Instructions to use brikdavies/qwen1.7B-MMLU-hint-following-RL with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use brikdavies/qwen1.7B-MMLU-hint-following-RL with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("brikdavies/qwen1.7B-MMLU-hint-following-RL", device_map="auto") - Notebooks
- Google Colab
- Kaggle
| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.021367521367521368, | |
| "eval_steps": 500, | |
| "global_step": 50, | |
| "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": 0.5555555820465088, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2837.0, | |
| "completions/mean_length": 2626.97216796875, | |
| "completions/mean_terminated_length": 2160.6875, | |
| "completions/min_length": 1285.0, | |
| "completions/min_terminated_length": 1285.0, | |
| "entropy": 0.17718915144602457, | |
| "epoch": 0.00042735042735042735, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.07994551956653595, | |
| "learning_rate": 0.0, | |
| "loss": 0.0468, | |
| "num_tokens": 102485.0, | |
| "reward": -1.153435230255127, | |
| "reward_std": 1.0195810794830322, | |
| "rewards/hint_following/mean": -0.2777777910232544, | |
| "rewards/hint_following/std": 0.8819171190261841, | |
| "rewards/length_penalty/mean": -0.8756573796272278, | |
| "rewards/length_penalty/std": 0.17644810676574707, | |
| "sampling/importance_sampling_ratio/max": 3.0, | |
| "sampling/importance_sampling_ratio/mean": 0.9937517642974854, | |
| "sampling/importance_sampling_ratio/min": 0.16920268535614014, | |
| "sampling/sampling_logp_difference/max": 1.7766579389572144, | |
| "sampling/sampling_logp_difference/mean": 0.02078823931515217, | |
| "step": 1, | |
| "step_time": 67.38753714214545 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 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": 3000.0, | |
| "completions/max_terminated_length": 2961.0, | |
| "completions/mean_length": 1809.4444580078125, | |
| "completions/mean_terminated_length": 1412.5926513671875, | |
| "completions/min_length": 858.0, | |
| "completions/min_terminated_length": 858.0, | |
| "entropy": 0.1950028215845426, | |
| "epoch": 0.0008547008547008547, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.07918090373277664, | |
| "learning_rate": 3.846153846153847e-06, | |
| "loss": 0.022, | |
| "num_tokens": 177279.0, | |
| "reward": -0.4642592668533325, | |
| "reward_std": 1.043834924697876, | |
| "rewards/hint_following/mean": 0.1388888955116272, | |
| "rewards/hint_following/std": 0.798311710357666, | |
| "rewards/length_penalty/mean": -0.6031482219696045, | |
| "rewards/length_penalty/std": 0.28368300199508667, | |
| "sampling/importance_sampling_ratio/max": 2.0890045166015625, | |
| "sampling/importance_sampling_ratio/mean": 0.9933527708053589, | |
| "sampling/importance_sampling_ratio/min": 0.29880645871162415, | |
| "sampling/sampling_logp_difference/max": 1.2079591751098633, | |
| "sampling/sampling_logp_difference/mean": 0.021868562325835228, | |
| "step": 2, | |
| "step_time": 59.25448451703414 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.007054112308348219, | |
| "clip_ratio/high_mean": 0.007054112308348219, | |
| "clip_ratio/low_mean": 0.0024281112127937376, | |
| "clip_ratio/low_min": 0.0024281112127937376, | |
| "clip_ratio/region_mean": 0.009482223695764938, | |
| "completions/clipped_ratio": 0.1666666716337204, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2411.0, | |
| "completions/mean_length": 1750.8055419921875, | |
| "completions/mean_terminated_length": 1500.966796875, | |
| "completions/min_length": 939.0, | |
| "completions/min_terminated_length": 939.0, | |
| "entropy": 0.1983068659901619, | |
| "epoch": 0.001282051282051282, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.06695634126663208, | |
| "learning_rate": 7.692307692307694e-06, | |
| "loss": 0.0035, | |
| "num_tokens": 245288.0, | |
| "reward": -0.25026851892471313, | |
| "reward_std": 0.9555131196975708, | |
| "rewards/hint_following/mean": 0.3333333432674408, | |
| "rewards/hint_following/std": 0.7559289932250977, | |
| "rewards/length_penalty/mean": -0.5836018323898315, | |
| "rewards/length_penalty/std": 0.22866536676883698, | |
| "sampling/importance_sampling_ratio/max": 2.208913803100586, | |
| "sampling/importance_sampling_ratio/mean": 0.9931869506835938, | |
| "sampling/importance_sampling_ratio/min": 0.30834895372390747, | |
| "sampling/sampling_logp_difference/max": 1.176523208618164, | |
| "sampling/sampling_logp_difference/mean": 0.022443262860178947, | |
| "step": 3, | |
| "step_time": 55.47524960897863 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0025329499427850046, | |
| "clip_ratio/high_mean": 0.0025329499427850046, | |
| "clip_ratio/low_mean": 0.0070530143566429615, | |
| "clip_ratio/low_min": 0.0070530143566429615, | |
| "clip_ratio/region_mean": 0.009585964183012644, | |
| "completions/clipped_ratio": 0.472222238779068, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2857.0, | |
| "completions/mean_length": 2469.27783203125, | |
| "completions/mean_terminated_length": 1994.4210205078125, | |
| "completions/min_length": 627.0, | |
| "completions/min_terminated_length": 627.0, | |
| "entropy": 0.1949674834807714, | |
| "epoch": 0.0017094017094017094, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.08172055333852768, | |
| "learning_rate": 1.153846153846154e-05, | |
| "loss": 0.0355, | |
| "num_tokens": 339768.0, | |
| "reward": -1.0453147888183594, | |
| "reward_std": 0.8787956237792969, | |
| "rewards/hint_following/mean": -0.2222222238779068, | |
| "rewards/hint_following/std": 0.7601169347763062, | |
| "rewards/length_penalty/mean": -0.8230926394462585, | |
| "rewards/length_penalty/std": 0.2654260993003845, | |
| "sampling/importance_sampling_ratio/max": 3.0, | |
| "sampling/importance_sampling_ratio/mean": 0.993114173412323, | |
| "sampling/importance_sampling_ratio/min": 0.19437530636787415, | |
| "sampling/sampling_logp_difference/max": 1.6379644870758057, | |
| "sampling/sampling_logp_difference/mean": 0.022510718554258347, | |
| "step": 4, | |
| "step_time": 62.77226335310843 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.001851275311006854, | |
| "clip_ratio/high_mean": 0.001851275311006854, | |
| "clip_ratio/low_mean": 0.00705820966201524, | |
| "clip_ratio/low_min": 0.00705820966201524, | |
| "clip_ratio/region_mean": 0.008909484914814433, | |
| "completions/clipped_ratio": 0.4166666567325592, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2915.0, | |
| "completions/mean_length": 2362.8056640625, | |
| "completions/mean_terminated_length": 1907.666748046875, | |
| "completions/min_length": 1109.0, | |
| "completions/min_terminated_length": 1109.0, | |
| "entropy": 0.18256587783495584, | |
| "epoch": 0.002136752136752137, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.072978176176548, | |
| "learning_rate": 1.5384615384615387e-05, | |
| "loss": 0.1055, | |
| "num_tokens": 431687.0, | |
| "reward": -0.9264907836914062, | |
| "reward_std": 1.0278923511505127, | |
| "rewards/hint_following/mean": -0.1388888955116272, | |
| "rewards/hint_following/std": 0.8333333730697632, | |
| "rewards/length_penalty/mean": -0.787601888179779, | |
| "rewards/length_penalty/std": 0.2262839674949646, | |
| "sampling/importance_sampling_ratio/max": 2.5575737953186035, | |
| "sampling/importance_sampling_ratio/mean": 0.9932507872581482, | |
| "sampling/importance_sampling_ratio/min": 0.2952674627304077, | |
| "sampling/sampling_logp_difference/max": 1.2198736667633057, | |
| "sampling/sampling_logp_difference/mean": 0.020803121849894524, | |
| "step": 5, | |
| "step_time": 61.06088067602832 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0030960943549871445, | |
| "clip_ratio/high_mean": 0.0030960943549871445, | |
| "clip_ratio/low_mean": 0.006137795591106017, | |
| "clip_ratio/low_min": 0.006137795591106017, | |
| "clip_ratio/region_mean": 0.009233890101313591, | |
| "completions/clipped_ratio": 0.4166666567325592, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2929.0, | |
| "completions/mean_length": 2417.444580078125, | |
| "completions/mean_terminated_length": 2001.3333740234375, | |
| "completions/min_length": 1434.0, | |
| "completions/min_terminated_length": 1434.0, | |
| "entropy": 0.1919809157649676, | |
| "epoch": 0.002564102564102564, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.07703683525323868, | |
| "learning_rate": 1.923076923076923e-05, | |
| "loss": 0.1019, | |
| "num_tokens": 525555.0, | |
| "reward": -0.7224815487861633, | |
| "reward_std": 1.1388682126998901, | |
| "rewards/hint_following/mean": 0.0833333358168602, | |
| "rewards/hint_following/std": 0.9673232436180115, | |
| "rewards/length_penalty/mean": -0.8058148622512817, | |
| "rewards/length_penalty/std": 0.18899519741535187, | |
| "sampling/importance_sampling_ratio/max": 2.9181816577911377, | |
| "sampling/importance_sampling_ratio/mean": 0.9935172200202942, | |
| "sampling/importance_sampling_ratio/min": 0.14165793359279633, | |
| "sampling/sampling_logp_difference/max": 1.9543399810791016, | |
| "sampling/sampling_logp_difference/mean": 0.021658122539520264, | |
| "step": 6, | |
| "step_time": 64.93633897381369 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.004218692930104832, | |
| "clip_ratio/high_mean": 0.004218692930104832, | |
| "clip_ratio/low_mean": 0.0048820926652600365, | |
| "clip_ratio/low_min": 0.0048820926652600365, | |
| "clip_ratio/region_mean": 0.009100785556559762, | |
| "completions/clipped_ratio": 0.2222222238779068, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2933.0, | |
| "completions/mean_length": 2029.0833740234375, | |
| "completions/mean_terminated_length": 1751.6785888671875, | |
| "completions/min_length": 1009.0, | |
| "completions/min_terminated_length": 1009.0, | |
| "entropy": 0.20896865924199423, | |
| "epoch": 0.0029914529914529917, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.07187572866678238, | |
| "learning_rate": 2.307692307692308e-05, | |
| "loss": 0.0798, | |
| "num_tokens": 606186.0, | |
| "reward": -0.4819166362285614, | |
| "reward_std": 0.9970327019691467, | |
| "rewards/hint_following/mean": 0.1944444477558136, | |
| "rewards/hint_following/std": 0.7862913012504578, | |
| "rewards/length_penalty/mean": -0.676361083984375, | |
| "rewards/length_penalty/std": 0.2366827428340912, | |
| "sampling/importance_sampling_ratio/max": 2.6964023113250732, | |
| "sampling/importance_sampling_ratio/mean": 0.9927431344985962, | |
| "sampling/importance_sampling_ratio/min": 0.21346372365951538, | |
| "sampling/sampling_logp_difference/max": 1.5442883968353271, | |
| "sampling/sampling_logp_difference/mean": 0.022996528074145317, | |
| "step": 7, | |
| "step_time": 59.43198294797912 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.007654782539854447, | |
| "clip_ratio/high_mean": 0.007654782539854447, | |
| "clip_ratio/low_mean": 0.002519401100774606, | |
| "clip_ratio/low_min": 0.002519401100774606, | |
| "clip_ratio/region_mean": 0.010174183640629053, | |
| "completions/clipped_ratio": 0.1666666716337204, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2957.0, | |
| "completions/mean_length": 2227.97216796875, | |
| "completions/mean_terminated_length": 2073.56689453125, | |
| "completions/min_length": 1197.0, | |
| "completions/min_terminated_length": 1197.0, | |
| "entropy": 0.19751630226771036, | |
| "epoch": 0.003418803418803419, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.08499564230442047, | |
| "learning_rate": 2.6923076923076923e-05, | |
| "loss": 0.0572, | |
| "num_tokens": 692279.0, | |
| "reward": -0.4093240797519684, | |
| "reward_std": 0.8091796040534973, | |
| "rewards/hint_following/mean": 0.3333333432674408, | |
| "rewards/hint_following/std": 0.7171371579170227, | |
| "rewards/length_penalty/mean": -0.7426574230194092, | |
| "rewards/length_penalty/std": 0.19112928211688995, | |
| "sampling/importance_sampling_ratio/max": 2.491806745529175, | |
| "sampling/importance_sampling_ratio/mean": 0.9929631352424622, | |
| "sampling/importance_sampling_ratio/min": 0.3287244141101837, | |
| "sampling/sampling_logp_difference/max": 1.1125354766845703, | |
| "sampling/sampling_logp_difference/mean": 0.021952621638774872, | |
| "step": 8, | |
| "step_time": 60.087568536167964 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.007754642438764374, | |
| "clip_ratio/high_mean": 0.007754642438764374, | |
| "clip_ratio/low_mean": 0.0032427439582534134, | |
| "clip_ratio/low_min": 0.0032427439582534134, | |
| "clip_ratio/region_mean": 0.010997386649250984, | |
| "completions/clipped_ratio": 0.1111111119389534, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2948.0, | |
| "completions/mean_length": 2034.2222900390625, | |
| "completions/mean_terminated_length": 1913.5, | |
| "completions/min_length": 778.0, | |
| "completions/min_terminated_length": 778.0, | |
| "entropy": 0.21248381833235422, | |
| "epoch": 0.0038461538461538464, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.07896988093852997, | |
| "learning_rate": 3.0769230769230774e-05, | |
| "loss": 0.0952, | |
| "num_tokens": 771241.0, | |
| "reward": -0.28918519616127014, | |
| "reward_std": 0.7873722314834595, | |
| "rewards/hint_following/mean": 0.3888888955116272, | |
| "rewards/hint_following/std": 0.6877615451812744, | |
| "rewards/length_penalty/mean": -0.6780741214752197, | |
| "rewards/length_penalty/std": 0.22194787859916687, | |
| "sampling/importance_sampling_ratio/max": 2.757654905319214, | |
| "sampling/importance_sampling_ratio/mean": 0.9926807880401611, | |
| "sampling/importance_sampling_ratio/min": 0.19047939777374268, | |
| "sampling/sampling_logp_difference/max": 1.6582112312316895, | |
| "sampling/sampling_logp_difference/mean": 0.02367428131401539, | |
| "step": 9, | |
| "step_time": 58.2687593010487 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0022279491822700948, | |
| "clip_ratio/high_mean": 0.0022279491822700948, | |
| "clip_ratio/low_mean": 0.0019324645400047302, | |
| "clip_ratio/low_min": 0.0019324645400047302, | |
| "clip_ratio/region_mean": 0.004160413751378655, | |
| "completions/clipped_ratio": 0.3611111044883728, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2960.0, | |
| "completions/mean_length": 2266.111083984375, | |
| "completions/mean_terminated_length": 1851.3043212890625, | |
| "completions/min_length": 784.0, | |
| "completions/min_terminated_length": 784.0, | |
| "entropy": 0.20578551292419434, | |
| "epoch": 0.004273504273504274, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.06615690886974335, | |
| "learning_rate": 3.461538461538462e-05, | |
| "loss": 0.0543, | |
| "num_tokens": 858521.0, | |
| "reward": -0.5331481099128723, | |
| "reward_std": 1.1243869066238403, | |
| "rewards/hint_following/mean": 0.2222222238779068, | |
| "rewards/hint_following/std": 0.929242730140686, | |
| "rewards/length_penalty/mean": -0.7553703784942627, | |
| "rewards/length_penalty/std": 0.23448802530765533, | |
| "sampling/importance_sampling_ratio/max": 2.0015952587127686, | |
| "sampling/importance_sampling_ratio/mean": 0.9925298094749451, | |
| "sampling/importance_sampling_ratio/min": 0.30108708143234253, | |
| "sampling/sampling_logp_difference/max": 1.2003557682037354, | |
| "sampling/sampling_logp_difference/mean": 0.021938424557447433, | |
| "step": 10, | |
| "step_time": 60.113012868212536 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.002628828883947184, | |
| "clip_ratio/high_mean": 0.002628828883947184, | |
| "clip_ratio/low_mean": 0.0007913654941755036, | |
| "clip_ratio/low_min": 0.0007913654941755036, | |
| "clip_ratio/region_mean": 0.0034201944169277945, | |
| "completions/clipped_ratio": 0.3888888955116272, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2856.0, | |
| "completions/mean_length": 2308.111083984375, | |
| "completions/mean_terminated_length": 1867.8182373046875, | |
| "completions/min_length": 853.0, | |
| "completions/min_terminated_length": 853.0, | |
| "entropy": 0.1780667081475258, | |
| "epoch": 0.004700854700854701, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.06938029080629349, | |
| "learning_rate": 3.846153846153846e-05, | |
| "loss": 0.0506, | |
| "num_tokens": 948171.0, | |
| "reward": -1.0471482276916504, | |
| "reward_std": 0.824184238910675, | |
| "rewards/hint_following/mean": -0.2777777910232544, | |
| "rewards/hint_following/std": 0.6594851613044739, | |
| "rewards/length_penalty/mean": -0.7693703770637512, | |
| "rewards/length_penalty/std": 0.24463628232479095, | |
| "sampling/importance_sampling_ratio/max": 2.5170931816101074, | |
| "sampling/importance_sampling_ratio/mean": 0.9934613108634949, | |
| "sampling/importance_sampling_ratio/min": 0.29526859521865845, | |
| "sampling/sampling_logp_difference/max": 1.21986985206604, | |
| "sampling/sampling_logp_difference/mean": 0.021038096398115158, | |
| "step": 11, | |
| "step_time": 61.20216279185843 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.009717889750997225, | |
| "clip_ratio/high_mean": 0.009717889750997225, | |
| "clip_ratio/low_mean": 0.001428690991209199, | |
| "clip_ratio/low_min": 0.001428690991209199, | |
| "clip_ratio/region_mean": 0.011146580955634514, | |
| "completions/clipped_ratio": 0.2777777910232544, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2997.0, | |
| "completions/mean_length": 2494.944580078125, | |
| "completions/mean_terminated_length": 2300.6923828125, | |
| "completions/min_length": 1195.0, | |
| "completions/min_terminated_length": 1195.0, | |
| "entropy": 0.21741948276758194, | |
| "epoch": 0.005128205128205128, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.08027122914791107, | |
| "learning_rate": 4.230769230769231e-05, | |
| "loss": 0.0271, | |
| "num_tokens": 1045165.0, | |
| "reward": -0.6094259023666382, | |
| "reward_std": 0.9179477691650391, | |
| "rewards/hint_following/mean": 0.2222222238779068, | |
| "rewards/hint_following/std": 0.8655670881271362, | |
| "rewards/length_penalty/mean": -0.8316481113433838, | |
| "rewards/length_penalty/std": 0.1910414844751358, | |
| "sampling/importance_sampling_ratio/max": 2.7227704524993896, | |
| "sampling/importance_sampling_ratio/mean": 0.9923471808433533, | |
| "sampling/importance_sampling_ratio/min": 0.309160441160202, | |
| "sampling/sampling_logp_difference/max": 1.1738948822021484, | |
| "sampling/sampling_logp_difference/mean": 0.02328832633793354, | |
| "step": 12, | |
| "step_time": 63.202891704044305 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0020594243348265686, | |
| "clip_ratio/high_mean": 0.0020594243348265686, | |
| "clip_ratio/low_mean": 0.0059360921538124485, | |
| "clip_ratio/low_min": 0.0059360921538124485, | |
| "clip_ratio/region_mean": 0.007995516527444124, | |
| "completions/clipped_ratio": 0.2222222238779068, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2981.0, | |
| "completions/mean_length": 2085.638916015625, | |
| "completions/mean_terminated_length": 1824.3929443359375, | |
| "completions/min_length": 956.0, | |
| "completions/min_terminated_length": 956.0, | |
| "entropy": 0.18983352681001028, | |
| "epoch": 0.005555555555555556, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.2806412875652313, | |
| "learning_rate": 4.615384615384616e-05, | |
| "loss": 0.0294, | |
| "num_tokens": 1128414.0, | |
| "reward": -0.6674352288246155, | |
| "reward_std": 0.8722909092903137, | |
| "rewards/hint_following/mean": 0.02777777798473835, | |
| "rewards/hint_following/std": 0.6963624954223633, | |
| "rewards/length_penalty/mean": -0.6952130198478699, | |
| "rewards/length_penalty/std": 0.23719921708106995, | |
| "sampling/importance_sampling_ratio/max": 2.7140581607818604, | |
| "sampling/importance_sampling_ratio/mean": 0.9935481548309326, | |
| "sampling/importance_sampling_ratio/min": 0.19567115604877472, | |
| "sampling/sampling_logp_difference/max": 1.6313197612762451, | |
| "sampling/sampling_logp_difference/mean": 0.02119005285203457, | |
| "step": 13, | |
| "step_time": 61.55426321423147 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0014092368267786999, | |
| "clip_ratio/high_mean": 0.0014092368267786999, | |
| "clip_ratio/low_mean": 0.007215868448838592, | |
| "clip_ratio/low_min": 0.007215868448838592, | |
| "clip_ratio/region_mean": 0.008625105411435166, | |
| "completions/clipped_ratio": 0.1944444477558136, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2941.0, | |
| "completions/mean_length": 1680.3055419921875, | |
| "completions/mean_terminated_length": 1361.7586669921875, | |
| "completions/min_length": 398.0, | |
| "completions/min_terminated_length": 398.0, | |
| "entropy": 0.20419775694608688, | |
| "epoch": 0.005982905982905983, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.08586689084768295, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0681, | |
| "num_tokens": 1193759.0, | |
| "reward": -0.5878795981407166, | |
| "reward_std": 0.8454972505569458, | |
| "rewards/hint_following/mean": -0.02777777798473835, | |
| "rewards/hint_following/std": 0.6087979674339294, | |
| "rewards/length_penalty/mean": -0.5601018071174622, | |
| "rewards/length_penalty/std": 0.31150683760643005, | |
| "sampling/importance_sampling_ratio/max": 2.1082680225372314, | |
| "sampling/importance_sampling_ratio/mean": 0.992834746837616, | |
| "sampling/importance_sampling_ratio/min": 0.3020838499069214, | |
| "sampling/sampling_logp_difference/max": 1.1970505714416504, | |
| "sampling/sampling_logp_difference/mean": 0.02261131815612316, | |
| "step": 14, | |
| "step_time": 53.72449496493209 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.007441268069669604, | |
| "clip_ratio/high_mean": 0.007441268069669604, | |
| "clip_ratio/low_mean": 0.002903422418360909, | |
| "clip_ratio/low_min": 0.002903422418360909, | |
| "clip_ratio/region_mean": 0.01034469079847137, | |
| "completions/clipped_ratio": 0.25, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2906.0, | |
| "completions/mean_length": 2191.111083984375, | |
| "completions/mean_terminated_length": 1921.4814453125, | |
| "completions/min_length": 397.0, | |
| "completions/min_terminated_length": 397.0, | |
| "entropy": 0.22065167874097824, | |
| "epoch": 0.00641025641025641, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.07726756483316422, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0177, | |
| "num_tokens": 1279125.0, | |
| "reward": -0.7859259247779846, | |
| "reward_std": 0.7896111607551575, | |
| "rewards/hint_following/mean": -0.0555555559694767, | |
| "rewards/hint_following/std": 0.6737716794013977, | |
| "rewards/length_penalty/mean": -0.7303703427314758, | |
| "rewards/length_penalty/std": 0.2772618532180786, | |
| "sampling/importance_sampling_ratio/max": 2.1340627670288086, | |
| "sampling/importance_sampling_ratio/mean": 0.9917824268341064, | |
| "sampling/importance_sampling_ratio/min": 0.07570598274469376, | |
| "sampling/sampling_logp_difference/max": 2.5808980464935303, | |
| "sampling/sampling_logp_difference/mean": 0.023916445672512054, | |
| "step": 15, | |
| "step_time": 59.183635055902414 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.007959824210653702, | |
| "clip_ratio/high_mean": 0.007959824210653702, | |
| "clip_ratio/low_mean": 0.0026605394668877125, | |
| "clip_ratio/low_min": 0.0026605394668877125, | |
| "clip_ratio/region_mean": 0.010620363832761845, | |
| "completions/clipped_ratio": 0.1388888955116272, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2836.0, | |
| "completions/mean_length": 1801.9722900390625, | |
| "completions/mean_terminated_length": 1612.0, | |
| "completions/min_length": 660.0, | |
| "completions/min_terminated_length": 660.0, | |
| "entropy": 0.2382892519235611, | |
| "epoch": 0.006837606837606838, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.07441020756959915, | |
| "learning_rate": 5e-05, | |
| "loss": 0.1035, | |
| "num_tokens": 1351580.0, | |
| "reward": -0.2951018214225769, | |
| "reward_std": 0.9211878180503845, | |
| "rewards/hint_following/mean": 0.3055555522441864, | |
| "rewards/hint_following/std": 0.7099072337150574, | |
| "rewards/length_penalty/mean": -0.6006573438644409, | |
| "rewards/length_penalty/std": 0.2538762390613556, | |
| "sampling/importance_sampling_ratio/max": 2.141491413116455, | |
| "sampling/importance_sampling_ratio/mean": 0.9912582039833069, | |
| "sampling/importance_sampling_ratio/min": 0.3961084187030792, | |
| "sampling/sampling_logp_difference/max": 0.9260673522949219, | |
| "sampling/sampling_logp_difference/mean": 0.024969136342406273, | |
| "step": 16, | |
| "step_time": 58.414970112848096 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.00782025900358955, | |
| "clip_ratio/high_mean": 0.00782025900358955, | |
| "clip_ratio/low_mean": 0.0016357484661663573, | |
| "clip_ratio/low_min": 0.0016357484661663573, | |
| "clip_ratio/region_mean": 0.00945600758617123, | |
| "completions/clipped_ratio": 0.0833333358168602, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2959.0, | |
| "completions/mean_length": 1858.02783203125, | |
| "completions/mean_terminated_length": 1754.212158203125, | |
| "completions/min_length": 523.0, | |
| "completions/min_terminated_length": 523.0, | |
| "entropy": 0.2335476577281952, | |
| "epoch": 0.007264957264957265, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.06890398263931274, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0923, | |
| "num_tokens": 1425669.0, | |
| "reward": 0.10287962108850479, | |
| "reward_std": 0.7695233225822449, | |
| "rewards/hint_following/mean": 0.7222222089767456, | |
| "rewards/hint_following/std": 0.6146363019943237, | |
| "rewards/length_penalty/mean": -0.619342565536499, | |
| "rewards/length_penalty/std": 0.2540966868400574, | |
| "sampling/importance_sampling_ratio/max": 2.8489930629730225, | |
| "sampling/importance_sampling_ratio/mean": 0.9912996888160706, | |
| "sampling/importance_sampling_ratio/min": 0.3177354037761688, | |
| "sampling/sampling_logp_difference/max": 1.1465363502502441, | |
| "sampling/sampling_logp_difference/mean": 0.0239370446652174, | |
| "step": 17, | |
| "step_time": 56.41279660095461 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.008962325053289533, | |
| "clip_ratio/high_mean": 0.008962325053289533, | |
| "clip_ratio/low_mean": 0.001638414493451516, | |
| "clip_ratio/low_min": 0.001638414493451516, | |
| "clip_ratio/region_mean": 0.010600739469130835, | |
| "completions/clipped_ratio": 0.2222222238779068, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2681.0, | |
| "completions/mean_length": 1895.4722900390625, | |
| "completions/mean_terminated_length": 1579.8929443359375, | |
| "completions/min_length": 409.0, | |
| "completions/min_terminated_length": 409.0, | |
| "entropy": 0.2237540160616239, | |
| "epoch": 0.007692307692307693, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.07782972604036331, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0547, | |
| "num_tokens": 1501340.0, | |
| "reward": -0.3818241059780121, | |
| "reward_std": 0.9488492608070374, | |
| "rewards/hint_following/mean": 0.25, | |
| "rewards/hint_following/std": 0.8062257766723633, | |
| "rewards/length_penalty/mean": -0.6318240761756897, | |
| "rewards/length_penalty/std": 0.2734360098838806, | |
| "sampling/importance_sampling_ratio/max": 2.2928225994110107, | |
| "sampling/importance_sampling_ratio/mean": 0.9916783571243286, | |
| "sampling/importance_sampling_ratio/min": 0.3001212775707245, | |
| "sampling/sampling_logp_difference/max": 1.203568696975708, | |
| "sampling/sampling_logp_difference/mean": 0.023737527430057526, | |
| "step": 18, | |
| "step_time": 58.57279848307371 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0038812818626562753, | |
| "clip_ratio/high_mean": 0.0038812818626562753, | |
| "clip_ratio/low_mean": 0.005094039059864978, | |
| "clip_ratio/low_min": 0.005094039059864978, | |
| "clip_ratio/region_mean": 0.00897532096132636, | |
| "completions/clipped_ratio": 0.1666666716337204, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2937.0, | |
| "completions/mean_length": 2020.138916015625, | |
| "completions/mean_terminated_length": 1824.166748046875, | |
| "completions/min_length": 765.0, | |
| "completions/min_terminated_length": 765.0, | |
| "entropy": 0.2477496862411499, | |
| "epoch": 0.00811965811965812, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.07387188822031021, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0524, | |
| "num_tokens": 1581145.0, | |
| "reward": -0.39560189843177795, | |
| "reward_std": 0.9154050350189209, | |
| "rewards/hint_following/mean": 0.2777777910232544, | |
| "rewards/hint_following/std": 0.7410845756530762, | |
| "rewards/length_penalty/mean": -0.67337965965271, | |
| "rewards/length_penalty/std": 0.24809792637825012, | |
| "sampling/importance_sampling_ratio/max": 2.1414597034454346, | |
| "sampling/importance_sampling_ratio/mean": 0.9909990429878235, | |
| "sampling/importance_sampling_ratio/min": 0.250805139541626, | |
| "sampling/sampling_logp_difference/max": 1.383078932762146, | |
| "sampling/sampling_logp_difference/mean": 0.025834236294031143, | |
| "step": 19, | |
| "step_time": 58.03654656501021 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0034701917708540955, | |
| "clip_ratio/high_mean": 0.0034701917708540955, | |
| "clip_ratio/low_mean": 0.004899940686300397, | |
| "clip_ratio/low_min": 0.004899940686300397, | |
| "clip_ratio/region_mean": 0.00837013234073917, | |
| "completions/clipped_ratio": 0.2777777910232544, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2645.0, | |
| "completions/mean_length": 1926.1666259765625, | |
| "completions/mean_terminated_length": 1513.1539306640625, | |
| "completions/min_length": 1111.0, | |
| "completions/min_terminated_length": 1111.0, | |
| "entropy": 0.2643888195355733, | |
| "epoch": 0.008547008547008548, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.07248850166797638, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0748, | |
| "num_tokens": 1658131.0, | |
| "reward": -0.25316667556762695, | |
| "reward_std": 1.1320863962173462, | |
| "rewards/hint_following/mean": 0.3888888955116272, | |
| "rewards/hint_following/std": 0.903256893157959, | |
| "rewards/length_penalty/mean": -0.6420555114746094, | |
| "rewards/length_penalty/std": 0.2418719381093979, | |
| "sampling/importance_sampling_ratio/max": 2.639763593673706, | |
| "sampling/importance_sampling_ratio/mean": 0.990202009677887, | |
| "sampling/importance_sampling_ratio/min": 0.3388563394546509, | |
| "sampling/sampling_logp_difference/max": 1.082179069519043, | |
| "sampling/sampling_logp_difference/mean": 0.026784788817167282, | |
| "step": 20, | |
| "step_time": 59.42642685002647 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0018190126866102219, | |
| "clip_ratio/high_mean": 0.0018190126866102219, | |
| "clip_ratio/low_mean": 0.005009224289096892, | |
| "clip_ratio/low_min": 0.005009224289096892, | |
| "clip_ratio/region_mean": 0.006828237092122436, | |
| "completions/clipped_ratio": 0.3611111044883728, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2893.0, | |
| "completions/mean_length": 2087.0556640625, | |
| "completions/mean_terminated_length": 1571.04345703125, | |
| "completions/min_length": 621.0, | |
| "completions/min_terminated_length": 621.0, | |
| "entropy": 0.23874077697594961, | |
| "epoch": 0.008974358974358974, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.07083278149366379, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0432, | |
| "num_tokens": 1741971.0, | |
| "reward": -0.7512407302856445, | |
| "reward_std": 1.0566543340682983, | |
| "rewards/hint_following/mean": -0.0555555559694767, | |
| "rewards/hint_following/std": 0.826159656047821, | |
| "rewards/length_penalty/mean": -0.6956851482391357, | |
| "rewards/length_penalty/std": 0.29789718985557556, | |
| "sampling/importance_sampling_ratio/max": 2.2389557361602783, | |
| "sampling/importance_sampling_ratio/mean": 0.9909273386001587, | |
| "sampling/importance_sampling_ratio/min": 0.1903061866760254, | |
| "sampling/sampling_logp_difference/max": 1.659121036529541, | |
| "sampling/sampling_logp_difference/mean": 0.024532614275813103, | |
| "step": 21, | |
| "step_time": 62.53779820702039 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.002690806732668231, | |
| "clip_ratio/high_mean": 0.002690806732668231, | |
| "clip_ratio/low_mean": 0.0009356096270494163, | |
| "clip_ratio/low_min": 0.0009356096270494163, | |
| "clip_ratio/region_mean": 0.003626416379120201, | |
| "completions/clipped_ratio": 0.1111111119389534, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2897.0, | |
| "completions/mean_length": 1785.02783203125, | |
| "completions/mean_terminated_length": 1633.15625, | |
| "completions/min_length": 687.0, | |
| "completions/min_terminated_length": 687.0, | |
| "entropy": 0.2271458531419436, | |
| "epoch": 0.009401709401709401, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.06376982480287552, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0558, | |
| "num_tokens": 1812208.0, | |
| "reward": -0.12278702855110168, | |
| "reward_std": 0.8545697927474976, | |
| "rewards/hint_following/mean": 0.4722222089767456, | |
| "rewards/hint_following/std": 0.6963624358177185, | |
| "rewards/length_penalty/mean": -0.5950092673301697, | |
| "rewards/length_penalty/std": 0.2556186616420746, | |
| "sampling/importance_sampling_ratio/max": 2.0840954780578613, | |
| "sampling/importance_sampling_ratio/mean": 0.9917669892311096, | |
| "sampling/importance_sampling_ratio/min": 0.4361201226711273, | |
| "sampling/sampling_logp_difference/max": 0.8298375606536865, | |
| "sampling/sampling_logp_difference/mean": 0.02324368618428707, | |
| "step": 22, | |
| "step_time": 54.28889209416229 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.00034443634406973916, | |
| "clip_ratio/high_mean": 0.00034443634406973916, | |
| "clip_ratio/low_mean": 0.001613353689511617, | |
| "clip_ratio/low_min": 0.001613353689511617, | |
| "clip_ratio/region_mean": 0.001957789994776249, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 2755.0, | |
| "completions/max_terminated_length": 2755.0, | |
| "completions/mean_length": 1515.5555419921875, | |
| "completions/mean_terminated_length": 1515.5555419921875, | |
| "completions/min_length": 761.0, | |
| "completions/min_terminated_length": 761.0, | |
| "entropy": 0.26160697638988495, | |
| "epoch": 0.009829059829059829, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.05521637201309204, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0467, | |
| "num_tokens": 1873452.0, | |
| "reward": 0.18925924599170685, | |
| "reward_std": 0.6213830709457397, | |
| "rewards/hint_following/mean": 0.6944444179534912, | |
| "rewards/hint_following/std": 0.467176616191864, | |
| "rewards/length_penalty/mean": -0.5051851868629456, | |
| "rewards/length_penalty/std": 0.1804182529449463, | |
| "sampling/importance_sampling_ratio/max": 2.6960017681121826, | |
| "sampling/importance_sampling_ratio/mean": 0.9903501272201538, | |
| "sampling/importance_sampling_ratio/min": 0.2622704803943634, | |
| "sampling/sampling_logp_difference/max": 1.33837890625, | |
| "sampling/sampling_logp_difference/mean": 0.026966720819473267, | |
| "step": 23, | |
| "step_time": 49.74234241596423 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.009169099697222313, | |
| "clip_ratio/high_mean": 0.009169099697222313, | |
| "clip_ratio/low_mean": 0.0011019935482181609, | |
| "clip_ratio/low_min": 0.0011019935482181609, | |
| "clip_ratio/region_mean": 0.010271093342453241, | |
| "completions/clipped_ratio": 0.02777777798473835, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2438.0, | |
| "completions/mean_length": 1354.77783203125, | |
| "completions/mean_terminated_length": 1307.771484375, | |
| "completions/min_length": 358.0, | |
| "completions/min_terminated_length": 358.0, | |
| "entropy": 0.28231214980284375, | |
| "epoch": 0.010256410256410256, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.061877232044935226, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0513, | |
| "num_tokens": 1928578.0, | |
| "reward": 0.0761851817369461, | |
| "reward_std": 0.6364034414291382, | |
| "rewards/hint_following/mean": 0.5277777910232544, | |
| "rewards/hint_following/std": 0.559903621673584, | |
| "rewards/length_penalty/mean": -0.4515925645828247, | |
| "rewards/length_penalty/std": 0.2236867994070053, | |
| "sampling/importance_sampling_ratio/max": 2.3720901012420654, | |
| "sampling/importance_sampling_ratio/mean": 0.9896314144134521, | |
| "sampling/importance_sampling_ratio/min": 0.3866640329360962, | |
| "sampling/sampling_logp_difference/max": 0.9501991271972656, | |
| "sampling/sampling_logp_difference/mean": 0.028320252895355225, | |
| "step": 24, | |
| "step_time": 51.51962701487355 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.007340530709673961, | |
| "clip_ratio/high_mean": 0.007340530709673961, | |
| "clip_ratio/low_mean": 0.002632848802022636, | |
| "clip_ratio/low_min": 0.002632848802022636, | |
| "clip_ratio/region_mean": 0.00997337931767106, | |
| "completions/clipped_ratio": 0.0555555559694767, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2833.0, | |
| "completions/mean_length": 1369.52783203125, | |
| "completions/mean_terminated_length": 1273.61767578125, | |
| "completions/min_length": 489.0, | |
| "completions/min_terminated_length": 489.0, | |
| "entropy": 0.24688356121381125, | |
| "epoch": 0.010683760683760684, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.0653451606631279, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0562, | |
| "num_tokens": 1983563.0, | |
| "reward": -0.15095369517803192, | |
| "reward_std": 0.692315936088562, | |
| "rewards/hint_following/mean": 0.3055555522441864, | |
| "rewards/hint_following/std": 0.576662540435791, | |
| "rewards/length_penalty/mean": -0.4565092623233795, | |
| "rewards/length_penalty/std": 0.22569186985492706, | |
| "sampling/importance_sampling_ratio/max": 2.21685791015625, | |
| "sampling/importance_sampling_ratio/mean": 0.9907597899436951, | |
| "sampling/importance_sampling_ratio/min": 0.26015329360961914, | |
| "sampling/sampling_logp_difference/max": 1.3464841842651367, | |
| "sampling/sampling_logp_difference/mean": 0.02600344829261303, | |
| "step": 25, | |
| "step_time": 53.30105395393912 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.005985865951515734, | |
| "clip_ratio/high_mean": 0.005985865951515734, | |
| "clip_ratio/low_mean": 0.0026564535413247845, | |
| "clip_ratio/low_min": 0.0026564535413247845, | |
| "clip_ratio/region_mean": 0.008642319512243072, | |
| "completions/clipped_ratio": 0.0833333358168602, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2935.0, | |
| "completions/mean_length": 1874.6944580078125, | |
| "completions/mean_terminated_length": 1772.39404296875, | |
| "completions/min_length": 921.0, | |
| "completions/min_terminated_length": 921.0, | |
| "entropy": 0.24719314028819403, | |
| "epoch": 0.011111111111111112, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.07119674980640411, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0874, | |
| "num_tokens": 2058072.0, | |
| "reward": -0.06934259831905365, | |
| "reward_std": 0.8091042637825012, | |
| "rewards/hint_following/mean": 0.5555555820465088, | |
| "rewards/hint_following/std": 0.6522245407104492, | |
| "rewards/length_penalty/mean": -0.6248981952667236, | |
| "rewards/length_penalty/std": 0.21120469272136688, | |
| "sampling/importance_sampling_ratio/max": 2.3882269859313965, | |
| "sampling/importance_sampling_ratio/mean": 0.9909408688545227, | |
| "sampling/importance_sampling_ratio/min": 0.3095887005329132, | |
| "sampling/sampling_logp_difference/max": 1.1725106239318848, | |
| "sampling/sampling_logp_difference/mean": 0.024691704660654068, | |
| "step": 26, | |
| "step_time": 55.06593968113884 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.006906990543939173, | |
| "clip_ratio/high_mean": 0.006906990543939173, | |
| "clip_ratio/low_mean": 0.002326829261922588, | |
| "clip_ratio/low_min": 0.002326829261922588, | |
| "clip_ratio/region_mean": 0.009233820019289851, | |
| "completions/clipped_ratio": 0.1388888955116272, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2980.0, | |
| "completions/mean_length": 1767.5833740234375, | |
| "completions/mean_terminated_length": 1568.806396484375, | |
| "completions/min_length": 614.0, | |
| "completions/min_terminated_length": 614.0, | |
| "entropy": 0.25413808474938077, | |
| "epoch": 0.011538461538461539, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.06020393222570419, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0678, | |
| "num_tokens": 2126523.0, | |
| "reward": -0.3114166855812073, | |
| "reward_std": 0.8995095491409302, | |
| "rewards/hint_following/mean": 0.2777777910232544, | |
| "rewards/hint_following/std": 0.7014724016189575, | |
| "rewards/length_penalty/mean": -0.5891944169998169, | |
| "rewards/length_penalty/std": 0.24767503142356873, | |
| "sampling/importance_sampling_ratio/max": 2.9181816577911377, | |
| "sampling/importance_sampling_ratio/mean": 0.9907821416854858, | |
| "sampling/importance_sampling_ratio/min": 0.28627336025238037, | |
| "sampling/sampling_logp_difference/max": 1.2508081197738647, | |
| "sampling/sampling_logp_difference/mean": 0.025369901210069656, | |
| "step": 27, | |
| "step_time": 54.32182147912681 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.3333333432674408, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 1529.0, | |
| "completions/mean_length": 1610.5833740234375, | |
| "completions/mean_terminated_length": 915.875, | |
| "completions/min_length": 478.0, | |
| "completions/min_terminated_length": 478.0, | |
| "entropy": 0.19960021724303564, | |
| "epoch": 0.011965811965811967, | |
| "frac_reward_zero_std": 1.0, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0, | |
| "num_tokens": 2192742.0, | |
| "reward": -0.8701944351196289, | |
| "reward_std": 0.8150903582572937, | |
| "rewards/hint_following/mean": -0.3333333432674408, | |
| "rewards/hint_following/std": 0.47809144854545593, | |
| "rewards/length_penalty/mean": -0.5368611216545105, | |
| "rewards/length_penalty/std": 0.3438311517238617, | |
| "sampling/importance_sampling_ratio/max": 2.0817480087280273, | |
| "sampling/importance_sampling_ratio/mean": 0.9929863214492798, | |
| "sampling/importance_sampling_ratio/min": 0.09024836868047714, | |
| "sampling/sampling_logp_difference/max": 2.4051897525787354, | |
| "sampling/sampling_logp_difference/mean": 0.02168331854045391, | |
| "step": 28, | |
| "step_time": 56.78273802890908 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.001870962364288668, | |
| "clip_ratio/high_mean": 0.001870962364288668, | |
| "clip_ratio/low_mean": 0.004373128215471904, | |
| "clip_ratio/low_min": 0.004373128215471904, | |
| "clip_ratio/region_mean": 0.006244090540955464, | |
| "completions/clipped_ratio": 0.1944444477558136, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2978.0, | |
| "completions/mean_length": 1729.77783203125, | |
| "completions/mean_terminated_length": 1423.17236328125, | |
| "completions/min_length": 529.0, | |
| "completions/min_terminated_length": 529.0, | |
| "entropy": 0.28223206599553424, | |
| "epoch": 0.012393162393162393, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.07922516763210297, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0378, | |
| "num_tokens": 2262334.0, | |
| "reward": -0.3265925943851471, | |
| "reward_std": 0.9465053081512451, | |
| "rewards/hint_following/mean": 0.25, | |
| "rewards/hint_following/std": 0.7699722051620483, | |
| "rewards/length_penalty/mean": -0.5765926241874695, | |
| "rewards/length_penalty/std": 0.28062164783477783, | |
| "sampling/importance_sampling_ratio/max": 3.0, | |
| "sampling/importance_sampling_ratio/mean": 0.9894134402275085, | |
| "sampling/importance_sampling_ratio/min": 0.29406988620758057, | |
| "sampling/sampling_logp_difference/max": 1.760861873626709, | |
| "sampling/sampling_logp_difference/mean": 0.026971129700541496, | |
| "step": 29, | |
| "step_time": 55.79818298702594 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.001682481961324811, | |
| "clip_ratio/high_mean": 0.001682481961324811, | |
| "clip_ratio/low_mean": 0.00047758713481016457, | |
| "clip_ratio/low_min": 0.00047758713481016457, | |
| "clip_ratio/region_mean": 0.002160069086433699, | |
| "completions/clipped_ratio": 0.1666666716337204, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2747.0, | |
| "completions/mean_length": 1684.638916015625, | |
| "completions/mean_terminated_length": 1421.5667724609375, | |
| "completions/min_length": 805.0, | |
| "completions/min_terminated_length": 805.0, | |
| "entropy": 0.2508200804392497, | |
| "epoch": 0.01282051282051282, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.06276588886976242, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0312, | |
| "num_tokens": 2331153.0, | |
| "reward": 0.02178703434765339, | |
| "reward_std": 0.9737120270729065, | |
| "rewards/hint_following/mean": 0.5833333134651184, | |
| "rewards/hint_following/std": 0.7699722051620483, | |
| "rewards/length_penalty/mean": -0.5615463256835938, | |
| "rewards/length_penalty/std": 0.233141228556633, | |
| "sampling/importance_sampling_ratio/max": 2.9634652137756348, | |
| "sampling/importance_sampling_ratio/mean": 0.9905918836593628, | |
| "sampling/importance_sampling_ratio/min": 0.2952629625797272, | |
| "sampling/sampling_logp_difference/max": 1.2198889255523682, | |
| "sampling/sampling_logp_difference/mean": 0.024356598034501076, | |
| "step": 30, | |
| "step_time": 56.55290222284384 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.008109931523601214, | |
| "clip_ratio/high_mean": 0.008109931523601214, | |
| "clip_ratio/low_mean": 0.0004286583377203594, | |
| "clip_ratio/low_min": 0.0004286583377203594, | |
| "clip_ratio/region_mean": 0.008538589735204974, | |
| "completions/clipped_ratio": 0.0833333358168602, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2567.0, | |
| "completions/mean_length": 1786.2222900390625, | |
| "completions/mean_terminated_length": 1675.8787841796875, | |
| "completions/min_length": 468.0, | |
| "completions/min_terminated_length": 468.0, | |
| "entropy": 0.2908742278814316, | |
| "epoch": 0.013247863247863248, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.07696390151977539, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0561, | |
| "num_tokens": 2403953.0, | |
| "reward": -0.0954074040055275, | |
| "reward_std": 0.6877580285072327, | |
| "rewards/hint_following/mean": 0.5, | |
| "rewards/hint_following/std": 0.6546536684036255, | |
| "rewards/length_penalty/mean": -0.5954073667526245, | |
| "rewards/length_penalty/std": 0.22547654807567596, | |
| "sampling/importance_sampling_ratio/max": 2.491806745529175, | |
| "sampling/importance_sampling_ratio/mean": 0.9886121153831482, | |
| "sampling/importance_sampling_ratio/min": 0.16667605936527252, | |
| "sampling/sampling_logp_difference/max": 1.7917031049728394, | |
| "sampling/sampling_logp_difference/mean": 0.02771041914820671, | |
| "step": 31, | |
| "step_time": 58.1784061481012 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0027225150261074305, | |
| "clip_ratio/high_mean": 0.0027225150261074305, | |
| "clip_ratio/low_mean": 0.003243642975576222, | |
| "clip_ratio/low_min": 0.003243642975576222, | |
| "clip_ratio/region_mean": 0.00596615804048876, | |
| "completions/clipped_ratio": 0.1666666716337204, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2917.0, | |
| "completions/mean_length": 2085.916748046875, | |
| "completions/mean_terminated_length": 1903.10009765625, | |
| "completions/min_length": 1014.0, | |
| "completions/min_terminated_length": 1014.0, | |
| "entropy": 0.278283953666687, | |
| "epoch": 0.013675213675213675, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.06229299306869507, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0363, | |
| "num_tokens": 2486342.0, | |
| "reward": -0.36197224259376526, | |
| "reward_std": 0.8716595768928528, | |
| "rewards/hint_following/mean": 0.3333333432674408, | |
| "rewards/hint_following/std": 0.7559289932250977, | |
| "rewards/length_penalty/mean": -0.695305585861206, | |
| "rewards/length_penalty/std": 0.1993320733308792, | |
| "sampling/importance_sampling_ratio/max": 2.49179744720459, | |
| "sampling/importance_sampling_ratio/mean": 0.9892889261245728, | |
| "sampling/importance_sampling_ratio/min": 0.19078373908996582, | |
| "sampling/sampling_logp_difference/max": 1.6566147804260254, | |
| "sampling/sampling_logp_difference/mean": 0.02677111327648163, | |
| "step": 32, | |
| "step_time": 59.783586613833904 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0016072407306637615, | |
| "clip_ratio/high_mean": 0.0016072407306637615, | |
| "clip_ratio/low_mean": 0.0016927721832568448, | |
| "clip_ratio/low_min": 0.0016927721832568448, | |
| "clip_ratio/region_mean": 0.003300012855712945, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 2629.0, | |
| "completions/max_terminated_length": 2629.0, | |
| "completions/mean_length": 1464.5833740234375, | |
| "completions/mean_terminated_length": 1464.5833740234375, | |
| "completions/min_length": 577.0, | |
| "completions/min_terminated_length": 577.0, | |
| "entropy": 0.25622066855430603, | |
| "epoch": 0.014102564102564103, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.05397763103246689, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0059, | |
| "num_tokens": 2544023.0, | |
| "reward": -0.29375001788139343, | |
| "reward_std": 0.45689767599105835, | |
| "rewards/hint_following/mean": 0.1944444477558136, | |
| "rewards/hint_following/std": 0.4013865292072296, | |
| "rewards/length_penalty/mean": -0.48819446563720703, | |
| "rewards/length_penalty/std": 0.14601026475429535, | |
| "sampling/importance_sampling_ratio/max": 2.081763982772827, | |
| "sampling/importance_sampling_ratio/mean": 0.9904037117958069, | |
| "sampling/importance_sampling_ratio/min": 0.339209645986557, | |
| "sampling/sampling_logp_difference/max": 1.08113694190979, | |
| "sampling/sampling_logp_difference/mean": 0.02619028650224209, | |
| "step": 33, | |
| "step_time": 49.13755757326726 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.001959889098846664, | |
| "clip_ratio/high_mean": 0.001959889098846664, | |
| "clip_ratio/low_mean": 0.0017016992593804996, | |
| "clip_ratio/low_min": 0.0017016992593804996, | |
| "clip_ratio/region_mean": 0.0036615883776297173, | |
| "completions/clipped_ratio": 0.1666666716337204, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2814.0, | |
| "completions/mean_length": 1581.1666259765625, | |
| "completions/mean_terminated_length": 1297.4000244140625, | |
| "completions/min_length": 447.0, | |
| "completions/min_terminated_length": 447.0, | |
| "entropy": 0.25040922313928604, | |
| "epoch": 0.01452991452991453, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.06385543942451477, | |
| "learning_rate": 5e-05, | |
| "loss": 0.046, | |
| "num_tokens": 2606177.0, | |
| "reward": -0.0826110988855362, | |
| "reward_std": 0.9639437198638916, | |
| "rewards/hint_following/mean": 0.4444444477558136, | |
| "rewards/hint_following/std": 0.7725447416305542, | |
| "rewards/length_penalty/mean": -0.5270555019378662, | |
| "rewards/length_penalty/std": 0.2994069457054138, | |
| "sampling/importance_sampling_ratio/max": 3.0, | |
| "sampling/importance_sampling_ratio/mean": 0.9912077784538269, | |
| "sampling/importance_sampling_ratio/min": 0.31699931621551514, | |
| "sampling/sampling_logp_difference/max": 1.2634482383728027, | |
| "sampling/sampling_logp_difference/mean": 0.02445373497903347, | |
| "step": 34, | |
| "step_time": 54.58746225386858 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0011685320253794391, | |
| "clip_ratio/high_mean": 0.0011685320253794391, | |
| "clip_ratio/low_mean": 0.0014686054394890864, | |
| "clip_ratio/low_min": 0.0014686054394890864, | |
| "clip_ratio/region_mean": 0.0026371375230761864, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 2743.0, | |
| "completions/max_terminated_length": 2743.0, | |
| "completions/mean_length": 1370.4722900390625, | |
| "completions/mean_terminated_length": 1370.4722900390625, | |
| "completions/min_length": 650.0, | |
| "completions/min_terminated_length": 650.0, | |
| "entropy": 0.26418235152959824, | |
| "epoch": 0.014957264957264958, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.057452693581581116, | |
| "learning_rate": 5e-05, | |
| "loss": 0.01, | |
| "num_tokens": 2661580.0, | |
| "reward": 0.15428705513477325, | |
| "reward_std": 0.5879056453704834, | |
| "rewards/hint_following/mean": 0.6111111044883728, | |
| "rewards/hint_following/std": 0.49441322684288025, | |
| "rewards/length_penalty/mean": -0.45682409405708313, | |
| "rewards/length_penalty/std": 0.1867014318704605, | |
| "sampling/importance_sampling_ratio/max": 3.0, | |
| "sampling/importance_sampling_ratio/mean": 0.9900287985801697, | |
| "sampling/importance_sampling_ratio/min": 0.15399949252605438, | |
| "sampling/sampling_logp_difference/max": 1.8708059787750244, | |
| "sampling/sampling_logp_difference/mean": 0.027402520179748535, | |
| "step": 35, | |
| "step_time": 47.69401362503413 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0057743463742857175, | |
| "clip_ratio/high_mean": 0.0057743463742857175, | |
| "clip_ratio/low_mean": 0.0029491251916624606, | |
| "clip_ratio/low_min": 0.0029491251916624606, | |
| "clip_ratio/region_mean": 0.00872347146893541, | |
| "completions/clipped_ratio": 0.1111111119389534, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2813.0, | |
| "completions/mean_length": 1503.6944580078125, | |
| "completions/mean_terminated_length": 1316.65625, | |
| "completions/min_length": 456.0, | |
| "completions/min_terminated_length": 456.0, | |
| "entropy": 0.2601601282755534, | |
| "epoch": 0.015384615384615385, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.06052026152610779, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0217, | |
| "num_tokens": 2721131.0, | |
| "reward": -0.44567596912384033, | |
| "reward_std": 0.6641168594360352, | |
| "rewards/hint_following/mean": 0.0555555559694767, | |
| "rewards/hint_following/std": 0.5315446853637695, | |
| "rewards/length_penalty/mean": -0.5012314915657043, | |
| "rewards/length_penalty/std": 0.2637265622615814, | |
| "sampling/importance_sampling_ratio/max": 2.678779125213623, | |
| "sampling/importance_sampling_ratio/mean": 0.98993980884552, | |
| "sampling/importance_sampling_ratio/min": 0.3602158725261688, | |
| "sampling/sampling_logp_difference/max": 1.0210518836975098, | |
| "sampling/sampling_logp_difference/mean": 0.02683359384536743, | |
| "step": 36, | |
| "step_time": 52.9094661710551 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.002038249202693502, | |
| "clip_ratio/high_mean": 0.002038249202693502, | |
| "clip_ratio/low_mean": 0.00521218675809602, | |
| "clip_ratio/low_min": 0.00521218675809602, | |
| "clip_ratio/region_mean": 0.007250435883179307, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 2391.0, | |
| "completions/max_terminated_length": 2391.0, | |
| "completions/mean_length": 1279.638916015625, | |
| "completions/mean_terminated_length": 1279.638916015625, | |
| "completions/min_length": 508.0, | |
| "completions/min_terminated_length": 508.0, | |
| "entropy": 0.26387400925159454, | |
| "epoch": 0.01581196581196581, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.0660337284207344, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0072, | |
| "num_tokens": 2774056.0, | |
| "reward": 0.12900924682617188, | |
| "reward_std": 0.5431578159332275, | |
| "rewards/hint_following/mean": 0.5555555820465088, | |
| "rewards/hint_following/std": 0.5039526224136353, | |
| "rewards/length_penalty/mean": -0.4265463054180145, | |
| "rewards/length_penalty/std": 0.12839393317699432, | |
| "sampling/importance_sampling_ratio/max": 2.7058398723602295, | |
| "sampling/importance_sampling_ratio/mean": 0.9903796911239624, | |
| "sampling/importance_sampling_ratio/min": 0.1684948354959488, | |
| "sampling/sampling_logp_difference/max": 1.7808502912521362, | |
| "sampling/sampling_logp_difference/mean": 0.027670899406075478, | |
| "step": 37, | |
| "step_time": 42.42564612603746 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.005249147206389655, | |
| "clip_ratio/high_mean": 0.005249147206389655, | |
| "clip_ratio/low_mean": 0.0019288610492367297, | |
| "clip_ratio/low_min": 0.0019288610492367297, | |
| "clip_ratio/region_mean": 0.007178008245925109, | |
| "completions/clipped_ratio": 0.1388888955116272, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2773.0, | |
| "completions/mean_length": 1562.9166259765625, | |
| "completions/mean_terminated_length": 1331.1290283203125, | |
| "completions/min_length": 508.0, | |
| "completions/min_terminated_length": 508.0, | |
| "entropy": 0.2739727521936099, | |
| "epoch": 0.01623931623931624, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.05931999906897545, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0111, | |
| "num_tokens": 2836603.0, | |
| "reward": 0.034583330154418945, | |
| "reward_std": 0.9446364045143127, | |
| "rewards/hint_following/mean": 0.5555555820465088, | |
| "rewards/hint_following/std": 0.7346308827400208, | |
| "rewards/length_penalty/mean": -0.5209722518920898, | |
| "rewards/length_penalty/std": 0.2565736174583435, | |
| "sampling/importance_sampling_ratio/max": 2.9026551246643066, | |
| "sampling/importance_sampling_ratio/mean": 0.9896891117095947, | |
| "sampling/importance_sampling_ratio/min": 0.1531003713607788, | |
| "sampling/sampling_logp_difference/max": 1.8766615390777588, | |
| "sampling/sampling_logp_difference/mean": 0.02661670744419098, | |
| "step": 38, | |
| "step_time": 54.65668713499326 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0073317300993949175, | |
| "clip_ratio/high_mean": 0.0073317300993949175, | |
| "clip_ratio/low_mean": 0.0012036147963954136, | |
| "clip_ratio/low_min": 0.0012036147963954136, | |
| "clip_ratio/region_mean": 0.008535344852134585, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 2373.0, | |
| "completions/max_terminated_length": 2373.0, | |
| "completions/mean_length": 1587.7222900390625, | |
| "completions/mean_terminated_length": 1587.7222900390625, | |
| "completions/min_length": 887.0, | |
| "completions/min_terminated_length": 887.0, | |
| "entropy": 0.2784932255744934, | |
| "epoch": 0.016666666666666666, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.05303090810775757, | |
| "learning_rate": 5e-05, | |
| "loss": -0.0225, | |
| "num_tokens": 2900313.0, | |
| "reward": 0.22075927257537842, | |
| "reward_std": 0.3945108652114868, | |
| "rewards/hint_following/mean": 0.75, | |
| "rewards/hint_following/std": 0.43915504217147827, | |
| "rewards/length_penalty/mean": -0.5292407274246216, | |
| "rewards/length_penalty/std": 0.10378893464803696, | |
| "sampling/importance_sampling_ratio/max": 2.234417200088501, | |
| "sampling/importance_sampling_ratio/mean": 0.9896702170372009, | |
| "sampling/importance_sampling_ratio/min": 0.22993366420269012, | |
| "sampling/sampling_logp_difference/max": 1.4699643850326538, | |
| "sampling/sampling_logp_difference/mean": 0.027140147984027863, | |
| "step": 39, | |
| "step_time": 45.66687847697176 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.008284316320593158, | |
| "clip_ratio/high_mean": 0.008284316320593158, | |
| "clip_ratio/low_mean": 0.0006285767303779721, | |
| "clip_ratio/low_min": 0.0006285767303779721, | |
| "clip_ratio/region_mean": 0.008912893012166023, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 2490.0, | |
| "completions/max_terminated_length": 2490.0, | |
| "completions/mean_length": 1358.638916015625, | |
| "completions/mean_terminated_length": 1358.638916015625, | |
| "completions/min_length": 668.0, | |
| "completions/min_terminated_length": 668.0, | |
| "entropy": 0.2635483493407567, | |
| "epoch": 0.017094017094017096, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.061241693794727325, | |
| "learning_rate": 5e-05, | |
| "loss": -0.0001, | |
| "num_tokens": 2955062.0, | |
| "reward": 0.2693426012992859, | |
| "reward_std": 0.4795752167701721, | |
| "rewards/hint_following/mean": 0.7222222089767456, | |
| "rewards/hint_following/std": 0.45425674319267273, | |
| "rewards/length_penalty/mean": -0.4528796374797821, | |
| "rewards/length_penalty/std": 0.17493696510791779, | |
| "sampling/importance_sampling_ratio/max": 2.239511728286743, | |
| "sampling/importance_sampling_ratio/mean": 0.9903267621994019, | |
| "sampling/importance_sampling_ratio/min": 0.11606565117835999, | |
| "sampling/sampling_logp_difference/max": 2.153599262237549, | |
| "sampling/sampling_logp_difference/mean": 0.02778884768486023, | |
| "step": 40, | |
| "step_time": 44.78665240900591 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0024162703969826302, | |
| "clip_ratio/high_mean": 0.0024162703969826302, | |
| "clip_ratio/low_mean": 0.004380864285243054, | |
| "clip_ratio/low_min": 0.004380864285243054, | |
| "clip_ratio/region_mean": 0.006797134643420577, | |
| "completions/clipped_ratio": 0.02777777798473835, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2931.0, | |
| "completions/mean_length": 1570.75, | |
| "completions/mean_terminated_length": 1529.914306640625, | |
| "completions/min_length": 437.0, | |
| "completions/min_terminated_length": 437.0, | |
| "entropy": 0.2851356367270152, | |
| "epoch": 0.01752136752136752, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.06564202159643173, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0083, | |
| "num_tokens": 3020207.0, | |
| "reward": 0.1708611100912094, | |
| "reward_std": 0.6461936831474304, | |
| "rewards/hint_following/mean": 0.6944444179534912, | |
| "rewards/hint_following/std": 0.524782657623291, | |
| "rewards/length_penalty/mean": -0.5235832929611206, | |
| "rewards/length_penalty/std": 0.23531726002693176, | |
| "sampling/importance_sampling_ratio/max": 3.0, | |
| "sampling/importance_sampling_ratio/mean": 0.9896803498268127, | |
| "sampling/importance_sampling_ratio/min": 0.2426270991563797, | |
| "sampling/sampling_logp_difference/max": 1.4162296056747437, | |
| "sampling/sampling_logp_difference/mean": 0.02950497530400753, | |
| "step": 41, | |
| "step_time": 55.908626700984314 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0035200511144163706, | |
| "clip_ratio/high_mean": 0.0035200511144163706, | |
| "clip_ratio/low_mean": 0.0037633493387450776, | |
| "clip_ratio/low_min": 0.0037633493387450776, | |
| "clip_ratio/region_mean": 0.007283400433758895, | |
| "completions/clipped_ratio": 0.1944444477558136, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2916.0, | |
| "completions/mean_length": 1708.861083984375, | |
| "completions/mean_terminated_length": 1397.2069091796875, | |
| "completions/min_length": 601.0, | |
| "completions/min_terminated_length": 601.0, | |
| "entropy": 0.297603373726209, | |
| "epoch": 0.017948717948717947, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.06374821066856384, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0258, | |
| "num_tokens": 3087486.0, | |
| "reward": -0.40295371413230896, | |
| "reward_std": 0.9737837910652161, | |
| "rewards/hint_following/mean": 0.1666666716337204, | |
| "rewards/hint_following/std": 0.7367883920669556, | |
| "rewards/length_penalty/mean": -0.5696203708648682, | |
| "rewards/length_penalty/std": 0.28024521470069885, | |
| "sampling/importance_sampling_ratio/max": 2.632629871368408, | |
| "sampling/importance_sampling_ratio/mean": 0.9891336560249329, | |
| "sampling/importance_sampling_ratio/min": 0.16689430177211761, | |
| "sampling/sampling_logp_difference/max": 1.79039466381073, | |
| "sampling/sampling_logp_difference/mean": 0.029687926173210144, | |
| "step": 42, | |
| "step_time": 55.97696550388355 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.005934966184819738, | |
| "clip_ratio/high_mean": 0.005934966184819738, | |
| "clip_ratio/low_mean": 0.0023504976027955613, | |
| "clip_ratio/low_min": 0.0023504976027955613, | |
| "clip_ratio/region_mean": 0.008285463865225514, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 2141.0, | |
| "completions/max_terminated_length": 2141.0, | |
| "completions/mean_length": 1201.638916015625, | |
| "completions/mean_terminated_length": 1201.638916015625, | |
| "completions/min_length": 517.0, | |
| "completions/min_terminated_length": 517.0, | |
| "entropy": 0.29462653398513794, | |
| "epoch": 0.018376068376068377, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.05552098527550697, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0036, | |
| "num_tokens": 3135263.0, | |
| "reward": 0.12723149359226227, | |
| "reward_std": 0.5270769596099854, | |
| "rewards/hint_following/mean": 0.5277777910232544, | |
| "rewards/hint_following/std": 0.506309449672699, | |
| "rewards/length_penalty/mean": -0.40054628252983093, | |
| "rewards/length_penalty/std": 0.14752288162708282, | |
| "sampling/importance_sampling_ratio/max": 3.0, | |
| "sampling/importance_sampling_ratio/mean": 0.9892792701721191, | |
| "sampling/importance_sampling_ratio/min": 0.3035065531730652, | |
| "sampling/sampling_logp_difference/max": 1.192352056503296, | |
| "sampling/sampling_logp_difference/mean": 0.030870094895362854, | |
| "step": 43, | |
| "step_time": 37.3642987072235 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0028009160111347833, | |
| "clip_ratio/high_mean": 0.0028009160111347833, | |
| "clip_ratio/low_mean": 0.003960246841112773, | |
| "clip_ratio/low_min": 0.003960246841112773, | |
| "clip_ratio/region_mean": 0.0067611627746373415, | |
| "completions/clipped_ratio": 0.2777777910232544, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2767.0, | |
| "completions/mean_length": 1834.611083984375, | |
| "completions/mean_terminated_length": 1386.3846435546875, | |
| "completions/min_length": 439.0, | |
| "completions/min_terminated_length": 439.0, | |
| "entropy": 0.2577643742163976, | |
| "epoch": 0.018803418803418803, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.06315090507268906, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0313, | |
| "num_tokens": 3208011.0, | |
| "reward": -0.22264817357063293, | |
| "reward_std": 1.158618688583374, | |
| "rewards/hint_following/mean": 0.3888888955116272, | |
| "rewards/hint_following/std": 0.903256893157959, | |
| "rewards/length_penalty/mean": -0.6115370988845825, | |
| "rewards/length_penalty/std": 0.3052543103694916, | |
| "sampling/importance_sampling_ratio/max": 2.770780324935913, | |
| "sampling/importance_sampling_ratio/mean": 0.9908032417297363, | |
| "sampling/importance_sampling_ratio/min": 0.14089906215667725, | |
| "sampling/sampling_logp_difference/max": 1.9597115516662598, | |
| "sampling/sampling_logp_difference/mean": 0.026477616280317307, | |
| "step": 44, | |
| "step_time": 57.508703485131264 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0022431810890945294, | |
| "clip_ratio/high_mean": 0.0022431810890945294, | |
| "clip_ratio/low_mean": 0.0008466135962711027, | |
| "clip_ratio/low_min": 0.0008466135962711027, | |
| "clip_ratio/region_mean": 0.003089794685365632, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 2567.0, | |
| "completions/max_terminated_length": 2567.0, | |
| "completions/mean_length": 1403.611083984375, | |
| "completions/mean_terminated_length": 1403.611083984375, | |
| "completions/min_length": 791.0, | |
| "completions/min_terminated_length": 791.0, | |
| "entropy": 0.2975164105494817, | |
| "epoch": 0.019230769230769232, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.0667228028178215, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0231, | |
| "num_tokens": 3265633.0, | |
| "reward": 0.39324072003364563, | |
| "reward_std": 0.3923882842063904, | |
| "rewards/hint_following/mean": 0.8611111044883728, | |
| "rewards/hint_following/std": 0.35073620080947876, | |
| "rewards/length_penalty/mean": -0.4678703546524048, | |
| "rewards/length_penalty/std": 0.15880942344665527, | |
| "sampling/importance_sampling_ratio/max": 2.800001382827759, | |
| "sampling/importance_sampling_ratio/mean": 0.9887976050376892, | |
| "sampling/importance_sampling_ratio/min": 0.2656780183315277, | |
| "sampling/sampling_logp_difference/max": 1.325470209121704, | |
| "sampling/sampling_logp_difference/mean": 0.030811548233032227, | |
| "step": 45, | |
| "step_time": 45.27160808304325 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.007887479305888215, | |
| "clip_ratio/high_mean": 0.007887479305888215, | |
| "clip_ratio/low_mean": 0.000994703887651364, | |
| "clip_ratio/low_min": 0.000994703887651364, | |
| "clip_ratio/region_mean": 0.008882183115929365, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 2359.0, | |
| "completions/max_terminated_length": 2359.0, | |
| "completions/mean_length": 1282.52783203125, | |
| "completions/mean_terminated_length": 1282.52783203125, | |
| "completions/min_length": 561.0, | |
| "completions/min_terminated_length": 561.0, | |
| "entropy": 0.2848661094903946, | |
| "epoch": 0.019658119658119658, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.051431454718112946, | |
| "learning_rate": 5e-05, | |
| "loss": -0.0024, | |
| "num_tokens": 3318716.0, | |
| "reward": 0.37804630398750305, | |
| "reward_std": 0.3656369149684906, | |
| "rewards/hint_following/mean": 0.8055555820465088, | |
| "rewards/hint_following/std": 0.4013865292072296, | |
| "rewards/length_penalty/mean": -0.42750924825668335, | |
| "rewards/length_penalty/std": 0.16107404232025146, | |
| "sampling/importance_sampling_ratio/max": 3.0, | |
| "sampling/importance_sampling_ratio/mean": 0.9896759986877441, | |
| "sampling/importance_sampling_ratio/min": 0.09941968321800232, | |
| "sampling/sampling_logp_difference/max": 2.3084051609039307, | |
| "sampling/sampling_logp_difference/mean": 0.029690390452742577, | |
| "step": 46, | |
| "step_time": 41.876618986018 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.003261159232351929, | |
| "clip_ratio/high_mean": 0.003261159232351929, | |
| "clip_ratio/low_mean": 0.0030720558910009763, | |
| "clip_ratio/low_min": 0.0030720558910009763, | |
| "clip_ratio/region_mean": 0.0063332150457426906, | |
| "completions/clipped_ratio": 0.0833333358168602, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2990.0, | |
| "completions/mean_length": 1868.77783203125, | |
| "completions/mean_terminated_length": 1765.939453125, | |
| "completions/min_length": 947.0, | |
| "completions/min_terminated_length": 947.0, | |
| "entropy": 0.3000795841217041, | |
| "epoch": 0.020085470085470087, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.07183089107275009, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0154, | |
| "num_tokens": 3395160.0, | |
| "reward": -0.12292594462633133, | |
| "reward_std": 0.785350501537323, | |
| "rewards/hint_following/mean": 0.5, | |
| "rewards/hint_following/std": 0.6546536684036255, | |
| "rewards/length_penalty/mean": -0.6229259371757507, | |
| "rewards/length_penalty/std": 0.2120964676141739, | |
| "sampling/importance_sampling_ratio/max": 2.9097461700439453, | |
| "sampling/importance_sampling_ratio/mean": 0.988696277141571, | |
| "sampling/importance_sampling_ratio/min": 0.10777360945940018, | |
| "sampling/sampling_logp_difference/max": 2.227722406387329, | |
| "sampling/sampling_logp_difference/mean": 0.030243437737226486, | |
| "step": 47, | |
| "step_time": 58.67247669992503 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.008134961283455292, | |
| "clip_ratio/high_mean": 0.008134961283455292, | |
| "clip_ratio/low_mean": 0.0004609357565641403, | |
| "clip_ratio/low_min": 0.0004609357565641403, | |
| "clip_ratio/region_mean": 0.008595897040019432, | |
| "completions/clipped_ratio": 0.2222222238779068, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2543.0, | |
| "completions/mean_length": 1645.8333740234375, | |
| "completions/mean_terminated_length": 1258.9285888671875, | |
| "completions/min_length": 729.0, | |
| "completions/min_terminated_length": 729.0, | |
| "entropy": 0.24671275417009988, | |
| "epoch": 0.020512820512820513, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.06476566940546036, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0513, | |
| "num_tokens": 3461970.0, | |
| "reward": 0.006944431457668543, | |
| "reward_std": 1.0957543849945068, | |
| "rewards/hint_following/mean": 0.5555555820465088, | |
| "rewards/hint_following/std": 0.8432740569114685, | |
| "rewards/length_penalty/mean": -0.5486111044883728, | |
| "rewards/length_penalty/std": 0.2772509455680847, | |
| "sampling/importance_sampling_ratio/max": 2.0642147064208984, | |
| "sampling/importance_sampling_ratio/mean": 0.9907264113426208, | |
| "sampling/importance_sampling_ratio/min": 0.04710150137543678, | |
| "sampling/sampling_logp_difference/max": 3.055450439453125, | |
| "sampling/sampling_logp_difference/mean": 0.025603190064430237, | |
| "step": 48, | |
| "step_time": 55.27118606586009 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0011131278927981232, | |
| "clip_ratio/high_mean": 0.0011131278927981232, | |
| "clip_ratio/low_mean": 0.006329516724993785, | |
| "clip_ratio/low_min": 0.006329516724993785, | |
| "clip_ratio/region_mean": 0.007442644564434886, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 1867.0, | |
| "completions/max_terminated_length": 1867.0, | |
| "completions/mean_length": 1157.6666259765625, | |
| "completions/mean_terminated_length": 1157.6666259765625, | |
| "completions/min_length": 722.0, | |
| "completions/min_terminated_length": 722.0, | |
| "entropy": 0.2628251537680626, | |
| "epoch": 0.02094017094017094, | |
| "frac_reward_zero_std": 0.0, | |
| "grad_norm": 0.06659235060214996, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0356, | |
| "num_tokens": 3510510.0, | |
| "reward": 0.5863333344459534, | |
| "reward_std": 0.22435788810253143, | |
| "rewards/hint_following/mean": 0.9722222089767456, | |
| "rewards/hint_following/std": 0.1666666716337204, | |
| "rewards/length_penalty/mean": -0.3858889043331146, | |
| "rewards/length_penalty/std": 0.10318709164857864, | |
| "sampling/importance_sampling_ratio/max": 2.3799936771392822, | |
| "sampling/importance_sampling_ratio/mean": 0.9904940128326416, | |
| "sampling/importance_sampling_ratio/min": 0.06988711655139923, | |
| "sampling/sampling_logp_difference/max": 2.6608738899230957, | |
| "sampling/sampling_logp_difference/mean": 0.030541572719812393, | |
| "step": 49, | |
| "step_time": 33.86789354507346 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.1666666716337204, | |
| "completions/max_length": 3000.0, | |
| "completions/max_terminated_length": 2940.0, | |
| "completions/mean_length": 1683.0, | |
| "completions/mean_terminated_length": 1419.60009765625, | |
| "completions/min_length": 730.0, | |
| "completions/min_terminated_length": 730.0, | |
| "entropy": 0.24505582451820374, | |
| "epoch": 0.021367521367521368, | |
| "frac_reward_zero_std": 1.0, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0, | |
| "num_tokens": 3577332.0, | |
| "reward": 0.10566666722297668, | |
| "reward_std": 0.9825097322463989, | |
| "rewards/hint_following/mean": 0.6666666865348816, | |
| "rewards/hint_following/std": 0.7559289932250977, | |
| "rewards/length_penalty/mean": -0.5609999895095825, | |
| "rewards/length_penalty/std": 0.30474284291267395, | |
| "sampling/importance_sampling_ratio/max": 3.0, | |
| "sampling/importance_sampling_ratio/mean": 0.9918250441551208, | |
| "sampling/importance_sampling_ratio/min": 0.06838197261095047, | |
| "sampling/sampling_logp_difference/max": 2.6826460361480713, | |
| "sampling/sampling_logp_difference/mean": 0.028820084407925606, | |
| "step": 50, | |
| "step_time": 54.36680988012813 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 250, | |
| "num_input_tokens_seen": 3577332, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 6, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |