2735 lines
96 KiB
JSON
2735 lines
96 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.25,
|
|
"eval_steps": 500,
|
|
"global_step": 100,
|
|
"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": 0.015625,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 1290.0,
|
|
"completions/mean_length": 550.703125,
|
|
"completions/mean_terminated_length": 526.9365234375,
|
|
"completions/min_length": 82.0,
|
|
"completions/min_terminated_length": 82.0,
|
|
"epoch": 0.0025,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 1.7431169748306274,
|
|
"learning_rate": 0.0,
|
|
"loss": 0.0,
|
|
"num_tokens": 109261.0,
|
|
"reward": 0.14374998211860657,
|
|
"reward_std": 0.2529315948486328,
|
|
"rewards/format_reward/mean": 0.34375,
|
|
"rewards/format_reward/std": 0.279951810836792,
|
|
"rewards/mcq_exact_match_reward/mean": 0.109375,
|
|
"rewards/mcq_exact_match_reward/std": 0.3145764470100403,
|
|
"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": 0.0,
|
|
"completions/max_length": 1402.0,
|
|
"completions/max_terminated_length": 1402.0,
|
|
"completions/mean_length": 491.765625,
|
|
"completions/mean_terminated_length": 491.765625,
|
|
"completions/min_length": 10.0,
|
|
"completions/min_terminated_length": 10.0,
|
|
"epoch": 0.005,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 2.454085350036621,
|
|
"learning_rate": 2e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 226478.0,
|
|
"reward": 0.22421874105930328,
|
|
"reward_std": 0.28602027893066406,
|
|
"rewards/format_reward/mean": 0.3671875,
|
|
"rewards/format_reward/std": 0.22257846593856812,
|
|
"rewards/mcq_exact_match_reward/mean": 0.1875,
|
|
"rewards/mcq_exact_match_reward/std": 0.39339789748191833,
|
|
"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": 0.0,
|
|
"completions/max_length": 988.0,
|
|
"completions/max_terminated_length": 988.0,
|
|
"completions/mean_length": 486.984375,
|
|
"completions/mean_terminated_length": 486.984375,
|
|
"completions/min_length": 105.0,
|
|
"completions/min_terminated_length": 105.0,
|
|
"epoch": 0.0075,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 1.4292494058609009,
|
|
"learning_rate": 4e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 305653.0,
|
|
"reward": 0.2554687261581421,
|
|
"reward_std": 0.3255230784416199,
|
|
"rewards/format_reward/mean": 0.3671875,
|
|
"rewards/format_reward/std": 0.28510910272598267,
|
|
"rewards/mcq_exact_match_reward/mean": 0.21875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4166666865348816,
|
|
"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": 0.0,
|
|
"completions/max_length": 1923.0,
|
|
"completions/max_terminated_length": 1923.0,
|
|
"completions/mean_length": 480.484375,
|
|
"completions/mean_terminated_length": 480.484375,
|
|
"completions/min_length": 83.0,
|
|
"completions/min_terminated_length": 83.0,
|
|
"epoch": 0.01,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 1.423560380935669,
|
|
"learning_rate": 6e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 394828.0,
|
|
"reward": 0.23515623807907104,
|
|
"reward_std": 0.39392516016960144,
|
|
"rewards/format_reward/mean": 0.3203125,
|
|
"rewards/format_reward/std": 0.27265870571136475,
|
|
"rewards/mcq_exact_match_reward/mean": 0.203125,
|
|
"rewards/mcq_exact_match_reward/std": 0.40550529956817627,
|
|
"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": 0.0,
|
|
"completions/max_length": 1132.0,
|
|
"completions/max_terminated_length": 1132.0,
|
|
"completions/mean_length": 519.15625,
|
|
"completions/mean_terminated_length": 519.15625,
|
|
"completions/min_length": 3.0,
|
|
"completions/min_terminated_length": 3.0,
|
|
"epoch": 0.0125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 12.591778755187988,
|
|
"learning_rate": 8e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 502966.0,
|
|
"reward": 0.22499997913837433,
|
|
"reward_std": 0.3240741491317749,
|
|
"rewards/format_reward/mean": 0.375,
|
|
"rewards/format_reward/std": 0.2519763112068176,
|
|
"rewards/mcq_exact_match_reward/mean": 0.1875,
|
|
"rewards/mcq_exact_match_reward/std": 0.39339789748191833,
|
|
"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": 0.0,
|
|
"completions/max_length": 1365.0,
|
|
"completions/max_terminated_length": 1365.0,
|
|
"completions/mean_length": 366.71875,
|
|
"completions/mean_terminated_length": 366.71875,
|
|
"completions/min_length": 3.0,
|
|
"completions/min_terminated_length": 3.0,
|
|
"epoch": 0.015,
|
|
"frac_reward_zero_std": 0.25,
|
|
"grad_norm": 9.619667053222656,
|
|
"learning_rate": 1e-06,
|
|
"loss": -0.0,
|
|
"num_tokens": 618196.0,
|
|
"reward": 0.24531248211860657,
|
|
"reward_std": 0.2914085388183594,
|
|
"rewards/format_reward/mean": 0.265625,
|
|
"rewards/format_reward/std": 0.2514837086200714,
|
|
"rewards/mcq_exact_match_reward/mean": 0.21875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4166666865348816,
|
|
"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": 0.0,
|
|
"completions/max_length": 1357.0,
|
|
"completions/max_terminated_length": 1357.0,
|
|
"completions/mean_length": 462.75,
|
|
"completions/mean_terminated_length": 462.75,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0175,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 8.53075122833252,
|
|
"learning_rate": 9.99726628670463e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 713652.0,
|
|
"reward": 0.17343749105930328,
|
|
"reward_std": 0.28512582182884216,
|
|
"rewards/format_reward/mean": 0.328125,
|
|
"rewards/format_reward/std": 0.31140682101249695,
|
|
"rewards/mcq_exact_match_reward/mean": 0.140625,
|
|
"rewards/mcq_exact_match_reward/std": 0.3503824472427368,
|
|
"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": 0.03125,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 1703.0,
|
|
"completions/mean_length": 539.421875,
|
|
"completions/mean_terminated_length": 490.758056640625,
|
|
"completions/min_length": 3.0,
|
|
"completions/min_terminated_length": 3.0,
|
|
"epoch": 0.02,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 48.09510040283203,
|
|
"learning_rate": 9.989068136093872e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 835071.0,
|
|
"reward": 0.3335937261581421,
|
|
"reward_std": 0.34784799814224243,
|
|
"rewards/format_reward/mean": 0.3671875,
|
|
"rewards/format_reward/std": 0.28510910272598267,
|
|
"rewards/mcq_exact_match_reward/mean": 0.296875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4604927599430084,
|
|
"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": 0.0625,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 1130.0,
|
|
"completions/mean_length": 584.421875,
|
|
"completions/mean_terminated_length": 486.85003662109375,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0225,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 6.827512264251709,
|
|
"learning_rate": 9.975414512725056e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 973882.0,
|
|
"reward": 0.31484371423721313,
|
|
"reward_std": 0.36092233657836914,
|
|
"rewards/format_reward/mean": 0.3359375,
|
|
"rewards/format_reward/std": 0.2366211861371994,
|
|
"rewards/mcq_exact_match_reward/mean": 0.28125,
|
|
"rewards/mcq_exact_match_reward/std": 0.4531635046005249,
|
|
"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": 0.0,
|
|
"completions/max_length": 1573.0,
|
|
"completions/max_terminated_length": 1573.0,
|
|
"completions/mean_length": 465.84375,
|
|
"completions/mean_terminated_length": 465.84375,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.025,
|
|
"frac_reward_zero_std": 0.125,
|
|
"grad_norm": 5.056453227996826,
|
|
"learning_rate": 9.956320346634875e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 1084952.0,
|
|
"reward": 0.1484375,
|
|
"reward_std": 0.19536465406417847,
|
|
"rewards/format_reward/mean": 0.390625,
|
|
"rewards/format_reward/std": 0.22658175230026245,
|
|
"rewards/mcq_exact_match_reward/mean": 0.109375,
|
|
"rewards/mcq_exact_match_reward/std": 0.3145764470100403,
|
|
"step": 10
|
|
},
|
|
{
|
|
"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": 0.015625,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 1522.0,
|
|
"completions/mean_length": 531.671875,
|
|
"completions/mean_terminated_length": 507.60321044921875,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0275,
|
|
"frac_reward_zero_std": 0.125,
|
|
"grad_norm": 4.861358642578125,
|
|
"learning_rate": 9.931806517013612e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 1210491.0,
|
|
"reward": 0.19921875,
|
|
"reward_std": 0.28306934237480164,
|
|
"rewards/format_reward/mean": 0.2734375,
|
|
"rewards/format_reward/std": 0.29452258348464966,
|
|
"rewards/mcq_exact_match_reward/mean": 0.171875,
|
|
"rewards/mcq_exact_match_reward/std": 0.38025420904159546,
|
|
"step": 11
|
|
},
|
|
{
|
|
"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": 0.015625,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 1927.0,
|
|
"completions/mean_length": 574.953125,
|
|
"completions/mean_terminated_length": 551.5714721679688,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.03,
|
|
"frac_reward_zero_std": 0.125,
|
|
"grad_norm": 2.9517505168914795,
|
|
"learning_rate": 9.901899829374047e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 1334152.0,
|
|
"reward": 0.07187499850988388,
|
|
"reward_std": 0.10144344717264175,
|
|
"rewards/format_reward/mean": 0.40625,
|
|
"rewards/format_reward/std": 0.21593283116817474,
|
|
"rewards/mcq_exact_match_reward/mean": 0.03125,
|
|
"rewards/mcq_exact_match_reward/std": 0.17536810040473938,
|
|
"step": 12
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 1979.0,
|
|
"completions/max_terminated_length": 1979.0,
|
|
"completions/mean_length": 574.71875,
|
|
"completions/mean_terminated_length": 574.71875,
|
|
"completions/min_length": 139.0,
|
|
"completions/min_terminated_length": 139.0,
|
|
"epoch": 0.0325,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 1.2114769220352173,
|
|
"learning_rate": 9.866632986240029e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 1446742.0,
|
|
"reward": 0.19765624403953552,
|
|
"reward_std": 0.28626081347465515,
|
|
"rewards/format_reward/mean": 0.4140625,
|
|
"rewards/format_reward/std": 0.19012710452079773,
|
|
"rewards/mcq_exact_match_reward/mean": 0.15625,
|
|
"rewards/mcq_exact_match_reward/std": 0.36596253514289856,
|
|
"step": 13
|
|
},
|
|
{
|
|
"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": 0.015625,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 1785.0,
|
|
"completions/mean_length": 340.984375,
|
|
"completions/mean_terminated_length": 313.888916015625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.035,
|
|
"frac_reward_zero_std": 0.25,
|
|
"grad_norm": 14.763101577758789,
|
|
"learning_rate": 9.826044551386742e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 1552581.0,
|
|
"reward": 0.4867187440395355,
|
|
"reward_std": 0.27787378430366516,
|
|
"rewards/format_reward/mean": 0.4921875,
|
|
"rewards/format_reward/std": 0.18881812691688538,
|
|
"rewards/mcq_exact_match_reward/mean": 0.4375,
|
|
"rewards/mcq_exact_match_reward/std": 0.5,
|
|
"step": 14
|
|
},
|
|
{
|
|
"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": 0.03125,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 1424.0,
|
|
"completions/mean_length": 512.265625,
|
|
"completions/mean_terminated_length": 462.7257995605469,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0375,
|
|
"frac_reward_zero_std": 0.125,
|
|
"grad_norm": 15.209944725036621,
|
|
"learning_rate": 9.780178907671788e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 1687854.0,
|
|
"reward": 0.234375,
|
|
"reward_std": 0.2666613459587097,
|
|
"rewards/format_reward/mean": 0.46875,
|
|
"rewards/format_reward/std": 0.19669894874095917,
|
|
"rewards/mcq_exact_match_reward/mean": 0.1875,
|
|
"rewards/mcq_exact_match_reward/std": 0.39339789748191833,
|
|
"step": 15
|
|
},
|
|
{
|
|
"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": 0.015625,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 814.0,
|
|
"completions/mean_length": 193.796875,
|
|
"completions/mean_terminated_length": 164.36508178710938,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.04,
|
|
"frac_reward_zero_std": 0.25,
|
|
"grad_norm": 13.295024871826172,
|
|
"learning_rate": 9.729086208503173e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 1773385.0,
|
|
"reward": 0.30156248807907104,
|
|
"reward_std": 0.338266521692276,
|
|
"rewards/format_reward/mean": 0.515625,
|
|
"rewards/format_reward/std": 0.1985812783241272,
|
|
"rewards/mcq_exact_match_reward/mean": 0.25,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"step": 16
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 1566.0,
|
|
"completions/max_terminated_length": 1566.0,
|
|
"completions/mean_length": 298.90625,
|
|
"completions/mean_terminated_length": 298.90625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0425,
|
|
"frac_reward_zero_std": 0.125,
|
|
"grad_norm": 14.432733535766602,
|
|
"learning_rate": 9.672822322997304e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 1883539.0,
|
|
"reward": 0.5148437023162842,
|
|
"reward_std": 0.4220954179763794,
|
|
"rewards/format_reward/mean": 0.4609375,
|
|
"rewards/format_reward/std": 0.16194961965084076,
|
|
"rewards/mcq_exact_match_reward/mean": 0.46875,
|
|
"rewards/mcq_exact_match_reward/std": 0.5029674172401428,
|
|
"step": 17
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 815.0,
|
|
"completions/max_terminated_length": 815.0,
|
|
"completions/mean_length": 180.640625,
|
|
"completions/mean_terminated_length": 180.640625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.045,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 11.126655578613281,
|
|
"learning_rate": 9.611448774886923e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 2004268.0,
|
|
"reward": 0.20156249403953552,
|
|
"reward_std": 0.12757223844528198,
|
|
"rewards/format_reward/mean": 0.453125,
|
|
"rewards/format_reward/std": 0.14689241349697113,
|
|
"rewards/mcq_exact_match_reward/mean": 0.15625,
|
|
"rewards/mcq_exact_match_reward/std": 0.36596253514289856,
|
|
"step": 18
|
|
},
|
|
{
|
|
"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": 0.015625,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 1465.0,
|
|
"completions/mean_length": 334.421875,
|
|
"completions/mean_terminated_length": 307.22222900390625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0475,
|
|
"frac_reward_zero_std": 0.25,
|
|
"grad_norm": 16.768993377685547,
|
|
"learning_rate": 9.545032675245813e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 2124791.0,
|
|
"reward": 0.234375,
|
|
"reward_std": 0.3307138979434967,
|
|
"rewards/format_reward/mean": 0.46875,
|
|
"rewards/format_reward/std": 0.12198751419782639,
|
|
"rewards/mcq_exact_match_reward/mean": 0.1875,
|
|
"rewards/mcq_exact_match_reward/std": 0.39339789748191833,
|
|
"step": 19
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 1172.0,
|
|
"completions/max_terminated_length": 1172.0,
|
|
"completions/mean_length": 201.78125,
|
|
"completions/mean_terminated_length": 201.78125,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.05,
|
|
"frac_reward_zero_std": 0.125,
|
|
"grad_norm": 17.379011154174805,
|
|
"learning_rate": 9.473646649103817e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 2230857.0,
|
|
"reward": 0.4718749523162842,
|
|
"reward_std": 0.366149365901947,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.421875,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"step": 20
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 932.0,
|
|
"completions/max_terminated_length": 932.0,
|
|
"completions/mean_length": 85.90625,
|
|
"completions/mean_terminated_length": 85.90625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0525,
|
|
"frac_reward_zero_std": 0.375,
|
|
"grad_norm": 16.797849655151367,
|
|
"learning_rate": 9.397368756032444e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 2324731.0,
|
|
"reward": 0.4718749523162842,
|
|
"reward_std": 0.24831001460552216,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.421875,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"step": 21
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 972.0,
|
|
"completions/max_terminated_length": 972.0,
|
|
"completions/mean_length": 51.15625,
|
|
"completions/mean_terminated_length": 51.15625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.055,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 10.87809944152832,
|
|
"learning_rate": 9.316282404787869e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 2418309.0,
|
|
"reward": 0.4562499523162842,
|
|
"reward_std": 0.1462520956993103,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.40625,
|
|
"rewards/mcq_exact_match_reward/std": 0.49501484632492065,
|
|
"step": 22
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 826.0,
|
|
"completions/max_terminated_length": 826.0,
|
|
"completions/mean_length": 46.609375,
|
|
"completions/mean_terminated_length": 46.609375,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0575,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 24.02557373046875,
|
|
"learning_rate": 9.230476262104676e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 2476844.0,
|
|
"reward": 0.5968749523162842,
|
|
"reward_std": 0.43556106090545654,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.546875,
|
|
"rewards/mcq_exact_match_reward/std": 0.501733124256134,
|
|
"step": 23
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 417.0,
|
|
"completions/max_terminated_length": 417.0,
|
|
"completions/mean_length": 15.515625,
|
|
"completions/mean_terminated_length": 15.515625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.06,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 22.337867736816406,
|
|
"learning_rate": 9.1400441557401e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 2547765.0,
|
|
"reward": 0.4562499523162842,
|
|
"reward_std": 0.22201894223690033,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.40625,
|
|
"rewards/mcq_exact_match_reward/std": 0.49501484632492065,
|
|
"step": 24
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 245.0,
|
|
"completions/max_terminated_length": 245.0,
|
|
"completions/mean_length": 12.5,
|
|
"completions/mean_terminated_length": 12.5,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0625,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 25.489038467407227,
|
|
"learning_rate": 9.045084971874737e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 2617885.0,
|
|
"reward": 0.26874998211860657,
|
|
"reward_std": 0.2041158527135849,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.21875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4166666865348816,
|
|
"step": 25
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 208.0,
|
|
"completions/max_terminated_length": 208.0,
|
|
"completions/mean_length": 9.15625,
|
|
"completions/mean_terminated_length": 9.15625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.065,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 19.022188186645508,
|
|
"learning_rate": 8.945702546981968e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 2695711.0,
|
|
"reward": 0.28437498211860657,
|
|
"reward_std": 0.16887322068214417,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.234375,
|
|
"rewards/mcq_exact_match_reward/std": 0.42695629596710205,
|
|
"step": 26
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0675,
|
|
"frac_reward_zero_std": 0.375,
|
|
"grad_norm": 23.002853393554688,
|
|
"learning_rate": 8.842005554284295e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 2781607.0,
|
|
"reward": 0.37812498211860657,
|
|
"reward_std": 0.2688094973564148,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.328125,
|
|
"rewards/mcq_exact_match_reward/std": 0.4732423722743988,
|
|
"step": 27
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.07,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 10.243049621582031,
|
|
"learning_rate": 8.734107384920769e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 2859703.0,
|
|
"reward": 0.23749998211860657,
|
|
"reward_std": 0.06681530177593231,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.1875,
|
|
"rewards/mcq_exact_match_reward/std": 0.39339789748191833,
|
|
"step": 28
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 567.0,
|
|
"completions/max_terminated_length": 567.0,
|
|
"completions/mean_length": 14.765625,
|
|
"completions/mean_terminated_length": 14.765625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0725,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 26.013399124145508,
|
|
"learning_rate": 8.622126023955445e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 2932728.0,
|
|
"reward": 0.5656249523162842,
|
|
"reward_std": 0.2198972851037979,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.515625,
|
|
"rewards/mcq_exact_match_reward/std": 0.5037065148353577,
|
|
"step": 29
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 553.0,
|
|
"completions/max_terminated_length": 553.0,
|
|
"completions/mean_length": 19.015625,
|
|
"completions/mean_terminated_length": 19.015625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.075,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 17.547225952148438,
|
|
"learning_rate": 8.506183921362442e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 3034289.0,
|
|
"reward": 0.22187498211860657,
|
|
"reward_std": 0.16887322068214417,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.171875,
|
|
"rewards/mcq_exact_match_reward/std": 0.38025420904159546,
|
|
"step": 30
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0775,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 13.620226860046387,
|
|
"learning_rate": 8.386407858128706e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 3123505.0,
|
|
"reward": 0.4093749523162842,
|
|
"reward_std": 0.13258251547813416,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.359375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"step": 31
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.08,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 34.841583251953125,
|
|
"learning_rate": 8.262928807620843e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 3227713.0,
|
|
"reward": 0.4093749523162842,
|
|
"reward_std": 0.13258251547813416,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.359375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"step": 32
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0825,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 18.24964714050293,
|
|
"learning_rate": 8.135881792367685e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 3313889.0,
|
|
"reward": 0.4718749523162842,
|
|
"reward_std": 0.16887322068214417,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.421875,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"step": 33
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 84.0,
|
|
"completions/max_terminated_length": 84.0,
|
|
"completions/mean_length": 7.21875,
|
|
"completions/mean_terminated_length": 7.21875,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.085,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 28.051219940185547,
|
|
"learning_rate": 8.005405736415125e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 3410999.0,
|
|
"reward": 0.7062499523162842,
|
|
"reward_std": 0.1293872892856598,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.65625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4787135720252991,
|
|
"step": 34
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0875,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 28.1188907623291,
|
|
"learning_rate": 7.871643313414718e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 3498063.0,
|
|
"reward": 0.31562498211860657,
|
|
"reward_std": 0.17358146607875824,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.265625,
|
|
"rewards/mcq_exact_match_reward/std": 0.44515693187713623,
|
|
"step": 35
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.09,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 14.5210542678833,
|
|
"learning_rate": 7.734740790612136e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 3579023.0,
|
|
"reward": 0.44062498211860657,
|
|
"reward_std": 0.17358145117759705,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.390625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4917473793029785,
|
|
"step": 36
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 10.0,
|
|
"completions/max_terminated_length": 10.0,
|
|
"completions/mean_length": 6.0625,
|
|
"completions/mean_terminated_length": 6.0625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0925,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 20.82362174987793,
|
|
"learning_rate": 7.594847868906076e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 3669291.0,
|
|
"reward": 0.4406249523162842,
|
|
"reward_std": 0.10205792635679245,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.390625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4917473793029785,
|
|
"step": 37
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.095,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 9.409337043762207,
|
|
"learning_rate": 7.452117519152541e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 3757899.0,
|
|
"reward": 0.33124998211860657,
|
|
"reward_std": 0.0578637532889843,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.28125,
|
|
"rewards/mcq_exact_match_reward/std": 0.4531635046005249,
|
|
"step": 38
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0975,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 30.472230911254883,
|
|
"learning_rate": 7.306705814893439e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 3842683.0,
|
|
"reward": 0.5031249523162842,
|
|
"reward_std": 0.0646936446428299,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.453125,
|
|
"rewards/mcq_exact_match_reward/std": 0.501733124256134,
|
|
"step": 39
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1,
|
|
"frac_reward_zero_std": 0.375,
|
|
"grad_norm": 48.22329330444336,
|
|
"learning_rate": 7.158771761692464e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 3934219.0,
|
|
"reward": 0.6749999523162842,
|
|
"reward_std": 0.3029785752296448,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 40
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1025,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 37.72251892089844,
|
|
"learning_rate": 7.008477123264847e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 4000867.0,
|
|
"reward": 0.6906249523162842,
|
|
"reward_std": 0.1804211586713791,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.640625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"step": 41
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 69.0,
|
|
"completions/max_terminated_length": 69.0,
|
|
"completions/mean_length": 6.984375,
|
|
"completions/mean_terminated_length": 6.984375,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.105,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 29.031400680541992,
|
|
"learning_rate": 6.855986244591103e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 4077722.0,
|
|
"reward": 0.3937499523162842,
|
|
"reward_std": 0.12467905879020691,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.34375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4787135720252991,
|
|
"step": 42
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1075,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 6.701465872208216e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 4143178.0,
|
|
"reward": 0.7999999523162842,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"step": 43
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.11,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 20.480224609375,
|
|
"learning_rate": 6.545084971874736e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 4232802.0,
|
|
"reward": 0.6906249523162842,
|
|
"reward_std": 0.13258251547813416,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.640625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"step": 44
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1125,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 10.604009628295898,
|
|
"learning_rate": 6.387014543809223e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 4317394.0,
|
|
"reward": 0.7999999523162842,
|
|
"reward_std": 0.0883883386850357,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"step": 45
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.115,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 20.595806121826172,
|
|
"learning_rate": 6.227427435703995e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 4418666.0,
|
|
"reward": 0.4093749523162842,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.359375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"step": 46
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1175,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 20.427709579467773,
|
|
"learning_rate": 6.066498153718734e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 4500402.0,
|
|
"reward": 0.7062499523162842,
|
|
"reward_std": 0.10888782143592834,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.65625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4787135720252991,
|
|
"step": 47
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.12,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 26.35272979736328,
|
|
"learning_rate": 5.90440267166055e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 4577146.0,
|
|
"reward": 0.8312499523162842,
|
|
"reward_std": 0.12467905879020691,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.78125,
|
|
"rewards/mcq_exact_match_reward/std": 0.4166666865348816,
|
|
"step": 48
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1225,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 5.741318238559209e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 4648418.0,
|
|
"reward": 0.5499999523162842,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"step": 49
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.125,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 32.905418395996094,
|
|
"learning_rate": 5.577423184847931e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 4725514.0,
|
|
"reward": 0.6281249523162842,
|
|
"reward_std": 0.16887322068214417,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.578125,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"step": 50
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 103.0,
|
|
"completions/max_terminated_length": 103.0,
|
|
"completions/mean_length": 7.515625,
|
|
"completions/mean_terminated_length": 7.515625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1275,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 10.500630378723145,
|
|
"learning_rate": 5.412896727361662e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 4798115.0,
|
|
"reward": 1.0343749523162842,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.984375,
|
|
"rewards/mcq_exact_match_reward/std": 0.125,
|
|
"step": 51
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.13,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 25.986589431762695,
|
|
"learning_rate": 5.247918773366111e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 4866267.0,
|
|
"reward": 0.6593749523162842,
|
|
"reward_std": 0.16887322068214417,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.609375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4917473793029785,
|
|
"step": 52
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1325,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 8.775482177734375,
|
|
"learning_rate": 5.082669723831793e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 4950787.0,
|
|
"reward": 0.6281249523162842,
|
|
"reward_std": 0.11100947111845016,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.578125,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"step": 53
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.135,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 4.917330276168208e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 5026387.0,
|
|
"reward": 0.6749999523162842,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 54
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1375,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 9.289410591125488,
|
|
"learning_rate": 4.752081226633888e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 5094203.0,
|
|
"reward": 0.4093749523162842,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.359375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"step": 55
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.14,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 17.342201232910156,
|
|
"learning_rate": 4.5871032726383385e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 5167995.0,
|
|
"reward": 0.4406249523162842,
|
|
"reward_std": 0.04419416934251785,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.390625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4917473793029785,
|
|
"step": 56
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1425,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 9.582889556884766,
|
|
"learning_rate": 4.4225768151520694e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 5246723.0,
|
|
"reward": 0.6906249523162842,
|
|
"reward_std": 0.04419416934251785,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.640625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"step": 57
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.145,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 39.130271911621094,
|
|
"learning_rate": 4.258681761440789e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 5312019.0,
|
|
"reward": 0.6124999523162842,
|
|
"reward_std": 0.06681530177593231,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5625,
|
|
"rewards/mcq_exact_match_reward/std": 0.5,
|
|
"step": 58
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1475,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 8.925666809082031,
|
|
"learning_rate": 4.095597328339452e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 5402123.0,
|
|
"reward": 0.5187499523162842,
|
|
"reward_std": 0.0578637532889843,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.46875,
|
|
"rewards/mcq_exact_match_reward/std": 0.5029674172401428,
|
|
"step": 59
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.15,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 29.714693069458008,
|
|
"learning_rate": 3.9335018462812664e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 5487995.0,
|
|
"reward": 0.6281249523162842,
|
|
"reward_std": 0.12255740165710449,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.578125,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"step": 60
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1525,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 66.1559066772461,
|
|
"learning_rate": 3.772572564296004e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 5576715.0,
|
|
"reward": 0.7999999523162842,
|
|
"reward_std": 0.13363060355186462,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"step": 61
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.155,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 26.079076766967773,
|
|
"learning_rate": 3.612985456190778e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 5672243.0,
|
|
"reward": 0.4406249523162842,
|
|
"reward_std": 0.04419416934251785,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.390625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4917473793029785,
|
|
"step": 62
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1575,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 13.241816520690918,
|
|
"learning_rate": 3.454915028125263e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 5771299.0,
|
|
"reward": 0.6593749523162842,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.609375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4917473793029785,
|
|
"step": 63
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.16,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 35.15870666503906,
|
|
"learning_rate": 3.2985341277917846e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 5849115.0,
|
|
"reward": 0.5031249523162842,
|
|
"reward_std": 0.10205792635679245,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.453125,
|
|
"rewards/mcq_exact_match_reward/std": 0.501733124256134,
|
|
"step": 64
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1625,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 64.22975158691406,
|
|
"learning_rate": 3.1440137554088953e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 5933275.0,
|
|
"reward": 0.5968749523162842,
|
|
"reward_std": 0.1804211586713791,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.546875,
|
|
"rewards/mcq_exact_match_reward/std": 0.501733124256134,
|
|
"step": 65
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.165,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 33.53551483154297,
|
|
"learning_rate": 2.9915228767351535e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6006987.0,
|
|
"reward": 0.25312498211860657,
|
|
"reward_std": 0.1530819833278656,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.203125,
|
|
"rewards/mcq_exact_match_reward/std": 0.40550529956817627,
|
|
"step": 66
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1675,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 11.71260929107666,
|
|
"learning_rate": 2.841228238307536e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6105795.0,
|
|
"reward": 0.31562498211860657,
|
|
"reward_std": 0.04419416934251785,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.265625,
|
|
"rewards/mcq_exact_match_reward/std": 0.44515693187713623,
|
|
"step": 67
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.17,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 2.6932941851065615e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6194587.0,
|
|
"reward": 0.6749999523162842,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 68
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1725,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 31.892173767089844,
|
|
"learning_rate": 2.547882480847461e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 6277163.0,
|
|
"reward": 0.6749999523162842,
|
|
"reward_std": 0.1157275065779686,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 69
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.175,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 26.409669876098633,
|
|
"learning_rate": 2.4051521310939254e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6367835.0,
|
|
"reward": 0.7843749523162842,
|
|
"reward_std": 0.13258251547813416,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.734375,
|
|
"rewards/mcq_exact_match_reward/std": 0.44515693187713623,
|
|
"step": 70
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 150.0,
|
|
"completions/max_terminated_length": 150.0,
|
|
"completions/mean_length": 8.25,
|
|
"completions/mean_terminated_length": 8.25,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1775,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 40.09273147583008,
|
|
"learning_rate": 2.2652592093878665e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 6433763.0,
|
|
"reward": 0.4406249523162842,
|
|
"reward_std": 0.1804211586713791,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.390625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4917473793029785,
|
|
"step": 71
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.18,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 15.880814552307129,
|
|
"learning_rate": 2.128356686585282e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6513155.0,
|
|
"reward": 0.8156249523162842,
|
|
"reward_std": 0.04419416934251785,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.765625,
|
|
"rewards/mcq_exact_match_reward/std": 0.42695629596710205,
|
|
"step": 72
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1825,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 17.474756240844727,
|
|
"learning_rate": 1.9945942635848745e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 6573699.0,
|
|
"reward": 0.5187499523162842,
|
|
"reward_std": 0.0578637532889843,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.46875,
|
|
"rewards/mcq_exact_match_reward/std": 0.5029674172401428,
|
|
"step": 73
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.185,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 29.215700149536133,
|
|
"learning_rate": 1.8641182076323148e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6658051.0,
|
|
"reward": 0.5499999523162842,
|
|
"reward_std": 0.1767766922712326,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"step": 74
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1875,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 23.075483322143555,
|
|
"learning_rate": 1.7370711923791564e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 6754411.0,
|
|
"reward": 0.6281249523162842,
|
|
"reward_std": 0.0646936446428299,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.578125,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"step": 75
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.19,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 28.799482345581055,
|
|
"learning_rate": 1.6135921418712955e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 6821875.0,
|
|
"reward": 0.6281249523162842,
|
|
"reward_std": 0.0646936446428299,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.578125,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"step": 76
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1925,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.493816078637557e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6906739.0,
|
|
"reward": 0.6749999523162842,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 77
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.195,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.3778739760445552e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6991147.0,
|
|
"reward": 0.6749999523162842,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 78
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.1975,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 7.471078395843506,
|
|
"learning_rate": 1.2658926150792322e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 7071699.0,
|
|
"reward": 0.31562498211860657,
|
|
"reward_std": 0.04419416934251785,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.265625,
|
|
"rewards/mcq_exact_match_reward/std": 0.44515693187713623,
|
|
"step": 79
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.2,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 27.225492477416992,
|
|
"learning_rate": 1.1579944457157059e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 7142763.0,
|
|
"reward": 0.7218749523162842,
|
|
"reward_std": 0.0646936446428299,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.671875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4732423722743988,
|
|
"step": 80
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.2025,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 29.588693618774414,
|
|
"learning_rate": 1.0542974530180327e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 7219091.0,
|
|
"reward": 0.6281249523162842,
|
|
"reward_std": 0.16887323558330536,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.578125,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"step": 81
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 8.0,
|
|
"completions/max_terminated_length": 8.0,
|
|
"completions/mean_length": 6.03125,
|
|
"completions/mean_terminated_length": 6.03125,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.205,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 4.567900657653809,
|
|
"learning_rate": 9.549150281252632e-08,
|
|
"loss": -0.0,
|
|
"num_tokens": 7299973.0,
|
|
"reward": 0.5648437142372131,
|
|
"reward_std": 0.04456353932619095,
|
|
"rewards/format_reward/mean": 0.4921875,
|
|
"rewards/format_reward/std": 0.0625,
|
|
"rewards/mcq_exact_match_reward/mean": 0.515625,
|
|
"rewards/mcq_exact_match_reward/std": 0.5037065148353577,
|
|
"step": 82
|
|
},
|
|
{
|
|
"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": 0.015625,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 37.90625,
|
|
"completions/mean_terminated_length": 6.000000476837158,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.2075,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 0.5184043645858765,
|
|
"learning_rate": 8.599558442598998e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 7415463.0,
|
|
"reward": 0.42421871423721313,
|
|
"reward_std": 0.00220970856025815,
|
|
"rewards/format_reward/mean": 0.4921875,
|
|
"rewards/format_reward/std": 0.0625,
|
|
"rewards/mcq_exact_match_reward/mean": 0.375,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 83
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.21,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 39.599239349365234,
|
|
"learning_rate": 7.695237378953224e-08,
|
|
"loss": -0.0,
|
|
"num_tokens": 7483199.0,
|
|
"reward": 0.45624998211860657,
|
|
"reward_std": 0.1293872892856598,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.40625,
|
|
"rewards/mcq_exact_match_reward/std": 0.49501484632492065,
|
|
"step": 84
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.2125,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 8.963640213012695,
|
|
"learning_rate": 6.837175952121304e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 7579639.0,
|
|
"reward": 0.8156249523162842,
|
|
"reward_std": 0.04419416934251785,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.765625,
|
|
"rewards/mcq_exact_match_reward/std": 0.42695629596710205,
|
|
"step": 85
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.215,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 6.026312439675551e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 7652935.0,
|
|
"reward": 0.6749999523162842,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 86
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 8.0,
|
|
"completions/max_terminated_length": 8.0,
|
|
"completions/mean_length": 6.125,
|
|
"completions/mean_terminated_length": 6.125,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.2175,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 31.712818145751953,
|
|
"learning_rate": 5.263533508961826e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 7734383.0,
|
|
"reward": 0.546875,
|
|
"reward_std": 0.09172910451889038,
|
|
"rewards/format_reward/mean": 0.46875,
|
|
"rewards/format_reward/std": 0.12198751419782639,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"step": 87
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.22,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 16.61139488220215,
|
|
"learning_rate": 4.549673247541874e-08,
|
|
"loss": -0.0,
|
|
"num_tokens": 7842743.0,
|
|
"reward": 0.6906249523162842,
|
|
"reward_std": 0.10205792635679245,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.640625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"step": 88
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.2225,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 7.404222011566162,
|
|
"learning_rate": 3.8855122511307626e-08,
|
|
"loss": -0.0,
|
|
"num_tokens": 7957647.0,
|
|
"reward": 0.31484371423721313,
|
|
"reward_std": 0.04456353932619095,
|
|
"rewards/format_reward/mean": 0.4921875,
|
|
"rewards/format_reward/std": 0.0625,
|
|
"rewards/mcq_exact_match_reward/mean": 0.265625,
|
|
"rewards/mcq_exact_match_reward/std": 0.44515693187713623,
|
|
"step": 89
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.225,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 56.98033905029297,
|
|
"learning_rate": 3.271776770026963e-08,
|
|
"loss": -0.0,
|
|
"num_tokens": 8044399.0,
|
|
"reward": 0.7218749523162842,
|
|
"reward_std": 0.11100947856903076,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.671875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4732423722743988,
|
|
"step": 90
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.2275,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 35.87734603881836,
|
|
"learning_rate": 2.7091379149682682e-08,
|
|
"loss": -0.0,
|
|
"num_tokens": 8113327.0,
|
|
"reward": 0.7999999523162842,
|
|
"reward_std": 0.17570310831069946,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"step": 91
|
|
},
|
|
{
|
|
"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": 0.03125,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 1450.0,
|
|
"completions/mean_length": 174.078125,
|
|
"completions/mean_terminated_length": 113.6290283203125,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.23,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 17.009075164794922,
|
|
"learning_rate": 2.1982109232821176e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 8227052.0,
|
|
"reward": 0.7937499284744263,
|
|
"reward_std": 0.0883883386850357,
|
|
"rewards/format_reward/mean": 0.4375,
|
|
"rewards/format_reward/std": 0.1666666716337204,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"step": 92
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.2325,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 21.320192337036133,
|
|
"learning_rate": 1.7395544861325718e-08,
|
|
"loss": -0.0,
|
|
"num_tokens": 8318356.0,
|
|
"reward": 0.8781249523162842,
|
|
"reward_std": 0.0646936446428299,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.828125,
|
|
"rewards/mcq_exact_match_reward/std": 0.38025420904159546,
|
|
"step": 93
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.235,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 19.22350311279297,
|
|
"learning_rate": 1.3336701375997127e-08,
|
|
"loss": -0.0,
|
|
"num_tokens": 8394124.0,
|
|
"reward": 0.4562499523162842,
|
|
"reward_std": 0.0578637532889843,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.40625,
|
|
"rewards/mcq_exact_match_reward/std": 0.49501484632492065,
|
|
"step": 94
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.2375,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 9.810017062595321e-09,
|
|
"loss": 0.0,
|
|
"num_tokens": 8472076.0,
|
|
"reward": 0.6749999523162842,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 95
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.24,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 6.819348298638839e-09,
|
|
"loss": 0.0,
|
|
"num_tokens": 8542732.0,
|
|
"reward": 0.6749999523162842,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 96
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.2425,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 18.509517669677734,
|
|
"learning_rate": 4.367965336512403e-09,
|
|
"loss": -0.0,
|
|
"num_tokens": 8623028.0,
|
|
"reward": 0.4562499523162842,
|
|
"reward_std": 0.0578637532889843,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.40625,
|
|
"rewards/mcq_exact_match_reward/std": 0.49501484632492065,
|
|
"step": 97
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 6.0,
|
|
"completions/max_terminated_length": 6.0,
|
|
"completions/mean_length": 6.0,
|
|
"completions/mean_terminated_length": 6.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.245,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 2.458548727494292e-09,
|
|
"loss": 0.0,
|
|
"num_tokens": 8696916.0,
|
|
"reward": 0.5499999523162842,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"step": 98
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 96.0,
|
|
"completions/max_terminated_length": 96.0,
|
|
"completions/mean_length": 7.40625,
|
|
"completions/mean_terminated_length": 7.40625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.2475,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 23.50063705444336,
|
|
"learning_rate": 1.0931863906127325e-09,
|
|
"loss": -0.0,
|
|
"num_tokens": 8773158.0,
|
|
"reward": 0.6124999523162842,
|
|
"reward_std": 0.06681530177593231,
|
|
"rewards/format_reward/mean": 0.5,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5625,
|
|
"rewards/mcq_exact_match_reward/std": 0.5,
|
|
"step": 99
|
|
},
|
|
{
|
|
"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": 0.0,
|
|
"completions/max_length": 276.0,
|
|
"completions/max_terminated_length": 276.0,
|
|
"completions/mean_length": 10.21875,
|
|
"completions/mean_terminated_length": 10.21875,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.25,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 25.21071434020996,
|
|
"learning_rate": 2.733713295369755e-10,
|
|
"loss": 0.0,
|
|
"num_tokens": 8887276.0,
|
|
"reward": 0.3882812261581421,
|
|
"reward_std": 0.11109752207994461,
|
|
"rewards/format_reward/mean": 0.4453125,
|
|
"rewards/format_reward/std": 0.15728822350502014,
|
|
"rewards/mcq_exact_match_reward/mean": 0.34375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4787135720252991,
|
|
"step": 100
|
|
}
|
|
],
|
|
"logging_steps": 1,
|
|
"max_steps": 100,
|
|
"num_input_tokens_seen": 8887276,
|
|
"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
|
|
}
|