oh_v3-1_only_evol_instruct_140k / trainer_log.jsonl
esfrankel17's picture
Training in progress, epoch 1
668a021 verified
raw
history blame
2.05 kB
{"current_steps": 10, "total_steps": 135, "loss": 1.0138, "learning_rate": 5e-06, "epoch": 0.21798365122615804, "percentage": 7.41, "elapsed_time": "0:05:23", "remaining_time": "1:07:21"}
{"current_steps": 20, "total_steps": 135, "loss": 0.9311, "learning_rate": 5e-06, "epoch": 0.4359673024523161, "percentage": 14.81, "elapsed_time": "0:10:36", "remaining_time": "1:00:59"}
{"current_steps": 30, "total_steps": 135, "loss": 0.902, "learning_rate": 5e-06, "epoch": 0.6539509536784741, "percentage": 22.22, "elapsed_time": "0:15:51", "remaining_time": "0:55:30"}
{"current_steps": 40, "total_steps": 135, "loss": 0.8817, "learning_rate": 5e-06, "epoch": 0.8719346049046321, "percentage": 29.63, "elapsed_time": "0:21:05", "remaining_time": "0:50:04"}
{"current_steps": 45, "total_steps": 135, "eval_loss": 0.8728658556938171, "epoch": 0.9809264305177112, "percentage": 33.33, "elapsed_time": "0:24:26", "remaining_time": "0:48:53"}
{"current_steps": 50, "total_steps": 135, "loss": 0.9119, "learning_rate": 5e-06, "epoch": 1.1008174386920981, "percentage": 37.04, "elapsed_time": "0:27:50", "remaining_time": "0:47:19"}
{"current_steps": 60, "total_steps": 135, "loss": 0.8201, "learning_rate": 5e-06, "epoch": 1.318801089918256, "percentage": 44.44, "elapsed_time": "0:33:02", "remaining_time": "0:41:18"}
{"current_steps": 70, "total_steps": 135, "loss": 0.8167, "learning_rate": 5e-06, "epoch": 1.5367847411444142, "percentage": 51.85, "elapsed_time": "0:38:15", "remaining_time": "0:35:31"}
{"current_steps": 80, "total_steps": 135, "loss": 0.8136, "learning_rate": 5e-06, "epoch": 1.7547683923705724, "percentage": 59.26, "elapsed_time": "0:43:27", "remaining_time": "0:29:52"}
{"current_steps": 90, "total_steps": 135, "loss": 0.8115, "learning_rate": 5e-06, "epoch": 1.9727520435967303, "percentage": 66.67, "elapsed_time": "0:48:38", "remaining_time": "0:24:19"}
{"current_steps": 90, "total_steps": 135, "eval_loss": 0.8485425710678101, "epoch": 1.9727520435967303, "percentage": 66.67, "elapsed_time": "0:49:31", "remaining_time": "0:24:45"}