Safetensors
qwen3_5
IterCAD-RL / trainer_state.json
TaoTao0216's picture
Upload IterCAD-RL
9512ae2 verified
Raw
History Blame Contribute Delete
94.3 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.5185185185185185,
"eval_steps": 10.0,
"global_step": 70,
"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": 0.03125,
"completions/max_length": 7193.0,
"completions/mean_length": 1925.9375,
"completions/min_length": 575.0,
"epoch": 0.007407407407407408,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.33120759452863396,
"gvpm/clamp_rate": 0.375,
"gvpm/mask_ratio": 0.375,
"learning_rate": 7.142857142857142e-08,
"loss": -0.050480857491493225,
"num_turns": 3.21875,
"reward": 0.9470400810241699,
"reward_std": 0.36403971910476685,
"rewards/CADCDValueReward/mean": 0.007383444812148809,
"rewards/CADCDValueReward/std": 0.013863335829228163,
"rewards/CADChamferReward/mean": 0.44704006612300873,
"rewards/CADChamferReward/std": 0.3833623081445694,
"rewards/CADFormatReward/mean": 0.96875,
"rewards/CADFormatReward/std": 0.12198751419782639,
"rewards/CADInvalidReward/mean": 0.0546875,
"rewards/CADInvalidReward/std": 0.22850853204727173,
"rewards/CADProgressReward/mean": 0.03125,
"rewards/CADProgressReward/std": 0.0730636678636074,
"step": 1,
"step_time": 978.6295343909878
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0546875,
"completions/max_length": 8451.5,
"completions/mean_length": 2523.6796875,
"completions/min_length": 638.0,
"epoch": 0.014814814814814815,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.25568444681206004,
"gvpm/clamp_rate": 0.25,
"gvpm/mask_ratio": 0.25,
"learning_rate": 1.4285714285714285e-07,
"loss": -0.04256395250558853,
"num_turns": 3.3828125,
"reward": 0.705535352230072,
"reward_std": 0.32588590681552887,
"rewards/CADCDValueReward/mean": 0.00952278426848352,
"rewards/CADCDValueReward/std": 0.012697386788204312,
"rewards/CADChamferReward/mean": 0.2305353358387947,
"rewards/CADChamferReward/std": 0.3032306879758835,
"rewards/CADFormatReward/mean": 0.9453125,
"rewards/CADFormatReward/std": 0.22850853204727173,
"rewards/CADInvalidReward/mean": 0.1171875,
"rewards/CADInvalidReward/std": 0.32395489513874054,
"rewards/CADProgressReward/mean": 0.004687500069849193,
"rewards/CADProgressReward/std": 0.030036812648177147,
"step": 2,
"step_time": 748.078124585445
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0546875,
"completions/max_length": 8543.5,
"completions/mean_length": 2519.375,
"completions/min_length": 698.5,
"epoch": 0.022222222222222223,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.2393744403089944,
"gvpm/clamp_rate": 0.375,
"gvpm/mask_ratio": 0.375,
"learning_rate": 2.1428571428571426e-07,
"loss": -0.08637851476669312,
"num_turns": 3.65625,
"reward": 1.0107279419898987,
"reward_std": 0.39696645736694336,
"rewards/CADCDValueReward/mean": 0.006078975275158882,
"rewards/CADCDValueReward/std": 0.015850992407649755,
"rewards/CADChamferReward/mean": 0.5208841860294342,
"rewards/CADChamferReward/std": 0.41364483535289764,
"rewards/CADFormatReward/mean": 0.9453125,
"rewards/CADFormatReward/std": 0.22850853204727173,
"rewards/CADInvalidReward/mean": 0.1484375,
"rewards/CADInvalidReward/std": 0.3567937761545181,
"rewards/CADProgressReward/mean": 0.034375001676380634,
"rewards/CADProgressReward/std": 0.07499999925494194,
"step": 3,
"step_time": 815.9138611884555
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 5940.5,
"completions/mean_length": 1743.0546875,
"completions/min_length": 569.5,
"epoch": 0.02962962962962963,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.2627046329091631,
"gvpm/clamp_rate": 0.3125,
"gvpm/mask_ratio": 0.3125,
"learning_rate": 2.857142857142857e-07,
"loss": -0.0766700804233551,
"num_turns": 3.296875,
"reward": 0.8810432255268097,
"reward_std": 0.3356814980506897,
"rewards/CADCDValueReward/mean": 0.008348886156454682,
"rewards/CADCDValueReward/std": 0.012812133878469467,
"rewards/CADChamferReward/mean": 0.36854322254657745,
"rewards/CADChamferReward/std": 0.3755342662334442,
"rewards/CADFormatReward/mean": 1.0,
"rewards/CADFormatReward/std": 0.0,
"rewards/CADInvalidReward/mean": 0.0390625,
"rewards/CADInvalidReward/std": 0.13524486124515533,
"rewards/CADProgressReward/mean": 0.02500000037252903,
"rewards/CADProgressReward/std": 0.06597474217414856,
"step": 4,
"step_time": 744.047703415039
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 6256.5,
"completions/mean_length": 1847.3828125,
"completions/min_length": 481.5,
"epoch": 0.037037037037037035,
"frac_reward_zero_std": 0.0625,
"grad_norm": 0.2720512949299147,
"gvpm/clamp_rate": 0.125,
"gvpm/mask_ratio": 0.125,
"learning_rate": 3.5714285714285716e-07,
"loss": -0.0666135624051094,
"num_turns": 3.3046875,
"reward": 0.896435558795929,
"reward_std": 0.29569074511528015,
"rewards/CADCDValueReward/mean": 0.01122749038040638,
"rewards/CADCDValueReward/std": 0.02446559350937605,
"rewards/CADChamferReward/mean": 0.3894043266773224,
"rewards/CADChamferReward/std": 0.4212627708911896,
"rewards/CADFormatReward/mean": 0.9921875,
"rewards/CADFormatReward/std": 0.0625,
"rewards/CADInvalidReward/mean": 0.078125,
"rewards/CADInvalidReward/std": 0.27048972249031067,
"rewards/CADProgressReward/mean": 0.021875000558793545,
"rewards/CADProgressReward/std": 0.06271181628108025,
"step": 5,
"step_time": 705.8345783235272
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 8192.0,
"completions/mean_length": 1825.5078125,
"completions/min_length": 478.5,
"epoch": 0.044444444444444446,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.2598129047513917,
"gvpm/clamp_rate": 0.375,
"gvpm/mask_ratio": 0.375,
"learning_rate": 4.285714285714285e-07,
"loss": -0.06650832295417786,
"num_turns": 3.1796875,
"reward": 0.9514619708061218,
"reward_std": 0.3243105411529541,
"rewards/CADCDValueReward/mean": 0.018256386276334524,
"rewards/CADCDValueReward/std": 0.03519853297621012,
"rewards/CADChamferReward/mean": 0.44208694994449615,
"rewards/CADChamferReward/std": 0.4390777051448822,
"rewards/CADFormatReward/mean": 0.9765625,
"rewards/CADFormatReward/std": 0.1501840502023697,
"rewards/CADInvalidReward/mean": 0.0546875,
"rewards/CADInvalidReward/std": 0.22850853204727173,
"rewards/CADProgressReward/mean": 0.04218750260770321,
"rewards/CADProgressReward/std": 0.08166900649666786,
"step": 6,
"step_time": 681.5786384145031
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 5864.0,
"completions/mean_length": 2106.703125,
"completions/min_length": 715.5,
"epoch": 0.05185185185185185,
"frac_reward_zero_std": 0.125,
"grad_norm": 0.20583619133227324,
"gvpm/clamp_rate": 0.375,
"gvpm/mask_ratio": 0.375,
"learning_rate": 5e-07,
"loss": -0.009767329320311546,
"num_turns": 3.484375,
"reward": 0.6793689131736755,
"reward_std": 0.28771600127220154,
"rewards/CADCDValueReward/mean": 0.019346370361745358,
"rewards/CADCDValueReward/std": 0.024582237005233765,
"rewards/CADChamferReward/mean": 0.1903063803911209,
"rewards/CADChamferReward/std": 0.31679461896419525,
"rewards/CADFormatReward/mean": 0.96875,
"rewards/CADFormatReward/std": 0.12198751419782639,
"rewards/CADInvalidReward/mean": 0.125,
"rewards/CADInvalidReward/std": 0.3333333432674408,
"rewards/CADProgressReward/mean": 0.009375000023283064,
"rewards/CADProgressReward/std": 0.039548974484205246,
"step": 7,
"step_time": 712.6712970329681
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 5095.5,
"completions/mean_length": 1699.3203125,
"completions/min_length": 465.5,
"epoch": 0.05925925925925926,
"frac_reward_zero_std": 0.0625,
"grad_norm": 0.26601978424045747,
"gvpm/clamp_rate": 0.3125,
"gvpm/mask_ratio": 0.3125,
"learning_rate": 5.714285714285714e-07,
"loss": -0.030686555430293083,
"num_turns": 3.2109375,
"reward": 0.971459150314331,
"reward_std": 0.2936087101697922,
"rewards/CADCDValueReward/mean": 0.00972110778093338,
"rewards/CADCDValueReward/std": 0.014380504842847586,
"rewards/CADChamferReward/mean": 0.4503653943538666,
"rewards/CADChamferReward/std": 0.4412277638912201,
"rewards/CADFormatReward/mean": 1.0,
"rewards/CADFormatReward/std": 0.0,
"rewards/CADInvalidReward/mean": 0.0234375,
"rewards/CADInvalidReward/std": 0.1501840502023697,
"rewards/CADProgressReward/mean": 0.04218749888241291,
"rewards/CADProgressReward/std": 0.08035459369421005,
"step": 8,
"step_time": 700.7545191425015
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0390625,
"completions/max_length": 6546.0,
"completions/mean_length": 2595.953125,
"completions/min_length": 897.0,
"epoch": 0.06666666666666667,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.27073053298895466,
"gvpm/clamp_rate": 0.3125,
"gvpm/mask_ratio": 0.3125,
"learning_rate": 6.428571428571429e-07,
"loss": -0.07535895705223083,
"num_turns": 3.6875,
"reward": 0.7963191270828247,
"reward_std": 0.33240535855293274,
"rewards/CADCDValueReward/mean": 0.009158202446997166,
"rewards/CADCDValueReward/std": 0.020484509877860546,
"rewards/CADChamferReward/mean": 0.3119441270828247,
"rewards/CADChamferReward/std": 0.3180331885814667,
"rewards/CADFormatReward/mean": 0.9609375,
"rewards/CADFormatReward/std": 0.13524486124515533,
"rewards/CADInvalidReward/mean": 0.1171875,
"rewards/CADInvalidReward/std": 0.3121146187186241,
"rewards/CADProgressReward/mean": 0.007812500232830644,
"rewards/CADProgressReward/std": 0.038841014727950096,
"step": 9,
"step_time": 813.0814969030325
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 8192.0,
"completions/mean_length": 2162.5234375,
"completions/min_length": 805.5,
"epoch": 0.07407407407407407,
"frac_reward_zero_std": 0.0625,
"grad_norm": 0.36699575014626085,
"gvpm/clamp_rate": 0.5625,
"gvpm/mask_ratio": 0.5625,
"learning_rate": 7.142857142857143e-07,
"loss": -0.15485888719558716,
"num_turns": 3.4765625,
"reward": 0.8366019427776337,
"reward_std": 0.38951922953128815,
"rewards/CADCDValueReward/mean": 0.015387039165943861,
"rewards/CADCDValueReward/std": 0.026676760986447334,
"rewards/CADChamferReward/mean": 0.3311332166194916,
"rewards/CADChamferReward/std": 0.40434587001800537,
"rewards/CADFormatReward/mean": 0.9765625,
"rewards/CADFormatReward/std": 0.1501840502023697,
"rewards/CADInvalidReward/mean": 0.1171875,
"rewards/CADInvalidReward/std": 0.32395489513874054,
"rewards/CADProgressReward/mean": 0.03437500074505806,
"rewards/CADProgressReward/std": 0.07593604549765587,
"step": 10,
"step_time": 735.5644311354845
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 4231.5,
"completions/mean_length": 1854.859375,
"completions/min_length": 562.0,
"epoch": 0.08148148148148149,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.2706968612002178,
"gvpm/clamp_rate": 0.4375,
"gvpm/mask_ratio": 0.4375,
"learning_rate": 7.857142857142856e-07,
"loss": -0.05570477992296219,
"num_turns": 3.375,
"reward": 0.8794412016868591,
"reward_std": 0.27011872828006744,
"rewards/CADCDValueReward/mean": 0.005848402855917811,
"rewards/CADCDValueReward/std": 0.00792743917554617,
"rewards/CADChamferReward/mean": 0.37397244572639465,
"rewards/CADChamferReward/std": 0.3982246369123459,
"rewards/CADFormatReward/mean": 1.0,
"rewards/CADFormatReward/std": 0.0,
"rewards/CADInvalidReward/mean": 0.09375,
"rewards/CADInvalidReward/std": 0.2925330847501755,
"rewards/CADProgressReward/mean": 0.010937500279396772,
"rewards/CADProgressReward/std": 0.045701704919338226,
"step": 11,
"step_time": 704.3677121474757
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 7880.0,
"completions/mean_length": 1946.6953125,
"completions/min_length": 541.0,
"epoch": 0.08888888888888889,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.27335119228951726,
"gvpm/clamp_rate": 0.125,
"gvpm/mask_ratio": 0.125,
"learning_rate": 8.57142857142857e-07,
"loss": -0.11574423313140869,
"num_turns": 3.25,
"reward": 0.9926949441432953,
"reward_std": 0.3828935921192169,
"rewards/CADCDValueReward/mean": 0.013384777121245861,
"rewards/CADCDValueReward/std": 0.023667567409574986,
"rewards/CADChamferReward/mean": 0.4809761345386505,
"rewards/CADChamferReward/std": 0.43272586166858673,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.08768405020236969,
"rewards/CADInvalidReward/mean": 0.0546875,
"rewards/CADInvalidReward/std": 0.22292891144752502,
"rewards/CADProgressReward/mean": 0.0390625,
"rewards/CADProgressReward/std": 0.07929188758134842,
"step": 12,
"step_time": 641.6010130060604
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 6145.0,
"completions/mean_length": 1847.7109375,
"completions/min_length": 546.5,
"epoch": 0.0962962962962963,
"frac_reward_zero_std": 0.125,
"grad_norm": 0.321418259989506,
"gvpm/clamp_rate": 0.375,
"gvpm/mask_ratio": 0.375,
"learning_rate": 9.285714285714285e-07,
"loss": -0.0536077544093132,
"num_turns": 3.3046875,
"reward": 0.8580175042152405,
"reward_std": 0.29149574041366577,
"rewards/CADCDValueReward/mean": 0.017112423665821552,
"rewards/CADCDValueReward/std": 0.027811897918581963,
"rewards/CADChamferReward/mean": 0.3533300310373306,
"rewards/CADChamferReward/std": 0.42514142394065857,
"rewards/CADFormatReward/mean": 0.9921875,
"rewards/CADFormatReward/std": 0.0625,
"rewards/CADInvalidReward/mean": 0.0234375,
"rewards/CADInvalidReward/std": 0.1501840502023697,
"rewards/CADProgressReward/mean": 0.01718750037252903,
"rewards/CADProgressReward/std": 0.05642745830118656,
"step": 13,
"step_time": 648.2220064480789
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 5874.5,
"completions/mean_length": 1644.7265625,
"completions/min_length": 559.0,
"epoch": 0.1037037037037037,
"frac_reward_zero_std": 0.0625,
"grad_norm": 0.3468778785457278,
"gvpm/clamp_rate": 0.125,
"gvpm/mask_ratio": 0.125,
"learning_rate": 1e-06,
"loss": -0.11274628341197968,
"num_turns": 3.0859375,
"reward": 0.9221363961696625,
"reward_std": 0.3210814744234085,
"rewards/CADCDValueReward/mean": 0.009630282409489155,
"rewards/CADCDValueReward/std": 0.018014184664934874,
"rewards/CADChamferReward/mean": 0.4127614051103592,
"rewards/CADChamferReward/std": 0.39582359790802,
"rewards/CADFormatReward/mean": 0.9921875,
"rewards/CADFormatReward/std": 0.0625,
"rewards/CADInvalidReward/mean": 0.0625,
"rewards/CADInvalidReward/std": 0.21978822350502014,
"rewards/CADProgressReward/mean": 0.02656250074505806,
"rewards/CADProgressReward/std": 0.06805390119552612,
"step": 14,
"step_time": 563.1225659624906
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 6446.0,
"completions/mean_length": 1554.78125,
"completions/min_length": 406.5,
"epoch": 0.1111111111111111,
"frac_reward_zero_std": 0.0625,
"grad_norm": 0.38267769429680915,
"gvpm/clamp_rate": 0.125,
"gvpm/mask_ratio": 0.125,
"learning_rate": 9.999623509195722e-07,
"loss": -0.10565665364265442,
"num_turns": 3.0859375,
"reward": 0.8347965478897095,
"reward_std": 0.3865255117416382,
"rewards/CADCDValueReward/mean": 0.02146594040095806,
"rewards/CADCDValueReward/std": 0.026481594890356064,
"rewards/CADChamferReward/mean": 0.3308902978897095,
"rewards/CADChamferReward/std": 0.4167514145374298,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.08768405020236969,
"rewards/CADInvalidReward/mean": 0.0703125,
"rewards/CADInvalidReward/std": 0.2572323754429817,
"rewards/CADProgressReward/mean": 0.023437500931322575,
"rewards/CADProgressReward/std": 0.06479097530245781,
"step": 15,
"step_time": 674.7883856399567
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 8192.0,
"completions/mean_length": 1760.765625,
"completions/min_length": 592.5,
"epoch": 0.11851851851851852,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.3103175454784033,
"gvpm/clamp_rate": 0.1875,
"gvpm/mask_ratio": 0.1875,
"learning_rate": 9.99849409348102e-07,
"loss": -0.060198090970516205,
"num_turns": 3.2578125,
"reward": 0.8692543208599091,
"reward_std": 0.30393216013908386,
"rewards/CADCDValueReward/mean": 0.012020858936011791,
"rewards/CADCDValueReward/std": 0.020353389903903008,
"rewards/CADChamferReward/mean": 0.36925433576107025,
"rewards/CADChamferReward/std": 0.4049799144268036,
"rewards/CADFormatReward/mean": 0.9765625,
"rewards/CADFormatReward/std": 0.1501840502023697,
"rewards/CADInvalidReward/mean": 0.109375,
"rewards/CADInvalidReward/std": 0.31043608486652374,
"rewards/CADProgressReward/mean": 0.023437500931322575,
"rewards/CADProgressReward/std": 0.06479097530245781,
"step": 16,
"step_time": 877.825865128485
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 4194.5,
"completions/mean_length": 1559.34375,
"completions/min_length": 447.5,
"epoch": 0.1259259259259259,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.3371184986908411,
"gvpm/clamp_rate": 0.25,
"gvpm/mask_ratio": 0.25,
"learning_rate": 9.996611922941747e-07,
"loss": -0.05865233764052391,
"num_turns": 3.1015625,
"reward": 1.0996561646461487,
"reward_std": 0.33785510063171387,
"rewards/CADCDValueReward/mean": 0.00933207105845213,
"rewards/CADCDValueReward/std": 0.02142926584929228,
"rewards/CADChamferReward/mean": 0.5777811706066132,
"rewards/CADChamferReward/std": 0.4316692054271698,
"rewards/CADFormatReward/mean": 1.0,
"rewards/CADFormatReward/std": 0.0,
"rewards/CADInvalidReward/mean": 0.0234375,
"rewards/CADInvalidReward/std": 0.1501840502023697,
"rewards/CADProgressReward/mean": 0.04375000111758709,
"rewards/CADProgressReward/std": 0.08254111930727959,
"step": 17,
"step_time": 645.3785339095048
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 8694.5,
"completions/mean_length": 2302.1484375,
"completions/min_length": 492.5,
"epoch": 0.13333333333333333,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.28539059561280866,
"gvpm/clamp_rate": 0.1875,
"gvpm/mask_ratio": 0.1875,
"learning_rate": 9.993977281025862e-07,
"loss": -0.076561838388443,
"num_turns": 3.484375,
"reward": 0.983812689781189,
"reward_std": 0.3841312676668167,
"rewards/CADCDValueReward/mean": 0.006533455103635788,
"rewards/CADCDValueReward/std": 0.014584028162062168,
"rewards/CADChamferReward/mean": 0.48537518084049225,
"rewards/CADChamferReward/std": 0.38363128900527954,
"rewards/CADFormatReward/mean": 0.96875,
"rewards/CADFormatReward/std": 0.17536810040473938,
"rewards/CADInvalidReward/mean": 0.109375,
"rewards/CADInvalidReward/std": 0.3145764470100403,
"rewards/CADProgressReward/mean": 0.02812500111758709,
"rewards/CADProgressReward/std": 0.07007648795843124,
"step": 18,
"step_time": 750.2256326780189
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 8826.0,
"completions/mean_length": 1763.71875,
"completions/min_length": 552.0,
"epoch": 0.14074074074074075,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.2941094542903187,
"gvpm/clamp_rate": 0.5,
"gvpm/mask_ratio": 0.5,
"learning_rate": 9.990590564500746e-07,
"loss": -0.056258171796798706,
"num_turns": 3.125,
"reward": 1.011709988117218,
"reward_std": 0.35970689356327057,
"rewards/CADCDValueReward/mean": 0.011950051411986351,
"rewards/CADCDValueReward/std": 0.02325180172920227,
"rewards/CADChamferReward/mean": 0.5046786963939667,
"rewards/CADChamferReward/std": 0.42988260090351105,
"rewards/CADFormatReward/mean": 0.9765625,
"rewards/CADFormatReward/std": 0.1501840502023697,
"rewards/CADInvalidReward/mean": 0.0625,
"rewards/CADInvalidReward/std": 0.24176587909460068,
"rewards/CADProgressReward/mean": 0.03749999962747097,
"rewards/CADProgressReward/std": 0.07857595384120941,
"step": 19,
"step_time": 689.71919516247
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 8192.0,
"completions/mean_length": 1588.125,
"completions/min_length": 488.5,
"epoch": 0.14814814814814814,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.2945812214682008,
"gvpm/clamp_rate": 0.3125,
"gvpm/mask_ratio": 0.3125,
"learning_rate": 9.986452283393451e-07,
"loss": -0.09122254699468613,
"num_turns": 3.0390625,
"reward": 0.8520500063896179,
"reward_std": 0.3548283874988556,
"rewards/CADCDValueReward/mean": 0.01876841951161623,
"rewards/CADCDValueReward/std": 0.030105404555797577,
"rewards/CADChamferReward/mean": 0.3512687683105469,
"rewards/CADChamferReward/std": 0.40571722388267517,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.125,
"rewards/CADInvalidReward/mean": 0.046875,
"rewards/CADInvalidReward/std": 0.20967156440019608,
"rewards/CADProgressReward/mean": 0.01718750037252903,
"rewards/CADProgressReward/std": 0.05642745643854141,
"step": 20,
"step_time": 668.8609365549055
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 5942.0,
"completions/mean_length": 1437.421875,
"completions/min_length": 425.5,
"epoch": 0.15555555555555556,
"frac_reward_zero_std": 0.1875,
"grad_norm": 0.2422842709739306,
"gvpm/clamp_rate": 0.375,
"gvpm/mask_ratio": 0.375,
"learning_rate": 9.981563060913889e-07,
"loss": -0.03578227385878563,
"num_turns": 2.859375,
"reward": 1.0359516143798828,
"reward_std": 0.25851575285196304,
"rewards/CADCDValueReward/mean": 0.010054457932710648,
"rewards/CADCDValueReward/std": 0.020625630393624306,
"rewards/CADChamferReward/mean": 0.5242328643798828,
"rewards/CADChamferReward/std": 0.46253058314323425,
"rewards/CADFormatReward/mean": 0.9921875,
"rewards/CADFormatReward/std": 0.0625,
"rewards/CADInvalidReward/mean": 0.015625,
"rewards/CADInvalidReward/std": 0.08768405020236969,
"rewards/CADProgressReward/mean": 0.031249999068677425,
"rewards/CADProgressReward/std": 0.0720081739127636,
"step": 21,
"step_time": 565.7626083085197
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 8192.0,
"completions/mean_length": 2412.03125,
"completions/min_length": 924.0,
"epoch": 0.16296296296296298,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.2306572381778335,
"gvpm/clamp_rate": 0.3125,
"gvpm/mask_ratio": 0.3125,
"learning_rate": 9.975923633360984e-07,
"loss": -0.07813145220279694,
"num_turns": 3.6171875,
"reward": 0.8644551336765289,
"reward_std": 0.3148357570171356,
"rewards/CADCDValueReward/mean": 0.0105775217525661,
"rewards/CADCDValueReward/std": 0.019866307266056538,
"rewards/CADChamferReward/mean": 0.3644551634788513,
"rewards/CADChamferReward/std": 0.37886635959148407,
"rewards/CADFormatReward/mean": 0.9765625,
"rewards/CADFormatReward/std": 0.1501840502023697,
"rewards/CADInvalidReward/mean": 0.0859375,
"rewards/CADInvalidReward/std": 0.2792757377028465,
"rewards/CADProgressReward/mean": 0.023437500931322575,
"rewards/CADProgressReward/std": 0.06441672891378403,
"step": 22,
"step_time": 894.0683379225084
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 4819.5,
"completions/mean_length": 1781.421875,
"completions/min_length": 390.0,
"epoch": 0.17037037037037037,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.2569209241985606,
"gvpm/clamp_rate": 0.375,
"gvpm/mask_ratio": 0.375,
"learning_rate": 9.96953485001178e-07,
"loss": -0.13267382979393005,
"num_turns": 3.3828125,
"reward": 1.056757390499115,
"reward_std": 0.30872842669487,
"rewards/CADCDValueReward/mean": 0.008921411354094744,
"rewards/CADCDValueReward/std": 0.018665974959731102,
"rewards/CADChamferReward/mean": 0.5434761941432953,
"rewards/CADChamferReward/std": 0.4205426275730133,
"rewards/CADFormatReward/mean": 1.0,
"rewards/CADFormatReward/std": 0.0,
"rewards/CADInvalidReward/mean": 0.0546875,
"rewards/CADInvalidReward/std": 0.22850853204727173,
"rewards/CADProgressReward/mean": 0.02656250074505806,
"rewards/CADProgressReward/std": 0.06837157532572746,
"step": 23,
"step_time": 838.0536323420238
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 6497.0,
"completions/mean_length": 2079.6484375,
"completions/min_length": 720.0,
"epoch": 0.17777777777777778,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.25247172099517995,
"gvpm/clamp_rate": 0.3125,
"gvpm/mask_ratio": 0.3125,
"learning_rate": 9.96239767299355e-07,
"loss": -0.07227814197540283,
"num_turns": 3.640625,
"reward": 1.0185624957084656,
"reward_std": 0.3580661565065384,
"rewards/CADCDValueReward/mean": 0.006263577379286289,
"rewards/CADCDValueReward/std": 0.01340128155425191,
"rewards/CADChamferReward/mean": 0.5115312337875366,
"rewards/CADChamferReward/std": 0.3675832748413086,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.08768405020236969,
"rewards/CADInvalidReward/mean": 0.078125,
"rewards/CADInvalidReward/std": 0.2543507218360901,
"rewards/CADProgressReward/mean": 0.029687500558793545,
"rewards/CADProgressReward/std": 0.07079743221402168,
"step": 24,
"step_time": 723.5349406970199
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 6039.0,
"completions/mean_length": 1637.6484375,
"completions/min_length": 402.5,
"epoch": 0.18518518518518517,
"frac_reward_zero_std": 0.0625,
"grad_norm": 0.33133232062197165,
"gvpm/clamp_rate": 0.5625,
"gvpm/mask_ratio": 0.5625,
"learning_rate": 9.9545131771389e-07,
"loss": -0.11317317932844162,
"num_turns": 3.1875,
"reward": 1.029203176498413,
"reward_std": 0.3615288585424423,
"rewards/CADCDValueReward/mean": 0.01225523091852665,
"rewards/CADCDValueReward/std": 0.02145096193999052,
"rewards/CADChamferReward/mean": 0.515140637755394,
"rewards/CADChamferReward/std": 0.4296441823244095,
"rewards/CADFormatReward/mean": 0.9921875,
"rewards/CADFormatReward/std": 0.0625,
"rewards/CADInvalidReward/mean": 0.0546875,
"rewards/CADInvalidReward/std": 0.22292891144752502,
"rewards/CADProgressReward/mean": 0.03593750111758709,
"rewards/CADProgressReward/std": 0.07714678719639778,
"step": 25,
"step_time": 616.4760901704431
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 8192.0,
"completions/mean_length": 1895.828125,
"completions/min_length": 482.0,
"epoch": 0.1925925925925926,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.3176523032612676,
"gvpm/clamp_rate": 0.75,
"gvpm/mask_ratio": 0.75,
"learning_rate": 9.945882549823904e-07,
"loss": -0.05554371699690819,
"num_turns": 3.3515625,
"reward": 0.8108876049518585,
"reward_std": 0.3796968460083008,
"rewards/CADCDValueReward/mean": 0.01634395821020007,
"rewards/CADCDValueReward/std": 0.027299873530864716,
"rewards/CADChamferReward/mean": 0.3187001198530197,
"rewards/CADChamferReward/std": 0.4155423492193222,
"rewards/CADFormatReward/mean": 0.96875,
"rewards/CADFormatReward/std": 0.16902101784944534,
"rewards/CADInvalidReward/mean": 0.1171875,
"rewards/CADInvalidReward/std": 0.32395489513874054,
"rewards/CADProgressReward/mean": 0.015625,
"rewards/CADProgressReward/std": 0.054097944870591164,
"step": 26,
"step_time": 784.9924379324657
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.046875,
"completions/max_length": 8604.0,
"completions/mean_length": 2056.6640625,
"completions/min_length": 621.0,
"epoch": 0.2,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.264799714555237,
"gvpm/clamp_rate": 0.0,
"gvpm/mask_ratio": 0.0,
"learning_rate": 9.936507090789292e-07,
"loss": -0.061725594103336334,
"num_turns": 3.28125,
"reward": 1.0490643382072449,
"reward_std": 0.3362731784582138,
"rewards/CADCDValueReward/mean": 0.006162935635074973,
"rewards/CADCDValueReward/std": 0.014397912658751011,
"rewards/CADChamferReward/mean": 0.5553143322467804,
"rewards/CADChamferReward/std": 0.38265109062194824,
"rewards/CADFormatReward/mean": 0.953125,
"rewards/CADFormatReward/std": 0.20967156440019608,
"rewards/CADInvalidReward/mean": 0.09375,
"rewards/CADInvalidReward/std": 0.2925330847501755,
"rewards/CADProgressReward/mean": 0.03437500074505806,
"rewards/CADProgressReward/std": 0.07605084776878357,
"step": 27,
"step_time": 719.3276593934861
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 8192.0,
"completions/mean_length": 2122.7421875,
"completions/min_length": 783.5,
"epoch": 0.2074074074074074,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.31592059791818744,
"gvpm/clamp_rate": 0.0625,
"gvpm/mask_ratio": 0.0625,
"learning_rate": 9.926388211944704e-07,
"loss": -0.08546764403581619,
"num_turns": 3.53125,
"reward": 0.9773533344268799,
"reward_std": 0.38023094832897186,
"rewards/CADCDValueReward/mean": 0.012124743778258562,
"rewards/CADCDValueReward/std": 0.025397044606506824,
"rewards/CADChamferReward/mean": 0.47813452780246735,
"rewards/CADChamferReward/std": 0.4035791605710983,
"rewards/CADFormatReward/mean": 0.9765625,
"rewards/CADFormatReward/std": 0.1501840502023697,
"rewards/CADInvalidReward/mean": 0.109375,
"rewards/CADInvalidReward/std": 0.31043608486652374,
"rewards/CADProgressReward/mean": 0.021875000093132257,
"rewards/CADProgressReward/std": 0.06099376082420349,
"step": 28,
"step_time": 725.8918347665458
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 8192.0,
"completions/mean_length": 2083.890625,
"completions/min_length": 564.0,
"epoch": 0.21481481481481482,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.30025494404400316,
"gvpm/clamp_rate": 0.0625,
"gvpm/mask_ratio": 0.0625,
"learning_rate": 9.915527437156081e-07,
"loss": -0.0366211012005806,
"num_turns": 3.5078125,
"reward": 0.9832450449466705,
"reward_std": 0.35620221495628357,
"rewards/CADCDValueReward/mean": 0.0032473006285727024,
"rewards/CADCDValueReward/std": 0.004252071725204587,
"rewards/CADChamferReward/mean": 0.47933876514434814,
"rewards/CADChamferReward/std": 0.37349840998649597,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.125,
"rewards/CADInvalidReward/mean": 0.09375,
"rewards/CADInvalidReward/std": 0.2925330847501755,
"rewards/CADProgressReward/mean": 0.023437500931322575,
"rewards/CADProgressReward/std": 0.06479097902774811,
"step": 29,
"step_time": 787.6940262585413
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 4964.5,
"completions/mean_length": 1841.96875,
"completions/min_length": 529.5,
"epoch": 0.2222222222222222,
"frac_reward_zero_std": 0.0625,
"grad_norm": 0.27354791426998043,
"gvpm/clamp_rate": 0.1875,
"gvpm/mask_ratio": 0.1875,
"learning_rate": 9.90392640201615e-07,
"loss": -0.059005916118621826,
"num_turns": 3.3984375,
"reward": 0.9467656910419464,
"reward_std": 0.2836601436138153,
"rewards/CADCDValueReward/mean": 0.010562224313616753,
"rewards/CADCDValueReward/std": 0.02116239257156849,
"rewards/CADChamferReward/mean": 0.43817196786403656,
"rewards/CADChamferReward/std": 0.390131875872612,
"rewards/CADFormatReward/mean": 1.0,
"rewards/CADFormatReward/std": 0.0,
"rewards/CADInvalidReward/mean": 0.015625,
"rewards/CADInvalidReward/std": 0.08768405020236969,
"rewards/CADProgressReward/mean": 0.017187499906867743,
"rewards/CADProgressReward/std": 0.055855147540569305,
"step": 30,
"step_time": 644.6225378446397
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 6864.5,
"completions/mean_length": 1962.7890625,
"completions/min_length": 524.5,
"epoch": 0.22962962962962963,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.258167449583024,
"gvpm/clamp_rate": 0.25,
"gvpm/mask_ratio": 0.25,
"learning_rate": 9.891586853598138e-07,
"loss": -0.07502081990242004,
"num_turns": 3.1328125,
"reward": 0.9736192226409912,
"reward_std": 0.31906336545944214,
"rewards/CADCDValueReward/mean": 0.005501617211848497,
"rewards/CADCDValueReward/std": 0.012062105350196362,
"rewards/CADChamferReward/mean": 0.47986921668052673,
"rewards/CADChamferReward/std": 0.37294983863830566,
"rewards/CADFormatReward/mean": 0.9765625,
"rewards/CADFormatReward/std": 0.10652101784944534,
"rewards/CADInvalidReward/mean": 0.0625,
"rewards/CADInvalidReward/std": 0.24176587909460068,
"rewards/CADProgressReward/mean": 0.010937500046566129,
"rewards/CADProgressReward/std": 0.044585783034563065,
"step": 31,
"step_time": 652.3193422019831
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 5661.0,
"completions/mean_length": 1726.6953125,
"completions/min_length": 604.5,
"epoch": 0.23703703703703705,
"frac_reward_zero_std": 0.125,
"grad_norm": 0.2729273570380658,
"gvpm/clamp_rate": 0.25,
"gvpm/mask_ratio": 0.25,
"learning_rate": 9.878510650192642e-07,
"loss": -0.05360779911279678,
"num_turns": 3.1328125,
"reward": 1.009830892086029,
"reward_std": 0.27592554688453674,
"rewards/CADCDValueReward/mean": 0.010024767369031906,
"rewards/CADCDValueReward/std": 0.022714214399456978,
"rewards/CADChamferReward/mean": 0.5020184218883514,
"rewards/CADChamferReward/std": 0.4194524884223938,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.08768405020236969,
"rewards/CADInvalidReward/mean": 0.0234375,
"rewards/CADInvalidReward/std": 0.10652101784944534,
"rewards/CADProgressReward/mean": 0.031250000931322575,
"rewards/CADProgressReward/std": 0.0730636678636074,
"step": 32,
"step_time": 610.3340677325032
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 6038.5,
"completions/mean_length": 1980.5625,
"completions/min_length": 351.0,
"epoch": 0.24444444444444444,
"frac_reward_zero_std": 0.125,
"grad_norm": 0.3199160524824189,
"gvpm/clamp_rate": 0.1875,
"gvpm/mask_ratio": 0.1875,
"learning_rate": 9.8646997610278e-07,
"loss": -0.06718609482049942,
"num_turns": 3.375,
"reward": 0.8588527143001556,
"reward_std": 0.29866546392440796,
"rewards/CADCDValueReward/mean": 0.005571319488808513,
"rewards/CADCDValueReward/std": 0.008845038712024689,
"rewards/CADChamferReward/mean": 0.3557277321815491,
"rewards/CADChamferReward/std": 0.344767764210701,
"rewards/CADFormatReward/mean": 0.9921875,
"rewards/CADFormatReward/std": 0.0625,
"rewards/CADInvalidReward/mean": 0.0546875,
"rewards/CADInvalidReward/std": 0.20939241349697113,
"rewards/CADProgressReward/mean": 0.014062500093132257,
"rewards/CADProgressReward/std": 0.051446475088596344,
"step": 33,
"step_time": 721.2837582180509
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 6548.5,
"completions/mean_length": 1799.359375,
"completions/min_length": 566.5,
"epoch": 0.2518518518518518,
"frac_reward_zero_std": 0.0625,
"grad_norm": 0.28745768475077366,
"gvpm/clamp_rate": 0.5625,
"gvpm/mask_ratio": 0.5625,
"learning_rate": 9.85015626597272e-07,
"loss": -0.0849800705909729,
"num_turns": 3.3046875,
"reward": 0.8399341702461243,
"reward_std": 0.28047215938568115,
"rewards/CADCDValueReward/mean": 0.012512448243796825,
"rewards/CADCDValueReward/std": 0.020015444606542587,
"rewards/CADChamferReward/mean": 0.33212170004844666,
"rewards/CADChamferReward/std": 0.41149912774562836,
"rewards/CADFormatReward/mean": 0.9921875,
"rewards/CADFormatReward/std": 0.0625,
"rewards/CADInvalidReward/mean": 0.0625,
"rewards/CADInvalidReward/std": 0.24176587909460068,
"rewards/CADProgressReward/mean": 0.0234375,
"rewards/CADProgressReward/std": 0.06441672891378403,
"step": 34,
"step_time": 743.4361940279487
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 8192.0,
"completions/mean_length": 1665.359375,
"completions/min_length": 444.5,
"epoch": 0.25925925925925924,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.31312486042551846,
"gvpm/clamp_rate": 0.25,
"gvpm/mask_ratio": 0.25,
"learning_rate": 9.83488235522426e-07,
"loss": -0.0926455706357956,
"num_turns": 3.015625,
"reward": 1.0441208481788635,
"reward_std": 0.43992879986763,
"rewards/CADCDValueReward/mean": 0.010189782828092575,
"rewards/CADCDValueReward/std": 0.020550732035189867,
"rewards/CADChamferReward/mean": 0.5284958332777023,
"rewards/CADChamferReward/std": 0.42446738481521606,
"rewards/CADFormatReward/mean": 0.9765625,
"rewards/CADFormatReward/std": 0.1501840502023697,
"rewards/CADInvalidReward/mean": 0.0546875,
"rewards/CADInvalidReward/std": 0.22850853204727173,
"rewards/CADProgressReward/mean": 0.05468750186264515,
"rewards/CADProgressReward/std": 0.08842188864946365,
"step": 35,
"step_time": 662.2956334860064
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 7357.5,
"completions/mean_length": 1978.1484375,
"completions/min_length": 509.5,
"epoch": 0.26666666666666666,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.29081684621406195,
"gvpm/clamp_rate": 0.125,
"gvpm/mask_ratio": 0.125,
"learning_rate": 9.8188803289772e-07,
"loss": -0.08506050705909729,
"num_turns": 3.265625,
"reward": 1.035053789615631,
"reward_std": 0.295712873339653,
"rewards/CADCDValueReward/mean": 0.006636728765442967,
"rewards/CADCDValueReward/std": 0.017200497444719076,
"rewards/CADChamferReward/mean": 0.5288038104772568,
"rewards/CADChamferReward/std": 0.438412606716156,
"rewards/CADFormatReward/mean": 0.9765625,
"rewards/CADFormatReward/std": 0.10652101784944534,
"rewards/CADInvalidReward/mean": 0.0859375,
"rewards/CADInvalidReward/std": 0.2628752738237381,
"rewards/CADProgressReward/mean": 0.03593749925494194,
"rewards/CADProgressReward/std": 0.07714678719639778,
"step": 36,
"step_time": 808.6509420100483
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 8671.5,
"completions/mean_length": 2072.84375,
"completions/min_length": 559.5,
"epoch": 0.2740740740740741,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.3353417284638203,
"gvpm/clamp_rate": 0.25,
"gvpm/mask_ratio": 0.25,
"learning_rate": 9.802152597077828e-07,
"loss": -0.07194460928440094,
"num_turns": 3.3671875,
"reward": 1.0173083543777466,
"reward_std": 0.40730439126491547,
"rewards/CADCDValueReward/mean": 0.004399771336466074,
"rewards/CADCDValueReward/std": 0.008308042772114277,
"rewards/CADChamferReward/mean": 0.5134020745754242,
"rewards/CADChamferReward/std": 0.3815064877271652,
"rewards/CADFormatReward/mean": 0.96875,
"rewards/CADFormatReward/std": 0.17536810040473938,
"rewards/CADInvalidReward/mean": 0.09375,
"rewards/CADInvalidReward/std": 0.29378482699394226,
"rewards/CADProgressReward/mean": 0.03906250186264515,
"rewards/CADProgressReward/std": 0.07969209179282188,
"step": 37,
"step_time": 753.216583872505
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 3918.0,
"completions/mean_length": 1757.5703125,
"completions/min_length": 453.5,
"epoch": 0.2814814814814815,
"frac_reward_zero_std": 0.125,
"grad_norm": 0.2589390600255359,
"gvpm/clamp_rate": 0.4375,
"gvpm/mask_ratio": 0.4375,
"learning_rate": 9.784701678661044e-07,
"loss": -0.05084498971700668,
"num_turns": 3.2890625,
"reward": 0.8897316157817841,
"reward_std": 0.24061861634254456,
"rewards/CADCDValueReward/mean": 0.01003743102774024,
"rewards/CADCDValueReward/std": 0.01767158042639494,
"rewards/CADChamferReward/mean": 0.3787941187620163,
"rewards/CADChamferReward/std": 0.3944964110851288,
"rewards/CADFormatReward/mean": 1.0,
"rewards/CADFormatReward/std": 0.0,
"rewards/CADInvalidReward/mean": 0.03125,
"rewards/CADInvalidReward/std": 0.16902101784944534,
"rewards/CADProgressReward/mean": 0.021875000558793545,
"rewards/CADProgressReward/std": 0.06271182000637054,
"step": 38,
"step_time": 626.9985265335417
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 5941.5,
"completions/mean_length": 1885.8515625,
"completions/min_length": 486.0,
"epoch": 0.28888888888888886,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.33316093398613267,
"gvpm/clamp_rate": 0.375,
"gvpm/mask_ratio": 0.375,
"learning_rate": 9.766530201770967e-07,
"loss": -0.10036209970712662,
"num_turns": 3.4609375,
"reward": 0.9943822026252747,
"reward_std": 0.38469935953617096,
"rewards/CADCDValueReward/mean": 0.010812665335834026,
"rewards/CADCDValueReward/std": 0.023920376785099506,
"rewards/CADChamferReward/mean": 0.48422594368457794,
"rewards/CADChamferReward/std": 0.4070392996072769,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.08768405020236969,
"rewards/CADInvalidReward/mean": 0.0859375,
"rewards/CADInvalidReward/std": 0.28213727474212646,
"rewards/CADProgressReward/mean": 0.03593750111758709,
"rewards/CADProgressReward/std": 0.07670491188764572,
"step": 39,
"step_time": 718.7788579615299
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 8192.0,
"completions/mean_length": 1818.8515625,
"completions/min_length": 481.0,
"epoch": 0.2962962962962963,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.31176581838680095,
"gvpm/clamp_rate": 0.0625,
"gvpm/mask_ratio": 0.0625,
"learning_rate": 9.747640902965182e-07,
"loss": -0.04899011552333832,
"num_turns": 3.109375,
"reward": 1.0272006392478943,
"reward_std": 0.39842282235622406,
"rewards/CADCDValueReward/mean": 0.006507785525172949,
"rewards/CADCDValueReward/std": 0.017087630927562714,
"rewards/CADChamferReward/mean": 0.5279819071292877,
"rewards/CADChamferReward/std": 0.4242442846298218,
"rewards/CADFormatReward/mean": 0.9765625,
"rewards/CADFormatReward/std": 0.1501840502023697,
"rewards/CADInvalidReward/mean": 0.09375,
"rewards/CADInvalidReward/std": 0.2925330847501755,
"rewards/CADProgressReward/mean": 0.02187499962747097,
"rewards/CADProgressReward/std": 0.06208721734583378,
"step": 40,
"step_time": 768.1774763509748
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 5956.0,
"completions/mean_length": 1333.609375,
"completions/min_length": 453.0,
"epoch": 0.3037037037037037,
"frac_reward_zero_std": 0.125,
"grad_norm": 0.24401950758696947,
"gvpm/clamp_rate": 0.1875,
"gvpm/mask_ratio": 0.1875,
"learning_rate": 9.728036626902607e-07,
"loss": -0.03415040299296379,
"num_turns": 2.6875,
"reward": 1.1090074181556702,
"reward_std": 0.2634388953447342,
"rewards/CADCDValueReward/mean": 0.005830263951793313,
"rewards/CADCDValueReward/std": 0.009733580984175205,
"rewards/CADChamferReward/mean": 0.6004136800765991,
"rewards/CADChamferReward/std": 0.4362885504961014,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.08768405020236969,
"rewards/CADInvalidReward/mean": 0.0234375,
"rewards/CADInvalidReward/std": 0.10652101784944534,
"rewards/CADProgressReward/mean": 0.03281250037252903,
"rewards/CADProgressReward/std": 0.07437803223729134,
"step": 41,
"step_time": 581.5830892570084
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 8192.0,
"completions/mean_length": 1627.2265625,
"completions/min_length": 508.5,
"epoch": 0.3111111111111111,
"frac_reward_zero_std": 0.0625,
"grad_norm": 0.2133396915352835,
"gvpm/clamp_rate": 0.1875,
"gvpm/mask_ratio": 0.1875,
"learning_rate": 9.707720325915103e-07,
"loss": -0.028079399839043617,
"num_turns": 2.921875,
"reward": 0.9877127707004547,
"reward_std": 0.24002548307180405,
"rewards/CADCDValueReward/mean": 0.00834473967552185,
"rewards/CADCDValueReward/std": 0.015562626533210278,
"rewards/CADChamferReward/mean": 0.48458781838417053,
"rewards/CADChamferReward/std": 0.42006009817123413,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.125,
"rewards/CADInvalidReward/mean": 0.03125,
"rewards/CADInvalidReward/std": 0.16902101784944534,
"rewards/CADProgressReward/mean": 0.021875000558793545,
"rewards/CADProgressReward/std": 0.06271181628108025,
"step": 42,
"step_time": 611.8008063075249
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 8680.0,
"completions/mean_length": 1774.984375,
"completions/min_length": 567.0,
"epoch": 0.31851851851851853,
"frac_reward_zero_std": 0.125,
"grad_norm": 0.3286262792890542,
"gvpm/clamp_rate": 0.3125,
"gvpm/mask_ratio": 0.3125,
"learning_rate": 9.686695059562874e-07,
"loss": -0.06646941602230072,
"num_turns": 3.265625,
"reward": 0.9156647622585297,
"reward_std": 0.3191141039133072,
"rewards/CADCDValueReward/mean": 0.009640823118388653,
"rewards/CADCDValueReward/std": 0.016673089005053043,
"rewards/CADChamferReward/mean": 0.40550853312015533,
"rewards/CADChamferReward/std": 0.42234233021736145,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.125,
"rewards/CADInvalidReward/mean": 0.03125,
"rewards/CADInvalidReward/std": 0.17536810040473938,
"rewards/CADProgressReward/mean": 0.03593750111758709,
"rewards/CADProgressReward/std": 0.07714678719639778,
"step": 43,
"step_time": 663.2222139424994
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.046875,
"completions/max_length": 8192.0,
"completions/mean_length": 2027.703125,
"completions/min_length": 562.0,
"epoch": 0.32592592592592595,
"frac_reward_zero_std": 0.0625,
"grad_norm": 0.3098968483848029,
"gvpm/clamp_rate": 0.3125,
"gvpm/mask_ratio": 0.3125,
"learning_rate": 9.664963994173693e-07,
"loss": -0.05407789349555969,
"num_turns": 3.1171875,
"reward": 0.907115250825882,
"reward_std": 0.3342723697423935,
"rewards/CADCDValueReward/mean": 0.009987911209464073,
"rewards/CADCDValueReward/std": 0.015833669807761908,
"rewards/CADChamferReward/mean": 0.41649025678634644,
"rewards/CADChamferReward/std": 0.3898242264986038,
"rewards/CADFormatReward/mean": 0.953125,
"rewards/CADFormatReward/std": 0.21304203569889069,
"rewards/CADInvalidReward/mean": 0.09375,
"rewards/CADInvalidReward/std": 0.2925330847501755,
"rewards/CADProgressReward/mean": 0.02812500111758709,
"rewards/CADProgressReward/std": 0.06948306784033775,
"step": 44,
"step_time": 686.5821760474937
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 5765.5,
"completions/mean_length": 1786.7109375,
"completions/min_length": 556.5,
"epoch": 0.3333333333333333,
"frac_reward_zero_std": 0.0625,
"grad_norm": 0.29315669725592414,
"gvpm/clamp_rate": 0.0625,
"gvpm/mask_ratio": 0.0625,
"learning_rate": 9.642530402366078e-07,
"loss": 0.0072112735360860825,
"num_turns": 3.3125,
"reward": 0.990590512752533,
"reward_std": 0.32899370789527893,
"rewards/CADCDValueReward/mean": 0.010082908440381289,
"rewards/CADCDValueReward/std": 0.01896606106311083,
"rewards/CADChamferReward/mean": 0.4819967597723007,
"rewards/CADChamferReward/std": 0.4468446969985962,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.08768405020236969,
"rewards/CADInvalidReward/mean": 0.09375,
"rewards/CADInvalidReward/std": 0.2925330847501755,
"rewards/CADProgressReward/mean": 0.03281250037252903,
"rewards/CADProgressReward/std": 0.07462168112397194,
"step": 45,
"step_time": 670.6647546880995
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 9148.0,
"completions/mean_length": 2010.7734375,
"completions/min_length": 527.5,
"epoch": 0.34074074074074073,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.26375500508995686,
"gvpm/clamp_rate": 0.0625,
"gvpm/mask_ratio": 0.0625,
"learning_rate": 9.619397662556433e-07,
"loss": -0.0605773963034153,
"num_turns": 3.3203125,
"reward": 1.0370360612869263,
"reward_std": 0.34883715212345123,
"rewards/CADCDValueReward/mean": 0.005773921264335513,
"rewards/CADCDValueReward/std": 0.01515254145488143,
"rewards/CADChamferReward/mean": 0.5339110493659973,
"rewards/CADChamferReward/std": 0.3714523911476135,
"rewards/CADFormatReward/mean": 0.9765625,
"rewards/CADFormatReward/std": 0.1501840502023697,
"rewards/CADInvalidReward/mean": 0.0546875,
"rewards/CADInvalidReward/std": 0.22850853204727173,
"rewards/CADProgressReward/mean": 0.02968750149011612,
"rewards/CADProgressReward/std": 0.06992901489138603,
"step": 46,
"step_time": 781.8355874035042
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 6850.5,
"completions/mean_length": 2329.1640625,
"completions/min_length": 578.5,
"epoch": 0.34814814814814815,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.25034575161954825,
"gvpm/clamp_rate": 0.4375,
"gvpm/mask_ratio": 0.4375,
"learning_rate": 9.595569258450289e-07,
"loss": -0.03825327008962631,
"num_turns": 3.515625,
"reward": 0.7986271977424622,
"reward_std": 0.287222221493721,
"rewards/CADCDValueReward/mean": 0.011657494585961103,
"rewards/CADCDValueReward/std": 0.02588548604398966,
"rewards/CADChamferReward/mean": 0.30878348648548126,
"rewards/CADChamferReward/std": 0.3143278509378433,
"rewards/CADFormatReward/mean": 0.9765625,
"rewards/CADFormatReward/std": 0.10652101784944534,
"rewards/CADInvalidReward/mean": 0.109375,
"rewards/CADInvalidReward/std": 0.3145764470100403,
"rewards/CADProgressReward/mean": 0.0031250000465661287,
"rewards/CADProgressReward/std": 0.02500000223517418,
"step": 47,
"step_time": 685.2994031485287
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 6124.5,
"completions/mean_length": 1663.7734375,
"completions/min_length": 458.0,
"epoch": 0.35555555555555557,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.28837179688759407,
"gvpm/clamp_rate": 0.0625,
"gvpm/mask_ratio": 0.0625,
"learning_rate": 9.571048778517652e-07,
"loss": -0.0845719426870346,
"num_turns": 3.140625,
"reward": 1.0179510712623596,
"reward_std": 0.28746291995048523,
"rewards/CADCDValueReward/mean": 0.010810192907229066,
"rewards/CADCDValueReward/std": 0.023981335572898388,
"rewards/CADChamferReward/mean": 0.5140447914600372,
"rewards/CADChamferReward/std": 0.38752706348896027,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.08768405020236969,
"rewards/CADInvalidReward/mean": 0.0234375,
"rewards/CADInvalidReward/std": 0.1501840502023697,
"rewards/CADProgressReward/mean": 0.0234375,
"rewards/CADProgressReward/std": 0.06364523060619831,
"step": 48,
"step_time": 639.4493279859889
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 6029.5,
"completions/mean_length": 1708.6953125,
"completions/min_length": 528.0,
"epoch": 0.362962962962963,
"frac_reward_zero_std": 0.125,
"grad_norm": 0.2624397697673845,
"gvpm/clamp_rate": 0.375,
"gvpm/mask_ratio": 0.375,
"learning_rate": 9.545839915452611e-07,
"loss": 0.0031496845185756683,
"num_turns": 3.078125,
"reward": 0.8401894569396973,
"reward_std": 0.2741198167204857,
"rewards/CADCDValueReward/mean": 0.00806389469653368,
"rewards/CADCDValueReward/std": 0.011322741396725178,
"rewards/CADChamferReward/mean": 0.3370644748210907,
"rewards/CADChamferReward/std": 0.39350321888923645,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.125,
"rewards/CADInvalidReward/mean": 0.0703125,
"rewards/CADInvalidReward/std": 0.24497227370738983,
"rewards/CADProgressReward/mean": 0.021875000558793545,
"rewards/CADProgressReward/std": 0.06208721548318863,
"step": 49,
"step_time": 654.7832974030753
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 4623.0,
"completions/mean_length": 1640.65625,
"completions/min_length": 515.5,
"epoch": 0.37037037037037035,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.33510029382490514,
"gvpm/clamp_rate": 0.1875,
"gvpm/mask_ratio": 0.1875,
"learning_rate": 9.519946465617217e-07,
"loss": -0.0020774416625499725,
"num_turns": 3.09375,
"reward": 1.146395742893219,
"reward_std": 0.36798277497291565,
"rewards/CADCDValueReward/mean": 0.005812318995594978,
"rewards/CADCDValueReward/std": 0.017129408195614815,
"rewards/CADChamferReward/mean": 0.6221769750118256,
"rewards/CADChamferReward/std": 0.3998716175556183,
"rewards/CADFormatReward/mean": 1.0,
"rewards/CADFormatReward/std": 0.0,
"rewards/CADInvalidReward/mean": 0.0546875,
"rewards/CADInvalidReward/std": 0.22850853204727173,
"rewards/CADProgressReward/mean": 0.04843750037252903,
"rewards/CADProgressReward/std": 0.0861823633313179,
"step": 50,
"step_time": 639.4267683760263
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 7139.5,
"completions/mean_length": 1918.0390625,
"completions/min_length": 487.0,
"epoch": 0.37777777777777777,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.3791754505354697,
"gvpm/clamp_rate": 0.1875,
"gvpm/mask_ratio": 0.1875,
"learning_rate": 9.493372328469769e-07,
"loss": -0.10320444405078888,
"num_turns": 3.1796875,
"reward": 0.8491643965244293,
"reward_std": 0.3237500786781311,
"rewards/CADCDValueReward/mean": 0.007575992261990905,
"rewards/CADCDValueReward/std": 0.01029264833778143,
"rewards/CADChamferReward/mean": 0.3530706316232681,
"rewards/CADChamferReward/std": 0.3840659260749817,
"rewards/CADFormatReward/mean": 0.96875,
"rewards/CADFormatReward/std": 0.12198751419782639,
"rewards/CADInvalidReward/mean": 0.0625,
"rewards/CADInvalidReward/std": 0.24176587909460068,
"rewards/CADProgressReward/mean": 0.02343750186264515,
"rewards/CADProgressReward/std": 0.06479097902774811,
"step": 51,
"step_time": 592.0421478134813
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 8761.0,
"completions/mean_length": 1802.5625,
"completions/min_length": 428.5,
"epoch": 0.3851851851851852,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.31925761072226416,
"gvpm/clamp_rate": 0.5,
"gvpm/mask_ratio": 0.5,
"learning_rate": 9.466121505977576e-07,
"loss": -0.0613069087266922,
"num_turns": 3.1015625,
"reward": 0.9655038118362427,
"reward_std": 0.340903177857399,
"rewards/CADCDValueReward/mean": 0.007181169930845499,
"rewards/CADCDValueReward/std": 0.012167679611593485,
"rewards/CADChamferReward/mean": 0.45925378799438477,
"rewards/CADChamferReward/std": 0.43096397817134857,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.125,
"rewards/CADInvalidReward/mean": 0.0703125,
"rewards/CADInvalidReward/std": 0.2534134313464165,
"rewards/CADProgressReward/mean": 0.02812500111758709,
"rewards/CADProgressReward/std": 0.06871827319264412,
"step": 52,
"step_time": 752.4245209319633
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 6691.0,
"completions/mean_length": 1669.578125,
"completions/min_length": 589.0,
"epoch": 0.3925925925925926,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.2601323689782601,
"gvpm/clamp_rate": 0.1875,
"gvpm/mask_ratio": 0.1875,
"learning_rate": 9.43819810201427e-07,
"loss": -0.06911112368106842,
"num_turns": 2.921875,
"reward": 1.141762375831604,
"reward_std": 0.28559741377830505,
"rewards/CADCDValueReward/mean": 0.007342529483139515,
"rewards/CADCDValueReward/std": 0.01919363532215357,
"rewards/CADChamferReward/mean": 0.6245749294757843,
"rewards/CADChamferReward/std": 0.3975260406732559,
"rewards/CADFormatReward/mean": 0.9921875,
"rewards/CADFormatReward/std": 0.0625,
"rewards/CADInvalidReward/mean": 0.015625,
"rewards/CADInvalidReward/std": 0.08768405020236969,
"rewards/CADProgressReward/mean": 0.04218750074505806,
"rewards/CADProgressReward/std": 0.08166900277137756,
"step": 53,
"step_time": 562.2636092515313
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 8192.0,
"completions/mean_length": 1722.53125,
"completions/min_length": 392.5,
"epoch": 0.4,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.3375879590417734,
"gvpm/clamp_rate": 0.0,
"gvpm/mask_ratio": 0.0,
"learning_rate": 9.409606321741774e-07,
"loss": -0.013662035576999187,
"num_turns": 3.03125,
"reward": 1.099386751651764,
"reward_std": 0.2912728488445282,
"rewards/CADCDValueReward/mean": 0.006153033580631018,
"rewards/CADCDValueReward/std": 0.015778191853314638,
"rewards/CADChamferReward/mean": 0.592355489730835,
"rewards/CADChamferReward/std": 0.4214763343334198,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.125,
"rewards/CADInvalidReward/mean": 0.03125,
"rewards/CADInvalidReward/std": 0.16902101784944534,
"rewards/CADProgressReward/mean": 0.029687500558793545,
"rewards/CADProgressReward/std": 0.07135875523090363,
"step": 54,
"step_time": 650.6152248754515
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 8643.5,
"completions/mean_length": 2059.3203125,
"completions/min_length": 632.5,
"epoch": 0.4074074074074074,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.3226498289188323,
"gvpm/clamp_rate": 0.4375,
"gvpm/mask_ratio": 0.4375,
"learning_rate": 9.380350470977032e-07,
"loss": -0.07155250757932663,
"num_turns": 3.5,
"reward": 0.9070218205451965,
"reward_std": 0.3886219561100006,
"rewards/CADCDValueReward/mean": 0.011722303926944733,
"rewards/CADCDValueReward/std": 0.025446307845413685,
"rewards/CADChamferReward/mean": 0.41092802584171295,
"rewards/CADChamferReward/std": 0.38270294666290283,
"rewards/CADFormatReward/mean": 0.96875,
"rewards/CADFormatReward/std": 0.16902101784944534,
"rewards/CADInvalidReward/mean": 0.140625,
"rewards/CADInvalidReward/std": 0.3496479392051697,
"rewards/CADProgressReward/mean": 0.023437500931322575,
"rewards/CADProgressReward/std": 0.06479097902774811,
"step": 55,
"step_time": 770.8545176874613
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 7938.0,
"completions/mean_length": 2245.65625,
"completions/min_length": 591.5,
"epoch": 0.4148148148148148,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.3003170703202294,
"gvpm/clamp_rate": 0.75,
"gvpm/mask_ratio": 0.75,
"learning_rate": 9.350434955543557e-07,
"loss": -0.06011253967881203,
"num_turns": 3.5859375,
"reward": 0.8453920781612396,
"reward_std": 0.3145531862974167,
"rewards/CADCDValueReward/mean": 0.01054031541571021,
"rewards/CADCDValueReward/std": 0.02487563155591488,
"rewards/CADChamferReward/mean": 0.33757956326007843,
"rewards/CADChamferReward/std": 0.33816616237163544,
"rewards/CADFormatReward/mean": 0.9921875,
"rewards/CADFormatReward/std": 0.0625,
"rewards/CADInvalidReward/mean": 0.125,
"rewards/CADInvalidReward/std": 0.3333333432674408,
"rewards/CADProgressReward/mean": 0.0234375,
"rewards/CADProgressReward/std": 0.06364523060619831,
"step": 56,
"step_time": 909.908546074992
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 7579.5,
"completions/mean_length": 1823.109375,
"completions/min_length": 408.0,
"epoch": 0.4222222222222222,
"frac_reward_zero_std": 0.0625,
"grad_norm": 0.3200746381649447,
"gvpm/clamp_rate": 0.25,
"gvpm/mask_ratio": 0.25,
"learning_rate": 9.319864280607934e-07,
"loss": -0.05908238887786865,
"num_turns": 3.25,
"reward": 1.0055773258209229,
"reward_std": 0.3037809878587723,
"rewards/CADCDValueReward/mean": 0.011222207453101873,
"rewards/CADCDValueReward/std": 0.024534993804991245,
"rewards/CADChamferReward/mean": 0.5055772811174393,
"rewards/CADChamferReward/std": 0.3964270204305649,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.125,
"rewards/CADInvalidReward/mean": 0.046875,
"rewards/CADInvalidReward/std": 0.20967156440019608,
"rewards/CADProgressReward/mean": 0.015625,
"rewards/CADProgressReward/std": 0.05409794673323631,
"step": 57,
"step_time": 756.2379550259211
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 8471.5,
"completions/mean_length": 1748.890625,
"completions/min_length": 585.5,
"epoch": 0.42962962962962964,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.3429335487801308,
"gvpm/clamp_rate": 0.1875,
"gvpm/mask_ratio": 0.1875,
"learning_rate": 9.28864305000136e-07,
"loss": -0.0015911739319562912,
"num_turns": 3.0390625,
"reward": 1.049176573753357,
"reward_std": 0.3722991943359375,
"rewards/CADCDValueReward/mean": 0.00502880965359509,
"rewards/CADCDValueReward/std": 0.00911300745792687,
"rewards/CADChamferReward/mean": 0.5390203297138214,
"rewards/CADChamferReward/std": 0.3947935700416565,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.125,
"rewards/CADInvalidReward/mean": 0.0234375,
"rewards/CADInvalidReward/std": 0.1501840502023697,
"rewards/CADProgressReward/mean": 0.03593750111758709,
"rewards/CADProgressReward/std": 0.07714678719639778,
"step": 58,
"step_time": 714.4565632169833
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 6798.0,
"completions/mean_length": 1776.546875,
"completions/min_length": 473.5,
"epoch": 0.43703703703703706,
"frac_reward_zero_std": 0.125,
"grad_norm": 0.2650050271247873,
"gvpm/clamp_rate": 0.0625,
"gvpm/mask_ratio": 0.0625,
"learning_rate": 9.256775965526326e-07,
"loss": -0.050642773509025574,
"num_turns": 3.3046875,
"reward": 0.9662282764911652,
"reward_std": 0.24029946327209473,
"rewards/CADCDValueReward/mean": 0.007690779631957412,
"rewards/CADCDValueReward/std": 0.016633108258247375,
"rewards/CADChamferReward/mean": 0.45841577649116516,
"rewards/CADChamferReward/std": 0.42017658054828644,
"rewards/CADFormatReward/mean": 0.9921875,
"rewards/CADFormatReward/std": 0.0625,
"rewards/CADInvalidReward/mean": 0.046875,
"rewards/CADInvalidReward/std": 0.21304203569889069,
"rewards/CADProgressReward/mean": 0.023437500931322575,
"rewards/CADProgressReward/std": 0.06479097530245781,
"step": 59,
"step_time": 703.0263737409841
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 5316.5,
"completions/mean_length": 1924.828125,
"completions/min_length": 593.5,
"epoch": 0.4444444444444444,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.2728441543145627,
"gvpm/clamp_rate": 0.3125,
"gvpm/mask_ratio": 0.3125,
"learning_rate": 9.224267826248536e-07,
"loss": -0.0760185718536377,
"num_turns": 3.3046875,
"reward": 0.9995194375514984,
"reward_std": 0.36176206171512604,
"rewards/CADCDValueReward/mean": 0.008715164847671986,
"rewards/CADCDValueReward/std": 0.022952506318688393,
"rewards/CADChamferReward/mean": 0.4815506935119629,
"rewards/CADChamferReward/std": 0.40026241540908813,
"rewards/CADFormatReward/mean": 1.0,
"rewards/CADFormatReward/std": 0.0,
"rewards/CADInvalidReward/mean": 0.03125,
"rewards/CADInvalidReward/std": 0.16902101784944534,
"rewards/CADProgressReward/mean": 0.03593750111758709,
"rewards/CADProgressReward/std": 0.07714678719639778,
"step": 60,
"step_time": 644.8002563769696
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 8472.0,
"completions/mean_length": 2038.0390625,
"completions/min_length": 575.0,
"epoch": 0.45185185185185184,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.21574088610134093,
"gvpm/clamp_rate": 0.4375,
"gvpm/mask_ratio": 0.4375,
"learning_rate": 9.191123527774189e-07,
"loss": -0.06047581881284714,
"num_turns": 3.3828125,
"reward": 0.9609400033950806,
"reward_std": 0.3012053519487381,
"rewards/CADCDValueReward/mean": 0.007246244233101606,
"rewards/CADCDValueReward/std": 0.015529151540249586,
"rewards/CADChamferReward/mean": 0.45859625935554504,
"rewards/CADChamferReward/std": 0.414530873298645,
"rewards/CADFormatReward/mean": 0.9765625,
"rewards/CADFormatReward/std": 0.1501840502023697,
"rewards/CADInvalidReward/mean": 0.0859375,
"rewards/CADInvalidReward/std": 0.2792757377028465,
"rewards/CADProgressReward/mean": 0.028125000186264515,
"rewards/CADProgressReward/std": 0.069929588586092,
"step": 61,
"step_time": 705.3993218655232
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 5964.0,
"completions/mean_length": 1686.34375,
"completions/min_length": 568.5,
"epoch": 0.45925925925925926,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.28915098737436373,
"gvpm/clamp_rate": 0.3125,
"gvpm/mask_ratio": 0.3125,
"learning_rate": 9.157348061512726e-07,
"loss": -0.024678653106093407,
"num_turns": 3.1796875,
"reward": 0.964135080575943,
"reward_std": 0.319868266582489,
"rewards/CADCDValueReward/mean": 0.005077847046777606,
"rewards/CADCDValueReward/std": 0.008623983478173614,
"rewards/CADChamferReward/mean": 0.45944756269454956,
"rewards/CADChamferReward/std": 0.3756243586540222,
"rewards/CADFormatReward/mean": 0.9921875,
"rewards/CADFormatReward/std": 0.0625,
"rewards/CADInvalidReward/mean": 0.0546875,
"rewards/CADInvalidReward/std": 0.22292891144752502,
"rewards/CADProgressReward/mean": 0.01718750037252903,
"rewards/CADProgressReward/std": 0.05642745830118656,
"step": 62,
"step_time": 673.9347062259912
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 3676.0,
"completions/mean_length": 1735.6171875,
"completions/min_length": 523.0,
"epoch": 0.4666666666666667,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.3370947637179292,
"gvpm/clamp_rate": 0.25,
"gvpm/mask_ratio": 0.25,
"learning_rate": 9.122946513925126e-07,
"loss": -0.058004140853881836,
"num_turns": 3.3203125,
"reward": 0.9642046689987183,
"reward_std": 0.37482963502407074,
"rewards/CADCDValueReward/mean": 0.015951286535710096,
"rewards/CADCDValueReward/std": 0.037708403542637825,
"rewards/CADChamferReward/mean": 0.4509234130382538,
"rewards/CADChamferReward/std": 0.3834534287452698,
"rewards/CADFormatReward/mean": 1.0,
"rewards/CADFormatReward/std": 0.0,
"rewards/CADInvalidReward/mean": 0.1015625,
"rewards/CADInvalidReward/std": 0.30191153287887573,
"rewards/CADProgressReward/mean": 0.026562499813735485,
"rewards/CADProgressReward/std": 0.06837157532572746,
"step": 63,
"step_time": 662.9919982755673
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 3434.0,
"completions/mean_length": 1800.2421875,
"completions/min_length": 700.0,
"epoch": 0.4740740740740741,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.34419399180551935,
"gvpm/clamp_rate": 0.1875,
"gvpm/mask_ratio": 0.1875,
"learning_rate": 9.087924065757918e-07,
"loss": -0.02677270956337452,
"num_turns": 3.25,
"reward": 1.0374482870101929,
"reward_std": 0.334428071975708,
"rewards/CADCDValueReward/mean": 0.00554040796123445,
"rewards/CADCDValueReward/std": 0.01146679138764739,
"rewards/CADChamferReward/mean": 0.5218233019113541,
"rewards/CADChamferReward/std": 0.34231896698474884,
"rewards/CADFormatReward/mean": 1.0,
"rewards/CADFormatReward/std": 0.0,
"rewards/CADInvalidReward/mean": 0.0078125,
"rewards/CADInvalidReward/std": 0.0625,
"rewards/CADProgressReward/mean": 0.03125,
"rewards/CADProgressReward/std": 0.06974460370838642,
"step": 64,
"step_time": 589.4235910624848
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 6190.0,
"completions/mean_length": 2119.671875,
"completions/min_length": 610.5,
"epoch": 0.48148148148148145,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.2546129693778692,
"gvpm/clamp_rate": 0.25,
"gvpm/mask_ratio": 0.25,
"learning_rate": 9.052285991262974e-07,
"loss": -0.03790409862995148,
"num_turns": 3.5234375,
"reward": 0.9496057033538818,
"reward_std": 0.28532589972019196,
"rewards/CADCDValueReward/mean": 0.004761902382597327,
"rewards/CADCDValueReward/std": 0.009021487087011337,
"rewards/CADChamferReward/mean": 0.4394494444131851,
"rewards/CADChamferReward/std": 0.39276182651519775,
"rewards/CADFormatReward/mean": 0.9921875,
"rewards/CADFormatReward/std": 0.0625,
"rewards/CADInvalidReward/mean": 0.078125,
"rewards/CADInvalidReward/std": 0.27048972249031067,
"rewards/CADProgressReward/mean": 0.02812500111758709,
"rewards/CADProgressReward/std": 0.06948306784033775,
"step": 65,
"step_time": 666.4829357709968
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 5897.5,
"completions/mean_length": 1852.3515625,
"completions/min_length": 582.0,
"epoch": 0.4888888888888889,
"frac_reward_zero_std": 0.0625,
"grad_norm": 0.32647079367496457,
"gvpm/clamp_rate": 0.3125,
"gvpm/mask_ratio": 0.3125,
"learning_rate": 9.016037657403223e-07,
"loss": -0.07062443345785141,
"num_turns": 3.3671875,
"reward": 1.0007648468017578,
"reward_std": 0.320096418261528,
"rewards/CADCDValueReward/mean": 0.003650993457995355,
"rewards/CADCDValueReward/std": 0.004244861658662558,
"rewards/CADChamferReward/mean": 0.4859210401773453,
"rewards/CADChamferReward/std": 0.3747712969779968,
"rewards/CADFormatReward/mean": 0.9921875,
"rewards/CADFormatReward/std": 0.0625,
"rewards/CADInvalidReward/mean": 0.03125,
"rewards/CADInvalidReward/std": 0.16902101784944534,
"rewards/CADProgressReward/mean": 0.037500000558793545,
"rewards/CADProgressReward/std": 0.07773387432098389,
"step": 66,
"step_time": 747.0061523441109
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 3447.5,
"completions/mean_length": 1537.765625,
"completions/min_length": 559.5,
"epoch": 0.4962962962962963,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.3684932883824493,
"gvpm/clamp_rate": 0.4375,
"gvpm/mask_ratio": 0.4375,
"learning_rate": 8.979184523044418e-07,
"loss": -0.08982193470001221,
"num_turns": 3.203125,
"reward": 1.005136787891388,
"reward_std": 0.4044913351535797,
"rewards/CADCDValueReward/mean": 0.015445917844772339,
"rewards/CADCDValueReward/std": 0.03328438941389322,
"rewards/CADChamferReward/mean": 0.4887305647134781,
"rewards/CADChamferReward/std": 0.41610924899578094,
"rewards/CADFormatReward/mean": 1.0,
"rewards/CADFormatReward/std": 0.0,
"rewards/CADInvalidReward/mean": 0.0625,
"rewards/CADInvalidReward/std": 0.24176587909460068,
"rewards/CADProgressReward/mean": 0.03281250037252903,
"rewards/CADProgressReward/std": 0.07462168112397194,
"step": 67,
"step_time": 612.9340318380273
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 5851.0,
"completions/mean_length": 1642.3359375,
"completions/min_length": 412.0,
"epoch": 0.5037037037037037,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.32089011997272,
"gvpm/clamp_rate": 0.0625,
"gvpm/mask_ratio": 0.0625,
"learning_rate": 8.941732138133031e-07,
"loss": -0.06648235768079758,
"num_turns": 2.953125,
"reward": 1.0746761560440063,
"reward_std": 0.3606071174144745,
"rewards/CADCDValueReward/mean": 0.00935483630746603,
"rewards/CADCDValueReward/std": 0.022447858937084675,
"rewards/CADChamferReward/mean": 0.570769876241684,
"rewards/CADChamferReward/std": 0.41204652190208435,
"rewards/CADFormatReward/mean": 0.9765625,
"rewards/CADFormatReward/std": 0.10652101784944534,
"rewards/CADInvalidReward/mean": 0.03125,
"rewards/CADInvalidReward/std": 0.16902101784944534,
"rewards/CADProgressReward/mean": 0.031250000931322575,
"rewards/CADProgressReward/std": 0.07267312332987785,
"step": 68,
"step_time": 626.2576917275437
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 8577.0,
"completions/mean_length": 1886.4140625,
"completions/min_length": 558.5,
"epoch": 0.5111111111111111,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.3337271280587593,
"gvpm/clamp_rate": 0.0625,
"gvpm/mask_ratio": 0.0625,
"learning_rate": 8.903686142860471e-07,
"loss": -0.07734221965074539,
"num_turns": 3.484375,
"reward": 0.8702357113361359,
"reward_std": 0.33238084614276886,
"rewards/CADCDValueReward/mean": 0.010502251796424389,
"rewards/CADCDValueReward/std": 0.023212259635329247,
"rewards/CADChamferReward/mean": 0.3733607083559036,
"rewards/CADChamferReward/std": 0.40685878694057465,
"rewards/CADFormatReward/mean": 0.9765625,
"rewards/CADFormatReward/std": 0.1501840502023697,
"rewards/CADInvalidReward/mean": 0.1484375,
"rewards/CADInvalidReward/std": 0.3567937761545181,
"rewards/CADProgressReward/mean": 0.01718750037252903,
"rewards/CADProgressReward/std": 0.05642745830118656,
"step": 69,
"step_time": 859.1491852009203
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 8192.0,
"completions/mean_length": 1653.1640625,
"completions/min_length": 455.0,
"epoch": 0.5185185185185185,
"frac_reward_zero_std": 0.0625,
"grad_norm": 0.3461346197894174,
"gvpm/clamp_rate": 0.1875,
"gvpm/mask_ratio": 0.1875,
"learning_rate": 8.865052266813685e-07,
"loss": -0.10868415236473083,
"num_turns": 3.21875,
"reward": 1.0690314769744873,
"reward_std": 0.35482974350452423,
"rewards/CADCDValueReward/mean": 0.00583080283831805,
"rewards/CADCDValueReward/std": 0.012201638892292976,
"rewards/CADChamferReward/mean": 0.5573127865791321,
"rewards/CADChamferReward/std": 0.41786521673202515,
"rewards/CADFormatReward/mean": 0.984375,
"rewards/CADFormatReward/std": 0.125,
"rewards/CADInvalidReward/mean": 0.0546875,
"rewards/CADInvalidReward/std": 0.22850853204727173,
"rewards/CADProgressReward/mean": 0.0390625,
"rewards/CADProgressReward/std": 0.07989032194018364,
"step": 70,
"step_time": 692.2213933565654
}
],
"logging_steps": 1,
"max_steps": 270,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 10,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}