12 lines
2.8 KiB
JSON
12 lines
2.8 KiB
JSON
{"loss": 1.11289322, "grad_norm": 12.19196497, "learning_rate": 5e-06, "epoch": 0.11594203, "global_step/max_steps": "1/27", "elapsed_time": "47s", "remaining_time": "20m 28s", "memory(GiB)": 18.14, "train_speed(s/it)": 47.220165}
|
|
{"loss": 0.9293499, "grad_norm": 1.77222941, "learning_rate": 9.65e-06, "epoch": 0.57971014, "global_step/max_steps": "5/27", "elapsed_time": "3m 39s", "remaining_time": "16m 4s", "memory(GiB)": 22.18, "train_speed(s/it)": 43.834819}
|
|
{"eval_loss": 0.66912872, "eval_runtime": 3.5432, "eval_samples_per_second": 1.693, "eval_steps_per_second": 0.282, "eval_token_acc": 0.80555405, "epoch": 1.0, "global_step/max_steps": "9/27", "elapsed_time": "6m 18s", "remaining_time": "12m 36s", "memory(GiB)": 53.57, "train_speed(s/it)": 41.9777}
|
|
{"loss": 0.72368984, "grad_norm": 0.87008269, "learning_rate": 7.68e-06, "epoch": 1.11594203, "global_step/max_steps": "10/27", "elapsed_time": "7m 20s", "remaining_time": "12m 28s", "memory(GiB)": 53.57, "train_speed(s/it)": 44.023439}
|
|
{"loss": 0.63697853, "grad_norm": 0.48984571, "learning_rate": 4.69e-06, "epoch": 1.69565217, "global_step/max_steps": "15/27", "elapsed_time": "10m 54s", "remaining_time": "8m 43s", "memory(GiB)": 53.57, "train_speed(s/it)": 43.585551}
|
|
{"eval_loss": 0.60824573, "eval_runtime": 3.4969, "eval_samples_per_second": 1.716, "eval_steps_per_second": 0.286, "eval_token_acc": 0.81819661, "epoch": 2.0, "global_step/max_steps": "18/27", "elapsed_time": "12m 50s", "remaining_time": "6m 25s", "memory(GiB)": 53.57, "train_speed(s/it)": 42.756697}
|
|
{"loss": 0.61085949, "grad_norm": 0.31489104, "learning_rate": 1.81e-06, "epoch": 2.23188406, "global_step/max_steps": "20/27", "elapsed_time": "15m 7s", "remaining_time": "5m 18s", "memory(GiB)": 53.57, "train_speed(s/it)": 45.363137}
|
|
{"loss": 0.59106207, "grad_norm": 0.28574588, "learning_rate": 1.6e-07, "epoch": 2.8115942, "global_step/max_steps": "25/27", "elapsed_time": "18m 41s", "remaining_time": "1m 30s", "memory(GiB)": 53.57, "train_speed(s/it)": 44.837141}
|
|
{"eval_loss": 0.59811389, "eval_runtime": 3.4377, "eval_samples_per_second": 1.745, "eval_steps_per_second": 0.291, "eval_token_acc": 0.8204016, "epoch": 3.0, "global_step/max_steps": "27/27", "elapsed_time": "19m 54s", "remaining_time": "0s", "memory(GiB)": 53.57, "train_speed(s/it)": 44.223312}
|
|
{"eval_loss": 0.59811389, "eval_runtime": 4.1001, "eval_samples_per_second": 1.463, "eval_steps_per_second": 0.244, "eval_token_acc": 0.8204016, "epoch": 3.0, "global_step/max_steps": "27/27", "elapsed_time": "21m 48s", "remaining_time": "0s", "memory(GiB)": 53.57, "train_speed(s/it)": 48.433686}
|
|
{"train_runtime": 1326.3373, "train_samples_per_second": 1.258, "train_steps_per_second": 0.02, "total_flos": 61857115668480.0, "train_loss": 0.6955941, "epoch": 3.0, "global_step/max_steps": "27/27", "elapsed_time": "22m 4s", "remaining_time": "0s", "memory(GiB)": 53.57, "train_speed(s/it)": 49.040129}
|