| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 1287, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06, | |
| "eval_loss": 2.3746426105499268, | |
| "eval_runtime": 1198.5011, | |
| "eval_samples_per_second": 10.489, | |
| "eval_steps_per_second": 0.164, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "eval_loss": 2.335777521133423, | |
| "eval_runtime": 1198.5148, | |
| "eval_samples_per_second": 10.489, | |
| "eval_steps_per_second": 0.164, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "eval_loss": 2.3206024169921875, | |
| "eval_runtime": 1198.298, | |
| "eval_samples_per_second": 10.491, | |
| "eval_steps_per_second": 0.164, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "eval_loss": 2.3067309856414795, | |
| "eval_runtime": 1198.4272, | |
| "eval_samples_per_second": 10.49, | |
| "eval_steps_per_second": 0.164, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "eval_loss": 2.295646905899048, | |
| "eval_runtime": 1198.0238, | |
| "eval_samples_per_second": 10.493, | |
| "eval_steps_per_second": 0.164, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "eval_loss": 2.288886308670044, | |
| "eval_runtime": 1198.6077, | |
| "eval_samples_per_second": 10.488, | |
| "eval_steps_per_second": 0.164, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "eval_loss": 2.2820334434509277, | |
| "eval_runtime": 1197.653, | |
| "eval_samples_per_second": 10.496, | |
| "eval_steps_per_second": 0.164, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "eval_loss": 2.275322675704956, | |
| "eval_runtime": 1197.9481, | |
| "eval_samples_per_second": 10.494, | |
| "eval_steps_per_second": 0.164, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "eval_loss": 2.269848108291626, | |
| "eval_runtime": 1197.9985, | |
| "eval_samples_per_second": 10.493, | |
| "eval_steps_per_second": 0.164, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "eval_loss": 2.266172170639038, | |
| "eval_runtime": 1198.9744, | |
| "eval_samples_per_second": 10.485, | |
| "eval_steps_per_second": 0.164, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "eval_loss": 2.2608513832092285, | |
| "eval_runtime": 1198.2986, | |
| "eval_samples_per_second": 10.491, | |
| "eval_steps_per_second": 0.164, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "eval_loss": 2.256772756576538, | |
| "eval_runtime": 1198.6541, | |
| "eval_samples_per_second": 10.488, | |
| "eval_steps_per_second": 0.164, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "eval_loss": 2.2526705265045166, | |
| "eval_runtime": 1198.4379, | |
| "eval_samples_per_second": 10.489, | |
| "eval_steps_per_second": 0.164, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "eval_loss": 2.2487242221832275, | |
| "eval_runtime": 1198.434, | |
| "eval_samples_per_second": 10.49, | |
| "eval_steps_per_second": 0.164, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "eval_loss": 2.2462055683135986, | |
| "eval_runtime": 1198.2638, | |
| "eval_samples_per_second": 10.491, | |
| "eval_steps_per_second": 0.164, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "eval_loss": 2.2430307865142822, | |
| "eval_runtime": 1198.979, | |
| "eval_samples_per_second": 10.485, | |
| "eval_steps_per_second": 0.164, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "eval_loss": 2.240046501159668, | |
| "eval_runtime": 1198.9558, | |
| "eval_samples_per_second": 10.485, | |
| "eval_steps_per_second": 0.164, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "eval_loss": 2.239072799682617, | |
| "eval_runtime": 1199.2167, | |
| "eval_samples_per_second": 10.483, | |
| "eval_steps_per_second": 0.164, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "eval_loss": 2.235506296157837, | |
| "eval_runtime": 1199.5308, | |
| "eval_samples_per_second": 10.48, | |
| "eval_steps_per_second": 0.164, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5e-06, | |
| "loss": 2.1094, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "eval_loss": 2.233208656311035, | |
| "eval_runtime": 1198.8451, | |
| "eval_samples_per_second": 10.486, | |
| "eval_steps_per_second": 0.164, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "eval_loss": 2.2311644554138184, | |
| "eval_runtime": 1198.7414, | |
| "eval_samples_per_second": 10.487, | |
| "eval_steps_per_second": 0.164, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "eval_loss": 2.2293059825897217, | |
| "eval_runtime": 1198.5095, | |
| "eval_samples_per_second": 10.489, | |
| "eval_steps_per_second": 0.164, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "eval_loss": 2.226186990737915, | |
| "eval_runtime": 1198.3579, | |
| "eval_samples_per_second": 10.49, | |
| "eval_steps_per_second": 0.164, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "eval_loss": 2.224607467651367, | |
| "eval_runtime": 1198.8254, | |
| "eval_samples_per_second": 10.486, | |
| "eval_steps_per_second": 0.164, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "eval_loss": 2.222740888595581, | |
| "eval_runtime": 1198.8308, | |
| "eval_samples_per_second": 10.486, | |
| "eval_steps_per_second": 0.164, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "eval_loss": 2.2205824851989746, | |
| "eval_runtime": 1197.986, | |
| "eval_samples_per_second": 10.493, | |
| "eval_steps_per_second": 0.164, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "eval_loss": 2.2183074951171875, | |
| "eval_runtime": 1198.7617, | |
| "eval_samples_per_second": 10.487, | |
| "eval_steps_per_second": 0.164, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "eval_loss": 2.216313123703003, | |
| "eval_runtime": 1197.8492, | |
| "eval_samples_per_second": 10.495, | |
| "eval_steps_per_second": 0.164, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "eval_loss": 2.214667320251465, | |
| "eval_runtime": 1198.1314, | |
| "eval_samples_per_second": 10.492, | |
| "eval_steps_per_second": 0.164, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "eval_loss": 2.2129483222961426, | |
| "eval_runtime": 1198.8701, | |
| "eval_samples_per_second": 10.486, | |
| "eval_steps_per_second": 0.164, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "eval_loss": 2.2119433879852295, | |
| "eval_runtime": 1198.3317, | |
| "eval_samples_per_second": 10.49, | |
| "eval_steps_per_second": 0.164, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "eval_loss": 2.210693836212158, | |
| "eval_runtime": 1198.0145, | |
| "eval_samples_per_second": 10.493, | |
| "eval_steps_per_second": 0.164, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "eval_loss": 2.209263801574707, | |
| "eval_runtime": 1198.6585, | |
| "eval_samples_per_second": 10.488, | |
| "eval_steps_per_second": 0.164, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "eval_loss": 2.20792293548584, | |
| "eval_runtime": 1198.1192, | |
| "eval_samples_per_second": 10.492, | |
| "eval_steps_per_second": 0.164, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "eval_loss": 2.2082931995391846, | |
| "eval_runtime": 1197.8097, | |
| "eval_samples_per_second": 10.495, | |
| "eval_steps_per_second": 0.164, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "eval_loss": 2.205972194671631, | |
| "eval_runtime": 1198.0467, | |
| "eval_samples_per_second": 10.493, | |
| "eval_steps_per_second": 0.164, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "eval_loss": 2.2038400173187256, | |
| "eval_runtime": 1197.5865, | |
| "eval_samples_per_second": 10.497, | |
| "eval_steps_per_second": 0.164, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "eval_loss": 2.2014477252960205, | |
| "eval_runtime": 1198.4564, | |
| "eval_samples_per_second": 10.489, | |
| "eval_steps_per_second": 0.164, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "eval_loss": 2.202155113220215, | |
| "eval_runtime": 1198.0855, | |
| "eval_samples_per_second": 10.493, | |
| "eval_steps_per_second": 0.164, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 5e-06, | |
| "loss": 2.007, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "eval_loss": 2.2006430625915527, | |
| "eval_runtime": 1198.775, | |
| "eval_samples_per_second": 10.487, | |
| "eval_steps_per_second": 0.164, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "eval_loss": 2.1995391845703125, | |
| "eval_runtime": 1198.5076, | |
| "eval_samples_per_second": 10.489, | |
| "eval_steps_per_second": 0.164, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "eval_loss": 2.199397087097168, | |
| "eval_runtime": 1198.3593, | |
| "eval_samples_per_second": 10.49, | |
| "eval_steps_per_second": 0.164, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "eval_loss": 2.19655179977417, | |
| "eval_runtime": 1198.8863, | |
| "eval_samples_per_second": 10.486, | |
| "eval_steps_per_second": 0.164, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "eval_loss": 2.196131467819214, | |
| "eval_runtime": 1198.1451, | |
| "eval_samples_per_second": 10.492, | |
| "eval_steps_per_second": 0.164, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "eval_loss": 2.195343255996704, | |
| "eval_runtime": 1198.7177, | |
| "eval_samples_per_second": 10.487, | |
| "eval_steps_per_second": 0.164, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "eval_loss": 2.194007158279419, | |
| "eval_runtime": 1199.5879, | |
| "eval_samples_per_second": 10.479, | |
| "eval_steps_per_second": 0.164, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "eval_loss": 2.192256212234497, | |
| "eval_runtime": 1198.4679, | |
| "eval_samples_per_second": 10.489, | |
| "eval_steps_per_second": 0.164, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "eval_loss": 2.1922357082366943, | |
| "eval_runtime": 1198.561, | |
| "eval_samples_per_second": 10.488, | |
| "eval_steps_per_second": 0.164, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "eval_loss": 2.190948724746704, | |
| "eval_runtime": 1198.5536, | |
| "eval_samples_per_second": 10.488, | |
| "eval_steps_per_second": 0.164, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "eval_loss": 2.188807487487793, | |
| "eval_runtime": 1199.0987, | |
| "eval_samples_per_second": 10.484, | |
| "eval_steps_per_second": 0.164, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "eval_loss": 2.1877503395080566, | |
| "eval_runtime": 1199.1927, | |
| "eval_samples_per_second": 10.483, | |
| "eval_steps_per_second": 0.164, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 1287, | |
| "total_flos": 5.915384239745925e+18, | |
| "train_loss": 2.0404130036088044, | |
| "train_runtime": 190042.9753, | |
| "train_samples_per_second": 0.867, | |
| "train_steps_per_second": 0.007 | |
| } | |
| ], | |
| "max_steps": 1287, | |
| "num_train_epochs": 3, | |
| "total_flos": 5.915384239745925e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |