| { | |
| "best_metric": 0.7275755258839849, | |
| "best_model_checkpoint": "./checkpoints/clip-2560-wikispan-all/checkpoint-240", | |
| "epoch": 0.5982799451576717, | |
| "global_step": 14400, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.8e-06, | |
| "loss": 1.8723, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.6e-06, | |
| "loss": 1.0969, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.4e-06, | |
| "loss": 0.8982, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.2e-06, | |
| "loss": 0.802, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9e-06, | |
| "loss": 0.7304, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.08e-05, | |
| "loss": 0.6815, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.26e-05, | |
| "loss": 0.6435, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.44e-05, | |
| "loss": 0.6089, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.62e-05, | |
| "loss": 0.581, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.8e-05, | |
| "loss": 0.5637, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.98e-05, | |
| "loss": 0.538, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.16e-05, | |
| "loss": 0.5223, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.3400000000000003e-05, | |
| "loss": 0.5123, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.52e-05, | |
| "loss": 0.4984, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 0.4848, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.88e-05, | |
| "loss": 0.4763, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.9945750452079567e-05, | |
| "loss": 0.468, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.978300180831826e-05, | |
| "loss": 0.4532, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.9620253164556963e-05, | |
| "loss": 0.445, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 2.945750452079566e-05, | |
| "loss": 0.4382, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 2.929475587703436e-05, | |
| "loss": 0.4289, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 2.9132007233273057e-05, | |
| "loss": 0.4216, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 2.8969258589511756e-05, | |
| "loss": 0.4134, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.8806509945750454e-05, | |
| "loss": 0.4084, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.864376130198915e-05, | |
| "loss": 0.3988, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.8481012658227846e-05, | |
| "loss": 0.3938, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.8318264014466548e-05, | |
| "loss": 0.3886, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.8155515370705246e-05, | |
| "loss": 0.3845, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.7992766726943944e-05, | |
| "loss": 0.378, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 2.7830018083182642e-05, | |
| "loss": 0.375, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 2.766726943942134e-05, | |
| "loss": 0.3665, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 2.7504520795660035e-05, | |
| "loss": 0.3644, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 2.7341772151898733e-05, | |
| "loss": 0.36, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 2.717902350813743e-05, | |
| "loss": 0.3597, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 2.701627486437613e-05, | |
| "loss": 0.3549, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 2.685352622061483e-05, | |
| "loss": 0.3545, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 2.669077757685353e-05, | |
| "loss": 0.3496, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 2.6528028933092224e-05, | |
| "loss": 0.3442, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 2.6365280289330922e-05, | |
| "loss": 0.3424, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 2.620253164556962e-05, | |
| "loss": 0.3414, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 2.6039783001808318e-05, | |
| "loss": 0.3361, | |
| "step": 4920 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.5877034358047016e-05, | |
| "loss": 0.3378, | |
| "step": 5040 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.5714285714285714e-05, | |
| "loss": 0.3316, | |
| "step": 5160 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 2.5551537070524416e-05, | |
| "loss": 0.3302, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 2.538878842676311e-05, | |
| "loss": 0.3245, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.522603978300181e-05, | |
| "loss": 0.3254, | |
| "step": 5520 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.5063291139240507e-05, | |
| "loss": 0.3206, | |
| "step": 5640 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.4900542495479205e-05, | |
| "loss": 0.3212, | |
| "step": 5760 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.4737793851717903e-05, | |
| "loss": 0.3184, | |
| "step": 5880 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.45750452079566e-05, | |
| "loss": 0.3141, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.44122965641953e-05, | |
| "loss": 0.3185, | |
| "step": 6120 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 2.4249547920433994e-05, | |
| "loss": 0.3115, | |
| "step": 6240 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 2.4086799276672696e-05, | |
| "loss": 0.3093, | |
| "step": 6360 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 2.3924050632911394e-05, | |
| "loss": 0.3116, | |
| "step": 6480 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 2.3761301989150092e-05, | |
| "loss": 0.3084, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 2.359855334538879e-05, | |
| "loss": 0.3071, | |
| "step": 6720 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 2.3435804701627488e-05, | |
| "loss": 0.3047, | |
| "step": 6840 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 2.3273056057866183e-05, | |
| "loss": 0.302, | |
| "step": 6960 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 2.311030741410488e-05, | |
| "loss": 0.3027, | |
| "step": 7080 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 2.294755877034358e-05, | |
| "loss": 0.3019, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 2.278481012658228e-05, | |
| "loss": 0.3, | |
| "step": 7320 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 2.262206148282098e-05, | |
| "loss": 0.2961, | |
| "step": 7440 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 2.2459312839059677e-05, | |
| "loss": 0.2967, | |
| "step": 7560 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.2296564195298375e-05, | |
| "loss": 0.2944, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.213381555153707e-05, | |
| "loss": 0.2937, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.1971066907775768e-05, | |
| "loss": 0.2914, | |
| "step": 7920 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.1808318264014466e-05, | |
| "loss": 0.2902, | |
| "step": 8040 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.1645569620253164e-05, | |
| "loss": 0.2889, | |
| "step": 8160 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.1482820976491862e-05, | |
| "loss": 0.2894, | |
| "step": 8280 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.1320072332730564e-05, | |
| "loss": 0.286, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 2.1157323688969262e-05, | |
| "loss": 0.2847, | |
| "step": 8520 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 2.0994575045207956e-05, | |
| "loss": 0.2838, | |
| "step": 8640 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 2.0831826401446655e-05, | |
| "loss": 0.2822, | |
| "step": 8760 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 2.0669077757685353e-05, | |
| "loss": 0.2814, | |
| "step": 8880 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 2.050632911392405e-05, | |
| "loss": 0.2818, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 2.034358047016275e-05, | |
| "loss": 0.2779, | |
| "step": 9120 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 2.0180831826401447e-05, | |
| "loss": 0.2794, | |
| "step": 9240 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 2.0018083182640145e-05, | |
| "loss": 0.2789, | |
| "step": 9360 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.9855334538878843e-05, | |
| "loss": 0.2757, | |
| "step": 9480 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.969258589511754e-05, | |
| "loss": 0.2758, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.952983725135624e-05, | |
| "loss": 0.277, | |
| "step": 9720 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.9367088607594938e-05, | |
| "loss": 0.2731, | |
| "step": 9840 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.9204339963833636e-05, | |
| "loss": 0.268, | |
| "step": 9960 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.9041591320072334e-05, | |
| "loss": 0.2702, | |
| "step": 10080 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.887884267631103e-05, | |
| "loss": 0.2699, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.8716094032549727e-05, | |
| "loss": 0.2707, | |
| "step": 10320 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.8553345388788428e-05, | |
| "loss": 0.2661, | |
| "step": 10440 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.8390596745027126e-05, | |
| "loss": 0.2668, | |
| "step": 10560 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.8227848101265824e-05, | |
| "loss": 0.2689, | |
| "step": 10680 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.8065099457504523e-05, | |
| "loss": 0.2695, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.790235081374322e-05, | |
| "loss": 0.2637, | |
| "step": 10920 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.7739602169981915e-05, | |
| "loss": 0.2645, | |
| "step": 11040 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.7576853526220614e-05, | |
| "loss": 0.2618, | |
| "step": 11160 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.741410488245931e-05, | |
| "loss": 0.2626, | |
| "step": 11280 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.7251356238698013e-05, | |
| "loss": 0.2581, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.708860759493671e-05, | |
| "loss": 0.2597, | |
| "step": 11520 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.692585895117541e-05, | |
| "loss": 0.2594, | |
| "step": 11640 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.6763110307414104e-05, | |
| "loss": 0.2577, | |
| "step": 11760 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.6600361663652802e-05, | |
| "loss": 0.2585, | |
| "step": 11880 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.64376130198915e-05, | |
| "loss": 0.2565, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.62748643761302e-05, | |
| "loss": 0.2562, | |
| "step": 12120 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.6112115732368897e-05, | |
| "loss": 0.2559, | |
| "step": 12240 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.5949367088607595e-05, | |
| "loss": 0.2551, | |
| "step": 12360 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.5786618444846296e-05, | |
| "loss": 0.2528, | |
| "step": 12480 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.562386980108499e-05, | |
| "loss": 0.2539, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.546112115732369e-05, | |
| "loss": 0.2525, | |
| "step": 12720 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.5298372513562387e-05, | |
| "loss": 0.2534, | |
| "step": 12840 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.5135623869801085e-05, | |
| "loss": 0.2518, | |
| "step": 12960 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.4972875226039783e-05, | |
| "loss": 0.2506, | |
| "step": 13080 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.4810126582278482e-05, | |
| "loss": 0.2517, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.464737793851718e-05, | |
| "loss": 0.2483, | |
| "step": 13320 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.4484629294755878e-05, | |
| "loss": 0.2471, | |
| "step": 13440 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.4321880650994574e-05, | |
| "loss": 0.2484, | |
| "step": 13560 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.4159132007233274e-05, | |
| "loss": 0.2474, | |
| "step": 13680 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.3996383363471972e-05, | |
| "loss": 0.25, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.383363471971067e-05, | |
| "loss": 0.2458, | |
| "step": 13920 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.3670886075949367e-05, | |
| "loss": 0.2466, | |
| "step": 14040 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.3508137432188065e-05, | |
| "loss": 0.2454, | |
| "step": 14160 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.3345388788426765e-05, | |
| "loss": 0.2461, | |
| "step": 14280 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.3182640144665461e-05, | |
| "loss": 0.2427, | |
| "step": 14400 | |
| } | |
| ], | |
| "max_steps": 24120, | |
| "num_train_epochs": 2, | |
| "total_flos": 0.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |