stage2-2560 / trainer_state.json
gowitheflow1998
upload checkpoint
7dc606c
raw
history blame
18.9 kB
{
"best_metric": 0.7675637669011621,
"best_model_checkpoint": "./checkpoints/clip-stage2-2560/checkpoint-1950",
"epoch": 1.0044351682755022,
"global_step": 7700,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 3.75e-06,
"loss": 3.5265,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 7.5e-06,
"loss": 1.4349,
"step": 100
},
{
"epoch": 0.02,
"learning_rate": 1.125e-05,
"loss": 1.1217,
"step": 150
},
{
"epoch": 0.03,
"learning_rate": 1.5e-05,
"loss": 0.9885,
"step": 200
},
{
"epoch": 0.03,
"learning_rate": 1.8750000000000002e-05,
"loss": 0.8975,
"step": 250
},
{
"epoch": 0.04,
"learning_rate": 2.25e-05,
"loss": 0.8384,
"step": 300
},
{
"epoch": 0.05,
"learning_rate": 2.625e-05,
"loss": 0.7769,
"step": 350
},
{
"epoch": 0.05,
"learning_rate": 3e-05,
"loss": 0.7357,
"step": 400
},
{
"epoch": 0.06,
"learning_rate": 2.9794520547945206e-05,
"loss": 0.7051,
"step": 450
},
{
"epoch": 0.07,
"learning_rate": 2.958904109589041e-05,
"loss": 0.6722,
"step": 500
},
{
"epoch": 0.07,
"learning_rate": 2.938356164383562e-05,
"loss": 0.647,
"step": 550
},
{
"epoch": 0.08,
"learning_rate": 2.9178082191780824e-05,
"loss": 0.6236,
"step": 600
},
{
"epoch": 0.08,
"learning_rate": 2.8972602739726026e-05,
"loss": 0.6035,
"step": 650
},
{
"epoch": 0.09,
"learning_rate": 2.876712328767123e-05,
"loss": 0.5885,
"step": 700
},
{
"epoch": 0.1,
"learning_rate": 2.856164383561644e-05,
"loss": 0.5622,
"step": 750
},
{
"epoch": 0.1,
"learning_rate": 2.8356164383561644e-05,
"loss": 0.5613,
"step": 800
},
{
"epoch": 0.11,
"learning_rate": 2.815068493150685e-05,
"loss": 0.5414,
"step": 850
},
{
"epoch": 0.12,
"learning_rate": 2.7945205479452054e-05,
"loss": 0.536,
"step": 900
},
{
"epoch": 0.12,
"learning_rate": 2.7739726027397263e-05,
"loss": 0.5227,
"step": 950
},
{
"epoch": 0.13,
"learning_rate": 2.7534246575342468e-05,
"loss": 0.5099,
"step": 1000
},
{
"epoch": 0.14,
"learning_rate": 2.732876712328767e-05,
"loss": 0.4992,
"step": 1050
},
{
"epoch": 0.14,
"learning_rate": 2.7123287671232875e-05,
"loss": 0.4937,
"step": 1100
},
{
"epoch": 0.15,
"learning_rate": 2.6917808219178083e-05,
"loss": 0.4914,
"step": 1150
},
{
"epoch": 0.16,
"learning_rate": 2.6712328767123288e-05,
"loss": 0.4758,
"step": 1200
},
{
"epoch": 0.16,
"learning_rate": 2.6506849315068493e-05,
"loss": 0.4722,
"step": 1250
},
{
"epoch": 0.17,
"learning_rate": 2.6301369863013698e-05,
"loss": 0.4673,
"step": 1300
},
{
"epoch": 0.18,
"learning_rate": 2.6095890410958907e-05,
"loss": 0.4639,
"step": 1350
},
{
"epoch": 0.18,
"learning_rate": 2.589041095890411e-05,
"loss": 0.4499,
"step": 1400
},
{
"epoch": 0.19,
"learning_rate": 2.5684931506849317e-05,
"loss": 0.4427,
"step": 1450
},
{
"epoch": 0.2,
"learning_rate": 2.5479452054794518e-05,
"loss": 0.4489,
"step": 1500
},
{
"epoch": 0.2,
"learning_rate": 2.5273972602739727e-05,
"loss": 0.441,
"step": 1550
},
{
"epoch": 0.21,
"learning_rate": 2.5068493150684932e-05,
"loss": 0.4287,
"step": 1600
},
{
"epoch": 0.22,
"learning_rate": 2.4863013698630137e-05,
"loss": 0.4337,
"step": 1650
},
{
"epoch": 0.22,
"learning_rate": 2.4657534246575342e-05,
"loss": 0.4283,
"step": 1700
},
{
"epoch": 0.23,
"learning_rate": 2.445205479452055e-05,
"loss": 0.4194,
"step": 1750
},
{
"epoch": 0.23,
"learning_rate": 2.4246575342465755e-05,
"loss": 0.4202,
"step": 1800
},
{
"epoch": 0.24,
"learning_rate": 2.404109589041096e-05,
"loss": 0.4106,
"step": 1850
},
{
"epoch": 0.25,
"learning_rate": 2.3835616438356165e-05,
"loss": 0.4123,
"step": 1900
},
{
"epoch": 0.25,
"learning_rate": 2.363013698630137e-05,
"loss": 0.4021,
"step": 1950
},
{
"epoch": 0.26,
"learning_rate": 2.3424657534246575e-05,
"loss": 0.4012,
"step": 2000
},
{
"epoch": 0.27,
"learning_rate": 2.321917808219178e-05,
"loss": 0.3981,
"step": 2050
},
{
"epoch": 0.27,
"learning_rate": 2.3013698630136985e-05,
"loss": 0.3972,
"step": 2100
},
{
"epoch": 0.28,
"learning_rate": 2.2808219178082194e-05,
"loss": 0.3949,
"step": 2150
},
{
"epoch": 0.29,
"learning_rate": 2.26027397260274e-05,
"loss": 0.3895,
"step": 2200
},
{
"epoch": 0.29,
"learning_rate": 2.2397260273972604e-05,
"loss": 0.3884,
"step": 2250
},
{
"epoch": 0.3,
"learning_rate": 2.219178082191781e-05,
"loss": 0.3849,
"step": 2300
},
{
"epoch": 0.31,
"learning_rate": 2.1986301369863014e-05,
"loss": 0.3824,
"step": 2350
},
{
"epoch": 0.31,
"learning_rate": 2.178082191780822e-05,
"loss": 0.3773,
"step": 2400
},
{
"epoch": 0.32,
"learning_rate": 2.1575342465753424e-05,
"loss": 0.3798,
"step": 2450
},
{
"epoch": 0.33,
"learning_rate": 2.136986301369863e-05,
"loss": 0.3724,
"step": 2500
},
{
"epoch": 0.33,
"learning_rate": 2.1164383561643838e-05,
"loss": 0.3721,
"step": 2550
},
{
"epoch": 0.34,
"learning_rate": 2.0958904109589043e-05,
"loss": 0.3658,
"step": 2600
},
{
"epoch": 0.35,
"learning_rate": 2.0753424657534248e-05,
"loss": 0.3718,
"step": 2650
},
{
"epoch": 0.35,
"learning_rate": 2.0547945205479453e-05,
"loss": 0.3642,
"step": 2700
},
{
"epoch": 0.36,
"learning_rate": 2.034246575342466e-05,
"loss": 0.3568,
"step": 2750
},
{
"epoch": 0.37,
"learning_rate": 2.0136986301369863e-05,
"loss": 0.3645,
"step": 2800
},
{
"epoch": 0.37,
"learning_rate": 1.9931506849315068e-05,
"loss": 0.3551,
"step": 2850
},
{
"epoch": 0.38,
"learning_rate": 1.9726027397260273e-05,
"loss": 0.3568,
"step": 2900
},
{
"epoch": 0.38,
"learning_rate": 1.952054794520548e-05,
"loss": 0.3523,
"step": 2950
},
{
"epoch": 0.39,
"learning_rate": 1.9315068493150686e-05,
"loss": 0.3525,
"step": 3000
},
{
"epoch": 0.4,
"learning_rate": 1.910958904109589e-05,
"loss": 0.3521,
"step": 3050
},
{
"epoch": 0.4,
"learning_rate": 1.8904109589041096e-05,
"loss": 0.348,
"step": 3100
},
{
"epoch": 0.41,
"learning_rate": 1.8698630136986305e-05,
"loss": 0.3475,
"step": 3150
},
{
"epoch": 0.42,
"learning_rate": 1.8493150684931506e-05,
"loss": 0.3402,
"step": 3200
},
{
"epoch": 0.42,
"learning_rate": 1.828767123287671e-05,
"loss": 0.3432,
"step": 3250
},
{
"epoch": 0.43,
"learning_rate": 1.8082191780821916e-05,
"loss": 0.3437,
"step": 3300
},
{
"epoch": 0.44,
"learning_rate": 1.7876712328767125e-05,
"loss": 0.3427,
"step": 3350
},
{
"epoch": 0.44,
"learning_rate": 1.767123287671233e-05,
"loss": 0.3354,
"step": 3400
},
{
"epoch": 0.45,
"learning_rate": 1.7465753424657535e-05,
"loss": 0.3422,
"step": 3450
},
{
"epoch": 0.46,
"learning_rate": 1.726027397260274e-05,
"loss": 0.3389,
"step": 3500
},
{
"epoch": 0.46,
"learning_rate": 1.705479452054795e-05,
"loss": 0.3297,
"step": 3550
},
{
"epoch": 0.47,
"learning_rate": 1.6849315068493153e-05,
"loss": 0.3331,
"step": 3600
},
{
"epoch": 0.48,
"learning_rate": 1.6643835616438355e-05,
"loss": 0.3288,
"step": 3650
},
{
"epoch": 0.48,
"learning_rate": 1.643835616438356e-05,
"loss": 0.3261,
"step": 3700
},
{
"epoch": 0.49,
"learning_rate": 1.623287671232877e-05,
"loss": 0.3303,
"step": 3750
},
{
"epoch": 0.5,
"learning_rate": 1.6027397260273974e-05,
"loss": 0.3302,
"step": 3800
},
{
"epoch": 0.5,
"learning_rate": 1.582191780821918e-05,
"loss": 0.3191,
"step": 3850
},
{
"epoch": 0.51,
"learning_rate": 1.5616438356164384e-05,
"loss": 0.3221,
"step": 3900
},
{
"epoch": 0.52,
"learning_rate": 1.5410958904109592e-05,
"loss": 0.322,
"step": 3950
},
{
"epoch": 0.52,
"learning_rate": 1.5205479452054795e-05,
"loss": 0.3159,
"step": 4000
},
{
"epoch": 0.53,
"learning_rate": 1.5e-05,
"loss": 0.3182,
"step": 4050
},
{
"epoch": 0.53,
"learning_rate": 1.4794520547945205e-05,
"loss": 0.3178,
"step": 4100
},
{
"epoch": 0.54,
"learning_rate": 1.4589041095890412e-05,
"loss": 0.3181,
"step": 4150
},
{
"epoch": 0.55,
"learning_rate": 1.4383561643835615e-05,
"loss": 0.3175,
"step": 4200
},
{
"epoch": 0.55,
"learning_rate": 1.4178082191780822e-05,
"loss": 0.3137,
"step": 4250
},
{
"epoch": 0.56,
"learning_rate": 1.3972602739726027e-05,
"loss": 0.3143,
"step": 4300
},
{
"epoch": 0.57,
"learning_rate": 1.3767123287671234e-05,
"loss": 0.3139,
"step": 4350
},
{
"epoch": 0.57,
"learning_rate": 1.3561643835616437e-05,
"loss": 0.3101,
"step": 4400
},
{
"epoch": 0.58,
"learning_rate": 1.3356164383561644e-05,
"loss": 0.3055,
"step": 4450
},
{
"epoch": 0.59,
"learning_rate": 1.3150684931506849e-05,
"loss": 0.3059,
"step": 4500
},
{
"epoch": 0.59,
"learning_rate": 1.2945205479452056e-05,
"loss": 0.3083,
"step": 4550
},
{
"epoch": 0.6,
"learning_rate": 1.2739726027397259e-05,
"loss": 0.3083,
"step": 4600
},
{
"epoch": 0.61,
"learning_rate": 1.2534246575342466e-05,
"loss": 0.2982,
"step": 4650
},
{
"epoch": 0.61,
"learning_rate": 1.2328767123287671e-05,
"loss": 0.3078,
"step": 4700
},
{
"epoch": 0.62,
"learning_rate": 1.2123287671232878e-05,
"loss": 0.3015,
"step": 4750
},
{
"epoch": 0.63,
"learning_rate": 1.1917808219178083e-05,
"loss": 0.3042,
"step": 4800
},
{
"epoch": 0.63,
"learning_rate": 1.1712328767123288e-05,
"loss": 0.3009,
"step": 4850
},
{
"epoch": 0.64,
"learning_rate": 1.1506849315068493e-05,
"loss": 0.3006,
"step": 4900
},
{
"epoch": 0.65,
"learning_rate": 1.13013698630137e-05,
"loss": 0.3002,
"step": 4950
},
{
"epoch": 0.65,
"learning_rate": 1.1095890410958904e-05,
"loss": 0.2929,
"step": 5000
},
{
"epoch": 0.66,
"learning_rate": 1.089041095890411e-05,
"loss": 0.3001,
"step": 5050
},
{
"epoch": 0.67,
"learning_rate": 1.0684931506849315e-05,
"loss": 0.2982,
"step": 5100
},
{
"epoch": 0.67,
"learning_rate": 1.0479452054794521e-05,
"loss": 0.2952,
"step": 5150
},
{
"epoch": 0.68,
"learning_rate": 1.0273972602739726e-05,
"loss": 0.292,
"step": 5200
},
{
"epoch": 0.68,
"learning_rate": 1.0068493150684931e-05,
"loss": 0.2954,
"step": 5250
},
{
"epoch": 0.69,
"learning_rate": 9.863013698630136e-06,
"loss": 0.2968,
"step": 5300
},
{
"epoch": 0.7,
"learning_rate": 9.657534246575343e-06,
"loss": 0.2896,
"step": 5350
},
{
"epoch": 0.7,
"learning_rate": 9.452054794520548e-06,
"loss": 0.2874,
"step": 5400
},
{
"epoch": 0.71,
"learning_rate": 9.246575342465753e-06,
"loss": 0.2948,
"step": 5450
},
{
"epoch": 0.72,
"learning_rate": 9.041095890410958e-06,
"loss": 0.2891,
"step": 5500
},
{
"epoch": 0.72,
"learning_rate": 8.835616438356165e-06,
"loss": 0.2986,
"step": 5550
},
{
"epoch": 0.73,
"learning_rate": 8.63013698630137e-06,
"loss": 0.2903,
"step": 5600
},
{
"epoch": 0.74,
"learning_rate": 8.424657534246577e-06,
"loss": 0.2878,
"step": 5650
},
{
"epoch": 0.74,
"learning_rate": 8.21917808219178e-06,
"loss": 0.2873,
"step": 5700
},
{
"epoch": 0.75,
"learning_rate": 8.013698630136987e-06,
"loss": 0.2879,
"step": 5750
},
{
"epoch": 0.76,
"learning_rate": 7.808219178082192e-06,
"loss": 0.2866,
"step": 5800
},
{
"epoch": 0.76,
"learning_rate": 7.602739726027398e-06,
"loss": 0.2869,
"step": 5850
},
{
"epoch": 0.77,
"learning_rate": 7.397260273972603e-06,
"loss": 0.2841,
"step": 5900
},
{
"epoch": 0.78,
"learning_rate": 7.191780821917808e-06,
"loss": 0.2814,
"step": 5950
},
{
"epoch": 0.78,
"learning_rate": 6.986301369863014e-06,
"loss": 0.2833,
"step": 6000
},
{
"epoch": 0.79,
"learning_rate": 6.780821917808219e-06,
"loss": 0.2845,
"step": 6050
},
{
"epoch": 0.8,
"learning_rate": 6.5753424657534245e-06,
"loss": 0.2823,
"step": 6100
},
{
"epoch": 0.8,
"learning_rate": 6.3698630136986296e-06,
"loss": 0.28,
"step": 6150
},
{
"epoch": 0.81,
"learning_rate": 6.1643835616438354e-06,
"loss": 0.2837,
"step": 6200
},
{
"epoch": 0.82,
"learning_rate": 5.958904109589041e-06,
"loss": 0.2739,
"step": 6250
},
{
"epoch": 0.82,
"learning_rate": 5.753424657534246e-06,
"loss": 0.2854,
"step": 6300
},
{
"epoch": 0.83,
"learning_rate": 5.547945205479452e-06,
"loss": 0.2797,
"step": 6350
},
{
"epoch": 0.83,
"learning_rate": 5.342465753424657e-06,
"loss": 0.2797,
"step": 6400
},
{
"epoch": 0.84,
"learning_rate": 5.136986301369863e-06,
"loss": 0.2759,
"step": 6450
},
{
"epoch": 0.85,
"learning_rate": 4.931506849315068e-06,
"loss": 0.2803,
"step": 6500
},
{
"epoch": 0.85,
"learning_rate": 4.726027397260274e-06,
"loss": 0.2772,
"step": 6550
},
{
"epoch": 0.86,
"learning_rate": 4.520547945205479e-06,
"loss": 0.2787,
"step": 6600
},
{
"epoch": 0.87,
"learning_rate": 4.315068493150685e-06,
"loss": 0.269,
"step": 6650
},
{
"epoch": 0.87,
"learning_rate": 4.10958904109589e-06,
"loss": 0.2719,
"step": 6700
},
{
"epoch": 0.88,
"learning_rate": 3.904109589041096e-06,
"loss": 0.2751,
"step": 6750
},
{
"epoch": 0.89,
"learning_rate": 3.6986301369863014e-06,
"loss": 0.2764,
"step": 6800
},
{
"epoch": 0.89,
"learning_rate": 3.493150684931507e-06,
"loss": 0.2747,
"step": 6850
},
{
"epoch": 0.9,
"learning_rate": 3.2876712328767123e-06,
"loss": 0.2728,
"step": 6900
},
{
"epoch": 0.91,
"learning_rate": 3.0821917808219177e-06,
"loss": 0.2708,
"step": 6950
},
{
"epoch": 0.91,
"learning_rate": 2.876712328767123e-06,
"loss": 0.2708,
"step": 7000
},
{
"epoch": 0.92,
"learning_rate": 2.6712328767123286e-06,
"loss": 0.2689,
"step": 7050
},
{
"epoch": 0.93,
"learning_rate": 2.465753424657534e-06,
"loss": 0.2743,
"step": 7100
},
{
"epoch": 0.93,
"learning_rate": 2.2602739726027396e-06,
"loss": 0.266,
"step": 7150
},
{
"epoch": 0.94,
"learning_rate": 2.054794520547945e-06,
"loss": 0.2697,
"step": 7200
},
{
"epoch": 0.95,
"learning_rate": 1.8493150684931507e-06,
"loss": 0.2712,
"step": 7250
},
{
"epoch": 0.95,
"learning_rate": 1.6438356164383561e-06,
"loss": 0.2721,
"step": 7300
},
{
"epoch": 0.96,
"learning_rate": 1.4383561643835616e-06,
"loss": 0.2699,
"step": 7350
},
{
"epoch": 0.97,
"learning_rate": 1.232876712328767e-06,
"loss": 0.2703,
"step": 7400
},
{
"epoch": 0.97,
"learning_rate": 1.0273972602739725e-06,
"loss": 0.2692,
"step": 7450
},
{
"epoch": 0.98,
"learning_rate": 8.219178082191781e-07,
"loss": 0.2675,
"step": 7500
},
{
"epoch": 0.98,
"learning_rate": 6.164383561643835e-07,
"loss": 0.2645,
"step": 7550
},
{
"epoch": 0.99,
"learning_rate": 4.1095890410958903e-07,
"loss": 0.2691,
"step": 7600
},
{
"epoch": 1.0,
"learning_rate": 2.0547945205479452e-07,
"loss": 0.2748,
"step": 7650
},
{
"epoch": 1.0,
"learning_rate": 0.0,
"loss": 0.2504,
"step": 7700
}
],
"max_steps": 7700,
"num_train_epochs": 2,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}