{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.008502949460594144, "eval_steps": 500, "global_step": 40, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "clip_ratio": 0.0, "completion_length": 216.796875, "epoch": 0.00021257373651485358, "grad_norm": 0.4854881763458252, "kl": 9.614229202270508e-05, "learning_rate": 9.997874149659865e-07, "loss": 0.0, "reward": 2.732285737991333, "reward_std": 0.02619727296405472, "rewards/format_reward_hoi_key": 0.9139583259820938, "rewards/format_reward_hoi_object_label": 0.8222222253680229, "rewards/format_reward_hoi_verb_label": 0.3161458373069763, "rewards/hoi_iou_reward": 0.6799592822790146, "step": 1 }, { "clip_ratio": 0.0, "completion_length": 173.3125, "epoch": 0.00042514747302970716, "grad_norm": 0.6831408739089966, "kl": 1.3329088687896729e-05, "learning_rate": 9.995748299319728e-07, "loss": 0.0, "reward": 2.8274163007736206, "reward_std": 0.03815040903282352, "rewards/format_reward_hoi_key": 0.8166666775941849, "rewards/format_reward_hoi_object_label": 0.7916666567325592, "rewards/format_reward_hoi_verb_label": 0.5974702388048172, "rewards/hoi_iou_reward": 0.6216127127408981, "step": 2 }, { "clip_ratio": 0.0, "completion_length": 166.03125, "epoch": 0.0006377212095445608, "grad_norm": 0.8407193422317505, "kl": 0.00014454126358032227, "learning_rate": 9.99362244897959e-07, "loss": 0.0, "reward": 2.986231029033661, "reward_std": 0.0052611194987548515, "rewards/format_reward_hoi_key": 0.8208333402872086, "rewards/format_reward_hoi_object_label": 0.84375, "rewards/format_reward_hoi_verb_label": 0.6927083432674408, "rewards/hoi_iou_reward": 0.6289393231272697, "step": 3 }, { "clip_ratio": 0.0, "completion_length": 185.703125, "epoch": 0.0008502949460594143, "grad_norm": 2.3168516159057617, "kl": 0.00014531612396240234, "learning_rate": 9.991496598639456e-07, "loss": 0.0, "reward": 2.3956105709075928, "reward_std": 0.045728508091997355, "rewards/format_reward_hoi_key": 0.7395220696926117, "rewards/format_reward_hoi_object_label": 0.59375, "rewards/format_reward_hoi_verb_label": 0.5073784738779068, "rewards/hoi_iou_reward": 0.5549599975347519, "step": 4 }, { "clip_ratio": 0.0, "completion_length": 180.421875, "epoch": 0.001062868682574268, "grad_norm": 0.5881515741348267, "kl": 0.00014841556549072266, "learning_rate": 9.989370748299319e-07, "loss": 0.0, "reward": 2.2462641298770905, "reward_std": 0.14320564700756222, "rewards/format_reward_hoi_key": 0.7350446432828903, "rewards/format_reward_hoi_object_label": 0.4899553433060646, "rewards/format_reward_hoi_verb_label": 0.5563345961272717, "rewards/hoi_iou_reward": 0.46492957696318626, "step": 5 }, { "clip_ratio": 0.0, "completion_length": 208.28125, "epoch": 0.0012754424190891216, "grad_norm": 0.29585161805152893, "kl": 0.0001379847526550293, "learning_rate": 9.987244897959182e-07, "loss": 0.0, "reward": 2.1843446791172028, "reward_std": 0.005820542646688409, "rewards/format_reward_hoi_key": 0.8457291722297668, "rewards/format_reward_hoi_object_label": 0.6000000089406967, "rewards/format_reward_hoi_verb_label": 0.1180555634200573, "rewards/hoi_iou_reward": 0.6205599009990692, "step": 6 }, { "clip_ratio": 0.0, "completion_length": 164.84375, "epoch": 0.0014880161556039752, "grad_norm": 0.5830075144767761, "kl": 0.00010955333709716797, "learning_rate": 9.985119047619047e-07, "loss": -0.0, "reward": 2.5442887246608734, "reward_std": 0.11149050580570474, "rewards/format_reward_hoi_key": 0.7979166656732559, "rewards/format_reward_hoi_object_label": 0.7083333358168602, "rewards/format_reward_hoi_verb_label": 0.4583333358168602, "rewards/hoi_iou_reward": 0.5797053650021553, "step": 7 }, { "clip_ratio": 0.0, "completion_length": 167.234375, "epoch": 0.0017005898921188286, "grad_norm": 0.35756170749664307, "kl": 8.910894393920898e-05, "learning_rate": 9.982993197278912e-07, "loss": 0.0, "reward": 2.5064347982406616, "reward_std": 0.0026310062530683354, "rewards/format_reward_hoi_key": 0.7702381014823914, "rewards/format_reward_hoi_object_label": 0.595362103311345, "rewards/format_reward_hoi_verb_label": 0.5941220238455571, "rewards/hoi_iou_reward": 0.546712551265955, "step": 8 }, { "clip_ratio": 0.0, "completion_length": 188.484375, "epoch": 0.0019131636286336823, "grad_norm": 1.1468232870101929, "kl": 0.00023877620697021484, "learning_rate": 9.980867346938775e-07, "loss": 0.0, "reward": 2.93448406457901, "reward_std": 0.07516021025367081, "rewards/format_reward_hoi_key": 0.90625, "rewards/format_reward_hoi_object_label": 0.79296875, "rewards/format_reward_hoi_verb_label": 0.447916679084301, "rewards/hoi_iou_reward": 0.7873486280441284, "step": 9 }, { "clip_ratio": 0.0, "completion_length": 188.921875, "epoch": 0.002125737365148536, "grad_norm": 0.608834445476532, "kl": 0.0003757476806640625, "learning_rate": 9.97874149659864e-07, "loss": -0.0, "reward": 2.309541165828705, "reward_std": 0.04332686646375805, "rewards/format_reward_hoi_key": 0.7756249904632568, "rewards/format_reward_hoi_object_label": 0.5166666656732559, "rewards/format_reward_hoi_verb_label": 0.46510415710508823, "rewards/hoi_iou_reward": 0.5521453768014908, "step": 10 }, { "clip_ratio": 0.0, "completion_length": 156.34375, "epoch": 0.0023383111016633895, "grad_norm": 1.079801321029663, "kl": 0.0002917051315307617, "learning_rate": 9.976615646258503e-07, "loss": -0.0, "reward": 2.9021179378032684, "reward_std": 0.06573383091017604, "rewards/format_reward_hoi_key": 0.9125000089406967, "rewards/format_reward_hoi_object_label": 0.75, "rewards/format_reward_hoi_verb_label": 0.5, "rewards/hoi_iou_reward": 0.7396180182695389, "step": 11 }, { "clip_ratio": 0.0, "completion_length": 174.46875, "epoch": 0.002550884838178243, "grad_norm": 0.6156663298606873, "kl": 0.0005776882171630859, "learning_rate": 9.974489795918366e-07, "loss": 0.0, "reward": 2.3791774213314056, "reward_std": 0.0850577435339801, "rewards/format_reward_hoi_key": 0.7312500178813934, "rewards/format_reward_hoi_object_label": 0.5208333358168602, "rewards/format_reward_hoi_verb_label": 0.5911458358168602, "rewards/hoi_iou_reward": 0.535948283970356, "step": 12 }, { "clip_ratio": 0.0, "completion_length": 167.578125, "epoch": 0.002763458574693097, "grad_norm": 3.2466673851013184, "kl": 0.0002658367156982422, "learning_rate": 9.972363945578231e-07, "loss": 0.0, "reward": 3.0418315529823303, "reward_std": 0.013055827002972364, "rewards/format_reward_hoi_key": 0.9000000059604645, "rewards/format_reward_hoi_object_label": 0.8125, "rewards/format_reward_hoi_verb_label": 0.625, "rewards/hoi_iou_reward": 0.704331636428833, "step": 13 }, { "clip_ratio": 0.0, "completion_length": 167.4375, "epoch": 0.0029760323112079505, "grad_norm": 0.5669279098510742, "kl": 0.0004019737243652344, "learning_rate": 9.970238095238094e-07, "loss": 0.0, "reward": 2.4804917573928833, "reward_std": 0.08349880830792245, "rewards/format_reward_hoi_key": 0.7427083253860474, "rewards/format_reward_hoi_object_label": 0.697916679084301, "rewards/format_reward_hoi_verb_label": 0.483333345502615, "rewards/hoi_iou_reward": 0.5565334260463715, "step": 14 }, { "clip_ratio": 0.0, "completion_length": 198.625, "epoch": 0.0031886060477228037, "grad_norm": 0.27379515767097473, "kl": 0.00033855438232421875, "learning_rate": 9.968112244897957e-07, "loss": 0.0, "reward": 2.1904609203338623, "reward_std": 0.06255148959462531, "rewards/format_reward_hoi_key": 0.7820312678813934, "rewards/format_reward_hoi_object_label": 0.6083984375, "rewards/format_reward_hoi_verb_label": 0.3639322891831398, "rewards/hoi_iou_reward": 0.4360988959670067, "step": 15 }, { "clip_ratio": 0.0, "completion_length": 242.234375, "epoch": 0.0034011797842376573, "grad_norm": 0.2515924870967865, "kl": 0.0006353855133056641, "learning_rate": 9.965986394557822e-07, "loss": 0.0, "reward": 2.716467797756195, "reward_std": 0.07810639549279585, "rewards/format_reward_hoi_key": 0.7664583474397659, "rewards/format_reward_hoi_object_label": 0.6187500059604645, "rewards/format_reward_hoi_verb_label": 0.6677083224058151, "rewards/hoi_iou_reward": 0.663551077246666, "step": 16 }, { "clip_ratio": 0.0, "completion_length": 226.859375, "epoch": 0.003613753520752511, "grad_norm": 0.5136963725090027, "kl": 0.0003933906555175781, "learning_rate": 9.963860544217688e-07, "loss": 0.0, "reward": 2.071069449186325, "reward_std": 0.06459418445592746, "rewards/format_reward_hoi_key": 0.6252120807766914, "rewards/format_reward_hoi_object_label": 0.5837053582072258, "rewards/format_reward_hoi_verb_label": 0.4394965320825577, "rewards/hoi_iou_reward": 0.4226554408669472, "step": 17 }, { "clip_ratio": 0.0, "completion_length": 277.625, "epoch": 0.0038263272572673646, "grad_norm": 0.6188202500343323, "kl": 0.0002378225326538086, "learning_rate": 9.96173469387755e-07, "loss": 0.0, "reward": 3.0354496240615845, "reward_std": 0.30482952669262886, "rewards/format_reward_hoi_key": 0.8430059552192688, "rewards/format_reward_hoi_object_label": 0.8227306753396988, "rewards/format_reward_hoi_verb_label": 0.5986328125, "rewards/hoi_iou_reward": 0.7710802108049393, "step": 18 }, { "clip_ratio": 0.0, "completion_length": 151.75, "epoch": 0.004038900993782219, "grad_norm": 0.27549490332603455, "kl": 0.0006313323974609375, "learning_rate": 9.959608843537416e-07, "loss": -0.0, "reward": 2.0183950662612915, "reward_std": 0.015180108457570896, "rewards/format_reward_hoi_key": 0.6604166775941849, "rewards/format_reward_hoi_object_label": 0.5416666716337204, "rewards/format_reward_hoi_verb_label": 0.3524305671453476, "rewards/hoi_iou_reward": 0.4638812467455864, "step": 19 }, { "clip_ratio": 0.0, "completion_length": 224.078125, "epoch": 0.004251474730297072, "grad_norm": 0.56353759765625, "kl": 0.0008664131164550781, "learning_rate": 9.957482993197279e-07, "loss": 0.0, "reward": 2.617310881614685, "reward_std": 0.185114907566458, "rewards/format_reward_hoi_key": 0.7604167088866234, "rewards/format_reward_hoi_object_label": 0.6744791641831398, "rewards/format_reward_hoi_verb_label": 0.5677083283662796, "rewards/hoi_iou_reward": 0.6147066801786423, "step": 20 }, { "clip_ratio": 0.0, "completion_length": 174.09375, "epoch": 0.004464048466811925, "grad_norm": 0.31322988867759705, "kl": 0.0007352828979492188, "learning_rate": 9.955357142857142e-07, "loss": 0.0, "reward": 2.9305796921253204, "reward_std": 0.01013911364134401, "rewards/format_reward_hoi_key": 0.8696428686380386, "rewards/format_reward_hoi_object_label": 0.7857142835855484, "rewards/format_reward_hoi_verb_label": 0.552300363779068, "rewards/hoi_iou_reward": 0.7229221612215042, "step": 21 }, { "clip_ratio": 0.0, "completion_length": 162.0, "epoch": 0.004676622203326779, "grad_norm": 0.4496309161186218, "kl": 0.001049041748046875, "learning_rate": 9.953231292517007e-07, "loss": 0.0, "reward": 2.2096868455410004, "reward_std": 0.0113821976701729, "rewards/format_reward_hoi_key": 0.7333928644657135, "rewards/format_reward_hoi_object_label": 0.6169642880558968, "rewards/format_reward_hoi_verb_label": 0.2777777761220932, "rewards/hoi_iou_reward": 0.5815519690513611, "step": 22 }, { "clip_ratio": 0.0, "completion_length": 154.609375, "epoch": 0.004889195939841632, "grad_norm": 0.8822689652442932, "kl": 0.0013833045959472656, "learning_rate": 9.95110544217687e-07, "loss": 0.0, "reward": 3.247895896434784, "reward_std": 0.04373934442992322, "rewards/format_reward_hoi_key": 0.9250000268220901, "rewards/format_reward_hoi_object_label": 0.9583333283662796, "rewards/format_reward_hoi_verb_label": 0.6562499850988388, "rewards/hoi_iou_reward": 0.708312600851059, "step": 23 }, { "clip_ratio": 0.0, "completion_length": 410.328125, "epoch": 0.005101769676356486, "grad_norm": 0.7035483121871948, "kl": 0.0004572868347167969, "learning_rate": 9.948979591836735e-07, "loss": 0.0, "reward": 2.264761805534363, "reward_std": 0.28715356811881065, "rewards/format_reward_hoi_key": 0.6794504672288895, "rewards/format_reward_hoi_object_label": 0.5326923131942749, "rewards/format_reward_hoi_verb_label": 0.6280448734760284, "rewards/hoi_iou_reward": 0.42457417771220207, "step": 24 }, { "clip_ratio": 0.0, "completion_length": 227.59375, "epoch": 0.0053143434128713396, "grad_norm": 0.31094542145729065, "kl": 0.0009341239929199219, "learning_rate": 9.946853741496598e-07, "loss": 0.0, "reward": 2.356251895427704, "reward_std": 0.003799198704655282, "rewards/format_reward_hoi_key": 0.767708346247673, "rewards/format_reward_hoi_object_label": 0.4895833432674408, "rewards/format_reward_hoi_verb_label": 0.5043560639023781, "rewards/hoi_iou_reward": 0.5946041345596313, "step": 25 }, { "clip_ratio": 0.0, "completion_length": 181.375, "epoch": 0.005526917149386194, "grad_norm": 0.5995525121688843, "kl": 0.00150299072265625, "learning_rate": 9.944727891156463e-07, "loss": 0.0001, "reward": 2.6978970766067505, "reward_std": 0.13544296027976088, "rewards/format_reward_hoi_key": 0.8333333432674408, "rewards/format_reward_hoi_object_label": 0.6158854141831398, "rewards/format_reward_hoi_verb_label": 0.5898437350988388, "rewards/hoi_iou_reward": 0.6588345021009445, "step": 26 }, { "clip_ratio": 0.0, "completion_length": 189.03125, "epoch": 0.005739490885901047, "grad_norm": 0.5540001392364502, "kl": 0.000858306884765625, "learning_rate": 9.942602040816326e-07, "loss": 0.0001, "reward": 3.355882227420807, "reward_std": 0.005877207615412772, "rewards/format_reward_hoi_key": 0.9535714238882065, "rewards/format_reward_hoi_object_label": 0.9017857313156128, "rewards/format_reward_hoi_verb_label": 0.7232142835855484, "rewards/hoi_iou_reward": 0.7773108184337616, "step": 27 }, { "clip_ratio": 0.0, "completion_length": 211.90625, "epoch": 0.005952064622415901, "grad_norm": 2.0975677967071533, "kl": 0.001495361328125, "learning_rate": 9.940476190476191e-07, "loss": 0.0001, "reward": 2.007324628531933, "reward_std": 0.03993106237612665, "rewards/format_reward_hoi_key": 0.5873221457004547, "rewards/format_reward_hoi_object_label": 0.44114159047603607, "rewards/format_reward_hoi_verb_label": 0.5036415904760361, "rewards/hoi_iou_reward": 0.47521928139030933, "step": 28 }, { "clip_ratio": 0.0, "completion_length": 165.1875, "epoch": 0.006164638358930754, "grad_norm": 0.37749311327934265, "kl": 0.0019092559814453125, "learning_rate": 9.938350340136054e-07, "loss": 0.0001, "reward": 2.2582033574581146, "reward_std": 0.08065436300239526, "rewards/format_reward_hoi_key": 0.6932291835546494, "rewards/format_reward_hoi_object_label": 0.59375, "rewards/format_reward_hoi_verb_label": 0.3541666641831398, "rewards/hoi_iou_reward": 0.6170575618743896, "step": 29 }, { "clip_ratio": 0.0, "completion_length": 184.59375, "epoch": 0.006377212095445607, "grad_norm": 0.3330162763595581, "kl": 0.0014448165893554688, "learning_rate": 9.936224489795917e-07, "loss": 0.0, "reward": 2.6335054636001587, "reward_std": 0.0012341497422312386, "rewards/format_reward_hoi_key": 0.8750000149011612, "rewards/format_reward_hoi_object_label": 0.6875, "rewards/format_reward_hoi_verb_label": 0.3880208358168602, "rewards/hoi_iou_reward": 0.6829846650362015, "step": 30 }, { "clip_ratio": 0.0, "completion_length": 172.125, "epoch": 0.006589785831960461, "grad_norm": 0.8369670510292053, "kl": 0.0013284683227539062, "learning_rate": 9.934098639455782e-07, "loss": 0.0001, "reward": 2.4850784838199615, "reward_std": 0.02788396377582103, "rewards/format_reward_hoi_key": 0.8687500208616257, "rewards/format_reward_hoi_object_label": 0.4687500074505806, "rewards/format_reward_hoi_verb_label": 0.5, "rewards/hoi_iou_reward": 0.6475784331560135, "step": 31 }, { "clip_ratio": 0.0, "completion_length": 216.046875, "epoch": 0.006802359568475315, "grad_norm": 0.9224941730499268, "kl": 0.00140380859375, "learning_rate": 9.931972789115645e-07, "loss": 0.0, "reward": 2.751905083656311, "reward_std": 0.08277821098454297, "rewards/format_reward_hoi_key": 0.809895858168602, "rewards/format_reward_hoi_object_label": 0.5078125, "rewards/format_reward_hoi_verb_label": 0.6927083283662796, "rewards/hoi_iou_reward": 0.7414884492754936, "step": 32 }, { "clip_ratio": 0.0, "completion_length": 210.71875, "epoch": 0.007014933304990169, "grad_norm": 0.39392945170402527, "kl": 0.002300262451171875, "learning_rate": 9.92984693877551e-07, "loss": 0.0001, "reward": 2.1440170407295227, "reward_std": 0.02253561234101653, "rewards/format_reward_hoi_key": 0.9121875166893005, "rewards/format_reward_hoi_object_label": 0.3125, "rewards/format_reward_hoi_verb_label": 0.3333333358168602, "rewards/hoi_iou_reward": 0.5859961807727814, "step": 33 }, { "clip_ratio": 0.0, "completion_length": 194.265625, "epoch": 0.007227507041505022, "grad_norm": 0.5018682479858398, "kl": 0.0017986297607421875, "learning_rate": 9.927721088435373e-07, "loss": 0.0001, "reward": 2.5592292845249176, "reward_std": 0.00986732606543228, "rewards/format_reward_hoi_key": 0.7691666930913925, "rewards/format_reward_hoi_object_label": 0.6583333313465118, "rewards/format_reward_hoi_verb_label": 0.5562499985098839, "rewards/hoi_iou_reward": 0.5754793435335159, "step": 34 }, { "clip_ratio": 0.0, "completion_length": 205.03125, "epoch": 0.007440080778019876, "grad_norm": 0.6149845719337463, "kl": 0.0016880035400390625, "learning_rate": 9.925595238095238e-07, "loss": 0.0001, "reward": 2.778216004371643, "reward_std": 0.11917518911650404, "rewards/format_reward_hoi_key": 0.8614583313465118, "rewards/format_reward_hoi_object_label": 0.7333333194255829, "rewards/format_reward_hoi_verb_label": 0.5011574029922485, "rewards/hoi_iou_reward": 0.6822669506072998, "step": 35 }, { "clip_ratio": 0.0, "completion_length": 222.671875, "epoch": 0.007652654514534729, "grad_norm": 0.5013077259063721, "kl": 0.0017242431640625, "learning_rate": 9.923469387755101e-07, "loss": 0.0001, "reward": 2.724997416138649, "reward_std": 0.007125564094167203, "rewards/format_reward_hoi_key": 0.8181547522544861, "rewards/format_reward_hoi_object_label": 0.6875, "rewards/format_reward_hoi_verb_label": 0.625, "rewards/hoi_iou_reward": 0.5943426117300987, "step": 36 }, { "clip_ratio": 0.0, "completion_length": 249.65625, "epoch": 0.007865228251049582, "grad_norm": 0.4427626430988312, "kl": 0.00135040283203125, "learning_rate": 9.921343537414967e-07, "loss": 0.0001, "reward": 2.5683979988098145, "reward_std": 0.05630575024406426, "rewards/format_reward_hoi_key": 0.810416653752327, "rewards/format_reward_hoi_object_label": 0.625, "rewards/format_reward_hoi_verb_label": 0.3906250111758709, "rewards/hoi_iou_reward": 0.7423563152551651, "step": 37 }, { "clip_ratio": 0.0, "completion_length": 270.921875, "epoch": 0.008077801987564437, "grad_norm": 2.8067455291748047, "kl": 0.0019435882568359375, "learning_rate": 9.91921768707483e-07, "loss": 0.0001, "reward": 2.22263365983963, "reward_std": 0.20146464882418513, "rewards/format_reward_hoi_key": 0.6945772171020508, "rewards/format_reward_hoi_object_label": 0.3977022171020508, "rewards/format_reward_hoi_verb_label": 0.5460824370384216, "rewards/hoi_iou_reward": 0.5842718333005905, "step": 38 }, { "clip_ratio": 0.0, "completion_length": 211.40625, "epoch": 0.00829037572407929, "grad_norm": 0.5841067433357239, "kl": 0.00464630126953125, "learning_rate": 9.917091836734693e-07, "loss": 0.0002, "reward": 2.7285755276679993, "reward_std": 0.15019595221383497, "rewards/format_reward_hoi_key": 0.931383952498436, "rewards/format_reward_hoi_object_label": 0.5188244059681892, "rewards/format_reward_hoi_verb_label": 0.5774181559681892, "rewards/hoi_iou_reward": 0.7009490430355072, "step": 39 }, { "clip_ratio": 0.0, "completion_length": 235.15625, "epoch": 0.008502949460594144, "grad_norm": 0.3903954327106476, "kl": 0.0014748573303222656, "learning_rate": 9.914965986394558e-07, "loss": 0.0001, "reward": 2.41436231136322, "reward_std": 0.031614198378520086, "rewards/format_reward_hoi_key": 0.7393315136432648, "rewards/format_reward_hoi_object_label": 0.6360462605953217, "rewards/format_reward_hoi_verb_label": 0.5051649361848831, "rewards/hoi_iou_reward": 0.5338196456432343, "step": 40 } ], "logging_steps": 1.0, "max_steps": 4704, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 20, "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": 8, "trial_name": null, "trial_params": null }