Uploaded checkpoint-2000

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:97fd931ee1c953569f7578d2073bcb0dfd81e58de8e1c6e9f96f29e86ff0f1e4
+oid sha256:8329e3ca51818b885f32c67069e26927d537402948545a07b81991ace8ab70f2
 size 2836579040

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e7e8808d3d8b3481772a6218e0912adbff479415b7aab2f6b7b86db5f2bda0c4
+oid sha256:367af0856d06ed99fd2d97d7b3f62d67c5116d32036644f9fc9edfd2c59753fe
 size 5673376169

scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8ee74dc482c903af9fba39ea4b4791b513313a76c7f1d1d5b626abcc8f5556d0
+oid sha256:7e843611c4040f67481738e5b72625984afb6b26e66c8ec6bac2dab49b2072a4
 size 1064

trainer_state.json

@@ -10,1410 +10,1410 @@
   "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
     }
   ],

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0abc2d5a3f8110dd335d09eeee1d3a6bffae4cf678060f558ef1576fbb52f9b8
+oid sha256:8cc9b3ae127809c9aeb5eb3b39b51c0769ec1f142431988037c307a984d56968
 size 4856