whisper-it-small-augmented / trainer_state.json
Moreno La Quatra
End of training
595eccf
raw
history blame
No virus
63.8 kB
{
"best_metric": 8.749734212204975,
"best_model_checkpoint": "whisper-it-small-multids-augmented/checkpoint-25000",
"epoch": 10.046,
"global_step": 25000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.7e-06,
"loss": 2.817,
"step": 50
},
{
"epoch": 0.0,
"learning_rate": 9.7e-06,
"loss": 1.0618,
"step": 100
},
{
"epoch": 0.01,
"learning_rate": 1.47e-05,
"loss": 0.6346,
"step": 150
},
{
"epoch": 0.01,
"learning_rate": 1.97e-05,
"loss": 0.4643,
"step": 200
},
{
"epoch": 0.01,
"learning_rate": 2.47e-05,
"loss": 0.3788,
"step": 250
},
{
"epoch": 0.01,
"learning_rate": 2.97e-05,
"loss": 0.3593,
"step": 300
},
{
"epoch": 0.01,
"learning_rate": 3.4699999999999996e-05,
"loss": 0.3703,
"step": 350
},
{
"epoch": 0.02,
"learning_rate": 3.97e-05,
"loss": 0.3383,
"step": 400
},
{
"epoch": 0.02,
"learning_rate": 4.47e-05,
"loss": 0.3096,
"step": 450
},
{
"epoch": 0.02,
"learning_rate": 4.97e-05,
"loss": 0.32,
"step": 500
},
{
"epoch": 0.02,
"learning_rate": 4.9904081632653066e-05,
"loss": 0.3321,
"step": 550
},
{
"epoch": 0.02,
"learning_rate": 4.980204081632654e-05,
"loss": 0.3139,
"step": 600
},
{
"epoch": 0.03,
"learning_rate": 4.97e-05,
"loss": 0.3148,
"step": 650
},
{
"epoch": 0.03,
"learning_rate": 4.959795918367347e-05,
"loss": 0.3107,
"step": 700
},
{
"epoch": 0.03,
"learning_rate": 4.949591836734694e-05,
"loss": 0.2823,
"step": 750
},
{
"epoch": 0.03,
"learning_rate": 4.939387755102041e-05,
"loss": 0.2677,
"step": 800
},
{
"epoch": 0.03,
"learning_rate": 4.929183673469388e-05,
"loss": 0.2671,
"step": 850
},
{
"epoch": 0.04,
"learning_rate": 4.9189795918367346e-05,
"loss": 0.2386,
"step": 900
},
{
"epoch": 0.04,
"learning_rate": 4.908775510204082e-05,
"loss": 0.2467,
"step": 950
},
{
"epoch": 0.04,
"learning_rate": 4.898571428571429e-05,
"loss": 0.2315,
"step": 1000
},
{
"epoch": 0.04,
"learning_rate": 4.888367346938776e-05,
"loss": 0.2456,
"step": 1050
},
{
"epoch": 0.04,
"learning_rate": 4.8781632653061225e-05,
"loss": 0.2501,
"step": 1100
},
{
"epoch": 0.05,
"learning_rate": 4.8679591836734696e-05,
"loss": 0.241,
"step": 1150
},
{
"epoch": 0.05,
"learning_rate": 4.857755102040816e-05,
"loss": 0.2276,
"step": 1200
},
{
"epoch": 0.05,
"learning_rate": 4.847551020408164e-05,
"loss": 0.2373,
"step": 1250
},
{
"epoch": 0.05,
"learning_rate": 4.8373469387755104e-05,
"loss": 0.215,
"step": 1300
},
{
"epoch": 0.05,
"learning_rate": 4.8271428571428575e-05,
"loss": 0.2136,
"step": 1350
},
{
"epoch": 0.06,
"learning_rate": 4.816938775510204e-05,
"loss": 0.1935,
"step": 1400
},
{
"epoch": 0.06,
"learning_rate": 4.806734693877551e-05,
"loss": 0.2047,
"step": 1450
},
{
"epoch": 0.06,
"learning_rate": 4.796530612244898e-05,
"loss": 0.1958,
"step": 1500
},
{
"epoch": 0.06,
"learning_rate": 4.7863265306122454e-05,
"loss": 0.2057,
"step": 1550
},
{
"epoch": 0.06,
"learning_rate": 4.776122448979592e-05,
"loss": 0.1945,
"step": 1600
},
{
"epoch": 0.07,
"learning_rate": 4.765918367346939e-05,
"loss": 0.2094,
"step": 1650
},
{
"epoch": 0.07,
"learning_rate": 4.755714285714286e-05,
"loss": 0.2152,
"step": 1700
},
{
"epoch": 0.07,
"learning_rate": 4.745510204081633e-05,
"loss": 0.2172,
"step": 1750
},
{
"epoch": 0.07,
"learning_rate": 4.73530612244898e-05,
"loss": 0.2236,
"step": 1800
},
{
"epoch": 0.07,
"learning_rate": 4.725102040816326e-05,
"loss": 0.2218,
"step": 1850
},
{
"epoch": 0.08,
"learning_rate": 4.714897959183674e-05,
"loss": 0.2382,
"step": 1900
},
{
"epoch": 0.08,
"learning_rate": 4.7046938775510206e-05,
"loss": 0.2301,
"step": 1950
},
{
"epoch": 0.08,
"learning_rate": 4.694489795918368e-05,
"loss": 0.184,
"step": 2000
},
{
"epoch": 0.08,
"learning_rate": 4.684285714285714e-05,
"loss": 0.1744,
"step": 2050
},
{
"epoch": 0.08,
"learning_rate": 4.6740816326530614e-05,
"loss": 0.2076,
"step": 2100
},
{
"epoch": 0.09,
"learning_rate": 4.6638775510204085e-05,
"loss": 0.1962,
"step": 2150
},
{
"epoch": 0.09,
"learning_rate": 4.6536734693877556e-05,
"loss": 0.1845,
"step": 2200
},
{
"epoch": 0.09,
"learning_rate": 4.643469387755102e-05,
"loss": 0.1898,
"step": 2250
},
{
"epoch": 0.09,
"learning_rate": 4.633265306122449e-05,
"loss": 0.1786,
"step": 2300
},
{
"epoch": 0.09,
"learning_rate": 4.6230612244897964e-05,
"loss": 0.2101,
"step": 2350
},
{
"epoch": 1.0,
"learning_rate": 4.612857142857143e-05,
"loss": 0.2193,
"step": 2400
},
{
"epoch": 1.0,
"learning_rate": 4.60265306122449e-05,
"loss": 0.1793,
"step": 2450
},
{
"epoch": 1.0,
"learning_rate": 4.592448979591837e-05,
"loss": 0.1927,
"step": 2500
},
{
"epoch": 1.0,
"eval_loss": 0.25064852833747864,
"eval_runtime": 3140.9231,
"eval_samples_per_second": 4.777,
"eval_steps_per_second": 0.149,
"eval_wer": 14.999069742717413,
"step": 2500
},
{
"epoch": 1.01,
"learning_rate": 4.5822448979591836e-05,
"loss": 0.1881,
"step": 2550
},
{
"epoch": 1.01,
"learning_rate": 4.572040816326531e-05,
"loss": 0.1624,
"step": 2600
},
{
"epoch": 1.01,
"learning_rate": 4.561836734693878e-05,
"loss": 0.1401,
"step": 2650
},
{
"epoch": 1.01,
"learning_rate": 4.5516326530612244e-05,
"loss": 0.1551,
"step": 2700
},
{
"epoch": 1.01,
"learning_rate": 4.5414285714285716e-05,
"loss": 0.1483,
"step": 2750
},
{
"epoch": 1.02,
"learning_rate": 4.531224489795919e-05,
"loss": 0.1278,
"step": 2800
},
{
"epoch": 1.02,
"learning_rate": 4.521020408163266e-05,
"loss": 0.1246,
"step": 2850
},
{
"epoch": 1.02,
"learning_rate": 4.510816326530612e-05,
"loss": 0.1259,
"step": 2900
},
{
"epoch": 1.02,
"learning_rate": 4.5006122448979595e-05,
"loss": 0.1236,
"step": 2950
},
{
"epoch": 1.02,
"learning_rate": 4.490408163265306e-05,
"loss": 0.1249,
"step": 3000
},
{
"epoch": 1.03,
"learning_rate": 4.480204081632654e-05,
"loss": 0.1253,
"step": 3050
},
{
"epoch": 1.03,
"learning_rate": 4.47e-05,
"loss": 0.1134,
"step": 3100
},
{
"epoch": 1.03,
"learning_rate": 4.4597959183673474e-05,
"loss": 0.1045,
"step": 3150
},
{
"epoch": 1.03,
"learning_rate": 4.449591836734694e-05,
"loss": 0.1074,
"step": 3200
},
{
"epoch": 1.03,
"learning_rate": 4.439387755102041e-05,
"loss": 0.104,
"step": 3250
},
{
"epoch": 1.04,
"learning_rate": 4.429183673469388e-05,
"loss": 0.1027,
"step": 3300
},
{
"epoch": 1.04,
"learning_rate": 4.4189795918367346e-05,
"loss": 0.1002,
"step": 3350
},
{
"epoch": 1.04,
"learning_rate": 4.408775510204082e-05,
"loss": 0.0978,
"step": 3400
},
{
"epoch": 1.04,
"learning_rate": 4.398571428571428e-05,
"loss": 0.0989,
"step": 3450
},
{
"epoch": 1.04,
"learning_rate": 4.388367346938776e-05,
"loss": 0.1094,
"step": 3500
},
{
"epoch": 1.05,
"learning_rate": 4.3781632653061225e-05,
"loss": 0.1031,
"step": 3550
},
{
"epoch": 1.05,
"learning_rate": 4.36795918367347e-05,
"loss": 0.0977,
"step": 3600
},
{
"epoch": 1.05,
"learning_rate": 4.357755102040816e-05,
"loss": 0.0947,
"step": 3650
},
{
"epoch": 1.05,
"learning_rate": 4.347551020408163e-05,
"loss": 0.092,
"step": 3700
},
{
"epoch": 1.05,
"learning_rate": 4.3373469387755104e-05,
"loss": 0.0831,
"step": 3750
},
{
"epoch": 1.06,
"learning_rate": 4.3271428571428576e-05,
"loss": 0.0852,
"step": 3800
},
{
"epoch": 1.06,
"learning_rate": 4.316938775510204e-05,
"loss": 0.0893,
"step": 3850
},
{
"epoch": 1.06,
"learning_rate": 4.306734693877551e-05,
"loss": 0.0865,
"step": 3900
},
{
"epoch": 1.06,
"learning_rate": 4.2965306122448983e-05,
"loss": 0.0892,
"step": 3950
},
{
"epoch": 1.06,
"learning_rate": 4.2863265306122455e-05,
"loss": 0.0824,
"step": 4000
},
{
"epoch": 1.07,
"learning_rate": 4.2769387755102044e-05,
"loss": 0.4548,
"step": 4050
},
{
"epoch": 1.07,
"learning_rate": 4.266734693877551e-05,
"loss": 0.0999,
"step": 4100
},
{
"epoch": 1.07,
"learning_rate": 4.256530612244898e-05,
"loss": 0.1038,
"step": 4150
},
{
"epoch": 1.07,
"learning_rate": 4.246326530612245e-05,
"loss": 0.1018,
"step": 4200
},
{
"epoch": 1.07,
"learning_rate": 4.236122448979592e-05,
"loss": 0.104,
"step": 4250
},
{
"epoch": 1.08,
"learning_rate": 4.225918367346939e-05,
"loss": 0.1149,
"step": 4300
},
{
"epoch": 1.08,
"learning_rate": 4.215714285714286e-05,
"loss": 0.0969,
"step": 4350
},
{
"epoch": 1.08,
"learning_rate": 4.205510204081633e-05,
"loss": 0.0842,
"step": 4400
},
{
"epoch": 1.08,
"learning_rate": 4.1953061224489796e-05,
"loss": 0.0811,
"step": 4450
},
{
"epoch": 1.08,
"learning_rate": 4.185102040816327e-05,
"loss": 0.0988,
"step": 4500
},
{
"epoch": 1.09,
"learning_rate": 4.174897959183673e-05,
"loss": 0.0866,
"step": 4550
},
{
"epoch": 1.09,
"learning_rate": 4.164693877551021e-05,
"loss": 0.0881,
"step": 4600
},
{
"epoch": 1.09,
"learning_rate": 4.1544897959183675e-05,
"loss": 0.0889,
"step": 4650
},
{
"epoch": 1.09,
"learning_rate": 4.1442857142857146e-05,
"loss": 0.088,
"step": 4700
},
{
"epoch": 1.09,
"learning_rate": 4.134081632653061e-05,
"loss": 0.0985,
"step": 4750
},
{
"epoch": 2.0,
"learning_rate": 4.123877551020409e-05,
"loss": 0.1051,
"step": 4800
},
{
"epoch": 2.0,
"learning_rate": 4.1136734693877554e-05,
"loss": 0.08,
"step": 4850
},
{
"epoch": 2.01,
"learning_rate": 4.1034693877551025e-05,
"loss": 0.098,
"step": 4900
},
{
"epoch": 2.01,
"learning_rate": 4.093265306122449e-05,
"loss": 0.0877,
"step": 4950
},
{
"epoch": 2.01,
"learning_rate": 4.083061224489796e-05,
"loss": 0.0736,
"step": 5000
},
{
"epoch": 2.01,
"eval_loss": 0.2257717102766037,
"eval_runtime": 3250.772,
"eval_samples_per_second": 4.615,
"eval_steps_per_second": 0.144,
"eval_wer": 12.786386349138848,
"step": 5000
},
{
"epoch": 2.01,
"learning_rate": 4.072857142857143e-05,
"loss": 0.0666,
"step": 5050
},
{
"epoch": 2.01,
"learning_rate": 4.06265306122449e-05,
"loss": 0.0797,
"step": 5100
},
{
"epoch": 2.02,
"learning_rate": 4.052448979591837e-05,
"loss": 0.0671,
"step": 5150
},
{
"epoch": 2.02,
"learning_rate": 4.0422448979591834e-05,
"loss": 0.0624,
"step": 5200
},
{
"epoch": 2.02,
"learning_rate": 4.032040816326531e-05,
"loss": 0.0611,
"step": 5250
},
{
"epoch": 2.02,
"learning_rate": 4.021836734693878e-05,
"loss": 0.0698,
"step": 5300
},
{
"epoch": 2.02,
"learning_rate": 4.011632653061225e-05,
"loss": 0.0699,
"step": 5350
},
{
"epoch": 2.03,
"learning_rate": 4.001428571428571e-05,
"loss": 0.063,
"step": 5400
},
{
"epoch": 2.03,
"learning_rate": 3.9912244897959185e-05,
"loss": 0.0641,
"step": 5450
},
{
"epoch": 2.03,
"learning_rate": 3.9810204081632656e-05,
"loss": 0.0556,
"step": 5500
},
{
"epoch": 2.03,
"learning_rate": 3.970816326530613e-05,
"loss": 0.0562,
"step": 5550
},
{
"epoch": 2.03,
"learning_rate": 3.960612244897959e-05,
"loss": 0.0568,
"step": 5600
},
{
"epoch": 2.04,
"learning_rate": 3.9504081632653064e-05,
"loss": 0.0582,
"step": 5650
},
{
"epoch": 2.04,
"learning_rate": 3.940204081632653e-05,
"loss": 0.0545,
"step": 5700
},
{
"epoch": 2.04,
"learning_rate": 3.9300000000000007e-05,
"loss": 0.0549,
"step": 5750
},
{
"epoch": 2.04,
"learning_rate": 3.919795918367347e-05,
"loss": 0.0542,
"step": 5800
},
{
"epoch": 2.04,
"learning_rate": 3.909591836734694e-05,
"loss": 0.0561,
"step": 5850
},
{
"epoch": 2.05,
"learning_rate": 3.899387755102041e-05,
"loss": 0.0603,
"step": 5900
},
{
"epoch": 2.05,
"learning_rate": 3.889183673469388e-05,
"loss": 0.0553,
"step": 5950
},
{
"epoch": 2.05,
"learning_rate": 3.878979591836735e-05,
"loss": 0.0524,
"step": 6000
},
{
"epoch": 2.05,
"learning_rate": 3.8687755102040815e-05,
"loss": 0.0522,
"step": 6050
},
{
"epoch": 2.05,
"learning_rate": 3.8585714285714287e-05,
"loss": 0.049,
"step": 6100
},
{
"epoch": 2.06,
"learning_rate": 3.848367346938775e-05,
"loss": 0.0441,
"step": 6150
},
{
"epoch": 2.06,
"learning_rate": 3.838163265306123e-05,
"loss": 0.048,
"step": 6200
},
{
"epoch": 2.06,
"learning_rate": 3.8279591836734694e-05,
"loss": 0.0435,
"step": 6250
},
{
"epoch": 2.06,
"learning_rate": 3.8177551020408166e-05,
"loss": 0.052,
"step": 6300
},
{
"epoch": 2.06,
"learning_rate": 3.807551020408163e-05,
"loss": 0.0477,
"step": 6350
},
{
"epoch": 2.07,
"learning_rate": 3.797346938775511e-05,
"loss": 0.0498,
"step": 6400
},
{
"epoch": 2.07,
"learning_rate": 3.787142857142857e-05,
"loss": 0.0617,
"step": 6450
},
{
"epoch": 2.07,
"learning_rate": 3.7769387755102045e-05,
"loss": 0.0561,
"step": 6500
},
{
"epoch": 2.07,
"learning_rate": 3.766734693877551e-05,
"loss": 0.052,
"step": 6550
},
{
"epoch": 2.07,
"learning_rate": 3.756530612244898e-05,
"loss": 0.0543,
"step": 6600
},
{
"epoch": 2.08,
"learning_rate": 3.746326530612245e-05,
"loss": 0.0567,
"step": 6650
},
{
"epoch": 2.08,
"learning_rate": 3.7361224489795924e-05,
"loss": 0.059,
"step": 6700
},
{
"epoch": 2.08,
"learning_rate": 3.725918367346939e-05,
"loss": 0.0473,
"step": 6750
},
{
"epoch": 2.08,
"learning_rate": 3.715714285714286e-05,
"loss": 0.0439,
"step": 6800
},
{
"epoch": 2.08,
"learning_rate": 3.705510204081633e-05,
"loss": 0.0445,
"step": 6850
},
{
"epoch": 2.09,
"learning_rate": 3.6953061224489796e-05,
"loss": 0.0517,
"step": 6900
},
{
"epoch": 2.09,
"learning_rate": 3.685102040816327e-05,
"loss": 0.0423,
"step": 6950
},
{
"epoch": 2.09,
"learning_rate": 3.674897959183673e-05,
"loss": 0.047,
"step": 7000
},
{
"epoch": 2.09,
"learning_rate": 3.6646938775510204e-05,
"loss": 0.0448,
"step": 7050
},
{
"epoch": 2.09,
"learning_rate": 3.6544897959183675e-05,
"loss": 0.0436,
"step": 7100
},
{
"epoch": 2.1,
"learning_rate": 3.644285714285715e-05,
"loss": 0.0543,
"step": 7150
},
{
"epoch": 3.0,
"learning_rate": 3.634081632653061e-05,
"loss": 0.0458,
"step": 7200
},
{
"epoch": 3.0,
"learning_rate": 3.623877551020408e-05,
"loss": 0.0438,
"step": 7250
},
{
"epoch": 3.01,
"learning_rate": 3.6136734693877554e-05,
"loss": 0.0516,
"step": 7300
},
{
"epoch": 3.01,
"learning_rate": 3.6034693877551026e-05,
"loss": 0.0423,
"step": 7350
},
{
"epoch": 3.01,
"learning_rate": 3.593265306122449e-05,
"loss": 0.0355,
"step": 7400
},
{
"epoch": 3.01,
"learning_rate": 3.583061224489796e-05,
"loss": 0.0407,
"step": 7450
},
{
"epoch": 3.01,
"learning_rate": 3.572857142857143e-05,
"loss": 0.0413,
"step": 7500
},
{
"epoch": 3.01,
"eval_loss": 0.21440866589546204,
"eval_runtime": 3126.0959,
"eval_samples_per_second": 4.799,
"eval_steps_per_second": 0.15,
"eval_wer": 11.450802679140974,
"step": 7500
},
{
"epoch": 3.02,
"learning_rate": 3.5626530612244905e-05,
"loss": 0.0356,
"step": 7550
},
{
"epoch": 3.02,
"learning_rate": 3.552448979591837e-05,
"loss": 0.0316,
"step": 7600
},
{
"epoch": 3.02,
"learning_rate": 3.542244897959184e-05,
"loss": 0.0317,
"step": 7650
},
{
"epoch": 3.02,
"learning_rate": 3.5320408163265306e-05,
"loss": 0.0355,
"step": 7700
},
{
"epoch": 3.02,
"learning_rate": 3.521836734693878e-05,
"loss": 0.0374,
"step": 7750
},
{
"epoch": 3.03,
"learning_rate": 3.511632653061225e-05,
"loss": 0.0363,
"step": 7800
},
{
"epoch": 3.03,
"learning_rate": 3.5014285714285714e-05,
"loss": 0.0355,
"step": 7850
},
{
"epoch": 3.03,
"learning_rate": 3.4912244897959185e-05,
"loss": 0.0302,
"step": 7900
},
{
"epoch": 3.03,
"learning_rate": 3.481020408163265e-05,
"loss": 0.0294,
"step": 7950
},
{
"epoch": 3.03,
"learning_rate": 3.470816326530613e-05,
"loss": 0.0326,
"step": 8000
},
{
"epoch": 3.04,
"learning_rate": 3.460612244897959e-05,
"loss": 0.0288,
"step": 8050
},
{
"epoch": 3.04,
"learning_rate": 3.4504081632653064e-05,
"loss": 0.03,
"step": 8100
},
{
"epoch": 3.04,
"learning_rate": 3.440204081632653e-05,
"loss": 0.0275,
"step": 8150
},
{
"epoch": 3.04,
"learning_rate": 3.430000000000001e-05,
"loss": 0.0297,
"step": 8200
},
{
"epoch": 3.04,
"learning_rate": 3.419795918367347e-05,
"loss": 0.03,
"step": 8250
},
{
"epoch": 3.05,
"learning_rate": 3.409591836734694e-05,
"loss": 0.0302,
"step": 8300
},
{
"epoch": 3.05,
"learning_rate": 3.399387755102041e-05,
"loss": 0.0275,
"step": 8350
},
{
"epoch": 3.05,
"learning_rate": 3.389183673469388e-05,
"loss": 0.0295,
"step": 8400
},
{
"epoch": 3.05,
"learning_rate": 3.378979591836735e-05,
"loss": 0.0287,
"step": 8450
},
{
"epoch": 3.05,
"learning_rate": 3.3687755102040816e-05,
"loss": 0.0282,
"step": 8500
},
{
"epoch": 3.06,
"learning_rate": 3.358571428571429e-05,
"loss": 0.0231,
"step": 8550
},
{
"epoch": 3.06,
"learning_rate": 3.348367346938776e-05,
"loss": 0.0256,
"step": 8600
},
{
"epoch": 3.06,
"learning_rate": 3.338163265306122e-05,
"loss": 0.0245,
"step": 8650
},
{
"epoch": 3.06,
"learning_rate": 3.3279591836734695e-05,
"loss": 0.0262,
"step": 8700
},
{
"epoch": 3.06,
"learning_rate": 3.3177551020408166e-05,
"loss": 0.0259,
"step": 8750
},
{
"epoch": 3.07,
"learning_rate": 3.307551020408163e-05,
"loss": 0.031,
"step": 8800
},
{
"epoch": 3.07,
"learning_rate": 3.29734693877551e-05,
"loss": 0.0286,
"step": 8850
},
{
"epoch": 3.07,
"learning_rate": 3.2871428571428574e-05,
"loss": 0.0284,
"step": 8900
},
{
"epoch": 3.07,
"learning_rate": 3.2769387755102045e-05,
"loss": 0.0322,
"step": 8950
},
{
"epoch": 3.07,
"learning_rate": 3.266734693877551e-05,
"loss": 0.0323,
"step": 9000
},
{
"epoch": 3.08,
"learning_rate": 3.256530612244898e-05,
"loss": 0.0334,
"step": 9050
},
{
"epoch": 3.08,
"learning_rate": 3.2463265306122446e-05,
"loss": 0.0329,
"step": 9100
},
{
"epoch": 3.08,
"learning_rate": 3.2361224489795924e-05,
"loss": 0.0264,
"step": 9150
},
{
"epoch": 3.08,
"learning_rate": 3.225918367346939e-05,
"loss": 0.024,
"step": 9200
},
{
"epoch": 3.08,
"learning_rate": 3.215714285714286e-05,
"loss": 0.0306,
"step": 9250
},
{
"epoch": 3.09,
"learning_rate": 3.2055102040816325e-05,
"loss": 0.0252,
"step": 9300
},
{
"epoch": 3.09,
"learning_rate": 3.19530612244898e-05,
"loss": 0.0244,
"step": 9350
},
{
"epoch": 3.09,
"learning_rate": 3.185102040816327e-05,
"loss": 0.0244,
"step": 9400
},
{
"epoch": 3.09,
"learning_rate": 3.174897959183673e-05,
"loss": 0.0232,
"step": 9450
},
{
"epoch": 3.09,
"learning_rate": 3.1646938775510204e-05,
"loss": 0.0261,
"step": 9500
},
{
"epoch": 4.0,
"learning_rate": 3.1544897959183676e-05,
"loss": 0.0304,
"step": 9550
},
{
"epoch": 4.0,
"learning_rate": 3.144285714285715e-05,
"loss": 0.0266,
"step": 9600
},
{
"epoch": 4.0,
"learning_rate": 3.134081632653061e-05,
"loss": 0.0275,
"step": 9650
},
{
"epoch": 4.01,
"learning_rate": 3.1238775510204083e-05,
"loss": 0.028,
"step": 9700
},
{
"epoch": 4.01,
"learning_rate": 3.113673469387755e-05,
"loss": 0.0223,
"step": 9750
},
{
"epoch": 4.01,
"learning_rate": 3.1034693877551026e-05,
"loss": 0.0197,
"step": 9800
},
{
"epoch": 4.01,
"learning_rate": 3.093265306122449e-05,
"loss": 0.0249,
"step": 9850
},
{
"epoch": 4.01,
"learning_rate": 3.083061224489796e-05,
"loss": 0.0259,
"step": 9900
},
{
"epoch": 4.02,
"learning_rate": 3.072857142857143e-05,
"loss": 0.0201,
"step": 9950
},
{
"epoch": 4.02,
"learning_rate": 3.06265306122449e-05,
"loss": 0.0201,
"step": 10000
},
{
"epoch": 4.02,
"eval_loss": 0.21463686227798462,
"eval_runtime": 3076.5113,
"eval_samples_per_second": 4.877,
"eval_steps_per_second": 0.152,
"eval_wer": 10.877365511375718,
"step": 10000
},
{
"epoch": 4.02,
"learning_rate": 3.052448979591837e-05,
"loss": 0.0204,
"step": 10050
},
{
"epoch": 4.02,
"learning_rate": 3.0422448979591838e-05,
"loss": 0.0226,
"step": 10100
},
{
"epoch": 4.02,
"learning_rate": 3.0320408163265306e-05,
"loss": 0.0214,
"step": 10150
},
{
"epoch": 4.03,
"learning_rate": 3.0218367346938774e-05,
"loss": 0.0219,
"step": 10200
},
{
"epoch": 4.03,
"learning_rate": 3.011632653061225e-05,
"loss": 0.0201,
"step": 10250
},
{
"epoch": 4.03,
"learning_rate": 3.0014285714285717e-05,
"loss": 0.0189,
"step": 10300
},
{
"epoch": 4.03,
"learning_rate": 2.9912244897959185e-05,
"loss": 0.0201,
"step": 10350
},
{
"epoch": 4.03,
"learning_rate": 2.9810204081632653e-05,
"loss": 0.0202,
"step": 10400
},
{
"epoch": 4.04,
"learning_rate": 2.970816326530612e-05,
"loss": 0.0163,
"step": 10450
},
{
"epoch": 4.04,
"learning_rate": 2.9606122448979596e-05,
"loss": 0.0175,
"step": 10500
},
{
"epoch": 4.04,
"learning_rate": 2.9504081632653065e-05,
"loss": 0.0161,
"step": 10550
},
{
"epoch": 4.04,
"learning_rate": 2.9402040816326533e-05,
"loss": 0.0189,
"step": 10600
},
{
"epoch": 4.04,
"learning_rate": 2.93e-05,
"loss": 0.0182,
"step": 10650
},
{
"epoch": 4.05,
"learning_rate": 2.9197959183673472e-05,
"loss": 0.0174,
"step": 10700
},
{
"epoch": 4.05,
"learning_rate": 2.909591836734694e-05,
"loss": 0.0149,
"step": 10750
},
{
"epoch": 4.05,
"learning_rate": 2.899387755102041e-05,
"loss": 0.017,
"step": 10800
},
{
"epoch": 4.05,
"learning_rate": 2.8891836734693876e-05,
"loss": 0.015,
"step": 10850
},
{
"epoch": 4.05,
"learning_rate": 2.8789795918367344e-05,
"loss": 0.0132,
"step": 10900
},
{
"epoch": 4.06,
"learning_rate": 2.868775510204082e-05,
"loss": 0.0137,
"step": 10950
},
{
"epoch": 4.06,
"learning_rate": 2.8585714285714287e-05,
"loss": 0.0141,
"step": 11000
},
{
"epoch": 4.06,
"learning_rate": 2.8483673469387756e-05,
"loss": 0.0138,
"step": 11050
},
{
"epoch": 4.06,
"learning_rate": 2.8381632653061224e-05,
"loss": 0.0143,
"step": 11100
},
{
"epoch": 4.06,
"learning_rate": 2.82795918367347e-05,
"loss": 0.0167,
"step": 11150
},
{
"epoch": 4.07,
"learning_rate": 2.8177551020408167e-05,
"loss": 0.0195,
"step": 11200
},
{
"epoch": 4.07,
"learning_rate": 2.8075510204081635e-05,
"loss": 0.017,
"step": 11250
},
{
"epoch": 4.07,
"learning_rate": 2.7973469387755103e-05,
"loss": 0.0196,
"step": 11300
},
{
"epoch": 4.07,
"learning_rate": 2.787142857142857e-05,
"loss": 0.0169,
"step": 11350
},
{
"epoch": 4.07,
"learning_rate": 2.7769387755102046e-05,
"loss": 0.0192,
"step": 11400
},
{
"epoch": 4.08,
"learning_rate": 2.7667346938775514e-05,
"loss": 0.0201,
"step": 11450
},
{
"epoch": 4.08,
"learning_rate": 2.7565306122448982e-05,
"loss": 0.0166,
"step": 11500
},
{
"epoch": 4.08,
"learning_rate": 2.746326530612245e-05,
"loss": 0.0141,
"step": 11550
},
{
"epoch": 4.08,
"learning_rate": 2.7361224489795918e-05,
"loss": 0.0145,
"step": 11600
},
{
"epoch": 4.08,
"learning_rate": 2.725918367346939e-05,
"loss": 0.0175,
"step": 11650
},
{
"epoch": 4.09,
"learning_rate": 2.7157142857142858e-05,
"loss": 0.015,
"step": 11700
},
{
"epoch": 4.09,
"learning_rate": 2.7055102040816326e-05,
"loss": 0.0145,
"step": 11750
},
{
"epoch": 4.09,
"learning_rate": 2.6953061224489794e-05,
"loss": 0.0127,
"step": 11800
},
{
"epoch": 4.09,
"learning_rate": 2.685102040816327e-05,
"loss": 0.0135,
"step": 11850
},
{
"epoch": 4.09,
"learning_rate": 2.6748979591836737e-05,
"loss": 0.0166,
"step": 11900
},
{
"epoch": 5.0,
"learning_rate": 2.6646938775510205e-05,
"loss": 0.0181,
"step": 11950
},
{
"epoch": 5.0,
"learning_rate": 2.6544897959183673e-05,
"loss": 0.0147,
"step": 12000
},
{
"epoch": 5.0,
"learning_rate": 2.644285714285714e-05,
"loss": 0.0168,
"step": 12050
},
{
"epoch": 5.01,
"learning_rate": 2.6340816326530616e-05,
"loss": 0.0154,
"step": 12100
},
{
"epoch": 5.01,
"learning_rate": 2.6238775510204084e-05,
"loss": 0.0142,
"step": 12150
},
{
"epoch": 5.01,
"learning_rate": 2.6136734693877552e-05,
"loss": 0.0125,
"step": 12200
},
{
"epoch": 5.01,
"learning_rate": 2.603469387755102e-05,
"loss": 0.0159,
"step": 12250
},
{
"epoch": 5.01,
"learning_rate": 2.5932653061224495e-05,
"loss": 0.0126,
"step": 12300
},
{
"epoch": 5.02,
"learning_rate": 2.5830612244897963e-05,
"loss": 0.013,
"step": 12350
},
{
"epoch": 5.02,
"learning_rate": 2.572857142857143e-05,
"loss": 0.0103,
"step": 12400
},
{
"epoch": 5.02,
"learning_rate": 2.56265306122449e-05,
"loss": 0.0123,
"step": 12450
},
{
"epoch": 5.02,
"learning_rate": 2.5524489795918367e-05,
"loss": 0.0129,
"step": 12500
},
{
"epoch": 5.02,
"eval_loss": 0.2126896232366562,
"eval_runtime": 3157.9608,
"eval_samples_per_second": 4.751,
"eval_steps_per_second": 0.149,
"eval_wer": 10.691978524346162,
"step": 12500
},
{
"epoch": 5.03,
"learning_rate": 2.542244897959184e-05,
"loss": 0.0122,
"step": 12550
},
{
"epoch": 5.03,
"learning_rate": 2.5320408163265307e-05,
"loss": 0.0128,
"step": 12600
},
{
"epoch": 5.03,
"learning_rate": 2.5218367346938775e-05,
"loss": 0.0108,
"step": 12650
},
{
"epoch": 5.03,
"learning_rate": 2.5116326530612243e-05,
"loss": 0.0103,
"step": 12700
},
{
"epoch": 5.03,
"learning_rate": 2.5014285714285718e-05,
"loss": 0.0127,
"step": 12750
},
{
"epoch": 5.04,
"learning_rate": 2.4912244897959186e-05,
"loss": 0.0115,
"step": 12800
},
{
"epoch": 5.04,
"learning_rate": 2.4810204081632654e-05,
"loss": 0.0112,
"step": 12850
},
{
"epoch": 5.04,
"learning_rate": 2.4708163265306122e-05,
"loss": 0.0102,
"step": 12900
},
{
"epoch": 5.04,
"learning_rate": 2.4606122448979593e-05,
"loss": 0.0109,
"step": 12950
},
{
"epoch": 5.04,
"learning_rate": 2.450408163265306e-05,
"loss": 0.01,
"step": 13000
},
{
"epoch": 5.04,
"learning_rate": 2.4402040816326533e-05,
"loss": 0.0105,
"step": 13050
},
{
"epoch": 5.05,
"learning_rate": 2.43e-05,
"loss": 0.0104,
"step": 13100
},
{
"epoch": 5.05,
"learning_rate": 2.4197959183673473e-05,
"loss": 0.0109,
"step": 13150
},
{
"epoch": 5.05,
"learning_rate": 2.409591836734694e-05,
"loss": 0.0101,
"step": 13200
},
{
"epoch": 5.05,
"learning_rate": 2.3993877551020412e-05,
"loss": 0.0085,
"step": 13250
},
{
"epoch": 5.05,
"learning_rate": 2.389183673469388e-05,
"loss": 0.0087,
"step": 13300
},
{
"epoch": 5.06,
"learning_rate": 2.378979591836735e-05,
"loss": 0.0082,
"step": 13350
},
{
"epoch": 5.06,
"learning_rate": 2.3687755102040816e-05,
"loss": 0.0084,
"step": 13400
},
{
"epoch": 5.06,
"learning_rate": 2.3585714285714284e-05,
"loss": 0.0093,
"step": 13450
},
{
"epoch": 5.06,
"learning_rate": 2.3483673469387756e-05,
"loss": 0.0088,
"step": 13500
},
{
"epoch": 5.07,
"learning_rate": 2.3381632653061224e-05,
"loss": 0.009,
"step": 13550
},
{
"epoch": 5.07,
"learning_rate": 2.3279591836734696e-05,
"loss": 0.0113,
"step": 13600
},
{
"epoch": 5.07,
"learning_rate": 2.3177551020408164e-05,
"loss": 0.0102,
"step": 13650
},
{
"epoch": 5.07,
"learning_rate": 2.3075510204081635e-05,
"loss": 0.0109,
"step": 13700
},
{
"epoch": 5.07,
"learning_rate": 2.2973469387755103e-05,
"loss": 0.0121,
"step": 13750
},
{
"epoch": 5.08,
"learning_rate": 2.287142857142857e-05,
"loss": 0.0111,
"step": 13800
},
{
"epoch": 5.08,
"learning_rate": 2.2769387755102043e-05,
"loss": 0.0136,
"step": 13850
},
{
"epoch": 5.08,
"learning_rate": 2.266734693877551e-05,
"loss": 0.0105,
"step": 13900
},
{
"epoch": 5.08,
"learning_rate": 2.2565306122448982e-05,
"loss": 0.0083,
"step": 13950
},
{
"epoch": 5.08,
"learning_rate": 2.246326530612245e-05,
"loss": 0.0089,
"step": 14000
},
{
"epoch": 5.08,
"learning_rate": 2.2361224489795922e-05,
"loss": 0.0109,
"step": 14050
},
{
"epoch": 5.09,
"learning_rate": 2.225918367346939e-05,
"loss": 0.0087,
"step": 14100
},
{
"epoch": 5.09,
"learning_rate": 2.2157142857142858e-05,
"loss": 0.0091,
"step": 14150
},
{
"epoch": 5.09,
"learning_rate": 2.205510204081633e-05,
"loss": 0.0089,
"step": 14200
},
{
"epoch": 5.09,
"learning_rate": 2.1953061224489798e-05,
"loss": 0.0077,
"step": 14250
},
{
"epoch": 5.09,
"learning_rate": 2.1851020408163266e-05,
"loss": 0.0096,
"step": 14300
},
{
"epoch": 6.0,
"learning_rate": 2.1748979591836734e-05,
"loss": 0.0107,
"step": 14350
},
{
"epoch": 6.0,
"learning_rate": 2.1646938775510205e-05,
"loss": 0.0082,
"step": 14400
},
{
"epoch": 6.01,
"learning_rate": 2.1544897959183673e-05,
"loss": 0.0099,
"step": 14450
},
{
"epoch": 6.01,
"learning_rate": 2.1442857142857145e-05,
"loss": 0.0078,
"step": 14500
},
{
"epoch": 6.01,
"learning_rate": 2.1340816326530613e-05,
"loss": 0.0082,
"step": 14550
},
{
"epoch": 6.01,
"learning_rate": 2.123877551020408e-05,
"loss": 0.0087,
"step": 14600
},
{
"epoch": 6.01,
"learning_rate": 2.1136734693877552e-05,
"loss": 0.009,
"step": 14650
},
{
"epoch": 6.02,
"learning_rate": 2.103469387755102e-05,
"loss": 0.0074,
"step": 14700
},
{
"epoch": 6.02,
"learning_rate": 2.0932653061224492e-05,
"loss": 0.0073,
"step": 14750
},
{
"epoch": 6.02,
"learning_rate": 2.083061224489796e-05,
"loss": 0.0079,
"step": 14800
},
{
"epoch": 6.02,
"learning_rate": 2.072857142857143e-05,
"loss": 0.0076,
"step": 14850
},
{
"epoch": 6.02,
"learning_rate": 2.06265306122449e-05,
"loss": 0.0071,
"step": 14900
},
{
"epoch": 6.03,
"learning_rate": 2.052448979591837e-05,
"loss": 0.0072,
"step": 14950
},
{
"epoch": 6.03,
"learning_rate": 2.042244897959184e-05,
"loss": 0.0091,
"step": 15000
},
{
"epoch": 6.03,
"eval_loss": 0.21168826520442963,
"eval_runtime": 2934.4723,
"eval_samples_per_second": 5.113,
"eval_steps_per_second": 0.16,
"eval_wer": 10.28665213693387,
"step": 15000
},
{
"epoch": 6.03,
"learning_rate": 2.0320408163265307e-05,
"loss": 0.0063,
"step": 15050
},
{
"epoch": 6.03,
"learning_rate": 2.0218367346938775e-05,
"loss": 0.007,
"step": 15100
},
{
"epoch": 6.03,
"learning_rate": 2.0116326530612243e-05,
"loss": 0.0078,
"step": 15150
},
{
"epoch": 6.04,
"learning_rate": 2.0014285714285715e-05,
"loss": 0.0057,
"step": 15200
},
{
"epoch": 6.04,
"learning_rate": 1.9912244897959183e-05,
"loss": 0.0051,
"step": 15250
},
{
"epoch": 6.04,
"learning_rate": 1.9810204081632654e-05,
"loss": 0.0063,
"step": 15300
},
{
"epoch": 6.04,
"learning_rate": 1.9708163265306122e-05,
"loss": 0.0049,
"step": 15350
},
{
"epoch": 6.04,
"learning_rate": 1.9606122448979594e-05,
"loss": 0.0061,
"step": 15400
},
{
"epoch": 6.05,
"learning_rate": 1.9504081632653062e-05,
"loss": 0.0059,
"step": 15450
},
{
"epoch": 6.05,
"learning_rate": 1.940204081632653e-05,
"loss": 0.0066,
"step": 15500
},
{
"epoch": 6.05,
"learning_rate": 1.93e-05,
"loss": 0.0074,
"step": 15550
},
{
"epoch": 6.05,
"learning_rate": 1.919795918367347e-05,
"loss": 0.0062,
"step": 15600
},
{
"epoch": 6.05,
"learning_rate": 1.909591836734694e-05,
"loss": 0.0052,
"step": 15650
},
{
"epoch": 6.06,
"learning_rate": 1.899387755102041e-05,
"loss": 0.0051,
"step": 15700
},
{
"epoch": 6.06,
"learning_rate": 1.889183673469388e-05,
"loss": 0.0054,
"step": 15750
},
{
"epoch": 6.06,
"learning_rate": 1.878979591836735e-05,
"loss": 0.0045,
"step": 15800
},
{
"epoch": 6.06,
"learning_rate": 1.8687755102040817e-05,
"loss": 0.0057,
"step": 15850
},
{
"epoch": 6.06,
"learning_rate": 1.858571428571429e-05,
"loss": 0.0048,
"step": 15900
},
{
"epoch": 6.07,
"learning_rate": 1.8483673469387756e-05,
"loss": 0.0057,
"step": 15950
},
{
"epoch": 6.07,
"learning_rate": 1.8381632653061224e-05,
"loss": 0.0064,
"step": 16000
},
{
"epoch": 6.07,
"learning_rate": 1.8279591836734693e-05,
"loss": 0.0077,
"step": 16050
},
{
"epoch": 6.07,
"learning_rate": 1.8177551020408164e-05,
"loss": 0.007,
"step": 16100
},
{
"epoch": 6.07,
"learning_rate": 1.8075510204081632e-05,
"loss": 0.0068,
"step": 16150
},
{
"epoch": 6.08,
"learning_rate": 1.7973469387755104e-05,
"loss": 0.0081,
"step": 16200
},
{
"epoch": 6.08,
"learning_rate": 1.787142857142857e-05,
"loss": 0.0072,
"step": 16250
},
{
"epoch": 6.08,
"learning_rate": 1.776938775510204e-05,
"loss": 0.0052,
"step": 16300
},
{
"epoch": 6.08,
"learning_rate": 1.766734693877551e-05,
"loss": 0.0047,
"step": 16350
},
{
"epoch": 6.08,
"learning_rate": 1.756530612244898e-05,
"loss": 0.0066,
"step": 16400
},
{
"epoch": 6.09,
"learning_rate": 1.746326530612245e-05,
"loss": 0.0055,
"step": 16450
},
{
"epoch": 6.09,
"learning_rate": 1.736122448979592e-05,
"loss": 0.0056,
"step": 16500
},
{
"epoch": 6.09,
"learning_rate": 1.725918367346939e-05,
"loss": 0.0052,
"step": 16550
},
{
"epoch": 6.09,
"learning_rate": 1.715714285714286e-05,
"loss": 0.0059,
"step": 16600
},
{
"epoch": 6.09,
"learning_rate": 1.705510204081633e-05,
"loss": 0.0055,
"step": 16650
},
{
"epoch": 7.0,
"learning_rate": 1.6953061224489798e-05,
"loss": 0.0063,
"step": 16700
},
{
"epoch": 7.0,
"learning_rate": 1.6851020408163266e-05,
"loss": 0.0049,
"step": 16750
},
{
"epoch": 7.0,
"learning_rate": 1.6748979591836738e-05,
"loss": 0.0049,
"step": 16800
},
{
"epoch": 7.01,
"learning_rate": 1.6646938775510206e-05,
"loss": 0.0056,
"step": 16850
},
{
"epoch": 7.01,
"learning_rate": 1.6544897959183674e-05,
"loss": 0.0047,
"step": 16900
},
{
"epoch": 7.01,
"learning_rate": 1.6442857142857142e-05,
"loss": 0.0046,
"step": 16950
},
{
"epoch": 7.01,
"learning_rate": 1.6340816326530613e-05,
"loss": 0.0053,
"step": 17000
},
{
"epoch": 7.01,
"learning_rate": 1.623877551020408e-05,
"loss": 0.0059,
"step": 17050
},
{
"epoch": 7.02,
"learning_rate": 1.613673469387755e-05,
"loss": 0.005,
"step": 17100
},
{
"epoch": 7.02,
"learning_rate": 1.603469387755102e-05,
"loss": 0.0054,
"step": 17150
},
{
"epoch": 7.02,
"learning_rate": 1.593265306122449e-05,
"loss": 0.0039,
"step": 17200
},
{
"epoch": 7.02,
"learning_rate": 1.583061224489796e-05,
"loss": 0.0051,
"step": 17250
},
{
"epoch": 7.02,
"learning_rate": 1.572857142857143e-05,
"loss": 0.0039,
"step": 17300
},
{
"epoch": 7.03,
"learning_rate": 1.56265306122449e-05,
"loss": 0.0049,
"step": 17350
},
{
"epoch": 7.03,
"learning_rate": 1.5524489795918368e-05,
"loss": 0.0049,
"step": 17400
},
{
"epoch": 7.03,
"learning_rate": 1.542244897959184e-05,
"loss": 0.004,
"step": 17450
},
{
"epoch": 7.03,
"learning_rate": 1.5320408163265308e-05,
"loss": 0.0043,
"step": 17500
},
{
"epoch": 7.03,
"eval_loss": 0.20759069919586182,
"eval_runtime": 3054.4242,
"eval_samples_per_second": 4.912,
"eval_steps_per_second": 0.154,
"eval_wer": 9.68597172017861,
"step": 17500
},
{
"epoch": 7.03,
"learning_rate": 1.5218367346938776e-05,
"loss": 0.0046,
"step": 17550
},
{
"epoch": 7.04,
"learning_rate": 1.5116326530612245e-05,
"loss": 0.0029,
"step": 17600
},
{
"epoch": 7.04,
"learning_rate": 1.5016326530612246e-05,
"loss": 0.0035,
"step": 17650
},
{
"epoch": 7.04,
"learning_rate": 1.4914285714285715e-05,
"loss": 0.0036,
"step": 17700
},
{
"epoch": 7.04,
"learning_rate": 1.4812244897959184e-05,
"loss": 0.0042,
"step": 17750
},
{
"epoch": 7.04,
"learning_rate": 1.4710204081632652e-05,
"loss": 0.0036,
"step": 17800
},
{
"epoch": 7.05,
"learning_rate": 1.4608163265306124e-05,
"loss": 0.0036,
"step": 17850
},
{
"epoch": 7.05,
"learning_rate": 1.4506122448979592e-05,
"loss": 0.0043,
"step": 17900
},
{
"epoch": 7.05,
"learning_rate": 1.4404081632653063e-05,
"loss": 0.0039,
"step": 17950
},
{
"epoch": 7.05,
"learning_rate": 1.4302040816326531e-05,
"loss": 0.0034,
"step": 18000
},
{
"epoch": 7.05,
"learning_rate": 1.42e-05,
"loss": 0.0035,
"step": 18050
},
{
"epoch": 7.06,
"learning_rate": 1.4097959183673471e-05,
"loss": 0.0024,
"step": 18100
},
{
"epoch": 7.06,
"learning_rate": 1.3995918367346939e-05,
"loss": 0.0024,
"step": 18150
},
{
"epoch": 7.06,
"learning_rate": 1.3893877551020409e-05,
"loss": 0.0031,
"step": 18200
},
{
"epoch": 7.06,
"learning_rate": 1.3791836734693877e-05,
"loss": 0.0036,
"step": 18250
},
{
"epoch": 7.06,
"learning_rate": 1.3689795918367348e-05,
"loss": 0.0031,
"step": 18300
},
{
"epoch": 7.07,
"learning_rate": 1.3587755102040817e-05,
"loss": 0.0042,
"step": 18350
},
{
"epoch": 7.07,
"learning_rate": 1.3485714285714288e-05,
"loss": 0.0029,
"step": 18400
},
{
"epoch": 7.07,
"learning_rate": 1.3383673469387756e-05,
"loss": 0.0041,
"step": 18450
},
{
"epoch": 7.07,
"learning_rate": 1.3281632653061224e-05,
"loss": 0.0034,
"step": 18500
},
{
"epoch": 7.07,
"learning_rate": 1.3179591836734696e-05,
"loss": 0.0033,
"step": 18550
},
{
"epoch": 7.08,
"learning_rate": 1.3077551020408164e-05,
"loss": 0.0062,
"step": 18600
},
{
"epoch": 7.08,
"learning_rate": 1.2975510204081634e-05,
"loss": 0.0041,
"step": 18650
},
{
"epoch": 7.08,
"learning_rate": 1.2873469387755102e-05,
"loss": 0.0032,
"step": 18700
},
{
"epoch": 7.08,
"learning_rate": 1.2771428571428573e-05,
"loss": 0.0038,
"step": 18750
},
{
"epoch": 7.08,
"learning_rate": 1.2669387755102041e-05,
"loss": 0.0041,
"step": 18800
},
{
"epoch": 7.09,
"learning_rate": 1.2567346938775513e-05,
"loss": 0.0029,
"step": 18850
},
{
"epoch": 7.09,
"learning_rate": 1.246530612244898e-05,
"loss": 0.003,
"step": 18900
},
{
"epoch": 7.09,
"learning_rate": 1.236326530612245e-05,
"loss": 0.0037,
"step": 18950
},
{
"epoch": 7.09,
"learning_rate": 1.226122448979592e-05,
"loss": 0.0029,
"step": 19000
},
{
"epoch": 7.09,
"learning_rate": 1.2159183673469388e-05,
"loss": 0.0028,
"step": 19050
},
{
"epoch": 8.0,
"learning_rate": 1.2057142857142856e-05,
"loss": 0.0038,
"step": 19100
},
{
"epoch": 8.0,
"learning_rate": 1.1955102040816326e-05,
"loss": 0.003,
"step": 19150
},
{
"epoch": 8.0,
"learning_rate": 1.1853061224489796e-05,
"loss": 0.0032,
"step": 19200
},
{
"epoch": 8.01,
"learning_rate": 1.1751020408163266e-05,
"loss": 0.0041,
"step": 19250
},
{
"epoch": 8.01,
"learning_rate": 1.1648979591836736e-05,
"loss": 0.0034,
"step": 19300
},
{
"epoch": 8.01,
"learning_rate": 1.1546938775510205e-05,
"loss": 0.0026,
"step": 19350
},
{
"epoch": 8.01,
"learning_rate": 1.1444897959183675e-05,
"loss": 0.0031,
"step": 19400
},
{
"epoch": 8.01,
"learning_rate": 1.1342857142857143e-05,
"loss": 0.0026,
"step": 19450
},
{
"epoch": 8.02,
"learning_rate": 1.1240816326530613e-05,
"loss": 0.003,
"step": 19500
},
{
"epoch": 8.02,
"learning_rate": 1.1138775510204081e-05,
"loss": 0.0024,
"step": 19550
},
{
"epoch": 8.02,
"learning_rate": 1.103673469387755e-05,
"loss": 0.0025,
"step": 19600
},
{
"epoch": 8.02,
"learning_rate": 1.093469387755102e-05,
"loss": 0.0025,
"step": 19650
},
{
"epoch": 8.02,
"learning_rate": 1.083265306122449e-05,
"loss": 0.0024,
"step": 19700
},
{
"epoch": 8.03,
"learning_rate": 1.073061224489796e-05,
"loss": 0.0021,
"step": 19750
},
{
"epoch": 8.03,
"learning_rate": 1.062857142857143e-05,
"loss": 0.0018,
"step": 19800
},
{
"epoch": 8.03,
"learning_rate": 1.05265306122449e-05,
"loss": 0.0023,
"step": 19850
},
{
"epoch": 8.03,
"learning_rate": 1.0424489795918368e-05,
"loss": 0.0023,
"step": 19900
},
{
"epoch": 8.03,
"learning_rate": 1.0322448979591836e-05,
"loss": 0.0023,
"step": 19950
},
{
"epoch": 8.04,
"learning_rate": 1.0220408163265306e-05,
"loss": 0.0018,
"step": 20000
},
{
"epoch": 8.04,
"eval_loss": 0.20652559399604797,
"eval_runtime": 3091.4094,
"eval_samples_per_second": 4.853,
"eval_steps_per_second": 0.152,
"eval_wer": 9.423506272591963,
"step": 20000
},
{
"epoch": 8.04,
"learning_rate": 1.0118367346938775e-05,
"loss": 0.0021,
"step": 20050
},
{
"epoch": 8.04,
"learning_rate": 1.0016326530612245e-05,
"loss": 0.002,
"step": 20100
},
{
"epoch": 8.04,
"learning_rate": 9.914285714285715e-06,
"loss": 0.0016,
"step": 20150
},
{
"epoch": 8.04,
"learning_rate": 9.812244897959185e-06,
"loss": 0.0018,
"step": 20200
},
{
"epoch": 8.05,
"learning_rate": 9.710204081632655e-06,
"loss": 0.0022,
"step": 20250
},
{
"epoch": 8.05,
"learning_rate": 9.608163265306124e-06,
"loss": 0.0021,
"step": 20300
},
{
"epoch": 8.05,
"learning_rate": 9.506122448979592e-06,
"loss": 0.0022,
"step": 20350
},
{
"epoch": 8.05,
"learning_rate": 9.40408163265306e-06,
"loss": 0.0018,
"step": 20400
},
{
"epoch": 8.05,
"learning_rate": 9.30204081632653e-06,
"loss": 0.0017,
"step": 20450
},
{
"epoch": 8.06,
"learning_rate": 9.2e-06,
"loss": 0.0018,
"step": 20500
},
{
"epoch": 8.06,
"learning_rate": 9.09795918367347e-06,
"loss": 0.0014,
"step": 20550
},
{
"epoch": 8.06,
"learning_rate": 8.99591836734694e-06,
"loss": 0.0013,
"step": 20600
},
{
"epoch": 8.06,
"learning_rate": 8.89387755102041e-06,
"loss": 0.0017,
"step": 20650
},
{
"epoch": 8.06,
"learning_rate": 8.791836734693879e-06,
"loss": 0.0015,
"step": 20700
},
{
"epoch": 8.07,
"learning_rate": 8.689795918367347e-06,
"loss": 0.0021,
"step": 20750
},
{
"epoch": 8.07,
"learning_rate": 8.587755102040817e-06,
"loss": 0.0019,
"step": 20800
},
{
"epoch": 8.07,
"learning_rate": 8.485714285714285e-06,
"loss": 0.0019,
"step": 20850
},
{
"epoch": 8.07,
"learning_rate": 8.383673469387755e-06,
"loss": 0.0015,
"step": 20900
},
{
"epoch": 8.07,
"learning_rate": 8.281632653061225e-06,
"loss": 0.0019,
"step": 20950
},
{
"epoch": 8.08,
"learning_rate": 8.179591836734694e-06,
"loss": 0.0026,
"step": 21000
},
{
"epoch": 8.08,
"learning_rate": 8.077551020408164e-06,
"loss": 0.0021,
"step": 21050
},
{
"epoch": 8.08,
"learning_rate": 7.975510204081634e-06,
"loss": 0.002,
"step": 21100
},
{
"epoch": 8.08,
"learning_rate": 7.873469387755104e-06,
"loss": 0.0021,
"step": 21150
},
{
"epoch": 8.08,
"learning_rate": 7.771428571428572e-06,
"loss": 0.0019,
"step": 21200
},
{
"epoch": 8.09,
"learning_rate": 7.66938775510204e-06,
"loss": 0.0015,
"step": 21250
},
{
"epoch": 8.09,
"learning_rate": 7.5673469387755105e-06,
"loss": 0.0025,
"step": 21300
},
{
"epoch": 8.09,
"learning_rate": 7.4653061224489794e-06,
"loss": 0.0018,
"step": 21350
},
{
"epoch": 8.09,
"learning_rate": 7.363265306122449e-06,
"loss": 0.0014,
"step": 21400
},
{
"epoch": 8.09,
"learning_rate": 7.261224489795919e-06,
"loss": 0.0014,
"step": 21450
},
{
"epoch": 9.0,
"learning_rate": 7.159183673469389e-06,
"loss": 0.0017,
"step": 21500
},
{
"epoch": 9.0,
"learning_rate": 7.057142857142858e-06,
"loss": 0.0017,
"step": 21550
},
{
"epoch": 9.01,
"learning_rate": 6.9551020408163275e-06,
"loss": 0.0023,
"step": 21600
},
{
"epoch": 9.01,
"learning_rate": 6.8530612244897956e-06,
"loss": 0.0017,
"step": 21650
},
{
"epoch": 9.01,
"learning_rate": 6.751020408163265e-06,
"loss": 0.0013,
"step": 21700
},
{
"epoch": 9.01,
"learning_rate": 6.648979591836735e-06,
"loss": 0.0017,
"step": 21750
},
{
"epoch": 9.01,
"learning_rate": 6.546938775510204e-06,
"loss": 0.0015,
"step": 21800
},
{
"epoch": 9.02,
"learning_rate": 6.444897959183674e-06,
"loss": 0.0013,
"step": 21850
},
{
"epoch": 9.02,
"learning_rate": 6.342857142857144e-06,
"loss": 0.0014,
"step": 21900
},
{
"epoch": 9.02,
"learning_rate": 6.2408163265306125e-06,
"loss": 0.0014,
"step": 21950
},
{
"epoch": 9.02,
"learning_rate": 6.1387755102040815e-06,
"loss": 0.0014,
"step": 22000
},
{
"epoch": 9.02,
"learning_rate": 6.036734693877551e-06,
"loss": 0.0014,
"step": 22050
},
{
"epoch": 9.03,
"learning_rate": 5.934693877551021e-06,
"loss": 0.0013,
"step": 22100
},
{
"epoch": 9.03,
"learning_rate": 5.83265306122449e-06,
"loss": 0.0014,
"step": 22150
},
{
"epoch": 9.03,
"learning_rate": 5.73061224489796e-06,
"loss": 0.0008,
"step": 22200
},
{
"epoch": 9.03,
"learning_rate": 5.628571428571429e-06,
"loss": 0.0012,
"step": 22250
},
{
"epoch": 9.03,
"learning_rate": 5.5265306122448984e-06,
"loss": 0.0015,
"step": 22300
},
{
"epoch": 9.04,
"learning_rate": 5.424489795918367e-06,
"loss": 0.0011,
"step": 22350
},
{
"epoch": 9.04,
"learning_rate": 5.322448979591837e-06,
"loss": 0.0009,
"step": 22400
},
{
"epoch": 9.04,
"learning_rate": 5.220408163265306e-06,
"loss": 0.0009,
"step": 22450
},
{
"epoch": 9.04,
"learning_rate": 5.118367346938776e-06,
"loss": 0.0013,
"step": 22500
},
{
"epoch": 9.04,
"eval_loss": 0.20026759803295135,
"eval_runtime": 3101.8937,
"eval_samples_per_second": 4.837,
"eval_steps_per_second": 0.151,
"eval_wer": 8.910535828194769,
"step": 22500
},
{
"epoch": 9.04,
"learning_rate": 5.016326530612246e-06,
"loss": 0.0018,
"step": 22550
},
{
"epoch": 9.05,
"learning_rate": 4.9142857142857145e-06,
"loss": 0.0008,
"step": 22600
},
{
"epoch": 9.05,
"learning_rate": 4.8122448979591835e-06,
"loss": 0.0016,
"step": 22650
},
{
"epoch": 9.05,
"learning_rate": 4.710204081632653e-06,
"loss": 0.0011,
"step": 22700
},
{
"epoch": 9.05,
"learning_rate": 4.608163265306123e-06,
"loss": 0.0014,
"step": 22750
},
{
"epoch": 9.05,
"learning_rate": 4.506122448979592e-06,
"loss": 0.0008,
"step": 22800
},
{
"epoch": 9.06,
"learning_rate": 4.404081632653062e-06,
"loss": 0.0009,
"step": 22850
},
{
"epoch": 9.06,
"learning_rate": 4.302040816326531e-06,
"loss": 0.0009,
"step": 22900
},
{
"epoch": 9.06,
"learning_rate": 4.2000000000000004e-06,
"loss": 0.0007,
"step": 22950
},
{
"epoch": 9.06,
"learning_rate": 4.097959183673469e-06,
"loss": 0.0012,
"step": 23000
},
{
"epoch": 9.06,
"learning_rate": 3.995918367346939e-06,
"loss": 0.0008,
"step": 23050
},
{
"epoch": 9.07,
"learning_rate": 3.893877551020408e-06,
"loss": 0.0009,
"step": 23100
},
{
"epoch": 9.07,
"learning_rate": 3.791836734693878e-06,
"loss": 0.001,
"step": 23150
},
{
"epoch": 9.07,
"learning_rate": 3.689795918367347e-06,
"loss": 0.0011,
"step": 23200
},
{
"epoch": 9.07,
"learning_rate": 3.5877551020408166e-06,
"loss": 0.0008,
"step": 23250
},
{
"epoch": 9.07,
"learning_rate": 3.485714285714286e-06,
"loss": 0.0007,
"step": 23300
},
{
"epoch": 9.08,
"learning_rate": 3.3836734693877553e-06,
"loss": 0.0014,
"step": 23350
},
{
"epoch": 9.08,
"learning_rate": 3.281632653061224e-06,
"loss": 0.0019,
"step": 23400
},
{
"epoch": 9.08,
"learning_rate": 3.179591836734694e-06,
"loss": 0.0011,
"step": 23450
},
{
"epoch": 9.08,
"learning_rate": 3.0775510204081633e-06,
"loss": 0.0007,
"step": 23500
},
{
"epoch": 9.08,
"learning_rate": 2.975510204081633e-06,
"loss": 0.0013,
"step": 23550
},
{
"epoch": 9.09,
"learning_rate": 2.873469387755102e-06,
"loss": 0.0006,
"step": 23600
},
{
"epoch": 9.09,
"learning_rate": 2.771428571428572e-06,
"loss": 0.0009,
"step": 23650
},
{
"epoch": 9.09,
"learning_rate": 2.6693877551020407e-06,
"loss": 0.001,
"step": 23700
},
{
"epoch": 9.09,
"learning_rate": 2.5673469387755105e-06,
"loss": 0.0008,
"step": 23750
},
{
"epoch": 9.09,
"learning_rate": 2.4653061224489795e-06,
"loss": 0.001,
"step": 23800
},
{
"epoch": 9.1,
"learning_rate": 2.3632653061224492e-06,
"loss": 0.0013,
"step": 23850
},
{
"epoch": 10.0,
"learning_rate": 2.2612244897959186e-06,
"loss": 0.0011,
"step": 23900
},
{
"epoch": 10.0,
"learning_rate": 2.159183673469388e-06,
"loss": 0.0012,
"step": 23950
},
{
"epoch": 10.01,
"learning_rate": 2.0571428571428573e-06,
"loss": 0.0009,
"step": 24000
},
{
"epoch": 10.01,
"learning_rate": 1.9551020408163266e-06,
"loss": 0.0007,
"step": 24050
},
{
"epoch": 10.01,
"learning_rate": 1.8530612244897958e-06,
"loss": 0.0007,
"step": 24100
},
{
"epoch": 10.01,
"learning_rate": 1.7510204081632654e-06,
"loss": 0.001,
"step": 24150
},
{
"epoch": 10.01,
"learning_rate": 1.6489795918367347e-06,
"loss": 0.0009,
"step": 24200
},
{
"epoch": 10.02,
"learning_rate": 1.5469387755102043e-06,
"loss": 0.0007,
"step": 24250
},
{
"epoch": 10.02,
"learning_rate": 1.4448979591836736e-06,
"loss": 0.0008,
"step": 24300
},
{
"epoch": 10.02,
"learning_rate": 1.3448979591836735e-06,
"loss": 0.0008,
"step": 24350
},
{
"epoch": 10.02,
"learning_rate": 1.2428571428571429e-06,
"loss": 0.0009,
"step": 24400
},
{
"epoch": 10.02,
"learning_rate": 1.1408163265306124e-06,
"loss": 0.0007,
"step": 24450
},
{
"epoch": 10.03,
"learning_rate": 1.0387755102040818e-06,
"loss": 0.0005,
"step": 24500
},
{
"epoch": 10.03,
"learning_rate": 9.367346938775511e-07,
"loss": 0.0007,
"step": 24550
},
{
"epoch": 10.03,
"learning_rate": 8.346938775510205e-07,
"loss": 0.0008,
"step": 24600
},
{
"epoch": 10.03,
"learning_rate": 7.326530612244899e-07,
"loss": 0.0007,
"step": 24650
},
{
"epoch": 10.03,
"learning_rate": 6.306122448979592e-07,
"loss": 0.0007,
"step": 24700
},
{
"epoch": 10.04,
"learning_rate": 5.285714285714287e-07,
"loss": 0.0006,
"step": 24750
},
{
"epoch": 10.04,
"learning_rate": 4.2653061224489797e-07,
"loss": 0.0006,
"step": 24800
},
{
"epoch": 10.04,
"learning_rate": 3.244897959183674e-07,
"loss": 0.0006,
"step": 24850
},
{
"epoch": 10.04,
"learning_rate": 2.2244897959183673e-07,
"loss": 0.0008,
"step": 24900
},
{
"epoch": 10.04,
"learning_rate": 1.2040816326530614e-07,
"loss": 0.0007,
"step": 24950
},
{
"epoch": 10.05,
"learning_rate": 1.836734693877551e-08,
"loss": 0.0009,
"step": 25000
},
{
"epoch": 10.05,
"eval_loss": 0.19784961640834808,
"eval_runtime": 3107.915,
"eval_samples_per_second": 4.827,
"eval_steps_per_second": 0.151,
"eval_wer": 8.749734212204975,
"step": 25000
},
{
"epoch": 10.05,
"step": 25000,
"total_flos": 4.616385214758912e+20,
"train_loss": 0.05547703759074211,
"train_runtime": 125201.2716,
"train_samples_per_second": 12.779,
"train_steps_per_second": 0.2
}
],
"max_steps": 25000,
"num_train_epochs": 9223372036854775807,
"total_flos": 4.616385214758912e+20,
"trial_name": null,
"trial_params": null
}