bc796b7b06
Model: huseyinatahaninan/appworld_distillation_sft_v2-SFT-Qwen3-8B Source: Original Platform
594 lines
14 KiB
JSON
594 lines
14 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 25.0,
|
|
"eval_steps": 500,
|
|
"global_step": 50,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.5714285714285714,
|
|
"grad_norm": 27.379572188798917,
|
|
"learning_rate": 0.0,
|
|
"loss": 1.2369,
|
|
"step": 1
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"grad_norm": 27.379572188798917,
|
|
"learning_rate": 1.0000000000000002e-06,
|
|
"loss": 1.1697,
|
|
"step": 2
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"eval_loss": 1.2943594455718994,
|
|
"eval_runtime": 13.9947,
|
|
"eval_samples_per_second": 2.501,
|
|
"eval_steps_per_second": 0.357,
|
|
"step": 2
|
|
},
|
|
{
|
|
"epoch": 1.5714285714285714,
|
|
"grad_norm": 31.930876552830178,
|
|
"learning_rate": 2.0000000000000003e-06,
|
|
"loss": 1.2156,
|
|
"step": 3
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"grad_norm": 24.003107200324983,
|
|
"learning_rate": 3e-06,
|
|
"loss": 1.0781,
|
|
"step": 4
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"eval_loss": 1.1406848430633545,
|
|
"eval_runtime": 12.0925,
|
|
"eval_samples_per_second": 2.894,
|
|
"eval_steps_per_second": 0.413,
|
|
"step": 4
|
|
},
|
|
{
|
|
"epoch": 2.571428571428571,
|
|
"grad_norm": 11.488887862580302,
|
|
"learning_rate": 4.000000000000001e-06,
|
|
"loss": 0.9982,
|
|
"step": 5
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"grad_norm": 4.84397184489544,
|
|
"learning_rate": 5e-06,
|
|
"loss": 0.9806,
|
|
"step": 6
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"eval_loss": 1.0041135549545288,
|
|
"eval_runtime": 12.0103,
|
|
"eval_samples_per_second": 2.914,
|
|
"eval_steps_per_second": 0.416,
|
|
"step": 6
|
|
},
|
|
{
|
|
"epoch": 3.571428571428571,
|
|
"grad_norm": 3.7919066688172225,
|
|
"learning_rate": 4.993910125649561e-06,
|
|
"loss": 0.9273,
|
|
"step": 7
|
|
},
|
|
{
|
|
"epoch": 4.0,
|
|
"grad_norm": 2.816005416926627,
|
|
"learning_rate": 4.975670171853926e-06,
|
|
"loss": 0.9093,
|
|
"step": 8
|
|
},
|
|
{
|
|
"epoch": 4.0,
|
|
"eval_loss": 0.9050876498222351,
|
|
"eval_runtime": 12.0336,
|
|
"eval_samples_per_second": 2.909,
|
|
"eval_steps_per_second": 0.416,
|
|
"step": 8
|
|
},
|
|
{
|
|
"epoch": 4.571428571428571,
|
|
"grad_norm": 1.8900642576089526,
|
|
"learning_rate": 4.9453690018345144e-06,
|
|
"loss": 0.86,
|
|
"step": 9
|
|
},
|
|
{
|
|
"epoch": 5.0,
|
|
"grad_norm": 1.8900642576089526,
|
|
"learning_rate": 4.903154239845798e-06,
|
|
"loss": 0.8595,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 5.0,
|
|
"eval_loss": 0.8866418600082397,
|
|
"eval_runtime": 12.0514,
|
|
"eval_samples_per_second": 2.904,
|
|
"eval_steps_per_second": 0.415,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 5.571428571428571,
|
|
"grad_norm": 2.388409790695317,
|
|
"learning_rate": 4.849231551964771e-06,
|
|
"loss": 0.7986,
|
|
"step": 11
|
|
},
|
|
{
|
|
"epoch": 6.0,
|
|
"grad_norm": 1.9351948554401477,
|
|
"learning_rate": 4.783863644106502e-06,
|
|
"loss": 0.7688,
|
|
"step": 12
|
|
},
|
|
{
|
|
"epoch": 6.0,
|
|
"eval_loss": 0.8012941479682922,
|
|
"eval_runtime": 12.0289,
|
|
"eval_samples_per_second": 2.91,
|
|
"eval_steps_per_second": 0.416,
|
|
"step": 12
|
|
},
|
|
{
|
|
"epoch": 6.571428571428571,
|
|
"grad_norm": 1.9475837570279504,
|
|
"learning_rate": 4.707368982147318e-06,
|
|
"loss": 0.7405,
|
|
"step": 13
|
|
},
|
|
{
|
|
"epoch": 7.0,
|
|
"grad_norm": 1.7205847922715594,
|
|
"learning_rate": 4.620120240391065e-06,
|
|
"loss": 0.7223,
|
|
"step": 14
|
|
},
|
|
{
|
|
"epoch": 7.0,
|
|
"eval_loss": 0.7613986730575562,
|
|
"eval_runtime": 12.0755,
|
|
"eval_samples_per_second": 2.898,
|
|
"eval_steps_per_second": 0.414,
|
|
"step": 14
|
|
},
|
|
{
|
|
"epoch": 7.571428571428571,
|
|
"grad_norm": 1.3353634306020017,
|
|
"learning_rate": 4.522542485937369e-06,
|
|
"loss": 0.7031,
|
|
"step": 15
|
|
},
|
|
{
|
|
"epoch": 8.0,
|
|
"grad_norm": 1.3800382206930182,
|
|
"learning_rate": 4.415111107797445e-06,
|
|
"loss": 0.689,
|
|
"step": 16
|
|
},
|
|
{
|
|
"epoch": 8.0,
|
|
"eval_loss": 0.727152407169342,
|
|
"eval_runtime": 11.966,
|
|
"eval_samples_per_second": 2.925,
|
|
"eval_steps_per_second": 0.418,
|
|
"step": 16
|
|
},
|
|
{
|
|
"epoch": 8.571428571428571,
|
|
"grad_norm": 0.9321182321446594,
|
|
"learning_rate": 4.2983495008466285e-06,
|
|
"loss": 0.6511,
|
|
"step": 17
|
|
},
|
|
{
|
|
"epoch": 9.0,
|
|
"grad_norm": 0.9321182321446594,
|
|
"learning_rate": 4.172826515897146e-06,
|
|
"loss": 0.6641,
|
|
"step": 18
|
|
},
|
|
{
|
|
"epoch": 9.0,
|
|
"eval_loss": 0.7126601934432983,
|
|
"eval_runtime": 12.059,
|
|
"eval_samples_per_second": 2.902,
|
|
"eval_steps_per_second": 0.415,
|
|
"step": 18
|
|
},
|
|
{
|
|
"epoch": 9.571428571428571,
|
|
"grad_norm": 1.284468249112213,
|
|
"learning_rate": 4.039153688314146e-06,
|
|
"loss": 0.6349,
|
|
"step": 19
|
|
},
|
|
{
|
|
"epoch": 10.0,
|
|
"grad_norm": 1.209401710296371,
|
|
"learning_rate": 3.897982258676867e-06,
|
|
"loss": 0.5795,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 10.0,
|
|
"eval_loss": 0.6720197796821594,
|
|
"eval_runtime": 12.0161,
|
|
"eval_samples_per_second": 2.913,
|
|
"eval_steps_per_second": 0.416,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 10.571428571428571,
|
|
"grad_norm": 1.5344113073991625,
|
|
"learning_rate": 3.7500000000000005e-06,
|
|
"loss": 0.5678,
|
|
"step": 21
|
|
},
|
|
{
|
|
"epoch": 11.0,
|
|
"grad_norm": 1.086081561802662,
|
|
"learning_rate": 3.595927866972694e-06,
|
|
"loss": 0.5451,
|
|
"step": 22
|
|
},
|
|
{
|
|
"epoch": 11.0,
|
|
"eval_loss": 0.655051589012146,
|
|
"eval_runtime": 12.1049,
|
|
"eval_samples_per_second": 2.891,
|
|
"eval_steps_per_second": 0.413,
|
|
"step": 22
|
|
},
|
|
{
|
|
"epoch": 11.571428571428571,
|
|
"grad_norm": 1.3512413194371768,
|
|
"learning_rate": 3.436516483539781e-06,
|
|
"loss": 0.5435,
|
|
"step": 23
|
|
},
|
|
{
|
|
"epoch": 12.0,
|
|
"grad_norm": 1.1140031619845532,
|
|
"learning_rate": 3.272542485937369e-06,
|
|
"loss": 0.5059,
|
|
"step": 24
|
|
},
|
|
{
|
|
"epoch": 12.0,
|
|
"eval_loss": 0.6408958435058594,
|
|
"eval_runtime": 12.063,
|
|
"eval_samples_per_second": 2.901,
|
|
"eval_steps_per_second": 0.414,
|
|
"step": 24
|
|
},
|
|
{
|
|
"epoch": 12.571428571428571,
|
|
"grad_norm": 0.9212734048420463,
|
|
"learning_rate": 3.1048047389991693e-06,
|
|
"loss": 0.5038,
|
|
"step": 25
|
|
},
|
|
{
|
|
"epoch": 13.0,
|
|
"grad_norm": 0.9212734048420463,
|
|
"learning_rate": 2.9341204441673267e-06,
|
|
"loss": 0.5035,
|
|
"step": 26
|
|
},
|
|
{
|
|
"epoch": 13.0,
|
|
"eval_loss": 0.6351635456085205,
|
|
"eval_runtime": 12.0596,
|
|
"eval_samples_per_second": 2.902,
|
|
"eval_steps_per_second": 0.415,
|
|
"step": 26
|
|
},
|
|
{
|
|
"epoch": 13.571428571428571,
|
|
"grad_norm": 1.1395169936217417,
|
|
"learning_rate": 2.761321158169134e-06,
|
|
"loss": 0.489,
|
|
"step": 27
|
|
},
|
|
{
|
|
"epoch": 14.0,
|
|
"grad_norm": 0.8506966891650993,
|
|
"learning_rate": 2.587248741756253e-06,
|
|
"loss": 0.484,
|
|
"step": 28
|
|
},
|
|
{
|
|
"epoch": 14.0,
|
|
"eval_loss": 0.6280523538589478,
|
|
"eval_runtime": 12.0505,
|
|
"eval_samples_per_second": 2.904,
|
|
"eval_steps_per_second": 0.415,
|
|
"step": 28
|
|
},
|
|
{
|
|
"epoch": 14.571428571428571,
|
|
"grad_norm": 0.9887201931904582,
|
|
"learning_rate": 2.4127512582437486e-06,
|
|
"loss": 0.4683,
|
|
"step": 29
|
|
},
|
|
{
|
|
"epoch": 15.0,
|
|
"grad_norm": 0.9155854743888291,
|
|
"learning_rate": 2.238678841830867e-06,
|
|
"loss": 0.4436,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 15.0,
|
|
"eval_loss": 0.6252209544181824,
|
|
"eval_runtime": 12.0701,
|
|
"eval_samples_per_second": 2.9,
|
|
"eval_steps_per_second": 0.414,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 15.571428571428571,
|
|
"grad_norm": 1.01480533363197,
|
|
"learning_rate": 2.0658795558326745e-06,
|
|
"loss": 0.4466,
|
|
"step": 31
|
|
},
|
|
{
|
|
"epoch": 16.0,
|
|
"grad_norm": 0.940925597770304,
|
|
"learning_rate": 1.895195261000831e-06,
|
|
"loss": 0.4347,
|
|
"step": 32
|
|
},
|
|
{
|
|
"epoch": 16.0,
|
|
"eval_loss": 0.6249921321868896,
|
|
"eval_runtime": 12.0063,
|
|
"eval_samples_per_second": 2.915,
|
|
"eval_steps_per_second": 0.416,
|
|
"step": 32
|
|
},
|
|
{
|
|
"epoch": 16.571428571428573,
|
|
"grad_norm": 0.745543391044636,
|
|
"learning_rate": 1.7274575140626318e-06,
|
|
"loss": 0.4275,
|
|
"step": 33
|
|
},
|
|
{
|
|
"epoch": 17.0,
|
|
"grad_norm": 0.745543391044636,
|
|
"learning_rate": 1.56348351646022e-06,
|
|
"loss": 0.4139,
|
|
"step": 34
|
|
},
|
|
{
|
|
"epoch": 17.0,
|
|
"eval_loss": 0.6253374218940735,
|
|
"eval_runtime": 12.0584,
|
|
"eval_samples_per_second": 2.903,
|
|
"eval_steps_per_second": 0.415,
|
|
"step": 34
|
|
},
|
|
{
|
|
"epoch": 17.571428571428573,
|
|
"grad_norm": 1.240672425729672,
|
|
"learning_rate": 1.4040721330273063e-06,
|
|
"loss": 0.4033,
|
|
"step": 35
|
|
},
|
|
{
|
|
"epoch": 18.0,
|
|
"grad_norm": 0.8138297069132504,
|
|
"learning_rate": 1.2500000000000007e-06,
|
|
"loss": 0.4108,
|
|
"step": 36
|
|
},
|
|
{
|
|
"epoch": 18.0,
|
|
"eval_loss": 0.6265431642532349,
|
|
"eval_runtime": 12.0931,
|
|
"eval_samples_per_second": 2.894,
|
|
"eval_steps_per_second": 0.413,
|
|
"step": 36
|
|
},
|
|
{
|
|
"epoch": 18.571428571428573,
|
|
"grad_norm": 1.0257247393101925,
|
|
"learning_rate": 1.1020177413231334e-06,
|
|
"loss": 0.3948,
|
|
"step": 37
|
|
},
|
|
{
|
|
"epoch": 19.0,
|
|
"grad_norm": 0.8730540573897112,
|
|
"learning_rate": 9.608463116858544e-07,
|
|
"loss": 0.3969,
|
|
"step": 38
|
|
},
|
|
{
|
|
"epoch": 19.0,
|
|
"eval_loss": 0.6287420392036438,
|
|
"eval_runtime": 12.0278,
|
|
"eval_samples_per_second": 2.91,
|
|
"eval_steps_per_second": 0.416,
|
|
"step": 38
|
|
},
|
|
{
|
|
"epoch": 19.571428571428573,
|
|
"grad_norm": 0.9053129240846155,
|
|
"learning_rate": 8.271734841028553e-07,
|
|
"loss": 0.3965,
|
|
"step": 39
|
|
},
|
|
{
|
|
"epoch": 20.0,
|
|
"grad_norm": 0.9013824277926773,
|
|
"learning_rate": 7.016504991533727e-07,
|
|
"loss": 0.3825,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 20.0,
|
|
"eval_loss": 0.6303004026412964,
|
|
"eval_runtime": 12.1106,
|
|
"eval_samples_per_second": 2.89,
|
|
"eval_steps_per_second": 0.413,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 20.571428571428573,
|
|
"grad_norm": 0.6882077843880618,
|
|
"learning_rate": 5.848888922025553e-07,
|
|
"loss": 0.3825,
|
|
"step": 41
|
|
},
|
|
{
|
|
"epoch": 21.0,
|
|
"grad_norm": 0.6882077843880618,
|
|
"learning_rate": 4.774575140626317e-07,
|
|
"loss": 0.3839,
|
|
"step": 42
|
|
},
|
|
{
|
|
"epoch": 21.0,
|
|
"eval_loss": 0.6313385367393494,
|
|
"eval_runtime": 11.905,
|
|
"eval_samples_per_second": 2.94,
|
|
"eval_steps_per_second": 0.42,
|
|
"step": 42
|
|
},
|
|
{
|
|
"epoch": 21.571428571428573,
|
|
"grad_norm": 1.0040896510913107,
|
|
"learning_rate": 3.798797596089351e-07,
|
|
"loss": 0.3821,
|
|
"step": 43
|
|
},
|
|
{
|
|
"epoch": 22.0,
|
|
"grad_norm": 0.8312442360781623,
|
|
"learning_rate": 2.9263101785268253e-07,
|
|
"loss": 0.3699,
|
|
"step": 44
|
|
},
|
|
{
|
|
"epoch": 22.0,
|
|
"eval_loss": 0.6325893998146057,
|
|
"eval_runtime": 11.9284,
|
|
"eval_samples_per_second": 2.934,
|
|
"eval_steps_per_second": 0.419,
|
|
"step": 44
|
|
},
|
|
{
|
|
"epoch": 22.571428571428573,
|
|
"grad_norm": 0.8906586628247445,
|
|
"learning_rate": 2.1613635589349756e-07,
|
|
"loss": 0.3652,
|
|
"step": 45
|
|
},
|
|
{
|
|
"epoch": 23.0,
|
|
"grad_norm": 0.8407663191838467,
|
|
"learning_rate": 1.507684480352292e-07,
|
|
"loss": 0.3871,
|
|
"step": 46
|
|
},
|
|
{
|
|
"epoch": 23.0,
|
|
"eval_loss": 0.6336179971694946,
|
|
"eval_runtime": 11.9514,
|
|
"eval_samples_per_second": 2.929,
|
|
"eval_steps_per_second": 0.418,
|
|
"step": 46
|
|
},
|
|
{
|
|
"epoch": 23.571428571428573,
|
|
"grad_norm": 0.8820384716444405,
|
|
"learning_rate": 9.684576015420277e-08,
|
|
"loss": 0.3612,
|
|
"step": 47
|
|
},
|
|
{
|
|
"epoch": 24.0,
|
|
"grad_norm": 0.9165057903043542,
|
|
"learning_rate": 5.463099816548578e-08,
|
|
"loss": 0.382,
|
|
"step": 48
|
|
},
|
|
{
|
|
"epoch": 24.0,
|
|
"eval_loss": 0.6337578892707825,
|
|
"eval_runtime": 11.8972,
|
|
"eval_samples_per_second": 2.942,
|
|
"eval_steps_per_second": 0.42,
|
|
"step": 48
|
|
},
|
|
{
|
|
"epoch": 24.571428571428573,
|
|
"grad_norm": 0.7173736876587623,
|
|
"learning_rate": 2.4329828146074096e-08,
|
|
"loss": 0.3611,
|
|
"step": 49
|
|
},
|
|
{
|
|
"epoch": 25.0,
|
|
"grad_norm": 0.7173736876587623,
|
|
"learning_rate": 6.089874350439507e-09,
|
|
"loss": 0.3879,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 25.0,
|
|
"eval_loss": 0.6342211961746216,
|
|
"eval_runtime": 12.0027,
|
|
"eval_samples_per_second": 2.916,
|
|
"eval_steps_per_second": 0.417,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 25.0,
|
|
"step": 50,
|
|
"total_flos": 9.581916505872794e+16,
|
|
"train_loss": 0.6022347152233124,
|
|
"train_runtime": 1896.7685,
|
|
"train_samples_per_second": 0.685,
|
|
"train_steps_per_second": 0.026
|
|
}
|
|
],
|
|
"logging_steps": 1,
|
|
"max_steps": 50,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 25,
|
|
"save_steps": 2000000,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": true
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 9.581916505872794e+16,
|
|
"train_batch_size": 1,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|