842105263e-06,
+      "loss": 1.0776,
       "step": 5570
     },
     {
       "epoch": 0.28,
-      "grad_norm": 5.3125,
-      "learning_rate": 6.978947368421051e-05,
-      "loss": 1.4087,
+      "grad_norm": 24.875,
+      "learning_rate": 6.9789473684210525e-06,
+      "loss": 1.2104,
       "step": 5580
     },
     {
       "epoch": 0.28,
-      "grad_norm": 5.53125,
-      "learning_rate": 6.963157894736841e-05,
-      "loss": 1.4641,
+      "grad_norm": 33.25,
+      "learning_rate": 6.963157894736842e-06,
+      "loss": 1.3473,
       "step": 5590
     },
     {
       "epoch": 0.28,
-      "grad_norm": 6.78125,
-      "learning_rate": 6.947368421052631e-05,
-      "loss": 1.2938,
+      "grad_norm": 20.0,
+      "learning_rate": 6.9473684210526315e-06,
+      "loss": 1.2039,
       "step": 5600
     },
     {
       "epoch": 0.28,
-      "grad_norm": 3.390625,
-      "learning_rate": 6.93157894736842e-05,
-      "loss": 1.2211,
+      "grad_norm": 10.125,
+      "learning_rate": 6.931578947368421e-06,
+      "loss": 1.1147,
       "step": 5610
     },
     {
       "epoch": 0.28,
-      "grad_norm": 4.40625,
-      "learning_rate": 6.91578947368421e-05,
-      "loss": 1.239,
+      "grad_norm": 14.25,
+      "learning_rate": 6.9157894736842105e-06,
+      "loss": 1.0561,
       "step": 5620
     },
     {
       "epoch": 0.28,
-      "grad_norm": 6.375,
-      "learning_rate": 6.9e-05,
-      "loss": 1.4557,
+      "grad_norm": 11.9375,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 1.3285,
       "step": 5630
     },
     {
       "epoch": 0.28,
-      "grad_norm": 10.1875,
-      "learning_rate": 6.884210526315788e-05,
-      "loss": 1.5778,
+      "grad_norm": 68.5,
+      "learning_rate": 6.8842105263157895e-06,
+      "loss": 1.4085,
       "step": 5640
     },
     {
       "epoch": 0.28,
-      "grad_norm": 6.1875,
-      "learning_rate": 6.868421052631578e-05,
-      "loss": 1.5556,
+      "grad_norm": 26.0,
+      "learning_rate": 6.868421052631579e-06,
+      "loss": 1.4786,
       "step": 5650
     },
     {
       "epoch": 0.28,
-      "grad_norm": 4.15625,
-      "learning_rate": 6.852631578947367e-05,
-      "loss": 1.2642,
+      "grad_norm": 20.125,
+      "learning_rate": 6.8526315789473685e-06,
+      "loss": 1.1576,
       "step": 5660
     },
     {
       "epoch": 0.28,
-      "grad_norm": 4.53125,
-      "learning_rate": 6.836842105263157e-05,
-      "loss": 1.3846,
+      "grad_norm": 15.5,
+      "learning_rate": 6.836842105263158e-06,
+      "loss": 1.2392,
       "step": 5670
     },
     {
       "epoch": 0.28,
-      "grad_norm": 4.84375,
-      "learning_rate": 6.821052631578947e-05,
-      "loss": 1.6962,
+      "grad_norm": 24.125,
+      "learning_rate": 6.8210526315789475e-06,
+      "loss": 1.5732,
       "step": 5680
     },
     {
       "epoch": 0.28,
-      "grad_norm": 7.03125,
-      "learning_rate": 6.805263157894736e-05,
-      "loss": 1.7847,
+      "grad_norm": 28.0,
+      "learning_rate": 6.805263157894737e-06,
+      "loss": 1.6376,
       "step": 5690
     },
     {
       "epoch": 0.28,
-      "grad_norm": 3.734375,
-      "learning_rate": 6.789473684210526e-05,
-      "loss": 1.618,
+      "grad_norm": 8.4375,
+      "learning_rate": 6.7894736842105264e-06,
+      "loss": 1.4348,
       "step": 5700
     },
     {
       "epoch": 0.29,
-      "grad_norm": 7.59375,
-      "learning_rate": 6.773684210526316e-05,
-      "loss": 1.2276,
+      "grad_norm": 36.5,
+      "learning_rate": 6.773684210526317e-06,
+      "loss": 1.1923,
       "step": 5710
     },
     {
       "epoch": 0.29,
-      "grad_norm": 4.0,
-      "learning_rate": 6.757894736842104e-05,
-      "loss": 1.4253,
+      "grad_norm": 16.625,
+      "learning_rate": 6.7578947368421054e-06,
+      "loss": 1.3055,
       "step": 5720
     },
     {
       "epoch": 0.29,
-      "grad_norm": 6.34375,
-      "learning_rate": 6.742105263157894e-05,
-      "loss": 1.5658,
+      "grad_norm": 16.5,
+      "learning_rate": 6.742105263157895e-06,
+      "loss": 1.5263,
       "step": 5730
     },
     {
       "epoch": 0.29,
-      "grad_norm": 6.875,
-      "learning_rate": 6.726315789473683e-05,
-      "loss": 1.6033,
+      "grad_norm": 21.25,
+      "learning_rate": 6.726315789473684e-06,
+      "loss": 1.4074,
       "step": 5740
     },
     {
       "epoch": 0.29,
-      "grad_norm": 5.90625,
-      "learning_rate": 6.710526315789473e-05,
-      "loss": 1.3583,
+      "grad_norm": 11.0,
+      "learning_rate": 6.710526315789474e-06,
+      "loss": 1.2683,
       "step": 5750
     },
     {
       "epoch": 0.29,
-      "grad_norm": 6.15625,
-      "learning_rate": 6.694736842105263e-05,
-      "loss": 1.4659,
+      "grad_norm": 25.125,
+      "learning_rate": 6.694736842105263e-06,
+      "loss": 1.2666,
       "step": 5760
     },
     {
       "epoch": 0.29,
-      "grad_norm": 5.09375,
-      "learning_rate": 6.678947368421052e-05,
-      "loss": 1.7884,
+      "grad_norm": 17.875,
+      "learning_rate": 6.678947368421053e-06,
+      "loss": 1.6824,
       "step": 5770
     },
     {
       "epoch": 0.29,
-      "grad_norm": 5.0625,
-      "learning_rate": 6.663157894736842e-05,
-      "loss": 2.0443,
+      "grad_norm": 22.0,
+      "learning_rate": 6.663157894736842e-06,
+      "loss": 1.9028,
       "step": 5780
     },
     {
       "epoch": 0.29,
-      "grad_norm": 6.90625,
-      "learning_rate": 6.64736842105263e-05,
-      "loss": 1.7151,
+      "grad_norm": 29.0,
+      "learning_rate": 6.647368421052631e-06,
+      "loss": 1.5481,
       "step": 5790
     },
     {
       "epoch": 0.29,
-      "grad_norm": 3.453125,
-      "learning_rate": 6.63157894736842e-05,
-      "loss": 1.6927,
+      "grad_norm": 15.3125,
+      "learning_rate": 6.631578947368421e-06,
+      "loss": 1.4571,
       "step": 5800
     },
     {
       "epoch": 0.29,
-      "grad_norm": 5.0625,
-      "learning_rate": 6.61578947368421e-05,
-      "loss": 1.4678,
+      "grad_norm": 19.5,
+      "learning_rate": 6.615789473684211e-06,
+      "loss": 1.3493,
       "step": 5810
     },
     {
       "epoch": 0.29,
-      "grad_norm": 3.171875,
-      "learning_rate": 6.599999999999999e-05,
-      "loss": 1.2995,
+      "grad_norm": 18.125,
+      "learning_rate": 6.6e-06,
+      "loss": 1.1466,
       "step": 5820
     },
     {
       "epoch": 0.29,
-      "grad_norm": 5.0625,
-      "learning_rate": 6.584210526315789e-05,
-      "loss": 1.503,
+      "grad_norm": 18.5,
+      "learning_rate": 6.58421052631579e-06,
+      "loss": 1.2788,
       "step": 5830
     },
     {
       "epoch": 0.29,
-      "grad_norm": 3.84375,
-      "learning_rate": 6.568421052631578e-05,
-      "loss": 1.3394,
+      "grad_norm": 23.5,
+      "learning_rate": 6.568421052631579e-06,
+      "loss": 1.2194,
       "step": 5840
     },
     {
       "epoch": 0.29,
-      "grad_norm": 4.46875,
-      "learning_rate": 6.552631578947368e-05,
-      "loss": 1.1829,
+      "grad_norm": 24.375,
+      "learning_rate": 6.552631578947369e-06,
+      "loss": 1.1311,
       "step": 5850
     },
     {
       "epoch": 0.29,
-      "grad_norm": 6.21875,
-      "learning_rate": 6.536842105263156e-05,
-      "loss": 1.1117,
+      "grad_norm": 15.9375,
+      "learning_rate": 6.5368421052631575e-06,
+      "loss": 0.91,
       "step": 5860
     },
     {
       "epoch": 0.29,
-      "grad_norm": 6.0,
-      "learning_rate": 6.521052631578946e-05,
-      "loss": 1.7312,
+      "grad_norm": 17.5,
+      "learning_rate": 6.521052631578947e-06,
+      "loss": 1.5186,
       "step": 5870
     },
     {
       "epoch": 0.29,
-      "grad_norm": 5.4375,
-      "learning_rate": 6.505263157894736e-05,
-      "loss": 1.459,
+      "grad_norm": 26.0,
+      "learning_rate": 6.505263157894737e-06,
+      "loss": 1.2989,
       "step": 5880
     },
     {
       "epoch": 0.29,
-      "grad_norm": 4.84375,
-      "learning_rate": 6.489473684210525e-05,
-      "loss": 1.179,
+      "grad_norm": 21.375,
+      "learning_rate": 6.489473684210527e-06,
+      "loss": 1.0283,
       "step": 5890
     },
     {
       "epoch": 0.29,
-      "grad_norm": 5.78125,
-      "learning_rate": 6.473684210526315e-05,
-      "loss": 1.4745,
+      "grad_norm": 25.5,
+      "learning_rate": 6.473684210526316e-06,
+      "loss": 1.4068,
       "step": 5900
     },
     {
       "epoch": 0.3,
-      "grad_norm": 4.4375,
-      "learning_rate": 6.457894736842105e-05,
-      "loss": 1.7963,
+      "grad_norm": 14.125,
+      "learning_rate": 6.457894736842106e-06,
+      "loss": 1.5789,
       "step": 5910
     },
     {
       "epoch": 0.3,
-      "grad_norm": 5.9375,
-      "learning_rate": 6.442105263157894e-05,
-      "loss": 1.23,
+      "grad_norm": 16.625,
+      "learning_rate": 6.442105263157895e-06,
+      "loss": 1.1244,
       "step": 5920
     },
     {
       "epoch": 0.3,
-      "grad_norm": 3.609375,
-      "learning_rate": 6.426315789473683e-05,
-      "loss": 1.3367,
+      "grad_norm": 19.75,
+      "learning_rate": 6.426315789473684e-06,
+      "loss": 1.2412,
       "step": 5930
     },
     {
       "epoch": 0.3,
-      "grad_norm": 4.15625,
-      "learning_rate": 6.410526315789472e-05,
-      "loss": 1.4904,
+      "grad_norm": 10.3125,
+      "learning_rate": 6.410526315789473e-06,
+      "loss": 1.336,
       "step": 5940
     },
     {
       "epoch": 0.3,
-      "grad_norm": 4.375,
-      "learning_rate": 6.394736842105262e-05,
-      "loss": 1.2374,
+      "grad_norm": 22.625,
+      "learning_rate": 6.394736842105263e-06,
+      "loss": 1.2082,
       "step": 5950
     },
     {
       "epoch": 0.3,
-      "grad_norm": 5.21875,
-      "learning_rate": 6.378947368421052e-05,
-      "loss": 1.4946,
+      "grad_norm": 22.375,
+      "learning_rate": 6.378947368421053e-06,
+      "loss": 1.3572,
       "step": 5960
     },
     {
       "epoch": 0.3,
-      "grad_norm": 4.6875,
-      "learning_rate": 6.363157894736841e-05,
-      "loss": 1.4075,
+      "grad_norm": 15.875,
+      "learning_rate": 6.363157894736843e-06,
+      "loss": 1.4054,
       "step": 5970
     },
     {
       "epoch": 0.3,
-      "grad_norm": 6.59375,
-      "learning_rate": 6.347368421052631e-05,
-      "loss": 1.6144,
+      "grad_norm": 25.625,
+      "learning_rate": 6.347368421052632e-06,
+      "loss": 1.4423,
       "step": 5980
     },
     {
       "epoch": 0.3,
-      "grad_norm": 5.71875,
-      "learning_rate": 6.33157894736842e-05,
-      "loss": 1.2361,
+      "grad_norm": 21.5,
+      "learning_rate": 6.331578947368422e-06,
+      "loss": 1.1326,
       "step": 5990
     },
     {
       "epoch": 0.3,
-      "grad_norm": 6.625,
-      "learning_rate": 6.315789473684209e-05,
-      "loss": 1.4764,
+      "grad_norm": 33.25,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 1.2669,
       "step": 6000
     },
     {
       "epoch": 0.3,
-      "eval_loss": 1.4902899265289307,
-      "eval_runtime": 30.9163,
-      "eval_samples_per_second": 32.345,
-      "eval_steps_per_second": 32.345,
+      "eval_loss": 1.3724089860916138,
+      "eval_runtime": 30.7072,
+      "eval_samples_per_second": 32.566,
+      "eval_steps_per_second": 32.566,
       "step": 6000
     },
     {
       "epoch": 0.3,
-      "grad_norm": 4.875,
-      "learning_rate": 6.299999999999999e-05,
-      "loss": 1.568,
+      "grad_norm": 24.375,
+      "learning_rate": 6.3e-06,
+      "loss": 1.2768,
       "step": 6010
     },
     {
       "epoch": 0.3,
-      "grad_norm": 5.8125,
-      "learning_rate": 6.284210526315788e-05,
-      "loss": 1.3994,
+      "grad_norm": 40.5,
+      "learning_rate": 6.284210526315789e-06,
+      "loss": 1.2205,
       "step": 6020
     },
     {
       "epoch": 0.3,
-      "grad_norm": 5.71875,
-      "learning_rate": 6.268421052631578e-05,
-      "loss": 1.5673,
+      "grad_norm": 28.875,
+      "learning_rate": 6.268421052631579e-06,
+      "loss": 1.4383,
       "step": 6030
     },
     {
       "epoch": 0.3,
-      "grad_norm": 5.5625,
-      "learning_rate": 6.252631578947368e-05,
-      "loss": 1.4264,
+      "grad_norm": 25.125,
+      "learning_rate": 6.252631578947368e-06,
+      "loss": 1.3299,
       "step": 6040
     },
     {
       "epoch": 0.3,
-      "grad_norm": 6.34375,
-      "learning_rate": 6.236842105263157e-05,
-      "loss": 1.6823,
+      "grad_norm": 39.5,
+      "learning_rate": 6.236842105263159e-06,
+      "loss": 1.5192,
       "step": 6050
     },
     {
       "epoch": 0.3,
-      "grad_norm": 3.84375,
-      "learning_rate": 6.221052631578947e-05,
-      "loss": 1.4544,
+      "grad_norm": 19.0,
+      "learning_rate": 6.221052631578948e-06,
+      "loss": 1.4088,
       "step": 6060
     },
     {
       "epoch": 0.3,
-      "grad_norm": 5.9375,
-      "learning_rate": 6.205263157894737e-05,
-      "loss": 1.4708,
+      "grad_norm": 26.5,
+      "learning_rate": 6.205263157894737e-06,
+      "loss": 1.4427,
       "step": 6070
     },
     {
       "epoch": 0.3,
-      "grad_norm": 5.0,
-      "learning_rate": 6.189473684210525e-05,
-      "loss": 1.4171,
+      "grad_norm": 18.25,
+      "learning_rate": 6.189473684210526e-06,
+      "loss": 1.2549,
       "step": 6080
     },
     {
       "epoch": 0.3,
-      "grad_norm": 6.90625,
-      "learning_rate": 6.173684210526315e-05,
-      "loss": 1.8525,
+      "grad_norm": 23.25,
+      "learning_rate": 6.173684210526316e-06,
+      "loss": 1.7159,
       "step": 6090
     },
     {
       "epoch": 0.3,
-      "grad_norm": 5.4375,
-      "learning_rate": 6.157894736842104e-05,
-      "loss": 1.386,
+      "grad_norm": 15.1875,
+      "learning_rate": 6.157894736842105e-06,
+      "loss": 1.2445,
       "step": 6100
     },
     {
       "epoch": 0.31,
-      "grad_norm": 5.09375,
-      "learning_rate": 6.142105263157894e-05,
-      "loss": 1.2856,
+      "grad_norm": 20.375,
+      "learning_rate": 6.142105263157895e-06,
+      "loss": 0.9629,
       "step": 6110
     },
     {
       "epoch": 0.31,
-      "grad_norm": 4.71875,
-      "learning_rate": 6.126315789473684e-05,
-      "loss": 1.4263,
+      "grad_norm": 11.125,
+      "learning_rate": 6.126315789473684e-06,
+      "loss": 1.3391,
       "step": 6120
     },
     {
       "epoch": 0.31,
-      "grad_norm": 8.0,
-      "learning_rate": 6.110526315789473e-05,
-      "loss": 1.4666,
+      "grad_norm": 17.75,
+      "learning_rate": 6.1105263157894746e-06,
+      "loss": 1.1964,
       "step": 6130
     },
     {
       "epoch": 0.31,
-      "grad_norm": 5.875,
-      "learning_rate": 6.094736842105262e-05,
-      "loss": 1.3614,
+      "grad_norm": 26.875,
+      "learning_rate": 6.094736842105263e-06,
+      "loss": 1.1751,
       "step": 6140
     },
     {
       "epoch": 0.31,
-      "grad_norm": 5.03125,
-      "learning_rate": 6.078947368421052e-05,
-      "loss": 1.3097,
+      "grad_norm": 14.0625,
+      "learning_rate": 6.078947368421053e-06,
+      "loss": 1.2314,
       "step": 6150
     },
     {
       "epoch": 0.31,
-      "grad_norm": 5.15625,
-      "learning_rate": 6.0631578947368415e-05,
-      "loss": 1.5042,
+      "grad_norm": 21.5,
+      "learning_rate": 6.063157894736842e-06,
+      "loss": 1.3675,
       "step": 6160
     },
     {
       "epoch": 0.31,
-      "grad_norm": 5.1875,
-      "learning_rate": 6.047368421052631e-05,
-      "loss": 1.0817,
+      "grad_norm": 30.75,
+      "learning_rate": 6.047368421052632e-06,
+      "loss": 1.0896,
       "step": 6170
     },
     {
       "epoch": 0.31,
-      "grad_norm": 5.6875,
-      "learning_rate": 6.031578947368421e-05,
-      "loss": 1.5722,
+      "grad_norm": 18.25,
+      "learning_rate": 6.031578947368421e-06,
+      "loss": 1.5274,
       "step": 6180
     },
     {
       "epoch": 0.31,
-      "grad_norm": 5.9375,
-      "learning_rate": 6.0157894736842105e-05,
-      "loss": 1.366,
+      "grad_norm": 17.25,
+      "learning_rate": 6.015789473684211e-06,
+      "loss": 1.2702,
       "step": 6190
     },
     {
       "epoch": 0.31,
-      "grad_norm": 4.1875,
-      "learning_rate": 5.9999999999999995e-05,
-      "loss": 1.6359,
+      "grad_norm": 21.5,
+      "learning_rate": 6e-06,
+      "loss": 1.5554,
       "step": 6200
     },
     {
       "epoch": 0.31,
-      "grad_norm": 5.125,
-      "learning_rate": 5.9842105263157885e-05,
-      "loss": 1.7341,
+      "grad_norm": 21.0,
+      "learning_rate": 5.98421052631579e-06,
+      "loss": 1.5431,
       "step": 6210
     },
     {
       "epoch": 0.31,
-      "grad_norm": 3.625,
-      "learning_rate": 5.968421052631578e-05,
-      "loss": 1.7589,
+      "grad_norm": 13.1875,
+      "learning_rate": 5.968421052631579e-06,
+      "loss": 1.6954,
       "step": 6220
     },
     {
       "epoch": 0.31,
-      "grad_norm": 5.21875,
-      "learning_rate": 5.952631578947368e-05,
-      "loss": 1.6357,
+      "grad_norm": 15.9375,
+      "learning_rate": 5.952631578947369e-06,
+      "loss": 1.5684,
       "step": 6230
     },
     {
       "epoch": 0.31,
-      "grad_norm": 7.25,
-      "learning_rate": 5.9368421052631574e-05,
-      "loss": 1.4806,
+      "grad_norm": 11.8125,
+      "learning_rate": 5.936842105263158e-06,
+      "loss": 1.2937,
       "step": 6240
     },
     {
       "epoch": 0.31,
-      "grad_norm": 5.09375,
-      "learning_rate": 5.921052631578947e-05,
-      "loss": 1.5461,
+      "grad_norm": 14.1875,
+      "learning_rate": 5.921052631578948e-06,
+      "loss": 1.4135,
       "step": 6250
     },
     {
       "epoch": 0.31,
-      "grad_norm": 6.5625,
-      "learning_rate": 5.905263157894737e-05,
-      "loss": 1.3267,
+      "grad_norm": 19.625,
+      "learning_rate": 5.905263157894737e-06,
+      "loss": 1.1792,
       "step": 6260
     },
     {
       "epoch": 0.31,
-      "grad_norm": 4.90625,
-      "learning_rate": 5.8894736842105264e-05,
-      "loss": 1.5908,
+      "grad_norm": 18.0,
+      "learning_rate": 5.889473684210527e-06,
+      "loss": 1.4643,
       "step": 6270
     },
     {
       "epoch": 0.31,
-      "grad_norm": 3.890625,
-      "learning_rate": 5.873684210526315e-05,
-      "loss": 1.5187,
+      "grad_norm": 20.5,
+      "learning_rate": 5.873684210526315e-06,
+      "loss": 1.4952,
       "step": 6280
     },
     {
       "epoch": 0.31,
-      "grad_norm": 5.875,
-      "learning_rate": 5.8578947368421044e-05,
-      "loss": 1.5094,
+      "grad_norm": 22.0,
+      "learning_rate": 5.857894736842105e-06,
+      "loss": 1.4588,
       "step": 6290
     },
     {
       "epoch": 0.32,
-      "grad_norm": 7.625,
-      "learning_rate": 5.842105263157894e-05,
-      "loss": 1.5578,
+      "grad_norm": 20.5,
+      "learning_rate": 5.842105263157895e-06,
+      "loss": 1.5374,
       "step": 6300
     },
     {
       "epoch": 0.32,
-      "grad_norm": 5.65625,
-      "learning_rate": 5.826315789473684e-05,
-      "loss": 1.5853,
+      "grad_norm": 26.625,
+      "learning_rate": 5.8263157894736846e-06,
+      "loss": 1.4717,
       "step": 6310
     },
     {
       "epoch": 0.32,
-      "grad_norm": 5.25,
-      "learning_rate": 5.8105263157894734e-05,
-      "loss": 1.0628,
+      "grad_norm": 20.0,
+      "learning_rate": 5.810526315789474e-06,
+      "loss": 1.1332,
       "step": 6320
     },
     {
       "epoch": 0.32,
-      "grad_norm": 6.0625,
-      "learning_rate": 5.794736842105263e-05,
-      "loss": 1.6518,
+      "grad_norm": 7.90625,
+      "learning_rate": 5.7947368421052635e-06,
+      "loss": 1.4775,
       "step": 6330
     },
     {
       "epoch": 0.32,
-      "grad_norm": 5.25,
-      "learning_rate": 5.778947368421053e-05,
-      "loss": 1.7742,
+      "grad_norm": 22.5,
+      "learning_rate": 5.778947368421053e-06,
+      "loss": 1.7034,
       "step": 6340
     },
     {
       "epoch": 0.32,
-      "grad_norm": 4.625,
-      "learning_rate": 5.763157894736841e-05,
-      "loss": 1.3831,
+      "grad_norm": 18.875,
+      "learning_rate": 5.763157894736842e-06,
+      "loss": 1.3331,
       "step": 6350
     },
     {
       "epoch": 0.32,
-      "grad_norm": 4.15625,
-      "learning_rate": 5.7473684210526307e-05,
-      "loss": 1.2819,
+      "grad_norm": 10.3125,
+      "learning_rate": 5.747368421052631e-06,
+      "loss": 1.1408,
       "step": 6360
     },
     {
       "epoch": 0.32,
-      "grad_norm": 7.6875,
-      "learning_rate": 5.73157894736842e-05,
-      "loss": 1.7622,
+      "grad_norm": 21.875,
+      "learning_rate": 5.731578947368421e-06,
+      "loss": 1.6183,
       "step": 6370
     },
     {
       "epoch": 0.32,
-      "grad_norm": 6.0625,
-      "learning_rate": 5.71578947368421e-05,
-      "loss": 1.3274,
+      "grad_norm": 27.375,
+      "learning_rate": 5.715789473684211e-06,
+      "loss": 1.282,
       "step": 6380
     },
     {
       "epoch": 0.32,
-      "grad_norm": 5.6875,
-      "learning_rate": 5.6999999999999996e-05,
-      "loss": 1.5017,
+      "grad_norm": 17.0,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 1.4119,
       "step": 6390
     },
     {
       "epoch": 0.32,
-      "grad_norm": 5.40625,
-      "learning_rate": 5.684210526315789e-05,
-      "loss": 1.5522,
+      "grad_norm": 19.125,
+      "learning_rate": 5.68421052631579e-06,
+      "loss": 1.3382,
       "step": 6400
     },
     {
       "epoch": 0.32,
-      "grad_norm": 6.21875,
-      "learning_rate": 5.668421052631579e-05,
-      "loss": 1.752,
+      "grad_norm": 17.0,
+      "learning_rate": 5.6684210526315795e-06,
+      "loss": 1.6847,
       "step": 6410
     },
     {
       "epoch": 0.32,
-      "grad_norm": 8.1875,
-      "learning_rate": 5.652631578947367e-05,
-      "loss": 1.3857,
+      "grad_norm": 18.75,
+      "learning_rate": 5.652631578947368e-06,
+      "loss": 1.3082,
       "step": 6420
     },
     {
       "epoch": 0.32,
-      "grad_norm": 7.4375,
-      "learning_rate": 5.636842105263157e-05,
-      "loss": 1.4945,
+      "grad_norm": 29.875,
+      "learning_rate": 5.636842105263158e-06,
+      "loss": 1.3222,
       "step": 6430
     },
     {
       "epoch": 0.32,
-      "grad_norm": 4.4375,
-      "learning_rate": 5.6210526315789466e-05,
-      "loss": 1.5736,
+      "grad_norm": 13.125,
+      "learning_rate": 5.621052631578947e-06,
+      "loss": 1.4203,
       "step": 6440
     },
     {
       "epoch": 0.32,
-      "grad_norm": 5.0625,
-      "learning_rate": 5.605263157894736e-05,
-      "loss": 1.4432,
+      "grad_norm": 22.5,
+      "learning_rate": 5.605263157894737e-06,
+      "loss": 1.3574,
       "step": 6450
     },
     {
       "epoch": 0.32,
-      "grad_norm": 9.625,
-      "learning_rate": 5.589473684210526e-05,
-      "loss": 1.499,
+      "grad_norm": 44.0,
+      "learning_rate": 5.589473684210527e-06,
+      "loss": 1.421,
       "step": 6460
     },
     {
       "epoch": 0.32,
-      "grad_norm": 5.34375,
-      "learning_rate": 5.5736842105263156e-05,
-      "loss": 1.54,
+      "grad_norm": 28.625,
+      "learning_rate": 5.573684210526316e-06,
+      "loss": 1.5993,
       "step": 6470
     },
     {
       "epoch": 0.32,
-      "grad_norm": 5.40625,
-      "learning_rate": 5.557894736842105e-05,
-      "loss": 1.3533,
+      "grad_norm": 19.375,
+      "learning_rate": 5.557894736842106e-06,
+      "loss": 1.2983,
       "step": 6480
     },
     {
       "epoch": 0.32,
-      "grad_norm": 8.0625,
-      "learning_rate": 5.542105263157895e-05,
-      "loss": 1.335,
+      "grad_norm": 23.625,
+      "learning_rate": 5.542105263157895e-06,
+      "loss": 1.3245,
       "step": 6490
     },
     {
       "epoch": 0.33,
-      "grad_norm": 5.875,
-      "learning_rate": 5.526315789473683e-05,
-      "loss": 1.498,
+      "grad_norm": 27.125,
+      "learning_rate": 5.526315789473684e-06,
+      "loss": 1.3728,
       "step": 6500
     },
     {
       "epoch": 0.33,
-      "grad_norm": 5.3125,
-      "learning_rate": 5.510526315789473e-05,
-      "loss": 1.37,
+      "grad_norm": 17.75,
+      "learning_rate": 5.5105263157894735e-06,
+      "loss": 1.2081,
       "step": 6510
     },
     {
       "epoch": 0.33,
-      "grad_norm": 6.3125,
-      "learning_rate": 5.4947368421052625e-05,
-      "loss": 1.3711,
+      "grad_norm": 27.875,
+      "learning_rate": 5.494736842105263e-06,
+      "loss": 1.2897,
       "step": 6520
     },
     {
       "epoch": 0.33,
-      "grad_norm": 5.5625,
-      "learning_rate": 5.478947368421052e-05,
-      "loss": 1.3383,
+      "grad_norm": 17.75,
+      "learning_rate": 5.4789473684210525e-06,
+      "loss": 1.373,
       "step": 6530
     },
     {
       "epoch": 0.33,
-      "grad_norm": 5.09375,
-      "learning_rate": 5.463157894736842e-05,
-      "loss": 1.5205,
+      "grad_norm": 19.5,
+      "learning_rate": 5.463157894736842e-06,
+      "loss": 1.4809,
       "step": 6540
     },
     {
       "epoch": 0.33,
-      "grad_norm": 7.09375,
-      "learning_rate": 5.4473684210526315e-05,
-      "loss": 1.4338,
+      "grad_norm": 43.25,
+      "learning_rate": 5.447368421052632e-06,
+      "loss": 1.4633,
       "step": 6550
     },
     {
       "epoch": 0.33,
-      "grad_norm": 7.6875,
-      "learning_rate": 5.431578947368421e-05,
-      "loss": 1.4553,
+      "grad_norm": 37.75,
+      "learning_rate": 5.431578947368422e-06,
+      "loss": 1.3994,
       "step": 6560
     },
     {
       "epoch": 0.33,
-      "grad_norm": 6.8125,
-      "learning_rate": 5.4157894736842095e-05,
-      "loss": 1.5879,
+      "grad_norm": 25.625,
+      "learning_rate": 5.4157894736842105e-06,
+      "loss": 1.3844,
       "step": 6570
     },
     {
       "epoch": 0.33,
-      "grad_norm": 5.0625,
-      "learning_rate": 5.399999999999999e-05,
-      "loss": 1.4582,
+      "grad_norm": 15.125,
+      "learning_rate": 5.4e-06,
+      "loss": 1.3222,
       "step": 6580
     },
     {
       "epoch": 0.33,
-      "grad_norm": 5.0,
-      "learning_rate": 5.384210526315789e-05,
-      "loss": 1.3241,
+      "grad_norm": 21.5,
+      "learning_rate": 5.3842105263157895e-06,
+      "loss": 1.2739,
       "step": 6590
     },
     {
       "epoch": 0.33,
-      "grad_norm": 6.96875,
-      "learning_rate": 5.3684210526315784e-05,
-      "loss": 1.7072,
+      "grad_norm": 76.0,
+      "learning_rate": 5.368421052631579e-06,
+      "loss": 1.6814,
       "step": 6600
     },
     {
       "epoch": 0.33,
-      "grad_norm": 7.34375,
-      "learning_rate": 5.352631578947368e-05,
-      "loss": 1.5041,
+      "grad_norm": 15.25,
+      "learning_rate": 5.3526315789473684e-06,
+      "loss": 1.3229,
       "step": 6610
     },
     {
       "epoch": 0.33,
-      "grad_norm": 5.28125,
-      "learning_rate": 5.336842105263158e-05,
-      "loss": 1.7467,
+      "grad_norm": 21.25,
+      "learning_rate": 5.336842105263158e-06,
+      "loss": 1.7966,
       "step": 6620
     },
     {
       "epoch": 0.33,
-      "grad_norm": 5.375,
-      "learning_rate": 5.3210526315789474e-05,
-      "loss": 1.3781,
+      "grad_norm": 28.625,
+      "learning_rate": 5.321052631578948e-06,
+      "loss": 1.2918,
       "step": 6630
     },
     {
       "epoch": 0.33,
-      "grad_norm": 5.0,
-      "learning_rate": 5.305263157894736e-05,
-      "loss": 1.3092,
+      "grad_norm": 20.75,
+      "learning_rate": 5.305263157894737e-06,
+      "loss": 1.2418,
       "step": 6640
     },
     {
       "epoch": 0.33,
-      "grad_norm": 2.796875,
-      "learning_rate": 5.2894736842105254e-05,
-      "loss": 1.2497,
+      "grad_norm": 10.125,
+      "learning_rate": 5.289473684210526e-06,
+      "loss": 1.0768,
       "step": 6650
     },
     {
       "epoch": 0.33,
-      "grad_norm": 4.9375,
-      "learning_rate": 5.273684210526315e-05,
-      "loss": 1.0076,
+      "grad_norm": 21.875,
+      "learning_rate": 5.273684210526316e-06,
+      "loss": 0.9163,
       "step": 6660
     },
     {
       "epoch": 0.33,
-      "grad_norm": 11.0625,
-      "learning_rate": 5.257894736842105e-05,
-      "loss": 1.4536,
+      "grad_norm": 21.375,
+      "learning_rate": 5.257894736842105e-06,
+      "loss": 1.3902,
       "step": 6670
     },
     {
       "epoch": 0.33,
-      "grad_norm": 7.09375,
-      "learning_rate": 5.2421052631578944e-05,
-      "loss": 1.3639,
+      "grad_norm": 24.0,
+      "learning_rate": 5.242105263157895e-06,
+      "loss": 1.2549,
       "step": 6680
     },
     {
       "epoch": 0.33,
-      "grad_norm": 8.9375,
-      "learning_rate": 5.226315789473684e-05,
-      "loss": 1.5545,
+      "grad_norm": 29.25,
+      "learning_rate": 5.226315789473684e-06,
+      "loss": 1.4466,
       "step": 6690
     },
     {
       "epoch": 0.34,
-      "grad_norm": 6.59375,
-      "learning_rate": 5.210526315789474e-05,
-      "loss": 1.2892,
+      "grad_norm": 20.375,
+      "learning_rate": 5.210526315789474e-06,
+      "loss": 1.2282,
       "step": 6700
     },
     {
       "epoch": 0.34,
-      "grad_norm": 5.125,
-      "learning_rate": 5.194736842105263e-05,
-      "loss": 1.4664,
+      "grad_norm": 28.5,
+      "learning_rate": 5.194736842105263e-06,
+      "loss": 1.5241,
       "step": 6710
     },
     {
       "epoch": 0.34,
-      "grad_norm": 6.875,
-      "learning_rate": 5.1789473684210517e-05,
-      "loss": 1.542,
+      "grad_norm": 18.75,
+      "learning_rate": 5.178947368421053e-06,
+      "loss": 1.4206,
       "step": 6720
     },
     {
       "epoch": 0.34,
-      "grad_norm": 5.15625,
-      "learning_rate": 5.163157894736841e-05,
-      "loss": 1.6154,
+      "grad_norm": 20.375,
+      "learning_rate": 5.163157894736842e-06,
+      "loss": 1.5183,
       "step": 6730
     },
     {
       "epoch": 0.34,
-      "grad_norm": 4.21875,
-      "learning_rate": 5.147368421052631e-05,
-      "loss": 1.289,
+      "grad_norm": 31.25,
+      "learning_rate": 5.147368421052632e-06,
+      "loss": 1.1815,
       "step": 6740
     },
     {
       "epoch": 0.34,
-      "grad_norm": 4.75,
-      "learning_rate": 5.1315789473684206e-05,
-      "loss": 1.4106,
+      "grad_norm": 24.375,
+      "learning_rate": 5.131578947368421e-06,
+      "loss": 1.2572,
       "step": 6750
     },
     {
       "epoch": 0.34,
-      "grad_norm": 5.625,
-      "learning_rate": 5.11578947368421e-05,
-      "loss": 1.4117,
+      "grad_norm": 14.375,
+      "learning_rate": 5.115789473684211e-06,
+      "loss": 1.4552,
       "step": 6760
     },
     {
       "epoch": 0.34,
-      "grad_norm": 6.875,
-      "learning_rate": 5.1e-05,
-      "loss": 1.5164,
+      "grad_norm": 19.25,
+      "learning_rate": 5.1e-06,
+      "loss": 1.5616,
       "step": 6770
     },
     {
       "epoch": 0.34,
-      "grad_norm": 6.25,
-      "learning_rate": 5.084210526315789e-05,
-      "loss": 1.3858,
+      "grad_norm": 35.5,
+      "learning_rate": 5.084210526315789e-06,
+      "loss": 1.37,
       "step": 6780
     },
     {
       "epoch": 0.34,
-      "grad_norm": 7.3125,
-      "learning_rate": 5.0684210526315786e-05,
-      "loss": 1.6531,
+      "grad_norm": 15.625,
+      "learning_rate": 5.0684210526315784e-06,
+      "loss": 1.6033,
       "step": 6790
     },
     {
       "epoch": 0.34,
-      "grad_norm": 6.125,
-      "learning_rate": 5.0526315789473676e-05,
-      "loss": 1.4197,
+      "grad_norm": 22.875,
+      "learning_rate": 5.052631578947369e-06,
+      "loss": 1.3356,
       "step": 6800
     },
     {
       "epoch": 0.34,
-      "grad_norm": 6.625,
-      "learning_rate": 5.036842105263157e-05,
-      "loss": 1.6369,
+      "grad_norm": 35.5,
+      "learning_rate": 5.036842105263158e-06,
+      "loss": 1.5591,
       "step": 6810
     },
     {
       "epoch": 0.34,
-      "grad_norm": 9.125,
-      "learning_rate": 5.021052631578947e-05,
-      "loss": 1.4371,
+      "grad_norm": 28.875,
+      "learning_rate": 5.021052631578948e-06,
+      "loss": 1.2637,
       "step": 6820
     },
     {
       "epoch": 0.34,
-      "grad_norm": 5.25,
-      "learning_rate": 5.0052631578947366e-05,
-      "loss": 1.3158,
+      "grad_norm": 20.125,
+      "learning_rate": 5.005263157894737e-06,
+      "loss": 1.2172,
       "step": 6830
     },
     {
       "epoch": 0.34,
-      "grad_norm": 5.25,
-      "learning_rate": 4.989473684210526e-05,
-      "loss": 1.421,
+      "grad_norm": 20.75,
+      "learning_rate": 4.989473684210527e-06,
+      "loss": 1.3636,
       "step": 6840
     },
     {
       "epoch": 0.34,
-      "grad_norm": 4.75,
-      "learning_rate": 4.973684210526315e-05,
-      "loss": 1.3778,
+      "grad_norm": 16.625,
+      "learning_rate": 4.973684210526315e-06,
+      "loss": 1.3217,
       "step": 6850
     },
     {
       "epoch": 0.34,
-      "grad_norm": 4.9375,
-      "learning_rate": 4.957894736842105e-05,
-      "loss": 1.4383,
+      "grad_norm": 23.75,
+      "learning_rate": 4.957894736842105e-06,
+      "loss": 1.4014,
       "step": 6860
     },
     {
       "epoch": 0.34,
-      "grad_norm": 4.6875,
-      "learning_rate": 4.942105263157894e-05,
-      "loss": 1.4528,
+      "grad_norm": 15.5,
+      "learning_rate": 4.942105263157894e-06,
+      "loss": 1.4636,
       "step": 6870
     },
     {
       "epoch": 0.34,
-      "grad_norm": 6.53125,
-      "learning_rate": 4.9263157894736835e-05,
-      "loss": 1.4387,
+      "grad_norm": 19.0,
+      "learning_rate": 4.926315789473685e-06,
+      "loss": 1.4897,
       "step": 6880
     },
     {
       "epoch": 0.34,
-      "grad_norm": 6.125,
-      "learning_rate": 4.910526315789473e-05,
-      "loss": 1.4136,
+      "grad_norm": 24.5,
+      "learning_rate": 4.910526315789474e-06,
+      "loss": 1.2786,
       "step": 6890
     },
     {
       "epoch": 0.34,
-      "grad_norm": 6.375,
-      "learning_rate": 4.894736842105263e-05,
-      "loss": 1.5636,
+      "grad_norm": 20.25,
+      "learning_rate": 4.894736842105264e-06,
+      "loss": 1.4247,
       "step": 6900
     },
     {
       "epoch": 0.35,
-      "grad_norm": 4.6875,
-      "learning_rate": 4.8789473684210525e-05,
-      "loss": 1.3611,
+      "grad_norm": 22.25,
+      "learning_rate": 4.878947368421053e-06,
+      "loss": 1.2238,
       "step": 6910
     },
     {
       "epoch": 0.35,
-      "grad_norm": 7.15625,
-      "learning_rate": 4.8631578947368415e-05,
-      "loss": 1.6587,
+      "grad_norm": 23.75,
+      "learning_rate": 4.863157894736842e-06,
+      "loss": 1.5511,
       "step": 6920
     },
     {
       "epoch": 0.35,
-      "grad_norm": 5.78125,
-      "learning_rate": 4.847368421052631e-05,
-      "loss": 1.4203,
+      "grad_norm": 14.1875,
+      "learning_rate": 4.847368421052631e-06,
+      "loss": 1.3549,
       "step": 6930
     },
     {
       "epoch": 0.35,
-      "grad_norm": 6.0,
-      "learning_rate": 4.831578947368421e-05,
-      "loss": 1.3195,
+      "grad_norm": 22.875,
+      "learning_rate": 4.831578947368421e-06,
+      "loss": 1.2673,
       "step": 6940
     },
     {
       "epoch": 0.35,
-      "grad_norm": 6.0625,
-      "learning_rate": 4.81578947368421e-05,
-      "loss": 1.5523,
+      "grad_norm": 19.0,
+      "learning_rate": 4.81578947368421e-06,
+      "loss": 1.4216,
       "step": 6950
     },
     {
       "epoch": 0.35,
-      "grad_norm": 6.0,
-      "learning_rate": 4.7999999999999994e-05,
-      "loss": 1.2051,
+      "grad_norm": 17.875,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.1628,
       "step": 6960
     },
     {
       "epoch": 0.35,
-      "grad_norm": 5.625,
-      "learning_rate": 4.784210526315789e-05,
-      "loss": 1.3228,
+      "grad_norm": 17.375,
+      "learning_rate": 4.78421052631579e-06,
+      "loss": 1.2581,
       "step": 6970
     },
     {
       "epoch": 0.35,
-      "grad_norm": 5.53125,
-      "learning_rate": 4.768421052631579e-05,
-      "loss": 1.4439,
+      "grad_norm": 18.375,
+      "learning_rate": 4.76842105263158e-06,
+      "loss": 1.3978,
       "step": 6980
     },
     {
       "epoch": 0.35,
-      "grad_norm": 4.875,
-      "learning_rate": 4.752631578947368e-05,
-      "loss": 1.2889,
+      "grad_norm": 20.25,
+      "learning_rate": 4.752631578947368e-06,
+      "loss": 1.1552,
       "step": 6990
     },
     {
       "epoch": 0.35,
-      "grad_norm": 5.78125,
-      "learning_rate": 4.7368421052631574e-05,
-      "loss": 1.6213,
+      "grad_norm": 19.75,
+      "learning_rate": 4.736842105263158e-06,
+      "loss": 1.5662,
       "step": 7000
     },
     {
       "epoch": 0.35,
-      "grad_norm": 6.8125,
-      "learning_rate": 4.721052631578947e-05,
-      "loss": 1.2401,
+      "grad_norm": 25.375,
+      "learning_rate": 4.721052631578947e-06,
+      "loss": 1.1584,
       "step": 7010
     },
     {
       "epoch": 0.35,
-      "grad_norm": 6.125,
-      "learning_rate": 4.705263157894737e-05,
-      "loss": 1.6266,
+      "grad_norm": 17.25,
+      "learning_rate": 4.705263157894737e-06,
+      "loss": 1.4341,
       "step": 7020
     },
     {
       "epoch": 0.35,
-      "grad_norm": 6.15625,
-      "learning_rate": 4.689473684210526e-05,
-      "loss": 1.3183,
+      "grad_norm": 33.5,
+      "learning_rate": 4.689473684210526e-06,
+      "loss": 1.3088,
       "step": 7030
     },
     {
       "epoch": 0.35,
-      "grad_norm": 4.71875,
-      "learning_rate": 4.6736842105263154e-05,
-      "loss": 1.4612,
+      "grad_norm": 17.5,
+      "learning_rate": 4.673684210526316e-06,
+      "loss": 1.4855,
       "step": 7040
     },
     {
       "epoch": 0.35,
-      "grad_norm": 7.8125,
-      "learning_rate": 4.657894736842105e-05,
-      "loss": 1.3345,
+      "grad_norm": 33.25,
+      "learning_rate": 4.657894736842106e-06,
+      "loss": 1.2726,
       "step": 7050
     },
     {
       "epoch": 0.35,
-      "grad_norm": 5.6875,
-      "learning_rate": 4.642105263157894e-05,
-      "loss": 1.3447,
+      "grad_norm": 24.75,
+      "learning_rate": 4.642105263157895e-06,
+      "loss": 1.2899,
       "step": 7060
     },
     {
       "epoch": 0.35,
-      "grad_norm": 6.53125,
-      "learning_rate": 4.626315789473684e-05,
-      "loss": 1.4236,
+      "grad_norm": 22.875,
+      "learning_rate": 4.626315789473684e-06,
+      "loss": 1.4073,
       "step": 7070
     },
     {
       "epoch": 0.35,
-      "grad_norm": 4.3125,
-      "learning_rate": 4.610526315789473e-05,
-      "loss": 0.9894,
+      "grad_norm": 16.5,
+      "learning_rate": 4.610526315789474e-06,
+      "loss": 0.8528,
       "step": 7080
     },
     {
       "epoch": 0.35,
-      "grad_norm": 4.9375,
-      "learning_rate": 4.594736842105263e-05,
-      "loss": 1.6438,
+      "grad_norm": 18.25,
+      "learning_rate": 4.594736842105263e-06,
+      "loss": 1.6042,
       "step": 7090
     },
     {
       "epoch": 0.35,
-      "grad_norm": 4.125,
-      "learning_rate": 4.578947368421052e-05,
-      "loss": 1.432,
+      "grad_norm": 21.625,
+      "learning_rate": 4.578947368421053e-06,
+      "loss": 1.3084,
       "step": 7100
     },
     {
       "epoch": 0.36,
-      "grad_norm": 8.0625,
-      "learning_rate": 4.5631578947368416e-05,
-      "loss": 1.4258,
+      "grad_norm": 31.5,
+      "learning_rate": 4.563157894736842e-06,
+      "loss": 1.3855,
       "step": 7110
     },
     {
       "epoch": 0.36,
-      "grad_norm": 5.03125,
-      "learning_rate": 4.547368421052631e-05,
-      "loss": 1.2104,
+      "grad_norm": 21.125,
+      "learning_rate": 4.547368421052632e-06,
+      "loss": 1.1414,
       "step": 7120
     },
     {
       "epoch": 0.36,
-      "grad_norm": 5.84375,
-      "learning_rate": 4.53157894736842e-05,
-      "loss": 1.5062,
+      "grad_norm": 18.5,
+      "learning_rate": 4.531578947368421e-06,
+      "loss": 1.4506,
       "step": 7130
     },
     {
       "epoch": 0.36,
-      "grad_norm": 4.4375,
-      "learning_rate": 4.51578947368421e-05,
-      "loss": 1.1665,
+      "grad_norm": 13.8125,
+      "learning_rate": 4.515789473684211e-06,
+      "loss": 1.1001,
       "step": 7140
     },
     {
       "epoch": 0.36,
-      "grad_norm": 2.25,
-      "learning_rate": 4.4999999999999996e-05,
-      "loss": 1.4222,
+      "grad_norm": 19.0,
+      "learning_rate": 4.5e-06,
+      "loss": 1.2032,
       "step": 7150
     },
     {
       "epoch": 0.36,
-      "grad_norm": 6.09375,
-      "learning_rate": 4.484210526315789e-05,
-      "loss": 1.2456,
+      "grad_norm": 24.25,
+      "learning_rate": 4.48421052631579e-06,
+      "loss": 1.2787,
       "step": 7160
     },
     {
       "epoch": 0.36,
-      "grad_norm": 9.5625,
-      "learning_rate": 4.468421052631579e-05,
-      "loss": 1.6563,
+      "grad_norm": 17.625,
+      "learning_rate": 4.468421052631579e-06,
+      "loss": 1.5579,
       "step": 7170
     },
     {
       "epoch": 0.36,
-      "grad_norm": 6.9375,
-      "learning_rate": 4.452631578947368e-05,
-      "loss": 1.4191,
+      "grad_norm": 28.5,
+      "learning_rate": 4.452631578947369e-06,
+      "loss": 1.3317,
       "step": 7180
     },
     {
       "epoch": 0.36,
-      "grad_norm": 5.125,
-      "learning_rate": 4.4368421052631576e-05,
-      "loss": 1.4417,
+      "grad_norm": 19.625,
+      "learning_rate": 4.436842105263158e-06,
+      "loss": 1.459,
       "step": 7190
     },
     {
       "epoch": 0.36,
-      "grad_norm": 5.34375,
-      "learning_rate": 4.4210526315789466e-05,
-      "loss": 1.2115,
+      "grad_norm": 25.5,
+      "learning_rate": 4.421052631578947e-06,
+      "loss": 1.1764,
       "step": 7200
     },
     {
       "epoch": 0.36,
-      "grad_norm": 6.0,
-      "learning_rate": 4.405263157894736e-05,
-      "loss": 1.5196,
+      "grad_norm": 19.5,
+      "learning_rate": 4.405263157894737e-06,
+      "loss": 1.5463,
       "step": 7210
     },
     {
       "epoch": 0.36,
-      "grad_norm": 5.375,
-      "learning_rate": 4.389473684210526e-05,
-      "loss": 1.4436,
+      "grad_norm": 22.25,
+      "learning_rate": 4.3894736842105266e-06,
+      "loss": 1.4262,
       "step": 7220
     },
     {
       "epoch": 0.36,
-      "grad_norm": 6.90625,
-      "learning_rate": 4.3736842105263155e-05,
-      "loss": 1.3976,
+      "grad_norm": 7.6875,
+      "learning_rate": 4.373684210526316e-06,
+      "loss": 1.2997,
       "step": 7230
     },
     {
       "epoch": 0.36,
-      "grad_norm": 7.8125,
-      "learning_rate": 4.357894736842105e-05,
-      "loss": 1.5432,
+      "grad_norm": 13.875,
+      "learning_rate": 4.3578947368421055e-06,
+      "loss": 1.5425,
       "step": 7240
     },
     {
       "epoch": 0.36,
-      "grad_norm": 3.234375,
-      "learning_rate": 4.342105263157895e-05,
-      "loss": 1.2969,
+      "grad_norm": 2.59375,
+      "learning_rate": 4.342105263157895e-06,
+      "loss": 1.2184,
       "step": 7250
     },
     {
       "epoch": 0.36,
-      "grad_norm": 6.8125,
-      "learning_rate": 4.326315789473684e-05,
-      "loss": 1.3246,
+      "grad_norm": 30.0,
+      "learning_rate": 4.3263157894736845e-06,
+      "loss": 1.2771,
       "step": 7260
     },
     {
       "epoch": 0.36,
-      "grad_norm": 9.8125,
-      "learning_rate": 4.3105263157894735e-05,
-      "loss": 1.3552,
+      "grad_norm": 20.125,
+      "learning_rate": 4.310526315789474e-06,
+      "loss": 1.3415,
       "step": 7270
     },
     {
       "epoch": 0.36,
-      "grad_norm": 3.84375,
-      "learning_rate": 4.2947368421052625e-05,
-      "loss": 1.1,
+      "grad_norm": 9.0625,
+      "learning_rate": 4.294736842105263e-06,
+      "loss": 1.0402,
       "step": 7280
     },
     {
       "epoch": 0.36,
-      "grad_norm": 6.25,
-      "learning_rate": 4.278947368421052e-05,
-      "loss": 1.449,
+      "grad_norm": 21.0,
+      "learning_rate": 4.278947368421052e-06,
+      "loss": 1.4532,
       "step": 7290
     },
     {
       "epoch": 0.36,
-      "grad_norm": 8.5,
-      "learning_rate": 4.263157894736842e-05,
-      "loss": 1.572,
+      "grad_norm": 25.75,
+      "learning_rate": 4.2631578947368425e-06,
+      "loss": 1.5666,
       "step": 7300
     },
     {
       "epoch": 0.37,
-      "grad_norm": 6.1875,
-      "learning_rate": 4.2473684210526315e-05,
-      "loss": 1.2444,
+      "grad_norm": 32.25,
+      "learning_rate": 4.247368421052632e-06,
+      "loss": 1.1727,
       "step": 7310
     },
     {
       "epoch": 0.37,
-      "grad_norm": 5.0,
-      "learning_rate": 4.231578947368421e-05,
-      "loss": 1.1988,
+      "grad_norm": 15.5625,
+      "learning_rate": 4.2315789473684215e-06,
+      "loss": 1.0616,
       "step": 7320
     },
     {
       "epoch": 0.37,
-      "grad_norm": 5.4375,
-      "learning_rate": 4.21578947368421e-05,
-      "loss": 1.6043,
+      "grad_norm": 19.5,
+      "learning_rate": 4.215789473684211e-06,
+      "loss": 1.4833,
       "step": 7330
     },
     {
       "epoch": 0.37,
-      "grad_norm": 5.3125,
-      "learning_rate": 4.2e-05,
-      "loss": 1.2237,
+      "grad_norm": 20.25,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 1.1481,
       "step": 7340
     },
     {
       "epoch": 0.37,
-      "grad_norm": 7.4375,
-      "learning_rate": 4.184210526315789e-05,
-      "loss": 1.6211,
+      "grad_norm": 25.25,
+      "learning_rate": 4.184210526315789e-06,
+      "loss": 1.5027,
       "step": 7350
     },
     {
       "epoch": 0.37,
-      "grad_norm": 5.28125,
-      "learning_rate": 4.1684210526315784e-05,
-      "loss": 1.5567,
+      "grad_norm": 17.75,
+      "learning_rate": 4.168421052631579e-06,
+      "loss": 1.4471,
       "step": 7360
     },
     {
       "epoch": 0.37,
-      "grad_norm": 7.375,
-      "learning_rate": 4.152631578947368e-05,
-      "loss": 1.3958,
+      "grad_norm": 31.625,
+      "learning_rate": 4.152631578947368e-06,
+      "loss": 1.2989,
       "step": 7370
     },
     {
       "epoch": 0.37,
-      "grad_norm": 5.5625,
-      "learning_rate": 4.136842105263158e-05,
-      "loss": 1.3257,
+      "grad_norm": 25.0,
+      "learning_rate": 4.136842105263158e-06,
+      "loss": 1.2379,
       "step": 7380
     },
     {
       "epoch": 0.37,
-      "grad_norm": 8.875,
-      "learning_rate": 4.1210526315789474e-05,
-      "loss": 1.3897,
+      "grad_norm": 25.0,
+      "learning_rate": 4.121052631578948e-06,
+      "loss": 1.2729,
       "step": 7390
     },
     {
       "epoch": 0.37,
-      "grad_norm": 5.0,
-      "learning_rate": 4.105263157894737e-05,
-      "loss": 1.3825,
+      "grad_norm": 33.0,
+      "learning_rate": 4.105263157894737e-06,
+      "loss": 1.3349,
       "step": 7400
     },
     {
       "epoch": 0.37,
-      "grad_norm": 4.75,
-      "learning_rate": 4.089473684210526e-05,
-      "loss": 1.7153,
+      "grad_norm": 20.75,
+      "learning_rate": 4.089473684210527e-06,
+      "loss": 1.6087,
       "step": 7410
     },
     {
       "epoch": 0.37,
-      "grad_norm": 5.21875,
-      "learning_rate": 4.073684210526315e-05,
-      "loss": 1.3951,
+      "grad_norm": 18.25,
+      "learning_rate": 4.0736842105263155e-06,
+      "loss": 1.3135,
       "step": 7420
     },
     {
       "epoch": 0.37,
-      "grad_norm": 7.34375,
-      "learning_rate": 4.057894736842105e-05,
-      "loss": 1.7072,
+      "grad_norm": 24.125,
+      "learning_rate": 4.057894736842105e-06,
+      "loss": 1.5614,
       "step": 7430
     },
     {
       "epoch": 0.37,
-      "grad_norm": 4.8125,
-      "learning_rate": 4.0421052631578943e-05,
-      "loss": 1.3874,
+      "grad_norm": 20.5,
+      "learning_rate": 4.0421052631578945e-06,
+      "loss": 1.4008,
       "step": 7440
     },
     {
       "epoch": 0.37,
-      "grad_norm": 7.09375,
-      "learning_rate": 4.026315789473684e-05,
-      "loss": 1.3483,
+      "grad_norm": 25.0,
+      "learning_rate": 4.026315789473684e-06,
+      "loss": 1.2597,
       "step": 7450
     },
     {
       "epoch": 0.37,
-      "grad_norm": 10.0625,
-      "learning_rate": 4.0105263157894737e-05,
-      "loss": 1.3684,
+      "grad_norm": 25.5,
+      "learning_rate": 4.010526315789474e-06,
+      "loss": 1.2063,
       "step": 7460
     },
     {
       "epoch": 0.37,
-      "grad_norm": 6.03125,
-      "learning_rate": 3.994736842105263e-05,
-      "loss": 1.1862,
+      "grad_norm": 17.625,
+      "learning_rate": 3.994736842105264e-06,
+      "loss": 1.1178,
       "step": 7470
     },
     {
       "epoch": 0.37,
-      "grad_norm": 6.53125,
-      "learning_rate": 3.978947368421053e-05,
-      "loss": 1.8852,
+      "grad_norm": 47.25,
+      "learning_rate": 3.978947368421053e-06,
+      "loss": 1.7909,
       "step": 7480
     },
     {
       "epoch": 0.37,
-      "grad_norm": 5.5,
-      "learning_rate": 3.963157894736841e-05,
-      "loss": 1.5302,
+      "grad_norm": 20.5,
+      "learning_rate": 3.963157894736842e-06,
+      "loss": 1.5256,
       "step": 7490
     },
     {
       "epoch": 0.38,
-      "grad_norm": 6.1875,
-      "learning_rate": 3.947368421052631e-05,
-      "loss": 1.2026,
+      "grad_norm": 17.75,
+      "learning_rate": 3.9473684210526315e-06,
+      "loss": 1.14,
       "step": 7500
     },
     {
       "epoch": 0.38,
-      "grad_norm": 5.5,
-      "learning_rate": 3.9315789473684206e-05,
-      "loss": 1.3831,
+      "grad_norm": 20.625,
+      "learning_rate": 3.931578947368421e-06,
+      "loss": 1.2839,
       "step": 7510
     },
     {
       "epoch": 0.38,
-      "grad_norm": 6.5625,
-      "learning_rate": 3.91578947368421e-05,
-      "loss": 1.4271,
+      "grad_norm": 22.125,
+      "learning_rate": 3.9157894736842104e-06,
+      "loss": 1.3771,
       "step": 7520
     },
     {
       "epoch": 0.38,
-      "grad_norm": 4.5,
-      "learning_rate": 3.9e-05,
-      "loss": 1.2867,
+      "grad_norm": 10.625,
+      "learning_rate": 3.9e-06,
+      "loss": 1.3,
       "step": 7530
     },
     {
       "epoch": 0.38,
-      "grad_norm": 2.828125,
-      "learning_rate": 3.8842105263157896e-05,
-      "loss": 0.9484,
+      "grad_norm": 11.0,
+      "learning_rate": 3.884210526315789e-06,
+      "loss": 0.8222,
       "step": 7540
     },
     {
       "epoch": 0.38,
-      "grad_norm": 8.875,
-      "learning_rate": 3.868421052631579e-05,
-      "loss": 1.4929,
+      "grad_norm": 16.375,
+      "learning_rate": 3.86842105263158e-06,
+      "loss": 1.5128,
       "step": 7550
     },
     {
       "epoch": 0.38,
-      "grad_norm": 4.09375,
-      "learning_rate": 3.8526315789473676e-05,
-      "loss": 1.4549,
+      "grad_norm": 12.0625,
+      "learning_rate": 3.852631578947368e-06,
+      "loss": 1.4502,
       "step": 7560
     },
     {
       "epoch": 0.38,
-      "grad_norm": 5.75,
-      "learning_rate": 3.836842105263157e-05,
-      "loss": 1.1515,
+      "grad_norm": 16.375,
+      "learning_rate": 3.836842105263158e-06,
+      "loss": 1.0498,
       "step": 7570
     },
     {
       "epoch": 0.38,
-      "grad_norm": 5.3125,
-      "learning_rate": 3.821052631578947e-05,
-      "loss": 1.3083,
+      "grad_norm": 16.5,
+      "learning_rate": 3.821052631578947e-06,
+      "loss": 1.2714,
       "step": 7580
     },
     {
       "epoch": 0.38,
-      "grad_norm": 5.8125,
-      "learning_rate": 3.8052631578947365e-05,
-      "loss": 1.1697,
+      "grad_norm": 24.5,
+      "learning_rate": 3.805263157894737e-06,
+      "loss": 1.0995,
       "step": 7590
     },
     {
       "epoch": 0.38,
-      "grad_norm": 5.0625,
-      "learning_rate": 3.789473684210526e-05,
-      "loss": 1.4502,
+      "grad_norm": 31.5,
+      "learning_rate": 3.7894736842105264e-06,
+      "loss": 1.3249,
       "step": 7600
     },
     {
       "epoch": 0.38,
-      "grad_norm": 5.3125,
-      "learning_rate": 3.773684210526316e-05,
-      "loss": 1.4979,
+      "grad_norm": 21.5,
+      "learning_rate": 3.7736842105263163e-06,
+      "loss": 1.4243,
       "step": 7610
     },
     {
       "epoch": 0.38,
-      "grad_norm": 5.09375,
-      "learning_rate": 3.7578947368421055e-05,
-      "loss": 1.4675,
+      "grad_norm": 29.625,
+      "learning_rate": 3.7578947368421058e-06,
+      "loss": 1.4749,
       "step": 7620
     },
     {
       "epoch": 0.38,
-      "grad_norm": 7.1875,
-      "learning_rate": 3.7421052631578945e-05,
-      "loss": 1.3028,
+      "grad_norm": 31.625,
+      "learning_rate": 3.742105263157895e-06,
+      "loss": 1.3472,
       "step": 7630
     },
     {
       "epoch": 0.38,
-      "grad_norm": 4.71875,
-      "learning_rate": 3.726315789473684e-05,
-      "loss": 1.337,
+      "grad_norm": 13.125,
+      "learning_rate": 3.7263157894736843e-06,
+      "loss": 1.2162,
       "step": 7640
     },
     {
       "epoch": 0.38,
-      "grad_norm": 8.4375,
-      "learning_rate": 3.710526315789473e-05,
-      "loss": 1.5052,
+      "grad_norm": 23.125,
+      "learning_rate": 3.710526315789474e-06,
+      "loss": 1.539,
       "step": 7650
     },
     {
       "epoch": 0.38,
-      "grad_norm": 5.6875,
-      "learning_rate": 3.694736842105263e-05,
-      "loss": 1.4187,
+      "grad_norm": 21.625,
+      "learning_rate": 3.6947368421052633e-06,
+      "loss": 1.4024,
       "step": 7660
     },
     {
       "epoch": 0.38,
-      "grad_norm": 6.3125,
-      "learning_rate": 3.6789473684210525e-05,
-      "loss": 1.2143,
+      "grad_norm": 15.6875,
+      "learning_rate": 3.678947368421053e-06,
+      "loss": 1.2368,
       "step": 7670
     },
     {
       "epoch": 0.38,
-      "grad_norm": 7.8125,
-      "learning_rate": 3.6631578947368414e-05,
-      "loss": 1.4787,
+      "grad_norm": 43.0,
+      "learning_rate": 3.663157894736842e-06,
+      "loss": 1.3536,
       "step": 7680
     },
     {
       "epoch": 0.38,
-      "grad_norm": 5.34375,
-      "learning_rate": 3.647368421052631e-05,
-      "loss": 1.5673,
+      "grad_norm": 15.5625,
+      "learning_rate": 3.6473684210526318e-06,
+      "loss": 1.4893,
       "step": 7690
     },
     {
       "epoch": 0.39,
-      "grad_norm": 3.828125,
-      "learning_rate": 3.631578947368421e-05,
-      "loss": 1.2965,
+      "grad_norm": 12.25,
+      "learning_rate": 3.6315789473684213e-06,
+      "loss": 1.197,
       "step": 7700
     },
     {
       "epoch": 0.39,
-      "grad_norm": 5.46875,
-      "learning_rate": 3.6157894736842104e-05,
-      "loss": 1.3391,
+      "grad_norm": 23.375,
+      "learning_rate": 3.6157894736842108e-06,
+      "loss": 1.3298,
       "step": 7710
     },
     {
       "epoch": 0.39,
-      "grad_norm": 4.6875,
-      "learning_rate": 3.5999999999999994e-05,
-      "loss": 0.8979,
+      "grad_norm": 14.75,
+      "learning_rate": 3.6e-06,
+      "loss": 0.8856,
       "step": 7720
     },
     {
       "epoch": 0.39,
-      "grad_norm": 6.25,
-      "learning_rate": 3.584210526315789e-05,
-      "loss": 1.3974,
+      "grad_norm": 27.875,
+      "learning_rate": 3.5842105263157898e-06,
+      "loss": 1.3668,
       "step": 7730
     },
     {
       "epoch": 0.39,
-      "grad_norm": 4.53125,
-      "learning_rate": 3.568421052631579e-05,
-      "loss": 1.562,
+      "grad_norm": 13.0,
+      "learning_rate": 3.5684210526315792e-06,
+      "loss": 1.6462,
       "step": 7740
     },
     {
       "epoch": 0.39,
-      "grad_norm": 7.84375,
-      "learning_rate": 3.552631578947368e-05,
-      "loss": 1.4908,
+      "grad_norm": 19.625,
+      "learning_rate": 3.5526315789473683e-06,
+      "loss": 1.3835,
       "step": 7750
     },
     {
       "epoch": 0.39,
-      "grad_norm": 5.03125,
-      "learning_rate": 3.5368421052631574e-05,
-      "loss": 1.5491,
+      "grad_norm": 14.1875,
+      "learning_rate": 3.536842105263158e-06,
+      "loss": 1.3977,
       "step": 7760
     },
     {
       "epoch": 0.39,
-      "grad_norm": 7.59375,
-      "learning_rate": 3.521052631578947e-05,
-      "loss": 1.5424,
+      "grad_norm": 27.75,
+      "learning_rate": 3.5210526315789473e-06,
+      "loss": 1.5368,
       "step": 7770
     },
     {
       "epoch": 0.39,
-      "grad_norm": 5.0,
-      "learning_rate": 3.505263157894737e-05,
-      "loss": 1.7824,
+      "grad_norm": 12.25,
+      "learning_rate": 3.505263157894737e-06,
+      "loss": 1.781,
       "step": 7780
     },
     {
       "epoch": 0.39,
-      "grad_norm": 5.84375,
-      "learning_rate": 3.489473684210526e-05,
-      "loss": 1.5703,
+      "grad_norm": 22.5,
+      "learning_rate": 3.4894736842105263e-06,
+      "loss": 1.6009,
       "step": 7790
     },
     {
       "epoch": 0.39,
-      "grad_norm": 8.4375,
-      "learning_rate": 3.4736842105263153e-05,
-      "loss": 1.079,
+      "grad_norm": 45.25,
+      "learning_rate": 3.4736842105263158e-06,
+      "loss": 0.9749,
       "step": 7800
     },
     {
       "epoch": 0.39,
-      "grad_norm": 5.5,
-      "learning_rate": 3.457894736842105e-05,
-      "loss": 1.438,
+      "grad_norm": 23.875,
+      "learning_rate": 3.4578947368421053e-06,
+      "loss": 1.4825,
       "step": 7810
     },
     {
       "epoch": 0.39,
-      "grad_norm": 6.28125,
-      "learning_rate": 3.442105263157894e-05,
-      "loss": 1.5293,
+      "grad_norm": 17.125,
+      "learning_rate": 3.4421052631578947e-06,
+      "loss": 1.4344,
       "step": 7820
     },
     {
       "epoch": 0.39,
-      "grad_norm": 6.09375,
-      "learning_rate": 3.4263157894736836e-05,
-      "loss": 1.2791,
+      "grad_norm": 21.5,
+      "learning_rate": 3.4263157894736842e-06,
+      "loss": 1.2112,
       "step": 7830
     },
     {
       "epoch": 0.39,
-      "grad_norm": 4.71875,
-      "learning_rate": 3.410526315789473e-05,
-      "loss": 1.2439,
+      "grad_norm": 12.25,
+      "learning_rate": 3.4105263157894737e-06,
+      "loss": 1.1979,
       "step": 7840
     },
     {
       "epoch": 0.39,
-      "grad_norm": 3.140625,
-      "learning_rate": 3.394736842105263e-05,
-      "loss": 1.1694,
+      "grad_norm": 7.8125,
+      "learning_rate": 3.3947368421052632e-06,
+      "loss": 1.1706,
       "step": 7850
     },
     {
       "epoch": 0.39,
-      "grad_norm": 6.40625,
-      "learning_rate": 3.378947368421052e-05,
-      "loss": 1.5024,
+      "grad_norm": 10.8125,
+      "learning_rate": 3.3789473684210527e-06,
+      "loss": 1.4322,
       "step": 7860
     },
     {
       "epoch": 0.39,
-      "grad_norm": 5.5625,
-      "learning_rate": 3.3631578947368416e-05,
-      "loss": 1.2139,
+      "grad_norm": 25.875,
+      "learning_rate": 3.363157894736842e-06,
+      "loss": 1.2244,
       "step": 7870
     },
     {
       "epoch": 0.39,
-      "grad_norm": 6.9375,
-      "learning_rate": 3.347368421052631e-05,
-      "loss": 1.5477,
+      "grad_norm": 18.875,
+      "learning_rate": 3.3473684210526317e-06,
+      "loss": 1.4572,
       "step": 7880
     },
     {
       "epoch": 0.39,
-      "grad_norm": 5.625,
-      "learning_rate": 3.331578947368421e-05,
-      "loss": 1.4138,
+      "grad_norm": 20.375,
+      "learning_rate": 3.331578947368421e-06,
+      "loss": 1.3651,
       "step": 7890
     },
     {
       "epoch": 0.4,
-      "grad_norm": 6.8125,
-      "learning_rate": 3.31578947368421e-05,
-      "loss": 1.2124,
+      "grad_norm": 18.625,
+      "learning_rate": 3.3157894736842107e-06,
+      "loss": 1.1065,
       "step": 7900
     },
     {
       "epoch": 0.4,
-      "grad_norm": 9.6875,
-      "learning_rate": 3.2999999999999996e-05,
-      "loss": 1.3678,
+      "grad_norm": 41.75,
+      "learning_rate": 3.3e-06,
+      "loss": 1.3439,
       "step": 7910
     },
     {
       "epoch": 0.4,
-      "grad_norm": 3.703125,
-      "learning_rate": 3.284210526315789e-05,
-      "loss": 1.1943,
+      "grad_norm": 14.25,
+      "learning_rate": 3.2842105263157897e-06,
+      "loss": 1.2267,
       "step": 7920
     },
     {
       "epoch": 0.4,
-      "grad_norm": 10.5625,
-      "learning_rate": 3.268421052631578e-05,
-      "loss": 1.3115,
+      "grad_norm": 37.25,
+      "learning_rate": 3.2684210526315787e-06,
+      "loss": 1.2631,
       "step": 7930
     },
     {
       "epoch": 0.4,
-      "grad_norm": 7.5,
-      "learning_rate": 3.252631578947368e-05,
-      "loss": 1.3972,
+      "grad_norm": 19.125,
+      "learning_rate": 3.2526315789473686e-06,
+      "loss": 1.3705,
       "step": 7940
     },
     {
       "epoch": 0.4,
-      "grad_norm": 5.21875,
-      "learning_rate": 3.2368421052631575e-05,
-      "loss": 1.1899,
+      "grad_norm": 19.75,
+      "learning_rate": 3.236842105263158e-06,
+      "loss": 1.0897,
       "step": 7950
     },
     {
       "epoch": 0.4,
-      "grad_norm": 8.125,
-      "learning_rate": 3.221052631578947e-05,
-      "loss": 1.439,
+      "grad_norm": 25.125,
+      "learning_rate": 3.2210526315789476e-06,
+      "loss": 1.3476,
       "step": 7960
     },
     {
       "epoch": 0.4,
-      "grad_norm": 6.09375,
-      "learning_rate": 3.205263157894736e-05,
-      "loss": 1.55,
+      "grad_norm": 19.75,
+      "learning_rate": 3.2052631578947367e-06,
+      "loss": 1.5058,
       "step": 7970
     },
     {
       "epoch": 0.4,
-      "grad_norm": 6.625,
-      "learning_rate": 3.189473684210526e-05,
-      "loss": 1.6196,
+      "grad_norm": 23.625,
+      "learning_rate": 3.1894736842105266e-06,
+      "loss": 1.6535,
       "step": 7980
     },
     {
       "epoch": 0.4,
-      "grad_norm": 7.96875,
-      "learning_rate": 3.1736842105263155e-05,
-      "loss": 1.1113,
+      "grad_norm": 11.5625,
+      "learning_rate": 3.173684210526316e-06,
+      "loss": 1.0905,
       "step": 7990
     },
     {
       "epoch": 0.4,
-      "grad_norm": 5.1875,
-      "learning_rate": 3.1578947368421045e-05,
-      "loss": 1.3067,
+      "grad_norm": 20.125,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 1.2503,
       "step": 8000
     },
     {
       "epoch": 0.4,
-      "eval_loss": 1.3947522640228271,
-      "eval_runtime": 30.8209,
-      "eval_samples_per_second": 32.445,
-      "eval_steps_per_second": 32.445,
+      "eval_loss": 1.3293771743774414,
+      "eval_runtime": 30.7768,
+      "eval_samples_per_second": 32.492,
+      "eval_steps_per_second": 32.492,
       "step": 8000
     }
   ],