Qwen2.5-1.5B-Open-R1-Code-GRPO / trainer_state.json
MinhPhuoc's picture
Model save
1ad2387 verified
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.0008395072092681596,
"eval_steps": 500,
"global_step": 10,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": -2.0,
"completions/max_length": 928.0,
"completions/max_terminated_length": 928.0,
"completions/mean_length": 519.3333740234375,
"completions/mean_terminated_length": 519.3333740234375,
"completions/min_length": 306.0,
"completions/min_terminated_length": 306.0,
"epoch": 8.395072092681596e-05,
"frac_reward_zero_std": 0.6666666865348816,
"grad_norm": 1.2832554578781128,
"kl": 0.0,
"learning_rate": 0.0,
"loss": 0.0,
"num_tokens": 14500.0,
"reward": 0.008333333767950535,
"reward_std": 0.01666666753590107,
"rewards/code_reward/mean": 0.0,
"rewards/code_reward/std": 0.0,
"rewards/format_reward/mean": 0.0833333358168602,
"rewards/format_reward/std": 0.28867512941360474,
"step": 1
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": -2.0,
"completions/max_length": 925.0,
"completions/max_terminated_length": 925.0,
"completions/mean_length": 334.75,
"completions/mean_terminated_length": 334.75,
"completions/min_length": 109.0,
"completions/min_terminated_length": 109.0,
"epoch": 0.00016790144185363192,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.0,
"kl": 0.0,
"learning_rate": 5e-06,
"loss": 0.0,
"num_tokens": 24765.0,
"reward": 0.0,
"reward_std": 0.0,
"rewards/code_reward/mean": 0.0,
"rewards/code_reward/std": 0.0,
"rewards/format_reward/mean": 0.0,
"rewards/format_reward/std": 0.0,
"step": 2
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": -2.0,
"completions/max_length": 760.0,
"completions/max_terminated_length": 760.0,
"completions/mean_length": 443.25,
"completions/mean_terminated_length": 443.25,
"completions/min_length": 251.0,
"completions/min_terminated_length": 251.0,
"epoch": 0.00025185216278044787,
"frac_reward_zero_std": 0.6666666865348816,
"grad_norm": 1.4658262729644775,
"kl": 0.0008602142333984375,
"learning_rate": 4.864308396768294e-06,
"loss": 0.0,
"num_tokens": 35224.0,
"reward": 0.008333333767950535,
"reward_std": 0.01666666753590107,
"rewards/code_reward/mean": 0.0,
"rewards/code_reward/std": 0.0,
"rewards/format_reward/mean": 0.0833333358168602,
"rewards/format_reward/std": 0.28867512941360474,
"step": 3
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": -1.75,
"completions/max_length": 2048.0,
"completions/max_terminated_length": 709.0,
"completions/mean_length": 528.1666870117188,
"completions/mean_terminated_length": 390.0,
"completions/min_length": 150.0,
"completions/min_terminated_length": 150.0,
"epoch": 0.00033580288370726384,
"frac_reward_zero_std": 0.6666666865348816,
"grad_norm": 1.4903650283813477,
"kl": 0.0033283233642578125,
"learning_rate": 4.473599997017701e-06,
"loss": 0.0,
"num_tokens": 49222.0,
"reward": 0.008333333767950535,
"reward_std": 0.01666666753590107,
"rewards/code_reward/mean": 0.0,
"rewards/code_reward/std": 0.0,
"rewards/format_reward/mean": 0.0833333358168602,
"rewards/format_reward/std": 0.28867512941360474,
"step": 4
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": -2.0,
"completions/max_length": 431.0,
"completions/max_terminated_length": 431.0,
"completions/mean_length": 299.3333435058594,
"completions/mean_terminated_length": 299.3333435058594,
"completions/min_length": 198.0,
"completions/min_terminated_length": 198.0,
"epoch": 0.0004197536046340798,
"frac_reward_zero_std": 0.0,
"grad_norm": 3.9364893436431885,
"kl": 0.005321502685546875,
"learning_rate": 3.875e-06,
"loss": 0.0001,
"num_tokens": 59278.0,
"reward": 0.05000000447034836,
"reward_std": 0.05257834494113922,
"rewards/code_reward/mean": 0.0,
"rewards/code_reward/std": 0.0,
"rewards/format_reward/mean": 0.5,
"rewards/format_reward/std": 0.5222329497337341,
"step": 5
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": -2.0,
"completions/max_length": 700.0,
"completions/max_terminated_length": 700.0,
"completions/mean_length": 377.41668701171875,
"completions/mean_terminated_length": 377.41668701171875,
"completions/min_length": 125.0,
"completions/min_terminated_length": 125.0,
"epoch": 0.0005037043255608957,
"frac_reward_zero_std": 0.6666666865348816,
"grad_norm": 2.257307291030884,
"kl": 0.005146026611328125,
"learning_rate": 3.140708399750594e-06,
"loss": 0.0001,
"num_tokens": 72151.0,
"reward": 0.05000000447034836,
"reward_std": 0.019245009869337082,
"rewards/code_reward/mean": 0.0,
"rewards/code_reward/std": 0.0,
"rewards/format_reward/mean": 0.5,
"rewards/format_reward/std": 0.5222329497337341,
"step": 6
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": -2.0,
"completions/max_length": 755.0,
"completions/max_terminated_length": 755.0,
"completions/mean_length": 463.41668701171875,
"completions/mean_terminated_length": 463.41668701171875,
"completions/min_length": 235.0,
"completions/min_terminated_length": 235.0,
"epoch": 0.0005876550464877118,
"frac_reward_zero_std": 0.0,
"grad_norm": 2.7628111839294434,
"kl": 0.00887298583984375,
"learning_rate": 2.359291600249407e-06,
"loss": 0.0001,
"num_tokens": 84448.0,
"reward": 0.06666667014360428,
"reward_std": 0.05257834494113922,
"rewards/code_reward/mean": 0.0,
"rewards/code_reward/std": 0.0,
"rewards/format_reward/mean": 0.6666666865348816,
"rewards/format_reward/std": 0.4923659861087799,
"step": 7
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": -2.0,
"completions/max_length": 956.0,
"completions/max_terminated_length": 956.0,
"completions/mean_length": 340.5,
"completions/mean_terminated_length": 340.5,
"completions/min_length": 17.0,
"completions/min_terminated_length": 17.0,
"epoch": 0.0006716057674145277,
"frac_reward_zero_std": 0.0,
"grad_norm": 12.499695777893066,
"kl": 0.0798187255859375,
"learning_rate": 1.6250000000000007e-06,
"loss": 0.0008,
"num_tokens": 96646.0,
"reward": 0.07500000298023224,
"reward_std": 0.05000000447034836,
"rewards/code_reward/mean": 0.0,
"rewards/code_reward/std": 0.0,
"rewards/format_reward/mean": 0.75,
"rewards/format_reward/std": 0.45226702094078064,
"step": 8
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": -1.75,
"completions/max_length": 2048.0,
"completions/max_terminated_length": 1160.0,
"completions/mean_length": 593.4166870117188,
"completions/mean_terminated_length": 461.18182373046875,
"completions/min_length": 208.0,
"completions/min_terminated_length": 208.0,
"epoch": 0.0007555564883413436,
"frac_reward_zero_std": 0.3333333432674408,
"grad_norm": 2.6419012546539307,
"kl": 0.03733062744140625,
"learning_rate": 1.0264000029823e-06,
"loss": 0.0004,
"num_tokens": 111479.0,
"reward": 0.05000000447034836,
"reward_std": 0.03333333507180214,
"rewards/code_reward/mean": 0.0,
"rewards/code_reward/std": 0.0,
"rewards/format_reward/mean": 0.5,
"rewards/format_reward/std": 0.5222329497337341,
"step": 9
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": -2.0,
"completions/max_length": 677.0,
"completions/max_terminated_length": 677.0,
"completions/mean_length": 376.41668701171875,
"completions/mean_terminated_length": 376.41668701171875,
"completions/min_length": 160.0,
"completions/min_terminated_length": 160.0,
"epoch": 0.0008395072092681596,
"frac_reward_zero_std": 0.3333333432674408,
"grad_norm": 14.455758094787598,
"kl": 0.0934295654296875,
"learning_rate": 6.356916032317063e-07,
"loss": 0.0009,
"num_tokens": 122756.0,
"reward": 0.06666667014360428,
"reward_std": 0.038490019738674164,
"rewards/code_reward/mean": 0.0,
"rewards/code_reward/std": 0.0,
"rewards/format_reward/mean": 0.6666666865348816,
"rewards/format_reward/std": 0.4923659861087799,
"step": 10
},
{
"epoch": 0.0008395072092681596,
"step": 10,
"total_flos": 0.0,
"train_loss": 0.00023389634034174379,
"train_runtime": 1466.2797,
"train_samples_per_second": 0.082,
"train_steps_per_second": 0.007
}
],
"logging_steps": 1.0,
"max_steps": 10,
"num_input_tokens_seen": 122756,
"num_train_epochs": 1,
"save_steps": 50,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}