unsup-wa-s64-bs128-lr6 / trainer_state.json
Zhuoxu Huang
init
894a9e8
{
"best_metric": 0.5954492153614562,
"best_model_checkpoint": "contrastive-unsup-wa-pixel-base-mean-64-128-1-3e-6-7600-42-eval/checkpoint-6000",
"epoch": 0.9986859395532195,
"global_step": 7600,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 6.000000000000001e-07,
"loss": 316.45,
"step": 100
},
{
"epoch": 0.03,
"learning_rate": 1.2000000000000002e-06,
"loss": 163.4776,
"step": 200
},
{
"epoch": 0.04,
"learning_rate": 1.8e-06,
"loss": 104.5736,
"step": 300
},
{
"epoch": 0.05,
"learning_rate": 2.4000000000000003e-06,
"loss": 76.69,
"step": 400
},
{
"epoch": 0.07,
"learning_rate": 3e-06,
"loss": 47.9191,
"step": 500
},
{
"epoch": 0.08,
"learning_rate": 2.9577464788732396e-06,
"loss": 18.6751,
"step": 600
},
{
"epoch": 0.09,
"learning_rate": 2.915492957746479e-06,
"loss": 7.4468,
"step": 700
},
{
"epoch": 0.11,
"learning_rate": 2.8732394366197183e-06,
"loss": 4.3921,
"step": 800
},
{
"epoch": 0.12,
"learning_rate": 2.830985915492958e-06,
"loss": 2.7528,
"step": 900
},
{
"epoch": 0.13,
"learning_rate": 2.7887323943661974e-06,
"loss": 2.5501,
"step": 1000
},
{
"epoch": 0.14,
"learning_rate": 2.7464788732394365e-06,
"loss": 1.8815,
"step": 1100
},
{
"epoch": 0.16,
"learning_rate": 2.704225352112676e-06,
"loss": 1.7108,
"step": 1200
},
{
"epoch": 0.17,
"learning_rate": 2.6619718309859156e-06,
"loss": 1.3322,
"step": 1300
},
{
"epoch": 0.18,
"learning_rate": 2.6197183098591547e-06,
"loss": 1.1365,
"step": 1400
},
{
"epoch": 0.2,
"learning_rate": 2.5774647887323943e-06,
"loss": 1.2889,
"step": 1500
},
{
"epoch": 0.21,
"learning_rate": 2.535211267605634e-06,
"loss": 0.9945,
"step": 1600
},
{
"epoch": 0.22,
"learning_rate": 2.4929577464788734e-06,
"loss": 0.9728,
"step": 1700
},
{
"epoch": 0.24,
"learning_rate": 2.450704225352113e-06,
"loss": 0.8538,
"step": 1800
},
{
"epoch": 0.25,
"learning_rate": 2.4084507042253525e-06,
"loss": 1.0389,
"step": 1900
},
{
"epoch": 0.26,
"learning_rate": 2.3661971830985916e-06,
"loss": 0.7072,
"step": 2000
},
{
"epoch": 0.28,
"learning_rate": 2.3239436619718307e-06,
"loss": 0.6604,
"step": 2100
},
{
"epoch": 0.29,
"learning_rate": 2.2816901408450703e-06,
"loss": 0.7912,
"step": 2200
},
{
"epoch": 0.3,
"learning_rate": 2.23943661971831e-06,
"loss": 0.6953,
"step": 2300
},
{
"epoch": 0.32,
"learning_rate": 2.1971830985915494e-06,
"loss": 0.5999,
"step": 2400
},
{
"epoch": 0.33,
"learning_rate": 2.154929577464789e-06,
"loss": 0.6738,
"step": 2500
},
{
"epoch": 0.34,
"learning_rate": 2.112676056338028e-06,
"loss": 0.5731,
"step": 2600
},
{
"epoch": 0.35,
"learning_rate": 2.0704225352112676e-06,
"loss": 0.5415,
"step": 2700
},
{
"epoch": 0.37,
"learning_rate": 2.028169014084507e-06,
"loss": 0.5263,
"step": 2800
},
{
"epoch": 0.38,
"learning_rate": 1.9859154929577467e-06,
"loss": 0.5921,
"step": 2900
},
{
"epoch": 0.39,
"learning_rate": 1.943661971830986e-06,
"loss": 0.5879,
"step": 3000
},
{
"epoch": 0.41,
"learning_rate": 1.9014084507042256e-06,
"loss": 0.4654,
"step": 3100
},
{
"epoch": 0.42,
"learning_rate": 1.8591549295774647e-06,
"loss": 0.4695,
"step": 3200
},
{
"epoch": 0.43,
"learning_rate": 1.8169014084507043e-06,
"loss": 0.4647,
"step": 3300
},
{
"epoch": 0.45,
"learning_rate": 1.7746478873239436e-06,
"loss": 0.4196,
"step": 3400
},
{
"epoch": 0.46,
"learning_rate": 1.7323943661971832e-06,
"loss": 0.4411,
"step": 3500
},
{
"epoch": 0.47,
"learning_rate": 1.6901408450704227e-06,
"loss": 0.3096,
"step": 3600
},
{
"epoch": 0.49,
"learning_rate": 1.6478873239436623e-06,
"loss": 0.4328,
"step": 3700
},
{
"epoch": 0.5,
"learning_rate": 1.6056338028169014e-06,
"loss": 0.3828,
"step": 3800
},
{
"epoch": 0.51,
"learning_rate": 1.5633802816901407e-06,
"loss": 0.3464,
"step": 3900
},
{
"epoch": 0.53,
"learning_rate": 1.5211267605633803e-06,
"loss": 0.3906,
"step": 4000
},
{
"epoch": 0.54,
"learning_rate": 1.4788732394366198e-06,
"loss": 0.3156,
"step": 4100
},
{
"epoch": 0.55,
"learning_rate": 1.4366197183098591e-06,
"loss": 0.2605,
"step": 4200
},
{
"epoch": 0.57,
"learning_rate": 1.3943661971830987e-06,
"loss": 0.4624,
"step": 4300
},
{
"epoch": 0.58,
"learning_rate": 1.352112676056338e-06,
"loss": 0.3546,
"step": 4400
},
{
"epoch": 0.59,
"learning_rate": 1.3098591549295774e-06,
"loss": 0.2539,
"step": 4500
},
{
"epoch": 0.6,
"learning_rate": 1.267605633802817e-06,
"loss": 0.3064,
"step": 4600
},
{
"epoch": 0.62,
"learning_rate": 1.2253521126760565e-06,
"loss": 0.287,
"step": 4700
},
{
"epoch": 0.63,
"learning_rate": 1.1830985915492958e-06,
"loss": 0.305,
"step": 4800
},
{
"epoch": 0.64,
"learning_rate": 1.1408450704225351e-06,
"loss": 0.2211,
"step": 4900
},
{
"epoch": 0.66,
"learning_rate": 1.0985915492957747e-06,
"loss": 0.2849,
"step": 5000
},
{
"epoch": 0.67,
"learning_rate": 1.056338028169014e-06,
"loss": 0.2763,
"step": 5100
},
{
"epoch": 0.68,
"learning_rate": 1.0140845070422536e-06,
"loss": 0.3159,
"step": 5200
},
{
"epoch": 0.7,
"learning_rate": 9.71830985915493e-07,
"loss": 0.3447,
"step": 5300
},
{
"epoch": 0.71,
"learning_rate": 9.295774647887324e-07,
"loss": 0.2292,
"step": 5400
},
{
"epoch": 0.72,
"learning_rate": 8.873239436619718e-07,
"loss": 0.3273,
"step": 5500
},
{
"epoch": 0.74,
"learning_rate": 8.450704225352114e-07,
"loss": 0.247,
"step": 5600
},
{
"epoch": 0.75,
"learning_rate": 8.028169014084507e-07,
"loss": 0.2929,
"step": 5700
},
{
"epoch": 0.76,
"learning_rate": 7.605633802816901e-07,
"loss": 0.2082,
"step": 5800
},
{
"epoch": 0.78,
"learning_rate": 7.183098591549296e-07,
"loss": 0.1899,
"step": 5900
},
{
"epoch": 0.79,
"learning_rate": 6.76056338028169e-07,
"loss": 0.3238,
"step": 6000
},
{
"epoch": 0.8,
"learning_rate": 6.338028169014085e-07,
"loss": 0.3028,
"step": 6100
},
{
"epoch": 0.81,
"learning_rate": 5.915492957746479e-07,
"loss": 0.3298,
"step": 6200
},
{
"epoch": 0.83,
"learning_rate": 5.492957746478873e-07,
"loss": 0.1636,
"step": 6300
},
{
"epoch": 0.84,
"learning_rate": 5.070422535211268e-07,
"loss": 0.1326,
"step": 6400
},
{
"epoch": 0.85,
"learning_rate": 4.647887323943662e-07,
"loss": 0.311,
"step": 6500
},
{
"epoch": 0.87,
"learning_rate": 4.225352112676057e-07,
"loss": 0.1522,
"step": 6600
},
{
"epoch": 0.88,
"learning_rate": 3.8028169014084507e-07,
"loss": 0.2158,
"step": 6700
},
{
"epoch": 0.89,
"learning_rate": 3.380281690140845e-07,
"loss": 0.1998,
"step": 6800
},
{
"epoch": 0.91,
"learning_rate": 2.9577464788732395e-07,
"loss": 0.2165,
"step": 6900
},
{
"epoch": 0.92,
"learning_rate": 2.535211267605634e-07,
"loss": 0.2359,
"step": 7000
},
{
"epoch": 0.93,
"learning_rate": 2.1126760563380284e-07,
"loss": 0.1662,
"step": 7100
},
{
"epoch": 0.95,
"learning_rate": 1.6901408450704225e-07,
"loss": 0.1787,
"step": 7200
},
{
"epoch": 0.96,
"learning_rate": 1.267605633802817e-07,
"loss": 0.1832,
"step": 7300
},
{
"epoch": 0.97,
"learning_rate": 8.450704225352113e-08,
"loss": 0.2574,
"step": 7400
},
{
"epoch": 0.99,
"learning_rate": 4.2253521126760564e-08,
"loss": 0.2527,
"step": 7500
},
{
"epoch": 1.0,
"learning_rate": 0.0,
"loss": 0.2062,
"step": 7600
},
{
"epoch": 1.0,
"step": 7600,
"total_flos": 503654016614400.0,
"train_loss": 10.217359839113135,
"train_runtime": 6308.9115,
"train_samples_per_second": 154.195,
"train_steps_per_second": 1.205
}
],
"max_steps": 7600,
"num_train_epochs": 1,
"total_flos": 503654016614400.0,
"trial_name": null,
"trial_params": null
}