Training in progress, epoch 1, checkpoint

checkpoint-490/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1d0d305fe7d18469c699f8af09c6de8da15fc09dd157923c393d45244eee618a
+oid sha256:0835750c45a0fdc412ccb9446101231b5b9dcd0acaf48bffe9ee381cbc7a3f8e
 size 358419776

checkpoint-490/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e024c7520678d099db15655cafdb1a32b8075d7cec9982d084593288c14dee31
+oid sha256:7f6128b40a187ee43372539f6ae675199c9b67ce97305cb3bcdd36ab6bdd8fa3
 size 707744314

checkpoint-490/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e98eae4f9ae907154431d389ab6aacd1b74927fc40870899dd45e9190a76fa7c
+oid sha256:fa339e69c89f04f664bc5b5af50ad45cd97bd253a6b751985062b9e53256a30b
 size 14244

checkpoint-490/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8cecb4841553cbca36671362c8184bc7ac16a241bbc226dd8692e56cbe168e94
+oid sha256:3e7b81709c289c6baefb02b805f0934bac2ee2cf328fa78aef2e50bcf44e76cb
 size 1064

checkpoint-490/trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.8470824949698189,
+  "best_metric": 0.8883299798792756,
   "best_model_checkpoint": "PhoWhisper-small-vispeech-classifier-v4/checkpoint-490",
   "epoch": 1.0,
   "eval_steps": 500,
@@ -10,697 +10,697 @@
   "log_history": [
     {
       "epoch": 0.01020408163265306,
-      "grad_norm": 56193.578125,
-      "learning_rate": 2.5510204081632652e-08,
-      "loss": 2.0798,
+      "grad_norm": 51640.66015625,
+      "learning_rate": 5.1020408163265303e-08,
+      "loss": 2.0778,
       "step": 5
     },
     {
       "epoch": 0.02040816326530612,
-      "grad_norm": 50520.33203125,
-      "learning_rate": 5.1020408163265303e-08,
-      "loss": 2.0772,
+      "grad_norm": 46773.35546875,
+      "learning_rate": 1.0204081632653061e-07,
+      "loss": 2.0809,
       "step": 10
     },
     {
       "epoch": 0.030612244897959183,
-      "grad_norm": 47583.30859375,
-      "learning_rate": 7.653061224489796e-08,
-      "loss": 2.0762,
+      "grad_norm": 45343.12109375,
+      "learning_rate": 1.5306122448979592e-07,
+      "loss": 2.0771,
       "step": 15
     },
     {
       "epoch": 0.04081632653061224,
-      "grad_norm": 53174.3828125,
-      "learning_rate": 1.0204081632653061e-07,
-      "loss": 2.0754,
+      "grad_norm": 49285.91796875,
+      "learning_rate": 2.0408163265306121e-07,
+      "loss": 2.077,
       "step": 20
     },
     {
       "epoch": 0.05102040816326531,
-      "grad_norm": 65972.65625,
-      "learning_rate": 1.2755102040816328e-07,
+      "grad_norm": 63271.82421875,
+      "learning_rate": 2.5510204081632656e-07,
       "loss": 2.0755,
       "step": 25
     },
     {
       "epoch": 0.061224489795918366,
-      "grad_norm": 53920.14453125,
-      "learning_rate": 1.5306122448979592e-07,
-      "loss": 2.0745,
+      "grad_norm": 51649.79296875,
+      "learning_rate": 3.0612244897959183e-07,
+      "loss": 2.0764,
       "step": 30
     },
     {
       "epoch": 0.07142857142857142,
-      "grad_norm": 52966.375,
-      "learning_rate": 1.7857142857142858e-07,
-      "loss": 2.0734,
+      "grad_norm": 51633.98828125,
+      "learning_rate": 3.5714285714285716e-07,
+      "loss": 2.0763,
       "step": 35
     },
     {
       "epoch": 0.08163265306122448,
-      "grad_norm": 60549.94921875,
-      "learning_rate": 2.0408163265306121e-07,
-      "loss": 2.0778,
+      "grad_norm": 61777.23828125,
+      "learning_rate": 4.0816326530612243e-07,
+      "loss": 2.0772,
       "step": 40
     },
     {
       "epoch": 0.09183673469387756,
-      "grad_norm": 55827.60546875,
-      "learning_rate": 2.295918367346939e-07,
-      "loss": 2.0787,
+      "grad_norm": 58467.15625,
+      "learning_rate": 4.591836734693878e-07,
+      "loss": 2.0743,
       "step": 45
     },
     {
       "epoch": 0.10204081632653061,
-      "grad_norm": 51334.3359375,
-      "learning_rate": 2.5510204081632656e-07,
-      "loss": 2.0759,
+      "grad_norm": 48974.37890625,
+      "learning_rate": 5.102040816326531e-07,
+      "loss": 2.072,
       "step": 50
     },
     {
       "epoch": 0.11224489795918367,
-      "grad_norm": 70249.5078125,
-      "learning_rate": 2.806122448979592e-07,
-      "loss": 2.0698,
+      "grad_norm": 71007.515625,
+      "learning_rate": 5.612244897959184e-07,
+      "loss": 2.0751,
       "step": 55
     },
     {
       "epoch": 0.12244897959183673,
-      "grad_norm": 54763.26171875,
-      "learning_rate": 3.0612244897959183e-07,
-      "loss": 2.0762,
+      "grad_norm": 58237.77734375,
+      "learning_rate": 6.122448979591837e-07,
+      "loss": 2.0702,
       "step": 60
     },
     {
       "epoch": 0.1326530612244898,
-      "grad_norm": 58244.69921875,
-      "learning_rate": 3.3163265306122455e-07,
-      "loss": 2.0696,
+      "grad_norm": 58785.83203125,
+      "learning_rate": 6.632653061224491e-07,
+      "loss": 2.0713,
       "step": 65
     },
     {
       "epoch": 0.14285714285714285,
-      "grad_norm": 51894.46484375,
-      "learning_rate": 3.5714285714285716e-07,
-      "loss": 2.0702,
+      "grad_norm": 52771.265625,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": 2.069,
       "step": 70
     },
     {
       "epoch": 0.15306122448979592,
-      "grad_norm": 72845.734375,
-      "learning_rate": 3.826530612244898e-07,
-      "loss": 2.0676,
+      "grad_norm": 69647.078125,
+      "learning_rate": 7.653061224489796e-07,
+      "loss": 2.0684,
       "step": 75
     },
     {
       "epoch": 0.16326530612244897,
-      "grad_norm": 56673.85546875,
-      "learning_rate": 4.0816326530612243e-07,
-      "loss": 2.0693,
+      "grad_norm": 59148.47265625,
+      "learning_rate": 8.163265306122449e-07,
+      "loss": 2.0634,
       "step": 80
     },
     {
       "epoch": 0.17346938775510204,
-      "grad_norm": 61739.0546875,
-      "learning_rate": 4.3367346938775514e-07,
-      "loss": 2.0668,
+      "grad_norm": 66723.921875,
+      "learning_rate": 8.673469387755103e-07,
+      "loss": 2.064,
       "step": 85
     },
     {
       "epoch": 0.1836734693877551,
-      "grad_norm": 68551.390625,
-      "learning_rate": 4.591836734693878e-07,
-      "loss": 2.0651,
+      "grad_norm": 72545.203125,
+      "learning_rate": 9.183673469387756e-07,
+      "loss": 2.0593,
       "step": 90
     },
     {
       "epoch": 0.19387755102040816,
-      "grad_norm": 55835.78125,
-      "learning_rate": 4.846938775510205e-07,
-      "loss": 2.0701,
+      "grad_norm": 59112.55859375,
+      "learning_rate": 9.69387755102041e-07,
+      "loss": 2.0553,
       "step": 95
     },
     {
       "epoch": 0.20408163265306123,
-      "grad_norm": 60520.2734375,
-      "learning_rate": 5.102040816326531e-07,
-      "loss": 2.0654,
+      "grad_norm": 68369.265625,
+      "learning_rate": 1.0204081632653063e-06,
+      "loss": 2.0566,
       "step": 100
     },
     {
       "epoch": 0.21428571428571427,
-      "grad_norm": 45642.94921875,
-      "learning_rate": 5.357142857142857e-07,
-      "loss": 2.0672,
+      "grad_norm": 51080.62890625,
+      "learning_rate": 1.0714285714285714e-06,
+      "loss": 2.0505,
       "step": 105
     },
     {
       "epoch": 0.22448979591836735,
-      "grad_norm": 67603.8515625,
-      "learning_rate": 5.612244897959184e-07,
-      "loss": 2.0673,
+      "grad_norm": 78760.0859375,
+      "learning_rate": 1.122448979591837e-06,
+      "loss": 2.0496,
       "step": 110
     },
     {
       "epoch": 0.23469387755102042,
-      "grad_norm": 54359.4375,
-      "learning_rate": 5.867346938775511e-07,
-      "loss": 2.0604,
+      "grad_norm": 57198.7890625,
+      "learning_rate": 1.1734693877551022e-06,
+      "loss": 2.0458,
       "step": 115
     },
     {
       "epoch": 0.24489795918367346,
-      "grad_norm": 81289.5390625,
-      "learning_rate": 6.122448979591837e-07,
-      "loss": 2.061,
+      "grad_norm": 98369.4609375,
+      "learning_rate": 1.2244897959183673e-06,
+      "loss": 2.0336,
       "step": 120
     },
     {
       "epoch": 0.25510204081632654,
-      "grad_norm": 70781.5703125,
-      "learning_rate": 6.377551020408164e-07,
-      "loss": 2.0573,
+      "grad_norm": 80801.5703125,
+      "learning_rate": 1.2755102040816329e-06,
+      "loss": 2.0354,
       "step": 125
     },
     {
       "epoch": 0.2653061224489796,
-      "grad_norm": 73178.1875,
-      "learning_rate": 6.632653061224491e-07,
-      "loss": 2.0541,
+      "grad_norm": 82179.9765625,
+      "learning_rate": 1.3265306122448982e-06,
+      "loss": 2.0263,
       "step": 130
     },
     {
       "epoch": 0.2755102040816326,
-      "grad_norm": 69632.25,
-      "learning_rate": 6.887755102040817e-07,
-      "loss": 2.0603,
+      "grad_norm": 83225.078125,
+      "learning_rate": 1.3775510204081633e-06,
+      "loss": 2.0168,
       "step": 135
     },
     {
       "epoch": 0.2857142857142857,
-      "grad_norm": 54642.47265625,
-      "learning_rate": 7.142857142857143e-07,
-      "loss": 2.0587,
+      "grad_norm": 62451.64453125,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 2.023,
       "step": 140
     },
     {
       "epoch": 0.29591836734693877,
-      "grad_norm": 64884.2890625,
-      "learning_rate": 7.39795918367347e-07,
-      "loss": 2.0574,
+      "grad_norm": 75567.9453125,
+      "learning_rate": 1.479591836734694e-06,
+      "loss": 2.0206,
       "step": 145
     },
     {
       "epoch": 0.30612244897959184,
-      "grad_norm": 74709.9609375,
-      "learning_rate": 7.653061224489796e-07,
-      "loss": 2.0466,
+      "grad_norm": 98461.484375,
+      "learning_rate": 1.5306122448979593e-06,
+      "loss": 1.9984,
       "step": 150
     },
     {
       "epoch": 0.3163265306122449,
-      "grad_norm": 59361.31640625,
-      "learning_rate": 7.908163265306124e-07,
-      "loss": 2.0473,
+      "grad_norm": 75078.1484375,
+      "learning_rate": 1.5816326530612248e-06,
+      "loss": 1.9899,
       "step": 155
     },
     {
       "epoch": 0.32653061224489793,
-      "grad_norm": 76060.171875,
-      "learning_rate": 8.163265306122449e-07,
-      "loss": 2.0413,
+      "grad_norm": 97895.671875,
+      "learning_rate": 1.6326530612244897e-06,
+      "loss": 1.9758,
       "step": 160
     },
     {
       "epoch": 0.336734693877551,
-      "grad_norm": 52821.26953125,
-      "learning_rate": 8.418367346938776e-07,
-      "loss": 2.0495,
+      "grad_norm": 65880.2421875,
+      "learning_rate": 1.6836734693877552e-06,
+      "loss": 1.9918,
       "step": 165
     },
     {
       "epoch": 0.3469387755102041,
-      "grad_norm": 55647.62109375,
-      "learning_rate": 8.673469387755103e-07,
-      "loss": 2.0361,
+      "grad_norm": 70675.578125,
+      "learning_rate": 1.7346938775510206e-06,
+      "loss": 1.9715,
       "step": 170
     },
     {
       "epoch": 0.35714285714285715,
-      "grad_norm": 49648.828125,
-      "learning_rate": 8.928571428571429e-07,
-      "loss": 2.0395,
+      "grad_norm": 64327.19921875,
+      "learning_rate": 1.7857142857142859e-06,
+      "loss": 1.9597,
       "step": 175
     },
     {
       "epoch": 0.3673469387755102,
-      "grad_norm": 63527.9140625,
-      "learning_rate": 9.183673469387756e-07,
-      "loss": 2.0334,
+      "grad_norm": 80610.6171875,
+      "learning_rate": 1.8367346938775512e-06,
+      "loss": 1.9632,
       "step": 180
     },
     {
       "epoch": 0.37755102040816324,
-      "grad_norm": 80417.828125,
-      "learning_rate": 9.438775510204082e-07,
-      "loss": 2.0331,
+      "grad_norm": 110899.375,
+      "learning_rate": 1.8877551020408163e-06,
+      "loss": 1.9476,
       "step": 185
     },
     {
       "epoch": 0.3877551020408163,
-      "grad_norm": 82782.7265625,
-      "learning_rate": 9.69387755102041e-07,
-      "loss": 2.0251,
+      "grad_norm": 111132.5703125,
+      "learning_rate": 1.938775510204082e-06,
+      "loss": 1.9369,
       "step": 190
     },
     {
       "epoch": 0.3979591836734694,
-      "grad_norm": 66964.8125,
-      "learning_rate": 9.948979591836735e-07,
-      "loss": 2.0289,
+      "grad_norm": 88722.625,
+      "learning_rate": 1.989795918367347e-06,
+      "loss": 1.9452,
       "step": 195
     },
     {
       "epoch": 0.40816326530612246,
-      "grad_norm": 73848.046875,
-      "learning_rate": 1.0204081632653063e-06,
-      "loss": 2.0259,
+      "grad_norm": 105874.296875,
+      "learning_rate": 2.0408163265306125e-06,
+      "loss": 1.9459,
       "step": 200
     },
     {
       "epoch": 0.41836734693877553,
-      "grad_norm": 70241.171875,
-      "learning_rate": 1.0459183673469388e-06,
-      "loss": 2.0082,
+      "grad_norm": 97286.0078125,
+      "learning_rate": 2.0918367346938776e-06,
+      "loss": 1.918,
       "step": 205
     },
     {
       "epoch": 0.42857142857142855,
-      "grad_norm": 63360.8125,
-      "learning_rate": 1.0714285714285714e-06,
-      "loss": 2.0091,
+      "grad_norm": 82848.0625,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 1.9135,
       "step": 210
     },
     {
       "epoch": 0.4387755102040816,
-      "grad_norm": 82052.796875,
-      "learning_rate": 1.0969387755102041e-06,
-      "loss": 2.0008,
+      "grad_norm": 114854.4609375,
+      "learning_rate": 2.1938775510204083e-06,
+      "loss": 1.8869,
       "step": 215
     },
     {
       "epoch": 0.4489795918367347,
-      "grad_norm": 63768.30078125,
-      "learning_rate": 1.122448979591837e-06,
-      "loss": 1.9958,
+      "grad_norm": 86661.25,
+      "learning_rate": 2.244897959183674e-06,
+      "loss": 1.8861,
       "step": 220
     },
     {
       "epoch": 0.45918367346938777,
-      "grad_norm": 90230.640625,
-      "learning_rate": 1.1479591836734695e-06,
-      "loss": 1.9816,
+      "grad_norm": 127591.3984375,
+      "learning_rate": 2.295918367346939e-06,
+      "loss": 1.8396,
       "step": 225
     },
     {
       "epoch": 0.46938775510204084,
-      "grad_norm": 48253.55078125,
-      "learning_rate": 1.1734693877551022e-06,
-      "loss": 2.0064,
+      "grad_norm": 71038.8671875,
+      "learning_rate": 2.3469387755102044e-06,
+      "loss": 1.8995,
       "step": 230
     },
     {
       "epoch": 0.47959183673469385,
-      "grad_norm": 73846.0859375,
-      "learning_rate": 1.1989795918367348e-06,
-      "loss": 1.996,
+      "grad_norm": 101402.421875,
+      "learning_rate": 2.3979591836734696e-06,
+      "loss": 1.8699,
       "step": 235
     },
     {
       "epoch": 0.4897959183673469,
-      "grad_norm": 86008.8046875,
-      "learning_rate": 1.2244897959183673e-06,
-      "loss": 1.9858,
+      "grad_norm": 113157.3828125,
+      "learning_rate": 2.4489795918367347e-06,
+      "loss": 1.8599,
       "step": 240
     },
     {
       "epoch": 0.5,
-      "grad_norm": 86437.734375,
-      "learning_rate": 1.25e-06,
-      "loss": 1.9696,
+      "grad_norm": 115485.0546875,
+      "learning_rate": 2.5e-06,
+      "loss": 1.8214,
       "step": 245
     },
     {
       "epoch": 0.5102040816326531,
-      "grad_norm": 82227.921875,
-      "learning_rate": 1.2755102040816329e-06,
-      "loss": 1.9803,
+      "grad_norm": 139369.8125,
+      "learning_rate": 2.5510204081632657e-06,
+      "loss": 1.8539,
       "step": 250
     },
     {
       "epoch": 0.5204081632653061,
-      "grad_norm": 96575.4765625,
-      "learning_rate": 1.3010204081632654e-06,
-      "loss": 1.9375,
+      "grad_norm": 141401.46875,
+      "learning_rate": 2.602040816326531e-06,
+      "loss": 1.7507,
       "step": 255
     },
     {
       "epoch": 0.5306122448979592,
-      "grad_norm": 89167.7265625,
-      "learning_rate": 1.3265306122448982e-06,
-      "loss": 1.9585,
+      "grad_norm": 125767.3984375,
+      "learning_rate": 2.6530612244897964e-06,
+      "loss": 1.7948,
       "step": 260
     },
     {
       "epoch": 0.5408163265306123,
-      "grad_norm": 101032.5,
-      "learning_rate": 1.3520408163265307e-06,
-      "loss": 1.947,
+      "grad_norm": 148848.5625,
+      "learning_rate": 2.7040816326530615e-06,
+      "loss": 1.7871,
       "step": 265
     },
     {
       "epoch": 0.5510204081632653,
-      "grad_norm": 105424.25,
-      "learning_rate": 1.3775510204081633e-06,
-      "loss": 1.9482,
+      "grad_norm": 152509.4375,
+      "learning_rate": 2.7551020408163266e-06,
+      "loss": 1.8054,
       "step": 270
     },
     {
       "epoch": 0.5612244897959183,
-      "grad_norm": 84136.484375,
-      "learning_rate": 1.4030612244897959e-06,
-      "loss": 1.9386,
+      "grad_norm": 114681.078125,
+      "learning_rate": 2.8061224489795917e-06,
+      "loss": 1.7711,
       "step": 275
     },
     {
       "epoch": 0.5714285714285714,
-      "grad_norm": 83034.078125,
-      "learning_rate": 1.4285714285714286e-06,
-      "loss": 1.9245,
+      "grad_norm": 119104.984375,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.7139,
       "step": 280
     },
     {
       "epoch": 0.5816326530612245,
-      "grad_norm": 89200.4609375,
-      "learning_rate": 1.4540816326530614e-06,
-      "loss": 1.9138,
+      "grad_norm": 123160.5,
+      "learning_rate": 2.908163265306123e-06,
+      "loss": 1.703,
       "step": 285
     },
     {
       "epoch": 0.5918367346938775,
-      "grad_norm": 103667.9609375,
-      "learning_rate": 1.479591836734694e-06,
-      "loss": 1.9283,
+      "grad_norm": 157218.203125,
+      "learning_rate": 2.959183673469388e-06,
+      "loss": 1.7685,
       "step": 290
     },
     {
       "epoch": 0.6020408163265306,
-      "grad_norm": 98443.2734375,
-      "learning_rate": 1.5051020408163267e-06,
-      "loss": 1.9025,
+      "grad_norm": 134804.828125,
+      "learning_rate": 3.0102040816326534e-06,
+      "loss": 1.7053,
       "step": 295
     },
     {
       "epoch": 0.6122448979591837,
-      "grad_norm": 99525.0,
-      "learning_rate": 1.5306122448979593e-06,
-      "loss": 1.8933,
+      "grad_norm": 153977.625,
+      "learning_rate": 3.0612244897959185e-06,
+      "loss": 1.6711,
       "step": 300
     },
     {
       "epoch": 0.6224489795918368,
-      "grad_norm": 145139.5625,
-      "learning_rate": 1.556122448979592e-06,
-      "loss": 1.8563,
+      "grad_norm": 204690.125,
+      "learning_rate": 3.112244897959184e-06,
+      "loss": 1.5873,
       "step": 305
     },
     {
       "epoch": 0.6326530612244898,
-      "grad_norm": 116158.859375,
-      "learning_rate": 1.5816326530612248e-06,
-      "loss": 1.8931,
+      "grad_norm": 158604.1875,
+      "learning_rate": 3.1632653061224496e-06,
+      "loss": 1.6642,
       "step": 310
     },
     {
       "epoch": 0.6428571428571429,
-      "grad_norm": 131590.09375,
-      "learning_rate": 1.6071428571428574e-06,
-      "loss": 1.8599,
+      "grad_norm": 181877.484375,
+      "learning_rate": 3.2142857142857147e-06,
+      "loss": 1.6204,
       "step": 315
     },
     {
       "epoch": 0.6530612244897959,
-      "grad_norm": 112483.40625,
-      "learning_rate": 1.6326530612244897e-06,
-      "loss": 1.8596,
+      "grad_norm": 157735.734375,
+      "learning_rate": 3.2653061224489794e-06,
+      "loss": 1.6152,
       "step": 320
     },
     {
       "epoch": 0.6632653061224489,
-      "grad_norm": 82979.0859375,
-      "learning_rate": 1.6581632653061225e-06,
-      "loss": 1.883,
+      "grad_norm": 121935.3828125,
+      "learning_rate": 3.316326530612245e-06,
+      "loss": 1.6357,
       "step": 325
     },
     {
       "epoch": 0.673469387755102,
-      "grad_norm": 96548.5703125,
-      "learning_rate": 1.6836734693877552e-06,
-      "loss": 1.8586,
+      "grad_norm": 133579.859375,
+      "learning_rate": 3.3673469387755105e-06,
+      "loss": 1.6085,
       "step": 330
     },
     {
       "epoch": 0.6836734693877551,
-      "grad_norm": 113625.171875,
-      "learning_rate": 1.7091836734693878e-06,
-      "loss": 1.8375,
+      "grad_norm": 154886.953125,
+      "learning_rate": 3.4183673469387756e-06,
+      "loss": 1.5609,
       "step": 335
     },
     {
       "epoch": 0.6938775510204082,
-      "grad_norm": 115079.0078125,
-      "learning_rate": 1.7346938775510206e-06,
-      "loss": 1.838,
+      "grad_norm": 151655.0625,
+      "learning_rate": 3.469387755102041e-06,
+      "loss": 1.5905,
       "step": 340
     },
     {
       "epoch": 0.7040816326530612,
-      "grad_norm": 139868.21875,
-      "learning_rate": 1.7602040816326531e-06,
-      "loss": 1.8013,
+      "grad_norm": 198471.609375,
+      "learning_rate": 3.5204081632653062e-06,
+      "loss": 1.5052,
       "step": 345
     },
     {
       "epoch": 0.7142857142857143,
-      "grad_norm": 126350.546875,
-      "learning_rate": 1.7857142857142859e-06,
-      "loss": 1.847,
+      "grad_norm": 191516.453125,
+      "learning_rate": 3.5714285714285718e-06,
+      "loss": 1.5798,
       "step": 350
     },
     {
       "epoch": 0.7244897959183674,
-      "grad_norm": 149690.375,
-      "learning_rate": 1.8112244897959187e-06,
-      "loss": 1.8203,
+      "grad_norm": 207511.296875,
+      "learning_rate": 3.6224489795918373e-06,
+      "loss": 1.5382,
       "step": 355
     },
     {
       "epoch": 0.7346938775510204,
-      "grad_norm": 132708.578125,
-      "learning_rate": 1.8367346938775512e-06,
-      "loss": 1.7931,
+      "grad_norm": 170820.609375,
+      "learning_rate": 3.6734693877551024e-06,
+      "loss": 1.485,
       "step": 360
     },
     {
       "epoch": 0.7448979591836735,
-      "grad_norm": 128203.3203125,
-      "learning_rate": 1.862244897959184e-06,
-      "loss": 1.7533,
+      "grad_norm": 183925.796875,
+      "learning_rate": 3.724489795918368e-06,
+      "loss": 1.427,
       "step": 365
     },
     {
       "epoch": 0.7551020408163265,
-      "grad_norm": 118367.4765625,
-      "learning_rate": 1.8877551020408163e-06,
-      "loss": 1.7662,
+      "grad_norm": 302829.75,
+      "learning_rate": 3.7755102040816327e-06,
+      "loss": 1.4242,
       "step": 370
     },
     {
       "epoch": 0.7653061224489796,
-      "grad_norm": 162099.25,
-      "learning_rate": 1.913265306122449e-06,
-      "loss": 1.7775,
+      "grad_norm": 292990.5625,
+      "learning_rate": 3.826530612244898e-06,
+      "loss": 1.484,
       "step": 375
     },
     {
       "epoch": 0.7755102040816326,
-      "grad_norm": 132369.234375,
-      "learning_rate": 1.938775510204082e-06,
-      "loss": 1.7908,
+      "grad_norm": 189517.421875,
+      "learning_rate": 3.877551020408164e-06,
+      "loss": 1.5039,
       "step": 380
     },
     {
       "epoch": 0.7857142857142857,
-      "grad_norm": 161412.09375,
-      "learning_rate": 1.9642857142857144e-06,
-      "loss": 1.7045,
+      "grad_norm": 239271.890625,
+      "learning_rate": 3.928571428571429e-06,
+      "loss": 1.3332,
       "step": 385
     },
     {
       "epoch": 0.7959183673469388,
-      "grad_norm": 117275.65625,
-      "learning_rate": 1.989795918367347e-06,
-      "loss": 1.74,
+      "grad_norm": 167386.28125,
+      "learning_rate": 3.979591836734694e-06,
+      "loss": 1.429,
       "step": 390
     },
     {
       "epoch": 0.8061224489795918,
-      "grad_norm": 134175.890625,
-      "learning_rate": 2.01530612244898e-06,
-      "loss": 1.7041,
+      "grad_norm": 165701.9375,
+      "learning_rate": 4.03061224489796e-06,
+      "loss": 1.3167,
       "step": 395
     },
     {
       "epoch": 0.8163265306122449,
-      "grad_norm": 165093.34375,
-      "learning_rate": 2.0408163265306125e-06,
-      "loss": 1.7217,
+      "grad_norm": 284793.46875,
+      "learning_rate": 4.081632653061225e-06,
+      "loss": 1.3867,
       "step": 400
     },
     {
       "epoch": 0.826530612244898,
-      "grad_norm": 108558.96875,
-      "learning_rate": 2.066326530612245e-06,
-      "loss": 1.7211,
+      "grad_norm": 159807.671875,
+      "learning_rate": 4.13265306122449e-06,
+      "loss": 1.404,
       "step": 405
     },
     {
       "epoch": 0.8367346938775511,
-      "grad_norm": 159761.328125,
-      "learning_rate": 2.0918367346938776e-06,
-      "loss": 1.7056,
+      "grad_norm": 225698.859375,
+      "learning_rate": 4.183673469387755e-06,
+      "loss": 1.3842,
       "step": 410
     },
     {
       "epoch": 0.8469387755102041,
-      "grad_norm": 131097.640625,
-      "learning_rate": 2.1173469387755106e-06,
-      "loss": 1.7289,
+      "grad_norm": 148945.90625,
+      "learning_rate": 4.234693877551021e-06,
+      "loss": 1.4605,
       "step": 415
     },
     {
       "epoch": 0.8571428571428571,
-      "grad_norm": 162492.6875,
-      "learning_rate": 2.1428571428571427e-06,
-      "loss": 1.7312,
+      "grad_norm": 301350.15625,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 1.4279,
       "step": 420
     },
     {
       "epoch": 0.8673469387755102,
-      "grad_norm": 122069.125,
-      "learning_rate": 2.1683673469387757e-06,
-      "loss": 1.6696,
+      "grad_norm": 216519.640625,
+      "learning_rate": 4.336734693877551e-06,
+      "loss": 1.3122,
       "step": 425
     },
     {
       "epoch": 0.8775510204081632,
-      "grad_norm": 194332.5625,
-      "learning_rate": 2.1938775510204083e-06,
-      "loss": 1.6226,
+      "grad_norm": 243937.1875,
+      "learning_rate": 4.3877551020408165e-06,
+      "loss": 1.2084,
       "step": 430
     },
     {
       "epoch": 0.8877551020408163,
-      "grad_norm": 132711.28125,
-      "learning_rate": 2.219387755102041e-06,
-      "loss": 1.663,
+      "grad_norm": 195260.390625,
+      "learning_rate": 4.438775510204082e-06,
+      "loss": 1.3024,
       "step": 435
     },
     {
       "epoch": 0.8979591836734694,
-      "grad_norm": 166239.375,
-      "learning_rate": 2.244897959183674e-06,
-      "loss": 1.6127,
+      "grad_norm": 218123.65625,
+      "learning_rate": 4.489795918367348e-06,
+      "loss": 1.2303,
       "step": 440
     },
     {
       "epoch": 0.9081632653061225,
-      "grad_norm": 132459.359375,
-      "learning_rate": 2.2704081632653064e-06,
-      "loss": 1.6042,
+      "grad_norm": 343990.03125,
+      "learning_rate": 4.540816326530613e-06,
+      "loss": 1.2012,
       "step": 445
     },
     {
       "epoch": 0.9183673469387755,
-      "grad_norm": 114693.3046875,
-      "learning_rate": 2.295918367346939e-06,
-      "loss": 1.6276,
+      "grad_norm": 157196.390625,
+      "learning_rate": 4.591836734693878e-06,
+      "loss": 1.2632,
       "step": 450
     },
     {
       "epoch": 0.9285714285714286,
-      "grad_norm": 184974.53125,
-      "learning_rate": 2.321428571428572e-06,
-      "loss": 1.6448,
+      "grad_norm": 238103.46875,
+      "learning_rate": 4.642857142857144e-06,
+      "loss": 1.3131,
       "step": 455
     },
     {
       "epoch": 0.9387755102040817,
-      "grad_norm": 134113.625,
-      "learning_rate": 2.3469387755102044e-06,
-      "loss": 1.6329,
+      "grad_norm": 158442.625,
+      "learning_rate": 4.693877551020409e-06,
+      "loss": 1.2849,
       "step": 460
     },
     {
       "epoch": 0.9489795918367347,
-      "grad_norm": 137895.78125,
-      "learning_rate": 2.372448979591837e-06,
-      "loss": 1.5441,
+      "grad_norm": 178316.640625,
+      "learning_rate": 4.744897959183674e-06,
+      "loss": 1.0987,
       "step": 465
     },
     {
       "epoch": 0.9591836734693877,
-      "grad_norm": 158595.234375,
-      "learning_rate": 2.3979591836734696e-06,
-      "loss": 1.5524,
+      "grad_norm": 240848.03125,
+      "learning_rate": 4.795918367346939e-06,
+      "loss": 1.154,
       "step": 470
     },
     {
       "epoch": 0.9693877551020408,
-      "grad_norm": 162135.703125,
-      "learning_rate": 2.423469387755102e-06,
-      "loss": 1.5796,
+      "grad_norm": 677730.8125,
+      "learning_rate": 4.846938775510204e-06,
+      "loss": 1.1682,
       "step": 475
     },
     {
       "epoch": 0.9795918367346939,
-      "grad_norm": 205519.296875,
-      "learning_rate": 2.4489795918367347e-06,
-      "loss": 1.5343,
+      "grad_norm": 300622.875,
+      "learning_rate": 4.897959183673469e-06,
+      "loss": 1.1212,
       "step": 480
     },
     {
       "epoch": 0.9897959183673469,
-      "grad_norm": 193819.75,
-      "learning_rate": 2.4744897959183676e-06,
-      "loss": 1.5776,
+      "grad_norm": 234745.03125,
+      "learning_rate": 4.948979591836735e-06,
+      "loss": 1.1859,
       "step": 485
     },
     {
       "epoch": 1.0,
-      "grad_norm": 225175.609375,
-      "learning_rate": 2.5e-06,
-      "loss": 1.5384,
+      "grad_norm": 254966.984375,
+      "learning_rate": 5e-06,
+      "loss": 1.1763,
       "step": 490
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.8470824949698189,
-      "eval_loss": 1.4336239099502563,
-      "eval_runtime": 214.7827,
-      "eval_samples_per_second": 13.884,
-      "eval_steps_per_second": 0.582,
+      "eval_accuracy": 0.8883299798792756,
+      "eval_loss": 0.968666136264801,
+      "eval_runtime": 208.5533,
+      "eval_samples_per_second": 14.299,
+      "eval_steps_per_second": 0.599,
       "step": 490
     }
   ],

checkpoint-490/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0feb04296f2d882ffa509dd7c6583732c9fea0f6c1d003c8bddb1fbd32ad29ca
+oid sha256:65e022e2b943d9a83eff2469e42d3ed5d9859bd381fb675a108709af3c02f25f
 size 5176