5435 lines
192 KiB
JSON
5435 lines
192 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.25,
|
|
"eval_steps": 500,
|
|
"global_step": 200,
|
|
"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": 1470.0,
|
|
"completions/mean_length": 444.328125,
|
|
"completions/mean_terminated_length": 418.873046875,
|
|
"completions/min_length": 3.0,
|
|
"completions/min_terminated_length": 3.0,
|
|
"epoch": 0.00125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 12.492976188659668,
|
|
"learning_rate": 0.0,
|
|
"loss": -0.0,
|
|
"num_tokens": 111893.0,
|
|
"reward": 0.28359371423721313,
|
|
"reward_std": 0.38594749569892883,
|
|
"rewards/format_reward/mean": 0.3359375,
|
|
"rewards/format_reward/std": 0.321882963180542,
|
|
"rewards/mcq_exact_match_reward/mean": 0.25,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"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": 1021.0,
|
|
"completions/max_terminated_length": 1021.0,
|
|
"completions/mean_length": 391.3125,
|
|
"completions/mean_terminated_length": 391.3125,
|
|
"completions/min_length": 11.0,
|
|
"completions/min_terminated_length": 11.0,
|
|
"epoch": 0.0025,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 6.380683898925781,
|
|
"learning_rate": 1e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 200761.0,
|
|
"reward": 0.1484375,
|
|
"reward_std": 0.16173049807548523,
|
|
"rewards/format_reward/mean": 0.390625,
|
|
"rewards/format_reward/std": 0.37234511971473694,
|
|
"rewards/mcq_exact_match_reward/mean": 0.109375,
|
|
"rewards/mcq_exact_match_reward/std": 0.3145764470100403,
|
|
"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": 0.015625,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 1195.0,
|
|
"completions/mean_length": 412.703125,
|
|
"completions/mean_terminated_length": 386.7460632324219,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.00375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 7.972509860992432,
|
|
"learning_rate": 2e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 308846.0,
|
|
"reward": 0.4562499523162842,
|
|
"reward_std": 0.4088757038116455,
|
|
"rewards/format_reward/mean": 0.34375,
|
|
"rewards/format_reward/std": 0.25,
|
|
"rewards/mcq_exact_match_reward/mean": 0.421875,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"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": 1460.0,
|
|
"completions/max_terminated_length": 1460.0,
|
|
"completions/mean_length": 404.453125,
|
|
"completions/mean_terminated_length": 404.453125,
|
|
"completions/min_length": 63.0,
|
|
"completions/min_terminated_length": 63.0,
|
|
"epoch": 0.005,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 1.6674329042434692,
|
|
"learning_rate": 3e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 403371.0,
|
|
"reward": 0.3695312440395355,
|
|
"reward_std": 0.43215417861938477,
|
|
"rewards/format_reward/mean": 0.4140625,
|
|
"rewards/format_reward/std": 0.3398726284503937,
|
|
"rewards/mcq_exact_match_reward/mean": 0.328125,
|
|
"rewards/mcq_exact_match_reward/std": 0.4732423722743988,
|
|
"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": 1727.0,
|
|
"completions/max_terminated_length": 1727.0,
|
|
"completions/mean_length": 491.453125,
|
|
"completions/mean_terminated_length": 491.453125,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.00625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 4.789487838745117,
|
|
"learning_rate": 4e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 512176.0,
|
|
"reward": 0.27265623211860657,
|
|
"reward_std": 0.37521791458129883,
|
|
"rewards/format_reward/mean": 0.3828125,
|
|
"rewards/format_reward/std": 0.2313210517168045,
|
|
"rewards/mcq_exact_match_reward/mean": 0.234375,
|
|
"rewards/mcq_exact_match_reward/std": 0.42695629596710205,
|
|
"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": 1386.0,
|
|
"completions/max_terminated_length": 1386.0,
|
|
"completions/mean_length": 511.53125,
|
|
"completions/mean_terminated_length": 511.53125,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.0075,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 6.271266460418701,
|
|
"learning_rate": 5e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 618002.0,
|
|
"reward": 0.3203124701976776,
|
|
"reward_std": 0.4231240153312683,
|
|
"rewards/format_reward/mean": 0.390625,
|
|
"rewards/format_reward/std": 0.28824523091316223,
|
|
"rewards/mcq_exact_match_reward/mean": 0.28125,
|
|
"rewards/mcq_exact_match_reward/std": 0.4531635046005249,
|
|
"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": 1506.0,
|
|
"completions/max_terminated_length": 1506.0,
|
|
"completions/mean_length": 363.5,
|
|
"completions/mean_terminated_length": 363.5,
|
|
"completions/min_length": 2.0,
|
|
"completions/min_terminated_length": 2.0,
|
|
"epoch": 0.00875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 33.56215286254883,
|
|
"learning_rate": 6e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 720306.0,
|
|
"reward": 0.39687496423721313,
|
|
"reward_std": 0.4708775281906128,
|
|
"rewards/format_reward/mean": 0.375,
|
|
"rewards/format_reward/std": 0.28171807527542114,
|
|
"rewards/mcq_exact_match_reward/mean": 0.359375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"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": 0.0,
|
|
"completions/max_length": 1197.0,
|
|
"completions/max_terminated_length": 1197.0,
|
|
"completions/mean_length": 460.0,
|
|
"completions/mean_terminated_length": 460.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.01,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 8.380439758300781,
|
|
"learning_rate": 7e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 830122.0,
|
|
"reward": 0.3749999701976776,
|
|
"reward_std": 0.48552989959716797,
|
|
"rewards/format_reward/mean": 0.3125,
|
|
"rewards/format_reward/std": 0.24397502839565277,
|
|
"rewards/mcq_exact_match_reward/mean": 0.34375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4787135720252991,
|
|
"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": 0.0,
|
|
"completions/max_length": 1763.0,
|
|
"completions/max_terminated_length": 1763.0,
|
|
"completions/mean_length": 383.03125,
|
|
"completions/mean_terminated_length": 383.03125,
|
|
"completions/min_length": 3.0,
|
|
"completions/min_terminated_length": 3.0,
|
|
"epoch": 0.01125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 5.271634578704834,
|
|
"learning_rate": 8e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 924628.0,
|
|
"reward": 0.19374999403953552,
|
|
"reward_std": 0.2776850759983063,
|
|
"rewards/format_reward/mean": 0.375,
|
|
"rewards/format_reward/std": 0.3333333432674408,
|
|
"rewards/mcq_exact_match_reward/mean": 0.15625,
|
|
"rewards/mcq_exact_match_reward/std": 0.36596253514289856,
|
|
"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": 1895.0,
|
|
"completions/max_terminated_length": 1895.0,
|
|
"completions/mean_length": 612.125,
|
|
"completions/mean_terminated_length": 612.125,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0125,
|
|
"frac_reward_zero_std": 0.25,
|
|
"grad_norm": 10.379047393798828,
|
|
"learning_rate": 9e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 1064932.0,
|
|
"reward": 0.20703125,
|
|
"reward_std": 0.2409304976463318,
|
|
"rewards/format_reward/mean": 0.3515625,
|
|
"rewards/format_reward/std": 0.2302463799715042,
|
|
"rewards/mcq_exact_match_reward/mean": 0.171875,
|
|
"rewards/mcq_exact_match_reward/std": 0.38025420904159546,
|
|
"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": 0.0,
|
|
"completions/max_length": 1811.0,
|
|
"completions/max_terminated_length": 1811.0,
|
|
"completions/mean_length": 483.625,
|
|
"completions/mean_terminated_length": 483.625,
|
|
"completions/min_length": 3.0,
|
|
"completions/min_terminated_length": 3.0,
|
|
"epoch": 0.01375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 11.598185539245605,
|
|
"learning_rate": 1e-06,
|
|
"loss": 0.0,
|
|
"num_tokens": 1171476.0,
|
|
"reward": 0.12734374403953552,
|
|
"reward_std": 0.177528977394104,
|
|
"rewards/format_reward/mean": 0.3359375,
|
|
"rewards/format_reward/std": 0.2824873626232147,
|
|
"rewards/mcq_exact_match_reward/mean": 0.09375,
|
|
"rewards/mcq_exact_match_reward/std": 0.29378482699394226,
|
|
"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": 0.0,
|
|
"completions/max_length": 958.0,
|
|
"completions/max_terminated_length": 958.0,
|
|
"completions/mean_length": 329.09375,
|
|
"completions/mean_terminated_length": 329.09375,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.015,
|
|
"frac_reward_zero_std": 0.25,
|
|
"grad_norm": 5.369181156158447,
|
|
"learning_rate": 9.999316524962345e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 1260026.0,
|
|
"reward": 0.21171873807907104,
|
|
"reward_std": 0.2331954836845398,
|
|
"rewards/format_reward/mean": 0.2421875,
|
|
"rewards/format_reward/std": 0.3211115300655365,
|
|
"rewards/mcq_exact_match_reward/mean": 0.1875,
|
|
"rewards/mcq_exact_match_reward/std": 0.39339789748191833,
|
|
"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": 1264.0,
|
|
"completions/max_terminated_length": 1264.0,
|
|
"completions/mean_length": 389.453125,
|
|
"completions/mean_terminated_length": 389.453125,
|
|
"completions/min_length": 13.0,
|
|
"completions/min_terminated_length": 13.0,
|
|
"epoch": 0.01625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 8.79492473602295,
|
|
"learning_rate": 9.99726628670463e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 1357183.0,
|
|
"reward": 0.3945312201976776,
|
|
"reward_std": 0.4425230622291565,
|
|
"rewards/format_reward/mean": 0.3515625,
|
|
"rewards/format_reward/std": 0.3172261714935303,
|
|
"rewards/mcq_exact_match_reward/mean": 0.359375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"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": 0.0,
|
|
"completions/max_length": 1744.0,
|
|
"completions/max_terminated_length": 1744.0,
|
|
"completions/mean_length": 576.140625,
|
|
"completions/mean_terminated_length": 576.140625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0175,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 1.769424557685852,
|
|
"learning_rate": 9.993849845741523e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 1488248.0,
|
|
"reward": 0.3843749761581421,
|
|
"reward_std": 0.39364832639694214,
|
|
"rewards/format_reward/mean": 0.40625,
|
|
"rewards/format_reward/std": 0.265398770570755,
|
|
"rewards/mcq_exact_match_reward/mean": 0.34375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4787135720252991,
|
|
"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": 0.015625,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 1769.0,
|
|
"completions/mean_length": 675.703125,
|
|
"completions/mean_terminated_length": 653.920654296875,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.01875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 3.2238948345184326,
|
|
"learning_rate": 9.989068136093872e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 1621333.0,
|
|
"reward": 0.23124998807907104,
|
|
"reward_std": 0.39479735493659973,
|
|
"rewards/format_reward/mean": 0.4375,
|
|
"rewards/format_reward/std": 0.208927720785141,
|
|
"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": 0.0,
|
|
"completions/max_length": 1123.0,
|
|
"completions/max_terminated_length": 1123.0,
|
|
"completions/mean_length": 436.640625,
|
|
"completions/mean_terminated_length": 436.640625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.02,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 7.444122314453125,
|
|
"learning_rate": 9.982922465033348e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 1706278.0,
|
|
"reward": 0.2320312261581421,
|
|
"reward_std": 0.30927249789237976,
|
|
"rewards/format_reward/mean": 0.4453125,
|
|
"rewards/format_reward/std": 0.3347255289554596,
|
|
"rewards/mcq_exact_match_reward/mean": 0.1875,
|
|
"rewards/mcq_exact_match_reward/std": 0.39339789748191833,
|
|
"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": 1502.0,
|
|
"completions/max_terminated_length": 1502.0,
|
|
"completions/mean_length": 445.421875,
|
|
"completions/mean_terminated_length": 445.421875,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.02125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 11.103367805480957,
|
|
"learning_rate": 9.975414512725056e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 1816705.0,
|
|
"reward": 0.2890625,
|
|
"reward_std": 0.25972992181777954,
|
|
"rewards/format_reward/mean": 0.390625,
|
|
"rewards/format_reward/std": 0.2741328477859497,
|
|
"rewards/mcq_exact_match_reward/mean": 0.25,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"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": 0.015625,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 978.0,
|
|
"completions/mean_length": 469.125,
|
|
"completions/mean_terminated_length": 444.0635070800781,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0225,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 5.469852447509766,
|
|
"learning_rate": 9.966546331768192e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 1919785.0,
|
|
"reward": 0.375,
|
|
"reward_std": 0.20704269409179688,
|
|
"rewards/format_reward/mean": 0.46875,
|
|
"rewards/format_reward/std": 0.33184191584587097,
|
|
"rewards/mcq_exact_match_reward/mean": 0.328125,
|
|
"rewards/mcq_exact_match_reward/std": 0.4732423722743988,
|
|
"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": 0.0,
|
|
"completions/max_length": 1197.0,
|
|
"completions/max_terminated_length": 1197.0,
|
|
"completions/mean_length": 389.359375,
|
|
"completions/mean_terminated_length": 389.359375,
|
|
"completions/min_length": 18.0,
|
|
"completions/min_terminated_length": 18.0,
|
|
"epoch": 0.02375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 1.9016674757003784,
|
|
"learning_rate": 9.956320346634875e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 2012040.0,
|
|
"reward": 0.1914062350988388,
|
|
"reward_std": 0.26537030935287476,
|
|
"rewards/format_reward/mean": 0.6640625,
|
|
"rewards/format_reward/std": 0.29620200395584106,
|
|
"rewards/mcq_exact_match_reward/mean": 0.125,
|
|
"rewards/mcq_exact_match_reward/std": 0.3333333432674408,
|
|
"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": 1039.0,
|
|
"completions/max_terminated_length": 1039.0,
|
|
"completions/mean_length": 317.625,
|
|
"completions/mean_terminated_length": 317.625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.025,
|
|
"frac_reward_zero_std": 0.125,
|
|
"grad_norm": 6.792388439178467,
|
|
"learning_rate": 9.944739353007341e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 2113872.0,
|
|
"reward": 0.42265623807907104,
|
|
"reward_std": 0.23530299961566925,
|
|
"rewards/format_reward/mean": 0.6328125,
|
|
"rewards/format_reward/std": 0.34780704975128174,
|
|
"rewards/mcq_exact_match_reward/mean": 0.359375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"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": 0.03125,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 1429.0,
|
|
"completions/mean_length": 353.8125,
|
|
"completions/mean_terminated_length": 299.1612854003906,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.02625,
|
|
"frac_reward_zero_std": 0.125,
|
|
"grad_norm": 8.842331886291504,
|
|
"learning_rate": 9.931806517013612e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 2226644.0,
|
|
"reward": 0.17812499403953552,
|
|
"reward_std": 0.24255229532718658,
|
|
"rewards/format_reward/mean": 0.53125,
|
|
"rewards/format_reward/std": 0.3549536168575287,
|
|
"rewards/mcq_exact_match_reward/mean": 0.125,
|
|
"rewards/mcq_exact_match_reward/std": 0.3333333432674408,
|
|
"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": 0.015625,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 1069.0,
|
|
"completions/mean_length": 283.078125,
|
|
"completions/mean_terminated_length": 255.06350708007812,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0275,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 11.046497344970703,
|
|
"learning_rate": 9.917525374361911e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 2323505.0,
|
|
"reward": 0.36640623211860657,
|
|
"reward_std": 0.3313905894756317,
|
|
"rewards/format_reward/mean": 0.6953125,
|
|
"rewards/format_reward/std": 0.3406015634536743,
|
|
"rewards/mcq_exact_match_reward/mean": 0.296875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4604927599430084,
|
|
"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": 1919.0,
|
|
"completions/max_terminated_length": 1919.0,
|
|
"completions/mean_length": 372.9375,
|
|
"completions/mean_terminated_length": 372.9375,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.02875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 12.129836082458496,
|
|
"learning_rate": 9.901899829374047e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 2440597.0,
|
|
"reward": 0.35624998807907104,
|
|
"reward_std": 0.4260106682777405,
|
|
"rewards/format_reward/mean": 0.59375,
|
|
"rewards/format_reward/std": 0.3435921370983124,
|
|
"rewards/mcq_exact_match_reward/mean": 0.296875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4604927599430084,
|
|
"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": 0.015625,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 1689.0,
|
|
"completions/mean_length": 222.78125,
|
|
"completions/mean_terminated_length": 193.80953979492188,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.03,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 18.581729888916016,
|
|
"learning_rate": 9.884934153917996e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 2543559.0,
|
|
"reward": 0.39921873807907104,
|
|
"reward_std": 0.3741254508495331,
|
|
"rewards/format_reward/mean": 0.7109375,
|
|
"rewards/format_reward/std": 0.42542871832847595,
|
|
"rewards/mcq_exact_match_reward/mean": 0.328125,
|
|
"rewards/mcq_exact_match_reward/std": 0.4732423722743988,
|
|
"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": 0.015625,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 1669.0,
|
|
"completions/mean_length": 219.9375,
|
|
"completions/mean_terminated_length": 190.92063903808594,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.03125,
|
|
"frac_reward_zero_std": 0.25,
|
|
"grad_norm": 5.54977560043335,
|
|
"learning_rate": 9.866632986240029e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 2672467.0,
|
|
"reward": 0.265625,
|
|
"reward_std": 0.2895887494087219,
|
|
"rewards/format_reward/mean": 0.78125,
|
|
"rewards/format_reward/std": 0.3549536168575287,
|
|
"rewards/mcq_exact_match_reward/mean": 0.1875,
|
|
"rewards/mcq_exact_match_reward/std": 0.39339789748191833,
|
|
"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": 0.03125,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 470.0,
|
|
"completions/mean_length": 162.234375,
|
|
"completions/mean_terminated_length": 101.4032211303711,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0325,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 20.450454711914062,
|
|
"learning_rate": 9.847001329696652e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 2745754.0,
|
|
"reward": 0.5093749761581421,
|
|
"reward_std": 0.4139128029346466,
|
|
"rewards/format_reward/mean": 0.875,
|
|
"rewards/format_reward/std": 0.29546841979026794,
|
|
"rewards/mcq_exact_match_reward/mean": 0.421875,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"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": 1237.0,
|
|
"completions/max_terminated_length": 1237.0,
|
|
"completions/mean_length": 95.015625,
|
|
"completions/mean_terminated_length": 95.015625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.03375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 14.90458869934082,
|
|
"learning_rate": 9.826044551386742e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 2831627.0,
|
|
"reward": 0.7265625,
|
|
"reward_std": 0.38551604747772217,
|
|
"rewards/format_reward/mean": 0.859375,
|
|
"rewards/format_reward/std": 0.28824523091316223,
|
|
"rewards/mcq_exact_match_reward/mean": 0.640625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"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": 0.03125,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 1642.0,
|
|
"completions/mean_length": 135.203125,
|
|
"completions/mean_terminated_length": 73.5,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.035,
|
|
"frac_reward_zero_std": 0.25,
|
|
"grad_norm": 16.59746551513672,
|
|
"learning_rate": 9.803768380684242e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 2915136.0,
|
|
"reward": 0.40312501788139343,
|
|
"reward_std": 0.2227361798286438,
|
|
"rewards/format_reward/mean": 0.90625,
|
|
"rewards/format_reward/std": 0.279951810836792,
|
|
"rewards/mcq_exact_match_reward/mean": 0.3125,
|
|
"rewards/mcq_exact_match_reward/std": 0.467176616191864,
|
|
"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": 1399.0,
|
|
"completions/max_terminated_length": 1399.0,
|
|
"completions/mean_length": 109.875,
|
|
"completions/mean_terminated_length": 109.875,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.03625,
|
|
"frac_reward_zero_std": 0.375,
|
|
"grad_norm": 8.776859283447266,
|
|
"learning_rate": 9.780178907671788e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 2999896.0,
|
|
"reward": 0.15156251192092896,
|
|
"reward_std": 0.1783457100391388,
|
|
"rewards/format_reward/mean": 0.890625,
|
|
"rewards/format_reward/std": 0.24346621334552765,
|
|
"rewards/mcq_exact_match_reward/mean": 0.0625,
|
|
"rewards/mcq_exact_match_reward/std": 0.24397502839565277,
|
|
"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": 380.0,
|
|
"completions/max_terminated_length": 380.0,
|
|
"completions/mean_length": 25.40625,
|
|
"completions/mean_terminated_length": 25.40625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0375,
|
|
"frac_reward_zero_std": 0.375,
|
|
"grad_norm": 13.594465255737305,
|
|
"learning_rate": 9.755282581475767e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 3087370.0,
|
|
"reward": 0.5171875357627869,
|
|
"reward_std": 0.21780578792095184,
|
|
"rewards/format_reward/mean": 0.953125,
|
|
"rewards/format_reward/std": 0.1717960685491562,
|
|
"rewards/mcq_exact_match_reward/mean": 0.421875,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"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": 0.03125,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 449.0,
|
|
"completions/mean_length": 96.09375,
|
|
"completions/mean_terminated_length": 33.129032135009766,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.03875,
|
|
"frac_reward_zero_std": 0.375,
|
|
"grad_norm": 9.822609901428223,
|
|
"learning_rate": 9.729086208503173e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 3177416.0,
|
|
"reward": 0.52734375,
|
|
"reward_std": 0.2801167666912079,
|
|
"rewards/format_reward/mean": 0.8984375,
|
|
"rewards/format_reward/std": 0.2387082874774933,
|
|
"rewards/mcq_exact_match_reward/mean": 0.4375,
|
|
"rewards/mcq_exact_match_reward/std": 0.5,
|
|
"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": 187.0,
|
|
"completions/max_terminated_length": 187.0,
|
|
"completions/mean_length": 17.640625,
|
|
"completions/mean_terminated_length": 17.640625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.04,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 21.135448455810547,
|
|
"learning_rate": 9.701596950580807e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 3255833.0,
|
|
"reward": 0.6664062738418579,
|
|
"reward_std": 0.12476710975170135,
|
|
"rewards/format_reward/mean": 0.8828125,
|
|
"rewards/format_reward/std": 0.21347814798355103,
|
|
"rewards/mcq_exact_match_reward/mean": 0.578125,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"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": 30.0,
|
|
"completions/max_terminated_length": 30.0,
|
|
"completions/mean_length": 12.71875,
|
|
"completions/mean_terminated_length": 12.71875,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.04125,
|
|
"frac_reward_zero_std": 0.375,
|
|
"grad_norm": 18.746240615844727,
|
|
"learning_rate": 9.672822322997304e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 3356175.0,
|
|
"reward": 0.5960937738418579,
|
|
"reward_std": 0.2415097951889038,
|
|
"rewards/format_reward/mean": 0.9609375,
|
|
"rewards/format_reward/std": 0.13524486124515533,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"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": 372.0,
|
|
"completions/max_terminated_length": 372.0,
|
|
"completions/mean_length": 18.125,
|
|
"completions/mean_terminated_length": 18.125,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0425,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 10.819052696228027,
|
|
"learning_rate": 9.642770192448535e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 3463031.0,
|
|
"reward": 0.34453126788139343,
|
|
"reward_std": 0.14846181869506836,
|
|
"rewards/format_reward/mean": 0.9453125,
|
|
"rewards/format_reward/std": 0.15728822350502014,
|
|
"rewards/mcq_exact_match_reward/mean": 0.25,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"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": 510.0,
|
|
"completions/max_terminated_length": 510.0,
|
|
"completions/mean_length": 27.015625,
|
|
"completions/mean_terminated_length": 27.015625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.04375,
|
|
"frac_reward_zero_std": 0.375,
|
|
"grad_norm": 15.864429473876953,
|
|
"learning_rate": 9.611448774886923e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 3537936.0,
|
|
"reward": 0.2835937738418579,
|
|
"reward_std": 0.2292654812335968,
|
|
"rewards/format_reward/mean": 0.9609375,
|
|
"rewards/format_reward/std": 0.16194961965084076,
|
|
"rewards/mcq_exact_match_reward/mean": 0.1875,
|
|
"rewards/mcq_exact_match_reward/std": 0.39339789748191833,
|
|
"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": 13.0,
|
|
"completions/max_terminated_length": 13.0,
|
|
"completions/mean_length": 11.03125,
|
|
"completions/mean_terminated_length": 11.03125,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.045,
|
|
"frac_reward_zero_std": 0.125,
|
|
"grad_norm": 24.202913284301758,
|
|
"learning_rate": 9.578866633275286e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 3640722.0,
|
|
"reward": 0.45078128576278687,
|
|
"reward_std": 0.37580984830856323,
|
|
"rewards/format_reward/mean": 0.9140625,
|
|
"rewards/format_reward/std": 0.19012710452079773,
|
|
"rewards/mcq_exact_match_reward/mean": 0.359375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 11.0625,
|
|
"completions/mean_terminated_length": 11.0625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.04625,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 13.204147338867188,
|
|
"learning_rate": 9.545032675245813e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 3717014.0,
|
|
"reward": 0.5140625238418579,
|
|
"reward_std": 0.15902264416217804,
|
|
"rewards/format_reward/mean": 0.921875,
|
|
"rewards/format_reward/std": 0.18298126757144928,
|
|
"rewards/mcq_exact_match_reward/mean": 0.421875,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"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": 0.015625,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 320.0,
|
|
"completions/mean_length": 48.71875,
|
|
"completions/mean_terminated_length": 16.984127044677734,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.0475,
|
|
"frac_reward_zero_std": 0.25,
|
|
"grad_norm": 12.726435661315918,
|
|
"learning_rate": 9.509956150664795e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 3796628.0,
|
|
"reward": 0.48906251788139343,
|
|
"reward_std": 0.25272947549819946,
|
|
"rewards/format_reward/mean": 0.984375,
|
|
"rewards/format_reward/std": 0.125,
|
|
"rewards/mcq_exact_match_reward/mean": 0.390625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4917473793029785,
|
|
"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": 0.015625,
|
|
"completions/max_length": 2048.0,
|
|
"completions/max_terminated_length": 222.0,
|
|
"completions/mean_length": 53.46875,
|
|
"completions/mean_terminated_length": 21.809524536132812,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.04875,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 21.27250099182129,
|
|
"learning_rate": 9.473646649103817e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 3869618.0,
|
|
"reward": 0.811718761920929,
|
|
"reward_std": 0.21845673024654388,
|
|
"rewards/format_reward/mean": 0.9296875,
|
|
"rewards/format_reward/std": 0.19654127955436707,
|
|
"rewards/mcq_exact_match_reward/mean": 0.71875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4531635046005249,
|
|
"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": 54.0,
|
|
"completions/max_terminated_length": 54.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.05,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 23.1184139251709,
|
|
"learning_rate": 9.436114097218058e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 3952802.0,
|
|
"reward": 0.15703126788139343,
|
|
"reward_std": 0.14846181869506836,
|
|
"rewards/format_reward/mean": 0.9453125,
|
|
"rewards/format_reward/std": 0.15728822350502014,
|
|
"rewards/mcq_exact_match_reward/mean": 0.0625,
|
|
"rewards/mcq_exact_match_reward/std": 0.24397502839565277,
|
|
"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": 312.0,
|
|
"completions/max_terminated_length": 312.0,
|
|
"completions/mean_length": 16.09375,
|
|
"completions/mean_terminated_length": 16.09375,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.05125,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 12.768502235412598,
|
|
"learning_rate": 9.397368756032444e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 4031320.0,
|
|
"reward": 0.6101562976837158,
|
|
"reward_std": 0.1323750764131546,
|
|
"rewards/format_reward/mean": 0.9453125,
|
|
"rewards/format_reward/std": 0.15728822350502014,
|
|
"rewards/mcq_exact_match_reward/mean": 0.515625,
|
|
"rewards/mcq_exact_match_reward/std": 0.5037065148353577,
|
|
"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": 1512.0,
|
|
"completions/max_terminated_length": 1512.0,
|
|
"completions/mean_length": 50.265625,
|
|
"completions/mean_terminated_length": 50.265625,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.0525,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 23.840425491333008,
|
|
"learning_rate": 9.357421218136386e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 4117881.0,
|
|
"reward": 0.5843750238418579,
|
|
"reward_std": 0.12255740165710449,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.484375,
|
|
"rewards/mcq_exact_match_reward/std": 0.5037065148353577,
|
|
"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": 419.0,
|
|
"completions/max_terminated_length": 419.0,
|
|
"completions/mean_length": 24.75,
|
|
"completions/mean_terminated_length": 24.75,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.05375,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 9.856094360351562,
|
|
"learning_rate": 9.316282404787869e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 4199609.0,
|
|
"reward": 0.3890625238418579,
|
|
"reward_std": 0.17584913969039917,
|
|
"rewards/format_reward/mean": 0.921875,
|
|
"rewards/format_reward/std": 0.20351573824882507,
|
|
"rewards/mcq_exact_match_reward/mean": 0.296875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4604927599430084,
|
|
"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": 13.0,
|
|
"completions/max_terminated_length": 13.0,
|
|
"completions/mean_length": 12.015625,
|
|
"completions/mean_terminated_length": 12.015625,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.055,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 13.109899520874023,
|
|
"learning_rate": 9.273963562927694e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 4271338.0,
|
|
"reward": 0.3656250238418579,
|
|
"reward_std": 0.15992169082164764,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.265625,
|
|
"rewards/mcq_exact_match_reward/std": 0.44515693187713623,
|
|
"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": 21.0,
|
|
"completions/max_terminated_length": 21.0,
|
|
"completions/mean_length": 11.515625,
|
|
"completions/mean_terminated_length": 11.515625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.05625,
|
|
"frac_reward_zero_std": 0.375,
|
|
"grad_norm": 26.21106719970703,
|
|
"learning_rate": 9.230476262104676e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 4351443.0,
|
|
"reward": 0.5679687261581421,
|
|
"reward_std": 0.19368097186088562,
|
|
"rewards/format_reward/mean": 0.8359375,
|
|
"rewards/format_reward/std": 0.30930912494659424,
|
|
"rewards/mcq_exact_match_reward/mean": 0.484375,
|
|
"rewards/mcq_exact_match_reward/std": 0.5037065148353577,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 11.25,
|
|
"completions/mean_terminated_length": 11.25,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.0575,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 21.321683883666992,
|
|
"learning_rate": 9.185832391312642e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 4459083.0,
|
|
"reward": 0.546875,
|
|
"reward_std": 0.24039676785469055,
|
|
"rewards/format_reward/mean": 0.9375,
|
|
"rewards/format_reward/std": 0.1666666716337204,
|
|
"rewards/mcq_exact_match_reward/mean": 0.453125,
|
|
"rewards/mcq_exact_match_reward/std": 0.501733124256134,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 11.90625,
|
|
"completions/mean_terminated_length": 11.90625,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.05875,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 25.36126708984375,
|
|
"learning_rate": 9.1400441557401e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 4542701.0,
|
|
"reward": 0.2554687559604645,
|
|
"reward_std": 0.06007346510887146,
|
|
"rewards/format_reward/mean": 0.9921875,
|
|
"rewards/format_reward/std": 0.0625,
|
|
"rewards/mcq_exact_match_reward/mean": 0.15625,
|
|
"rewards/mcq_exact_match_reward/std": 0.36596253514289856,
|
|
"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": 33.0,
|
|
"completions/max_terminated_length": 33.0,
|
|
"completions/mean_length": 14.078125,
|
|
"completions/mean_terminated_length": 14.078125,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"epoch": 0.06,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 26.997413635253906,
|
|
"learning_rate": 9.093124073433462e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 4657762.0,
|
|
"reward": 0.39531251788139343,
|
|
"reward_std": 0.15809684991836548,
|
|
"rewards/format_reward/mean": 0.828125,
|
|
"rewards/format_reward/std": 0.3475395441055298,
|
|
"rewards/mcq_exact_match_reward/mean": 0.3125,
|
|
"rewards/mcq_exact_match_reward/std": 0.467176616191864,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.06125,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 13.210586547851562,
|
|
"learning_rate": 9.045084971874737e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 4725234.0,
|
|
"reward": 0.6000000238418579,
|
|
"reward_std": 0.17570313811302185,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"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": 13.0,
|
|
"completions/max_terminated_length": 13.0,
|
|
"completions/mean_length": 12.03125,
|
|
"completions/mean_terminated_length": 12.03125,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.0625,
|
|
"frac_reward_zero_std": 0.375,
|
|
"grad_norm": 14.75271987915039,
|
|
"learning_rate": 8.995939984474623e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 4791620.0,
|
|
"reward": 0.4281250238418579,
|
|
"reward_std": 0.2414703369140625,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.328125,
|
|
"rewards/mcq_exact_match_reward/std": 0.4732423722743988,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.06375,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 16.48665428161621,
|
|
"learning_rate": 8.945702546981968e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 4872452.0,
|
|
"reward": 0.6000000238418579,
|
|
"reward_std": 0.0883883461356163,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"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": 16.0,
|
|
"completions/max_terminated_length": 16.0,
|
|
"completions/mean_length": 12.0625,
|
|
"completions/mean_terminated_length": 12.0625,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.065,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 15.117328643798828,
|
|
"learning_rate": 8.894386393810562e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 4941224.0,
|
|
"reward": 0.2718750238418579,
|
|
"reward_std": 0.21778544783592224,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.171875,
|
|
"rewards/mcq_exact_match_reward/std": 0.38025420904159546,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.06625,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 14.067106246948242,
|
|
"learning_rate": 8.842005554284295e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 5024888.0,
|
|
"reward": 0.5843750238418579,
|
|
"reward_std": 0.10205793380737305,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.484375,
|
|
"rewards/mcq_exact_match_reward/std": 0.5037065148353577,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.0675,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 7.727897644042969,
|
|
"learning_rate": 8.788574348801674e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 5133184.0,
|
|
"reward": 0.6781250238418579,
|
|
"reward_std": 0.0646936446428299,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.578125,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"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": 13.0,
|
|
"completions/max_terminated_length": 13.0,
|
|
"completions/mean_length": 12.015625,
|
|
"completions/mean_terminated_length": 12.015625,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.06875,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 5.476644515991211,
|
|
"learning_rate": 8.734107384920769e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 5193313.0,
|
|
"reward": 0.4906250238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.390625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4917473793029785,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.07,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 17.98200225830078,
|
|
"learning_rate": 8.678619553365658e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 5265001.0,
|
|
"reward": 0.4593750238418579,
|
|
"reward_std": 0.1804211586713791,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.359375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.07125,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 7.063903331756592,
|
|
"learning_rate": 8.622126023955445e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 5342409.0,
|
|
"reward": 0.9593750238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.859375,
|
|
"rewards/mcq_exact_match_reward/std": 0.3503824472427368,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.0725,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 13.756085395812988,
|
|
"learning_rate": 8.564642241456986e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 5442433.0,
|
|
"reward": 0.5531250238418579,
|
|
"reward_std": 0.22673700749874115,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.453125,
|
|
"rewards/mcq_exact_match_reward/std": 0.501733124256134,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.07375,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 10.853642463684082,
|
|
"learning_rate": 8.506183921362442e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 5526609.0,
|
|
"reward": 0.7250000238418579,
|
|
"reward_std": 0.1293872892856598,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.075,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 8.446767045592829e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 5633697.0,
|
|
"reward": 0.8500000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"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": 13.0,
|
|
"completions/max_terminated_length": 13.0,
|
|
"completions/mean_length": 12.015625,
|
|
"completions/mean_terminated_length": 12.015625,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.07625,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 28.784507751464844,
|
|
"learning_rate": 8.386407858128706e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 5703498.0,
|
|
"reward": 0.5375000238418579,
|
|
"reward_std": 0.10888782143592834,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.4375,
|
|
"rewards/mcq_exact_match_reward/std": 0.5,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.0775,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 9.740281105041504,
|
|
"learning_rate": 8.325122860569241e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 5761730.0,
|
|
"reward": 0.3343750238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.234375,
|
|
"rewards/mcq_exact_match_reward/std": 0.42695629596710205,
|
|
"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": 13.0,
|
|
"completions/max_terminated_length": 13.0,
|
|
"completions/mean_length": 12.015625,
|
|
"completions/mean_terminated_length": 12.015625,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.07875,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 18.035011291503906,
|
|
"learning_rate": 8.262928807620843e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 5838435.0,
|
|
"reward": 0.7562500238418579,
|
|
"reward_std": 0.1293872892856598,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.65625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4787135720252991,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.08,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 7.4544501304626465,
|
|
"learning_rate": 8.199842702516582e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 5908339.0,
|
|
"reward": 0.6156250238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.515625,
|
|
"rewards/mcq_exact_match_reward/std": 0.5037065148353577,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.08125,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 21.87096405029297,
|
|
"learning_rate": 8.135881792367685e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6001347.0,
|
|
"reward": 0.5218750238418579,
|
|
"reward_std": 0.23144522309303284,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.421875,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"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": 13.0,
|
|
"completions/max_terminated_length": 13.0,
|
|
"completions/mean_length": 12.015625,
|
|
"completions/mean_terminated_length": 12.015625,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.0825,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 5.694606304168701,
|
|
"learning_rate": 8.071063563448339e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6077524.0,
|
|
"reward": 0.6312500238418579,
|
|
"reward_std": 0.0578637570142746,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.53125,
|
|
"rewards/mcq_exact_match_reward/std": 0.5029674172401428,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.08375,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 10.359199523925781,
|
|
"learning_rate": 8.005405736415125e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6166844.0,
|
|
"reward": 0.5218750238418579,
|
|
"reward_std": 0.10205793380737305,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.421875,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.085,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 7.938926261462365e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6251940.0,
|
|
"reward": 0.8500000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"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": 13.0,
|
|
"completions/max_terminated_length": 13.0,
|
|
"completions/mean_length": 12.015625,
|
|
"completions/mean_terminated_length": 12.015625,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.08625,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 15.241501808166504,
|
|
"learning_rate": 7.871643313414718e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6330485.0,
|
|
"reward": 0.5375000238418579,
|
|
"reward_std": 0.1462521106004715,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.4375,
|
|
"rewards/mcq_exact_match_reward/std": 0.5,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.0875,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 7.803575286758363e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6404309.0,
|
|
"reward": 0.7250000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.08875,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 16.783206939697266,
|
|
"learning_rate": 7.734740790612136e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6484293.0,
|
|
"reward": 0.5062500238418579,
|
|
"reward_std": 0.1293872892856598,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.40625,
|
|
"rewards/mcq_exact_match_reward/std": 0.49501484632492065,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.09,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 7.665158643639969e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6556669.0,
|
|
"reward": 0.8500000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.09125,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 30.312788009643555,
|
|
"learning_rate": 7.594847868906076e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6630717.0,
|
|
"reward": 0.7406250238418579,
|
|
"reward_std": 0.2109457552433014,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.640625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.0925,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 11.27795696258545,
|
|
"learning_rate": 7.523827688674219e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 6709965.0,
|
|
"reward": 0.5062500238418579,
|
|
"reward_std": 0.10888782143592834,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.40625,
|
|
"rewards/mcq_exact_match_reward/std": 0.49501484632492065,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.09375,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 6.554471015930176,
|
|
"learning_rate": 7.452117519152541e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 6841037.0,
|
|
"reward": 0.5531250238418579,
|
|
"reward_std": 0.11100947856903076,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.453125,
|
|
"rewards/mcq_exact_match_reward/std": 0.501733124256134,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.095,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 9.212408065795898,
|
|
"learning_rate": 7.379736965185368e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6919005.0,
|
|
"reward": 0.6937500238418579,
|
|
"reward_std": 0.0578637570142746,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.59375,
|
|
"rewards/mcq_exact_match_reward/std": 0.49501484632492065,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.09625,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 37.92858123779297,
|
|
"learning_rate": 7.306705814893439e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 6996109.0,
|
|
"reward": 0.5062500238418579,
|
|
"reward_std": 0.16675157845020294,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.40625,
|
|
"rewards/mcq_exact_match_reward/std": 0.49501484632492065,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.0975,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 17.949735641479492,
|
|
"learning_rate": 7.233044034264033e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 7091661.0,
|
|
"reward": 0.6937500238418579,
|
|
"reward_std": 0.1872510462999344,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.59375,
|
|
"rewards/mcq_exact_match_reward/std": 0.49501484632492065,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.09875,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 7.158771761692464e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 7178109.0,
|
|
"reward": 0.8500000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 20.589990615844727,
|
|
"learning_rate": 7.083909302476452e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 7252685.0,
|
|
"reward": 0.8031250238418579,
|
|
"reward_std": 0.16887323558330536,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.703125,
|
|
"rewards/mcq_exact_match_reward/std": 0.4604927599430084,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.10125,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 7.008477123264847e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 7330925.0,
|
|
"reward": 0.8500000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1025,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 34.49396514892578,
|
|
"learning_rate": 6.932495846462261e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 7403133.0,
|
|
"reward": 0.6781250238418579,
|
|
"reward_std": 0.12255740165710449,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.578125,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"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": 1735.0,
|
|
"completions/mean_length": 173.828125,
|
|
"completions/mean_terminated_length": 144.07937622070312,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.10375,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 20.851499557495117,
|
|
"learning_rate": 6.855986244591103e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 7509546.0,
|
|
"reward": 0.5359375476837158,
|
|
"reward_std": 0.07116301357746124,
|
|
"rewards/format_reward/mean": 0.828125,
|
|
"rewards/format_reward/std": 0.38025420904159546,
|
|
"rewards/mcq_exact_match_reward/mean": 0.453125,
|
|
"rewards/mcq_exact_match_reward/std": 0.501733124256134,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.105,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 8.781059265136719,
|
|
"learning_rate": 6.778969234612583e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 7580522.0,
|
|
"reward": 0.8343750238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.734375,
|
|
"rewards/mcq_exact_match_reward/std": 0.44515693187713623,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.10625,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 7.650911331176758,
|
|
"learning_rate": 6.701465872208216e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 7690706.0,
|
|
"reward": 0.6000000238418579,
|
|
"reward_std": 0.0883883461356163,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1075,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 12.82839298248291,
|
|
"learning_rate": 6.623497346023417e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 7771410.0,
|
|
"reward": 0.6140625476837158,
|
|
"reward_std": 0.04861358925700188,
|
|
"rewards/format_reward/mean": 0.984375,
|
|
"rewards/format_reward/std": 0.125,
|
|
"rewards/mcq_exact_match_reward/mean": 0.515625,
|
|
"rewards/mcq_exact_match_reward/std": 0.5037065148353577,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.10875,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 26.970367431640625,
|
|
"learning_rate": 6.545084971874736e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 7840058.0,
|
|
"reward": 0.5531250238418579,
|
|
"reward_std": 0.19939783215522766,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.453125,
|
|
"rewards/mcq_exact_match_reward/std": 0.501733124256134,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.11,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 21.771242141723633,
|
|
"learning_rate": 6.466250186922324e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 7942954.0,
|
|
"reward": 1.006250023841858,
|
|
"reward_std": 0.1293872892856598,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.90625,
|
|
"rewards/mcq_exact_match_reward/std": 0.29378482699394226,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.11125,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 15.462240219116211,
|
|
"learning_rate": 6.387014543809223e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 8039274.0,
|
|
"reward": 0.7406250238418579,
|
|
"reward_std": 0.10205793380737305,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.640625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1125,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 9.880819320678711,
|
|
"learning_rate": 6.307399704769098e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 8116786.0,
|
|
"reward": 0.6156250238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.515625,
|
|
"rewards/mcq_exact_match_reward/std": 0.5037065148353577,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.11375,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 15.82265853881836,
|
|
"learning_rate": 6.227427435703995e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 8196418.0,
|
|
"reward": 0.4593750238418579,
|
|
"reward_std": 0.10205793380737305,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.359375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"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": 0.0,
|
|
"completions/max_length": 17.0,
|
|
"completions/max_terminated_length": 17.0,
|
|
"completions/mean_length": 12.078125,
|
|
"completions/mean_terminated_length": 12.078125,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.115,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 11.207847595214844,
|
|
"learning_rate": 6.147119600233758e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 8268607.0,
|
|
"reward": 0.9421875476837158,
|
|
"reward_std": 0.09280776977539062,
|
|
"rewards/format_reward/mean": 0.984375,
|
|
"rewards/format_reward/std": 0.125,
|
|
"rewards/mcq_exact_match_reward/mean": 0.84375,
|
|
"rewards/mcq_exact_match_reward/std": 0.36596253514289856,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.11625,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 6.866385459899902,
|
|
"learning_rate": 6.066498153718734e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 8372607.0,
|
|
"reward": 0.7875000238418579,
|
|
"reward_std": 0.06681530922651291,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.6875,
|
|
"rewards/mcq_exact_match_reward/std": 0.467176616191864,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1175,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 15.375765800476074,
|
|
"learning_rate": 5.985585137257401e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 8444655.0,
|
|
"reward": 0.8187500238418579,
|
|
"reward_std": 0.0578637570142746,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.71875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4531635046005249,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.11875,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 15.74787425994873,
|
|
"learning_rate": 5.90440267166055e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 8522247.0,
|
|
"reward": 0.6781250238418579,
|
|
"reward_std": 0.0646936446428299,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.578125,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.12,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 19.906932830810547,
|
|
"learning_rate": 5.82297295140367e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 8618319.0,
|
|
"reward": 0.7718750238418579,
|
|
"reward_std": 0.11100947856903076,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.671875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4732423722743988,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.12125,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 5.741318238559209e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 8705847.0,
|
|
"reward": 0.7250000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1225,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 4.607022762298584,
|
|
"learning_rate": 5.659460856710345e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 8818503.0,
|
|
"reward": 0.4593750238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.359375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.12375,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 5.152970790863037,
|
|
"learning_rate": 5.577423184847931e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 8897183.0,
|
|
"reward": 0.8343750238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.734375,
|
|
"rewards/mcq_exact_match_reward/std": 0.44515693187713623,
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.125,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 5.495227651252315e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 8980063.0,
|
|
"reward": 0.7250000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 100
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.12625,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 9.869670867919922,
|
|
"learning_rate": 5.412896727361662e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 9056007.0,
|
|
"reward": 0.8968750238418579,
|
|
"reward_std": 0.0646936446428299,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.796875,
|
|
"rewards/mcq_exact_match_reward/std": 0.40550529956817627,
|
|
"step": 101
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1275,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 9.079934120178223,
|
|
"learning_rate": 5.330452921628497e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 9127807.0,
|
|
"reward": 0.8656250238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.765625,
|
|
"rewards/mcq_exact_match_reward/std": 0.42695629596710205,
|
|
"step": 102
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.12875,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 5.247918773366111e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 9209351.0,
|
|
"reward": 0.8500000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"step": 103
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.13,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 7.6732635498046875,
|
|
"learning_rate": 5.165316846586541e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 9302183.0,
|
|
"reward": 0.5218750238418579,
|
|
"reward_std": 0.10205793380737305,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.421875,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"step": 104
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.13125,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 10.727198600769043,
|
|
"learning_rate": 5.082669723831793e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 9374207.0,
|
|
"reward": 0.3656250238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.265625,
|
|
"rewards/mcq_exact_match_reward/std": 0.44515693187713623,
|
|
"step": 105
|
|
},
|
|
{
|
|
"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": 18.0,
|
|
"completions/max_terminated_length": 18.0,
|
|
"completions/mean_length": 12.09375,
|
|
"completions/mean_terminated_length": 12.09375,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1325,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 11.330345153808594,
|
|
"learning_rate": 5e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 9494293.0,
|
|
"reward": 0.6453125476837158,
|
|
"reward_std": 0.10647734999656677,
|
|
"rewards/format_reward/mean": 0.984375,
|
|
"rewards/format_reward/std": 0.125,
|
|
"rewards/mcq_exact_match_reward/mean": 0.546875,
|
|
"rewards/mcq_exact_match_reward/std": 0.501733124256134,
|
|
"step": 106
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.13375,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 12.694079399108887,
|
|
"learning_rate": 4.917330276168208e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 9590661.0,
|
|
"reward": 0.7093750238418579,
|
|
"reward_std": 0.12255740165710449,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.609375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4917473793029785,
|
|
"step": 107
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.135,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 39.67001724243164,
|
|
"learning_rate": 4.834683153413459e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 9689293.0,
|
|
"reward": 0.5218750238418579,
|
|
"reward_std": 0.23568856716156006,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.421875,
|
|
"rewards/mcq_exact_match_reward/std": 0.49776285886764526,
|
|
"step": 108
|
|
},
|
|
{
|
|
"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": 948.0,
|
|
"completions/max_terminated_length": 948.0,
|
|
"completions/mean_length": 82.1875,
|
|
"completions/mean_terminated_length": 82.1875,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.13625,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 29.97638702392578,
|
|
"learning_rate": 4.752081226633888e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 9806377.0,
|
|
"reward": 0.40234375,
|
|
"reward_std": 0.15686902403831482,
|
|
"rewards/format_reward/mean": 0.8984375,
|
|
"rewards/format_reward/std": 0.28423789143562317,
|
|
"rewards/mcq_exact_match_reward/mean": 0.3125,
|
|
"rewards/mcq_exact_match_reward/std": 0.467176616191864,
|
|
"step": 109
|
|
},
|
|
{
|
|
"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": 13.0,
|
|
"completions/max_terminated_length": 13.0,
|
|
"completions/mean_length": 12.015625,
|
|
"completions/mean_terminated_length": 12.015625,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1375,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 23.60181427001953,
|
|
"learning_rate": 4.669547078371503e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 9905698.0,
|
|
"reward": 0.8343750238418579,
|
|
"reward_std": 0.1315089464187622,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.734375,
|
|
"rewards/mcq_exact_match_reward/std": 0.44515693187713623,
|
|
"step": 110
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.13875,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 4.5871032726383385e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 9984194.0,
|
|
"reward": 0.4750000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.375,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 111
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.14,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 4.5047723487476864e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 10064970.0,
|
|
"reward": 0.8500000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"step": 112
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.14125,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 4.4225768151520694e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 10164954.0,
|
|
"reward": 0.7250000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 113
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1425,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 9.757353782653809,
|
|
"learning_rate": 4.340539143289655e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 10220466.0,
|
|
"reward": 0.5375000238418579,
|
|
"reward_std": 0.06681530922651291,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.4375,
|
|
"rewards/mcq_exact_match_reward/std": 0.5,
|
|
"step": 114
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.14375,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 9.696577072143555,
|
|
"learning_rate": 4.258681761440789e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 10299706.0,
|
|
"reward": 0.9593750238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.859375,
|
|
"rewards/mcq_exact_match_reward/std": 0.3503824472427368,
|
|
"step": 115
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.145,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 6.446499347686768,
|
|
"learning_rate": 4.1770270485963294e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 10391202.0,
|
|
"reward": 1.084375023841858,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.984375,
|
|
"rewards/mcq_exact_match_reward/std": 0.125,
|
|
"step": 116
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.14625,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 4.095597328339452e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 10471786.0,
|
|
"reward": 0.7250000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 117
|
|
},
|
|
{
|
|
"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": 13.0,
|
|
"completions/max_terminated_length": 13.0,
|
|
"completions/mean_length": 12.015625,
|
|
"completions/mean_terminated_length": 12.015625,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1475,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 4.0144148627425986e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 10565315.0,
|
|
"reward": 0.8500000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"step": 118
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.14875,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 4.166144371032715,
|
|
"learning_rate": 3.9335018462812664e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 10662483.0,
|
|
"reward": 0.7406250238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.640625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"step": 119
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.15,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 18.16716957092285,
|
|
"learning_rate": 3.8528803997662423e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 10740347.0,
|
|
"reward": 0.7718750238418579,
|
|
"reward_std": 0.11100947856903076,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.671875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4732423722743988,
|
|
"step": 120
|
|
},
|
|
{
|
|
"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": 17.0,
|
|
"completions/max_terminated_length": 17.0,
|
|
"completions/mean_length": 12.234375,
|
|
"completions/mean_terminated_length": 12.234375,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.15125,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 17.088464736938477,
|
|
"learning_rate": 3.772572564296004e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 10837266.0,
|
|
"reward": 0.8921875357627869,
|
|
"reward_std": 0.10436524450778961,
|
|
"rewards/format_reward/mean": 0.953125,
|
|
"rewards/format_reward/std": 0.21304203569889069,
|
|
"rewards/mcq_exact_match_reward/mean": 0.796875,
|
|
"rewards/mcq_exact_match_reward/std": 0.40550529956817627,
|
|
"step": 121
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1525,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 11.872878074645996,
|
|
"learning_rate": 3.692600295230901e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 10922410.0,
|
|
"reward": 0.8343750238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.734375,
|
|
"rewards/mcq_exact_match_reward/std": 0.44515693187713623,
|
|
"step": 122
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.15375,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 10.134330749511719,
|
|
"learning_rate": 3.612985456190778e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 11027762.0,
|
|
"reward": 0.4125000238418579,
|
|
"reward_std": 0.06681530922651291,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.3125,
|
|
"rewards/mcq_exact_match_reward/std": 0.467176616191864,
|
|
"step": 123
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.155,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 3.533749813077677e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 11108994.0,
|
|
"reward": 0.8500000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"step": 124
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.15625,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 19.913314819335938,
|
|
"learning_rate": 3.454915028125263e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 11188714.0,
|
|
"reward": 0.7718750238418579,
|
|
"reward_std": 0.1530819833278656,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.671875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4732423722743988,
|
|
"step": 125
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1575,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 3.3765026539765827e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 11261826.0,
|
|
"reward": 0.6000000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"step": 126
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.15875,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 16.5478572845459,
|
|
"learning_rate": 3.2985341277917846e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 11359498.0,
|
|
"reward": 0.7406250238418579,
|
|
"reward_std": 0.12255740165710449,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.640625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"step": 127
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.16,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 21.59103775024414,
|
|
"learning_rate": 3.221030765387417e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 11455802.0,
|
|
"reward": 0.8343750238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.734375,
|
|
"rewards/mcq_exact_match_reward/std": 0.44515693187713623,
|
|
"step": 128
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.16125,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 28.282245635986328,
|
|
"learning_rate": 3.1440137554088953e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 11528338.0,
|
|
"reward": 0.7718750238418579,
|
|
"reward_std": 0.11100947856903076,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.671875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4732423722743988,
|
|
"step": 129
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1625,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 3.06750415353774e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 11609754.0,
|
|
"reward": 0.8500000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"step": 130
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.16375,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 2.9915228767351535e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 11706114.0,
|
|
"reward": 0.6000000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"step": 131
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.165,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 2.916090697523549e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 11790474.0,
|
|
"reward": 0.7250000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 132
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.16625,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 18.859106063842773,
|
|
"learning_rate": 2.841228238307536e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 11873762.0,
|
|
"reward": 0.8187500238418579,
|
|
"reward_std": 0.10888782143592834,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.71875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4531635046005249,
|
|
"step": 133
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1675,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 2.7669559657359673e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 11950250.0,
|
|
"reward": 0.8500000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"step": 134
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.16875,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 18.018556594848633,
|
|
"learning_rate": 2.6932941851065615e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 12048482.0,
|
|
"reward": 0.6625000238418579,
|
|
"reward_std": 0.10888782143592834,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5625,
|
|
"rewards/mcq_exact_match_reward/std": 0.5,
|
|
"step": 135
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.17,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 17.110048294067383,
|
|
"learning_rate": 2.620263034814632e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 12124546.0,
|
|
"reward": 0.4437500238418579,
|
|
"reward_std": 0.10888782143592834,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.34375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4787135720252991,
|
|
"step": 136
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.17125,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 2.547882480847461e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 12215346.0,
|
|
"reward": 0.6000000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"step": 137
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1725,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 2.476172311325783e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 12298682.0,
|
|
"reward": 0.6000000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"step": 138
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.17375,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 6.204960346221924,
|
|
"learning_rate": 2.4051521310939254e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 12381378.0,
|
|
"reward": 0.6156250238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.515625,
|
|
"rewards/mcq_exact_match_reward/std": 0.5037065148353577,
|
|
"step": 139
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.175,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 29.57842445373535,
|
|
"learning_rate": 2.3348413563600323e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 12453050.0,
|
|
"reward": 0.8031250238418579,
|
|
"reward_std": 0.16887323558330536,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.703125,
|
|
"rewards/mcq_exact_match_reward/std": 0.4604927599430084,
|
|
"step": 140
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.17625,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 11.079608917236328,
|
|
"learning_rate": 2.2652592093878665e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 12550858.0,
|
|
"reward": 0.6156250238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.515625,
|
|
"rewards/mcq_exact_match_reward/std": 0.5037065148353577,
|
|
"step": 141
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1775,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 41.790828704833984,
|
|
"learning_rate": 2.1964247132416368e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 12639426.0,
|
|
"reward": 0.9906250238418579,
|
|
"reward_std": 0.12255740165710449,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.890625,
|
|
"rewards/mcq_exact_match_reward/std": 0.3145764470100403,
|
|
"step": 142
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.17875,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 2.128356686585282e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 12710242.0,
|
|
"reward": 0.8500000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"step": 143
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.18,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 8.774580001831055,
|
|
"learning_rate": 2.0610737385376348e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 12783826.0,
|
|
"reward": 0.8656250238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.765625,
|
|
"rewards/mcq_exact_match_reward/std": 0.42695629596710205,
|
|
"step": 144
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.18125,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 31.257766723632812,
|
|
"learning_rate": 1.9945942635848745e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 12872354.0,
|
|
"reward": 0.5375000238418579,
|
|
"reward_std": 0.1157275140285492,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.4375,
|
|
"rewards/mcq_exact_match_reward/std": 0.5,
|
|
"step": 145
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1825,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 6.080259799957275,
|
|
"learning_rate": 1.9289364365516607e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 12948058.0,
|
|
"reward": 0.6156250238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.515625,
|
|
"rewards/mcq_exact_match_reward/std": 0.5037065148353577,
|
|
"step": 146
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.18375,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.8641182076323148e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 13037682.0,
|
|
"reward": 0.6000000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"step": 147
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.185,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.8001572974834168e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 13112626.0,
|
|
"reward": 0.9750000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.875,
|
|
"rewards/mcq_exact_match_reward/std": 0.3333333432674408,
|
|
"step": 148
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.18625,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 6.417766571044922,
|
|
"learning_rate": 1.7370711923791564e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 13196250.0,
|
|
"reward": 0.8343750238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.734375,
|
|
"rewards/mcq_exact_match_reward/std": 0.44515693187713623,
|
|
"step": 149
|
|
},
|
|
{
|
|
"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": 13.0,
|
|
"completions/max_terminated_length": 13.0,
|
|
"completions/mean_length": 12.015625,
|
|
"completions/mean_terminated_length": 12.015625,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1875,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 7.4770307540893555,
|
|
"learning_rate": 1.674877139430758e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 13281091.0,
|
|
"reward": 1.068750023841858,
|
|
"reward_std": 0.0578637570142746,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.96875,
|
|
"rewards/mcq_exact_match_reward/std": 0.17536810040473938,
|
|
"step": 150
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.18875,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 18.69822120666504,
|
|
"learning_rate": 1.6135921418712955e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 13352979.0,
|
|
"reward": 0.7093750238418579,
|
|
"reward_std": 0.12255740165710449,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.609375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4917473793029785,
|
|
"step": 151
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.19,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.553232954407171e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 13427595.0,
|
|
"reward": 0.9750000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.875,
|
|
"rewards/mcq_exact_match_reward/std": 0.3333333432674408,
|
|
"step": 152
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.19125,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 22.641624450683594,
|
|
"learning_rate": 1.493816078637557e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 13495283.0,
|
|
"reward": 0.7875000238418579,
|
|
"reward_std": 0.06681530922651291,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.6875,
|
|
"rewards/mcq_exact_match_reward/std": 0.467176616191864,
|
|
"step": 153
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1925,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 28.363882064819336,
|
|
"learning_rate": 1.435357758543015e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 13558827.0,
|
|
"reward": 0.8343750238418579,
|
|
"reward_std": 0.1315089464187622,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.734375,
|
|
"rewards/mcq_exact_match_reward/std": 0.44515693187713623,
|
|
"step": 154
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.19375,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.3778739760445552e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 13672043.0,
|
|
"reward": 0.6000000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"step": 155
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.195,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 20.779916763305664,
|
|
"learning_rate": 1.321380446634342e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 13759379.0,
|
|
"reward": 0.9906250238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.890625,
|
|
"rewards/mcq_exact_match_reward/std": 0.3145764470100403,
|
|
"step": 156
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.19625,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.2658926150792322e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 13870963.0,
|
|
"reward": 0.6000000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"step": 157
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.1975,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.2114256511983274e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 13945003.0,
|
|
"reward": 0.7250000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 158
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.19875,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.1579944457157059e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 14008835.0,
|
|
"reward": 0.6000000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"step": 159
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.2,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.1056136061894384e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 14119467.0,
|
|
"reward": 0.6000000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"step": 160
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.20125,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 9.324119567871094,
|
|
"learning_rate": 1.0542974530180327e-07,
|
|
"loss": 0.0,
|
|
"num_tokens": 14189547.0,
|
|
"reward": 0.8812500238418579,
|
|
"reward_std": 0.0578637570142746,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.78125,
|
|
"rewards/mcq_exact_match_reward/std": 0.4166666865348816,
|
|
"step": 161
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.2025,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 21.598480224609375,
|
|
"learning_rate": 1.0040600155253764e-07,
|
|
"loss": -0.0,
|
|
"num_tokens": 14260019.0,
|
|
"reward": 0.9437500238418579,
|
|
"reward_std": 0.0883883461356163,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.84375,
|
|
"rewards/mcq_exact_match_reward/std": 0.36596253514289856,
|
|
"step": 162
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.20375,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 9.549150281252632e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 14322027.0,
|
|
"reward": 0.7250000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 163
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.205,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 3.2285122871398926,
|
|
"learning_rate": 9.068759265665382e-08,
|
|
"loss": -0.0,
|
|
"num_tokens": 14447307.0,
|
|
"reward": 0.7406250238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.640625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"step": 164
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.20625,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 16.710023880004883,
|
|
"learning_rate": 8.599558442598998e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 14545395.0,
|
|
"reward": 0.5062500238418579,
|
|
"reward_std": 0.0578637570142746,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.40625,
|
|
"rewards/mcq_exact_match_reward/std": 0.49501484632492065,
|
|
"step": 165
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.2075,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 7.041760444641113,
|
|
"learning_rate": 8.141676086873573e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 14627403.0,
|
|
"reward": 0.7093750238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.609375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4917473793029785,
|
|
"step": 166
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.20875,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 9.780082702636719,
|
|
"learning_rate": 7.695237378953224e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 14721723.0,
|
|
"reward": 0.3343750238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.234375,
|
|
"rewards/mcq_exact_match_reward/std": 0.42695629596710205,
|
|
"step": 167
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.21,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 48.45919418334961,
|
|
"learning_rate": 7.260364370723043e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 14791971.0,
|
|
"reward": 0.4750000238418579,
|
|
"reward_std": 0.1293872892856598,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.375,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 168
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.21125,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 14.335803031921387,
|
|
"learning_rate": 6.837175952121304e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 14881179.0,
|
|
"reward": 0.8187500238418579,
|
|
"reward_std": 0.0578637570142746,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.71875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4531635046005249,
|
|
"step": 169
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.2125,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 6.42578781863613e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 14968803.0,
|
|
"reward": 0.6000000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"step": 170
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.21375,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 11.465300559997559,
|
|
"learning_rate": 6.026312439675551e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 15058523.0,
|
|
"reward": 0.8656250238418579,
|
|
"reward_std": 0.10205793380737305,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.765625,
|
|
"rewards/mcq_exact_match_reward/std": 0.42695629596710205,
|
|
"step": 171
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.215,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 26.202821731567383,
|
|
"learning_rate": 5.638859027819409e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 15151699.0,
|
|
"reward": 0.8343750238418579,
|
|
"reward_std": 0.12255740165710449,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.734375,
|
|
"rewards/mcq_exact_match_reward/std": 0.44515693187713623,
|
|
"step": 172
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.21625,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 17.447301864624023,
|
|
"learning_rate": 5.263533508961826e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 15227115.0,
|
|
"reward": 0.6937500238418579,
|
|
"reward_std": 0.0578637570142746,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.59375,
|
|
"rewards/mcq_exact_match_reward/std": 0.49501484632492065,
|
|
"step": 173
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.2175,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 4.9004384933520547e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 15294499.0,
|
|
"reward": 0.8500000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.75,
|
|
"rewards/mcq_exact_match_reward/std": 0.4364357888698578,
|
|
"step": 174
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.21875,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 10.398773193359375,
|
|
"learning_rate": 4.549673247541874e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 15368139.0,
|
|
"reward": 0.8343750238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.734375,
|
|
"rewards/mcq_exact_match_reward/std": 0.44515693187713623,
|
|
"step": 175
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.22,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 17.357410430908203,
|
|
"learning_rate": 4.2113336672471245e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 15455195.0,
|
|
"reward": 0.3968750238418579,
|
|
"reward_std": 0.10205793380737305,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.296875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4604927599430084,
|
|
"step": 176
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.22125,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 3.8855122511307626e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 15535539.0,
|
|
"reward": 0.7250000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.625,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 177
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.2225,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 25.104820251464844,
|
|
"learning_rate": 3.572298075514652e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 15615915.0,
|
|
"reward": 0.6468750238418579,
|
|
"reward_std": 0.0646936446428299,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.546875,
|
|
"rewards/mcq_exact_match_reward/std": 0.501733124256134,
|
|
"step": 178
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.22375,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 12.439480781555176,
|
|
"learning_rate": 3.271776770026963e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 15708275.0,
|
|
"reward": 0.8031250238418579,
|
|
"reward_std": 0.0646936446428299,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.703125,
|
|
"rewards/mcq_exact_match_reward/std": 0.4604927599430084,
|
|
"step": 179
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.225,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 13.589136123657227,
|
|
"learning_rate": 2.9840304941919416e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 15775899.0,
|
|
"reward": 0.7406250238418579,
|
|
"reward_std": 0.04419417306780815,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.640625,
|
|
"rewards/mcq_exact_match_reward/std": 0.4836103618144989,
|
|
"step": 180
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.22625,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 2.7091379149682682e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 15862011.0,
|
|
"reward": 0.4750000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.375,
|
|
"rewards/mcq_exact_match_reward/std": 0.48795005679130554,
|
|
"step": 181
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.2275,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 20.426326751708984,
|
|
"learning_rate": 2.4471741852423233e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 15973723.0,
|
|
"reward": 0.5062500238418579,
|
|
"reward_std": 0.1462520956993103,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.40625,
|
|
"rewards/mcq_exact_match_reward/std": 0.49501484632492065,
|
|
"step": 182
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.22875,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 31.343536376953125,
|
|
"learning_rate": 2.1982109232821176e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 16071523.0,
|
|
"reward": 0.7718750238418579,
|
|
"reward_std": 0.15992169082164764,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.671875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4732423722743988,
|
|
"step": 183
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.23,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.9623161931575926e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 16173651.0,
|
|
"reward": 1.100000023841858,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 1.0,
|
|
"rewards/mcq_exact_match_reward/std": 0.0,
|
|
"step": 184
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.23125,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 35.29838180541992,
|
|
"learning_rate": 1.7395544861325718e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 16246931.0,
|
|
"reward": 0.8031250238418579,
|
|
"reward_std": 0.1530819982290268,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.703125,
|
|
"rewards/mcq_exact_match_reward/std": 0.4604927599430084,
|
|
"step": 185
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.2325,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.5299867030334813e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 16313603.0,
|
|
"reward": 0.6000000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"step": 186
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.23375,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 12.993623733520508,
|
|
"learning_rate": 1.3336701375997127e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 16395667.0,
|
|
"reward": 0.6625000238418579,
|
|
"reward_std": 0.06681530922651291,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5625,
|
|
"rewards/mcq_exact_match_reward/std": 0.5,
|
|
"step": 187
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.235,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 5.812836170196533,
|
|
"learning_rate": 1.1506584608200364e-08,
|
|
"loss": 0.0,
|
|
"num_tokens": 16465627.0,
|
|
"reward": 0.8812500238418579,
|
|
"reward_std": 0.0578637570142746,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.78125,
|
|
"rewards/mcq_exact_match_reward/std": 0.4166666865348816,
|
|
"step": 188
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.23625,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 21.743375778198242,
|
|
"learning_rate": 9.810017062595321e-09,
|
|
"loss": -0.0,
|
|
"num_tokens": 16543987.0,
|
|
"reward": 0.49375003576278687,
|
|
"reward_std": 0.0883883461356163,
|
|
"rewards/format_reward/mean": 0.875,
|
|
"rewards/format_reward/std": 0.3333333432674408,
|
|
"rewards/mcq_exact_match_reward/mean": 0.40625,
|
|
"rewards/mcq_exact_match_reward/std": 0.49501484632492065,
|
|
"step": 189
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.2375,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 8.247462563808816e-09,
|
|
"loss": 0.0,
|
|
"num_tokens": 16606083.0,
|
|
"reward": 0.6000000238418579,
|
|
"reward_std": 0.0,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.5,
|
|
"rewards/mcq_exact_match_reward/std": 0.5039526224136353,
|
|
"step": 190
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.23875,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 59.542694091796875,
|
|
"learning_rate": 6.819348298638839e-09,
|
|
"loss": 0.0,
|
|
"num_tokens": 16679619.0,
|
|
"reward": 0.8187500238418579,
|
|
"reward_std": 0.0578637570142746,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.71875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4531635046005249,
|
|
"step": 191
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.24,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 24.91984748840332,
|
|
"learning_rate": 5.526064699265753e-09,
|
|
"loss": 0.0,
|
|
"num_tokens": 16754675.0,
|
|
"reward": 0.8968750238418579,
|
|
"reward_std": 0.10205793380737305,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.796875,
|
|
"rewards/mcq_exact_match_reward/std": 0.40550529956817627,
|
|
"step": 192
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.24125,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 20.199951171875,
|
|
"learning_rate": 4.367965336512403e-09,
|
|
"loss": 0.0,
|
|
"num_tokens": 16822067.0,
|
|
"reward": 0.7718750238418579,
|
|
"reward_std": 0.1530819982290268,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.671875,
|
|
"rewards/mcq_exact_match_reward/std": 0.4732423722743988,
|
|
"step": 193
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.2425,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 7.043610095977783,
|
|
"learning_rate": 3.3453668231809283e-09,
|
|
"loss": 0.0,
|
|
"num_tokens": 16907883.0,
|
|
"reward": 0.5062500238418579,
|
|
"reward_std": 0.0578637570142746,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.40625,
|
|
"rewards/mcq_exact_match_reward/std": 0.49501484632492065,
|
|
"step": 194
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.24375,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 16.697999954223633,
|
|
"learning_rate": 2.458548727494292e-09,
|
|
"loss": 0.0,
|
|
"num_tokens": 17002739.0,
|
|
"reward": 0.6156250238418579,
|
|
"reward_std": 0.10205793380737305,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.515625,
|
|
"rewards/mcq_exact_match_reward/std": 0.5037065148353577,
|
|
"step": 195
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.245,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 29.16732406616211,
|
|
"learning_rate": 1.7077534966650765e-09,
|
|
"loss": 0.0,
|
|
"num_tokens": 17087267.0,
|
|
"reward": 0.7093750238418579,
|
|
"reward_std": 0.16887323558330536,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.609375,
|
|
"rewards/mcq_exact_match_reward/std": 0.4917473793029785,
|
|
"step": 196
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.24625,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 24.210670471191406,
|
|
"learning_rate": 1.0931863906127325e-09,
|
|
"loss": 0.0,
|
|
"num_tokens": 17156259.0,
|
|
"reward": 0.5375000238418579,
|
|
"reward_std": 0.06681530922651291,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.4375,
|
|
"rewards/mcq_exact_match_reward/std": 0.5,
|
|
"step": 197
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.2475,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 22.674039840698242,
|
|
"learning_rate": 6.150154258476314e-10,
|
|
"loss": -0.0,
|
|
"num_tokens": 17241187.0,
|
|
"reward": 0.6312500238418579,
|
|
"reward_std": 0.16675157845020294,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.53125,
|
|
"rewards/mcq_exact_match_reward/std": 0.5029674172401428,
|
|
"step": 198
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.24875,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 16.04971694946289,
|
|
"learning_rate": 2.733713295369755e-10,
|
|
"loss": -0.0,
|
|
"num_tokens": 17320643.0,
|
|
"reward": 0.5375000238418579,
|
|
"reward_std": 0.06681530922651291,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.4375,
|
|
"rewards/mcq_exact_match_reward/std": 0.5,
|
|
"step": 199
|
|
},
|
|
{
|
|
"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": 12.0,
|
|
"completions/max_terminated_length": 12.0,
|
|
"completions/mean_length": 12.0,
|
|
"completions/mean_terminated_length": 12.0,
|
|
"completions/min_length": 12.0,
|
|
"completions/min_terminated_length": 12.0,
|
|
"epoch": 0.25,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 40.62717056274414,
|
|
"learning_rate": 6.834750376549791e-11,
|
|
"loss": 0.0,
|
|
"num_tokens": 17383411.0,
|
|
"reward": 0.8031250238418579,
|
|
"reward_std": 0.10205793380737305,
|
|
"rewards/format_reward/mean": 1.0,
|
|
"rewards/format_reward/std": 0.0,
|
|
"rewards/mcq_exact_match_reward/mean": 0.703125,
|
|
"rewards/mcq_exact_match_reward/std": 0.4604927599430084,
|
|
"step": 200
|
|
}
|
|
],
|
|
"logging_steps": 1,
|
|
"max_steps": 200,
|
|
"num_input_tokens_seen": 17383411,
|
|
"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
|
|
}
|