397 lines
8.0 KiB
JSON
397 lines
8.0 KiB
JSON
|
{
|
||
|
|
"best_metric": 3.468397378921509,
|
||
|
|
"best_model_checkpoint": "/datadrive/disk1/pierpaolo/llm/itlangadapt/bloom-1b3_it/checkpoint-50000",
|
||
|
|
"epoch": 5.236680193742637,
|
||
|
|
"global_step": 50000,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.1,
|
||
|
|
"learning_rate": 0.00098,
|
||
|
|
"loss": 3.4734,
|
||
|
|
"step": 1000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21,
|
||
|
|
"learning_rate": 0.00096,
|
||
|
|
"loss": 3.466,
|
||
|
|
"step": 2000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.31,
|
||
|
|
"learning_rate": 0.00094,
|
||
|
|
"loss": 3.461,
|
||
|
|
"step": 3000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42,
|
||
|
|
"learning_rate": 0.00092,
|
||
|
|
"loss": 3.4725,
|
||
|
|
"step": 4000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.52,
|
||
|
|
"learning_rate": 0.0009000000000000001,
|
||
|
|
"loss": 3.4614,
|
||
|
|
"step": 5000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.52,
|
||
|
|
"eval_loss": 3.469589948654175,
|
||
|
|
"eval_runtime": 358.4452,
|
||
|
|
"eval_samples_per_second": 10.624,
|
||
|
|
"eval_steps_per_second": 5.312,
|
||
|
|
"step": 5000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.63,
|
||
|
|
"learning_rate": 0.00088,
|
||
|
|
"loss": 3.4653,
|
||
|
|
"step": 6000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.73,
|
||
|
|
"learning_rate": 0.00086,
|
||
|
|
"loss": 3.4618,
|
||
|
|
"step": 7000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.84,
|
||
|
|
"learning_rate": 0.00084,
|
||
|
|
"loss": 3.4707,
|
||
|
|
"step": 8000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.94,
|
||
|
|
"learning_rate": 0.00082002,
|
||
|
|
"loss": 3.4706,
|
||
|
|
"step": 9000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.05,
|
||
|
|
"learning_rate": 0.0008000199999999999,
|
||
|
|
"loss": 3.4694,
|
||
|
|
"step": 10000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.05,
|
||
|
|
"eval_loss": 3.4690916538238525,
|
||
|
|
"eval_runtime": 358.2981,
|
||
|
|
"eval_samples_per_second": 10.628,
|
||
|
|
"eval_steps_per_second": 5.314,
|
||
|
|
"step": 10000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.15,
|
||
|
|
"learning_rate": 0.0007800200000000001,
|
||
|
|
"loss": 3.4686,
|
||
|
|
"step": 11000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.26,
|
||
|
|
"learning_rate": 0.00076002,
|
||
|
|
"loss": 3.4636,
|
||
|
|
"step": 12000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.36,
|
||
|
|
"learning_rate": 0.0007400400000000001,
|
||
|
|
"loss": 3.463,
|
||
|
|
"step": 13000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.47,
|
||
|
|
"learning_rate": 0.00072006,
|
||
|
|
"loss": 3.4694,
|
||
|
|
"step": 14000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.57,
|
||
|
|
"learning_rate": 0.0007000600000000001,
|
||
|
|
"loss": 3.4709,
|
||
|
|
"step": 15000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.57,
|
||
|
|
"eval_loss": 3.4688496589660645,
|
||
|
|
"eval_runtime": 358.397,
|
||
|
|
"eval_samples_per_second": 10.625,
|
||
|
|
"eval_steps_per_second": 5.313,
|
||
|
|
"step": 15000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.68,
|
||
|
|
"learning_rate": 0.00068006,
|
||
|
|
"loss": 3.4627,
|
||
|
|
"step": 16000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.78,
|
||
|
|
"learning_rate": 0.0006600800000000001,
|
||
|
|
"loss": 3.4658,
|
||
|
|
"step": 17000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.89,
|
||
|
|
"learning_rate": 0.00064008,
|
||
|
|
"loss": 3.4661,
|
||
|
|
"step": 18000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.99,
|
||
|
|
"learning_rate": 0.00062008,
|
||
|
|
"loss": 3.4615,
|
||
|
|
"step": 19000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.09,
|
||
|
|
"learning_rate": 0.0006001,
|
||
|
|
"loss": 3.481,
|
||
|
|
"step": 20000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.09,
|
||
|
|
"eval_loss": 3.468804359436035,
|
||
|
|
"eval_runtime": 359.4028,
|
||
|
|
"eval_samples_per_second": 10.595,
|
||
|
|
"eval_steps_per_second": 5.298,
|
||
|
|
"step": 20000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2,
|
||
|
|
"learning_rate": 0.0005801,
|
||
|
|
"loss": 3.465,
|
||
|
|
"step": 21000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3,
|
||
|
|
"learning_rate": 0.00056012,
|
||
|
|
"loss": 3.4656,
|
||
|
|
"step": 22000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.41,
|
||
|
|
"learning_rate": 0.00054012,
|
||
|
|
"loss": 3.472,
|
||
|
|
"step": 23000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.51,
|
||
|
|
"learning_rate": 0.00052012,
|
||
|
|
"loss": 3.4574,
|
||
|
|
"step": 24000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.62,
|
||
|
|
"learning_rate": 0.0005001200000000001,
|
||
|
|
"loss": 3.4638,
|
||
|
|
"step": 25000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.62,
|
||
|
|
"eval_loss": 3.468665599822998,
|
||
|
|
"eval_runtime": 358.6249,
|
||
|
|
"eval_samples_per_second": 10.618,
|
||
|
|
"eval_steps_per_second": 5.309,
|
||
|
|
"step": 25000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.72,
|
||
|
|
"learning_rate": 0.00048012,
|
||
|
|
"loss": 3.4611,
|
||
|
|
"step": 26000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.83,
|
||
|
|
"learning_rate": 0.00046012,
|
||
|
|
"loss": 3.4631,
|
||
|
|
"step": 27000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.93,
|
||
|
|
"learning_rate": 0.00044014,
|
||
|
|
"loss": 3.4626,
|
||
|
|
"step": 28000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.04,
|
||
|
|
"learning_rate": 0.00042014,
|
||
|
|
"loss": 3.469,
|
||
|
|
"step": 29000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.14,
|
||
|
|
"learning_rate": 0.00040018,
|
||
|
|
"loss": 3.4704,
|
||
|
|
"step": 30000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.14,
|
||
|
|
"eval_loss": 3.468623399734497,
|
||
|
|
"eval_runtime": 359.7336,
|
||
|
|
"eval_samples_per_second": 10.586,
|
||
|
|
"eval_steps_per_second": 5.293,
|
||
|
|
"step": 30000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.25,
|
||
|
|
"learning_rate": 0.00038018000000000004,
|
||
|
|
"loss": 3.4659,
|
||
|
|
"step": 31000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.35,
|
||
|
|
"learning_rate": 0.00036018,
|
||
|
|
"loss": 3.4706,
|
||
|
|
"step": 32000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.46,
|
||
|
|
"learning_rate": 0.00034018,
|
||
|
|
"loss": 3.4659,
|
||
|
|
"step": 33000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.56,
|
||
|
|
"learning_rate": 0.00032018000000000004,
|
||
|
|
"loss": 3.4669,
|
||
|
|
"step": 34000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.67,
|
||
|
|
"learning_rate": 0.00030018,
|
||
|
|
"loss": 3.4617,
|
||
|
|
"step": 35000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.67,
|
||
|
|
"eval_loss": 3.4684813022613525,
|
||
|
|
"eval_runtime": 361.1986,
|
||
|
|
"eval_samples_per_second": 10.543,
|
||
|
|
"eval_steps_per_second": 5.271,
|
||
|
|
"step": 35000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.77,
|
||
|
|
"learning_rate": 0.00028020000000000003,
|
||
|
|
"loss": 3.4703,
|
||
|
|
"step": 36000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.88,
|
||
|
|
"learning_rate": 0.00026024,
|
||
|
|
"loss": 3.4606,
|
||
|
|
"step": 37000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.98,
|
||
|
|
"learning_rate": 0.00024026,
|
||
|
|
"loss": 3.4549,
|
||
|
|
"step": 38000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.08,
|
||
|
|
"learning_rate": 0.00022026,
|
||
|
|
"loss": 3.4701,
|
||
|
|
"step": 39000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.19,
|
||
|
|
"learning_rate": 0.00020026,
|
||
|
|
"loss": 3.4636,
|
||
|
|
"step": 40000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.19,
|
||
|
|
"eval_loss": 3.468475103378296,
|
||
|
|
"eval_runtime": 358.6453,
|
||
|
|
"eval_samples_per_second": 10.618,
|
||
|
|
"eval_steps_per_second": 5.309,
|
||
|
|
"step": 40000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.29,
|
||
|
|
"learning_rate": 0.00018026,
|
||
|
|
"loss": 3.4541,
|
||
|
|
"step": 41000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.4,
|
||
|
|
"learning_rate": 0.00016026000000000001,
|
||
|
|
"loss": 3.4724,
|
||
|
|
"step": 42000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.5,
|
||
|
|
"learning_rate": 0.00014026,
|
||
|
|
"loss": 3.4683,
|
||
|
|
"step": 43000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.61,
|
||
|
|
"learning_rate": 0.00012026,
|
||
|
|
"loss": 3.463,
|
||
|
|
"step": 44000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.71,
|
||
|
|
"learning_rate": 0.00010026,
|
||
|
|
"loss": 3.4668,
|
||
|
|
"step": 45000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.71,
|
||
|
|
"eval_loss": 3.4684269428253174,
|
||
|
|
"eval_runtime": 358.1004,
|
||
|
|
"eval_samples_per_second": 10.634,
|
||
|
|
"eval_steps_per_second": 5.317,
|
||
|
|
"step": 45000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.82,
|
||
|
|
"learning_rate": 8.028000000000001e-05,
|
||
|
|
"loss": 3.4696,
|
||
|
|
"step": 46000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.92,
|
||
|
|
"learning_rate": 6.028e-05,
|
||
|
|
"loss": 3.4605,
|
||
|
|
"step": 47000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.03,
|
||
|
|
"learning_rate": 4.028000000000001e-05,
|
||
|
|
"loss": 3.4645,
|
||
|
|
"step": 48000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.13,
|
||
|
|
"learning_rate": 2.028e-05,
|
||
|
|
"loss": 3.4704,
|
||
|
|
"step": 49000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.24,
|
||
|
|
"learning_rate": 2.7999999999999997e-07,
|
||
|
|
"loss": 3.4623,
|
||
|
|
"step": 50000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.24,
|
||
|
|
"eval_loss": 3.468397378921509,
|
||
|
|
"eval_runtime": 358.1322,
|
||
|
|
"eval_samples_per_second": 10.633,
|
||
|
|
"eval_steps_per_second": 5.316,
|
||
|
|
"step": 50000
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"max_steps": 50000,
|
||
|
|
"num_train_epochs": 6,
|
||
|
|
"total_flos": 2.9704946454626304e+18,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|