{ "best_metric": 15.510316437482013, "best_model_checkpoint": "./checkpoint-5000", "epoch": 1.0458063166701528, "global_step": 5000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 2.5e-06, "loss": 0.5694, "step": 25 }, { "epoch": 0.01, "learning_rate": 5e-06, "loss": 0.4515, "step": 50 }, { "epoch": 0.02, "learning_rate": 7.500000000000001e-06, "loss": 0.3885, "step": 75 }, { "epoch": 0.02, "learning_rate": 1e-05, "loss": 0.4144, "step": 100 }, { "epoch": 0.03, "learning_rate": 9.948979591836737e-06, "loss": 0.4179, "step": 125 }, { "epoch": 0.03, "learning_rate": 9.89795918367347e-06, "loss": 0.3702, "step": 150 }, { "epoch": 0.04, "learning_rate": 9.846938775510205e-06, "loss": 0.4184, "step": 175 }, { "epoch": 0.04, "learning_rate": 9.795918367346939e-06, "loss": 0.3545, "step": 200 }, { "epoch": 0.05, "learning_rate": 9.744897959183674e-06, "loss": 0.3848, "step": 225 }, { "epoch": 0.05, "learning_rate": 9.693877551020408e-06, "loss": 0.3634, "step": 250 }, { "epoch": 0.06, "learning_rate": 9.642857142857144e-06, "loss": 0.3809, "step": 275 }, { "epoch": 0.06, "learning_rate": 9.591836734693878e-06, "loss": 0.4118, "step": 300 }, { "epoch": 0.07, "learning_rate": 9.540816326530612e-06, "loss": 0.3991, "step": 325 }, { "epoch": 0.07, "learning_rate": 9.489795918367348e-06, "loss": 0.3951, "step": 350 }, { "epoch": 0.08, "learning_rate": 9.438775510204082e-06, "loss": 0.3518, "step": 375 }, { "epoch": 0.08, "learning_rate": 9.387755102040818e-06, "loss": 0.3586, "step": 400 }, { "epoch": 0.09, "learning_rate": 9.336734693877552e-06, "loss": 0.3896, "step": 425 }, { "epoch": 0.09, "learning_rate": 9.285714285714288e-06, "loss": 0.362, "step": 450 }, { "epoch": 0.1, "learning_rate": 9.234693877551022e-06, "loss": 0.3496, "step": 475 }, { "epoch": 0.1, "learning_rate": 9.183673469387756e-06, "loss": 0.3838, "step": 500 }, { "epoch": 0.11, "learning_rate": 9.13265306122449e-06, "loss": 0.3118, "step": 525 }, { "epoch": 0.12, "learning_rate": 9.081632653061225e-06, "loss": 0.3437, "step": 550 }, { "epoch": 0.12, "learning_rate": 9.03061224489796e-06, "loss": 0.3354, "step": 575 }, { "epoch": 0.13, "learning_rate": 8.979591836734695e-06, "loss": 0.3159, "step": 600 }, { "epoch": 0.13, "learning_rate": 8.92857142857143e-06, "loss": 0.3255, "step": 625 }, { "epoch": 0.14, "learning_rate": 8.877551020408163e-06, "loss": 0.3229, "step": 650 }, { "epoch": 0.14, "learning_rate": 8.826530612244899e-06, "loss": 0.3157, "step": 675 }, { "epoch": 0.15, "learning_rate": 8.775510204081633e-06, "loss": 0.3425, "step": 700 }, { "epoch": 0.15, "learning_rate": 8.724489795918369e-06, "loss": 0.3573, "step": 725 }, { "epoch": 0.16, "learning_rate": 8.673469387755103e-06, "loss": 0.3077, "step": 750 }, { "epoch": 0.16, "learning_rate": 8.622448979591837e-06, "loss": 0.3559, "step": 775 }, { "epoch": 0.17, "learning_rate": 8.571428571428571e-06, "loss": 0.35, "step": 800 }, { "epoch": 0.17, "learning_rate": 8.520408163265307e-06, "loss": 0.3363, "step": 825 }, { "epoch": 0.18, "learning_rate": 8.469387755102042e-06, "loss": 0.3092, "step": 850 }, { "epoch": 0.18, "learning_rate": 8.418367346938776e-06, "loss": 0.3278, "step": 875 }, { "epoch": 0.19, "learning_rate": 8.36734693877551e-06, "loss": 0.2868, "step": 900 }, { "epoch": 0.19, "learning_rate": 8.316326530612246e-06, "loss": 0.3615, "step": 925 }, { "epoch": 0.2, "learning_rate": 8.26530612244898e-06, "loss": 0.3296, "step": 950 }, { "epoch": 0.2, "learning_rate": 8.214285714285714e-06, "loss": 0.3642, "step": 975 }, { "epoch": 0.21, "learning_rate": 8.16326530612245e-06, "loss": 0.3465, "step": 1000 }, { "epoch": 0.21, "learning_rate": 8.112244897959184e-06, "loss": 0.315, "step": 1025 }, { "epoch": 0.22, "learning_rate": 8.06122448979592e-06, "loss": 0.3054, "step": 1050 }, { "epoch": 0.22, "learning_rate": 8.010204081632654e-06, "loss": 0.322, "step": 1075 }, { "epoch": 0.23, "learning_rate": 7.959183673469388e-06, "loss": 0.3274, "step": 1100 }, { "epoch": 0.24, "learning_rate": 7.908163265306124e-06, "loss": 0.3102, "step": 1125 }, { "epoch": 0.24, "learning_rate": 7.857142857142858e-06, "loss": 0.3205, "step": 1150 }, { "epoch": 0.25, "learning_rate": 7.806122448979593e-06, "loss": 0.2956, "step": 1175 }, { "epoch": 0.25, "learning_rate": 7.755102040816327e-06, "loss": 0.3074, "step": 1200 }, { "epoch": 0.26, "learning_rate": 7.704081632653061e-06, "loss": 0.2963, "step": 1225 }, { "epoch": 0.26, "learning_rate": 7.653061224489796e-06, "loss": 0.2707, "step": 1250 }, { "epoch": 0.27, "learning_rate": 7.602040816326531e-06, "loss": 0.2859, "step": 1275 }, { "epoch": 0.27, "learning_rate": 7.551020408163265e-06, "loss": 0.3162, "step": 1300 }, { "epoch": 0.28, "learning_rate": 7.500000000000001e-06, "loss": 0.2726, "step": 1325 }, { "epoch": 0.28, "learning_rate": 7.448979591836736e-06, "loss": 0.2982, "step": 1350 }, { "epoch": 0.29, "learning_rate": 7.39795918367347e-06, "loss": 0.2991, "step": 1375 }, { "epoch": 0.29, "learning_rate": 7.346938775510205e-06, "loss": 0.3111, "step": 1400 }, { "epoch": 0.3, "learning_rate": 7.295918367346939e-06, "loss": 0.2848, "step": 1425 }, { "epoch": 0.3, "learning_rate": 7.244897959183675e-06, "loss": 0.2626, "step": 1450 }, { "epoch": 0.31, "learning_rate": 7.193877551020409e-06, "loss": 0.311, "step": 1475 }, { "epoch": 0.31, "learning_rate": 7.1428571428571436e-06, "loss": 0.2579, "step": 1500 }, { "epoch": 0.32, "learning_rate": 7.091836734693878e-06, "loss": 0.3058, "step": 1525 }, { "epoch": 0.32, "learning_rate": 7.0408163265306125e-06, "loss": 0.3455, "step": 1550 }, { "epoch": 0.33, "learning_rate": 6.989795918367348e-06, "loss": 0.29, "step": 1575 }, { "epoch": 0.33, "learning_rate": 6.938775510204082e-06, "loss": 0.2895, "step": 1600 }, { "epoch": 0.34, "learning_rate": 6.887755102040817e-06, "loss": 0.2734, "step": 1625 }, { "epoch": 0.35, "learning_rate": 6.836734693877551e-06, "loss": 0.3142, "step": 1650 }, { "epoch": 0.35, "learning_rate": 6.785714285714287e-06, "loss": 0.2921, "step": 1675 }, { "epoch": 0.36, "learning_rate": 6.734693877551021e-06, "loss": 0.291, "step": 1700 }, { "epoch": 0.36, "learning_rate": 6.683673469387756e-06, "loss": 0.2835, "step": 1725 }, { "epoch": 0.37, "learning_rate": 6.63265306122449e-06, "loss": 0.2948, "step": 1750 }, { "epoch": 0.37, "learning_rate": 6.581632653061225e-06, "loss": 0.2803, "step": 1775 }, { "epoch": 0.38, "learning_rate": 6.530612244897959e-06, "loss": 0.2953, "step": 1800 }, { "epoch": 0.38, "learning_rate": 6.4795918367346946e-06, "loss": 0.2619, "step": 1825 }, { "epoch": 0.39, "learning_rate": 6.4285714285714295e-06, "loss": 0.2513, "step": 1850 }, { "epoch": 0.39, "learning_rate": 6.3775510204081635e-06, "loss": 0.3014, "step": 1875 }, { "epoch": 0.4, "learning_rate": 6.326530612244899e-06, "loss": 0.2828, "step": 1900 }, { "epoch": 0.4, "learning_rate": 6.277551020408164e-06, "loss": 0.2965, "step": 1925 }, { "epoch": 0.41, "learning_rate": 6.2265306122448985e-06, "loss": 0.2888, "step": 1950 }, { "epoch": 0.41, "learning_rate": 6.1755102040816325e-06, "loss": 0.2891, "step": 1975 }, { "epoch": 0.42, "learning_rate": 6.124489795918368e-06, "loss": 0.2899, "step": 2000 }, { "epoch": 0.42, "learning_rate": 6.073469387755102e-06, "loss": 0.3153, "step": 2025 }, { "epoch": 0.43, "learning_rate": 6.022448979591837e-06, "loss": 0.2886, "step": 2050 }, { "epoch": 0.43, "learning_rate": 5.971428571428572e-06, "loss": 0.3079, "step": 2075 }, { "epoch": 0.44, "learning_rate": 5.920408163265306e-06, "loss": 0.2838, "step": 2100 }, { "epoch": 0.44, "learning_rate": 5.869387755102042e-06, "loss": 0.2683, "step": 2125 }, { "epoch": 0.45, "learning_rate": 5.818367346938776e-06, "loss": 0.2598, "step": 2150 }, { "epoch": 0.45, "learning_rate": 5.767346938775511e-06, "loss": 0.2734, "step": 2175 }, { "epoch": 0.46, "learning_rate": 5.716326530612245e-06, "loss": 0.2561, "step": 2200 }, { "epoch": 0.47, "learning_rate": 5.665306122448981e-06, "loss": 0.2801, "step": 2225 }, { "epoch": 0.47, "learning_rate": 5.614285714285715e-06, "loss": 0.289, "step": 2250 }, { "epoch": 0.48, "learning_rate": 5.5632653061224495e-06, "loss": 0.2856, "step": 2275 }, { "epoch": 0.48, "learning_rate": 5.5122448979591836e-06, "loss": 0.2792, "step": 2300 }, { "epoch": 0.49, "learning_rate": 5.4612244897959184e-06, "loss": 0.2714, "step": 2325 }, { "epoch": 0.49, "learning_rate": 5.410204081632654e-06, "loss": 0.2613, "step": 2350 }, { "epoch": 0.5, "learning_rate": 5.359183673469388e-06, "loss": 0.2459, "step": 2375 }, { "epoch": 0.5, "learning_rate": 5.308163265306123e-06, "loss": 0.2633, "step": 2400 }, { "epoch": 0.51, "learning_rate": 5.257142857142857e-06, "loss": 0.2788, "step": 2425 }, { "epoch": 0.51, "learning_rate": 5.206122448979593e-06, "loss": 0.2613, "step": 2450 }, { "epoch": 0.52, "learning_rate": 5.155102040816327e-06, "loss": 0.2533, "step": 2475 }, { "epoch": 0.52, "learning_rate": 5.104081632653062e-06, "loss": 0.2499, "step": 2500 }, { "epoch": 0.53, "learning_rate": 5.053061224489796e-06, "loss": 0.2576, "step": 2525 }, { "epoch": 0.53, "learning_rate": 5.002040816326531e-06, "loss": 0.2903, "step": 2550 }, { "epoch": 0.54, "learning_rate": 4.951020408163266e-06, "loss": 0.2976, "step": 2575 }, { "epoch": 0.54, "learning_rate": 4.9000000000000005e-06, "loss": 0.2575, "step": 2600 }, { "epoch": 0.55, "learning_rate": 4.8489795918367346e-06, "loss": 0.2653, "step": 2625 }, { "epoch": 0.55, "learning_rate": 4.7979591836734694e-06, "loss": 0.2723, "step": 2650 }, { "epoch": 0.56, "learning_rate": 4.746938775510204e-06, "loss": 0.289, "step": 2675 }, { "epoch": 0.56, "learning_rate": 4.695918367346939e-06, "loss": 0.2807, "step": 2700 }, { "epoch": 0.57, "learning_rate": 4.644897959183674e-06, "loss": 0.2523, "step": 2725 }, { "epoch": 0.58, "learning_rate": 4.593877551020409e-06, "loss": 0.2644, "step": 2750 }, { "epoch": 0.58, "learning_rate": 4.542857142857143e-06, "loss": 0.2345, "step": 2775 }, { "epoch": 0.59, "learning_rate": 4.491836734693878e-06, "loss": 0.2555, "step": 2800 }, { "epoch": 0.59, "learning_rate": 4.440816326530613e-06, "loss": 0.2306, "step": 2825 }, { "epoch": 0.6, "learning_rate": 4.389795918367347e-06, "loss": 0.2736, "step": 2850 }, { "epoch": 0.6, "learning_rate": 4.338775510204082e-06, "loss": 0.2209, "step": 2875 }, { "epoch": 0.61, "learning_rate": 4.287755102040817e-06, "loss": 0.2407, "step": 2900 }, { "epoch": 0.61, "learning_rate": 4.236734693877551e-06, "loss": 0.2537, "step": 2925 }, { "epoch": 0.62, "learning_rate": 4.185714285714286e-06, "loss": 0.2529, "step": 2950 }, { "epoch": 0.62, "learning_rate": 4.134693877551021e-06, "loss": 0.2502, "step": 2975 }, { "epoch": 0.63, "learning_rate": 4.083673469387755e-06, "loss": 0.2772, "step": 3000 }, { "epoch": 0.63, "learning_rate": 4.03265306122449e-06, "loss": 0.2715, "step": 3025 }, { "epoch": 0.64, "learning_rate": 3.981632653061225e-06, "loss": 0.2633, "step": 3050 }, { "epoch": 0.64, "learning_rate": 3.930612244897959e-06, "loss": 0.2332, "step": 3075 }, { "epoch": 0.65, "learning_rate": 3.879591836734694e-06, "loss": 0.2396, "step": 3100 }, { "epoch": 0.65, "learning_rate": 3.828571428571429e-06, "loss": 0.2246, "step": 3125 }, { "epoch": 0.66, "learning_rate": 3.7775510204081634e-06, "loss": 0.2437, "step": 3150 }, { "epoch": 0.66, "learning_rate": 3.7265306122448987e-06, "loss": 0.2788, "step": 3175 }, { "epoch": 0.67, "learning_rate": 3.675510204081633e-06, "loss": 0.2517, "step": 3200 }, { "epoch": 0.67, "learning_rate": 3.6244897959183676e-06, "loss": 0.2378, "step": 3225 }, { "epoch": 0.68, "learning_rate": 3.5734693877551025e-06, "loss": 0.2245, "step": 3250 }, { "epoch": 0.69, "learning_rate": 3.522448979591837e-06, "loss": 0.28, "step": 3275 }, { "epoch": 0.69, "learning_rate": 3.471428571428572e-06, "loss": 0.2331, "step": 3300 }, { "epoch": 0.7, "learning_rate": 3.4204081632653064e-06, "loss": 0.2485, "step": 3325 }, { "epoch": 0.7, "learning_rate": 3.369387755102041e-06, "loss": 0.2598, "step": 3350 }, { "epoch": 0.71, "learning_rate": 3.3183673469387757e-06, "loss": 0.2515, "step": 3375 }, { "epoch": 0.71, "learning_rate": 3.26734693877551e-06, "loss": 0.2514, "step": 3400 }, { "epoch": 0.72, "learning_rate": 3.2163265306122455e-06, "loss": 0.2154, "step": 3425 }, { "epoch": 0.72, "learning_rate": 3.16530612244898e-06, "loss": 0.2505, "step": 3450 }, { "epoch": 0.73, "learning_rate": 3.114285714285715e-06, "loss": 0.2575, "step": 3475 }, { "epoch": 0.73, "learning_rate": 3.0632653061224493e-06, "loss": 0.2308, "step": 3500 }, { "epoch": 0.74, "learning_rate": 3.0122448979591838e-06, "loss": 0.2551, "step": 3525 }, { "epoch": 0.74, "learning_rate": 2.9612244897959187e-06, "loss": 0.2545, "step": 3550 }, { "epoch": 0.75, "learning_rate": 2.910204081632653e-06, "loss": 0.2756, "step": 3575 }, { "epoch": 0.75, "learning_rate": 2.859183673469388e-06, "loss": 0.2581, "step": 3600 }, { "epoch": 0.76, "learning_rate": 2.8081632653061225e-06, "loss": 0.2214, "step": 3625 }, { "epoch": 0.76, "learning_rate": 2.757142857142857e-06, "loss": 0.2317, "step": 3650 }, { "epoch": 0.77, "learning_rate": 2.7061224489795923e-06, "loss": 0.2664, "step": 3675 }, { "epoch": 0.77, "learning_rate": 2.655102040816327e-06, "loss": 0.2683, "step": 3700 }, { "epoch": 0.78, "learning_rate": 2.6040816326530616e-06, "loss": 0.2261, "step": 3725 }, { "epoch": 0.78, "learning_rate": 2.553061224489796e-06, "loss": 0.2777, "step": 3750 }, { "epoch": 0.79, "learning_rate": 2.502040816326531e-06, "loss": 0.2201, "step": 3775 }, { "epoch": 0.79, "learning_rate": 2.4510204081632654e-06, "loss": 0.2276, "step": 3800 }, { "epoch": 0.8, "learning_rate": 2.4000000000000003e-06, "loss": 0.2746, "step": 3825 }, { "epoch": 0.81, "learning_rate": 2.348979591836735e-06, "loss": 0.2471, "step": 3850 }, { "epoch": 0.81, "learning_rate": 2.2979591836734697e-06, "loss": 0.2143, "step": 3875 }, { "epoch": 0.82, "learning_rate": 2.246938775510204e-06, "loss": 0.2548, "step": 3900 }, { "epoch": 0.82, "learning_rate": 2.195918367346939e-06, "loss": 0.2196, "step": 3925 }, { "epoch": 0.83, "learning_rate": 2.1448979591836735e-06, "loss": 0.1985, "step": 3950 }, { "epoch": 0.83, "learning_rate": 2.0938775510204084e-06, "loss": 0.2331, "step": 3975 }, { "epoch": 0.84, "learning_rate": 2.0428571428571433e-06, "loss": 0.212, "step": 4000 }, { "epoch": 0.84, "learning_rate": 1.9918367346938777e-06, "loss": 0.2581, "step": 4025 }, { "epoch": 0.85, "learning_rate": 1.940816326530612e-06, "loss": 0.2457, "step": 4050 }, { "epoch": 0.85, "learning_rate": 1.889795918367347e-06, "loss": 0.2307, "step": 4075 }, { "epoch": 0.86, "learning_rate": 1.838775510204082e-06, "loss": 0.2239, "step": 4100 }, { "epoch": 0.86, "learning_rate": 1.7877551020408164e-06, "loss": 0.1908, "step": 4125 }, { "epoch": 0.87, "learning_rate": 1.7367346938775511e-06, "loss": 0.203, "step": 4150 }, { "epoch": 0.87, "learning_rate": 1.6857142857142858e-06, "loss": 0.2384, "step": 4175 }, { "epoch": 0.88, "learning_rate": 1.6346938775510205e-06, "loss": 0.2539, "step": 4200 }, { "epoch": 0.88, "learning_rate": 1.5836734693877554e-06, "loss": 0.2381, "step": 4225 }, { "epoch": 0.89, "learning_rate": 1.53265306122449e-06, "loss": 0.2171, "step": 4250 }, { "epoch": 0.89, "learning_rate": 1.4816326530612247e-06, "loss": 0.2334, "step": 4275 }, { "epoch": 0.9, "learning_rate": 1.4306122448979592e-06, "loss": 0.224, "step": 4300 }, { "epoch": 0.9, "learning_rate": 1.3795918367346938e-06, "loss": 0.2217, "step": 4325 }, { "epoch": 0.91, "learning_rate": 1.3285714285714287e-06, "loss": 0.2293, "step": 4350 }, { "epoch": 0.92, "learning_rate": 1.2775510204081634e-06, "loss": 0.2361, "step": 4375 }, { "epoch": 0.92, "learning_rate": 1.226530612244898e-06, "loss": 0.2361, "step": 4400 }, { "epoch": 0.93, "learning_rate": 1.1755102040816328e-06, "loss": 0.2612, "step": 4425 }, { "epoch": 0.93, "learning_rate": 1.1244897959183674e-06, "loss": 0.2425, "step": 4450 }, { "epoch": 0.94, "learning_rate": 1.0734693877551021e-06, "loss": 0.2548, "step": 4475 }, { "epoch": 0.94, "learning_rate": 1.0224489795918368e-06, "loss": 0.2, "step": 4500 }, { "epoch": 0.95, "learning_rate": 9.714285714285715e-07, "loss": 0.2096, "step": 4525 }, { "epoch": 0.95, "learning_rate": 9.204081632653062e-07, "loss": 0.1915, "step": 4550 }, { "epoch": 0.96, "learning_rate": 8.693877551020409e-07, "loss": 0.2663, "step": 4575 }, { "epoch": 0.96, "learning_rate": 8.183673469387755e-07, "loss": 0.2209, "step": 4600 }, { "epoch": 0.97, "learning_rate": 7.673469387755103e-07, "loss": 0.2243, "step": 4625 }, { "epoch": 0.97, "learning_rate": 7.16326530612245e-07, "loss": 0.2337, "step": 4650 }, { "epoch": 0.98, "learning_rate": 6.653061224489797e-07, "loss": 0.2093, "step": 4675 }, { "epoch": 0.98, "learning_rate": 6.142857142857143e-07, "loss": 0.2325, "step": 4700 }, { "epoch": 0.99, "learning_rate": 5.632653061224491e-07, "loss": 0.2381, "step": 4725 }, { "epoch": 0.99, "learning_rate": 5.122448979591837e-07, "loss": 0.2464, "step": 4750 }, { "epoch": 1.0, "learning_rate": 4.612244897959184e-07, "loss": 0.2377, "step": 4775 }, { "epoch": 1.0, "learning_rate": 4.102040816326531e-07, "loss": 0.1863, "step": 4800 }, { "epoch": 1.01, "learning_rate": 3.591836734693878e-07, "loss": 0.1817, "step": 4825 }, { "epoch": 1.01, "learning_rate": 3.0816326530612243e-07, "loss": 0.1573, "step": 4850 }, { "epoch": 1.02, "learning_rate": 2.5714285714285716e-07, "loss": 0.1797, "step": 4875 }, { "epoch": 1.02, "learning_rate": 2.0612244897959186e-07, "loss": 0.1789, "step": 4900 }, { "epoch": 1.03, "learning_rate": 1.5510204081632654e-07, "loss": 0.1416, "step": 4925 }, { "epoch": 1.04, "learning_rate": 1.0408163265306123e-07, "loss": 0.1457, "step": 4950 }, { "epoch": 1.04, "learning_rate": 5.306122448979593e-08, "loss": 0.1499, "step": 4975 }, { "epoch": 1.05, "learning_rate": 2.0408163265306126e-09, "loss": 0.1845, "step": 5000 }, { "epoch": 1.05, "eval_loss": 0.12443096935749054, "eval_runtime": 3978.521, "eval_samples_per_second": 2.747, "eval_steps_per_second": 0.344, "eval_wer": 15.510316437482013, "step": 5000 }, { "epoch": 1.05, "step": 5000, "total_flos": 1.698473121509376e+20, "train_loss": 0.27737956771850586, "train_runtime": 13393.8488, "train_samples_per_second": 5.973, "train_steps_per_second": 0.373 } ], "max_steps": 5000, "num_train_epochs": 2, "total_flos": 1.698473121509376e+20, "trial_name": null, "trial_params": null }