brysgo/gol-grpo-fixed-validation-37156495
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

初始化项目，由ModelHub XC社区提供模型

Browse Source 
Model: brysgo/gol-grpo-fixed-validation-37156495
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-06-25 22:11:24 +08:00
commit 6bf7a7c528
17 changed files with 1277 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

844
checkpoint-30/trainer_state.json Normal file
View File

@@ -0,0 +1,844 @@
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.19736842105263158,
"eval_steps": 500,
"global_step": 30,
"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": 0.125,
"completions/max_length": 192.0,
"completions/max_terminated_length": 175.0,
"completions/mean_length": 135.25,
"completions/mean_terminated_length": 127.14286041259766,
"completions/min_length": 73.0,
"completions/min_terminated_length": 73.0,
"entropy": 0.7073924690485001,
"epoch": 0.006578947368421052,
"frac_reward_zero_std": 0.0,
"grad_norm": 11.996140480041504,
"learning_rate": 1e-06,
"loss": -0.16150084137916565,
"num_tokens": 2426.0,
"reward": -2.15625,
"reward_std": 0.5827506184577942,
"rewards/gol_tool_reward/mean": -2.15625,
"rewards/gol_tool_reward/std": 0.5827506184577942,
"step": 1,
"step_time": 5.788707780069672
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 172.0,
"completions/max_terminated_length": 172.0,
"completions/mean_length": 126.75,
"completions/mean_terminated_length": 126.75,
"completions/min_length": 101.0,
"completions/min_terminated_length": 101.0,
"entropy": 0.6544792205095291,
"epoch": 0.013157894736842105,
"frac_reward_zero_std": 0.0,
"grad_norm": 12.064927101135254,
"learning_rate": 9.666666666666666e-07,
"loss": -0.05266939848661423,
"num_tokens": 4784.0,
"reward": -2.28125,
"reward_std": 0.7454326152801514,
"rewards/gol_tool_reward/mean": -2.28125,
"rewards/gol_tool_reward/std": 0.7454325556755066,
"step": 2,
"step_time": 4.945590557064861
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.125,
"completions/max_length": 192.0,
"completions/max_terminated_length": 150.0,
"completions/mean_length": 141.125,
"completions/mean_terminated_length": 133.85714721679688,
"completions/min_length": 113.0,
"completions/min_terminated_length": 113.0,
"entropy": 0.7222911566495895,
"epoch": 0.019736842105263157,
"frac_reward_zero_std": 0.5,
"grad_norm": 8.669416427612305,
"learning_rate": 9.333333333333333e-07,
"loss": 0.061358727514743805,
"num_tokens": 7257.0,
"reward": -1.7750000953674316,
"reward_std": 0.6943651437759399,
"rewards/gol_tool_reward/mean": -1.7750000953674316,
"rewards/gol_tool_reward/std": 0.6943651437759399,
"step": 3,
"step_time": 5.346994869993068
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.125,
"completions/max_length": 192.0,
"completions/max_terminated_length": 178.0,
"completions/mean_length": 137.75,
"completions/mean_terminated_length": 130.0,
"completions/min_length": 101.0,
"completions/min_terminated_length": 101.0,
"entropy": 0.584687352180481,
"epoch": 0.02631578947368421,
"frac_reward_zero_std": 0.0,
"grad_norm": 11.648664474487305,
"learning_rate": 9e-07,
"loss": 0.0426417738199234,
"num_tokens": 9703.0,
"reward": -1.6624999046325684,
"reward_std": 0.4860555827617645,
"rewards/gol_tool_reward/mean": -1.6624999046325684,
"rewards/gol_tool_reward/std": 0.4860555827617645,
"step": 4,
"step_time": 5.309761939919554
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.125,
"completions/max_length": 192.0,
"completions/max_terminated_length": 164.0,
"completions/mean_length": 131.75,
"completions/mean_terminated_length": 123.14286041259766,
"completions/min_length": 83.0,
"completions/min_terminated_length": 83.0,
"entropy": 0.6980996131896973,
"epoch": 0.03289473684210526,
"frac_reward_zero_std": 0.0,
"grad_norm": 10.72757625579834,
"learning_rate": 8.666666666666667e-07,
"loss": 0.014450110495090485,
"num_tokens": 12101.0,
"reward": -1.8328126668930054,
"reward_std": 1.0132369995117188,
"rewards/gol_tool_reward/mean": -1.8328126668930054,
"rewards/gol_tool_reward/std": 1.0132369995117188,
"step": 5,
"step_time": 5.347554020117968
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.25,
"completions/max_length": 192.0,
"completions/max_terminated_length": 136.0,
"completions/mean_length": 123.75,
"completions/mean_terminated_length": 101.0,
"completions/min_length": 69.0,
"completions/min_terminated_length": 69.0,
"entropy": 0.5499220415949821,
"epoch": 0.039473684210526314,
"frac_reward_zero_std": 0.5,
"grad_norm": 9.949792861938477,
"learning_rate": 8.333333333333333e-07,
"loss": 0.14735683798789978,
"num_tokens": 14435.0,
"reward": -1.431249976158142,
"reward_std": 0.0883883461356163,
"rewards/gol_tool_reward/mean": -1.431249976158142,
"rewards/gol_tool_reward/std": 0.0883883461356163,
"step": 6,
"step_time": 5.282774350955151
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.25,
"completions/max_length": 192.0,
"completions/max_terminated_length": 141.0,
"completions/mean_length": 134.625,
"completions/mean_terminated_length": 115.5,
"completions/min_length": 86.0,
"completions/min_terminated_length": 86.0,
"entropy": 0.5586348995566368,
"epoch": 0.046052631578947366,
"frac_reward_zero_std": 0.5,
"grad_norm": 7.489627361297607,
"learning_rate": 8e-07,
"loss": 0.10210999846458435,
"num_tokens": 16856.0,
"reward": -1.7750000953674316,
"reward_std": 0.6943651437759399,
"rewards/gol_tool_reward/mean": -1.7750000953674316,
"rewards/gol_tool_reward/std": 0.6943651437759399,
"step": 7,
"step_time": 5.278500100132078
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.125,
"completions/max_length": 192.0,
"completions/max_terminated_length": 171.0,
"completions/mean_length": 146.375,
"completions/mean_terminated_length": 139.85714721679688,
"completions/min_length": 92.0,
"completions/min_terminated_length": 92.0,
"entropy": 0.6310850158333778,
"epoch": 0.05263157894736842,
"frac_reward_zero_std": 0.0,
"grad_norm": 10.776825904846191,
"learning_rate": 7.666666666666667e-07,
"loss": 0.019737035036087036,
"num_tokens": 19371.0,
"reward": -1.568750023841858,
"reward_std": 0.9996205568313599,
"rewards/gol_tool_reward/mean": -1.568750023841858,
"rewards/gol_tool_reward/std": 0.9996204972267151,
"step": 8,
"step_time": 5.483975553070195
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.125,
"completions/max_length": 192.0,
"completions/max_terminated_length": 175.0,
"completions/mean_length": 136.75,
"completions/mean_terminated_length": 128.85714721679688,
"completions/min_length": 102.0,
"completions/min_terminated_length": 102.0,
"entropy": 0.5831628292798996,
"epoch": 0.05921052631578947,
"frac_reward_zero_std": 0.5,
"grad_norm": 8.288241386413574,
"learning_rate": 7.333333333333332e-07,
"loss": 0.06289202719926834,
"num_tokens": 21809.0,
"reward": -1.537500023841858,
"reward_std": 0.5680480599403381,
"rewards/gol_tool_reward/mean": -1.537500023841858,
"rewards/gol_tool_reward/std": 0.5680480599403381,
"step": 9,
"step_time": 5.505063942982815
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.125,
"completions/max_length": 192.0,
"completions/max_terminated_length": 163.0,
"completions/mean_length": 126.25,
"completions/mean_terminated_length": 116.85714721679688,
"completions/min_length": 76.0,
"completions/min_terminated_length": 76.0,
"entropy": 0.4830511435866356,
"epoch": 0.06578947368421052,
"frac_reward_zero_std": 0.0,
"grad_norm": 11.556458473205566,
"learning_rate": 7e-07,
"loss": 0.0357169508934021,
"num_tokens": 24163.0,
"reward": -1.7000000476837158,
"reward_std": 0.769044041633606,
"rewards/gol_tool_reward/mean": -1.7000000476837158,
"rewards/gol_tool_reward/std": 0.7690439820289612,
"step": 10,
"step_time": 5.544649249990471
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.25,
"completions/max_length": 192.0,
"completions/max_terminated_length": 186.0,
"completions/mean_length": 165.0,
"completions/mean_terminated_length": 156.0,
"completions/min_length": 125.0,
"completions/min_terminated_length": 125.0,
"entropy": 0.5843875631690025,
"epoch": 0.07236842105263158,
"frac_reward_zero_std": 0.0,
"grad_norm": 12.836280822753906,
"learning_rate": 6.666666666666666e-07,
"loss": 0.047385718673467636,
"num_tokens": 26827.0,
"reward": -1.1484375,
"reward_std": 1.135318636894226,
"rewards/gol_tool_reward/mean": -1.1484375,
"rewards/gol_tool_reward/std": 1.135318636894226,
"step": 11,
"step_time": 5.533237581024878
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.125,
"completions/max_length": 192.0,
"completions/max_terminated_length": 164.0,
"completions/mean_length": 143.0,
"completions/mean_terminated_length": 136.0,
"completions/min_length": 104.0,
"completions/min_terminated_length": 104.0,
"entropy": 0.510637991130352,
"epoch": 0.07894736842105263,
"frac_reward_zero_std": 0.0,
"grad_norm": 10.66059684753418,
"learning_rate": 6.333333333333332e-07,
"loss": 0.08006405830383301,
"num_tokens": 29315.0,
"reward": -0.04531252384185791,
"reward_std": 1.7028460502624512,
"rewards/gol_tool_reward/mean": -0.04531252384185791,
"rewards/gol_tool_reward/std": 1.7028461694717407,
"step": 12,
"step_time": 5.5445326251210645
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 180.0,
"completions/max_terminated_length": 180.0,
"completions/mean_length": 134.0,
"completions/mean_terminated_length": 134.0,
"completions/min_length": 108.0,
"completions/min_terminated_length": 108.0,
"entropy": 0.548699863255024,
"epoch": 0.08552631578947369,
"frac_reward_zero_std": 0.0,
"grad_norm": 11.632511138916016,
"learning_rate": 6e-07,
"loss": -0.1205173134803772,
"num_tokens": 31731.0,
"reward": -0.0703125149011612,
"reward_std": 1.1454672813415527,
"rewards/gol_tool_reward/mean": -0.0703125149011612,
"rewards/gol_tool_reward/std": 1.1454674005508423,
"step": 13,
"step_time": 5.114862979855388
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.125,
"completions/max_length": 192.0,
"completions/max_terminated_length": 170.0,
"completions/mean_length": 130.25,
"completions/mean_terminated_length": 121.42857360839844,
"completions/min_length": 79.0,
"completions/min_terminated_length": 79.0,
"entropy": 0.595326341688633,
"epoch": 0.09210526315789473,
"frac_reward_zero_std": 0.0,
"grad_norm": 12.228357315063477,
"learning_rate": 5.666666666666666e-07,
"loss": 0.16490532457828522,
"num_tokens": 34117.0,
"reward": -0.22187499701976776,
"reward_std": 1.3627170324325562,
"rewards/gol_tool_reward/mean": -0.22187499701976776,
"rewards/gol_tool_reward/std": 1.3627170324325562,
"step": 14,
"step_time": 5.458060751087032
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 181.0,
"completions/max_terminated_length": 181.0,
"completions/mean_length": 123.125,
"completions/mean_terminated_length": 123.125,
"completions/min_length": 94.0,
"completions/min_terminated_length": 94.0,
"entropy": 0.5212945640087128,
"epoch": 0.09868421052631579,
"frac_reward_zero_std": 0.0,
"grad_norm": 11.923970222473145,
"learning_rate": 5.333333333333333e-07,
"loss": -0.03577335178852081,
"num_tokens": 36446.0,
"reward": -0.27812498807907104,
"reward_std": 1.2783524990081787,
"rewards/gol_tool_reward/mean": -0.27812498807907104,
"rewards/gol_tool_reward/std": 1.2783526182174683,
"step": 15,
"step_time": 5.0575181999010965
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 160.0,
"completions/max_terminated_length": 160.0,
"completions/mean_length": 118.375,
"completions/mean_terminated_length": 118.375,
"completions/min_length": 51.0,
"completions/min_terminated_length": 51.0,
"entropy": 0.5411187782883644,
"epoch": 0.10526315789473684,
"frac_reward_zero_std": 0.0,
"grad_norm": 12.17962646484375,
"learning_rate": 5e-07,
"loss": -0.07631053775548935,
"num_tokens": 38737.0,
"reward": -0.42812496423721313,
"reward_std": 0.8981565237045288,
"rewards/gol_tool_reward/mean": -0.42812496423721313,
"rewards/gol_tool_reward/std": 0.8981566429138184,
"step": 16,
"step_time": 4.665566484094597
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.125,
"completions/max_length": 192.0,
"completions/max_terminated_length": 166.0,
"completions/mean_length": 133.375,
"completions/mean_terminated_length": 125.00000762939453,
"completions/min_length": 91.0,
"completions/min_terminated_length": 91.0,
"entropy": 0.5414882004261017,
"epoch": 0.1118421052631579,
"frac_reward_zero_std": 0.0,
"grad_norm": 10.49337387084961,
"learning_rate": 4.6666666666666666e-07,
"loss": 0.016712799668312073,
"num_tokens": 41148.0,
"reward": -0.5531250238418579,
"reward_std": 0.7563643455505371,
"rewards/gol_tool_reward/mean": -0.5531250238418579,
"rewards/gol_tool_reward/std": 0.7563644051551819,
"step": 17,
"step_time": 5.364008491043933
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 146.0,
"completions/max_terminated_length": 146.0,
"completions/mean_length": 117.25,
"completions/mean_terminated_length": 117.25,
"completions/min_length": 86.0,
"completions/min_terminated_length": 86.0,
"entropy": 0.4803786799311638,
"epoch": 0.11842105263157894,
"frac_reward_zero_std": 0.0,
"grad_norm": 8.492094993591309,
"learning_rate": 4.3333333333333335e-07,
"loss": 0.002299964427947998,
"num_tokens": 43430.0,
"reward": 0.09843747317790985,
"reward_std": 0.9447796940803528,
"rewards/gol_tool_reward/mean": 0.09843747317790985,
"rewards/gol_tool_reward/std": 0.9447796940803528,
"step": 18,
"step_time": 4.548883846960962
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 144.0,
"completions/max_terminated_length": 144.0,
"completions/mean_length": 127.75,
"completions/mean_terminated_length": 127.75,
"completions/min_length": 102.0,
"completions/min_terminated_length": 102.0,
"entropy": 0.5627511814236641,
"epoch": 0.125,
"frac_reward_zero_std": 0.0,
"grad_norm": 9.863720893859863,
"learning_rate": 4e-07,
"loss": -0.03161659836769104,
"num_tokens": 45796.0,
"reward": -0.21875,
"reward_std": 0.3750595152378082,
"rewards/gol_tool_reward/mean": -0.21875,
"rewards/gol_tool_reward/std": 0.3750595152378082,
"step": 19,
"step_time": 4.288126179017127
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.125,
"completions/max_length": 192.0,
"completions/max_terminated_length": 147.0,
"completions/mean_length": 126.5,
"completions/mean_terminated_length": 117.14286041259766,
"completions/min_length": 92.0,
"completions/min_terminated_length": 92.0,
"entropy": 0.6194332987070084,
"epoch": 0.13157894736842105,
"frac_reward_zero_std": 0.0,
"grad_norm": 13.363960266113281,
"learning_rate": 3.666666666666666e-07,
"loss": 0.08973874151706696,
"num_tokens": 48152.0,
"reward": -0.7171875238418579,
"reward_std": 0.9436095952987671,
"rewards/gol_tool_reward/mean": -0.7171875238418579,
"rewards/gol_tool_reward/std": 0.9436095356941223,
"step": 20,
"step_time": 5.38997302914504
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 168.0,
"completions/max_terminated_length": 168.0,
"completions/mean_length": 128.375,
"completions/mean_terminated_length": 128.375,
"completions/min_length": 95.0,
"completions/min_terminated_length": 95.0,
"entropy": 0.5529855117201805,
"epoch": 0.13815789473684212,
"frac_reward_zero_std": 0.0,
"grad_norm": 11.634380340576172,
"learning_rate": 3.333333333333333e-07,
"loss": 0.003947541117668152,
"num_tokens": 50523.0,
"reward": 0.3453124761581421,
"reward_std": 1.174352765083313,
"rewards/gol_tool_reward/mean": 0.3453124761581421,
"rewards/gol_tool_reward/std": 1.174352765083313,
"step": 21,
"step_time": 4.907234247075394
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.125,
"completions/max_length": 192.0,
"completions/max_terminated_length": 138.0,
"completions/mean_length": 126.0,
"completions/mean_terminated_length": 116.5714340209961,
"completions/min_length": 97.0,
"completions/min_terminated_length": 97.0,
"entropy": 0.554837092757225,
"epoch": 0.14473684210526316,
"frac_reward_zero_std": 0.0,
"grad_norm": 11.619768142700195,
"learning_rate": 3e-07,
"loss": 0.07526145875453949,
"num_tokens": 52875.0,
"reward": -0.6468750238418579,
"reward_std": 0.6682115793228149,
"rewards/gol_tool_reward/mean": -0.6468750238418579,
"rewards/gol_tool_reward/std": 0.6682115197181702,
"step": 22,
"step_time": 5.336878409842029
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 160.0,
"completions/max_terminated_length": 160.0,
"completions/mean_length": 131.25,
"completions/mean_terminated_length": 131.25,
"completions/min_length": 118.0,
"completions/min_terminated_length": 118.0,
"entropy": 0.5712674260139465,
"epoch": 0.1513157894736842,
"frac_reward_zero_std": 0.0,
"grad_norm": 10.13196849822998,
"learning_rate": 2.6666666666666667e-07,
"loss": -0.0519290491938591,
"num_tokens": 55269.0,
"reward": -0.23125000298023224,
"reward_std": 0.3797908425331116,
"rewards/gol_tool_reward/mean": -0.23125000298023224,
"rewards/gol_tool_reward/std": 0.3797908127307892,
"step": 23,
"step_time": 4.688064404879697
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 185.0,
"completions/max_terminated_length": 185.0,
"completions/mean_length": 119.125,
"completions/mean_terminated_length": 119.125,
"completions/min_length": 95.0,
"completions/min_terminated_length": 95.0,
"entropy": 0.5097240433096886,
"epoch": 0.15789473684210525,
"frac_reward_zero_std": 0.0,
"grad_norm": 9.0028076171875,
"learning_rate": 2.3333333333333333e-07,
"loss": -0.05878718942403793,
"num_tokens": 57566.0,
"reward": -0.02656250074505806,
"reward_std": 0.5322810411453247,
"rewards/gol_tool_reward/mean": -0.02656250074505806,
"rewards/gol_tool_reward/std": 0.5322810411453247,
"step": 24,
"step_time": 5.215061811963096
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 182.0,
"completions/max_terminated_length": 182.0,
"completions/mean_length": 120.5,
"completions/mean_terminated_length": 120.5,
"completions/min_length": 95.0,
"completions/min_terminated_length": 95.0,
"entropy": 0.5010545030236244,
"epoch": 0.16447368421052633,
"frac_reward_zero_std": 0.0,
"grad_norm": 10.190753936767578,
"learning_rate": 2e-07,
"loss": -0.00694732740521431,
"num_tokens": 59874.0,
"reward": 0.265625,
"reward_std": 1.212799072265625,
"rewards/gol_tool_reward/mean": 0.265625,
"rewards/gol_tool_reward/std": 1.2127991914749146,
"step": 25,
"step_time": 5.231085888110101
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 176.0,
"completions/max_terminated_length": 176.0,
"completions/mean_length": 126.375,
"completions/mean_terminated_length": 126.375,
"completions/min_length": 88.0,
"completions/min_terminated_length": 88.0,
"entropy": 0.5022949725389481,
"epoch": 0.17105263157894737,
"frac_reward_zero_std": 0.0,
"grad_norm": 11.172710418701172,
"learning_rate": 1.6666666666666665e-07,
"loss": -0.02744058519601822,
"num_tokens": 62229.0,
"reward": -0.359375,
"reward_std": 0.05966797471046448,
"rewards/gol_tool_reward/mean": -0.359375,
"rewards/gol_tool_reward/std": 0.05966798588633537,
"step": 26,
"step_time": 4.989802666124888
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 168.0,
"completions/max_terminated_length": 168.0,
"completions/mean_length": 123.75,
"completions/mean_terminated_length": 123.75,
"completions/min_length": 94.0,
"completions/min_terminated_length": 94.0,
"entropy": 0.5283998399972916,
"epoch": 0.17763157894736842,
"frac_reward_zero_std": 0.0,
"grad_norm": 9.951343536376953,
"learning_rate": 1.3333333333333334e-07,
"loss": 0.0610831081867218,
"num_tokens": 64563.0,
"reward": -0.0078125,
"reward_std": 1.0492969751358032,
"rewards/gol_tool_reward/mean": -0.0078125,
"rewards/gol_tool_reward/std": 1.0492969751358032,
"step": 27,
"step_time": 4.680415777955204
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 173.0,
"completions/max_terminated_length": 173.0,
"completions/mean_length": 118.875,
"completions/mean_terminated_length": 118.875,
"completions/min_length": 86.0,
"completions/min_terminated_length": 86.0,
"entropy": 0.4742962121963501,
"epoch": 0.18421052631578946,
"frac_reward_zero_std": 0.0,
"grad_norm": 12.151405334472656,
"learning_rate": 1e-07,
"loss": 0.058089837431907654,
"num_tokens": 66858.0,
"reward": -0.42031246423721313,
"reward_std": 0.11199040710926056,
"rewards/gol_tool_reward/mean": -0.42031246423721313,
"rewards/gol_tool_reward/std": 0.11199040710926056,
"step": 28,
"step_time": 4.9447458980139345
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 160.0,
"completions/max_terminated_length": 160.0,
"completions/mean_length": 134.125,
"completions/mean_terminated_length": 134.125,
"completions/min_length": 98.0,
"completions/min_terminated_length": 98.0,
"entropy": 0.5360276401042938,
"epoch": 0.19078947368421054,
"frac_reward_zero_std": 0.0,
"grad_norm": 10.772101402282715,
"learning_rate": 6.666666666666667e-08,
"loss": -0.08468331396579742,
"num_tokens": 69275.0,
"reward": 0.04843750596046448,
"reward_std": 0.7822938561439514,
"rewards/gol_tool_reward/mean": 0.04843750596046448,
"rewards/gol_tool_reward/std": 0.7822939157485962,
"step": 29,
"step_time": 4.704459578031674
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 146.0,
"completions/max_terminated_length": 146.0,
"completions/mean_length": 124.5,
"completions/mean_terminated_length": 124.5,
"completions/min_length": 107.0,
"completions/min_terminated_length": 107.0,
"entropy": 0.5363342761993408,
"epoch": 0.19736842105263158,
"frac_reward_zero_std": 0.0,
"grad_norm": 10.216547966003418,
"learning_rate": 3.3333333333333334e-08,
"loss": -0.06437680870294571,
"num_tokens": 71615.0,
"reward": -0.2890625,
"reward_std": 0.22573663294315338,
"rewards/gol_tool_reward/mean": -0.2890625,
"rewards/gol_tool_reward/std": 0.22573664784431458,
"step": 30,
"step_time": 4.333932535140775
}
],
"logging_steps": 1,
"max_steps": 30,
"num_input_tokens_seen": 71615,
"num_train_epochs": 1,
"save_steps": 30,
"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
}