7971 lines
304 KiB
JSON
7971 lines
304 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 1.0,
|
|
"eval_steps": 100.0,
|
|
"global_step": 256,
|
|
"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": 0.0,
|
|
"completions/max_length": 3062.0,
|
|
"completions/mean_length": 1676.234375,
|
|
"completions/min_length": 774.0,
|
|
"entropy/max": 0.65234375,
|
|
"entropy/mean": 0.4375,
|
|
"entropy/min": 0.341796875,
|
|
"epoch": 0.00390625,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 0.33669739961624146,
|
|
"kl": 0.0,
|
|
"learning_rate": 7.692307692307694e-07,
|
|
"loss": 6.51925802230835e-09,
|
|
"reward": 0.33529332280158997,
|
|
"reward_std": 0.41226130723953247,
|
|
"rewards/DenseCaptionF1/mean": 0.11390755325555801,
|
|
"rewards/DenseCaptionF1/std": 0.2861669063568115,
|
|
"rewards/DenseCaptionSodaM/mean": 0.08076076209545135,
|
|
"rewards/DenseCaptionSodaM/std": 0.2062472105026245,
|
|
"rewards/TiemstampCaptionLength/mean": 0.140625,
|
|
"rewards/TiemstampCaptionLength/std": 0.3503824472427368,
|
|
"rewards/TimestampFormat/mean": 0.140625,
|
|
"rewards/TimestampFormat/std": 0.3503824472427368,
|
|
"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": 3774.0,
|
|
"completions/mean_length": 2192.40625,
|
|
"completions/min_length": 1202.0,
|
|
"entropy/max": 0.671875,
|
|
"entropy/mean": 0.46875,
|
|
"entropy/min": 0.259765625,
|
|
"epoch": 0.0078125,
|
|
"frac_reward_zero_std": 1.0,
|
|
"grad_norm": 0.0,
|
|
"kl": 0.0,
|
|
"learning_rate": 1.5384615384615387e-06,
|
|
"loss": 0.0,
|
|
"reward": 0.0,
|
|
"reward_std": 0.0,
|
|
"rewards/DenseCaptionF1/mean": 0.0,
|
|
"rewards/DenseCaptionF1/std": 0.0,
|
|
"rewards/DenseCaptionSodaM/mean": 0.0,
|
|
"rewards/DenseCaptionSodaM/std": 0.0,
|
|
"rewards/TiemstampCaptionLength/mean": 0.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 0.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"step": 2
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3581.0,
|
|
"completions/mean_length": 1615.6875,
|
|
"completions/min_length": 381.0,
|
|
"entropy/max": 0.73046875,
|
|
"entropy/mean": 0.5234375,
|
|
"entropy/min": 0.298828125,
|
|
"epoch": 0.01171875,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 0.45485445857048035,
|
|
"kl": 7.989356527104974e-05,
|
|
"learning_rate": 2.307692307692308e-06,
|
|
"loss": 3.299224772490561e-06,
|
|
"reward": 0.6258076429367065,
|
|
"reward_std": 0.419928640127182,
|
|
"rewards/DenseCaptionF1/mean": 0.20682787895202637,
|
|
"rewards/DenseCaptionF1/std": 0.3422560691833496,
|
|
"rewards/DenseCaptionSodaM/mean": 0.12210478633642197,
|
|
"rewards/DenseCaptionSodaM/std": 0.19544678926467896,
|
|
"rewards/TiemstampCaptionLength/mean": 0.296875,
|
|
"rewards/TiemstampCaptionLength/std": 0.4604927599430084,
|
|
"rewards/TimestampFormat/mean": 0.296875,
|
|
"rewards/TimestampFormat/std": 0.4604927599430084,
|
|
"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": 3052.0,
|
|
"completions/mean_length": 1579.359375,
|
|
"completions/min_length": 889.0,
|
|
"entropy/max": 0.57421875,
|
|
"entropy/mean": 0.439453125,
|
|
"entropy/min": 0.2578125,
|
|
"epoch": 0.015625,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 0.3300647735595703,
|
|
"kl": 8.311930287163705e-05,
|
|
"learning_rate": 3.0769230769230774e-06,
|
|
"loss": 3.3535045531607466e-06,
|
|
"reward": 0.20333555340766907,
|
|
"reward_std": 0.22650524973869324,
|
|
"rewards/DenseCaptionF1/mean": 0.06781341135501862,
|
|
"rewards/DenseCaptionF1/std": 0.21572406589984894,
|
|
"rewards/DenseCaptionSodaM/mean": 0.041772134602069855,
|
|
"rewards/DenseCaptionSodaM/std": 0.13558875024318695,
|
|
"rewards/TiemstampCaptionLength/mean": 0.09375,
|
|
"rewards/TiemstampCaptionLength/std": 0.29378482699394226,
|
|
"rewards/TimestampFormat/mean": 0.09375,
|
|
"rewards/TimestampFormat/std": 0.29378482699394226,
|
|
"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": 3021.0,
|
|
"completions/mean_length": 1719.71875,
|
|
"completions/min_length": 498.0,
|
|
"entropy/max": 0.875,
|
|
"entropy/mean": 0.5234375,
|
|
"entropy/min": 0.359375,
|
|
"epoch": 0.01953125,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 0.35874465107917786,
|
|
"kl": 7.733127131359652e-05,
|
|
"learning_rate": 3.846153846153847e-06,
|
|
"loss": 3.127614263576106e-06,
|
|
"reward": 0.2088506668806076,
|
|
"reward_std": 0.3283998668193817,
|
|
"rewards/DenseCaptionF1/mean": 0.055382177233695984,
|
|
"rewards/DenseCaptionF1/std": 0.16304251551628113,
|
|
"rewards/DenseCaptionSodaM/mean": 0.04409348964691162,
|
|
"rewards/DenseCaptionSodaM/std": 0.12984584271907806,
|
|
"rewards/TiemstampCaptionLength/mean": 0.109375,
|
|
"rewards/TiemstampCaptionLength/std": 0.3145764470100403,
|
|
"rewards/TimestampFormat/mean": 0.109375,
|
|
"rewards/TimestampFormat/std": 0.3145764470100403,
|
|
"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": 3330.0,
|
|
"completions/mean_length": 1659.578125,
|
|
"completions/min_length": 450.0,
|
|
"entropy/max": 0.62890625,
|
|
"entropy/mean": 0.4453125,
|
|
"entropy/min": 0.330078125,
|
|
"epoch": 0.0234375,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 0.23100338876247406,
|
|
"kl": 8.260580943897367e-05,
|
|
"learning_rate": 4.615384615384616e-06,
|
|
"loss": 3.2648445085214917e-06,
|
|
"reward": 0.11252006888389587,
|
|
"reward_std": 0.12164957076311111,
|
|
"rewards/DenseCaptionF1/mean": 0.0279682707041502,
|
|
"rewards/DenseCaptionF1/std": 0.11443595588207245,
|
|
"rewards/DenseCaptionSodaM/mean": 0.022051798179745674,
|
|
"rewards/DenseCaptionSodaM/std": 0.08787863701581955,
|
|
"rewards/TiemstampCaptionLength/mean": 0.0625,
|
|
"rewards/TiemstampCaptionLength/std": 0.24397502839565277,
|
|
"rewards/TimestampFormat/mean": 0.0625,
|
|
"rewards/TimestampFormat/std": 0.24397502839565277,
|
|
"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": 2488.0,
|
|
"completions/mean_length": 1616.1875,
|
|
"completions/min_length": 914.0,
|
|
"entropy/max": 0.6171875,
|
|
"entropy/mean": 0.47265625,
|
|
"entropy/min": 0.341796875,
|
|
"epoch": 0.02734375,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 0.34800484776496887,
|
|
"kl": 9.036576375365257e-05,
|
|
"learning_rate": 5.384615384615385e-06,
|
|
"loss": 3.6178503250994254e-06,
|
|
"reward": 0.4976370930671692,
|
|
"reward_std": 0.3053871691226959,
|
|
"rewards/DenseCaptionF1/mean": 0.14112716913223267,
|
|
"rewards/DenseCaptionF1/std": 0.25632479786872864,
|
|
"rewards/DenseCaptionSodaM/mean": 0.10650992393493652,
|
|
"rewards/DenseCaptionSodaM/std": 0.18886621296405792,
|
|
"rewards/TiemstampCaptionLength/mean": 0.25,
|
|
"rewards/TiemstampCaptionLength/std": 0.4364357888698578,
|
|
"rewards/TimestampFormat/mean": 0.25,
|
|
"rewards/TimestampFormat/std": 0.4364357888698578,
|
|
"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": 2626.0,
|
|
"completions/mean_length": 1691.46875,
|
|
"completions/min_length": 959.0,
|
|
"entropy/max": 0.71484375,
|
|
"entropy/mean": 0.498046875,
|
|
"entropy/min": 0.2119140625,
|
|
"epoch": 0.03125,
|
|
"frac_reward_zero_std": 0.375,
|
|
"grad_norm": 0.45934543013572693,
|
|
"kl": 0.00012245975085534155,
|
|
"learning_rate": 6.153846153846155e-06,
|
|
"loss": 4.908712526230374e-06,
|
|
"reward": 0.38751348853111267,
|
|
"reward_std": 0.5993294715881348,
|
|
"rewards/DenseCaptionF1/mean": 0.1268688291311264,
|
|
"rewards/DenseCaptionF1/std": 0.2948600649833679,
|
|
"rewards/DenseCaptionSodaM/mean": 0.08876965939998627,
|
|
"rewards/DenseCaptionSodaM/std": 0.20143647491931915,
|
|
"rewards/TiemstampCaptionLength/mean": 0.171875,
|
|
"rewards/TiemstampCaptionLength/std": 0.38025420904159546,
|
|
"rewards/TimestampFormat/mean": 0.171875,
|
|
"rewards/TimestampFormat/std": 0.38025420904159546,
|
|
"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": 3861.0,
|
|
"completions/mean_length": 2073.40625,
|
|
"completions/min_length": 909.0,
|
|
"entropy/max": 0.7109375,
|
|
"entropy/mean": 0.5234375,
|
|
"entropy/min": 0.3984375,
|
|
"epoch": 0.03515625,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 0.34851229190826416,
|
|
"kl": 0.00021328622824512422,
|
|
"learning_rate": 6.923076923076923e-06,
|
|
"loss": 8.552353392587975e-06,
|
|
"reward": 0.2592480778694153,
|
|
"reward_std": 0.18327659368515015,
|
|
"rewards/DenseCaptionF1/mean": 0.0821901485323906,
|
|
"rewards/DenseCaptionF1/std": 0.2231617271900177,
|
|
"rewards/DenseCaptionSodaM/mean": 0.05205794796347618,
|
|
"rewards/DenseCaptionSodaM/std": 0.1404447704553604,
|
|
"rewards/TiemstampCaptionLength/mean": 0.125,
|
|
"rewards/TiemstampCaptionLength/std": 0.3333333432674408,
|
|
"rewards/TimestampFormat/mean": 0.125,
|
|
"rewards/TimestampFormat/std": 0.3333333432674408,
|
|
"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": 3575.0,
|
|
"completions/mean_length": 1887.984375,
|
|
"completions/min_length": 885.0,
|
|
"entropy/max": 0.65234375,
|
|
"entropy/mean": 0.48828125,
|
|
"entropy/min": 0.318359375,
|
|
"epoch": 0.0390625,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 0.32556527853012085,
|
|
"kl": 0.00043061739415861666,
|
|
"learning_rate": 7.692307692307694e-06,
|
|
"loss": 1.7255888451472856e-05,
|
|
"reward": 0.09155982732772827,
|
|
"reward_std": 0.25897032022476196,
|
|
"rewards/DenseCaptionF1/mean": 0.0298549123108387,
|
|
"rewards/DenseCaptionF1/std": 0.14728239178657532,
|
|
"rewards/DenseCaptionSodaM/mean": 0.01743408665060997,
|
|
"rewards/DenseCaptionSodaM/std": 0.08487895131111145,
|
|
"rewards/TiemstampCaptionLength/mean": 0.0416666679084301,
|
|
"rewards/TiemstampCaptionLength/std": 0.19245009124279022,
|
|
"rewards/TimestampFormat/mean": 0.046875,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 2205.0,
|
|
"completions/mean_length": 1440.625,
|
|
"completions/min_length": 647.0,
|
|
"entropy/max": 0.72265625,
|
|
"entropy/mean": 0.490234375,
|
|
"entropy/min": 0.302734375,
|
|
"epoch": 0.04296875,
|
|
"frac_reward_zero_std": 0.375,
|
|
"grad_norm": 0.4874301254749298,
|
|
"kl": 0.0008995528914965689,
|
|
"learning_rate": 8.461538461538462e-06,
|
|
"loss": 3.6125522456131876e-05,
|
|
"reward": 0.8666101098060608,
|
|
"reward_std": 0.5633249878883362,
|
|
"rewards/DenseCaptionF1/mean": 0.3042565584182739,
|
|
"rewards/DenseCaptionF1/std": 0.41655826568603516,
|
|
"rewards/DenseCaptionSodaM/mean": 0.18735358119010925,
|
|
"rewards/DenseCaptionSodaM/std": 0.2545086145401001,
|
|
"rewards/TiemstampCaptionLength/mean": 0.375,
|
|
"rewards/TiemstampCaptionLength/std": 0.48795005679130554,
|
|
"rewards/TimestampFormat/mean": 0.375,
|
|
"rewards/TimestampFormat/std": 0.48795005679130554,
|
|
"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": 3659.0,
|
|
"completions/mean_length": 2064.046875,
|
|
"completions/min_length": 376.0,
|
|
"entropy/max": 0.81640625,
|
|
"entropy/mean": 0.5078125,
|
|
"entropy/min": 0.30859375,
|
|
"epoch": 0.046875,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 0.36841216683387756,
|
|
"kl": 0.0014957406092435122,
|
|
"learning_rate": 9.230769230769232e-06,
|
|
"loss": 5.995495666866191e-05,
|
|
"reward": 0.22793494164943695,
|
|
"reward_std": 0.21785634756088257,
|
|
"rewards/DenseCaptionF1/mean": 0.06400325894355774,
|
|
"rewards/DenseCaptionF1/std": 0.1750815510749817,
|
|
"rewards/DenseCaptionSodaM/mean": 0.03893166407942772,
|
|
"rewards/DenseCaptionSodaM/std": 0.11452297121286392,
|
|
"rewards/TiemstampCaptionLength/mean": 0.125,
|
|
"rewards/TiemstampCaptionLength/std": 0.3333333432674408,
|
|
"rewards/TimestampFormat/mean": 0.125,
|
|
"rewards/TimestampFormat/std": 0.3333333432674408,
|
|
"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": 3706.0,
|
|
"completions/mean_length": 2285.765625,
|
|
"completions/min_length": 1077.0,
|
|
"entropy/max": 0.640625,
|
|
"entropy/mean": 0.50390625,
|
|
"entropy/min": 0.353515625,
|
|
"epoch": 0.05078125,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 0.26272639632225037,
|
|
"kl": 0.0025145045947283506,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.00010084829409606755,
|
|
"reward": 0.11095122992992401,
|
|
"reward_std": 0.24312539398670197,
|
|
"rewards/DenseCaptionF1/mean": 0.03828125074505806,
|
|
"rewards/DenseCaptionF1/std": 0.17428816854953766,
|
|
"rewards/DenseCaptionSodaM/mean": 0.02579498291015625,
|
|
"rewards/DenseCaptionSodaM/std": 0.11784151196479797,
|
|
"rewards/TiemstampCaptionLength/mean": 0.046875,
|
|
"rewards/TiemstampCaptionLength/std": 0.21304203569889069,
|
|
"rewards/TimestampFormat/mean": 0.046875,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 2938.0,
|
|
"completions/mean_length": 1778.921875,
|
|
"completions/min_length": 819.0,
|
|
"entropy/max": 0.6953125,
|
|
"entropy/mean": 0.515625,
|
|
"entropy/min": 0.349609375,
|
|
"epoch": 0.0546875,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 0.3603832423686981,
|
|
"kl": 0.003843991318717599,
|
|
"learning_rate": 9.999582149277188e-06,
|
|
"loss": 0.000153951725224033,
|
|
"reward": 0.2635659873485565,
|
|
"reward_std": 0.2455659955739975,
|
|
"rewards/DenseCaptionF1/mean": 0.07066268473863602,
|
|
"rewards/DenseCaptionF1/std": 0.18533751368522644,
|
|
"rewards/DenseCaptionSodaM/mean": 0.0522783026099205,
|
|
"rewards/DenseCaptionSodaM/std": 0.13549016416072845,
|
|
"rewards/TiemstampCaptionLength/mean": 0.140625,
|
|
"rewards/TiemstampCaptionLength/std": 0.3503824472427368,
|
|
"rewards/TimestampFormat/mean": 0.140625,
|
|
"rewards/TimestampFormat/std": 0.3503824472427368,
|
|
"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": 0.0,
|
|
"completions/max_length": 4184.0,
|
|
"completions/mean_length": 2201.046875,
|
|
"completions/min_length": 885.0,
|
|
"entropy/max": 0.609375,
|
|
"entropy/mean": 0.5,
|
|
"entropy/min": 0.404296875,
|
|
"epoch": 0.05859375,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 0.30636030435562134,
|
|
"kl": 0.006575712934136391,
|
|
"learning_rate": 9.998328666948437e-06,
|
|
"loss": 0.00026336696464568377,
|
|
"reward": 0.20778439939022064,
|
|
"reward_std": 0.3372444808483124,
|
|
"rewards/DenseCaptionF1/mean": 0.06573660671710968,
|
|
"rewards/DenseCaptionF1/std": 0.2090059369802475,
|
|
"rewards/DenseCaptionSodaM/mean": 0.04829777777194977,
|
|
"rewards/DenseCaptionSodaM/std": 0.15207083523273468,
|
|
"rewards/TiemstampCaptionLength/mean": 0.09375,
|
|
"rewards/TiemstampCaptionLength/std": 0.29378482699394226,
|
|
"rewards/TimestampFormat/mean": 0.09375,
|
|
"rewards/TimestampFormat/std": 0.29378482699394226,
|
|
"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": 4208.0,
|
|
"completions/mean_length": 2478.390625,
|
|
"completions/min_length": 863.0,
|
|
"entropy/max": 0.671875,
|
|
"entropy/mean": 0.51953125,
|
|
"entropy/min": 0.365234375,
|
|
"epoch": 0.0625,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 0.20259444415569305,
|
|
"kl": 0.008406754583120346,
|
|
"learning_rate": 9.996239762521152e-06,
|
|
"loss": 0.0003381561255082488,
|
|
"reward": 0.12115638703107834,
|
|
"reward_std": 0.16726306080818176,
|
|
"rewards/DenseCaptionF1/mean": 0.0455729179084301,
|
|
"rewards/DenseCaptionF1/std": 0.20730151236057281,
|
|
"rewards/DenseCaptionSodaM/mean": 0.028708472847938538,
|
|
"rewards/DenseCaptionSodaM/std": 0.1305268108844757,
|
|
"rewards/TiemstampCaptionLength/mean": 0.046875,
|
|
"rewards/TiemstampCaptionLength/std": 0.21304203569889069,
|
|
"rewards/TimestampFormat/mean": 0.046875,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 0.015625,
|
|
"completions/max_length": 9216.0,
|
|
"completions/mean_length": 2195.046875,
|
|
"completions/min_length": 728.0,
|
|
"entropy/max": 1.1875,
|
|
"entropy/mean": 0.609375,
|
|
"entropy/min": 0.1103515625,
|
|
"epoch": 0.06640625,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 0.2823851406574249,
|
|
"kl": 0.011962870135903358,
|
|
"learning_rate": 9.993315785135417e-06,
|
|
"loss": 0.00048333226004615426,
|
|
"reward": 0.35875755548477173,
|
|
"reward_std": 0.16727258265018463,
|
|
"rewards/DenseCaptionF1/mean": 0.11765454709529877,
|
|
"rewards/DenseCaptionF1/std": 0.29483330249786377,
|
|
"rewards/DenseCaptionSodaM/mean": 0.08485300838947296,
|
|
"rewards/DenseCaptionSodaM/std": 0.20221486687660217,
|
|
"rewards/TiemstampCaptionLength/mean": 0.15625,
|
|
"rewards/TiemstampCaptionLength/std": 0.36596253514289856,
|
|
"rewards/TimestampFormat/mean": 0.15625,
|
|
"rewards/TimestampFormat/std": 0.36596253514289856,
|
|
"step": 17
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3605.0,
|
|
"completions/mean_length": 2101.4375,
|
|
"completions/min_length": 880.0,
|
|
"entropy/max": 0.7265625,
|
|
"entropy/mean": 0.5078125,
|
|
"entropy/min": 0.3203125,
|
|
"epoch": 0.0703125,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 0.2937946319580078,
|
|
"kl": 0.015237608924508095,
|
|
"learning_rate": 9.989557223505661e-06,
|
|
"loss": 0.0006094181444495916,
|
|
"reward": 0.3275236189365387,
|
|
"reward_std": 0.23650048673152924,
|
|
"rewards/DenseCaptionF1/mean": 0.10937843471765518,
|
|
"rewards/DenseCaptionF1/std": 0.2602582275867462,
|
|
"rewards/DenseCaptionSodaM/mean": 0.06189517304301262,
|
|
"rewards/DenseCaptionSodaM/std": 0.14991170167922974,
|
|
"rewards/TiemstampCaptionLength/mean": 0.15625,
|
|
"rewards/TiemstampCaptionLength/std": 0.36596253514289856,
|
|
"rewards/TimestampFormat/mean": 0.15625,
|
|
"rewards/TimestampFormat/std": 0.36596253514289856,
|
|
"step": 18
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3604.0,
|
|
"completions/mean_length": 2388.359375,
|
|
"completions/min_length": 1166.0,
|
|
"entropy/max": 0.7734375,
|
|
"entropy/mean": 0.55078125,
|
|
"entropy/min": 0.380859375,
|
|
"epoch": 0.07421875,
|
|
"frac_reward_zero_std": 0.375,
|
|
"grad_norm": 0.37759771943092346,
|
|
"kl": 0.0160197950899601,
|
|
"learning_rate": 9.98496470583896e-06,
|
|
"loss": 0.0006415115785785019,
|
|
"reward": 0.4447316527366638,
|
|
"reward_std": 0.6170377731323242,
|
|
"rewards/DenseCaptionF1/mean": 0.1351507306098938,
|
|
"rewards/DenseCaptionF1/std": 0.27623483538627625,
|
|
"rewards/DenseCaptionSodaM/mean": 0.10775801539421082,
|
|
"rewards/DenseCaptionSodaM/std": 0.21980473399162292,
|
|
"rewards/TiemstampCaptionLength/mean": 0.2005208283662796,
|
|
"rewards/TiemstampCaptionLength/std": 0.4008145332336426,
|
|
"rewards/TimestampFormat/mean": 0.203125,
|
|
"rewards/TimestampFormat/std": 0.40550529956817627,
|
|
"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": 5027.0,
|
|
"completions/mean_length": 2083.15625,
|
|
"completions/min_length": 590.0,
|
|
"entropy/max": 0.90234375,
|
|
"entropy/mean": 0.54296875,
|
|
"entropy/min": 0.30078125,
|
|
"epoch": 0.078125,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 0.3374199867248535,
|
|
"kl": 0.016353856772184372,
|
|
"learning_rate": 9.979538999730047e-06,
|
|
"loss": 0.0006515942513942719,
|
|
"reward": 0.12008658796548843,
|
|
"reward_std": 0.2829945981502533,
|
|
"rewards/DenseCaptionF1/mean": 0.03498358279466629,
|
|
"rewards/DenseCaptionF1/std": 0.13844537734985352,
|
|
"rewards/DenseCaptionSodaM/mean": 0.022603008896112442,
|
|
"rewards/DenseCaptionSodaM/std": 0.09045752882957458,
|
|
"rewards/TiemstampCaptionLength/mean": 0.0625,
|
|
"rewards/TiemstampCaptionLength/std": 0.24397502839565277,
|
|
"rewards/TimestampFormat/mean": 0.0625,
|
|
"rewards/TimestampFormat/std": 0.24397502839565277,
|
|
"step": 20
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3595.0,
|
|
"completions/mean_length": 1994.25,
|
|
"completions/min_length": 964.0,
|
|
"entropy/max": 0.765625,
|
|
"entropy/mean": 0.55078125,
|
|
"entropy/min": 0.376953125,
|
|
"epoch": 0.08203125,
|
|
"frac_reward_zero_std": 0.625,
|
|
"grad_norm": 0.3366459012031555,
|
|
"kl": 0.01988409459590912,
|
|
"learning_rate": 9.973281012033009e-06,
|
|
"loss": 0.000799685250967741,
|
|
"reward": 0.33904898166656494,
|
|
"reward_std": 0.23382535576820374,
|
|
"rewards/DenseCaptionF1/mean": 0.09593891352415085,
|
|
"rewards/DenseCaptionF1/std": 0.21650290489196777,
|
|
"rewards/DenseCaptionSodaM/mean": 0.07123503088951111,
|
|
"rewards/DenseCaptionSodaM/std": 0.1632857322692871,
|
|
"rewards/TiemstampCaptionLength/mean": 0.171875,
|
|
"rewards/TiemstampCaptionLength/std": 0.38025420904159546,
|
|
"rewards/TimestampFormat/mean": 0.171875,
|
|
"rewards/TimestampFormat/std": 0.38025420904159546,
|
|
"step": 21
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3650.0,
|
|
"completions/mean_length": 1928.578125,
|
|
"completions/min_length": 816.0,
|
|
"entropy/max": 0.875,
|
|
"entropy/mean": 0.5859375,
|
|
"entropy/min": 0.43359375,
|
|
"epoch": 0.0859375,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 0.4066086709499359,
|
|
"kl": 0.022677309811115265,
|
|
"learning_rate": 9.966191788709716e-06,
|
|
"loss": 0.0009096093126572669,
|
|
"reward": 0.18491098284721375,
|
|
"reward_std": 0.40771812200546265,
|
|
"rewards/DenseCaptionF1/mean": 0.05379953980445862,
|
|
"rewards/DenseCaptionF1/std": 0.1785956472158432,
|
|
"rewards/DenseCaptionSodaM/mean": 0.037361450493335724,
|
|
"rewards/DenseCaptionSodaM/std": 0.1213492900133133,
|
|
"rewards/TiemstampCaptionLength/mean": 0.09375,
|
|
"rewards/TiemstampCaptionLength/std": 0.29378482699394226,
|
|
"rewards/TimestampFormat/mean": 0.09375,
|
|
"rewards/TimestampFormat/std": 0.29378482699394226,
|
|
"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": 3628.0,
|
|
"completions/mean_length": 2177.875,
|
|
"completions/min_length": 902.0,
|
|
"entropy/max": 0.70703125,
|
|
"entropy/mean": 0.51171875,
|
|
"entropy/min": 0.3828125,
|
|
"epoch": 0.08984375,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 0.35382241010665894,
|
|
"kl": 0.020987944677472115,
|
|
"learning_rate": 9.958272514655006e-06,
|
|
"loss": 0.000841582368593663,
|
|
"reward": 0.2189454734325409,
|
|
"reward_std": 0.4250008761882782,
|
|
"rewards/DenseCaptionF1/mean": 0.06233471632003784,
|
|
"rewards/DenseCaptionF1/std": 0.1853744387626648,
|
|
"rewards/DenseCaptionSodaM/mean": 0.047235749661922455,
|
|
"rewards/DenseCaptionSodaM/std": 0.13834145665168762,
|
|
"rewards/TiemstampCaptionLength/mean": 0.109375,
|
|
"rewards/TiemstampCaptionLength/std": 0.3145764470100403,
|
|
"rewards/TimestampFormat/mean": 0.109375,
|
|
"rewards/TimestampFormat/std": 0.3145764470100403,
|
|
"step": 23
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3190.0,
|
|
"completions/mean_length": 1853.296875,
|
|
"completions/min_length": 989.0,
|
|
"entropy/max": 0.79296875,
|
|
"entropy/mean": 0.56640625,
|
|
"entropy/min": 0.4375,
|
|
"epoch": 0.09375,
|
|
"frac_reward_zero_std": 0.375,
|
|
"grad_norm": 0.44761282205581665,
|
|
"kl": 0.029041863977909088,
|
|
"learning_rate": 9.949524513498636e-06,
|
|
"loss": 0.0011664638295769691,
|
|
"reward": 0.7176538705825806,
|
|
"reward_std": 0.5046995282173157,
|
|
"rewards/DenseCaptionF1/mean": 0.21744775772094727,
|
|
"rewards/DenseCaptionF1/std": 0.2997380197048187,
|
|
"rewards/DenseCaptionSodaM/mean": 0.1408311277627945,
|
|
"rewards/DenseCaptionSodaM/std": 0.1953917145729065,
|
|
"rewards/TiemstampCaptionLength/mean": 0.359375,
|
|
"rewards/TiemstampCaptionLength/std": 0.4836103618144989,
|
|
"rewards/TimestampFormat/mean": 0.359375,
|
|
"rewards/TimestampFormat/std": 0.4836103618144989,
|
|
"step": 24
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3667.0,
|
|
"completions/mean_length": 2230.59375,
|
|
"completions/min_length": 1048.0,
|
|
"entropy/max": 0.76953125,
|
|
"entropy/mean": 0.578125,
|
|
"entropy/min": 0.4296875,
|
|
"epoch": 0.09765625,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 0.2841798663139343,
|
|
"kl": 0.028013236820697784,
|
|
"learning_rate": 9.939949247384046e-06,
|
|
"loss": 0.0011283187195658684,
|
|
"reward": 0.3157457709312439,
|
|
"reward_std": 0.10166770219802856,
|
|
"rewards/DenseCaptionF1/mean": 0.09699587523937225,
|
|
"rewards/DenseCaptionF1/std": 0.24287192523479462,
|
|
"rewards/DenseCaptionSodaM/mean": 0.07812491804361343,
|
|
"rewards/DenseCaptionSodaM/std": 0.1968449503183365,
|
|
"rewards/TiemstampCaptionLength/mean": 0.140625,
|
|
"rewards/TiemstampCaptionLength/std": 0.3503824472427368,
|
|
"rewards/TimestampFormat/mean": 0.140625,
|
|
"rewards/TimestampFormat/std": 0.3503824472427368,
|
|
"step": 25
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 4428.0,
|
|
"completions/mean_length": 1915.203125,
|
|
"completions/min_length": 244.0,
|
|
"entropy/max": 0.90625,
|
|
"entropy/mean": 0.55859375,
|
|
"entropy/min": 0.33984375,
|
|
"epoch": 0.1015625,
|
|
"frac_reward_zero_std": 0.125,
|
|
"grad_norm": 0.5215222835540771,
|
|
"kl": 0.032491087913513184,
|
|
"learning_rate": 9.929548316723983e-06,
|
|
"loss": 0.0013178382068872452,
|
|
"reward": 0.6358413696289062,
|
|
"reward_std": 0.7758326530456543,
|
|
"rewards/DenseCaptionF1/mean": 0.18299464881420135,
|
|
"rewards/DenseCaptionF1/std": 0.28064069151878357,
|
|
"rewards/DenseCaptionSodaM/mean": 0.1305810809135437,
|
|
"rewards/DenseCaptionSodaM/std": 0.20247425138950348,
|
|
"rewards/TiemstampCaptionLength/mean": 0.31640625,
|
|
"rewards/TiemstampCaptionLength/std": 0.4612831473350525,
|
|
"rewards/TimestampFormat/mean": 0.328125,
|
|
"rewards/TimestampFormat/std": 0.4732423722743988,
|
|
"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": 4103.0,
|
|
"completions/mean_length": 2101.765625,
|
|
"completions/min_length": 952.0,
|
|
"entropy/max": 0.9375,
|
|
"entropy/mean": 0.578125,
|
|
"entropy/min": 0.28125,
|
|
"epoch": 0.10546875,
|
|
"frac_reward_zero_std": 0.375,
|
|
"grad_norm": 0.3893539607524872,
|
|
"kl": 0.02889391966164112,
|
|
"learning_rate": 9.918323459933006e-06,
|
|
"loss": 0.001159125124104321,
|
|
"reward": 0.35946089029312134,
|
|
"reward_std": 0.5790758728981018,
|
|
"rewards/DenseCaptionF1/mean": 0.12246093153953552,
|
|
"rewards/DenseCaptionF1/std": 0.27518293261528015,
|
|
"rewards/DenseCaptionSodaM/mean": 0.06981244683265686,
|
|
"rewards/DenseCaptionSodaM/std": 0.15849006175994873,
|
|
"rewards/TiemstampCaptionLength/mean": 0.16249999403953552,
|
|
"rewards/TiemstampCaptionLength/std": 0.36666667461395264,
|
|
"rewards/TimestampFormat/mean": 0.171875,
|
|
"rewards/TimestampFormat/std": 0.38025420904159546,
|
|
"step": 27
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3711.0,
|
|
"completions/mean_length": 1937.203125,
|
|
"completions/min_length": 752.0,
|
|
"entropy/max": 0.7734375,
|
|
"entropy/mean": 0.5546875,
|
|
"entropy/min": 0.365234375,
|
|
"epoch": 0.109375,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 0.4255591034889221,
|
|
"kl": 0.03484990447759628,
|
|
"learning_rate": 9.906276553136924e-06,
|
|
"loss": 0.0014025644632056355,
|
|
"reward": 0.1896531581878662,
|
|
"reward_std": 0.35658401250839233,
|
|
"rewards/DenseCaptionF1/mean": 0.0452314093708992,
|
|
"rewards/DenseCaptionF1/std": 0.14166533946990967,
|
|
"rewards/DenseCaptionSodaM/mean": 0.03504673391580582,
|
|
"rewards/DenseCaptionSodaM/std": 0.10599219799041748,
|
|
"rewards/TiemstampCaptionLength/mean": 0.109375,
|
|
"rewards/TiemstampCaptionLength/std": 0.3145764470100403,
|
|
"rewards/TimestampFormat/mean": 0.109375,
|
|
"rewards/TimestampFormat/std": 0.3145764470100403,
|
|
"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": 3660.0,
|
|
"completions/mean_length": 1987.65625,
|
|
"completions/min_length": 754.0,
|
|
"entropy/max": 0.80859375,
|
|
"entropy/mean": 0.55859375,
|
|
"entropy/min": 0.259765625,
|
|
"epoch": 0.11328125,
|
|
"frac_reward_zero_std": 0.25,
|
|
"grad_norm": 0.45167720317840576,
|
|
"kl": 0.0329359695315361,
|
|
"learning_rate": 9.893409609859221e-06,
|
|
"loss": 0.0013238743413239717,
|
|
"reward": 1.1427762508392334,
|
|
"reward_std": 0.45914119482040405,
|
|
"rewards/DenseCaptionF1/mean": 0.3499799966812134,
|
|
"rewards/DenseCaptionF1/std": 0.3501557409763336,
|
|
"rewards/DenseCaptionSodaM/mean": 0.2472234070301056,
|
|
"rewards/DenseCaptionSodaM/std": 0.24433213472366333,
|
|
"rewards/TiemstampCaptionLength/mean": 0.5442708134651184,
|
|
"rewards/TiemstampCaptionLength/std": 0.4997726082801819,
|
|
"rewards/TimestampFormat/mean": 0.546875,
|
|
"rewards/TimestampFormat/std": 0.501733124256134,
|
|
"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": 2918.0,
|
|
"completions/mean_length": 1711.234375,
|
|
"completions/min_length": 709.0,
|
|
"entropy/max": 0.859375,
|
|
"entropy/mean": 0.56640625,
|
|
"entropy/min": 0.443359375,
|
|
"epoch": 0.1171875,
|
|
"frac_reward_zero_std": 0.25,
|
|
"grad_norm": 0.49171510338783264,
|
|
"kl": 0.041421182453632355,
|
|
"learning_rate": 9.879724780684518e-06,
|
|
"loss": 0.0016657232772558928,
|
|
"reward": 0.6909555196762085,
|
|
"reward_std": 0.6424126625061035,
|
|
"rewards/DenseCaptionF1/mean": 0.2269957959651947,
|
|
"rewards/DenseCaptionF1/std": 0.33632996678352356,
|
|
"rewards/DenseCaptionSodaM/mean": 0.1358346790075302,
|
|
"rewards/DenseCaptionSodaM/std": 0.21048593521118164,
|
|
"rewards/TiemstampCaptionLength/mean": 0.328125,
|
|
"rewards/TiemstampCaptionLength/std": 0.4732423722743988,
|
|
"rewards/TimestampFormat/mean": 0.328125,
|
|
"rewards/TimestampFormat/std": 0.4732423722743988,
|
|
"step": 30
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3525.0,
|
|
"completions/mean_length": 1632.375,
|
|
"completions/min_length": 592.0,
|
|
"entropy/max": 0.75390625,
|
|
"entropy/mean": 0.484375,
|
|
"entropy/min": 0.28125,
|
|
"epoch": 0.12109375,
|
|
"frac_reward_zero_std": 0.375,
|
|
"grad_norm": 0.4527930021286011,
|
|
"kl": 0.03862081840634346,
|
|
"learning_rate": 9.86522435289912e-06,
|
|
"loss": 0.0015425803139805794,
|
|
"reward": 0.660457968711853,
|
|
"reward_std": 0.642650842666626,
|
|
"rewards/DenseCaptionF1/mean": 0.20808811485767365,
|
|
"rewards/DenseCaptionF1/std": 0.33126798272132874,
|
|
"rewards/DenseCaptionSodaM/mean": 0.15549485385417938,
|
|
"rewards/DenseCaptionSodaM/std": 0.2465606927871704,
|
|
"rewards/TiemstampCaptionLength/mean": 0.296875,
|
|
"rewards/TiemstampCaptionLength/std": 0.4604927599430084,
|
|
"rewards/TimestampFormat/mean": 0.296875,
|
|
"rewards/TimestampFormat/std": 0.4604927599430084,
|
|
"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": 3501.0,
|
|
"completions/mean_length": 1980.0625,
|
|
"completions/min_length": 639.0,
|
|
"entropy/max": 0.875,
|
|
"entropy/mean": 0.578125,
|
|
"entropy/min": 0.416015625,
|
|
"epoch": 0.125,
|
|
"frac_reward_zero_std": 0.375,
|
|
"grad_norm": 0.47621989250183105,
|
|
"kl": 0.045973040163517,
|
|
"learning_rate": 9.849910750108718e-06,
|
|
"loss": 0.0018451442010700703,
|
|
"reward": 0.6612102389335632,
|
|
"reward_std": 0.5066034197807312,
|
|
"rewards/DenseCaptionF1/mean": 0.18637947738170624,
|
|
"rewards/DenseCaptionF1/std": 0.3137284815311432,
|
|
"rewards/DenseCaptionSodaM/mean": 0.1515420526266098,
|
|
"rewards/DenseCaptionSodaM/std": 0.22664391994476318,
|
|
"rewards/TiemstampCaptionLength/mean": 0.318452388048172,
|
|
"rewards/TiemstampCaptionLength/std": 0.46229735016822815,
|
|
"rewards/TimestampFormat/mean": 0.328125,
|
|
"rewards/TimestampFormat/std": 0.4732423722743988,
|
|
"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": 3340.0,
|
|
"completions/mean_length": 1555.0625,
|
|
"completions/min_length": 903.0,
|
|
"entropy/max": 0.65625,
|
|
"entropy/mean": 0.50390625,
|
|
"entropy/min": 0.330078125,
|
|
"epoch": 0.12890625,
|
|
"frac_reward_zero_std": 0.25,
|
|
"grad_norm": 0.4809737801551819,
|
|
"kl": 0.04283065348863602,
|
|
"learning_rate": 9.833786531833311e-06,
|
|
"loss": 0.0017127534374594688,
|
|
"reward": 0.8187474012374878,
|
|
"reward_std": 0.5873106718063354,
|
|
"rewards/DenseCaptionF1/mean": 0.25816553831100464,
|
|
"rewards/DenseCaptionF1/std": 0.34912580251693726,
|
|
"rewards/DenseCaptionSodaM/mean": 0.19079023599624634,
|
|
"rewards/DenseCaptionSodaM/std": 0.25204429030418396,
|
|
"rewards/TiemstampCaptionLength/mean": 0.3645833134651184,
|
|
"rewards/TiemstampCaptionLength/std": 0.4814687669277191,
|
|
"rewards/TimestampFormat/mean": 0.375,
|
|
"rewards/TimestampFormat/std": 0.48795005679130554,
|
|
"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": 3063.0,
|
|
"completions/mean_length": 2028.6875,
|
|
"completions/min_length": 1019.0,
|
|
"entropy/max": 0.83984375,
|
|
"entropy/mean": 0.54296875,
|
|
"entropy/min": 0.40234375,
|
|
"epoch": 0.1328125,
|
|
"frac_reward_zero_std": 0.125,
|
|
"grad_norm": 0.4648730158805847,
|
|
"kl": 0.043154969811439514,
|
|
"learning_rate": 9.816854393079402e-06,
|
|
"loss": 0.0017284248024225235,
|
|
"reward": 0.8269856572151184,
|
|
"reward_std": 0.7138704061508179,
|
|
"rewards/DenseCaptionF1/mean": 0.2372475117444992,
|
|
"rewards/DenseCaptionF1/std": 0.30505022406578064,
|
|
"rewards/DenseCaptionSodaM/mean": 0.183488130569458,
|
|
"rewards/DenseCaptionSodaM/std": 0.23432087898254395,
|
|
"rewards/TiemstampCaptionLength/mean": 0.40625,
|
|
"rewards/TiemstampCaptionLength/std": 0.49501484632492065,
|
|
"rewards/TimestampFormat/mean": 0.40625,
|
|
"rewards/TimestampFormat/std": 0.49501484632492065,
|
|
"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": 3714.0,
|
|
"completions/mean_length": 1911.1875,
|
|
"completions/min_length": 325.0,
|
|
"entropy/max": 0.76953125,
|
|
"entropy/mean": 0.52734375,
|
|
"entropy/min": 0.35546875,
|
|
"epoch": 0.13671875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5330745577812195,
|
|
"kl": 0.04283313453197479,
|
|
"learning_rate": 9.79911716388956e-06,
|
|
"loss": 0.0017346041277050972,
|
|
"reward": 0.7788149118423462,
|
|
"reward_std": 0.6244890093803406,
|
|
"rewards/DenseCaptionF1/mean": 0.22027342021465302,
|
|
"rewards/DenseCaptionF1/std": 0.28901150822639465,
|
|
"rewards/DenseCaptionSodaM/mean": 0.17253299057483673,
|
|
"rewards/DenseCaptionSodaM/std": 0.2367001622915268,
|
|
"rewards/TiemstampCaptionLength/mean": 0.38139206171035767,
|
|
"rewards/TiemstampCaptionLength/std": 0.4841456711292267,
|
|
"rewards/TimestampFormat/mean": 0.390625,
|
|
"rewards/TimestampFormat/std": 0.4917473793029785,
|
|
"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": 3692.0,
|
|
"completions/mean_length": 1689.375,
|
|
"completions/min_length": 735.0,
|
|
"entropy/max": 0.8359375,
|
|
"entropy/mean": 0.5625,
|
|
"entropy/min": 0.3828125,
|
|
"epoch": 0.140625,
|
|
"frac_reward_zero_std": 0.25,
|
|
"grad_norm": 0.5430750846862793,
|
|
"kl": 0.04801669344305992,
|
|
"learning_rate": 9.7805778088694e-06,
|
|
"loss": 0.0019329048227518797,
|
|
"reward": 1.0940829515457153,
|
|
"reward_std": 0.47211867570877075,
|
|
"rewards/DenseCaptionF1/mean": 0.3414607346057892,
|
|
"rewards/DenseCaptionF1/std": 0.35802242159843445,
|
|
"rewards/DenseCaptionSodaM/mean": 0.24480977654457092,
|
|
"rewards/DenseCaptionSodaM/std": 0.2575652599334717,
|
|
"rewards/TiemstampCaptionLength/mean": 0.5,
|
|
"rewards/TiemstampCaptionLength/std": 0.4940117597579956,
|
|
"rewards/TimestampFormat/mean": 0.515625,
|
|
"rewards/TimestampFormat/std": 0.5037065148353577,
|
|
"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": 2703.0,
|
|
"completions/mean_length": 1723.71875,
|
|
"completions/min_length": 672.0,
|
|
"entropy/max": 0.92578125,
|
|
"entropy/mean": 0.5859375,
|
|
"entropy/min": 0.431640625,
|
|
"epoch": 0.14453125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5763617157936096,
|
|
"kl": 0.051824457943439484,
|
|
"learning_rate": 9.761239426692077e-06,
|
|
"loss": 0.0020843958482146263,
|
|
"reward": 1.4357333183288574,
|
|
"reward_std": 0.6296243667602539,
|
|
"rewards/DenseCaptionF1/mean": 0.44952523708343506,
|
|
"rewards/DenseCaptionF1/std": 0.3528907895088196,
|
|
"rewards/DenseCaptionSodaM/mean": 0.2830829918384552,
|
|
"rewards/DenseCaptionSodaM/std": 0.21533827483654022,
|
|
"rewards/TiemstampCaptionLength/mean": 0.703125,
|
|
"rewards/TiemstampCaptionLength/std": 0.4604927599430084,
|
|
"rewards/TimestampFormat/mean": 0.703125,
|
|
"rewards/TimestampFormat/std": 0.4604927599430084,
|
|
"step": 37
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3096.0,
|
|
"completions/mean_length": 1675.75,
|
|
"completions/min_length": 761.0,
|
|
"entropy/max": 0.76171875,
|
|
"entropy/mean": 0.53515625,
|
|
"entropy/min": 0.349609375,
|
|
"epoch": 0.1484375,
|
|
"frac_reward_zero_std": 0.125,
|
|
"grad_norm": 0.5313538312911987,
|
|
"kl": 0.053580090403556824,
|
|
"learning_rate": 9.741105249580383e-06,
|
|
"loss": 0.002145090838894248,
|
|
"reward": 1.1066089868545532,
|
|
"reward_std": 0.7426514029502869,
|
|
"rewards/DenseCaptionF1/mean": 0.35633063316345215,
|
|
"rewards/DenseCaptionF1/std": 0.36695289611816406,
|
|
"rewards/DenseCaptionSodaM/mean": 0.2346533089876175,
|
|
"rewards/DenseCaptionSodaM/std": 0.2370554357767105,
|
|
"rewards/TiemstampCaptionLength/mean": 0.515625,
|
|
"rewards/TiemstampCaptionLength/std": 0.5037065148353577,
|
|
"rewards/TimestampFormat/mean": 0.515625,
|
|
"rewards/TimestampFormat/std": 0.5037065148353577,
|
|
"step": 38
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3117.0,
|
|
"completions/mean_length": 1778.984375,
|
|
"completions/min_length": 917.0,
|
|
"entropy/max": 0.7734375,
|
|
"entropy/mean": 0.51953125,
|
|
"entropy/min": 0.26953125,
|
|
"epoch": 0.15234375,
|
|
"frac_reward_zero_std": 0.125,
|
|
"grad_norm": 0.4957989454269409,
|
|
"kl": 0.04684580862522125,
|
|
"learning_rate": 9.7201786427665e-06,
|
|
"loss": 0.0018801564583554864,
|
|
"reward": 1.1384104490280151,
|
|
"reward_std": 0.8322298526763916,
|
|
"rewards/DenseCaptionF1/mean": 0.3500368595123291,
|
|
"rewards/DenseCaptionF1/std": 0.35018593072891235,
|
|
"rewards/DenseCaptionSodaM/mean": 0.24345172941684723,
|
|
"rewards/DenseCaptionSodaM/std": 0.23924703896045685,
|
|
"rewards/TiemstampCaptionLength/mean": 0.54296875,
|
|
"rewards/TiemstampCaptionLength/std": 0.4991156756877899,
|
|
"rewards/TimestampFormat/mean": 0.546875,
|
|
"rewards/TimestampFormat/std": 0.501733124256134,
|
|
"step": 39
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 2655.0,
|
|
"completions/mean_length": 1472.078125,
|
|
"completions/min_length": 351.0,
|
|
"entropy/max": 0.609375,
|
|
"entropy/mean": 0.50390625,
|
|
"entropy/min": 0.2314453125,
|
|
"epoch": 0.15625,
|
|
"frac_reward_zero_std": 0.125,
|
|
"grad_norm": 0.5393358469009399,
|
|
"kl": 0.06385189294815063,
|
|
"learning_rate": 9.698463103929542e-06,
|
|
"loss": 0.002563423477113247,
|
|
"reward": 1.4312630891799927,
|
|
"reward_std": 0.44951027631759644,
|
|
"rewards/DenseCaptionF1/mean": 0.43007513880729675,
|
|
"rewards/DenseCaptionF1/std": 0.33957046270370483,
|
|
"rewards/DenseCaptionSodaM/mean": 0.2980629503726959,
|
|
"rewards/DenseCaptionSodaM/std": 0.2082088440656662,
|
|
"rewards/TiemstampCaptionLength/mean": 0.703125,
|
|
"rewards/TiemstampCaptionLength/std": 0.4604927599430084,
|
|
"rewards/TimestampFormat/mean": 0.703125,
|
|
"rewards/TimestampFormat/std": 0.4604927599430084,
|
|
"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": 2781.0,
|
|
"completions/mean_length": 1705.421875,
|
|
"completions/min_length": 748.0,
|
|
"entropy/max": 0.85546875,
|
|
"entropy/mean": 0.51953125,
|
|
"entropy/min": 0.326171875,
|
|
"epoch": 0.16015625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5383458733558655,
|
|
"kl": 0.05609148368239403,
|
|
"learning_rate": 9.67596226261095e-06,
|
|
"loss": 0.002261662855744362,
|
|
"reward": 1.6230489015579224,
|
|
"reward_std": 0.5145324468612671,
|
|
"rewards/DenseCaptionF1/mean": 0.4726840853691101,
|
|
"rewards/DenseCaptionF1/std": 0.2743079364299774,
|
|
"rewards/DenseCaptionSodaM/mean": 0.3404690623283386,
|
|
"rewards/DenseCaptionSodaM/std": 0.1955847442150116,
|
|
"rewards/TiemstampCaptionLength/mean": 0.8072916865348816,
|
|
"rewards/TiemstampCaptionLength/std": 0.39308255910873413,
|
|
"rewards/TimestampFormat/mean": 0.8125,
|
|
"rewards/TimestampFormat/std": 0.39339789748191833,
|
|
"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": 3672.0,
|
|
"completions/mean_length": 1794.90625,
|
|
"completions/min_length": 670.0,
|
|
"entropy/max": 0.7734375,
|
|
"entropy/mean": 0.51953125,
|
|
"entropy/min": 0.337890625,
|
|
"epoch": 0.1640625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5417736172676086,
|
|
"kl": 0.05722331255674362,
|
|
"learning_rate": 9.652679879607843e-06,
|
|
"loss": 0.0022972850129008293,
|
|
"reward": 1.420531988143921,
|
|
"reward_std": 0.9789804220199585,
|
|
"rewards/DenseCaptionF1/mean": 0.460341215133667,
|
|
"rewards/DenseCaptionF1/std": 0.37514445185661316,
|
|
"rewards/DenseCaptionSodaM/mean": 0.3039408326148987,
|
|
"rewards/DenseCaptionSodaM/std": 0.24638064205646515,
|
|
"rewards/TiemstampCaptionLength/mean": 0.65625,
|
|
"rewards/TiemstampCaptionLength/std": 0.4787135720252991,
|
|
"rewards/TimestampFormat/mean": 0.65625,
|
|
"rewards/TimestampFormat/std": 0.4787135720252991,
|
|
"step": 42
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 2397.0,
|
|
"completions/mean_length": 1448.8125,
|
|
"completions/min_length": 568.0,
|
|
"entropy/max": 0.734375,
|
|
"entropy/mean": 0.5546875,
|
|
"entropy/min": 0.357421875,
|
|
"epoch": 0.16796875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.6057270169258118,
|
|
"kl": 0.059576164931058884,
|
|
"learning_rate": 9.628619846344453e-06,
|
|
"loss": 0.0023994343355298042,
|
|
"reward": 1.9311561584472656,
|
|
"reward_std": 0.5075036287307739,
|
|
"rewards/DenseCaptionF1/mean": 0.6298707723617554,
|
|
"rewards/DenseCaptionF1/std": 0.2856041491031647,
|
|
"rewards/DenseCaptionSodaM/mean": 0.41066038608551025,
|
|
"rewards/DenseCaptionSodaM/std": 0.17797227203845978,
|
|
"rewards/TiemstampCaptionLength/mean": 0.890625,
|
|
"rewards/TiemstampCaptionLength/std": 0.3145764470100403,
|
|
"rewards/TimestampFormat/mean": 0.890625,
|
|
"rewards/TimestampFormat/std": 0.3145764470100403,
|
|
"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": 2799.0,
|
|
"completions/mean_length": 1738.515625,
|
|
"completions/min_length": 554.0,
|
|
"entropy/max": 0.77734375,
|
|
"entropy/mean": 0.54296875,
|
|
"entropy/min": 0.412109375,
|
|
"epoch": 0.171875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5635643005371094,
|
|
"kl": 0.05436231195926666,
|
|
"learning_rate": 9.603786184221693e-06,
|
|
"loss": 0.0021877912804484367,
|
|
"reward": 1.4041552543640137,
|
|
"reward_std": 0.8095611333847046,
|
|
"rewards/DenseCaptionF1/mean": 0.44313710927963257,
|
|
"rewards/DenseCaptionF1/std": 0.34724023938179016,
|
|
"rewards/DenseCaptionSodaM/mean": 0.2735181152820587,
|
|
"rewards/DenseCaptionSodaM/std": 0.19664731621742249,
|
|
"rewards/TiemstampCaptionLength/mean": 0.6875,
|
|
"rewards/TiemstampCaptionLength/std": 0.467176616191864,
|
|
"rewards/TimestampFormat/mean": 0.6875,
|
|
"rewards/TimestampFormat/std": 0.467176616191864,
|
|
"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": 3219.0,
|
|
"completions/mean_length": 1852.859375,
|
|
"completions/min_length": 1007.0,
|
|
"entropy/max": 0.73046875,
|
|
"entropy/mean": 0.56640625,
|
|
"entropy/min": 0.310546875,
|
|
"epoch": 0.17578125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5410345196723938,
|
|
"kl": 0.05509365350008011,
|
|
"learning_rate": 9.578183043945031e-06,
|
|
"loss": 0.0022120936773717403,
|
|
"reward": 1.625002384185791,
|
|
"reward_std": 0.6609989404678345,
|
|
"rewards/DenseCaptionF1/mean": 0.5345268249511719,
|
|
"rewards/DenseCaptionF1/std": 0.3231217563152313,
|
|
"rewards/DenseCaptionSodaM/mean": 0.32485055923461914,
|
|
"rewards/DenseCaptionSodaM/std": 0.2140761762857437,
|
|
"rewards/TiemstampCaptionLength/mean": 0.765625,
|
|
"rewards/TiemstampCaptionLength/std": 0.42695629596710205,
|
|
"rewards/TimestampFormat/mean": 0.765625,
|
|
"rewards/TimestampFormat/std": 0.42695629596710205,
|
|
"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": 3824.0,
|
|
"completions/mean_length": 2272.515625,
|
|
"completions/min_length": 706.0,
|
|
"entropy/max": 0.82421875,
|
|
"entropy/mean": 0.55859375,
|
|
"entropy/min": 0.376953125,
|
|
"epoch": 0.1796875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45778459310531616,
|
|
"kl": 0.0554184764623642,
|
|
"learning_rate": 9.551814704830734e-06,
|
|
"loss": 0.002217470668256283,
|
|
"reward": 1.5976332426071167,
|
|
"reward_std": 0.651961088180542,
|
|
"rewards/DenseCaptionF1/mean": 0.470289409160614,
|
|
"rewards/DenseCaptionF1/std": 0.2826341390609741,
|
|
"rewards/DenseCaptionSodaM/mean": 0.3460938334465027,
|
|
"rewards/DenseCaptionSodaM/std": 0.22078247368335724,
|
|
"rewards/TiemstampCaptionLength/mean": 0.78125,
|
|
"rewards/TiemstampCaptionLength/std": 0.4166666865348816,
|
|
"rewards/TimestampFormat/mean": 0.78125,
|
|
"rewards/TimestampFormat/std": 0.4166666865348816,
|
|
"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": 0.015625,
|
|
"completions/max_length": 9216.0,
|
|
"completions/mean_length": 2133.4375,
|
|
"completions/min_length": 886.0,
|
|
"entropy/max": 0.79296875,
|
|
"entropy/mean": 0.57421875,
|
|
"entropy/min": 0.396484375,
|
|
"epoch": 0.18359375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.495386004447937,
|
|
"kl": 0.06056925281882286,
|
|
"learning_rate": 9.524685574090627e-06,
|
|
"loss": 0.002464357763528824,
|
|
"reward": 1.3884316682815552,
|
|
"reward_std": 0.8333114385604858,
|
|
"rewards/DenseCaptionF1/mean": 0.41399046778678894,
|
|
"rewards/DenseCaptionF1/std": 0.31623101234436035,
|
|
"rewards/DenseCaptionSodaM/mean": 0.2882433235645294,
|
|
"rewards/DenseCaptionSodaM/std": 0.20920830965042114,
|
|
"rewards/TiemstampCaptionLength/mean": 0.6848958134651184,
|
|
"rewards/TiemstampCaptionLength/std": 0.4658696949481964,
|
|
"rewards/TimestampFormat/mean": 0.6875,
|
|
"rewards/TimestampFormat/std": 0.467176616191864,
|
|
"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": 3831.0,
|
|
"completions/mean_length": 1907.34375,
|
|
"completions/min_length": 650.0,
|
|
"entropy/max": 0.734375,
|
|
"entropy/mean": 0.55078125,
|
|
"entropy/min": 0.408203125,
|
|
"epoch": 0.1875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5320143103599548,
|
|
"kl": 0.06790857017040253,
|
|
"learning_rate": 9.496800186095466e-06,
|
|
"loss": 0.002733142115175724,
|
|
"reward": 1.5999373197555542,
|
|
"reward_std": 0.7917318344116211,
|
|
"rewards/DenseCaptionF1/mean": 0.45771533250808716,
|
|
"rewards/DenseCaptionF1/std": 0.3045267164707184,
|
|
"rewards/DenseCaptionSodaM/mean": 0.3453470468521118,
|
|
"rewards/DenseCaptionSodaM/std": 0.21235060691833496,
|
|
"rewards/TiemstampCaptionLength/mean": 0.796875,
|
|
"rewards/TiemstampCaptionLength/std": 0.40550529956817627,
|
|
"rewards/TimestampFormat/mean": 0.796875,
|
|
"rewards/TimestampFormat/std": 0.40550529956817627,
|
|
"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": 0.015625,
|
|
"completions/max_length": 9216.0,
|
|
"completions/mean_length": 2016.375,
|
|
"completions/min_length": 825.0,
|
|
"entropy/max": 0.8828125,
|
|
"entropy/mean": 0.69921875,
|
|
"entropy/min": 0.130859375,
|
|
"epoch": 0.19140625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5382195711135864,
|
|
"kl": 0.06226423755288124,
|
|
"learning_rate": 9.468163201617063e-06,
|
|
"loss": 0.002524332143366337,
|
|
"reward": 1.8110225200653076,
|
|
"reward_std": 0.4265463352203369,
|
|
"rewards/DenseCaptionF1/mean": 0.5875532627105713,
|
|
"rewards/DenseCaptionF1/std": 0.26714834570884705,
|
|
"rewards/DenseCaptionSodaM/mean": 0.35042253136634827,
|
|
"rewards/DenseCaptionSodaM/std": 0.16498103737831116,
|
|
"rewards/TiemstampCaptionLength/mean": 0.87109375,
|
|
"rewards/TiemstampCaptionLength/std": 0.3333100974559784,
|
|
"rewards/TimestampFormat/mean": 0.875,
|
|
"rewards/TimestampFormat/std": 0.3333333432674408,
|
|
"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": 4920.0,
|
|
"completions/mean_length": 2190.953125,
|
|
"completions/min_length": 965.0,
|
|
"entropy/max": 0.94921875,
|
|
"entropy/mean": 0.66796875,
|
|
"entropy/min": 0.4453125,
|
|
"epoch": 0.1953125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5177400708198547,
|
|
"kl": 0.06634458899497986,
|
|
"learning_rate": 9.438779407049282e-06,
|
|
"loss": 0.002665899693965912,
|
|
"reward": 1.6761609315872192,
|
|
"reward_std": 0.7608978748321533,
|
|
"rewards/DenseCaptionF1/mean": 0.530105471611023,
|
|
"rewards/DenseCaptionF1/std": 0.29278987646102905,
|
|
"rewards/DenseCaptionSodaM/mean": 0.35648149251937866,
|
|
"rewards/DenseCaptionSodaM/std": 0.19881953299045563,
|
|
"rewards/TiemstampCaptionLength/mean": 0.7822730541229248,
|
|
"rewards/TiemstampCaptionLength/std": 0.40720799565315247,
|
|
"rewards/TimestampFormat/mean": 0.796875,
|
|
"rewards/TimestampFormat/std": 0.40550529956817627,
|
|
"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": 0.015625,
|
|
"completions/max_length": 9216.0,
|
|
"completions/mean_length": 2464.9375,
|
|
"completions/min_length": 989.0,
|
|
"entropy/max": 0.96875,
|
|
"entropy/mean": 0.68359375,
|
|
"entropy/min": 0.1298828125,
|
|
"epoch": 0.19921875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4651657044887543,
|
|
"kl": 0.06646151840686798,
|
|
"learning_rate": 9.40865371360804e-06,
|
|
"loss": 0.002682502381503582,
|
|
"reward": 1.781751036643982,
|
|
"reward_std": 0.5297815799713135,
|
|
"rewards/DenseCaptionF1/mean": 0.5445922613143921,
|
|
"rewards/DenseCaptionF1/std": 0.24131445586681366,
|
|
"rewards/DenseCaptionSodaM/mean": 0.3472440540790558,
|
|
"rewards/DenseCaptionSodaM/std": 0.14093098044395447,
|
|
"rewards/TiemstampCaptionLength/mean": 0.8892045617103577,
|
|
"rewards/TiemstampCaptionLength/std": 0.31427985429763794,
|
|
"rewards/TimestampFormat/mean": 0.890625,
|
|
"rewards/TimestampFormat/std": 0.3145764470100403,
|
|
"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": 0.03125,
|
|
"completions/max_length": 9216.0,
|
|
"completions/mean_length": 2706.375,
|
|
"completions/min_length": 759.0,
|
|
"entropy/max": 0.90234375,
|
|
"entropy/mean": 0.6328125,
|
|
"entropy/min": 0.02294921875,
|
|
"epoch": 0.203125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.442674845457077,
|
|
"kl": 0.06238989159464836,
|
|
"learning_rate": 9.377791156510456e-06,
|
|
"loss": 0.002551492303609848,
|
|
"reward": 1.8445196151733398,
|
|
"reward_std": 0.3831322193145752,
|
|
"rewards/DenseCaptionF1/mean": 0.5301719903945923,
|
|
"rewards/DenseCaptionF1/std": 0.19613783061504364,
|
|
"rewards/DenseCaptionSodaM/mean": 0.3938821852207184,
|
|
"rewards/DenseCaptionSodaM/std": 0.14461737871170044,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9190559387207031,
|
|
"rewards/TiemstampCaptionLength/std": 0.2701304256916046,
|
|
"rewards/TimestampFormat/mean": 0.921875,
|
|
"rewards/TimestampFormat/std": 0.27048972249031067,
|
|
"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": 4269.0,
|
|
"completions/mean_length": 2093.890625,
|
|
"completions/min_length": 672.0,
|
|
"entropy/max": 0.97265625,
|
|
"entropy/mean": 0.69140625,
|
|
"entropy/min": 0.478515625,
|
|
"epoch": 0.20703125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5645370483398438,
|
|
"kl": 0.07713831961154938,
|
|
"learning_rate": 9.346196894133239e-06,
|
|
"loss": 0.003090888261795044,
|
|
"reward": 1.8269987106323242,
|
|
"reward_std": 0.6235939264297485,
|
|
"rewards/DenseCaptionF1/mean": 0.5599989891052246,
|
|
"rewards/DenseCaptionF1/std": 0.2531862258911133,
|
|
"rewards/DenseCaptionSodaM/mean": 0.37865591049194336,
|
|
"rewards/DenseCaptionSodaM/std": 0.15898536145687103,
|
|
"rewards/TiemstampCaptionLength/mean": 0.8860626220703125,
|
|
"rewards/TiemstampCaptionLength/std": 0.3145526647567749,
|
|
"rewards/TimestampFormat/mean": 0.890625,
|
|
"rewards/TimestampFormat/std": 0.3145764470100403,
|
|
"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": 4122.0,
|
|
"completions/mean_length": 2167.671875,
|
|
"completions/min_length": 780.0,
|
|
"entropy/max": 0.9296875,
|
|
"entropy/mean": 0.62890625,
|
|
"entropy/min": 0.419921875,
|
|
"epoch": 0.2109375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5097282528877258,
|
|
"kl": 0.06437567621469498,
|
|
"learning_rate": 9.313876207150544e-06,
|
|
"loss": 0.002588218078017235,
|
|
"reward": 1.8655548095703125,
|
|
"reward_std": 0.37404483556747437,
|
|
"rewards/DenseCaptionF1/mean": 0.5587145090103149,
|
|
"rewards/DenseCaptionF1/std": 0.19311966001987457,
|
|
"rewards/DenseCaptionSodaM/mean": 0.36938798427581787,
|
|
"rewards/DenseCaptionSodaM/std": 0.14692431688308716,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9374046325683594,
|
|
"rewards/TiemstampCaptionLength/std": 0.24395141005516052,
|
|
"rewards/TimestampFormat/mean": 0.9375,
|
|
"rewards/TimestampFormat/std": 0.24397502839565277,
|
|
"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": 3360.0,
|
|
"completions/mean_length": 1985.75,
|
|
"completions/min_length": 785.0,
|
|
"entropy/max": 0.890625,
|
|
"entropy/mean": 0.640625,
|
|
"entropy/min": 0.51171875,
|
|
"epoch": 0.21484375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5162221193313599,
|
|
"kl": 0.07181726396083832,
|
|
"learning_rate": 9.280834497651334e-06,
|
|
"loss": 0.0028788154013454914,
|
|
"reward": 2.0226664543151855,
|
|
"reward_std": 0.4651912450790405,
|
|
"rewards/DenseCaptionF1/mean": 0.6559849381446838,
|
|
"rewards/DenseCaptionF1/std": 0.23934565484523773,
|
|
"rewards/DenseCaptionSodaM/mean": 0.43699392676353455,
|
|
"rewards/DenseCaptionSodaM/std": 0.14905153214931488,
|
|
"rewards/TiemstampCaptionLength/mean": 0.921875,
|
|
"rewards/TiemstampCaptionLength/std": 0.25539806485176086,
|
|
"rewards/TimestampFormat/mean": 0.9375,
|
|
"rewards/TimestampFormat/std": 0.24397502839565277,
|
|
"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": 3461.0,
|
|
"completions/mean_length": 1875.28125,
|
|
"completions/min_length": 538.0,
|
|
"entropy/max": 0.8984375,
|
|
"entropy/mean": 0.68359375,
|
|
"entropy/min": 0.3984375,
|
|
"epoch": 0.21875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5501840114593506,
|
|
"kl": 0.08197345584630966,
|
|
"learning_rate": 9.247077288236488e-06,
|
|
"loss": 0.0032946080900728703,
|
|
"reward": 2.073521137237549,
|
|
"reward_std": 0.21284720301628113,
|
|
"rewards/DenseCaptionF1/mean": 0.6387024521827698,
|
|
"rewards/DenseCaptionF1/std": 0.1778193861246109,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4598187208175659,
|
|
"rewards/DenseCaptionSodaM/std": 0.13882185518741608,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9656250476837158,
|
|
"rewards/TiemstampCaptionLength/std": 0.1404162049293518,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4774.0,
|
|
"completions/mean_length": 2623.734375,
|
|
"completions/min_length": 1428.0,
|
|
"entropy/max": 0.953125,
|
|
"entropy/mean": 0.68359375,
|
|
"entropy/min": 0.498046875,
|
|
"epoch": 0.22265625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.43559393286705017,
|
|
"kl": 0.06752844899892807,
|
|
"learning_rate": 9.212610221095748e-06,
|
|
"loss": 0.0027080695144832134,
|
|
"reward": 1.931420087814331,
|
|
"reward_std": 0.3583586513996124,
|
|
"rewards/DenseCaptionF1/mean": 0.575497031211853,
|
|
"rewards/DenseCaptionF1/std": 0.21436206996440887,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4054606258869171,
|
|
"rewards/DenseCaptionSodaM/std": 0.12318224459886551,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9477996826171875,
|
|
"rewards/TiemstampCaptionLength/std": 0.21343040466308594,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 4218.0,
|
|
"completions/mean_length": 2411.5625,
|
|
"completions/min_length": 890.0,
|
|
"entropy/max": 1.0625,
|
|
"entropy/mean": 0.7109375,
|
|
"entropy/min": 0.56640625,
|
|
"epoch": 0.2265625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5348519682884216,
|
|
"kl": 0.0774393305182457,
|
|
"learning_rate": 9.177439057064684e-06,
|
|
"loss": 0.0031305921729654074,
|
|
"reward": 2.0812172889709473,
|
|
"reward_std": 0.22512415051460266,
|
|
"rewards/DenseCaptionF1/mean": 0.6437010765075684,
|
|
"rewards/DenseCaptionF1/std": 0.17044788599014282,
|
|
"rewards/DenseCaptionSodaM/mean": 0.45336809754371643,
|
|
"rewards/DenseCaptionSodaM/std": 0.11320245265960693,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9839210510253906,
|
|
"rewards/TiemstampCaptionLength/std": 0.12499511241912842,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4333.0,
|
|
"completions/mean_length": 2394.359375,
|
|
"completions/min_length": 1129.0,
|
|
"entropy/max": 0.90625,
|
|
"entropy/mean": 0.67578125,
|
|
"entropy/min": 0.478515625,
|
|
"epoch": 0.23046875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.47797057032585144,
|
|
"kl": 0.06732644885778427,
|
|
"learning_rate": 9.141569674661816e-06,
|
|
"loss": 0.0026968615129590034,
|
|
"reward": 1.9791253805160522,
|
|
"reward_std": 0.375088095664978,
|
|
"rewards/DenseCaptionF1/mean": 0.6192713379859924,
|
|
"rewards/DenseCaptionF1/std": 0.2090216875076294,
|
|
"rewards/DenseCaptionSodaM/mean": 0.40720587968826294,
|
|
"rewards/DenseCaptionSodaM/std": 0.1373559832572937,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9521713256835938,
|
|
"rewards/TiemstampCaptionLength/std": 0.21295078098773956,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 0.015625,
|
|
"completions/max_length": 9216.0,
|
|
"completions/mean_length": 2370.875,
|
|
"completions/min_length": 1364.0,
|
|
"entropy/max": 1.0703125,
|
|
"entropy/mean": 0.6640625,
|
|
"entropy/min": 0.1572265625,
|
|
"epoch": 0.234375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.44017094373703003,
|
|
"kl": 0.07527967542409897,
|
|
"learning_rate": 9.105008069106093e-06,
|
|
"loss": 0.0030458117835223675,
|
|
"reward": 2.054739475250244,
|
|
"reward_std": 0.28885501623153687,
|
|
"rewards/DenseCaptionF1/mean": 0.6691363453865051,
|
|
"rewards/DenseCaptionF1/std": 0.1983274221420288,
|
|
"rewards/DenseCaptionSodaM/mean": 0.41685330867767334,
|
|
"rewards/DenseCaptionSodaM/std": 0.1314084678888321,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3686.0,
|
|
"completions/mean_length": 2361.875,
|
|
"completions/min_length": 863.0,
|
|
"entropy/max": 0.7734375,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.3359375,
|
|
"epoch": 0.23828125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4437764883041382,
|
|
"kl": 0.07143499702215195,
|
|
"learning_rate": 9.067760351314838e-06,
|
|
"loss": 0.002866515889763832,
|
|
"reward": 2.105388641357422,
|
|
"reward_std": 0.20455285906791687,
|
|
"rewards/DenseCaptionF1/mean": 0.6408991813659668,
|
|
"rewards/DenseCaptionF1/std": 0.13143233954906464,
|
|
"rewards/DenseCaptionSodaM/mean": 0.48089566826820374,
|
|
"rewards/DenseCaptionSodaM/std": 0.11053228378295898,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9828125238418579,
|
|
"rewards/TiemstampCaptionLength/std": 0.12542586028575897,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 0.015625,
|
|
"completions/max_length": 9216.0,
|
|
"completions/mean_length": 1774.953125,
|
|
"completions/min_length": 279.0,
|
|
"entropy/max": 0.83203125,
|
|
"entropy/mean": 0.609375,
|
|
"entropy/min": 0.1376953125,
|
|
"epoch": 0.2421875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5490212440490723,
|
|
"kl": 0.08521997928619385,
|
|
"learning_rate": 9.029832746882372e-06,
|
|
"loss": 0.0035130027681589127,
|
|
"reward": 1.8784807920455933,
|
|
"reward_std": 0.3403272032737732,
|
|
"rewards/DenseCaptionF1/mean": 0.5604114532470703,
|
|
"rewards/DenseCaptionF1/std": 0.2598591148853302,
|
|
"rewards/DenseCaptionSodaM/mean": 0.35758763551712036,
|
|
"rewards/DenseCaptionSodaM/std": 0.13501794636249542,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9522132873535156,
|
|
"rewards/TiemstampCaptionLength/std": 0.21292772889137268,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3683.0,
|
|
"completions/mean_length": 1927.125,
|
|
"completions/min_length": 352.0,
|
|
"entropy/max": 0.9765625,
|
|
"entropy/mean": 0.6328125,
|
|
"entropy/min": 0.369140625,
|
|
"epoch": 0.24609375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5698288083076477,
|
|
"kl": 0.07601820677518845,
|
|
"learning_rate": 8.991231595039464e-06,
|
|
"loss": 0.003107120282948017,
|
|
"reward": 2.029665470123291,
|
|
"reward_std": 0.31099215149879456,
|
|
"rewards/DenseCaptionF1/mean": 0.6040011644363403,
|
|
"rewards/DenseCaptionF1/std": 0.23040224611759186,
|
|
"rewards/DenseCaptionSodaM/mean": 0.45886746048927307,
|
|
"rewards/DenseCaptionSodaM/std": 0.15016283094882965,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96484375,
|
|
"rewards/TiemstampCaptionLength/std": 0.17743313312530518,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 4567.0,
|
|
"completions/mean_length": 2457.421875,
|
|
"completions/min_length": 1296.0,
|
|
"entropy/max": 0.83984375,
|
|
"entropy/mean": 0.66015625,
|
|
"entropy/min": 0.5703125,
|
|
"epoch": 0.25,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.449542760848999,
|
|
"kl": 0.07092197239398956,
|
|
"learning_rate": 8.951963347593797e-06,
|
|
"loss": 0.0028461660258471966,
|
|
"reward": 1.8207156658172607,
|
|
"reward_std": 0.4419528841972351,
|
|
"rewards/DenseCaptionF1/mean": 0.5536099672317505,
|
|
"rewards/DenseCaptionF1/std": 0.237579807639122,
|
|
"rewards/DenseCaptionSodaM/mean": 0.362589567899704,
|
|
"rewards/DenseCaptionSodaM/std": 0.14053674042224884,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9027824401855469,
|
|
"rewards/TiemstampCaptionLength/std": 0.2932094633579254,
|
|
"rewards/TimestampFormat/mean": 0.90625,
|
|
"rewards/TimestampFormat/std": 0.29378482699394226,
|
|
"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": 3570.0,
|
|
"completions/mean_length": 2080.3125,
|
|
"completions/min_length": 927.0,
|
|
"entropy/max": 0.96875,
|
|
"entropy/mean": 0.66015625,
|
|
"entropy/min": 0.4609375,
|
|
"epoch": 0.25390625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5001714825630188,
|
|
"kl": 0.07675490528345108,
|
|
"learning_rate": 8.9120345678516e-06,
|
|
"loss": 0.003085065633058548,
|
|
"reward": 1.954641580581665,
|
|
"reward_std": 0.32957372069358826,
|
|
"rewards/DenseCaptionF1/mean": 0.6267664432525635,
|
|
"rewards/DenseCaptionF1/std": 0.21088743209838867,
|
|
"rewards/DenseCaptionSodaM/mean": 0.37475019693374634,
|
|
"rewards/DenseCaptionSodaM/std": 0.12253134697675705,
|
|
"rewards/TiemstampCaptionLength/mean": 0.953125,
|
|
"rewards/TiemstampCaptionLength/std": 0.21304203569889069,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 4629.0,
|
|
"completions/mean_length": 2415.90625,
|
|
"completions/min_length": 1047.0,
|
|
"entropy/max": 0.83984375,
|
|
"entropy/mean": 0.65625,
|
|
"entropy/min": 0.53515625,
|
|
"epoch": 0.2578125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.46256938576698303,
|
|
"kl": 0.08071467280387878,
|
|
"learning_rate": 8.871451929520662e-06,
|
|
"loss": 0.003244481049478054,
|
|
"reward": 2.0497398376464844,
|
|
"reward_std": 0.12369006872177124,
|
|
"rewards/DenseCaptionF1/mean": 0.6514697670936584,
|
|
"rewards/DenseCaptionF1/std": 0.14243650436401367,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4017102122306824,
|
|
"rewards/DenseCaptionSodaM/std": 0.07843148708343506,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9931193590164185,
|
|
"rewards/TiemstampCaptionLength/std": 0.032912302762269974,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 4181.0,
|
|
"completions/mean_length": 2213.46875,
|
|
"completions/min_length": 1139.0,
|
|
"entropy/max": 0.8046875,
|
|
"entropy/mean": 0.609375,
|
|
"entropy/min": 0.3671875,
|
|
"epoch": 0.26171875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4749186038970947,
|
|
"kl": 0.0748860090970993,
|
|
"learning_rate": 8.83022221559489e-06,
|
|
"loss": 0.003019839059561491,
|
|
"reward": 2.031412124633789,
|
|
"reward_std": 0.3107888102531433,
|
|
"rewards/DenseCaptionF1/mean": 0.65138179063797,
|
|
"rewards/DenseCaptionF1/std": 0.1926232874393463,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4115608334541321,
|
|
"rewards/DenseCaptionSodaM/std": 0.10604707151651382,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9681892395019531,
|
|
"rewards/TiemstampCaptionLength/std": 0.1752953976392746,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3339.0,
|
|
"completions/mean_length": 1957.53125,
|
|
"completions/min_length": 721.0,
|
|
"entropy/max": 0.7890625,
|
|
"entropy/mean": 0.53125,
|
|
"entropy/min": 0.310546875,
|
|
"epoch": 0.265625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4781399369239807,
|
|
"kl": 0.07253287732601166,
|
|
"learning_rate": 8.78835231722059e-06,
|
|
"loss": 0.002902414184063673,
|
|
"reward": 2.263742446899414,
|
|
"reward_std": 0.11524605751037598,
|
|
"rewards/DenseCaptionF1/mean": 0.7564222812652588,
|
|
"rewards/DenseCaptionF1/std": 0.16660180687904358,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5138305425643921,
|
|
"rewards/DenseCaptionSodaM/std": 0.14809079468250275,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9869791865348816,
|
|
"rewards/TiemstampCaptionLength/std": 0.07456332445144653,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 4310.0,
|
|
"completions/mean_length": 2461.234375,
|
|
"completions/min_length": 1231.0,
|
|
"entropy/max": 0.75390625,
|
|
"entropy/mean": 0.5625,
|
|
"entropy/min": 0.412109375,
|
|
"epoch": 0.26953125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.42899295687675476,
|
|
"kl": 0.07073007524013519,
|
|
"learning_rate": 8.74584923254468e-06,
|
|
"loss": 0.0028330343775451183,
|
|
"reward": 2.133619546890259,
|
|
"reward_std": 0.11275416612625122,
|
|
"rewards/DenseCaptionF1/mean": 0.683053731918335,
|
|
"rewards/DenseCaptionF1/std": 0.14080090820789337,
|
|
"rewards/DenseCaptionSodaM/mean": 0.45099109411239624,
|
|
"rewards/DenseCaptionSodaM/std": 0.1174091100692749,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9991493225097656,
|
|
"rewards/TiemstampCaptionLength/std": 0.0065025631338357925,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 4201.0,
|
|
"completions/mean_length": 2458.375,
|
|
"completions/min_length": 1269.0,
|
|
"entropy/max": 0.7890625,
|
|
"entropy/mean": 0.54296875,
|
|
"entropy/min": 0.390625,
|
|
"epoch": 0.2734375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4160049259662628,
|
|
"kl": 0.06441740691661835,
|
|
"learning_rate": 8.702720065545024e-06,
|
|
"loss": 0.0025962484069168568,
|
|
"reward": 2.0632200241088867,
|
|
"reward_std": 0.18613176047801971,
|
|
"rewards/DenseCaptionF1/mean": 0.6221563220024109,
|
|
"rewards/DenseCaptionF1/std": 0.17072495818138123,
|
|
"rewards/DenseCaptionSodaM/mean": 0.45693108439445496,
|
|
"rewards/DenseCaptionSodaM/std": 0.14009734988212585,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9838905334472656,
|
|
"rewards/TiemstampCaptionLength/std": 0.12498046457767487,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3662.0,
|
|
"completions/mean_length": 1988.75,
|
|
"completions/min_length": 703.0,
|
|
"entropy/max": 0.80078125,
|
|
"entropy/mean": 0.546875,
|
|
"entropy/min": 0.302734375,
|
|
"epoch": 0.27734375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4744465947151184,
|
|
"kl": 0.06727895885705948,
|
|
"learning_rate": 8.658972024843063e-06,
|
|
"loss": 0.0026999362744390965,
|
|
"reward": 2.0450263023376465,
|
|
"reward_std": 0.31568199396133423,
|
|
"rewards/DenseCaptionF1/mean": 0.6489371061325073,
|
|
"rewards/DenseCaptionF1/std": 0.20903915166854858,
|
|
"rewards/DenseCaptionSodaM/mean": 0.42799025774002075,
|
|
"rewards/DenseCaptionSodaM/std": 0.14429260790348053,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9674479365348816,
|
|
"rewards/TiemstampCaptionLength/std": 0.17544174194335938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 2955.0,
|
|
"completions/mean_length": 2171.765625,
|
|
"completions/min_length": 1291.0,
|
|
"entropy/max": 0.84375,
|
|
"entropy/mean": 0.50390625,
|
|
"entropy/min": 0.337890625,
|
|
"epoch": 0.28125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.41763022541999817,
|
|
"kl": 0.06876518577337265,
|
|
"learning_rate": 8.614612422498965e-06,
|
|
"loss": 0.0027553928084671497,
|
|
"reward": 2.162646532058716,
|
|
"reward_std": 0.1864587366580963,
|
|
"rewards/DenseCaptionF1/mean": 0.6895356178283691,
|
|
"rewards/DenseCaptionF1/std": 0.14416387677192688,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4887358844280243,
|
|
"rewards/DenseCaptionSodaM/std": 0.10739783942699432,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3092.0,
|
|
"completions/mean_length": 1879.375,
|
|
"completions/min_length": 714.0,
|
|
"entropy/max": 1.171875,
|
|
"entropy/mean": 0.625,
|
|
"entropy/min": 0.39453125,
|
|
"epoch": 0.28515625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5157968997955322,
|
|
"kl": 0.07641156017780304,
|
|
"learning_rate": 8.569648672789496e-06,
|
|
"loss": 0.0030631432309746742,
|
|
"reward": 2.1225454807281494,
|
|
"reward_std": 0.15138301253318787,
|
|
"rewards/DenseCaptionF1/mean": 0.6931498050689697,
|
|
"rewards/DenseCaptionF1/std": 0.1821095198392868,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4309581518173218,
|
|
"rewards/DenseCaptionSodaM/std": 0.08070362359285355,
|
|
"rewards/TiemstampCaptionLength/mean": 0.996874988079071,
|
|
"rewards/TiemstampCaptionLength/std": 0.02500000037252903,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3294.0,
|
|
"completions/mean_length": 2040.59375,
|
|
"completions/min_length": 702.0,
|
|
"entropy/max": 0.7890625,
|
|
"entropy/mean": 0.61328125,
|
|
"entropy/min": 0.421875,
|
|
"epoch": 0.2890625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4754391610622406,
|
|
"kl": 0.08198946714401245,
|
|
"learning_rate": 8.524088290968781e-06,
|
|
"loss": 0.003284250386059284,
|
|
"reward": 1.951261043548584,
|
|
"reward_std": 0.37935590744018555,
|
|
"rewards/DenseCaptionF1/mean": 0.6092362403869629,
|
|
"rewards/DenseCaptionF1/std": 0.22482743859291077,
|
|
"rewards/DenseCaptionSodaM/mean": 0.40452486276626587,
|
|
"rewards/DenseCaptionSodaM/std": 0.127100870013237,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9375,
|
|
"rewards/TiemstampCaptionLength/std": 0.24397502839565277,
|
|
"rewards/TimestampFormat/mean": 0.9375,
|
|
"rewards/TimestampFormat/std": 0.24397502839565277,
|
|
"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": 4295.0,
|
|
"completions/mean_length": 2238.734375,
|
|
"completions/min_length": 1171.0,
|
|
"entropy/max": 0.8828125,
|
|
"entropy/mean": 0.59375,
|
|
"entropy/min": 0.412109375,
|
|
"epoch": 0.29296875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.44021475315093994,
|
|
"kl": 0.07665577530860901,
|
|
"learning_rate": 8.477938892012209e-06,
|
|
"loss": 0.003067487385123968,
|
|
"reward": 1.9776023626327515,
|
|
"reward_std": 0.241469606757164,
|
|
"rewards/DenseCaptionF1/mean": 0.6269410252571106,
|
|
"rewards/DenseCaptionF1/std": 0.18612676858901978,
|
|
"rewards/DenseCaptionSodaM/mean": 0.3877677917480469,
|
|
"rewards/DenseCaptionSodaM/std": 0.13575981557369232,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9570369124412537,
|
|
"rewards/TiemstampCaptionLength/std": 0.1844019740819931,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 4222.0,
|
|
"completions/mean_length": 2058.890625,
|
|
"completions/min_length": 1063.0,
|
|
"entropy/max": 0.81640625,
|
|
"entropy/mean": 0.54296875,
|
|
"entropy/min": 0.31640625,
|
|
"epoch": 0.296875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45110055804252625,
|
|
"kl": 0.07034870237112045,
|
|
"learning_rate": 8.43120818934367e-06,
|
|
"loss": 0.002813545521348715,
|
|
"reward": 2.110698699951172,
|
|
"reward_std": 0.1968698799610138,
|
|
"rewards/DenseCaptionF1/mean": 0.7020023465156555,
|
|
"rewards/DenseCaptionF1/std": 0.15579208731651306,
|
|
"rewards/DenseCaptionSodaM/mean": 0.42824020981788635,
|
|
"rewards/DenseCaptionSodaM/std": 0.11006494611501694,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9765371084213257,
|
|
"rewards/TiemstampCaptionLength/std": 0.1290905475616455,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3763.0,
|
|
"completions/mean_length": 1922.3125,
|
|
"completions/min_length": 989.0,
|
|
"entropy/max": 0.796875,
|
|
"entropy/mean": 0.5625,
|
|
"entropy/min": 0.435546875,
|
|
"epoch": 0.30078125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.471979022026062,
|
|
"kl": 0.0769549310207367,
|
|
"learning_rate": 8.38390399354631e-06,
|
|
"loss": 0.003084153402596712,
|
|
"reward": 2.0593576431274414,
|
|
"reward_std": 0.3367141783237457,
|
|
"rewards/DenseCaptionF1/mean": 0.6514508128166199,
|
|
"rewards/DenseCaptionF1/std": 0.22580590844154358,
|
|
"rewards/DenseCaptionSodaM/mean": 0.454781711101532,
|
|
"rewards/DenseCaptionSodaM/std": 0.1670650690793991,
|
|
"rewards/TiemstampCaptionLength/mean": 0.953125,
|
|
"rewards/TiemstampCaptionLength/std": 0.21304203569889069,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 4218.0,
|
|
"completions/mean_length": 2163.421875,
|
|
"completions/min_length": 1350.0,
|
|
"entropy/max": 0.7421875,
|
|
"entropy/mean": 0.5390625,
|
|
"entropy/min": 0.333984375,
|
|
"epoch": 0.3046875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4647288918495178,
|
|
"kl": 0.076052725315094,
|
|
"learning_rate": 8.336034211057098e-06,
|
|
"loss": 0.003052730578929186,
|
|
"reward": 2.110725164413452,
|
|
"reward_std": 0.20956245064735413,
|
|
"rewards/DenseCaptionF1/mean": 0.6667389869689941,
|
|
"rewards/DenseCaptionF1/std": 0.18336373567581177,
|
|
"rewards/DenseCaptionSodaM/mean": 0.45984014868736267,
|
|
"rewards/DenseCaptionSodaM/std": 0.10147108137607574,
|
|
"rewards/TiemstampCaptionLength/mean": 0.983917236328125,
|
|
"rewards/TiemstampCaptionLength/std": 0.12499552220106125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4041.0,
|
|
"completions/mean_length": 2145.015625,
|
|
"completions/min_length": 942.0,
|
|
"entropy/max": 0.671875,
|
|
"entropy/mean": 0.52734375,
|
|
"entropy/min": 0.408203125,
|
|
"epoch": 0.30859375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4572528898715973,
|
|
"kl": 0.07140101492404938,
|
|
"learning_rate": 8.28760684284532e-06,
|
|
"loss": 0.0028764642775058746,
|
|
"reward": 2.0567917823791504,
|
|
"reward_std": 0.25641506910324097,
|
|
"rewards/DenseCaptionF1/mean": 0.6210831999778748,
|
|
"rewards/DenseCaptionF1/std": 0.1876806765794754,
|
|
"rewards/DenseCaptionSodaM/mean": 0.48258349299430847,
|
|
"rewards/DenseCaptionSodaM/std": 0.13528074324131012,
|
|
"rewards/TiemstampCaptionLength/mean": 0.953125,
|
|
"rewards/TiemstampCaptionLength/std": 0.21304203569889069,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 3495.0,
|
|
"completions/mean_length": 2219.125,
|
|
"completions/min_length": 1093.0,
|
|
"entropy/max": 0.81640625,
|
|
"entropy/mean": 0.5859375,
|
|
"entropy/min": 0.404296875,
|
|
"epoch": 0.3125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4375706911087036,
|
|
"kl": 0.08007296919822693,
|
|
"learning_rate": 8.238629983075296e-06,
|
|
"loss": 0.0032063666731119156,
|
|
"reward": 2.0028772354125977,
|
|
"reward_std": 0.19251595437526703,
|
|
"rewards/DenseCaptionF1/mean": 0.5738047361373901,
|
|
"rewards/DenseCaptionF1/std": 0.15946848690509796,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4459995925426483,
|
|
"rewards/DenseCaptionSodaM/std": 0.10410931706428528,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9817708134651184,
|
|
"rewards/TiemstampCaptionLength/std": 0.12639760971069336,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4308.0,
|
|
"completions/mean_length": 2390.5,
|
|
"completions/min_length": 923.0,
|
|
"entropy/max": 0.74609375,
|
|
"entropy/mean": 0.5390625,
|
|
"entropy/min": 0.3984375,
|
|
"epoch": 0.31640625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4354075789451599,
|
|
"kl": 0.06995181739330292,
|
|
"learning_rate": 8.18911181775353e-06,
|
|
"loss": 0.002805061638355255,
|
|
"reward": 2.0832748413085938,
|
|
"reward_std": 0.29736328125,
|
|
"rewards/DenseCaptionF1/mean": 0.6486762762069702,
|
|
"rewards/DenseCaptionF1/std": 0.17525248229503632,
|
|
"rewards/DenseCaptionSodaM/mean": 0.46625107526779175,
|
|
"rewards/DenseCaptionSodaM/std": 0.1081622987985611,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9679450988769531,
|
|
"rewards/TiemstampCaptionLength/std": 0.175339475274086,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 4986.0,
|
|
"completions/mean_length": 2003.109375,
|
|
"completions/min_length": 862.0,
|
|
"entropy/max": 0.76953125,
|
|
"entropy/mean": 0.61328125,
|
|
"entropy/min": 0.451171875,
|
|
"epoch": 0.3203125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5022936463356018,
|
|
"kl": 0.08349774777889252,
|
|
"learning_rate": 8.139060623360494e-06,
|
|
"loss": 0.0033505428582429886,
|
|
"reward": 2.0526070594787598,
|
|
"reward_std": 0.21284550428390503,
|
|
"rewards/DenseCaptionF1/mean": 0.6233960390090942,
|
|
"rewards/DenseCaptionF1/std": 0.1946212202310562,
|
|
"rewards/DenseCaptionSodaM/mean": 0.44978249073028564,
|
|
"rewards/DenseCaptionSodaM/std": 0.13011571764945984,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9744822382926941,
|
|
"rewards/TiemstampCaptionLength/std": 0.13103936612606049,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 0.0,
|
|
"completions/max_length": 3873.0,
|
|
"completions/mean_length": 2122.046875,
|
|
"completions/min_length": 704.0,
|
|
"entropy/max": 0.84765625,
|
|
"entropy/mean": 0.59375,
|
|
"entropy/min": 0.421875,
|
|
"epoch": 0.32421875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.46683013439178467,
|
|
"kl": 0.08198022097349167,
|
|
"learning_rate": 8.088484765467286e-06,
|
|
"loss": 0.0033003194257616997,
|
|
"reward": 2.23862886428833,
|
|
"reward_std": 0.21417886018753052,
|
|
"rewards/DenseCaptionF1/mean": 0.7569527626037598,
|
|
"rewards/DenseCaptionF1/std": 0.1854260116815567,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4982775151729584,
|
|
"rewards/DenseCaptionSodaM/std": 0.11217108368873596,
|
|
"rewards/TiemstampCaptionLength/mean": 0.982421875,
|
|
"rewards/TiemstampCaptionLength/std": 0.12572643160820007,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4125.0,
|
|
"completions/mean_length": 2011.546875,
|
|
"completions/min_length": 838.0,
|
|
"entropy/max": 0.8828125,
|
|
"entropy/mean": 0.6171875,
|
|
"entropy/min": 0.2890625,
|
|
"epoch": 0.328125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5108813643455505,
|
|
"kl": 0.08043907582759857,
|
|
"learning_rate": 8.037392697337418e-06,
|
|
"loss": 0.003254396840929985,
|
|
"reward": 2.064764976501465,
|
|
"reward_std": 0.13689574599266052,
|
|
"rewards/DenseCaptionF1/mean": 0.6362853050231934,
|
|
"rewards/DenseCaptionF1/std": 0.13021419942378998,
|
|
"rewards/DenseCaptionSodaM/mean": 0.42983517050743103,
|
|
"rewards/DenseCaptionSodaM/std": 0.12798793613910675,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9972890019416809,
|
|
"rewards/TiemstampCaptionLength/std": 0.020837295800447464,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3911.0,
|
|
"completions/mean_length": 2324.40625,
|
|
"completions/min_length": 1031.0,
|
|
"entropy/max": 0.87890625,
|
|
"entropy/mean": 0.62890625,
|
|
"entropy/min": 0.427734375,
|
|
"epoch": 0.33203125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4670025408267975,
|
|
"kl": 0.07925336062908173,
|
|
"learning_rate": 7.985792958513932e-06,
|
|
"loss": 0.003175109624862671,
|
|
"reward": 2.0984528064727783,
|
|
"reward_std": 0.23483282327651978,
|
|
"rewards/DenseCaptionF1/mean": 0.6740026473999023,
|
|
"rewards/DenseCaptionF1/std": 0.17317916452884674,
|
|
"rewards/DenseCaptionSodaM/mean": 0.440075159072876,
|
|
"rewards/DenseCaptionSodaM/std": 0.1310994029045105,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3151.0,
|
|
"completions/mean_length": 2096.640625,
|
|
"completions/min_length": 1305.0,
|
|
"entropy/max": 0.87890625,
|
|
"entropy/mean": 0.6328125,
|
|
"entropy/min": 0.39453125,
|
|
"epoch": 0.3359375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4730246365070343,
|
|
"kl": 0.07844884693622589,
|
|
"learning_rate": 7.93369417339209e-06,
|
|
"loss": 0.0031482032500207424,
|
|
"reward": 2.1352877616882324,
|
|
"reward_std": 0.09777705371379852,
|
|
"rewards/DenseCaptionF1/mean": 0.7098089456558228,
|
|
"rewards/DenseCaptionF1/std": 0.12179086357355118,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4286037087440491,
|
|
"rewards/DenseCaptionSodaM/std": 0.12594905495643616,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9937499761581421,
|
|
"rewards/TiemstampCaptionLength/std": 0.05000000074505806,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 0.015625,
|
|
"completions/max_length": 9216.0,
|
|
"completions/mean_length": 2228.828125,
|
|
"completions/min_length": 1055.0,
|
|
"entropy/max": 0.7890625,
|
|
"entropy/mean": 0.57421875,
|
|
"entropy/min": 0.044677734375,
|
|
"epoch": 0.33984375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4559997618198395,
|
|
"kl": 0.07476093620061874,
|
|
"learning_rate": 7.881105049777902e-06,
|
|
"loss": 0.0030210758559405804,
|
|
"reward": 2.0786261558532715,
|
|
"reward_std": 0.2752508223056793,
|
|
"rewards/DenseCaptionF1/mean": 0.6454404592514038,
|
|
"rewards/DenseCaptionF1/std": 0.17010398209095,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4657379388809204,
|
|
"rewards/DenseCaptionSodaM/std": 0.1222817599773407,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9661458134651184,
|
|
"rewards/TiemstampCaptionLength/std": 0.17613248527050018,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 4043.0,
|
|
"completions/mean_length": 2428.71875,
|
|
"completions/min_length": 1281.0,
|
|
"entropy/max": 0.80859375,
|
|
"entropy/mean": 0.62109375,
|
|
"entropy/min": 0.451171875,
|
|
"epoch": 0.34375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4348693788051605,
|
|
"kl": 0.08055365830659866,
|
|
"learning_rate": 7.828034377432694e-06,
|
|
"loss": 0.0032192887738347054,
|
|
"reward": 2.068355083465576,
|
|
"reward_std": 0.29523277282714844,
|
|
"rewards/DenseCaptionF1/mean": 0.6584528684616089,
|
|
"rewards/DenseCaptionF1/std": 0.19537685811519623,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4450583755970001,
|
|
"rewards/DenseCaptionSodaM/std": 0.11957451701164246,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9609375,
|
|
"rewards/TiemstampCaptionLength/std": 0.18000143766403198,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3186.0,
|
|
"completions/mean_length": 1970.84375,
|
|
"completions/min_length": 1282.0,
|
|
"entropy/max": 0.79296875,
|
|
"entropy/mean": 0.5703125,
|
|
"entropy/min": 0.40234375,
|
|
"epoch": 0.34765625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4701857268810272,
|
|
"kl": 0.07622550427913666,
|
|
"learning_rate": 7.774491026603985e-06,
|
|
"loss": 0.003049303777515888,
|
|
"reward": 2.1712543964385986,
|
|
"reward_std": 0.1918593943119049,
|
|
"rewards/DenseCaptionF1/mean": 0.6795837879180908,
|
|
"rewards/DenseCaptionF1/std": 0.13512182235717773,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5088581442832947,
|
|
"rewards/DenseCaptionSodaM/std": 0.10053939372301102,
|
|
"rewards/TiemstampCaptionLength/mean": 0.981249988079071,
|
|
"rewards/TiemstampCaptionLength/std": 0.12708577513694763,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3679.0,
|
|
"completions/mean_length": 2321.390625,
|
|
"completions/min_length": 969.0,
|
|
"entropy/max": 0.77734375,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.4609375,
|
|
"epoch": 0.3515625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4848327338695526,
|
|
"kl": 0.07753942906856537,
|
|
"learning_rate": 7.720483946542913e-06,
|
|
"loss": 0.0031137799378484488,
|
|
"reward": 2.0770950317382812,
|
|
"reward_std": 0.31855612993240356,
|
|
"rewards/DenseCaptionF1/mean": 0.6584451794624329,
|
|
"rewards/DenseCaptionF1/std": 0.1920374631881714,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4498997926712036,
|
|
"rewards/DenseCaptionSodaM/std": 0.12485680729150772,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 0.015625,
|
|
"completions/max_length": 9216.0,
|
|
"completions/mean_length": 2294.125,
|
|
"completions/min_length": 1317.0,
|
|
"entropy/max": 0.78125,
|
|
"entropy/mean": 0.609375,
|
|
"entropy/min": 0.1572265625,
|
|
"epoch": 0.35546875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4357386827468872,
|
|
"kl": 0.07910288125276566,
|
|
"learning_rate": 7.666022164008458e-06,
|
|
"loss": 0.0032052365131676197,
|
|
"reward": 2.1311159133911133,
|
|
"reward_std": 0.18095478415489197,
|
|
"rewards/DenseCaptionF1/mean": 0.6685265302658081,
|
|
"rewards/DenseCaptionF1/std": 0.1838243305683136,
|
|
"rewards/DenseCaptionSodaM/mean": 0.47953325510025024,
|
|
"rewards/DenseCaptionSodaM/std": 0.09859392791986465,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9817373752593994,
|
|
"rewards/TiemstampCaptionLength/std": 0.12643776834011078,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4702.0,
|
|
"completions/mean_length": 2567.328125,
|
|
"completions/min_length": 1335.0,
|
|
"entropy/max": 0.78515625,
|
|
"entropy/mean": 0.60546875,
|
|
"entropy/min": 0.478515625,
|
|
"epoch": 0.359375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4263266921043396,
|
|
"kl": 0.07881991565227509,
|
|
"learning_rate": 7.6111147817586925e-06,
|
|
"loss": 0.0031613344326615334,
|
|
"reward": 2.128614902496338,
|
|
"reward_std": 0.11585809290409088,
|
|
"rewards/DenseCaptionF1/mean": 0.6696683764457703,
|
|
"rewards/DenseCaptionF1/std": 0.1590665876865387,
|
|
"rewards/DenseCaptionSodaM/mean": 0.46010249853134155,
|
|
"rewards/DenseCaptionSodaM/std": 0.07637323439121246,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9976882934570312,
|
|
"rewards/TiemstampCaptionLength/std": 0.01849365234375,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3329.0,
|
|
"completions/mean_length": 2196.390625,
|
|
"completions/min_length": 1305.0,
|
|
"entropy/max": 0.78125,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.37890625,
|
|
"epoch": 0.36328125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.44893574714660645,
|
|
"kl": 0.0778866857290268,
|
|
"learning_rate": 7.5557709770293664e-06,
|
|
"loss": 0.003123055212199688,
|
|
"reward": 2.242305278778076,
|
|
"reward_std": 0.146044060587883,
|
|
"rewards/DenseCaptionF1/mean": 0.7382992506027222,
|
|
"rewards/DenseCaptionF1/std": 0.14254668354988098,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5162829160690308,
|
|
"rewards/DenseCaptionSodaM/std": 0.13311879336833954,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9754464626312256,
|
|
"rewards/TiemstampCaptionLength/std": 0.08280186355113983,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3506.0,
|
|
"completions/mean_length": 2073.765625,
|
|
"completions/min_length": 917.0,
|
|
"entropy/max": 0.91015625,
|
|
"entropy/mean": 0.59375,
|
|
"entropy/min": 0.478515625,
|
|
"epoch": 0.3671875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4797140955924988,
|
|
"kl": 0.07541140913963318,
|
|
"learning_rate": 7.500000000000001e-06,
|
|
"loss": 0.0030153607949614525,
|
|
"reward": 2.0866267681121826,
|
|
"reward_std": 0.23808912932872772,
|
|
"rewards/DenseCaptionF1/mean": 0.677537739276886,
|
|
"rewards/DenseCaptionF1/std": 0.20984452962875366,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4403391480445862,
|
|
"rewards/DenseCaptionSodaM/std": 0.10602159053087234,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 4109.0,
|
|
"completions/mean_length": 2264.859375,
|
|
"completions/min_length": 1144.0,
|
|
"entropy/max": 0.73046875,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.40234375,
|
|
"epoch": 0.37109375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4785752594470978,
|
|
"kl": 0.08089148998260498,
|
|
"learning_rate": 7.443811172247822e-06,
|
|
"loss": 0.0032392162829637527,
|
|
"reward": 2.1369659900665283,
|
|
"reward_std": 0.30026695132255554,
|
|
"rewards/DenseCaptionF1/mean": 0.6629904508590698,
|
|
"rewards/DenseCaptionF1/std": 0.18864844739437103,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5067400932312012,
|
|
"rewards/DenseCaptionSodaM/std": 0.12611261010169983,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9657212495803833,
|
|
"rewards/TiemstampCaptionLength/std": 0.17562004923820496,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3040.0,
|
|
"completions/mean_length": 2175.71875,
|
|
"completions/min_length": 1565.0,
|
|
"entropy/max": 0.78515625,
|
|
"entropy/mean": 0.63671875,
|
|
"entropy/min": 0.515625,
|
|
"epoch": 0.375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45520272850990295,
|
|
"kl": 0.09045030176639557,
|
|
"learning_rate": 7.387213885189746e-06,
|
|
"loss": 0.0036247065290808678,
|
|
"reward": 2.2175164222717285,
|
|
"reward_std": 0.20569007098674774,
|
|
"rewards/DenseCaptionF1/mean": 0.7261995673179626,
|
|
"rewards/DenseCaptionF1/std": 0.1654254049062729,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5082440972328186,
|
|
"rewards/DenseCaptionSodaM/std": 0.09253308922052383,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9817708134651184,
|
|
"rewards/TiemstampCaptionLength/std": 0.12639760971069336,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4051.0,
|
|
"completions/mean_length": 2542.875,
|
|
"completions/min_length": 986.0,
|
|
"entropy/max": 0.89453125,
|
|
"entropy/mean": 0.63671875,
|
|
"entropy/min": 0.447265625,
|
|
"epoch": 0.37890625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.43797022104263306,
|
|
"kl": 0.0741940289735794,
|
|
"learning_rate": 7.330217598512696e-06,
|
|
"loss": 0.0029716668650507927,
|
|
"reward": 2.086512327194214,
|
|
"reward_std": 0.23860614001750946,
|
|
"rewards/DenseCaptionF1/mean": 0.6499544978141785,
|
|
"rewards/DenseCaptionF1/std": 0.20542526245117188,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4521827697753906,
|
|
"rewards/DenseCaptionSodaM/std": 0.09262199699878693,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 2951.0,
|
|
"completions/mean_length": 2027.390625,
|
|
"completions/min_length": 807.0,
|
|
"entropy/max": 0.86328125,
|
|
"entropy/mean": 0.58984375,
|
|
"entropy/min": 0.390625,
|
|
"epoch": 0.3828125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4523578882217407,
|
|
"kl": 0.08195482194423676,
|
|
"learning_rate": 7.2728318385925035e-06,
|
|
"loss": 0.0032738554291427135,
|
|
"reward": 2.14823055267334,
|
|
"reward_std": 0.22445625066757202,
|
|
"rewards/DenseCaptionF1/mean": 0.6198608875274658,
|
|
"rewards/DenseCaptionF1/std": 0.17194251716136932,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5439946055412292,
|
|
"rewards/DenseCaptionSodaM/std": 0.12489025294780731,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4176.0,
|
|
"completions/mean_length": 2919.421875,
|
|
"completions/min_length": 1698.0,
|
|
"entropy/max": 0.90234375,
|
|
"entropy/mean": 0.60546875,
|
|
"entropy/min": 0.39453125,
|
|
"epoch": 0.38671875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.377668172121048,
|
|
"kl": 0.07640206813812256,
|
|
"learning_rate": 7.215066196901676e-06,
|
|
"loss": 0.00305812805891037,
|
|
"reward": 2.1023716926574707,
|
|
"reward_std": 0.17110249400138855,
|
|
"rewards/DenseCaptionF1/mean": 0.6405081152915955,
|
|
"rewards/DenseCaptionF1/std": 0.14213506877422333,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4818001985549927,
|
|
"rewards/DenseCaptionSodaM/std": 0.1384231597185135,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9757517576217651,
|
|
"rewards/TiemstampCaptionLength/std": 0.13296526670455933,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3982.0,
|
|
"completions/mean_length": 2404.359375,
|
|
"completions/min_length": 1510.0,
|
|
"entropy/max": 0.84765625,
|
|
"entropy/mean": 0.6328125,
|
|
"entropy/min": 0.47265625,
|
|
"epoch": 0.390625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.44833576679229736,
|
|
"kl": 0.07901492714881897,
|
|
"learning_rate": 7.156930328406268e-06,
|
|
"loss": 0.003162129782140255,
|
|
"reward": 2.113173007965088,
|
|
"reward_std": 0.10366939008235931,
|
|
"rewards/DenseCaptionF1/mean": 0.6593329310417175,
|
|
"rewards/DenseCaptionF1/std": 0.17583686113357544,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4554026424884796,
|
|
"rewards/DenseCaptionSodaM/std": 0.0941043421626091,
|
|
"rewards/TiemstampCaptionLength/mean": 0.996874988079071,
|
|
"rewards/TiemstampCaptionLength/std": 0.02500000037252903,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3739.0,
|
|
"completions/mean_length": 2175.03125,
|
|
"completions/min_length": 1281.0,
|
|
"entropy/max": 0.89453125,
|
|
"entropy/mean": 0.56640625,
|
|
"entropy/min": 0.40625,
|
|
"epoch": 0.39453125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45377618074417114,
|
|
"kl": 0.07455649226903915,
|
|
"learning_rate": 7.098433949952146e-06,
|
|
"loss": 0.00298325065523386,
|
|
"reward": 2.1517369747161865,
|
|
"reward_std": 0.2411295622587204,
|
|
"rewards/DenseCaptionF1/mean": 0.6895813941955566,
|
|
"rewards/DenseCaptionF1/std": 0.18673183023929596,
|
|
"rewards/DenseCaptionSodaM/mean": 0.49340564012527466,
|
|
"rewards/DenseCaptionSodaM/std": 0.10936973989009857,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 4237.0,
|
|
"completions/mean_length": 2150.0625,
|
|
"completions/min_length": 966.0,
|
|
"entropy/max": 0.70703125,
|
|
"entropy/mean": 0.55859375,
|
|
"entropy/min": 0.40234375,
|
|
"epoch": 0.3984375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.47996532917022705,
|
|
"kl": 0.08367238193750381,
|
|
"learning_rate": 7.039586838640918e-06,
|
|
"loss": 0.0033626602962613106,
|
|
"reward": 2.017162799835205,
|
|
"reward_std": 0.34179067611694336,
|
|
"rewards/DenseCaptionF1/mean": 0.6233688592910767,
|
|
"rewards/DenseCaptionF1/std": 0.22591255605220795,
|
|
"rewards/DenseCaptionSodaM/mean": 0.45874860882759094,
|
|
"rewards/DenseCaptionSodaM/std": 0.1617920696735382,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9325904846191406,
|
|
"rewards/TiemstampCaptionLength/std": 0.2447463572025299,
|
|
"rewards/TimestampFormat/mean": 0.9375,
|
|
"rewards/TimestampFormat/std": 0.24397502839565277,
|
|
"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": 4251.0,
|
|
"completions/mean_length": 2408.359375,
|
|
"completions/min_length": 984.0,
|
|
"entropy/max": 0.83984375,
|
|
"entropy/mean": 0.5703125,
|
|
"entropy/min": 0.3671875,
|
|
"epoch": 0.40234375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.46457499265670776,
|
|
"kl": 0.06997324526309967,
|
|
"learning_rate": 6.980398830195785e-06,
|
|
"loss": 0.0028030076064169407,
|
|
"reward": 2.2786498069763184,
|
|
"reward_std": 0.11027923226356506,
|
|
"rewards/DenseCaptionF1/mean": 0.7410256862640381,
|
|
"rewards/DenseCaptionF1/std": 0.16160745918750763,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5409583449363708,
|
|
"rewards/DenseCaptionSodaM/std": 0.07088367640972137,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9933311343193054,
|
|
"rewards/TiemstampCaptionLength/std": 0.0350026860833168,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3882.0,
|
|
"completions/mean_length": 2358.390625,
|
|
"completions/min_length": 1220.0,
|
|
"entropy/max": 0.6796875,
|
|
"entropy/mean": 0.55078125,
|
|
"entropy/min": 0.283203125,
|
|
"epoch": 0.40625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.412352979183197,
|
|
"kl": 0.07164886593818665,
|
|
"learning_rate": 6.920879817317588e-06,
|
|
"loss": 0.0028869202360510826,
|
|
"reward": 2.1052591800689697,
|
|
"reward_std": 0.19318059086799622,
|
|
"rewards/DenseCaptionF1/mean": 0.6278367042541504,
|
|
"rewards/DenseCaptionF1/std": 0.16896402835845947,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4961166977882385,
|
|
"rewards/DenseCaptionSodaM/std": 0.13075561821460724,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9782366156578064,
|
|
"rewards/TiemstampCaptionLength/std": 0.1292588710784912,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3634.0,
|
|
"completions/mean_length": 2189.46875,
|
|
"completions/min_length": 997.0,
|
|
"entropy/max": 0.8125,
|
|
"entropy/mean": 0.58984375,
|
|
"entropy/min": 0.400390625,
|
|
"epoch": 0.41015625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4769591987133026,
|
|
"kl": 0.07978053390979767,
|
|
"learning_rate": 6.861039748031351e-06,
|
|
"loss": 0.0031947405077517033,
|
|
"reward": 2.176485538482666,
|
|
"reward_std": 0.14867156744003296,
|
|
"rewards/DenseCaptionF1/mean": 0.7199992537498474,
|
|
"rewards/DenseCaptionF1/std": 0.15194834768772125,
|
|
"rewards/DenseCaptionSodaM/mean": 0.45648619532585144,
|
|
"rewards/DenseCaptionSodaM/std": 0.09429176896810532,
|
|
"rewards/TiemstampCaptionLength/mean": 1.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 0.015625,
|
|
"completions/max_length": 9216.0,
|
|
"completions/mean_length": 2370.90625,
|
|
"completions/min_length": 1174.0,
|
|
"entropy/max": 0.93359375,
|
|
"entropy/mean": 0.61328125,
|
|
"entropy/min": 0.453125,
|
|
"epoch": 0.4140625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45270076394081116,
|
|
"kl": 0.0827193409204483,
|
|
"learning_rate": 6.800888624023552e-06,
|
|
"loss": 0.0033212807029485703,
|
|
"reward": 2.1291093826293945,
|
|
"reward_std": 0.29201728105545044,
|
|
"rewards/DenseCaptionF1/mean": 0.679262638092041,
|
|
"rewards/DenseCaptionF1/std": 0.1974390596151352,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4819648861885071,
|
|
"rewards/DenseCaptionSodaM/std": 0.14247861504554749,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9670138955116272,
|
|
"rewards/TiemstampCaptionLength/std": 0.17560364305973053,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 0.015625,
|
|
"completions/max_length": 9216.0,
|
|
"completions/mean_length": 2507.890625,
|
|
"completions/min_length": 942.0,
|
|
"entropy/max": 0.82421875,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.2431640625,
|
|
"epoch": 0.41796875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4403754472732544,
|
|
"kl": 0.07606419920921326,
|
|
"learning_rate": 6.740436498970453e-06,
|
|
"loss": 0.0030570104718208313,
|
|
"reward": 2.0796656608581543,
|
|
"reward_std": 0.2234925925731659,
|
|
"rewards/DenseCaptionF1/mean": 0.6541966199874878,
|
|
"rewards/DenseCaptionF1/std": 0.21035099029541016,
|
|
"rewards/DenseCaptionSodaM/mean": 0.45958346128463745,
|
|
"rewards/DenseCaptionSodaM/std": 0.10667937248945236,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9630208015441895,
|
|
"rewards/TiemstampCaptionLength/std": 0.17729271948337555,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3815.0,
|
|
"completions/mean_length": 2447.953125,
|
|
"completions/min_length": 958.0,
|
|
"entropy/max": 0.84375,
|
|
"entropy/mean": 0.59375,
|
|
"entropy/min": 0.451171875,
|
|
"epoch": 0.421875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4195391535758972,
|
|
"kl": 0.075007364153862,
|
|
"learning_rate": 6.679693476857712e-06,
|
|
"loss": 0.0029998552054166794,
|
|
"reward": 2.121521472930908,
|
|
"reward_std": 0.19369511306285858,
|
|
"rewards/DenseCaptionF1/mean": 0.6594398617744446,
|
|
"rewards/DenseCaptionF1/std": 0.1779697984457016,
|
|
"rewards/DenseCaptionSodaM/mean": 0.47770658135414124,
|
|
"rewards/DenseCaptionSodaM/std": 0.13796070218086243,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 2951.0,
|
|
"completions/mean_length": 2098.671875,
|
|
"completions/min_length": 1053.0,
|
|
"entropy/max": 0.75,
|
|
"entropy/mean": 0.5546875,
|
|
"entropy/min": 0.2734375,
|
|
"epoch": 0.42578125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45432087779045105,
|
|
"kl": 0.07422572374343872,
|
|
"learning_rate": 6.618669710291607e-06,
|
|
"loss": 0.0029711276292800903,
|
|
"reward": 2.2597405910491943,
|
|
"reward_std": 0.10507716238498688,
|
|
"rewards/DenseCaptionF1/mean": 0.7595625519752502,
|
|
"rewards/DenseCaptionF1/std": 0.10846223682165146,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5001780986785889,
|
|
"rewards/DenseCaptionSodaM/std": 0.10454103350639343,
|
|
"rewards/TiemstampCaptionLength/mean": 1.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3640.0,
|
|
"completions/mean_length": 2099.5,
|
|
"completions/min_length": 1332.0,
|
|
"entropy/max": 0.765625,
|
|
"entropy/mean": 0.578125,
|
|
"entropy/min": 0.3984375,
|
|
"epoch": 0.4296875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4507025182247162,
|
|
"kl": 0.08030343800783157,
|
|
"learning_rate": 6.557375398802124e-06,
|
|
"loss": 0.0032133255153894424,
|
|
"reward": 2.052093029022217,
|
|
"reward_std": 0.4489860534667969,
|
|
"rewards/DenseCaptionF1/mean": 0.6731139421463013,
|
|
"rewards/DenseCaptionF1/std": 0.2409752607345581,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4456457197666168,
|
|
"rewards/DenseCaptionSodaM/std": 0.15500512719154358,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9291666746139526,
|
|
"rewards/TiemstampCaptionLength/std": 0.24656368792057037,
|
|
"rewards/TimestampFormat/mean": 0.9375,
|
|
"rewards/TimestampFormat/std": 0.24397502839565277,
|
|
"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": 4649.0,
|
|
"completions/mean_length": 2456.265625,
|
|
"completions/min_length": 984.0,
|
|
"entropy/max": 0.86328125,
|
|
"entropy/mean": 0.609375,
|
|
"entropy/min": 0.4375,
|
|
"epoch": 0.43359375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.43649446964263916,
|
|
"kl": 0.07889571785926819,
|
|
"learning_rate": 6.495820787138209e-06,
|
|
"loss": 0.003167032264173031,
|
|
"reward": 2.207371234893799,
|
|
"reward_std": 0.11937980353832245,
|
|
"rewards/DenseCaptionF1/mean": 0.7265190482139587,
|
|
"rewards/DenseCaptionF1/std": 0.1683252453804016,
|
|
"rewards/DenseCaptionSodaM/mean": 0.48906826972961426,
|
|
"rewards/DenseCaptionSodaM/std": 0.09366622567176819,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9835683107376099,
|
|
"rewards/TiemstampCaptionLength/std": 0.05088389292359352,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3276.0,
|
|
"completions/mean_length": 1968.25,
|
|
"completions/min_length": 1133.0,
|
|
"entropy/max": 0.94140625,
|
|
"entropy/mean": 0.62109375,
|
|
"entropy/min": 0.462890625,
|
|
"epoch": 0.4375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4778197109699249,
|
|
"kl": 0.08649688959121704,
|
|
"learning_rate": 6.434016163555452e-06,
|
|
"loss": 0.003462088294327259,
|
|
"reward": 2.2246928215026855,
|
|
"reward_std": 0.22315554320812225,
|
|
"rewards/DenseCaptionF1/mean": 0.7509918212890625,
|
|
"rewards/DenseCaptionF1/std": 0.1588038057088852,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4908885955810547,
|
|
"rewards/DenseCaptionSodaM/std": 0.10360200703144073,
|
|
"rewards/TiemstampCaptionLength/mean": 0.981249988079071,
|
|
"rewards/TiemstampCaptionLength/std": 0.12708577513694763,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4535.0,
|
|
"completions/mean_length": 2397.15625,
|
|
"completions/min_length": 985.0,
|
|
"entropy/max": 0.8359375,
|
|
"entropy/mean": 0.60546875,
|
|
"entropy/min": 0.44921875,
|
|
"epoch": 0.44140625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45221975445747375,
|
|
"kl": 0.07573950290679932,
|
|
"learning_rate": 6.371971858096509e-06,
|
|
"loss": 0.003038412658497691,
|
|
"reward": 2.1565451622009277,
|
|
"reward_std": 0.19371052086353302,
|
|
"rewards/DenseCaptionF1/mean": 0.6990419626235962,
|
|
"rewards/DenseCaptionF1/std": 0.19177736341953278,
|
|
"rewards/DenseCaptionSodaM/mean": 0.48057520389556885,
|
|
"rewards/DenseCaptionSodaM/std": 0.1356320083141327,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9694812297821045,
|
|
"rewards/TiemstampCaptionLength/std": 0.13477176427841187,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3674.0,
|
|
"completions/mean_length": 2573.8125,
|
|
"completions/min_length": 1192.0,
|
|
"entropy/max": 0.765625,
|
|
"entropy/mean": 0.609375,
|
|
"entropy/min": 0.4140625,
|
|
"epoch": 0.4453125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.41967564821243286,
|
|
"kl": 0.07641534507274628,
|
|
"learning_rate": 6.30969824086453e-06,
|
|
"loss": 0.0030584577471017838,
|
|
"reward": 2.154862403869629,
|
|
"reward_std": 0.10016033053398132,
|
|
"rewards/DenseCaptionF1/mean": 0.6651291847229004,
|
|
"rewards/DenseCaptionF1/std": 0.1349097192287445,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4897330105304718,
|
|
"rewards/DenseCaptionSodaM/std": 0.09691990911960602,
|
|
"rewards/TiemstampCaptionLength/mean": 1.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3362.0,
|
|
"completions/mean_length": 2246.75,
|
|
"completions/min_length": 740.0,
|
|
"entropy/max": 0.765625,
|
|
"entropy/mean": 0.625,
|
|
"entropy/min": 0.357421875,
|
|
"epoch": 0.44921875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4850248396396637,
|
|
"kl": 0.07770873606204987,
|
|
"learning_rate": 6.247205720289907e-06,
|
|
"loss": 0.0031074020080268383,
|
|
"reward": 2.105367422103882,
|
|
"reward_std": 0.2740570604801178,
|
|
"rewards/DenseCaptionF1/mean": 0.6702495813369751,
|
|
"rewards/DenseCaptionF1/std": 0.203532412648201,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4663679003715515,
|
|
"rewards/DenseCaptionSodaM/std": 0.13411854207515717,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3658.0,
|
|
"completions/mean_length": 2495.046875,
|
|
"completions/min_length": 1110.0,
|
|
"entropy/max": 0.796875,
|
|
"entropy/mean": 0.578125,
|
|
"entropy/min": 0.416015625,
|
|
"epoch": 0.453125,
|
|
"frac_reward_zero_std": 0.125,
|
|
"grad_norm": 0.39278051257133484,
|
|
"kl": 0.07527855038642883,
|
|
"learning_rate": 6.184504741390596e-06,
|
|
"loss": 0.003027956234291196,
|
|
"reward": 1.781051754951477,
|
|
"reward_std": 0.2833193838596344,
|
|
"rewards/DenseCaptionF1/mean": 0.5420864820480347,
|
|
"rewards/DenseCaptionF1/std": 0.2658858001232147,
|
|
"rewards/DenseCaptionSodaM/mean": 0.39781951904296875,
|
|
"rewards/DenseCaptionSodaM/std": 0.19784612953662872,
|
|
"rewards/TiemstampCaptionLength/mean": 0.83984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.3656023144721985,
|
|
"rewards/TimestampFormat/mean": 0.8424479365348816,
|
|
"rewards/TimestampFormat/std": 0.36554577946662903,
|
|
"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": 3945.0,
|
|
"completions/mean_length": 2073.890625,
|
|
"completions/min_length": 1240.0,
|
|
"entropy/max": 0.8046875,
|
|
"entropy/mean": 0.59375,
|
|
"entropy/min": 0.451171875,
|
|
"epoch": 0.45703125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.46272948384284973,
|
|
"kl": 0.0793612152338028,
|
|
"learning_rate": 6.121605784026339e-06,
|
|
"loss": 0.0031837821006774902,
|
|
"reward": 1.994809865951538,
|
|
"reward_std": 0.33968663215637207,
|
|
"rewards/DenseCaptionF1/mean": 0.5881167650222778,
|
|
"rewards/DenseCaptionF1/std": 0.2010265439748764,
|
|
"rewards/DenseCaptionSodaM/mean": 0.45356807112693787,
|
|
"rewards/DenseCaptionSodaM/std": 0.15971964597702026,
|
|
"rewards/TiemstampCaptionLength/mean": 0.953125,
|
|
"rewards/TiemstampCaptionLength/std": 0.21304203569889069,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 3265.0,
|
|
"completions/mean_length": 1820.046875,
|
|
"completions/min_length": 1171.0,
|
|
"entropy/max": 0.84765625,
|
|
"entropy/mean": 0.5859375,
|
|
"entropy/min": 0.421875,
|
|
"epoch": 0.4609375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.47077837586402893,
|
|
"kl": 0.08538872003555298,
|
|
"learning_rate": 6.058519361147055e-06,
|
|
"loss": 0.0034187142737209797,
|
|
"reward": 2.0582194328308105,
|
|
"reward_std": 0.2934507727622986,
|
|
"rewards/DenseCaptionF1/mean": 0.615313708782196,
|
|
"rewards/DenseCaptionF1/std": 0.20698460936546326,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4941520094871521,
|
|
"rewards/DenseCaptionSodaM/std": 0.1554199457168579,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9443824291229248,
|
|
"rewards/TiemstampCaptionLength/std": 0.21501989662647247,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 3361.0,
|
|
"completions/mean_length": 2210.484375,
|
|
"completions/min_length": 1170.0,
|
|
"entropy/max": 0.83203125,
|
|
"entropy/mean": 0.6484375,
|
|
"entropy/min": 0.38671875,
|
|
"epoch": 0.46484375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45301973819732666,
|
|
"kl": 0.0787690132856369,
|
|
"learning_rate": 5.995256017035703e-06,
|
|
"loss": 0.0031627845019102097,
|
|
"reward": 1.9711133241653442,
|
|
"reward_std": 0.27808910608291626,
|
|
"rewards/DenseCaptionF1/mean": 0.5771324634552002,
|
|
"rewards/DenseCaptionF1/std": 0.22329936921596527,
|
|
"rewards/DenseCaptionSodaM/mean": 0.42523080110549927,
|
|
"rewards/DenseCaptionSodaM/std": 0.1310344636440277,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3498.0,
|
|
"completions/mean_length": 2194.875,
|
|
"completions/min_length": 1333.0,
|
|
"entropy/max": 0.80078125,
|
|
"entropy/mean": 0.61328125,
|
|
"entropy/min": 0.486328125,
|
|
"epoch": 0.46875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45540276169776917,
|
|
"kl": 0.07793214917182922,
|
|
"learning_rate": 5.931826325545912e-06,
|
|
"loss": 0.0031153373420238495,
|
|
"reward": 2.0628232955932617,
|
|
"reward_std": 0.45016881823539734,
|
|
"rewards/DenseCaptionF1/mean": 0.6490108966827393,
|
|
"rewards/DenseCaptionF1/std": 0.20744843780994415,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4774283170700073,
|
|
"rewards/DenseCaptionSodaM/std": 0.15979807078838348,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9352678656578064,
|
|
"rewards/TiemstampCaptionLength/std": 0.2440476268529892,
|
|
"rewards/TimestampFormat/mean": 0.9375,
|
|
"rewards/TimestampFormat/std": 0.24397502839565277,
|
|
"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": 3085.0,
|
|
"completions/mean_length": 1896.515625,
|
|
"completions/min_length": 533.0,
|
|
"entropy/max": 1.0703125,
|
|
"entropy/mean": 0.61328125,
|
|
"entropy/min": 0.341796875,
|
|
"epoch": 0.47265625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5396003127098083,
|
|
"kl": 0.07848010212182999,
|
|
"learning_rate": 5.8682408883346535e-06,
|
|
"loss": 0.0031436989083886147,
|
|
"reward": 2.1525871753692627,
|
|
"reward_std": 0.28493666648864746,
|
|
"rewards/DenseCaptionF1/mean": 0.7212038636207581,
|
|
"rewards/DenseCaptionF1/std": 0.2505525052547455,
|
|
"rewards/DenseCaptionSodaM/mean": 0.46728357672691345,
|
|
"rewards/DenseCaptionSodaM/std": 0.15110737085342407,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9594494104385376,
|
|
"rewards/TiemstampCaptionLength/std": 0.17744947969913483,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3156.0,
|
|
"completions/mean_length": 2252.875,
|
|
"completions/min_length": 1010.0,
|
|
"entropy/max": 0.77734375,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.47265625,
|
|
"epoch": 0.4765625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45987099409103394,
|
|
"kl": 0.07764089852571487,
|
|
"learning_rate": 5.804510333090287e-06,
|
|
"loss": 0.003107200376689434,
|
|
"reward": 2.0881872177124023,
|
|
"reward_std": 0.21753573417663574,
|
|
"rewards/DenseCaptionF1/mean": 0.6521599292755127,
|
|
"rewards/DenseCaptionF1/std": 0.17497283220291138,
|
|
"rewards/DenseCaptionSodaM/mean": 0.45686084032058716,
|
|
"rewards/DenseCaptionSodaM/std": 0.1126566231250763,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9739583730697632,
|
|
"rewards/TiemstampCaptionLength/std": 0.1367896944284439,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3198.0,
|
|
"completions/mean_length": 1817.984375,
|
|
"completions/min_length": 1068.0,
|
|
"entropy/max": 0.89453125,
|
|
"entropy/mean": 0.61328125,
|
|
"entropy/min": 0.447265625,
|
|
"epoch": 0.48046875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5073015689849854,
|
|
"kl": 0.0798579603433609,
|
|
"learning_rate": 5.740645311756246e-06,
|
|
"loss": 0.0031996332108974457,
|
|
"reward": 2.1528115272521973,
|
|
"reward_std": 0.37162232398986816,
|
|
"rewards/DenseCaptionF1/mean": 0.7066170573234558,
|
|
"rewards/DenseCaptionF1/std": 0.2179267257452011,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4930696189403534,
|
|
"rewards/DenseCaptionSodaM/std": 0.15379729866981506,
|
|
"rewards/TiemstampCaptionLength/mean": 0.953125,
|
|
"rewards/TiemstampCaptionLength/std": 0.21304203569889069,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 3474.0,
|
|
"completions/mean_length": 2020.171875,
|
|
"completions/min_length": 1078.0,
|
|
"entropy/max": 0.9765625,
|
|
"entropy/mean": 0.62890625,
|
|
"entropy/min": 0.3671875,
|
|
"epoch": 0.484375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.46273958683013916,
|
|
"kl": 0.07388724386692047,
|
|
"learning_rate": 5.6766564987506564e-06,
|
|
"loss": 0.0029639480635523796,
|
|
"reward": 2.0951733589172363,
|
|
"reward_std": 0.36066603660583496,
|
|
"rewards/DenseCaptionF1/mean": 0.6725229024887085,
|
|
"rewards/DenseCaptionF1/std": 0.19911877810955048,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4695255160331726,
|
|
"rewards/DenseCaptionSodaM/std": 0.16270573437213898,
|
|
"rewards/TiemstampCaptionLength/mean": 0.953125,
|
|
"rewards/TiemstampCaptionLength/std": 0.21304203569889069,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 3616.0,
|
|
"completions/mean_length": 2323.375,
|
|
"completions/min_length": 1270.0,
|
|
"entropy/max": 0.8203125,
|
|
"entropy/mean": 0.625,
|
|
"entropy/min": 0.486328125,
|
|
"epoch": 0.48828125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45393383502960205,
|
|
"kl": 0.07782484591007233,
|
|
"learning_rate": 5.612554589182228e-06,
|
|
"loss": 0.003118175081908703,
|
|
"reward": 2.130305767059326,
|
|
"reward_std": 0.13083592057228088,
|
|
"rewards/DenseCaptionF1/mean": 0.6734137535095215,
|
|
"rewards/DenseCaptionF1/std": 0.13672827184200287,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4568920135498047,
|
|
"rewards/DenseCaptionSodaM/std": 0.091212198138237,
|
|
"rewards/TiemstampCaptionLength/mean": 1.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3029.0,
|
|
"completions/mean_length": 1909.359375,
|
|
"completions/min_length": 1267.0,
|
|
"entropy/max": 0.82421875,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.439453125,
|
|
"epoch": 0.4921875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.47897541522979736,
|
|
"kl": 0.0780659019947052,
|
|
"learning_rate": 5.548350297062659e-06,
|
|
"loss": 0.003124142996966839,
|
|
"reward": 2.068937063217163,
|
|
"reward_std": 0.3318726420402527,
|
|
"rewards/DenseCaptionF1/mean": 0.6577168703079224,
|
|
"rewards/DenseCaptionF1/std": 0.2262330800294876,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4737202823162079,
|
|
"rewards/DenseCaptionSodaM/std": 0.1527242511510849,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9375,
|
|
"rewards/TiemstampCaptionLength/std": 0.24397502839565277,
|
|
"rewards/TimestampFormat/mean": 0.9375,
|
|
"rewards/TimestampFormat/std": 0.24397502839565277,
|
|
"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": 7959.0,
|
|
"completions/mean_length": 2067.703125,
|
|
"completions/min_length": 752.0,
|
|
"entropy/max": 0.91796875,
|
|
"entropy/mean": 0.65625,
|
|
"entropy/min": 0.09228515625,
|
|
"epoch": 0.49609375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4522978365421295,
|
|
"kl": 0.08107860386371613,
|
|
"learning_rate": 5.484054353515896e-06,
|
|
"loss": 0.00327342189848423,
|
|
"reward": 2.1223559379577637,
|
|
"reward_std": 0.26210957765579224,
|
|
"rewards/DenseCaptionF1/mean": 0.7080155611038208,
|
|
"rewards/DenseCaptionF1/std": 0.23213255405426025,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4455903470516205,
|
|
"rewards/DenseCaptionSodaM/std": 0.11769149452447891,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3528.0,
|
|
"completions/mean_length": 1888.6875,
|
|
"completions/min_length": 1219.0,
|
|
"entropy/max": 1.0078125,
|
|
"entropy/mean": 0.6640625,
|
|
"entropy/min": 0.421875,
|
|
"epoch": 0.5,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5178694128990173,
|
|
"kl": 0.08069656789302826,
|
|
"learning_rate": 5.419677504984534e-06,
|
|
"loss": 0.0032322704792022705,
|
|
"reward": 1.9883326292037964,
|
|
"reward_std": 0.4129721224308014,
|
|
"rewards/DenseCaptionF1/mean": 0.6113201379776001,
|
|
"rewards/DenseCaptionF1/std": 0.24685245752334595,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4414656162261963,
|
|
"rewards/DenseCaptionSodaM/std": 0.19202734529972076,
|
|
"rewards/TiemstampCaptionLength/mean": 0.93359375,
|
|
"rewards/TiemstampCaptionLength/std": 0.244957834482193,
|
|
"rewards/TimestampFormat/mean": 0.9375,
|
|
"rewards/TimestampFormat/std": 0.24397502839565277,
|
|
"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": 3457.0,
|
|
"completions/mean_length": 2259.78125,
|
|
"completions/min_length": 1303.0,
|
|
"entropy/max": 0.87109375,
|
|
"entropy/mean": 0.6484375,
|
|
"entropy/min": 0.453125,
|
|
"epoch": 0.50390625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4596750736236572,
|
|
"kl": 0.07779379189014435,
|
|
"learning_rate": 5.3552305114336515e-06,
|
|
"loss": 0.003117215819656849,
|
|
"reward": 1.9733624458312988,
|
|
"reward_std": 0.28246039152145386,
|
|
"rewards/DenseCaptionF1/mean": 0.5915824174880981,
|
|
"rewards/DenseCaptionF1/std": 0.21047313511371613,
|
|
"rewards/DenseCaptionSodaM/mean": 0.41303005814552307,
|
|
"rewards/DenseCaptionSodaM/std": 0.10529915243387222,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3394.0,
|
|
"completions/mean_length": 2013.109375,
|
|
"completions/min_length": 1058.0,
|
|
"entropy/max": 0.71484375,
|
|
"entropy/mean": 0.59375,
|
|
"entropy/min": 0.45703125,
|
|
"epoch": 0.5078125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.47184762358665466,
|
|
"kl": 0.07777858525514603,
|
|
"learning_rate": 5.290724144552379e-06,
|
|
"loss": 0.003114067018032074,
|
|
"reward": 2.1788721084594727,
|
|
"reward_std": 0.3331061899662018,
|
|
"rewards/DenseCaptionF1/mean": 0.7039194107055664,
|
|
"rewards/DenseCaptionF1/std": 0.21520444750785828,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5233902931213379,
|
|
"rewards/DenseCaptionSodaM/std": 0.15150612592697144,
|
|
"rewards/TiemstampCaptionLength/mean": 0.949999988079071,
|
|
"rewards/TiemstampCaptionLength/std": 0.21380898356437683,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 3048.0,
|
|
"completions/mean_length": 1807.40625,
|
|
"completions/min_length": 1015.0,
|
|
"entropy/max": 0.8125,
|
|
"entropy/mean": 0.65234375,
|
|
"entropy/min": 0.416015625,
|
|
"epoch": 0.51171875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5258738994598389,
|
|
"kl": 0.07863827049732208,
|
|
"learning_rate": 5.2261691859535325e-06,
|
|
"loss": 0.00314484816044569,
|
|
"reward": 2.192899227142334,
|
|
"reward_std": 0.19421175122261047,
|
|
"rewards/DenseCaptionF1/mean": 0.7292016744613647,
|
|
"rewards/DenseCaptionF1/std": 0.21162056922912598,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4804387092590332,
|
|
"rewards/DenseCaptionSodaM/std": 0.12279528379440308,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9821428656578064,
|
|
"rewards/TiemstampCaptionLength/std": 0.1259881556034088,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3077.0,
|
|
"completions/mean_length": 2114.984375,
|
|
"completions/min_length": 1017.0,
|
|
"entropy/max": 0.828125,
|
|
"entropy/mean": 0.60546875,
|
|
"entropy/min": 0.4375,
|
|
"epoch": 0.515625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4472525715827942,
|
|
"kl": 0.07749108970165253,
|
|
"learning_rate": 5.161576425371554e-06,
|
|
"loss": 0.0031098229810595512,
|
|
"reward": 2.1546554565429688,
|
|
"reward_std": 0.2748919129371643,
|
|
"rewards/DenseCaptionF1/mean": 0.7203766703605652,
|
|
"rewards/DenseCaptionF1/std": 0.19003459811210632,
|
|
"rewards/DenseCaptionSodaM/mean": 0.46552878618240356,
|
|
"rewards/DenseCaptionSodaM/std": 0.12826240062713623,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3224.0,
|
|
"completions/mean_length": 1893.296875,
|
|
"completions/min_length": 1071.0,
|
|
"entropy/max": 0.8125,
|
|
"entropy/mean": 0.61328125,
|
|
"entropy/min": 0.40625,
|
|
"epoch": 0.51953125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4892449676990509,
|
|
"kl": 0.08212453871965408,
|
|
"learning_rate": 5.096956658859122e-06,
|
|
"loss": 0.003289756365120411,
|
|
"reward": 2.0817782878875732,
|
|
"reward_std": 0.20652133226394653,
|
|
"rewards/DenseCaptionF1/mean": 0.6309571266174316,
|
|
"rewards/DenseCaptionF1/std": 0.17405256628990173,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4690502882003784,
|
|
"rewards/DenseCaptionSodaM/std": 0.10827261954545975,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9791666865348816,
|
|
"rewards/TiemstampCaptionLength/std": 0.13113263249397278,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 2830.0,
|
|
"completions/mean_length": 1989.390625,
|
|
"completions/min_length": 1229.0,
|
|
"entropy/max": 1.0546875,
|
|
"entropy/mean": 0.6796875,
|
|
"entropy/min": 0.41796875,
|
|
"epoch": 0.5234375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.47863513231277466,
|
|
"kl": 0.07702918350696564,
|
|
"learning_rate": 5.032320686982697e-06,
|
|
"loss": 0.003087609075009823,
|
|
"reward": 2.080071449279785,
|
|
"reward_std": 0.21030884981155396,
|
|
"rewards/DenseCaptionF1/mean": 0.6409437656402588,
|
|
"rewards/DenseCaptionF1/std": 0.1598178744316101,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4649089574813843,
|
|
"rewards/DenseCaptionSodaM/std": 0.13791081309318542,
|
|
"rewards/TiemstampCaptionLength/mean": 0.964062511920929,
|
|
"rewards/TiemstampCaptionLength/std": 0.16937531530857086,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3882.0,
|
|
"completions/mean_length": 2071.984375,
|
|
"completions/min_length": 996.0,
|
|
"entropy/max": 0.85546875,
|
|
"entropy/mean": 0.625,
|
|
"entropy/min": 0.474609375,
|
|
"epoch": 0.52734375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.46847179532051086,
|
|
"kl": 0.07634493708610535,
|
|
"learning_rate": 4.967679313017304e-06,
|
|
"loss": 0.0030563147738575935,
|
|
"reward": 2.254507064819336,
|
|
"reward_std": 0.1252697855234146,
|
|
"rewards/DenseCaptionF1/mean": 0.745608389377594,
|
|
"rewards/DenseCaptionF1/std": 0.16144125163555145,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5104612708091736,
|
|
"rewards/DenseCaptionSodaM/std": 0.08606183528900146,
|
|
"rewards/TiemstampCaptionLength/mean": 0.996874988079071,
|
|
"rewards/TiemstampCaptionLength/std": 0.02500000037252903,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3804.0,
|
|
"completions/mean_length": 2212.9375,
|
|
"completions/min_length": 1171.0,
|
|
"entropy/max": 0.84765625,
|
|
"entropy/mean": 0.62109375,
|
|
"entropy/min": 0.458984375,
|
|
"epoch": 0.53125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4657875597476959,
|
|
"kl": 0.07665620744228363,
|
|
"learning_rate": 4.903043341140879e-06,
|
|
"loss": 0.0030647674575448036,
|
|
"reward": 2.2156639099121094,
|
|
"reward_std": 0.2053491771221161,
|
|
"rewards/DenseCaptionF1/mean": 0.7130922079086304,
|
|
"rewards/DenseCaptionF1/std": 0.164754718542099,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5197591185569763,
|
|
"rewards/DenseCaptionSodaM/std": 0.12633611261844635,
|
|
"rewards/TiemstampCaptionLength/mean": 0.981249988079071,
|
|
"rewards/TiemstampCaptionLength/std": 0.12708577513694763,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4474.0,
|
|
"completions/mean_length": 2508.359375,
|
|
"completions/min_length": 1452.0,
|
|
"entropy/max": 0.83203125,
|
|
"entropy/mean": 0.62890625,
|
|
"entropy/min": 0.44140625,
|
|
"epoch": 0.53515625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.40959247946739197,
|
|
"kl": 0.07662597298622131,
|
|
"learning_rate": 4.838423574628447e-06,
|
|
"loss": 0.003065949771553278,
|
|
"reward": 1.9942350387573242,
|
|
"reward_std": 0.4310782551765442,
|
|
"rewards/DenseCaptionF1/mean": 0.6403563022613525,
|
|
"rewards/DenseCaptionF1/std": 0.2952542304992676,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4634045362472534,
|
|
"rewards/DenseCaptionSodaM/std": 0.20146074891090393,
|
|
"rewards/TiemstampCaptionLength/mean": 0.8903236389160156,
|
|
"rewards/TiemstampCaptionLength/std": 0.3144792318344116,
|
|
"rewards/TimestampFormat/mean": 0.890625,
|
|
"rewards/TimestampFormat/std": 0.3145764470100403,
|
|
"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": 4055.0,
|
|
"completions/mean_length": 2282.65625,
|
|
"completions/min_length": 800.0,
|
|
"entropy/max": 0.78515625,
|
|
"entropy/mean": 0.6484375,
|
|
"entropy/min": 0.47265625,
|
|
"epoch": 0.5390625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.46899816393852234,
|
|
"kl": 0.07598844170570374,
|
|
"learning_rate": 4.773830814046469e-06,
|
|
"loss": 0.0030368417501449585,
|
|
"reward": 2.105253219604492,
|
|
"reward_std": 0.325213760137558,
|
|
"rewards/DenseCaptionF1/mean": 0.6453598737716675,
|
|
"rewards/DenseCaptionF1/std": 0.19740337133407593,
|
|
"rewards/DenseCaptionSodaM/mean": 0.49114352464675903,
|
|
"rewards/DenseCaptionSodaM/std": 0.13330049812793732,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 4706.0,
|
|
"completions/mean_length": 2358.6875,
|
|
"completions/min_length": 868.0,
|
|
"entropy/max": 0.8671875,
|
|
"entropy/mean": 0.67578125,
|
|
"entropy/min": 0.4375,
|
|
"epoch": 0.54296875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4552667438983917,
|
|
"kl": 0.08017595112323761,
|
|
"learning_rate": 4.7092758554476215e-06,
|
|
"loss": 0.003209669142961502,
|
|
"reward": 2.005809783935547,
|
|
"reward_std": 0.27194541692733765,
|
|
"rewards/DenseCaptionF1/mean": 0.617243766784668,
|
|
"rewards/DenseCaptionF1/std": 0.18094880878925323,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4219282567501068,
|
|
"rewards/DenseCaptionSodaM/std": 0.12984782457351685,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9645257592201233,
|
|
"rewards/TiemstampCaptionLength/std": 0.1759679913520813,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3173.0,
|
|
"completions/mean_length": 2263.890625,
|
|
"completions/min_length": 1349.0,
|
|
"entropy/max": 0.85546875,
|
|
"entropy/mean": 0.69140625,
|
|
"entropy/min": 0.490234375,
|
|
"epoch": 0.546875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.44491004943847656,
|
|
"kl": 0.08564892411231995,
|
|
"learning_rate": 4.644769488566351e-06,
|
|
"loss": 0.0034335777163505554,
|
|
"reward": 2.034367799758911,
|
|
"reward_std": 0.38397544622421265,
|
|
"rewards/DenseCaptionF1/mean": 0.6195927858352661,
|
|
"rewards/DenseCaptionF1/std": 0.21382057666778564,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4616501033306122,
|
|
"rewards/DenseCaptionSodaM/std": 0.14214344322681427,
|
|
"rewards/TiemstampCaptionLength/mean": 0.953125,
|
|
"rewards/TiemstampCaptionLength/std": 0.21304203569889069,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 3409.0,
|
|
"completions/mean_length": 2234.828125,
|
|
"completions/min_length": 1226.0,
|
|
"entropy/max": 0.85546875,
|
|
"entropy/mean": 0.65625,
|
|
"entropy/min": 0.49609375,
|
|
"epoch": 0.55078125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4594617486000061,
|
|
"kl": 0.0822211354970932,
|
|
"learning_rate": 4.580322495015466e-06,
|
|
"loss": 0.003294694237411022,
|
|
"reward": 2.1689605712890625,
|
|
"reward_std": 0.20229323208332062,
|
|
"rewards/DenseCaptionF1/mean": 0.6771668195724487,
|
|
"rewards/DenseCaptionF1/std": 0.14375855028629303,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5074186325073242,
|
|
"rewards/DenseCaptionSodaM/std": 0.1291685849428177,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4437.0,
|
|
"completions/mean_length": 2687.640625,
|
|
"completions/min_length": 1354.0,
|
|
"entropy/max": 0.921875,
|
|
"entropy/mean": 0.6953125,
|
|
"entropy/min": 0.50390625,
|
|
"epoch": 0.5546875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4368516206741333,
|
|
"kl": 0.08078444004058838,
|
|
"learning_rate": 4.515945646484105e-06,
|
|
"loss": 0.003236331045627594,
|
|
"reward": 2.0671610832214355,
|
|
"reward_std": 0.0964779257774353,
|
|
"rewards/DenseCaptionF1/mean": 0.6158434152603149,
|
|
"rewards/DenseCaptionF1/std": 0.10087098926305771,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4550629258155823,
|
|
"rewards/DenseCaptionSodaM/std": 0.07321523129940033,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9925091862678528,
|
|
"rewards/TiemstampCaptionLength/std": 0.02667914889752865,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 4109.0,
|
|
"completions/mean_length": 2490.921875,
|
|
"completions/min_length": 1337.0,
|
|
"entropy/max": 0.9296875,
|
|
"entropy/mean": 0.6796875,
|
|
"entropy/min": 0.474609375,
|
|
"epoch": 0.55859375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.43431416153907776,
|
|
"kl": 0.07664476335048676,
|
|
"learning_rate": 4.451649702937343e-06,
|
|
"loss": 0.003065924160182476,
|
|
"reward": 2.0921058654785156,
|
|
"reward_std": 0.19337058067321777,
|
|
"rewards/DenseCaptionF1/mean": 0.6772254705429077,
|
|
"rewards/DenseCaptionF1/std": 0.15021462738513947,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4391201138496399,
|
|
"rewards/DenseCaptionSodaM/std": 0.11458110064268112,
|
|
"rewards/TiemstampCaptionLength/mean": 0.967145562171936,
|
|
"rewards/TiemstampCaptionLength/std": 0.1480521559715271,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4213.0,
|
|
"completions/mean_length": 2231.84375,
|
|
"completions/min_length": 1017.0,
|
|
"entropy/max": 0.82421875,
|
|
"entropy/mean": 0.65625,
|
|
"entropy/min": 0.494140625,
|
|
"epoch": 0.5625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4718267023563385,
|
|
"kl": 0.08167318999767303,
|
|
"learning_rate": 4.387445410817774e-06,
|
|
"loss": 0.0032684034667909145,
|
|
"reward": 2.172900438308716,
|
|
"reward_std": 0.27565664052963257,
|
|
"rewards/DenseCaptionF1/mean": 0.7096633911132812,
|
|
"rewards/DenseCaptionF1/std": 0.19180156290531158,
|
|
"rewards/DenseCaptionSodaM/mean": 0.49491995573043823,
|
|
"rewards/DenseCaptionSodaM/std": 0.1180020347237587,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9678840637207031,
|
|
"rewards/TiemstampCaptionLength/std": 0.1752786934375763,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 4205.0,
|
|
"completions/mean_length": 1944.765625,
|
|
"completions/min_length": 656.0,
|
|
"entropy/max": 0.9140625,
|
|
"entropy/mean": 0.65234375,
|
|
"entropy/min": 0.515625,
|
|
"epoch": 0.56640625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4986732602119446,
|
|
"kl": 0.09024765342473984,
|
|
"learning_rate": 4.323343501249346e-06,
|
|
"loss": 0.003628816921263933,
|
|
"reward": 2.1517534255981445,
|
|
"reward_std": 0.2129601091146469,
|
|
"rewards/DenseCaptionF1/mean": 0.6971390247344971,
|
|
"rewards/DenseCaptionF1/std": 0.2230452001094818,
|
|
"rewards/DenseCaptionSodaM/mean": 0.47044554352760315,
|
|
"rewards/DenseCaptionSodaM/std": 0.1473081409931183,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9839630126953125,
|
|
"rewards/TiemstampCaptionLength/std": 0.12499114125967026,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3401.0,
|
|
"completions/mean_length": 2158.265625,
|
|
"completions/min_length": 1113.0,
|
|
"entropy/max": 1.1328125,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.326171875,
|
|
"epoch": 0.5703125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.46137720346450806,
|
|
"kl": 0.0771188735961914,
|
|
"learning_rate": 4.259354688243758e-06,
|
|
"loss": 0.0030826330184936523,
|
|
"reward": 2.2336673736572266,
|
|
"reward_std": 0.14201684296131134,
|
|
"rewards/DenseCaptionF1/mean": 0.7033727765083313,
|
|
"rewards/DenseCaptionF1/std": 0.15630793571472168,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5334195494651794,
|
|
"rewards/DenseCaptionSodaM/std": 0.0781983733177185,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9937499761581421,
|
|
"rewards/TiemstampCaptionLength/std": 0.05000000074505806,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3300.0,
|
|
"completions/mean_length": 2201.375,
|
|
"completions/min_length": 1103.0,
|
|
"entropy/max": 0.89453125,
|
|
"entropy/mean": 0.62890625,
|
|
"entropy/min": 0.427734375,
|
|
"epoch": 0.57421875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45495036244392395,
|
|
"kl": 0.07991325855255127,
|
|
"learning_rate": 4.195489666909714e-06,
|
|
"loss": 0.003197193145751953,
|
|
"reward": 2.087711811065674,
|
|
"reward_std": 0.3588419258594513,
|
|
"rewards/DenseCaptionF1/mean": 0.640007734298706,
|
|
"rewards/DenseCaptionF1/std": 0.19632108509540558,
|
|
"rewards/DenseCaptionSodaM/mean": 0.49457916617393494,
|
|
"rewards/DenseCaptionSodaM/std": 0.14636559784412384,
|
|
"rewards/TiemstampCaptionLength/mean": 0.953125,
|
|
"rewards/TiemstampCaptionLength/std": 0.21304203569889069,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 3105.0,
|
|
"completions/mean_length": 2104.15625,
|
|
"completions/min_length": 1452.0,
|
|
"entropy/max": 0.8125,
|
|
"entropy/mean": 0.66796875,
|
|
"entropy/min": 0.498046875,
|
|
"epoch": 0.578125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4562710225582123,
|
|
"kl": 0.0780017152428627,
|
|
"learning_rate": 4.131759111665349e-06,
|
|
"loss": 0.003115745261311531,
|
|
"reward": 2.1103854179382324,
|
|
"reward_std": 0.12313683331012726,
|
|
"rewards/DenseCaptionF1/mean": 0.65642249584198,
|
|
"rewards/DenseCaptionF1/std": 0.1891339123249054,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4539627432823181,
|
|
"rewards/DenseCaptionSodaM/std": 0.09051599353551865,
|
|
"rewards/TiemstampCaptionLength/mean": 1.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3722.0,
|
|
"completions/mean_length": 2194.5,
|
|
"completions/min_length": 1407.0,
|
|
"entropy/max": 0.81640625,
|
|
"entropy/mean": 0.60546875,
|
|
"entropy/min": 0.484375,
|
|
"epoch": 0.58203125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4287312626838684,
|
|
"kl": 0.0745185911655426,
|
|
"learning_rate": 4.06817367445409e-06,
|
|
"loss": 0.002987407147884369,
|
|
"reward": 2.203336477279663,
|
|
"reward_std": 0.18982213735580444,
|
|
"rewards/DenseCaptionF1/mean": 0.6999586820602417,
|
|
"rewards/DenseCaptionF1/std": 0.1766565889120102,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5190026164054871,
|
|
"rewards/DenseCaptionSodaM/std": 0.10568448901176453,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3458.0,
|
|
"completions/mean_length": 1971.984375,
|
|
"completions/min_length": 783.0,
|
|
"entropy/max": 1.09375,
|
|
"entropy/mean": 0.703125,
|
|
"entropy/min": 0.54296875,
|
|
"epoch": 0.5859375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.542759120464325,
|
|
"kl": 0.07930778712034225,
|
|
"learning_rate": 4.004743982964298e-06,
|
|
"loss": 0.0031747575849294662,
|
|
"reward": 2.123366355895996,
|
|
"reward_std": 0.1879543662071228,
|
|
"rewards/DenseCaptionF1/mean": 0.6784964203834534,
|
|
"rewards/DenseCaptionF1/std": 0.20692989230155945,
|
|
"rewards/DenseCaptionSodaM/mean": 0.46049508452415466,
|
|
"rewards/DenseCaptionSodaM/std": 0.13381001353263855,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4005.0,
|
|
"completions/mean_length": 2177.953125,
|
|
"completions/min_length": 1245.0,
|
|
"entropy/max": 0.8828125,
|
|
"entropy/mean": 0.625,
|
|
"entropy/min": 0.400390625,
|
|
"epoch": 0.58984375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.43816837668418884,
|
|
"kl": 0.08020733296871185,
|
|
"learning_rate": 3.941480638852948e-06,
|
|
"loss": 0.0032116477377712727,
|
|
"reward": 2.2209386825561523,
|
|
"reward_std": 0.28657346963882446,
|
|
"rewards/DenseCaptionF1/mean": 0.7376999855041504,
|
|
"rewards/DenseCaptionF1/std": 0.1764175295829773,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5164416432380676,
|
|
"rewards/DenseCaptionSodaM/std": 0.12051527202129364,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96484375,
|
|
"rewards/TiemstampCaptionLength/std": 0.17743313312530518,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3805.0,
|
|
"completions/mean_length": 2246.703125,
|
|
"completions/min_length": 1347.0,
|
|
"entropy/max": 1.0078125,
|
|
"entropy/mean": 0.67578125,
|
|
"entropy/min": 0.4609375,
|
|
"epoch": 0.59375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4635348320007324,
|
|
"kl": 0.07573550194501877,
|
|
"learning_rate": 3.878394215973663e-06,
|
|
"loss": 0.003034967929124832,
|
|
"reward": 2.129924774169922,
|
|
"reward_std": 0.13099989295005798,
|
|
"rewards/DenseCaptionF1/mean": 0.6344149708747864,
|
|
"rewards/DenseCaptionF1/std": 0.1651492863893509,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4981141686439514,
|
|
"rewards/DenseCaptionSodaM/std": 0.12987202405929565,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9947916865348816,
|
|
"rewards/TiemstampCaptionLength/std": 0.0416666679084301,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3872.0,
|
|
"completions/mean_length": 2001.15625,
|
|
"completions/min_length": 997.0,
|
|
"entropy/max": 0.78125,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.4375,
|
|
"epoch": 0.59765625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4990518093109131,
|
|
"kl": 0.07201633602380753,
|
|
"learning_rate": 3.815495258609404e-06,
|
|
"loss": 0.002882446628063917,
|
|
"reward": 2.2244858741760254,
|
|
"reward_std": 0.11344104260206223,
|
|
"rewards/DenseCaptionF1/mean": 0.7057707905769348,
|
|
"rewards/DenseCaptionF1/std": 0.12868660688400269,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5187150239944458,
|
|
"rewards/DenseCaptionSodaM/std": 0.1085120216012001,
|
|
"rewards/TiemstampCaptionLength/mean": 1.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 4950.0,
|
|
"completions/mean_length": 2356.203125,
|
|
"completions/min_length": 1256.0,
|
|
"entropy/max": 0.77734375,
|
|
"entropy/mean": 0.63671875,
|
|
"entropy/min": 0.53125,
|
|
"epoch": 0.6015625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45741304755210876,
|
|
"kl": 0.07944130897521973,
|
|
"learning_rate": 3.752794279710094e-06,
|
|
"loss": 0.0031835921108722687,
|
|
"reward": 2.009526252746582,
|
|
"reward_std": 0.20912650227546692,
|
|
"rewards/DenseCaptionF1/mean": 0.5993127822875977,
|
|
"rewards/DenseCaptionF1/std": 0.1969134509563446,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4295196831226349,
|
|
"rewards/DenseCaptionSodaM/std": 0.0874083936214447,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9770126342773438,
|
|
"rewards/TiemstampCaptionLength/std": 0.13045813143253326,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3683.0,
|
|
"completions/mean_length": 1949.875,
|
|
"completions/min_length": 667.0,
|
|
"entropy/max": 0.91796875,
|
|
"entropy/mean": 0.6328125,
|
|
"entropy/min": 0.451171875,
|
|
"epoch": 0.60546875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5073304176330566,
|
|
"kl": 0.07963970303535461,
|
|
"learning_rate": 3.690301759135471e-06,
|
|
"loss": 0.003191078081727028,
|
|
"reward": 2.1645946502685547,
|
|
"reward_std": 0.31681501865386963,
|
|
"rewards/DenseCaptionF1/mean": 0.7298088073730469,
|
|
"rewards/DenseCaptionF1/std": 0.20084446668624878,
|
|
"rewards/DenseCaptionSodaM/mean": 0.46896541118621826,
|
|
"rewards/DenseCaptionSodaM/std": 0.11861751228570938,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96484375,
|
|
"rewards/TiemstampCaptionLength/std": 0.17743313312530518,
|
|
"rewards/TimestampFormat/mean": 0.966796875,
|
|
"rewards/TimestampFormat/std": 0.1757102757692337,
|
|
"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": 2783.0,
|
|
"completions/mean_length": 1783.46875,
|
|
"completions/min_length": 895.0,
|
|
"entropy/max": 0.80078125,
|
|
"entropy/mean": 0.54296875,
|
|
"entropy/min": 0.3359375,
|
|
"epoch": 0.609375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4820803999900818,
|
|
"kl": 0.07591649889945984,
|
|
"learning_rate": 3.6280281419034934e-06,
|
|
"loss": 0.003037598915398121,
|
|
"reward": 2.2391483783721924,
|
|
"reward_std": 0.250728964805603,
|
|
"rewards/DenseCaptionF1/mean": 0.7175197601318359,
|
|
"rewards/DenseCaptionF1/std": 0.19149324297904968,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5372534990310669,
|
|
"rewards/DenseCaptionSodaM/std": 0.1222560852766037,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 2440.0,
|
|
"completions/mean_length": 1856.484375,
|
|
"completions/min_length": 1144.0,
|
|
"entropy/max": 0.90234375,
|
|
"entropy/mean": 0.6328125,
|
|
"entropy/min": 0.482421875,
|
|
"epoch": 0.61328125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4854549169540405,
|
|
"kl": 0.07825914025306702,
|
|
"learning_rate": 3.5659838364445505e-06,
|
|
"loss": 0.0031337589025497437,
|
|
"reward": 2.1887636184692383,
|
|
"reward_std": 0.22441458702087402,
|
|
"rewards/DenseCaptionF1/mean": 0.7078425884246826,
|
|
"rewards/DenseCaptionF1/std": 0.18556053936481476,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5082647800445557,
|
|
"rewards/DenseCaptionSodaM/std": 0.12181976437568665,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9700521230697632,
|
|
"rewards/TiemstampCaptionLength/std": 0.13957545161247253,
|
|
"rewards/TimestampFormat/mean": 0.9752604365348816,
|
|
"rewards/TimestampFormat/std": 0.13419009745121002,
|
|
"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": 3916.0,
|
|
"completions/mean_length": 2258.6875,
|
|
"completions/min_length": 531.0,
|
|
"entropy/max": 0.79296875,
|
|
"entropy/mean": 0.58984375,
|
|
"entropy/min": 0.400390625,
|
|
"epoch": 0.6171875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.44412437081336975,
|
|
"kl": 0.07344679534435272,
|
|
"learning_rate": 3.504179212861793e-06,
|
|
"loss": 0.0029507786966860294,
|
|
"reward": 2.173398971557617,
|
|
"reward_std": 0.14470039308071136,
|
|
"rewards/DenseCaptionF1/mean": 0.73322594165802,
|
|
"rewards/DenseCaptionF1/std": 0.17344404757022858,
|
|
"rewards/DenseCaptionSodaM/mean": 0.45319393277168274,
|
|
"rewards/DenseCaptionSodaM/std": 0.10041268914937973,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9739583730697632,
|
|
"rewards/TiemstampCaptionLength/std": 0.09375918656587601,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3890.0,
|
|
"completions/mean_length": 2661.3125,
|
|
"completions/min_length": 1681.0,
|
|
"entropy/max": 0.71484375,
|
|
"entropy/mean": 0.57421875,
|
|
"entropy/min": 0.3828125,
|
|
"epoch": 0.62109375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.3947578966617584,
|
|
"kl": 0.0687277615070343,
|
|
"learning_rate": 3.442624601197877e-06,
|
|
"loss": 0.002743840217590332,
|
|
"reward": 2.160353422164917,
|
|
"reward_std": 0.2935871481895447,
|
|
"rewards/DenseCaptionF1/mean": 0.7180777788162231,
|
|
"rewards/DenseCaptionF1/std": 0.18129633367061615,
|
|
"rewards/DenseCaptionSodaM/mean": 0.47352564334869385,
|
|
"rewards/DenseCaptionSodaM/std": 0.11926417797803879,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 4709.0,
|
|
"completions/mean_length": 2226.484375,
|
|
"completions/min_length": 1103.0,
|
|
"entropy/max": 0.8203125,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.443359375,
|
|
"epoch": 0.625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.46847644448280334,
|
|
"kl": 0.07471471279859543,
|
|
"learning_rate": 3.3813302897083955e-06,
|
|
"loss": 0.002989533357322216,
|
|
"reward": 2.1062912940979004,
|
|
"reward_std": 0.20779664814472198,
|
|
"rewards/DenseCaptionF1/mean": 0.6103988885879517,
|
|
"rewards/DenseCaptionF1/std": 0.18720845878124237,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5138769745826721,
|
|
"rewards/DenseCaptionSodaM/std": 0.1282425820827484,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9796562194824219,
|
|
"rewards/TiemstampCaptionLength/std": 0.12646403908729553,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3004.0,
|
|
"completions/mean_length": 2021.84375,
|
|
"completions/min_length": 1112.0,
|
|
"entropy/max": 0.69140625,
|
|
"entropy/mean": 0.578125,
|
|
"entropy/min": 0.4375,
|
|
"epoch": 0.62890625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4503817856311798,
|
|
"kl": 0.07432515919208527,
|
|
"learning_rate": 3.3203065231422904e-06,
|
|
"loss": 0.0029710540547966957,
|
|
"reward": 2.2090325355529785,
|
|
"reward_std": 0.1947193145751953,
|
|
"rewards/DenseCaptionF1/mean": 0.7223544120788574,
|
|
"rewards/DenseCaptionF1/std": 0.1703852266073227,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5023031830787659,
|
|
"rewards/DenseCaptionSodaM/std": 0.11408014595508575,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3613.0,
|
|
"completions/mean_length": 2257.78125,
|
|
"completions/min_length": 620.0,
|
|
"entropy/max": 0.75,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.4140625,
|
|
"epoch": 0.6328125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4628556966781616,
|
|
"kl": 0.07528138905763626,
|
|
"learning_rate": 3.259563501029548e-06,
|
|
"loss": 0.0030170101672410965,
|
|
"reward": 1.999921441078186,
|
|
"reward_std": 0.18532216548919678,
|
|
"rewards/DenseCaptionF1/mean": 0.5711463093757629,
|
|
"rewards/DenseCaptionF1/std": 0.13621442019939423,
|
|
"rewards/DenseCaptionSodaM/mean": 0.44440019130706787,
|
|
"rewards/DenseCaptionSodaM/std": 0.11345592886209488,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 2928.0,
|
|
"completions/mean_length": 1688.109375,
|
|
"completions/min_length": 590.0,
|
|
"entropy/max": 0.89453125,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.37890625,
|
|
"epoch": 0.63671875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5050703287124634,
|
|
"kl": 0.08230897784233093,
|
|
"learning_rate": 3.1991113759764493e-06,
|
|
"loss": 0.0032972879707813263,
|
|
"reward": 2.1300344467163086,
|
|
"reward_std": 0.2373379021883011,
|
|
"rewards/DenseCaptionF1/mean": 0.6795761585235596,
|
|
"rewards/DenseCaptionF1/std": 0.20928002893924713,
|
|
"rewards/DenseCaptionSodaM/mean": 0.466864675283432,
|
|
"rewards/DenseCaptionSodaM/std": 0.11047963798046112,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9828125238418579,
|
|
"rewards/TiemstampCaptionLength/std": 0.12542586028575897,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4313.0,
|
|
"completions/mean_length": 2701.3125,
|
|
"completions/min_length": 1418.0,
|
|
"entropy/max": 0.8359375,
|
|
"entropy/mean": 0.62890625,
|
|
"entropy/min": 0.431640625,
|
|
"epoch": 0.640625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4016496241092682,
|
|
"kl": 0.07297959923744202,
|
|
"learning_rate": 3.1389602519686515e-06,
|
|
"loss": 0.002920348197221756,
|
|
"reward": 2.137756109237671,
|
|
"reward_std": 0.12692490220069885,
|
|
"rewards/DenseCaptionF1/mean": 0.6812595129013062,
|
|
"rewards/DenseCaptionF1/std": 0.17293255031108856,
|
|
"rewards/DenseCaptionSodaM/mean": 0.457242488861084,
|
|
"rewards/DenseCaptionSodaM/std": 0.08693484961986542,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9985084533691406,
|
|
"rewards/TiemstampCaptionLength/std": 0.008417597971856594,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3828.0,
|
|
"completions/mean_length": 2167.09375,
|
|
"completions/min_length": 1130.0,
|
|
"entropy/max": 0.8046875,
|
|
"entropy/mean": 0.6015625,
|
|
"entropy/min": 0.427734375,
|
|
"epoch": 0.64453125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4546695649623871,
|
|
"kl": 0.07320497930049896,
|
|
"learning_rate": 3.0791201826824117e-06,
|
|
"loss": 0.0029290388338267803,
|
|
"reward": 2.1534037590026855,
|
|
"reward_std": 0.3382655084133148,
|
|
"rewards/DenseCaptionF1/mean": 0.7191801071166992,
|
|
"rewards/DenseCaptionF1/std": 0.21838988363742828,
|
|
"rewards/DenseCaptionSodaM/mean": 0.48109862208366394,
|
|
"rewards/DenseCaptionSodaM/std": 0.14587734639644623,
|
|
"rewards/TiemstampCaptionLength/mean": 0.953125,
|
|
"rewards/TiemstampCaptionLength/std": 0.21304203569889069,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 3871.0,
|
|
"completions/mean_length": 2094.71875,
|
|
"completions/min_length": 1055.0,
|
|
"entropy/max": 0.77734375,
|
|
"entropy/mean": 0.62109375,
|
|
"entropy/min": 0.447265625,
|
|
"epoch": 0.6484375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.47961291670799255,
|
|
"kl": 0.07660332322120667,
|
|
"learning_rate": 3.019601169804216e-06,
|
|
"loss": 0.00306691974401474,
|
|
"reward": 2.1812386512756348,
|
|
"reward_std": 0.12379679828882217,
|
|
"rewards/DenseCaptionF1/mean": 0.7198240160942078,
|
|
"rewards/DenseCaptionF1/std": 0.15444563329219818,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4633675813674927,
|
|
"rewards/DenseCaptionSodaM/std": 0.10771329700946808,
|
|
"rewards/TiemstampCaptionLength/mean": 0.99609375,
|
|
"rewards/TiemstampCaptionLength/std": 0.03125,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 2645.0,
|
|
"completions/mean_length": 1818.34375,
|
|
"completions/min_length": 1088.0,
|
|
"entropy/max": 0.9140625,
|
|
"entropy/mean": 0.671875,
|
|
"entropy/min": 0.46875,
|
|
"epoch": 0.65234375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.48787254095077515,
|
|
"kl": 0.07863400131464005,
|
|
"learning_rate": 2.9604131613590825e-06,
|
|
"loss": 0.00315242912620306,
|
|
"reward": 2.0957303047180176,
|
|
"reward_std": 0.3016930818557739,
|
|
"rewards/DenseCaptionF1/mean": 0.6641665697097778,
|
|
"rewards/DenseCaptionF1/std": 0.17183725535869598,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4628139138221741,
|
|
"rewards/DenseCaptionSodaM/std": 0.13749481737613678,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3935.0,
|
|
"completions/mean_length": 2586.671875,
|
|
"completions/min_length": 1089.0,
|
|
"entropy/max": 0.80078125,
|
|
"entropy/mean": 0.59375,
|
|
"entropy/min": 0.3671875,
|
|
"epoch": 0.65625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4076475203037262,
|
|
"kl": 0.07010465860366821,
|
|
"learning_rate": 2.901566050047855e-06,
|
|
"loss": 0.0028206678107380867,
|
|
"reward": 2.0712218284606934,
|
|
"reward_std": 0.2646124064922333,
|
|
"rewards/DenseCaptionF1/mean": 0.6718050241470337,
|
|
"rewards/DenseCaptionF1/std": 0.22563011944293976,
|
|
"rewards/DenseCaptionSodaM/mean": 0.44889605045318604,
|
|
"rewards/DenseCaptionSodaM/std": 0.1393977701663971,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9479166865348816,
|
|
"rewards/TiemstampCaptionLength/std": 0.21593283116817474,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 4262.0,
|
|
"completions/mean_length": 2314.046875,
|
|
"completions/min_length": 663.0,
|
|
"entropy/max": 1.0703125,
|
|
"entropy/mean": 0.6640625,
|
|
"entropy/min": 0.48828125,
|
|
"epoch": 0.66015625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.497470885515213,
|
|
"kl": 0.07848264276981354,
|
|
"learning_rate": 2.843069671593734e-06,
|
|
"loss": 0.0031392592936754227,
|
|
"reward": 2.099266529083252,
|
|
"reward_std": 0.24790078401565552,
|
|
"rewards/DenseCaptionF1/mean": 0.6861398220062256,
|
|
"rewards/DenseCaptionF1/std": 0.21495842933654785,
|
|
"rewards/DenseCaptionSodaM/mean": 0.44522881507873535,
|
|
"rewards/DenseCaptionSodaM/std": 0.1339079588651657,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9670459032058716,
|
|
"rewards/TiemstampCaptionLength/std": 0.17558936774730682,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3184.0,
|
|
"completions/mean_length": 2051.734375,
|
|
"completions/min_length": 1218.0,
|
|
"entropy/max": 0.8515625,
|
|
"entropy/mean": 0.62109375,
|
|
"entropy/min": 0.400390625,
|
|
"epoch": 0.6640625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4371926784515381,
|
|
"kl": 0.07437405735254288,
|
|
"learning_rate": 2.784933803098326e-06,
|
|
"loss": 0.0029774000868201256,
|
|
"reward": 2.052684783935547,
|
|
"reward_std": 0.26392990350723267,
|
|
"rewards/DenseCaptionF1/mean": 0.6261408925056458,
|
|
"rewards/DenseCaptionF1/std": 0.19510464370250702,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4609188139438629,
|
|
"rewards/DenseCaptionSodaM/std": 0.1267315149307251,
|
|
"rewards/TiemstampCaptionLength/mean": 0.965624988079071,
|
|
"rewards/TiemstampCaptionLength/std": 0.17658017575740814,
|
|
"rewards/TimestampFormat/mean": 0.965624988079071,
|
|
"rewards/TimestampFormat/std": 0.17658017575740814,
|
|
"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": 4037.0,
|
|
"completions/mean_length": 2418.453125,
|
|
"completions/min_length": 1151.0,
|
|
"entropy/max": 0.88671875,
|
|
"entropy/mean": 0.62109375,
|
|
"entropy/min": 0.443359375,
|
|
"epoch": 0.66796875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4279203712940216,
|
|
"kl": 0.07013701647520065,
|
|
"learning_rate": 2.7271681614074973e-06,
|
|
"loss": 0.0028105881065130234,
|
|
"reward": 2.040663003921509,
|
|
"reward_std": 0.2685054838657379,
|
|
"rewards/DenseCaptionF1/mean": 0.6230917572975159,
|
|
"rewards/DenseCaptionF1/std": 0.17188072204589844,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4505574107170105,
|
|
"rewards/DenseCaptionSodaM/std": 0.12054003775119781,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9652777910232544,
|
|
"rewards/TiemstampCaptionLength/std": 0.17693251371383667,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 4019.0,
|
|
"completions/mean_length": 2082.78125,
|
|
"completions/min_length": 1236.0,
|
|
"entropy/max": 0.84765625,
|
|
"entropy/mean": 0.59375,
|
|
"entropy/min": 0.431640625,
|
|
"epoch": 0.671875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4470483064651489,
|
|
"kl": 0.07786456495523453,
|
|
"learning_rate": 2.6697824014873076e-06,
|
|
"loss": 0.003116773907095194,
|
|
"reward": 2.1676316261291504,
|
|
"reward_std": 0.397072434425354,
|
|
"rewards/DenseCaptionF1/mean": 0.740259051322937,
|
|
"rewards/DenseCaptionF1/std": 0.23550359904766083,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4755496084690094,
|
|
"rewards/DenseCaptionSodaM/std": 0.14706814289093018,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9505208134651184,
|
|
"rewards/TiemstampCaptionLength/std": 0.21347814798355103,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 2941.0,
|
|
"completions/mean_length": 2070.40625,
|
|
"completions/min_length": 1096.0,
|
|
"entropy/max": 0.8203125,
|
|
"entropy/mean": 0.58984375,
|
|
"entropy/min": 0.3046875,
|
|
"epoch": 0.67578125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.44076287746429443,
|
|
"kl": 0.0743403285741806,
|
|
"learning_rate": 2.6127861148102552e-06,
|
|
"loss": 0.00297456793487072,
|
|
"reward": 2.1409144401550293,
|
|
"reward_std": 0.11926329135894775,
|
|
"rewards/DenseCaptionF1/mean": 0.6629953384399414,
|
|
"rewards/DenseCaptionF1/std": 0.17725545167922974,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4805231988430023,
|
|
"rewards/DenseCaptionSodaM/std": 0.12674783170223236,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9947916865348816,
|
|
"rewards/TiemstampCaptionLength/std": 0.0416666679084301,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3479.0,
|
|
"completions/mean_length": 2512.09375,
|
|
"completions/min_length": 1344.0,
|
|
"entropy/max": 0.75390625,
|
|
"entropy/mean": 0.55859375,
|
|
"entropy/min": 0.38671875,
|
|
"epoch": 0.6796875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.395829975605011,
|
|
"kl": 0.06750495731830597,
|
|
"learning_rate": 2.5561888277521797e-06,
|
|
"loss": 0.002698497148230672,
|
|
"reward": 2.242475748062134,
|
|
"reward_std": 0.12299925088882446,
|
|
"rewards/DenseCaptionF1/mean": 0.7270916700363159,
|
|
"rewards/DenseCaptionF1/std": 0.16096866130828857,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5153840184211731,
|
|
"rewards/DenseCaptionSodaM/std": 0.10852368921041489,
|
|
"rewards/TiemstampCaptionLength/mean": 1.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3797.0,
|
|
"completions/mean_length": 2257.265625,
|
|
"completions/min_length": 1065.0,
|
|
"entropy/max": 0.7890625,
|
|
"entropy/mean": 0.63671875,
|
|
"entropy/min": 0.484375,
|
|
"epoch": 0.68359375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4416440725326538,
|
|
"kl": 0.0740923136472702,
|
|
"learning_rate": 2.5000000000000015e-06,
|
|
"loss": 0.002964516170322895,
|
|
"reward": 2.0810370445251465,
|
|
"reward_std": 0.19657137989997864,
|
|
"rewards/DenseCaptionF1/mean": 0.6534109115600586,
|
|
"rewards/DenseCaptionF1/std": 0.15201124548912048,
|
|
"rewards/DenseCaptionSodaM/mean": 0.44987988471984863,
|
|
"rewards/DenseCaptionSodaM/std": 0.09995286166667938,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9711174368858337,
|
|
"rewards/TiemstampCaptionLength/std": 0.1381218135356903,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4328.0,
|
|
"completions/mean_length": 2506.09375,
|
|
"completions/min_length": 1061.0,
|
|
"entropy/max": 0.8046875,
|
|
"entropy/mean": 0.62109375,
|
|
"entropy/min": 0.451171875,
|
|
"epoch": 0.6875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4191126525402069,
|
|
"kl": 0.07020944356918335,
|
|
"learning_rate": 2.4442290229706344e-06,
|
|
"loss": 0.002813016064465046,
|
|
"reward": 2.133711099624634,
|
|
"reward_std": 0.10300801694393158,
|
|
"rewards/DenseCaptionF1/mean": 0.6849145293235779,
|
|
"rewards/DenseCaptionF1/std": 0.1559366136789322,
|
|
"rewards/DenseCaptionSodaM/mean": 0.44923704862594604,
|
|
"rewards/DenseCaptionSodaM/std": 0.1093151867389679,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9991188049316406,
|
|
"rewards/TiemstampCaptionLength/std": 0.007049560546875,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3366.0,
|
|
"completions/mean_length": 2043.234375,
|
|
"completions/min_length": 1205.0,
|
|
"entropy/max": 0.8671875,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.4375,
|
|
"epoch": 0.69140625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4535602033138275,
|
|
"kl": 0.07541535794734955,
|
|
"learning_rate": 2.3888852182413087e-06,
|
|
"loss": 0.003023270983248949,
|
|
"reward": 2.0415470600128174,
|
|
"reward_std": 0.13503248989582062,
|
|
"rewards/DenseCaptionF1/mean": 0.6082522869110107,
|
|
"rewards/DenseCaptionF1/std": 0.20383715629577637,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4332948923110962,
|
|
"rewards/DenseCaptionSodaM/std": 0.13324302434921265,
|
|
"rewards/TiemstampCaptionLength/mean": 1.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3483.0,
|
|
"completions/mean_length": 2130.96875,
|
|
"completions/min_length": 1160.0,
|
|
"entropy/max": 0.76171875,
|
|
"entropy/mean": 0.5703125,
|
|
"entropy/min": 0.431640625,
|
|
"epoch": 0.6953125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.44665437936782837,
|
|
"kl": 0.07155901193618774,
|
|
"learning_rate": 2.333977835991545e-06,
|
|
"loss": 0.002864268608391285,
|
|
"reward": 2.2150917053222656,
|
|
"reward_std": 0.32075563073158264,
|
|
"rewards/DenseCaptionF1/mean": 0.7311955094337463,
|
|
"rewards/DenseCaptionF1/std": 0.2124902456998825,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5227725505828857,
|
|
"rewards/DenseCaptionSodaM/std": 0.14286459982395172,
|
|
"rewards/TiemstampCaptionLength/mean": 0.953497052192688,
|
|
"rewards/TiemstampCaptionLength/std": 0.18869297206401825,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3955.0,
|
|
"completions/mean_length": 2342.3125,
|
|
"completions/min_length": 1224.0,
|
|
"entropy/max": 0.83984375,
|
|
"entropy/mean": 0.64453125,
|
|
"entropy/min": 0.435546875,
|
|
"epoch": 0.69921875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.44968166947364807,
|
|
"kl": 0.07600906491279602,
|
|
"learning_rate": 2.2795160534570866e-06,
|
|
"loss": 0.0030437661334872246,
|
|
"reward": 2.0526013374328613,
|
|
"reward_std": 0.19291798770427704,
|
|
"rewards/DenseCaptionF1/mean": 0.6455673575401306,
|
|
"rewards/DenseCaptionF1/std": 0.142412468791008,
|
|
"rewards/DenseCaptionSodaM/mean": 0.42265889048576355,
|
|
"rewards/DenseCaptionSodaM/std": 0.09118315577507019,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3451.0,
|
|
"completions/mean_length": 2206.578125,
|
|
"completions/min_length": 1333.0,
|
|
"entropy/max": 0.75390625,
|
|
"entropy/mean": 0.5859375,
|
|
"entropy/min": 0.431640625,
|
|
"epoch": 0.703125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.43218135833740234,
|
|
"kl": 0.07586587965488434,
|
|
"learning_rate": 2.2255089733960162e-06,
|
|
"loss": 0.0030368072912096977,
|
|
"reward": 2.200169563293457,
|
|
"reward_std": 0.20373526215553284,
|
|
"rewards/DenseCaptionF1/mean": 0.6856173872947693,
|
|
"rewards/DenseCaptionF1/std": 0.15119345486164093,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5301769971847534,
|
|
"rewards/DenseCaptionSodaM/std": 0.10045289248228073,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3360.0,
|
|
"completions/mean_length": 2379.890625,
|
|
"completions/min_length": 1621.0,
|
|
"entropy/max": 0.93359375,
|
|
"entropy/mean": 0.609375,
|
|
"entropy/min": 0.4140625,
|
|
"epoch": 0.70703125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4222281277179718,
|
|
"kl": 0.07173281908035278,
|
|
"learning_rate": 2.171965622567308e-06,
|
|
"loss": 0.0028846794739365578,
|
|
"reward": 2.0830845832824707,
|
|
"reward_std": 0.20682808756828308,
|
|
"rewards/DenseCaptionF1/mean": 0.645068883895874,
|
|
"rewards/DenseCaptionF1/std": 0.15822316706180573,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4556247293949127,
|
|
"rewards/DenseCaptionSodaM/std": 0.12347334623336792,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9804067611694336,
|
|
"rewards/TiemstampCaptionLength/std": 0.12650269269943237,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3016.0,
|
|
"completions/mean_length": 1821.28125,
|
|
"completions/min_length": 1047.0,
|
|
"entropy/max": 0.94921875,
|
|
"entropy/mean": 0.62109375,
|
|
"entropy/min": 0.4296875,
|
|
"epoch": 0.7109375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.49071553349494934,
|
|
"kl": 0.07316526770591736,
|
|
"learning_rate": 2.1188949502220987e-06,
|
|
"loss": 0.002941974438726902,
|
|
"reward": 2.1613147258758545,
|
|
"reward_std": 0.21143783628940582,
|
|
"rewards/DenseCaptionF1/mean": 0.6989374160766602,
|
|
"rewards/DenseCaptionF1/std": 0.1964447796344757,
|
|
"rewards/DenseCaptionSodaM/mean": 0.47995543479919434,
|
|
"rewards/DenseCaptionSodaM/std": 0.09392297267913818,
|
|
"rewards/TiemstampCaptionLength/mean": 0.98046875,
|
|
"rewards/TiemstampCaptionLength/std": 0.1283649355173111,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3147.0,
|
|
"completions/mean_length": 2061.71875,
|
|
"completions/min_length": 811.0,
|
|
"entropy/max": 0.96875,
|
|
"entropy/mean": 0.64453125,
|
|
"entropy/min": 0.44140625,
|
|
"epoch": 0.71484375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.48470523953437805,
|
|
"kl": 0.07881060987710953,
|
|
"learning_rate": 2.066305826607911e-06,
|
|
"loss": 0.0031569981947541237,
|
|
"reward": 2.147979736328125,
|
|
"reward_std": 0.22912800312042236,
|
|
"rewards/DenseCaptionF1/mean": 0.7178086042404175,
|
|
"rewards/DenseCaptionF1/std": 0.18792113661766052,
|
|
"rewards/DenseCaptionSodaM/mean": 0.45401784777641296,
|
|
"rewards/DenseCaptionSodaM/std": 0.12819407880306244,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9679315090179443,
|
|
"rewards/TiemstampCaptionLength/std": 0.13817273080348969,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 2962.0,
|
|
"completions/mean_length": 1975.1875,
|
|
"completions/min_length": 1135.0,
|
|
"entropy/max": 0.79296875,
|
|
"entropy/mean": 0.60546875,
|
|
"entropy/min": 0.4453125,
|
|
"epoch": 0.71875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4550570845603943,
|
|
"kl": 0.07269569486379623,
|
|
"learning_rate": 2.0142070414860704e-06,
|
|
"loss": 0.0029172184877097607,
|
|
"reward": 2.123661994934082,
|
|
"reward_std": 0.20257888734340668,
|
|
"rewards/DenseCaptionF1/mean": 0.6824607849121094,
|
|
"rewards/DenseCaptionF1/std": 0.19700302183628082,
|
|
"rewards/DenseCaptionSodaM/mean": 0.45682621002197266,
|
|
"rewards/DenseCaptionSodaM/std": 0.09912190586328506,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3525.0,
|
|
"completions/mean_length": 2330.828125,
|
|
"completions/min_length": 1014.0,
|
|
"entropy/max": 0.77734375,
|
|
"entropy/mean": 0.5859375,
|
|
"entropy/min": 0.412109375,
|
|
"epoch": 0.72265625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.42335712909698486,
|
|
"kl": 0.06866112351417542,
|
|
"learning_rate": 1.962607302662582e-06,
|
|
"loss": 0.002748417668044567,
|
|
"reward": 2.190992832183838,
|
|
"reward_std": 0.17999503016471863,
|
|
"rewards/DenseCaptionF1/mean": 0.728723406791687,
|
|
"rewards/DenseCaptionF1/std": 0.1993679702281952,
|
|
"rewards/DenseCaptionSodaM/mean": 0.47789451479911804,
|
|
"rewards/DenseCaptionSodaM/std": 0.10779637098312378,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4001.0,
|
|
"completions/mean_length": 2398.203125,
|
|
"completions/min_length": 1115.0,
|
|
"entropy/max": 0.72265625,
|
|
"entropy/mean": 0.58203125,
|
|
"entropy/min": 0.416015625,
|
|
"epoch": 0.7265625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4338615834712982,
|
|
"kl": 0.07361350208520889,
|
|
"learning_rate": 1.9115152345327154e-06,
|
|
"loss": 0.0029378142207860947,
|
|
"reward": 2.087932825088501,
|
|
"reward_std": 0.21515028178691864,
|
|
"rewards/DenseCaptionF1/mean": 0.6555083394050598,
|
|
"rewards/DenseCaptionF1/std": 0.18240214884281158,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4656275808811188,
|
|
"rewards/DenseCaptionSodaM/std": 0.13835589587688446,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96484375,
|
|
"rewards/TiemstampCaptionLength/std": 0.17743313312530518,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 2724.0,
|
|
"completions/mean_length": 2028.203125,
|
|
"completions/min_length": 1090.0,
|
|
"entropy/max": 0.84375,
|
|
"entropy/mean": 0.609375,
|
|
"entropy/min": 0.3671875,
|
|
"epoch": 0.73046875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45324060320854187,
|
|
"kl": 0.07853089272975922,
|
|
"learning_rate": 1.8609393766395083e-06,
|
|
"loss": 0.0031442586332559586,
|
|
"reward": 2.0174880027770996,
|
|
"reward_std": 0.3095704913139343,
|
|
"rewards/DenseCaptionF1/mean": 0.5845763683319092,
|
|
"rewards/DenseCaptionF1/std": 0.1757514774799347,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4676770567893982,
|
|
"rewards/DenseCaptionSodaM/std": 0.12563413381576538,
|
|
"rewards/TiemstampCaptionLength/mean": 0.961718738079071,
|
|
"rewards/TiemstampCaptionLength/std": 0.1785617619752884,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"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": 3534.0,
|
|
"completions/mean_length": 2102.265625,
|
|
"completions/min_length": 645.0,
|
|
"entropy/max": 0.87890625,
|
|
"entropy/mean": 0.62890625,
|
|
"entropy/min": 0.439453125,
|
|
"epoch": 0.734375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4552648961544037,
|
|
"kl": 0.07522042095661163,
|
|
"learning_rate": 1.8108881822464697e-06,
|
|
"loss": 0.003008980769664049,
|
|
"reward": 1.9012553691864014,
|
|
"reward_std": 0.26180797815322876,
|
|
"rewards/DenseCaptionF1/mean": 0.6142083406448364,
|
|
"rewards/DenseCaptionF1/std": 0.27197957038879395,
|
|
"rewards/DenseCaptionSodaM/mean": 0.41204696893692017,
|
|
"rewards/DenseCaptionSodaM/std": 0.18474657833576202,
|
|
"rewards/TiemstampCaptionLength/mean": 0.875,
|
|
"rewards/TiemstampCaptionLength/std": 0.3333333432674408,
|
|
"rewards/TimestampFormat/mean": 0.875,
|
|
"rewards/TimestampFormat/std": 0.3333333432674408,
|
|
"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": 3448.0,
|
|
"completions/mean_length": 1791.25,
|
|
"completions/min_length": 920.0,
|
|
"entropy/max": 0.890625,
|
|
"entropy/mean": 0.60546875,
|
|
"entropy/min": 0.416015625,
|
|
"epoch": 0.73828125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.50104820728302,
|
|
"kl": 0.07191494107246399,
|
|
"learning_rate": 1.7613700169247055e-06,
|
|
"loss": 0.0028807390481233597,
|
|
"reward": 2.0999393463134766,
|
|
"reward_std": 0.22087660431861877,
|
|
"rewards/DenseCaptionF1/mean": 0.6402316093444824,
|
|
"rewards/DenseCaptionF1/std": 0.14945225417613983,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4753328561782837,
|
|
"rewards/DenseCaptionSodaM/std": 0.09546773880720139,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3792.0,
|
|
"completions/mean_length": 2096.3125,
|
|
"completions/min_length": 1114.0,
|
|
"entropy/max": 0.83984375,
|
|
"entropy/mean": 0.6328125,
|
|
"entropy/min": 0.462890625,
|
|
"epoch": 0.7421875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4427258372306824,
|
|
"kl": 0.07179544121026993,
|
|
"learning_rate": 1.7123931571546826e-06,
|
|
"loss": 0.002877972088754177,
|
|
"reward": 2.050466299057007,
|
|
"reward_std": 0.21054741740226746,
|
|
"rewards/DenseCaptionF1/mean": 0.5769652724266052,
|
|
"rewards/DenseCaptionF1/std": 0.1798660010099411,
|
|
"rewards/DenseCaptionSodaM/mean": 0.49540388584136963,
|
|
"rewards/DenseCaptionSodaM/std": 0.13009656965732574,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9718191623687744,
|
|
"rewards/TiemstampCaptionLength/std": 0.14217542111873627,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4705.0,
|
|
"completions/mean_length": 2052.6875,
|
|
"completions/min_length": 1169.0,
|
|
"entropy/max": 0.85546875,
|
|
"entropy/mean": 0.6015625,
|
|
"entropy/min": 0.380859375,
|
|
"epoch": 0.74609375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.46291711926460266,
|
|
"kl": 0.07411293685436249,
|
|
"learning_rate": 1.6639657889429017e-06,
|
|
"loss": 0.0029668142087757587,
|
|
"reward": 2.1138978004455566,
|
|
"reward_std": 0.3824462890625,
|
|
"rewards/DenseCaptionF1/mean": 0.6552042961120605,
|
|
"rewards/DenseCaptionF1/std": 0.20194876194000244,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5076112747192383,
|
|
"rewards/DenseCaptionSodaM/std": 0.15611189603805542,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9490394592285156,
|
|
"rewards/TiemstampCaptionLength/std": 0.21312199532985687,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 3168.0,
|
|
"completions/mean_length": 1958.484375,
|
|
"completions/min_length": 1113.0,
|
|
"entropy/max": 0.796875,
|
|
"entropy/mean": 0.56640625,
|
|
"entropy/min": 0.3984375,
|
|
"epoch": 0.75,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45552197098731995,
|
|
"kl": 0.06913820654153824,
|
|
"learning_rate": 1.6160960064536907e-06,
|
|
"loss": 0.0027688471600413322,
|
|
"reward": 2.1777918338775635,
|
|
"reward_std": 0.10849934816360474,
|
|
"rewards/DenseCaptionF1/mean": 0.6881303787231445,
|
|
"rewards/DenseCaptionF1/std": 0.17405274510383606,
|
|
"rewards/DenseCaptionSodaM/mean": 0.48966145515441895,
|
|
"rewards/DenseCaptionSodaM/std": 0.13072755932807922,
|
|
"rewards/TiemstampCaptionLength/mean": 1.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 4225.0,
|
|
"completions/mean_length": 2162.96875,
|
|
"completions/min_length": 1045.0,
|
|
"entropy/max": 0.828125,
|
|
"entropy/mean": 0.57421875,
|
|
"entropy/min": 0.419921875,
|
|
"epoch": 0.75390625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.43679875135421753,
|
|
"kl": 0.0711248368024826,
|
|
"learning_rate": 1.5687918106563326e-06,
|
|
"loss": 0.0028449371457099915,
|
|
"reward": 2.109659194946289,
|
|
"reward_std": 0.23365890979766846,
|
|
"rewards/DenseCaptionF1/mean": 0.6740397214889526,
|
|
"rewards/DenseCaptionF1/std": 0.21991942822933197,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4846953749656677,
|
|
"rewards/DenseCaptionSodaM/std": 0.14436903595924377,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9487228393554688,
|
|
"rewards/TiemstampCaptionLength/std": 0.21437354385852814,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 3412.0,
|
|
"completions/mean_length": 2167.484375,
|
|
"completions/min_length": 1049.0,
|
|
"entropy/max": 0.91015625,
|
|
"entropy/mean": 0.62109375,
|
|
"entropy/min": 0.44140625,
|
|
"epoch": 0.7578125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4549855589866638,
|
|
"kl": 0.07042649388313293,
|
|
"learning_rate": 1.52206110798779e-06,
|
|
"loss": 0.0028215572237968445,
|
|
"reward": 2.158311128616333,
|
|
"reward_std": 0.1848626732826233,
|
|
"rewards/DenseCaptionF1/mean": 0.7175110578536987,
|
|
"rewards/DenseCaptionF1/std": 0.17548702657222748,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4609822928905487,
|
|
"rewards/DenseCaptionSodaM/std": 0.1038520559668541,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9752604365348816,
|
|
"rewards/TiemstampCaptionLength/std": 0.13419011235237122,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 3686.0,
|
|
"completions/mean_length": 2076.71875,
|
|
"completions/min_length": 1283.0,
|
|
"entropy/max": 0.85546875,
|
|
"entropy/mean": 0.61328125,
|
|
"entropy/min": 0.431640625,
|
|
"epoch": 0.76171875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4611286520957947,
|
|
"kl": 0.06882451474666595,
|
|
"learning_rate": 1.4759117090312197e-06,
|
|
"loss": 0.0027554999105632305,
|
|
"reward": 2.1432876586914062,
|
|
"reward_std": 0.09668301045894623,
|
|
"rewards/DenseCaptionF1/mean": 0.6383467316627502,
|
|
"rewards/DenseCaptionF1/std": 0.13292570412158966,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5071728825569153,
|
|
"rewards/DenseCaptionSodaM/std": 0.09460372477769852,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9955357313156128,
|
|
"rewards/TiemstampCaptionLength/std": 0.0357142835855484,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 4485.0,
|
|
"completions/mean_length": 2300.515625,
|
|
"completions/min_length": 1425.0,
|
|
"entropy/max": 0.78515625,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.421875,
|
|
"epoch": 0.765625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.43017178773880005,
|
|
"kl": 0.07375560700893402,
|
|
"learning_rate": 1.4303513272105057e-06,
|
|
"loss": 0.0029521677643060684,
|
|
"reward": 2.105917453765869,
|
|
"reward_std": 0.2647617757320404,
|
|
"rewards/DenseCaptionF1/mean": 0.6670179963111877,
|
|
"rewards/DenseCaptionF1/std": 0.21269913017749786,
|
|
"rewards/DenseCaptionSodaM/mean": 0.48807311058044434,
|
|
"rewards/DenseCaptionSodaM/std": 0.1310773491859436,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9485275149345398,
|
|
"rewards/TiemstampCaptionLength/std": 0.21378369629383087,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"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": 3448.0,
|
|
"completions/mean_length": 2158.984375,
|
|
"completions/min_length": 1004.0,
|
|
"entropy/max": 0.79296875,
|
|
"entropy/mean": 0.625,
|
|
"entropy/min": 0.439453125,
|
|
"epoch": 0.76953125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.46644216775894165,
|
|
"kl": 0.0757053792476654,
|
|
"learning_rate": 1.3853875775010355e-06,
|
|
"loss": 0.0030341013334691525,
|
|
"reward": 2.1779346466064453,
|
|
"reward_std": 0.1979537010192871,
|
|
"rewards/DenseCaptionF1/mean": 0.6974056959152222,
|
|
"rewards/DenseCaptionF1/std": 0.19552253186702728,
|
|
"rewards/DenseCaptionSodaM/mean": 0.49615389108657837,
|
|
"rewards/DenseCaptionSodaM/std": 0.11029209941625595,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"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": 4013.0,
|
|
"completions/mean_length": 2295.875,
|
|
"completions/min_length": 1025.0,
|
|
"entropy/max": 0.85546875,
|
|
"entropy/mean": 0.578125,
|
|
"entropy/min": 0.3359375,
|
|
"epoch": 0.7734375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4417324662208557,
|
|
"kl": 0.06748858094215393,
|
|
"learning_rate": 1.3410279751569399e-06,
|
|
"loss": 0.0026979809626936913,
|
|
"reward": 2.103792905807495,
|
|
"reward_std": 0.10410730540752411,
|
|
"rewards/DenseCaptionF1/mean": 0.6195416450500488,
|
|
"rewards/DenseCaptionF1/std": 0.13912326097488403,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4842512309551239,
|
|
"rewards/DenseCaptionSodaM/std": 0.06717878580093384,
|
|
"rewards/TiemstampCaptionLength/mean": 1.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 4025.0,
|
|
"completions/mean_length": 2578.0,
|
|
"completions/min_length": 1260.0,
|
|
"entropy/max": 0.7578125,
|
|
"entropy/mean": 0.5859375,
|
|
"entropy/min": 0.37109375,
|
|
"epoch": 0.77734375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.3920497000217438,
|
|
"kl": 0.07040741294622421,
|
|
"learning_rate": 1.297279934454978e-06,
|
|
"loss": 0.002820095978677273,
|
|
"reward": 2.163327217102051,
|
|
"reward_std": 0.12465543299913406,
|
|
"rewards/DenseCaptionF1/mean": 0.6730664968490601,
|
|
"rewards/DenseCaptionF1/std": 0.15805818140506744,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4933856725692749,
|
|
"rewards/DenseCaptionSodaM/std": 0.0710998922586441,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9937499761581421,
|
|
"rewards/TiemstampCaptionLength/std": 0.05000000074505806,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"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": 3922.0,
|
|
"completions/mean_length": 2440.03125,
|
|
"completions/min_length": 1408.0,
|
|
"entropy/max": 1.0234375,
|
|
"entropy/mean": 0.6484375,
|
|
"entropy/min": 0.50390625,
|
|
"epoch": 0.78125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.41854700446128845,
|
|
"kl": 0.07008274644613266,
|
|
"learning_rate": 1.25415076745532e-06,
|
|
"loss": 0.002804812043905258,
|
|
"reward": 1.972537875175476,
|
|
"reward_std": 0.36541616916656494,
|
|
"rewards/DenseCaptionF1/mean": 0.6237460374832153,
|
|
"rewards/DenseCaptionF1/std": 0.23509538173675537,
|
|
"rewards/DenseCaptionSodaM/mean": 0.41302791237831116,
|
|
"rewards/DenseCaptionSodaM/std": 0.14560307562351227,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9340277910232544,
|
|
"rewards/TiemstampCaptionLength/std": 0.24465179443359375,
|
|
"rewards/TimestampFormat/mean": 0.9375,
|
|
"rewards/TimestampFormat/std": 0.24397502839565277,
|
|
"step": 200
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 2842.0,
|
|
"completions/mean_length": 2028.125,
|
|
"completions/min_length": 1415.0,
|
|
"entropy/max": 0.75390625,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.455078125,
|
|
"epoch": 0.78515625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.44143903255462646,
|
|
"kl": 0.07384993135929108,
|
|
"learning_rate": 1.2116476827794104e-06,
|
|
"loss": 0.0029541028197854757,
|
|
"reward": 2.1639225482940674,
|
|
"reward_std": 0.3017212450504303,
|
|
"rewards/DenseCaptionF1/mean": 0.7080820798873901,
|
|
"rewards/DenseCaptionF1/std": 0.21013802289962769,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4882064759731293,
|
|
"rewards/DenseCaptionSodaM/std": 0.11897902935743332,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9665178656578064,
|
|
"rewards/TiemstampCaptionLength/std": 0.17587246000766754,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 201
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3123.0,
|
|
"completions/mean_length": 2220.015625,
|
|
"completions/min_length": 1318.0,
|
|
"entropy/max": 0.95703125,
|
|
"entropy/mean": 0.671875,
|
|
"entropy/min": 0.482421875,
|
|
"epoch": 0.7890625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4494209885597229,
|
|
"kl": 0.06831617653369904,
|
|
"learning_rate": 1.1697777844051105e-06,
|
|
"loss": 0.002736983820796013,
|
|
"reward": 2.0604043006896973,
|
|
"reward_std": 0.38972604274749756,
|
|
"rewards/DenseCaptionF1/mean": 0.6555015444755554,
|
|
"rewards/DenseCaptionF1/std": 0.2091411054134369,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4517776370048523,
|
|
"rewards/DenseCaptionSodaM/std": 0.12859277427196503,
|
|
"rewards/TiemstampCaptionLength/mean": 0.953125,
|
|
"rewards/TiemstampCaptionLength/std": 0.21304203569889069,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"step": 202
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 2567.0,
|
|
"completions/mean_length": 1999.109375,
|
|
"completions/min_length": 1406.0,
|
|
"entropy/max": 0.9296875,
|
|
"entropy/mean": 0.6015625,
|
|
"entropy/min": 0.216796875,
|
|
"epoch": 0.79296875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4470736086368561,
|
|
"kl": 0.0712474063038826,
|
|
"learning_rate": 1.1285480704793378e-06,
|
|
"loss": 0.002855958417057991,
|
|
"reward": 2.1464757919311523,
|
|
"reward_std": 0.2201235294342041,
|
|
"rewards/DenseCaptionF1/mean": 0.6772543787956238,
|
|
"rewards/DenseCaptionF1/std": 0.20828545093536377,
|
|
"rewards/DenseCaptionSodaM/mean": 0.48797160387039185,
|
|
"rewards/DenseCaptionSodaM/std": 0.15297161042690277,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9781249761581421,
|
|
"rewards/TiemstampCaptionLength/std": 0.13389021158218384,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"step": 203
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3641.0,
|
|
"completions/mean_length": 2079.015625,
|
|
"completions/min_length": 1053.0,
|
|
"entropy/max": 0.875,
|
|
"entropy/mean": 0.60546875,
|
|
"entropy/min": 0.455078125,
|
|
"epoch": 0.796875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.44605034589767456,
|
|
"kl": 0.06595249474048615,
|
|
"learning_rate": 1.0879654321484012e-06,
|
|
"loss": 0.0026347199454903603,
|
|
"reward": 2.2877116203308105,
|
|
"reward_std": 0.11608919501304626,
|
|
"rewards/DenseCaptionF1/mean": 0.7574422955513,
|
|
"rewards/DenseCaptionF1/std": 0.1552867442369461,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5302694439888,
|
|
"rewards/DenseCaptionSodaM/std": 0.12350013107061386,
|
|
"rewards/TiemstampCaptionLength/mean": 1.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"step": 204
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3713.0,
|
|
"completions/mean_length": 2055.5,
|
|
"completions/min_length": 1050.0,
|
|
"entropy/max": 0.80859375,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.47265625,
|
|
"epoch": 0.80078125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.46936070919036865,
|
|
"kl": 0.07235345244407654,
|
|
"learning_rate": 1.0480366524062041e-06,
|
|
"loss": 0.0029008351266384125,
|
|
"reward": 2.112664222717285,
|
|
"reward_std": 0.26479020714759827,
|
|
"rewards/DenseCaptionF1/mean": 0.6766363382339478,
|
|
"rewards/DenseCaptionF1/std": 0.24521754682064056,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4679291546344757,
|
|
"rewards/DenseCaptionSodaM/std": 0.11289465427398682,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9674479365348816,
|
|
"rewards/TiemstampCaptionLength/std": 0.17544174194335938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 205
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 4430.0,
|
|
"completions/mean_length": 2183.515625,
|
|
"completions/min_length": 945.0,
|
|
"entropy/max": 0.890625,
|
|
"entropy/mean": 0.63671875,
|
|
"entropy/min": 0.490234375,
|
|
"epoch": 0.8046875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4510256052017212,
|
|
"kl": 0.0715785101056099,
|
|
"learning_rate": 1.008768404960535e-06,
|
|
"loss": 0.002870047464966774,
|
|
"reward": 2.092207431793213,
|
|
"reward_std": 0.2838000953197479,
|
|
"rewards/DenseCaptionF1/mean": 0.653795063495636,
|
|
"rewards/DenseCaptionF1/std": 0.17665718495845795,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4728514552116394,
|
|
"rewards/DenseCaptionSodaM/std": 0.12559223175048828,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9623715281486511,
|
|
"rewards/TiemstampCaptionLength/std": 0.17747832834720612,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 206
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 2997.0,
|
|
"completions/mean_length": 2052.40625,
|
|
"completions/min_length": 1278.0,
|
|
"entropy/max": 0.7734375,
|
|
"entropy/mean": 0.60546875,
|
|
"entropy/min": 0.4921875,
|
|
"epoch": 0.80859375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45914798974990845,
|
|
"kl": 0.07530860602855682,
|
|
"learning_rate": 9.701672531176287e-07,
|
|
"loss": 0.003017054870724678,
|
|
"reward": 2.0713613033294678,
|
|
"reward_std": 0.36397799849510193,
|
|
"rewards/DenseCaptionF1/mean": 0.6464629173278809,
|
|
"rewards/DenseCaptionF1/std": 0.21029022336006165,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4873984158039093,
|
|
"rewards/DenseCaptionSodaM/std": 0.15191198885440826,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9375,
|
|
"rewards/TiemstampCaptionLength/std": 0.24397502839565277,
|
|
"rewards/TimestampFormat/mean": 0.9375,
|
|
"rewards/TimestampFormat/std": 0.24397502839565277,
|
|
"step": 207
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 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": 9216.0,
|
|
"completions/mean_length": 2205.6875,
|
|
"completions/min_length": 884.0,
|
|
"entropy/max": 0.88671875,
|
|
"entropy/mean": 0.609375,
|
|
"entropy/min": 0.1259765625,
|
|
"epoch": 0.8125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45285290479660034,
|
|
"kl": 0.06573300808668137,
|
|
"learning_rate": 9.322396486851626e-07,
|
|
"loss": 0.0026457724161446095,
|
|
"reward": 2.1070380210876465,
|
|
"reward_std": 0.2084147036075592,
|
|
"rewards/DenseCaptionF1/mean": 0.7000912427902222,
|
|
"rewards/DenseCaptionF1/std": 0.20154625177383423,
|
|
"rewards/DenseCaptionSodaM/mean": 0.43819671869277954,
|
|
"rewards/DenseCaptionSodaM/std": 0.1280461698770523,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 208
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3699.0,
|
|
"completions/mean_length": 1969.6875,
|
|
"completions/min_length": 1028.0,
|
|
"entropy/max": 0.82421875,
|
|
"entropy/mean": 0.64453125,
|
|
"entropy/min": 0.47265625,
|
|
"epoch": 0.81640625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.47152179479599,
|
|
"kl": 0.07548464089632034,
|
|
"learning_rate": 8.949919308939081e-07,
|
|
"loss": 0.0030218875035643578,
|
|
"reward": 2.082573652267456,
|
|
"reward_std": 0.3463110625743866,
|
|
"rewards/DenseCaptionF1/mean": 0.6819970607757568,
|
|
"rewards/DenseCaptionF1/std": 0.25651249289512634,
|
|
"rewards/DenseCaptionSodaM/mean": 0.44745153188705444,
|
|
"rewards/DenseCaptionSodaM/std": 0.1388605684041977,
|
|
"rewards/TiemstampCaptionLength/mean": 0.953125,
|
|
"rewards/TiemstampCaptionLength/std": 0.21304203569889069,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"step": 209
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3282.0,
|
|
"completions/mean_length": 2113.78125,
|
|
"completions/min_length": 1020.0,
|
|
"entropy/max": 0.8046875,
|
|
"entropy/mean": 0.609375,
|
|
"entropy/min": 0.38671875,
|
|
"epoch": 0.8203125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.44752559065818787,
|
|
"kl": 0.06798422336578369,
|
|
"learning_rate": 8.584303253381848e-07,
|
|
"loss": 0.0027289558202028275,
|
|
"reward": 2.0789904594421387,
|
|
"reward_std": 0.21277743577957153,
|
|
"rewards/DenseCaptionF1/mean": 0.6418662071228027,
|
|
"rewards/DenseCaptionF1/std": 0.15732134878635406,
|
|
"rewards/DenseCaptionSodaM/mean": 0.45274922251701355,
|
|
"rewards/DenseCaptionSodaM/std": 0.11441058665513992,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"step": 210
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3629.0,
|
|
"completions/mean_length": 2243.0625,
|
|
"completions/min_length": 1053.0,
|
|
"entropy/max": 0.96875,
|
|
"entropy/mean": 0.66015625,
|
|
"entropy/min": 0.486328125,
|
|
"epoch": 0.82421875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4434755742549896,
|
|
"kl": 0.07127964496612549,
|
|
"learning_rate": 8.225609429353187e-07,
|
|
"loss": 0.0028593293391168118,
|
|
"reward": 2.0070624351501465,
|
|
"reward_std": 0.2700831890106201,
|
|
"rewards/DenseCaptionF1/mean": 0.6369417905807495,
|
|
"rewards/DenseCaptionF1/std": 0.20550128817558289,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4148314595222473,
|
|
"rewards/DenseCaptionSodaM/std": 0.10662335902452469,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9418278932571411,
|
|
"rewards/TiemstampCaptionLength/std": 0.19645355641841888,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 211
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3176.0,
|
|
"completions/mean_length": 2029.359375,
|
|
"completions/min_length": 1116.0,
|
|
"entropy/max": 1.0390625,
|
|
"entropy/mean": 0.61328125,
|
|
"entropy/min": 0.416015625,
|
|
"epoch": 0.828125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.47614586353302,
|
|
"kl": 0.06632761657238007,
|
|
"learning_rate": 7.873897789042523e-07,
|
|
"loss": 0.0026550842449069023,
|
|
"reward": 2.3058950901031494,
|
|
"reward_std": 0.14569929242134094,
|
|
"rewards/DenseCaptionF1/mean": 0.7638165950775146,
|
|
"rewards/DenseCaptionF1/std": 0.14367879927158356,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5420786142349243,
|
|
"rewards/DenseCaptionSodaM/std": 0.09184540808200836,
|
|
"rewards/TiemstampCaptionLength/mean": 1.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"step": 212
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 2787.0,
|
|
"completions/mean_length": 1889.140625,
|
|
"completions/min_length": 855.0,
|
|
"entropy/max": 0.8203125,
|
|
"entropy/mean": 0.609375,
|
|
"entropy/min": 0.412109375,
|
|
"epoch": 0.83203125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.47536811232566833,
|
|
"kl": 0.07015581429004669,
|
|
"learning_rate": 7.529227117635135e-07,
|
|
"loss": 0.0027969004586338997,
|
|
"reward": 2.0682640075683594,
|
|
"reward_std": 0.4049973487854004,
|
|
"rewards/DenseCaptionF1/mean": 0.6707839965820312,
|
|
"rewards/DenseCaptionF1/std": 0.21859075129032135,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4658394157886505,
|
|
"rewards/DenseCaptionSodaM/std": 0.14330562949180603,
|
|
"rewards/TiemstampCaptionLength/mean": 0.92578125,
|
|
"rewards/TiemstampCaptionLength/std": 0.24672332406044006,
|
|
"rewards/TimestampFormat/mean": 0.9375,
|
|
"rewards/TimestampFormat/std": 0.24397502839565277,
|
|
"step": 213
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3565.0,
|
|
"completions/mean_length": 2083.625,
|
|
"completions/min_length": 1310.0,
|
|
"entropy/max": 0.83984375,
|
|
"entropy/mean": 0.63671875,
|
|
"entropy/min": 0.40234375,
|
|
"epoch": 0.8359375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4679231643676758,
|
|
"kl": 0.07617446035146713,
|
|
"learning_rate": 7.191655023486682e-07,
|
|
"loss": 0.0030477726832032204,
|
|
"reward": 2.1063032150268555,
|
|
"reward_std": 0.17786702513694763,
|
|
"rewards/DenseCaptionF1/mean": 0.6722162961959839,
|
|
"rewards/DenseCaptionF1/std": 0.1687920242547989,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4497119188308716,
|
|
"rewards/DenseCaptionSodaM/std": 0.13003011047840118,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"step": 214
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 4273.0,
|
|
"completions/mean_length": 2406.578125,
|
|
"completions/min_length": 1258.0,
|
|
"entropy/max": 0.8828125,
|
|
"entropy/mean": 0.60546875,
|
|
"entropy/min": 0.3671875,
|
|
"epoch": 0.83984375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4217173457145691,
|
|
"kl": 0.07177823781967163,
|
|
"learning_rate": 6.86123792849458e-07,
|
|
"loss": 0.002874116413295269,
|
|
"reward": 2.1588830947875977,
|
|
"reward_std": 0.1843496412038803,
|
|
"rewards/DenseCaptionF1/mean": 0.7002047896385193,
|
|
"rewards/DenseCaptionF1/std": 0.16961494088172913,
|
|
"rewards/DenseCaptionSodaM/mean": 0.47687089443206787,
|
|
"rewards/DenseCaptionSodaM/std": 0.09769587218761444,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9792393445968628,
|
|
"rewards/TiemstampCaptionLength/std": 0.12946131825447083,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"step": 215
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3752.0,
|
|
"completions/mean_length": 2368.515625,
|
|
"completions/min_length": 1501.0,
|
|
"entropy/max": 0.8359375,
|
|
"entropy/mean": 0.58984375,
|
|
"entropy/min": 0.453125,
|
|
"epoch": 0.84375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.39841553568840027,
|
|
"kl": 0.06628976762294769,
|
|
"learning_rate": 6.53803105866761e-07,
|
|
"loss": 0.0026589157059788704,
|
|
"reward": 2.0928306579589844,
|
|
"reward_std": 0.28863298892974854,
|
|
"rewards/DenseCaptionF1/mean": 0.6491104364395142,
|
|
"rewards/DenseCaptionF1/std": 0.2038692682981491,
|
|
"rewards/DenseCaptionSodaM/mean": 0.476923406124115,
|
|
"rewards/DenseCaptionSodaM/std": 0.11315272748470306,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96484375,
|
|
"rewards/TiemstampCaptionLength/std": 0.17743313312530518,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 216
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3498.0,
|
|
"completions/mean_length": 2042.828125,
|
|
"completions/min_length": 1063.0,
|
|
"entropy/max": 0.9296875,
|
|
"entropy/mean": 0.6328125,
|
|
"entropy/min": 0.482421875,
|
|
"epoch": 0.84765625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4707883298397064,
|
|
"kl": 0.0769767165184021,
|
|
"learning_rate": 6.222088434895462e-07,
|
|
"loss": 0.003077911213040352,
|
|
"reward": 2.0829648971557617,
|
|
"reward_std": 0.27355125546455383,
|
|
"rewards/DenseCaptionF1/mean": 0.6705271005630493,
|
|
"rewards/DenseCaptionF1/std": 0.1908894032239914,
|
|
"rewards/DenseCaptionSodaM/mean": 0.44564101099967957,
|
|
"rewards/DenseCaptionSodaM/std": 0.12517456710338593,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96484375,
|
|
"rewards/TiemstampCaptionLength/std": 0.17743313312530518,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 217
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3759.0,
|
|
"completions/mean_length": 2335.328125,
|
|
"completions/min_length": 883.0,
|
|
"entropy/max": 0.8125,
|
|
"entropy/mean": 0.60546875,
|
|
"entropy/min": 0.3046875,
|
|
"epoch": 0.8515625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.44571444392204285,
|
|
"kl": 0.06844501197338104,
|
|
"learning_rate": 5.9134628639196e-07,
|
|
"loss": 0.002740604802966118,
|
|
"reward": 2.1427454948425293,
|
|
"reward_std": 0.27938559651374817,
|
|
"rewards/DenseCaptionF1/mean": 0.6829416155815125,
|
|
"rewards/DenseCaptionF1/std": 0.17771215736865997,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4910537898540497,
|
|
"rewards/DenseCaptionSodaM/std": 0.15398336946964264,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 218
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3600.0,
|
|
"completions/mean_length": 2080.234375,
|
|
"completions/min_length": 991.0,
|
|
"entropy/max": 0.8828125,
|
|
"entropy/mean": 0.640625,
|
|
"entropy/min": 0.416015625,
|
|
"epoch": 0.85546875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4650964140892029,
|
|
"kl": 0.07336148619651794,
|
|
"learning_rate": 5.612205929507209e-07,
|
|
"loss": 0.002939282450824976,
|
|
"reward": 2.1364259719848633,
|
|
"reward_std": 0.28768667578697205,
|
|
"rewards/DenseCaptionF1/mean": 0.6924072504043579,
|
|
"rewards/DenseCaptionF1/std": 0.18149390816688538,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4752686321735382,
|
|
"rewards/DenseCaptionSodaM/std": 0.14733342826366425,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 219
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3113.0,
|
|
"completions/mean_length": 2066.921875,
|
|
"completions/min_length": 1293.0,
|
|
"entropy/max": 0.859375,
|
|
"entropy/mean": 0.625,
|
|
"entropy/min": 0.435546875,
|
|
"epoch": 0.859375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4474633038043976,
|
|
"kl": 0.07305897027254105,
|
|
"learning_rate": 5.318367983829393e-07,
|
|
"loss": 0.0029309550300240517,
|
|
"reward": 2.1579113006591797,
|
|
"reward_std": 0.22205166518688202,
|
|
"rewards/DenseCaptionF1/mean": 0.6792052388191223,
|
|
"rewards/DenseCaptionF1/std": 0.16498726606369019,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4943310022354126,
|
|
"rewards/DenseCaptionSodaM/std": 0.12913137674331665,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"step": 220
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3457.0,
|
|
"completions/mean_length": 2067.984375,
|
|
"completions/min_length": 930.0,
|
|
"entropy/max": 0.734375,
|
|
"entropy/mean": 0.59375,
|
|
"entropy/min": 0.41015625,
|
|
"epoch": 0.86328125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4524148404598236,
|
|
"kl": 0.07275983691215515,
|
|
"learning_rate": 5.031998139045352e-07,
|
|
"loss": 0.0029124626889824867,
|
|
"reward": 2.168494462966919,
|
|
"reward_std": 0.17572757601737976,
|
|
"rewards/DenseCaptionF1/mean": 0.6836122274398804,
|
|
"rewards/DenseCaptionF1/std": 0.14580166339874268,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5055854320526123,
|
|
"rewards/DenseCaptionSodaM/std": 0.10375098884105682,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9742187261581421,
|
|
"rewards/TiemstampCaptionLength/std": 0.13685591518878937,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"step": 221
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 4325.0,
|
|
"completions/mean_length": 2887.171875,
|
|
"completions/min_length": 1661.0,
|
|
"entropy/max": 0.82421875,
|
|
"entropy/mean": 0.6171875,
|
|
"entropy/min": 0.4296875,
|
|
"epoch": 0.8671875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.380627304315567,
|
|
"kl": 0.06471222639083862,
|
|
"learning_rate": 4.753144259093734e-07,
|
|
"loss": 0.0025849281810224056,
|
|
"reward": 2.122527599334717,
|
|
"reward_std": 0.09056463092565536,
|
|
"rewards/DenseCaptionF1/mean": 0.6412855386734009,
|
|
"rewards/DenseCaptionF1/std": 0.12630011141300201,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4820184111595154,
|
|
"rewards/DenseCaptionSodaM/std": 0.08255140483379364,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9984474182128906,
|
|
"rewards/TiemstampCaptionLength/std": 0.008777683600783348,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"step": 222
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3261.0,
|
|
"completions/mean_length": 2071.703125,
|
|
"completions/min_length": 796.0,
|
|
"entropy/max": 0.76171875,
|
|
"entropy/mean": 0.5703125,
|
|
"entropy/min": 0.42578125,
|
|
"epoch": 0.87109375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4588277339935303,
|
|
"kl": 0.06928963959217072,
|
|
"learning_rate": 4.481852951692672e-07,
|
|
"loss": 0.002766369841992855,
|
|
"reward": 2.2557458877563477,
|
|
"reward_std": 0.27973294258117676,
|
|
"rewards/DenseCaptionF1/mean": 0.7644034624099731,
|
|
"rewards/DenseCaptionF1/std": 0.19099953770637512,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5225926637649536,
|
|
"rewards/DenseCaptionSodaM/std": 0.12312100827693939,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 223
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 4333.0,
|
|
"completions/mean_length": 2400.609375,
|
|
"completions/min_length": 669.0,
|
|
"entropy/max": 0.87890625,
|
|
"entropy/mean": 0.62109375,
|
|
"entropy/min": 0.462890625,
|
|
"epoch": 0.875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.43985965847969055,
|
|
"kl": 0.07196754217147827,
|
|
"learning_rate": 4.2181695605497066e-07,
|
|
"loss": 0.0028892047703266144,
|
|
"reward": 2.0200648307800293,
|
|
"reward_std": 0.2255019247531891,
|
|
"rewards/DenseCaptionF1/mean": 0.6254401206970215,
|
|
"rewards/DenseCaptionF1/std": 0.1799243539571762,
|
|
"rewards/DenseCaptionSodaM/mean": 0.42632484436035156,
|
|
"rewards/DenseCaptionSodaM/std": 0.14487968385219574,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9678497314453125,
|
|
"rewards/TiemstampCaptionLength/std": 0.1753530204296112,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 224
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 4103.0,
|
|
"completions/mean_length": 2191.234375,
|
|
"completions/min_length": 1137.0,
|
|
"entropy/max": 0.9453125,
|
|
"entropy/mean": 0.62890625,
|
|
"entropy/min": 0.4609375,
|
|
"epoch": 0.87890625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4341824948787689,
|
|
"kl": 0.07162442803382874,
|
|
"learning_rate": 3.9621381577830855e-07,
|
|
"loss": 0.0028738011606037617,
|
|
"reward": 2.065830945968628,
|
|
"reward_std": 0.27442771196365356,
|
|
"rewards/DenseCaptionF1/mean": 0.6485331058502197,
|
|
"rewards/DenseCaptionF1/std": 0.2045939713716507,
|
|
"rewards/DenseCaptionSodaM/mean": 0.45051243901252747,
|
|
"rewards/DenseCaptionSodaM/std": 0.11047396063804626,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9648208618164062,
|
|
"rewards/TiemstampCaptionLength/std": 0.17742861807346344,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 225
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 4282.0,
|
|
"completions/mean_length": 2272.125,
|
|
"completions/min_length": 1127.0,
|
|
"entropy/max": 0.85546875,
|
|
"entropy/mean": 0.65625,
|
|
"entropy/min": 0.404296875,
|
|
"epoch": 0.8828125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.47349095344543457,
|
|
"kl": 0.07164978981018066,
|
|
"learning_rate": 3.7138015365554834e-07,
|
|
"loss": 0.0028734710067510605,
|
|
"reward": 2.250483989715576,
|
|
"reward_std": 0.17411521077156067,
|
|
"rewards/DenseCaptionF1/mean": 0.7337714433670044,
|
|
"rewards/DenseCaptionF1/std": 0.19994576275348663,
|
|
"rewards/DenseCaptionSodaM/mean": 0.532686710357666,
|
|
"rewards/DenseCaptionSodaM/std": 0.13804049789905548,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9836769104003906,
|
|
"rewards/TiemstampCaptionLength/std": 0.1250361055135727,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"step": 226
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3917.0,
|
|
"completions/mean_length": 2077.9375,
|
|
"completions/min_length": 1069.0,
|
|
"entropy/max": 0.9140625,
|
|
"entropy/mean": 0.625,
|
|
"entropy/min": 0.37890625,
|
|
"epoch": 0.88671875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4496707022190094,
|
|
"kl": 0.06940577924251556,
|
|
"learning_rate": 3.473201203921578e-07,
|
|
"loss": 0.002788152080029249,
|
|
"reward": 2.157864570617676,
|
|
"reward_std": 0.18150551617145538,
|
|
"rewards/DenseCaptionF1/mean": 0.678928792476654,
|
|
"rewards/DenseCaptionF1/std": 0.14928093552589417,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5075816512107849,
|
|
"rewards/DenseCaptionSodaM/std": 0.11503242701292038,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9583333730697632,
|
|
"rewards/TiemstampCaptionLength/std": 0.15141892433166504,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"step": 227
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3290.0,
|
|
"completions/mean_length": 2042.390625,
|
|
"completions/min_length": 1132.0,
|
|
"entropy/max": 0.88671875,
|
|
"entropy/mean": 0.61328125,
|
|
"entropy/min": 0.3828125,
|
|
"epoch": 0.890625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.464011549949646,
|
|
"kl": 0.069074347615242,
|
|
"learning_rate": 3.2403773738905185e-07,
|
|
"loss": 0.002766566351056099,
|
|
"reward": 2.10378098487854,
|
|
"reward_std": 0.2530044913291931,
|
|
"rewards/DenseCaptionF1/mean": 0.6559926271438599,
|
|
"rewards/DenseCaptionF1/std": 0.18993285298347473,
|
|
"rewards/DenseCaptionSodaM/mean": 0.484897643327713,
|
|
"rewards/DenseCaptionSodaM/std": 0.12367614358663559,
|
|
"rewards/TiemstampCaptionLength/mean": 0.95703125,
|
|
"rewards/TiemstampCaptionLength/std": 0.18123632669448853,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 228
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3662.0,
|
|
"completions/mean_length": 2453.53125,
|
|
"completions/min_length": 1540.0,
|
|
"entropy/max": 0.86328125,
|
|
"entropy/mean": 0.609375,
|
|
"entropy/min": 0.388671875,
|
|
"epoch": 0.89453125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.40985172986984253,
|
|
"kl": 0.06651058793067932,
|
|
"learning_rate": 3.015368960704584e-07,
|
|
"loss": 0.0026644468307495117,
|
|
"reward": 2.2062201499938965,
|
|
"reward_std": 0.1915397346019745,
|
|
"rewards/DenseCaptionF1/mean": 0.7498728036880493,
|
|
"rewards/DenseCaptionF1/std": 0.18998169898986816,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4719722270965576,
|
|
"rewards/DenseCaptionSodaM/std": 0.1014145016670227,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"step": 229
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3979.0,
|
|
"completions/mean_length": 2211.734375,
|
|
"completions/min_length": 1275.0,
|
|
"entropy/max": 0.765625,
|
|
"entropy/mean": 0.609375,
|
|
"entropy/min": 0.4375,
|
|
"epoch": 0.8984375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4526543915271759,
|
|
"kl": 0.06700141727924347,
|
|
"learning_rate": 2.798213572335001e-07,
|
|
"loss": 0.002682400867342949,
|
|
"reward": 2.118042230606079,
|
|
"reward_std": 0.19524793326854706,
|
|
"rewards/DenseCaptionF1/mean": 0.7293181419372559,
|
|
"rewards/DenseCaptionF1/std": 0.17116592824459076,
|
|
"rewards/DenseCaptionSodaM/mean": 0.40434902906417847,
|
|
"rewards/DenseCaptionSodaM/std": 0.08266216516494751,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"step": 230
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3042.0,
|
|
"completions/mean_length": 2002.90625,
|
|
"completions/min_length": 1046.0,
|
|
"entropy/max": 0.7578125,
|
|
"entropy/mean": 0.62109375,
|
|
"entropy/min": 0.423828125,
|
|
"epoch": 0.90234375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.46941038966178894,
|
|
"kl": 0.06912705302238464,
|
|
"learning_rate": 2.5889475041961767e-07,
|
|
"loss": 0.002772316802293062,
|
|
"reward": 2.170058250427246,
|
|
"reward_std": 0.3172154426574707,
|
|
"rewards/DenseCaptionF1/mean": 0.694161593914032,
|
|
"rewards/DenseCaptionF1/std": 0.17546270787715912,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5090996026992798,
|
|
"rewards/DenseCaptionSodaM/std": 0.13624192774295807,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96484375,
|
|
"rewards/TiemstampCaptionLength/std": 0.17743313312530518,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 231
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3563.0,
|
|
"completions/mean_length": 1987.46875,
|
|
"completions/min_length": 659.0,
|
|
"entropy/max": 0.828125,
|
|
"entropy/mean": 0.6171875,
|
|
"entropy/min": 0.33984375,
|
|
"epoch": 0.90625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45779043436050415,
|
|
"kl": 0.06570929288864136,
|
|
"learning_rate": 2.3876057330792344e-07,
|
|
"loss": 0.0026300763711333275,
|
|
"reward": 2.1797404289245605,
|
|
"reward_std": 0.17871052026748657,
|
|
"rewards/DenseCaptionF1/mean": 0.7018101811408997,
|
|
"rewards/DenseCaptionF1/std": 0.22568084299564362,
|
|
"rewards/DenseCaptionSodaM/mean": 0.49467143416404724,
|
|
"rewards/DenseCaptionSodaM/std": 0.12307889014482498,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9821428656578064,
|
|
"rewards/TiemstampCaptionLength/std": 0.1259881556034088,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"step": 232
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3909.0,
|
|
"completions/mean_length": 2162.15625,
|
|
"completions/min_length": 1290.0,
|
|
"entropy/max": 0.7734375,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.34375,
|
|
"epoch": 0.91015625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4510428309440613,
|
|
"kl": 0.07253089547157288,
|
|
"learning_rate": 2.1942219113060215e-07,
|
|
"loss": 0.002905457280576229,
|
|
"reward": 2.173069715499878,
|
|
"reward_std": 0.2984904646873474,
|
|
"rewards/DenseCaptionF1/mean": 0.6751847267150879,
|
|
"rewards/DenseCaptionF1/std": 0.1709144115447998,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5291350483894348,
|
|
"rewards/DenseCaptionSodaM/std": 0.13066641986370087,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 233
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 2951.0,
|
|
"completions/mean_length": 2158.0,
|
|
"completions/min_length": 1542.0,
|
|
"entropy/max": 0.8046875,
|
|
"entropy/mean": 0.640625,
|
|
"entropy/min": 0.51171875,
|
|
"epoch": 0.9140625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.43684422969818115,
|
|
"kl": 0.07163859903812408,
|
|
"learning_rate": 2.0088283611044034e-07,
|
|
"loss": 0.0028678132221102715,
|
|
"reward": 2.1748478412628174,
|
|
"reward_std": 0.09829949587583542,
|
|
"rewards/DenseCaptionF1/mean": 0.6855545043945312,
|
|
"rewards/DenseCaptionF1/std": 0.15107811987400055,
|
|
"rewards/DenseCaptionSodaM/mean": 0.48929333686828613,
|
|
"rewards/DenseCaptionSodaM/std": 0.06877553462982178,
|
|
"rewards/TiemstampCaptionLength/mean": 1.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"step": 234
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3572.0,
|
|
"completions/mean_length": 2506.125,
|
|
"completions/min_length": 1396.0,
|
|
"entropy/max": 0.859375,
|
|
"entropy/mean": 0.69921875,
|
|
"entropy/min": 0.5546875,
|
|
"epoch": 0.91796875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.41872698068618774,
|
|
"kl": 0.06576144695281982,
|
|
"learning_rate": 1.8314560692059836e-07,
|
|
"loss": 0.0026304367929697037,
|
|
"reward": 2.002267837524414,
|
|
"reward_std": 0.3426184356212616,
|
|
"rewards/DenseCaptionF1/mean": 0.633314847946167,
|
|
"rewards/DenseCaptionF1/std": 0.19844883680343628,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4158281981945038,
|
|
"rewards/DenseCaptionSodaM/std": 0.1418202966451645,
|
|
"rewards/TiemstampCaptionLength/mean": 0.953125,
|
|
"rewards/TiemstampCaptionLength/std": 0.21304203569889069,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"step": 235
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3937.0,
|
|
"completions/mean_length": 2436.171875,
|
|
"completions/min_length": 1402.0,
|
|
"entropy/max": 0.7890625,
|
|
"entropy/mean": 0.62109375,
|
|
"entropy/min": 0.5,
|
|
"epoch": 0.921875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4116588532924652,
|
|
"kl": 0.07316441833972931,
|
|
"learning_rate": 1.6621346816668993e-07,
|
|
"loss": 0.0029328539967536926,
|
|
"reward": 2.1459648609161377,
|
|
"reward_std": 0.21624067425727844,
|
|
"rewards/DenseCaptionF1/mean": 0.6648256182670593,
|
|
"rewards/DenseCaptionF1/std": 0.18039190769195557,
|
|
"rewards/DenseCaptionSodaM/mean": 0.49763229489326477,
|
|
"rewards/DenseCaptionSodaM/std": 0.10255026072263718,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9826388955116272,
|
|
"rewards/TiemstampCaptionLength/std": 0.12554994225502014,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"step": 236
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3285.0,
|
|
"completions/mean_length": 2254.84375,
|
|
"completions/min_length": 1356.0,
|
|
"entropy/max": 0.90625,
|
|
"entropy/mean": 0.65625,
|
|
"entropy/min": 0.3984375,
|
|
"epoch": 0.92578125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4395812451839447,
|
|
"kl": 0.070140540599823,
|
|
"learning_rate": 1.500892498912826e-07,
|
|
"loss": 0.0028052818961441517,
|
|
"reward": 2.2381887435913086,
|
|
"reward_std": 0.10850651562213898,
|
|
"rewards/DenseCaptionF1/mean": 0.7510522603988647,
|
|
"rewards/DenseCaptionF1/std": 0.1511983424425125,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4871364235877991,
|
|
"rewards/DenseCaptionSodaM/std": 0.08491774648427963,
|
|
"rewards/TiemstampCaptionLength/mean": 1.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"step": 237
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3432.0,
|
|
"completions/mean_length": 2316.765625,
|
|
"completions/min_length": 1277.0,
|
|
"entropy/max": 0.7890625,
|
|
"entropy/mean": 0.6015625,
|
|
"entropy/min": 0.4140625,
|
|
"epoch": 0.9296875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4198128879070282,
|
|
"kl": 0.06609868258237839,
|
|
"learning_rate": 1.3477564710088097e-07,
|
|
"loss": 0.0026463111862540245,
|
|
"reward": 2.1127736568450928,
|
|
"reward_std": 0.2268976867198944,
|
|
"rewards/DenseCaptionF1/mean": 0.7002504467964172,
|
|
"rewards/DenseCaptionF1/std": 0.202296644449234,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4750233292579651,
|
|
"rewards/DenseCaptionSodaM/std": 0.16162288188934326,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9375,
|
|
"rewards/TiemstampCaptionLength/std": 0.24397502839565277,
|
|
"rewards/TimestampFormat/mean": 0.9375,
|
|
"rewards/TimestampFormat/std": 0.24397502839565277,
|
|
"step": 238
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3404.0,
|
|
"completions/mean_length": 2235.015625,
|
|
"completions/min_length": 1098.0,
|
|
"entropy/max": 0.98828125,
|
|
"entropy/mean": 0.68359375,
|
|
"entropy/min": 0.390625,
|
|
"epoch": 0.93359375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.45057475566864014,
|
|
"kl": 0.07315898686647415,
|
|
"learning_rate": 1.2027521931548214e-07,
|
|
"loss": 0.0029359692707657814,
|
|
"reward": 2.0559120178222656,
|
|
"reward_std": 0.1921595335006714,
|
|
"rewards/DenseCaptionF1/mean": 0.6378650069236755,
|
|
"rewards/DenseCaptionF1/std": 0.1667502373456955,
|
|
"rewards/DenseCaptionSodaM/mean": 0.433671772480011,
|
|
"rewards/DenseCaptionSodaM/std": 0.11927897483110428,
|
|
"rewards/TiemstampCaptionLength/mean": 0.984375,
|
|
"rewards/TiemstampCaptionLength/std": 0.125,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"step": 239
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 4122.0,
|
|
"completions/mean_length": 2106.46875,
|
|
"completions/min_length": 668.0,
|
|
"entropy/max": 0.83203125,
|
|
"entropy/mean": 0.62890625,
|
|
"entropy/min": 0.41796875,
|
|
"epoch": 0.9375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5213373303413391,
|
|
"kl": 0.07656201720237732,
|
|
"learning_rate": 1.0659039014077943e-07,
|
|
"loss": 0.00306682288646698,
|
|
"reward": 2.0398974418640137,
|
|
"reward_std": 0.353277325630188,
|
|
"rewards/DenseCaptionF1/mean": 0.6235545873641968,
|
|
"rewards/DenseCaptionF1/std": 0.2217920571565628,
|
|
"rewards/DenseCaptionSodaM/mean": 0.46451741456985474,
|
|
"rewards/DenseCaptionSodaM/std": 0.13420319557189941,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9505256414413452,
|
|
"rewards/TiemstampCaptionLength/std": 0.21292540431022644,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"step": 240
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 4234.0,
|
|
"completions/mean_length": 2567.890625,
|
|
"completions/min_length": 1133.0,
|
|
"entropy/max": 0.84375,
|
|
"entropy/mean": 0.6484375,
|
|
"entropy/min": 0.435546875,
|
|
"epoch": 0.94140625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.42212745547294617,
|
|
"kl": 0.06747667491436005,
|
|
"learning_rate": 9.372344686307655e-08,
|
|
"loss": 0.0027042608708143234,
|
|
"reward": 2.0471315383911133,
|
|
"reward_std": 0.38125166296958923,
|
|
"rewards/DenseCaptionF1/mean": 0.617460310459137,
|
|
"rewards/DenseCaptionF1/std": 0.18913257122039795,
|
|
"rewards/DenseCaptionSodaM/mean": 0.47680580615997314,
|
|
"rewards/DenseCaptionSodaM/std": 0.12752464413642883,
|
|
"rewards/TiemstampCaptionLength/mean": 0.952606201171875,
|
|
"rewards/TiemstampCaptionLength/std": 0.21296648681163788,
|
|
"rewards/TimestampFormat/mean": 0.953125,
|
|
"rewards/TimestampFormat/std": 0.21304203569889069,
|
|
"step": 241
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3376.0,
|
|
"completions/mean_length": 2288.578125,
|
|
"completions/min_length": 1101.0,
|
|
"entropy/max": 1.0859375,
|
|
"entropy/mean": 0.59765625,
|
|
"entropy/min": 0.3828125,
|
|
"epoch": 0.9453125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.441020131111145,
|
|
"kl": 0.06421603262424469,
|
|
"learning_rate": 8.167654006699444e-08,
|
|
"loss": 0.0025690067559480667,
|
|
"reward": 2.1579861640930176,
|
|
"reward_std": 0.12455077469348907,
|
|
"rewards/DenseCaptionF1/mean": 0.6679261922836304,
|
|
"rewards/DenseCaptionF1/std": 0.14132031798362732,
|
|
"rewards/DenseCaptionSodaM/mean": 0.49005988240242004,
|
|
"rewards/DenseCaptionSodaM/std": 0.1283525824546814,
|
|
"rewards/TiemstampCaptionLength/mean": 1.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"step": 242
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 4263.0,
|
|
"completions/mean_length": 1993.25,
|
|
"completions/min_length": 1075.0,
|
|
"entropy/max": 0.80859375,
|
|
"entropy/mean": 0.6171875,
|
|
"entropy/min": 0.4609375,
|
|
"epoch": 0.94921875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.47914016246795654,
|
|
"kl": 0.07047290354967117,
|
|
"learning_rate": 7.04516832760177e-08,
|
|
"loss": 0.0028210515156388283,
|
|
"reward": 2.190624952316284,
|
|
"reward_std": 0.2937573790550232,
|
|
"rewards/DenseCaptionF1/mean": 0.7174718976020813,
|
|
"rewards/DenseCaptionF1/std": 0.20946037769317627,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5048742294311523,
|
|
"rewards/DenseCaptionSodaM/std": 0.1464603692293167,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9678077697753906,
|
|
"rewards/TiemstampCaptionLength/std": 0.17528696358203888,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 243
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3077.0,
|
|
"completions/mean_length": 2154.015625,
|
|
"completions/min_length": 1308.0,
|
|
"entropy/max": 0.73828125,
|
|
"entropy/mean": 0.57421875,
|
|
"entropy/min": 0.40625,
|
|
"epoch": 0.953125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4197753965854645,
|
|
"kl": 0.06408187001943588,
|
|
"learning_rate": 6.005075261595495e-08,
|
|
"loss": 0.0025652721524238586,
|
|
"reward": 2.2150726318359375,
|
|
"reward_std": 0.27494385838508606,
|
|
"rewards/DenseCaptionF1/mean": 0.7120701670646667,
|
|
"rewards/DenseCaptionF1/std": 0.20112037658691406,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5342526435852051,
|
|
"rewards/DenseCaptionSodaM/std": 0.13056960701942444,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 244
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3090.0,
|
|
"completions/mean_length": 1707.8125,
|
|
"completions/min_length": 397.0,
|
|
"entropy/max": 0.82421875,
|
|
"entropy/mean": 0.5859375,
|
|
"entropy/min": 0.41796875,
|
|
"epoch": 0.95703125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.5193677544593811,
|
|
"kl": 0.07979986071586609,
|
|
"learning_rate": 5.047548650136513e-08,
|
|
"loss": 0.003219340927898884,
|
|
"reward": 2.2161216735839844,
|
|
"reward_std": 0.22335359454154968,
|
|
"rewards/DenseCaptionF1/mean": 0.7086374163627625,
|
|
"rewards/DenseCaptionF1/std": 0.21759891510009766,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5244112610816956,
|
|
"rewards/DenseCaptionSodaM/std": 0.12199661880731583,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9817708134651184,
|
|
"rewards/TiemstampCaptionLength/std": 0.12639760971069336,
|
|
"rewards/TimestampFormat/mean": 0.984375,
|
|
"rewards/TimestampFormat/std": 0.125,
|
|
"step": 245
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 2493.0,
|
|
"completions/mean_length": 1680.015625,
|
|
"completions/min_length": 947.0,
|
|
"entropy/max": 0.83203125,
|
|
"entropy/mean": 0.62109375,
|
|
"entropy/min": 0.4609375,
|
|
"epoch": 0.9609375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.506488561630249,
|
|
"kl": 0.07751372456550598,
|
|
"learning_rate": 4.172748534499449e-08,
|
|
"loss": 0.0031061689369380474,
|
|
"reward": 2.1879899501800537,
|
|
"reward_std": 0.1074364110827446,
|
|
"rewards/DenseCaptionF1/mean": 0.6649491190910339,
|
|
"rewards/DenseCaptionF1/std": 0.19902512431144714,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5230408906936646,
|
|
"rewards/DenseCaptionSodaM/std": 0.08708979189395905,
|
|
"rewards/TiemstampCaptionLength/mean": 1.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"step": 246
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 4872.0,
|
|
"completions/mean_length": 2129.78125,
|
|
"completions/min_length": 640.0,
|
|
"entropy/max": 0.82421875,
|
|
"entropy/mean": 0.6015625,
|
|
"entropy/min": 0.451171875,
|
|
"epoch": 0.96484375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.48763999342918396,
|
|
"kl": 0.07435654103755951,
|
|
"learning_rate": 3.3808211290284886e-08,
|
|
"loss": 0.0029764333739876747,
|
|
"reward": 2.144674301147461,
|
|
"reward_std": 0.1485549360513687,
|
|
"rewards/DenseCaptionF1/mean": 0.6928300857543945,
|
|
"rewards/DenseCaptionF1/std": 0.16323688626289368,
|
|
"rewards/DenseCaptionSodaM/mean": 0.45737937092781067,
|
|
"rewards/DenseCaptionSodaM/std": 0.11042436212301254,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9889297485351562,
|
|
"rewards/TiemstampCaptionLength/std": 0.03645797818899155,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"step": 247
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3257.0,
|
|
"completions/mean_length": 1980.921875,
|
|
"completions/min_length": 905.0,
|
|
"entropy/max": 0.74609375,
|
|
"entropy/mean": 0.59375,
|
|
"entropy/min": 0.435546875,
|
|
"epoch": 0.96875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4555642604827881,
|
|
"kl": 0.07218649983406067,
|
|
"learning_rate": 2.6718987966992683e-08,
|
|
"loss": 0.0028909044340252876,
|
|
"reward": 2.045485734939575,
|
|
"reward_std": 0.2386668473482132,
|
|
"rewards/DenseCaptionF1/mean": 0.6260990500450134,
|
|
"rewards/DenseCaptionF1/std": 0.22993408143520355,
|
|
"rewards/DenseCaptionSodaM/mean": 0.45063668489456177,
|
|
"rewards/DenseCaptionSodaM/std": 0.10911618918180466,
|
|
"rewards/TiemstampCaptionLength/mean": 0.96875,
|
|
"rewards/TiemstampCaptionLength/std": 0.17536810040473938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 248
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3933.0,
|
|
"completions/mean_length": 2316.71875,
|
|
"completions/min_length": 1155.0,
|
|
"entropy/max": 0.84375,
|
|
"entropy/mean": 0.6328125,
|
|
"entropy/min": 0.470703125,
|
|
"epoch": 0.97265625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.42503464221954346,
|
|
"kl": 0.06713893264532089,
|
|
"learning_rate": 2.0461000269953457e-08,
|
|
"loss": 0.002684153150767088,
|
|
"reward": 2.1603641510009766,
|
|
"reward_std": 0.13414692878723145,
|
|
"rewards/DenseCaptionF1/mean": 0.6677990555763245,
|
|
"rewards/DenseCaptionF1/std": 0.17346404492855072,
|
|
"rewards/DenseCaptionSodaM/mean": 0.49451813101768494,
|
|
"rewards/DenseCaptionSodaM/std": 0.09743543714284897,
|
|
"rewards/TiemstampCaptionLength/mean": 0.99609375,
|
|
"rewards/TiemstampCaptionLength/std": 0.03125,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"step": 249
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 4524.0,
|
|
"completions/mean_length": 2455.328125,
|
|
"completions/min_length": 1329.0,
|
|
"entropy/max": 0.79296875,
|
|
"entropy/mean": 0.5859375,
|
|
"entropy/min": 0.376953125,
|
|
"epoch": 0.9765625,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.41289544105529785,
|
|
"kl": 0.06387481093406677,
|
|
"learning_rate": 1.5035294161039882e-08,
|
|
"loss": 0.002559009939432144,
|
|
"reward": 2.2636756896972656,
|
|
"reward_std": 0.12502282857894897,
|
|
"rewards/DenseCaptionF1/mean": 0.7034400701522827,
|
|
"rewards/DenseCaptionF1/std": 0.13659776747226715,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5678247213363647,
|
|
"rewards/DenseCaptionSodaM/std": 0.06698809564113617,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9848217964172363,
|
|
"rewards/TiemstampCaptionLength/std": 0.08733879774808884,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3199.0,
|
|
"completions/mean_length": 2053.171875,
|
|
"completions/min_length": 713.0,
|
|
"entropy/max": 0.96484375,
|
|
"entropy/mean": 0.5859375,
|
|
"entropy/min": 0.2578125,
|
|
"epoch": 0.98046875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4668898582458496,
|
|
"kl": 0.0702609121799469,
|
|
"learning_rate": 1.044277649433989e-08,
|
|
"loss": 0.002815072424709797,
|
|
"reward": 2.1282243728637695,
|
|
"reward_std": 0.24179862439632416,
|
|
"rewards/DenseCaptionF1/mean": 0.6647729873657227,
|
|
"rewards/DenseCaptionF1/std": 0.21551796793937683,
|
|
"rewards/DenseCaptionSodaM/mean": 0.497305691242218,
|
|
"rewards/DenseCaptionSodaM/std": 0.13626523315906525,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9635416865348816,
|
|
"rewards/TiemstampCaptionLength/std": 0.17933039367198944,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 251
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3601.0,
|
|
"completions/mean_length": 2079.109375,
|
|
"completions/min_length": 1178.0,
|
|
"entropy/max": 0.81640625,
|
|
"entropy/mean": 0.5546875,
|
|
"entropy/min": 0.35546875,
|
|
"epoch": 0.984375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4461975395679474,
|
|
"kl": 0.06337602436542511,
|
|
"learning_rate": 6.6842148645840374e-09,
|
|
"loss": 0.0025420826859772205,
|
|
"reward": 2.2069387435913086,
|
|
"reward_std": 0.11307065188884735,
|
|
"rewards/DenseCaptionF1/mean": 0.6973945498466492,
|
|
"rewards/DenseCaptionF1/std": 0.15259438753128052,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5109056830406189,
|
|
"rewards/DenseCaptionSodaM/std": 0.10468944162130356,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9972774982452393,
|
|
"rewards/TiemstampCaptionLength/std": 0.015293176285922527,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"step": 252
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3537.0,
|
|
"completions/mean_length": 1864.875,
|
|
"completions/min_length": 1038.0,
|
|
"entropy/max": 0.75390625,
|
|
"entropy/mean": 0.58984375,
|
|
"entropy/min": 0.298828125,
|
|
"epoch": 0.98828125,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4846988618373871,
|
|
"kl": 0.06949920952320099,
|
|
"learning_rate": 3.760237478849793e-09,
|
|
"loss": 0.0027765128761529922,
|
|
"reward": 2.2026705741882324,
|
|
"reward_std": 0.11286143213510513,
|
|
"rewards/DenseCaptionF1/mean": 0.6730291843414307,
|
|
"rewards/DenseCaptionF1/std": 0.14994429051876068,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5296412706375122,
|
|
"rewards/DenseCaptionSodaM/std": 0.10844793915748596,
|
|
"rewards/TiemstampCaptionLength/mean": 1.0,
|
|
"rewards/TiemstampCaptionLength/std": 0.0,
|
|
"rewards/TimestampFormat/mean": 1.0,
|
|
"rewards/TimestampFormat/std": 0.0,
|
|
"step": 253
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3433.0,
|
|
"completions/mean_length": 1971.46875,
|
|
"completions/min_length": 1014.0,
|
|
"entropy/max": 0.76953125,
|
|
"entropy/mean": 0.62109375,
|
|
"entropy/min": 0.494140625,
|
|
"epoch": 0.9921875,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4856790006160736,
|
|
"kl": 0.06836636364459991,
|
|
"learning_rate": 1.6713330515627512e-09,
|
|
"loss": 0.0027369549497962,
|
|
"reward": 2.1835498809814453,
|
|
"reward_std": 0.2145911157131195,
|
|
"rewards/DenseCaptionF1/mean": 0.7142390012741089,
|
|
"rewards/DenseCaptionF1/std": 0.21817132830619812,
|
|
"rewards/DenseCaptionSodaM/mean": 0.5012121200561523,
|
|
"rewards/DenseCaptionSodaM/std": 0.1440695971250534,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9674479365348816,
|
|
"rewards/TiemstampCaptionLength/std": 0.17544174194335938,
|
|
"rewards/TimestampFormat/mean": 0.96875,
|
|
"rewards/TimestampFormat/std": 0.17536810040473938,
|
|
"step": 254
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 4529.0,
|
|
"completions/mean_length": 2173.328125,
|
|
"completions/min_length": 1104.0,
|
|
"entropy/max": 0.79296875,
|
|
"entropy/mean": 0.61328125,
|
|
"entropy/min": 0.365234375,
|
|
"epoch": 0.99609375,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4448992908000946,
|
|
"kl": 0.07250708341598511,
|
|
"learning_rate": 4.178507228136397e-10,
|
|
"loss": 0.002903016284108162,
|
|
"reward": 2.0148417949676514,
|
|
"reward_std": 0.4292619526386261,
|
|
"rewards/DenseCaptionF1/mean": 0.6329973936080933,
|
|
"rewards/DenseCaptionF1/std": 0.2630157470703125,
|
|
"rewards/DenseCaptionSodaM/mean": 0.4614725112915039,
|
|
"rewards/DenseCaptionSodaM/std": 0.17066454887390137,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9188690185546875,
|
|
"rewards/TiemstampCaptionLength/std": 0.270029753446579,
|
|
"rewards/TimestampFormat/mean": 0.921875,
|
|
"rewards/TimestampFormat/std": 0.27048972249031067,
|
|
"step": 255
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completions/clipped_ratio": 0.0,
|
|
"completions/max_length": 3888.0,
|
|
"completions/mean_length": 2174.671875,
|
|
"completions/min_length": 576.0,
|
|
"entropy/max": 0.84765625,
|
|
"entropy/mean": 0.66015625,
|
|
"entropy/min": 0.404296875,
|
|
"epoch": 1.0,
|
|
"frac_reward_zero_std": 0.0,
|
|
"grad_norm": 0.4850475788116455,
|
|
"kl": 0.06848195195198059,
|
|
"learning_rate": 0.0,
|
|
"loss": 0.0027359342202544212,
|
|
"reward": 1.9704103469848633,
|
|
"reward_std": 0.35087159276008606,
|
|
"rewards/DenseCaptionF1/mean": 0.6019693613052368,
|
|
"rewards/DenseCaptionF1/std": 0.2335011512041092,
|
|
"rewards/DenseCaptionSodaM/mean": 0.43094098567962646,
|
|
"rewards/DenseCaptionSodaM/std": 0.17221741378307343,
|
|
"rewards/TiemstampCaptionLength/mean": 0.9375,
|
|
"rewards/TiemstampCaptionLength/std": 0.24397502839565277,
|
|
"rewards/TimestampFormat/mean": 0.9375,
|
|
"rewards/TimestampFormat/std": 0.24397502839565277,
|
|
"step": 256
|
|
}
|
|
],
|
|
"logging_steps": 1,
|
|
"max_steps": 256,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 1,
|
|
"save_steps": 10,
|
|
"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": 2,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|