{"loss/policy": -1.4901161193847656e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28377822041511536, "ppo/advantage_mean": 1.4901161193847656e-08, "loss/infonce": 1.2804733514785767, "step": 0, "loss": 0.3841419816017151, "reward/raw_mean": 0.4296875, "reward/gold_mean": 0.6875, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2960161566734314, "ppo/advantage_mean": 0.0, "loss/infonce": 1.3021833896636963, "step": 1, "loss": 0.3906550407409668, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -5.587935447692871e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30022379755973816, "ppo/advantage_mean": 5.587935447692871e-09, "loss/infonce": 1.5337605476379395, "step": 2, "loss": 0.46012818813323975, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.25} {"loss/policy": 1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28784847259521484, "ppo/advantage_mean": -1.862645149230957e-09, "loss/infonce": 0.8407899141311646, "step": 3, "loss": 0.2522369921207428, "reward/raw_mean": 0.1953125, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29157984256744385, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 1.194321870803833, "step": 4, "loss": 0.35829657316207886, "reward/raw_mean": 0.171875, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.25} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29695093631744385, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 1.189415454864502, "step": 5, "loss": 0.3568246364593506, "reward/raw_mean": 0.0390625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": 5.587935447692871e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28624576330184937, "ppo/advantage_mean": -5.587935447692871e-09, "loss/infonce": 1.0234150886535645, "step": 6, "loss": 0.3070245385169983, "reward/raw_mean": 0.375, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2834683060646057, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 0.9578173160552979, "step": 7, "loss": 0.28734520077705383, "reward/raw_mean": 0.2421875, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2963838577270508, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.9631607532501221, "step": 8, "loss": 0.2889482378959656, "reward/raw_mean": 0.0703125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29991787672042847, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.0154544115066528, "step": 9, "loss": 0.3046363294124603, "reward/raw_mean": 0.0078125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -1.4901161193847656e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.299377977848053, "ppo/advantage_mean": 1.4901161193847656e-08, "loss/infonce": 0.8297242522239685, "step": 10, "loss": 0.24891726672649384, "reward/raw_mean": 0.09375, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 1.0} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2945602834224701, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 1.110363483428955, "step": 11, "loss": 0.333109050989151, "reward/raw_mean": 0.078125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29909276962280273, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.9089183807373047, "step": 12, "loss": 0.2726755142211914, "reward/raw_mean": 0.0859375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.27861288189888, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.1297180652618408, "step": 13, "loss": 0.3389154374599457, "reward/raw_mean": 0.375, "reward/gold_mean": 0.6875, "retrieval/gold_in_topN": 0.5} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29326769709587097, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.40665268898010254, "step": 14, "loss": 0.12199580669403076, "reward/raw_mean": 0.1953125, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2854386568069458, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 2.077425956726074, "step": 15, "loss": 0.6232278347015381, "reward/raw_mean": 0.0078125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2912495732307434, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.9203202724456787, "step": 16, "loss": 0.2760961055755615, "reward/raw_mean": 0.2734375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2884179353713989, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 1.063979148864746, "step": 17, "loss": 0.3191937506198883, "reward/raw_mean": 0.1953125, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3038754463195801, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.5206839442253113, "step": 18, "loss": 0.1562051922082901, "reward/raw_mean": 0.3203125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 1.0} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2878909111022949, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 1.1057264804840088, "step": 19, "loss": 0.33171796798706055, "reward/raw_mean": 0.375, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28547605872154236, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 1.0917547941207886, "step": 20, "loss": 0.3275264501571655, "reward/raw_mean": 0.203125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3055182695388794, "ppo/advantage_mean": 0.0, "loss/infonce": 0.7741667628288269, "step": 21, "loss": 0.23225003480911255, "reward/raw_mean": 0.0625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29749149084091187, "ppo/advantage_mean": 0.0, "loss/infonce": 0.7596724629402161, "step": 22, "loss": 0.22790174186229706, "reward/raw_mean": 0.4375, "reward/gold_mean": 0.6875, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29362237453460693, "ppo/advantage_mean": 0.0, "loss/infonce": 1.7154160737991333, "step": 23, "loss": 0.514624834060669, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3045852780342102, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.6228291988372803, "step": 24, "loss": 0.18684875965118408, "reward/raw_mean": 0.0234375, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2928597331047058, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8470190167427063, "step": 25, "loss": 0.25410571694374084, "reward/raw_mean": 0.2265625, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28471559286117554, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.8555948734283447, "step": 26, "loss": 0.2566784620285034, "reward/raw_mean": 0.375, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2982333302497864, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.2568897008895874, "step": 27, "loss": 0.3770669400691986, "reward/raw_mean": 0.0234375, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.5} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29315808415412903, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.7110286951065063, "step": 28, "loss": 0.21330861747264862, "reward/raw_mean": 0.3359375, "reward/gold_mean": 0.625, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2916255295276642, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.1437945365905762, "step": 29, "loss": 0.34313836693763733, "reward/raw_mean": 0.171875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -9.313225746154785e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2936877906322479, "ppo/advantage_mean": 9.313225746154785e-09, "loss/infonce": 0.6056746244430542, "step": 30, "loss": 0.1817023754119873, "reward/raw_mean": 0.1171875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2939503490924835, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8093839287757874, "step": 31, "loss": 0.24281518161296844, "reward/raw_mean": 0.0078125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2923243045806885, "ppo/advantage_mean": 0.0, "loss/infonce": 1.034019947052002, "step": 32, "loss": 0.31020599603652954, "reward/raw_mean": 0.1796875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28896835446357727, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.9592291116714478, "step": 33, "loss": 0.28776875138282776, "reward/raw_mean": 0.15625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2977867126464844, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 0.6011627316474915, "step": 34, "loss": 0.18034884333610535, "reward/raw_mean": 0.140625, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": 1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31025731563568115, "ppo/advantage_mean": -1.862645149230957e-09, "loss/infonce": 0.4474892318248749, "step": 35, "loss": 0.13424678146839142, "reward/raw_mean": 0.1171875, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3020349144935608, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 1.0801982879638672, "step": 36, "loss": 0.32405948638916016, "reward/raw_mean": 0.21875, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3012961745262146, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.692383885383606, "step": 37, "loss": 0.20771516859531403, "reward/raw_mean": 0.1328125, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.75} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3129095137119293, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 0.5382456183433533, "step": 38, "loss": 0.16147367656230927, "reward/raw_mean": 0.09375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29801174998283386, "ppo/advantage_mean": 0.0, "loss/infonce": 1.6049963235855103, "step": 39, "loss": 0.48149892687797546, "reward/raw_mean": 0.0625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29712527990341187, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.0471761226654053, "step": 40, "loss": 0.3141528367996216, "reward/raw_mean": 0.28125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30611830949783325, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.6461142301559448, "step": 41, "loss": 0.19383427500724792, "reward/raw_mean": 0.296875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3020230829715729, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.9955165386199951, "step": 42, "loss": 0.2986549735069275, "reward/raw_mean": 0.109375, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3029899001121521, "ppo/advantage_mean": 0.0, "loss/infonce": 1.2207140922546387, "step": 43, "loss": 0.36621424555778503, "reward/raw_mean": 0.1875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3035963475704193, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.7558916807174683, "step": 44, "loss": 0.22676751017570496, "reward/raw_mean": 0.1875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3069925308227539, "ppo/advantage_mean": 0.0, "loss/infonce": 0.7986709475517273, "step": 45, "loss": 0.23960129916667938, "reward/raw_mean": 0.2265625, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.25} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29973262548446655, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 1.1247867345809937, "step": 46, "loss": 0.3374360203742981, "reward/raw_mean": 0.1015625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28782153129577637, "ppo/advantage_mean": 0.0, "loss/infonce": 1.0250768661499023, "step": 47, "loss": 0.30752307176589966, "reward/raw_mean": 0.25, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": -5.587935447692871e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2940424680709839, "ppo/advantage_mean": 5.587935447692871e-09, "loss/infonce": 1.1722584962844849, "step": 48, "loss": 0.3516775667667389, "reward/raw_mean": 0.171875, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28927046060562134, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.4632987976074219, "step": 49, "loss": 0.1389896273612976, "reward/raw_mean": 0.015625, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2969142198562622, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.9659773111343384, "step": 50, "loss": 0.2897931933403015, "reward/raw_mean": 0.09375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2968151867389679, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9152168035507202, "step": 51, "loss": 0.27456504106521606, "reward/raw_mean": 0.25, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31002557277679443, "ppo/advantage_mean": 0.0, "loss/infonce": 1.5634427070617676, "step": 52, "loss": 0.46903282403945923, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": 1.6763806343078613e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31526148319244385, "ppo/advantage_mean": -1.6763806343078613e-08, "loss/infonce": 1.0696029663085938, "step": 53, "loss": 0.3208809196949005, "reward/raw_mean": 0.1328125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2988527715206146, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.2557711601257324, "step": 54, "loss": 0.37673136591911316, "reward/raw_mean": 0.0703125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3046497702598572, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 1.1722512245178223, "step": 55, "loss": 0.3516753911972046, "reward/raw_mean": 0.1875, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2939862310886383, "ppo/advantage_mean": 0.0, "loss/infonce": 1.0866916179656982, "step": 56, "loss": 0.3260074853897095, "reward/raw_mean": 0.0625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29172053933143616, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.5822263956069946, "step": 57, "loss": 0.17466792464256287, "reward/raw_mean": 0.1875, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29788485169410706, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.5799582004547119, "step": 58, "loss": 0.1739874631166458, "reward/raw_mean": 0.359375, "reward/gold_mean": 0.625, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2942391633987427, "ppo/advantage_mean": 0.0, "loss/infonce": 1.2709952592849731, "step": 59, "loss": 0.38129860162734985, "reward/raw_mean": 0.4375, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.75} {"loss/policy": 1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3109416365623474, "ppo/advantage_mean": -1.862645149230957e-09, "loss/infonce": 1.3089568614959717, "step": 60, "loss": 0.3926870822906494, "reward/raw_mean": 0.0546875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30917051434516907, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 1.1768935918807983, "step": 61, "loss": 0.353068083524704, "reward/raw_mean": 0.21875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3059283494949341, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8507545590400696, "step": 62, "loss": 0.25522637367248535, "reward/raw_mean": 0.2421875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3017176687717438, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.6843401789665222, "step": 63, "loss": 0.20530205965042114, "reward/raw_mean": 0.3203125, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30538368225097656, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.4258801937103271, "step": 64, "loss": 0.42776408791542053, "reward/raw_mean": 0.3046875, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2889587879180908, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.6678872108459473, "step": 65, "loss": 0.20036616921424866, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -5.587935447692871e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.302694171667099, "ppo/advantage_mean": 5.587935447692871e-09, "loss/infonce": 1.1047286987304688, "step": 66, "loss": 0.33141863346099854, "reward/raw_mean": 0.125, "reward/gold_mean": 0.6875, "retrieval/gold_in_topN": 1.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2958329916000366, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.9291835427284241, "step": 67, "loss": 0.2787550687789917, "reward/raw_mean": 0.1796875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28293377161026, "ppo/advantage_mean": 0.0, "loss/infonce": 0.6445352435112, "step": 68, "loss": 0.1933605819940567, "reward/raw_mean": 0.0625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2963000535964966, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 1.555242657661438, "step": 69, "loss": 0.4665728211402893, "reward/raw_mean": 0.0625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3056275248527527, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 0.9753903150558472, "step": 70, "loss": 0.2926171123981476, "reward/raw_mean": 0.1796875, "reward/gold_mean": 0.625, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2982618808746338, "ppo/advantage_mean": 0.0, "loss/infonce": 0.644096851348877, "step": 71, "loss": 0.1932290643453598, "reward/raw_mean": 0.5625, "reward/gold_mean": 0.5625, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3004736304283142, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8557281494140625, "step": 72, "loss": 0.2567184567451477, "reward/raw_mean": 0.1328125, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.75} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31117355823516846, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 1.183401107788086, "step": 73, "loss": 0.35502034425735474, "reward/raw_mean": 0.1171875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29853716492652893, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.40232622623443604, "step": 74, "loss": 0.12069786339998245, "reward/raw_mean": 0.0390625, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2936002016067505, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 1.371297001838684, "step": 75, "loss": 0.4113891124725342, "reward/raw_mean": 0.078125, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30853530764579773, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8923385143280029, "step": 76, "loss": 0.26770156621932983, "reward/raw_mean": 0.125, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30014634132385254, "ppo/advantage_mean": 0.0, "loss/infonce": 0.6062502264976501, "step": 77, "loss": 0.181875079870224, "reward/raw_mean": 0.0859375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2867533564567566, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.2375799417495728, "step": 78, "loss": 0.3712739944458008, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.287644624710083, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.1448618173599243, "step": 79, "loss": 0.3434585630893707, "reward/raw_mean": 0.4921875, "reward/gold_mean": 0.75, "retrieval/gold_in_topN": 1.0} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29129624366760254, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 1.2753970623016357, "step": 80, "loss": 0.38261914253234863, "reward/raw_mean": 0.0390625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": 1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31388217210769653, "ppo/advantage_mean": -1.862645149230957e-09, "loss/infonce": 1.2731232643127441, "step": 81, "loss": 0.3819369971752167, "reward/raw_mean": 0.3828125, "reward/gold_mean": 0.6875, "retrieval/gold_in_topN": 1.0} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29518598318099976, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 1.1243858337402344, "step": 82, "loss": 0.33731576800346375, "reward/raw_mean": 0.1171875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2906361222267151, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.1181989908218384, "step": 83, "loss": 0.33545970916748047, "reward/raw_mean": 0.0390625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -5.587935447692871e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29591354727745056, "ppo/advantage_mean": 5.587935447692871e-09, "loss/infonce": 0.488870769739151, "step": 84, "loss": 0.14666123688220978, "reward/raw_mean": 0.0546875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -2.7939677238464355e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29692602157592773, "ppo/advantage_mean": 2.7939677238464355e-09, "loss/infonce": 1.1132899522781372, "step": 85, "loss": 0.3339869976043701, "reward/raw_mean": 0.1796875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2868953347206116, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.3279916048049927, "step": 86, "loss": 0.3983975052833557, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28691840171813965, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.8623455762863159, "step": 87, "loss": 0.25870367884635925, "reward/raw_mean": 0.1171875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2979664206504822, "ppo/advantage_mean": 0.0, "loss/infonce": 1.1582379341125488, "step": 88, "loss": 0.3474713861942291, "reward/raw_mean": 0.3125, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31436699628829956, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 1.2001008987426758, "step": 89, "loss": 0.36003029346466064, "reward/raw_mean": 0.078125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29970914125442505, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.8763506412506104, "step": 90, "loss": 0.26290521025657654, "reward/raw_mean": 0.4375, "reward/gold_mean": 0.5625, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3033410906791687, "ppo/advantage_mean": 0.0, "loss/infonce": 1.092583179473877, "step": 91, "loss": 0.3277749717235565, "reward/raw_mean": 0.25, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2886977791786194, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.5961170196533203, "step": 92, "loss": 0.4788351356983185, "reward/raw_mean": 0.09375, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2828144431114197, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.982757568359375, "step": 93, "loss": 0.29482728242874146, "reward/raw_mean": 0.3984375, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3054089844226837, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.6551581621170044, "step": 94, "loss": 0.1965474635362625, "reward/raw_mean": 0.078125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29749003052711487, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 1.8805361986160278, "step": 95, "loss": 0.5641608834266663, "reward/raw_mean": 0.0546875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2990729808807373, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.33973705768585205, "step": 96, "loss": 0.10192111879587173, "reward/raw_mean": 0.15625, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.309353768825531, "ppo/advantage_mean": 0.0, "loss/infonce": 1.3590779304504395, "step": 97, "loss": 0.40772339701652527, "reward/raw_mean": 0.390625, "reward/gold_mean": 0.625, "retrieval/gold_in_topN": 0.5} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29804760217666626, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 1.0972014665603638, "step": 98, "loss": 0.3291604518890381, "reward/raw_mean": 0.046875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": 5.587935447692871e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2983657717704773, "ppo/advantage_mean": -5.587935447692871e-09, "loss/infonce": 0.573727548122406, "step": 99, "loss": 0.17211827635765076, "reward/raw_mean": 0.140625, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 1.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2904968559741974, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.9996287822723389, "step": 100, "loss": 0.29988864064216614, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -9.313225746154785e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.292494535446167, "ppo/advantage_mean": 9.313225746154785e-09, "loss/infonce": 0.8377481698989868, "step": 101, "loss": 0.25132447481155396, "reward/raw_mean": 0.1171875, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2885867655277252, "ppo/advantage_mean": 0.0, "loss/infonce": 1.1172924041748047, "step": 102, "loss": 0.33518773317337036, "reward/raw_mean": 0.25, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.25} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2975724935531616, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.5042436122894287, "step": 103, "loss": 0.15127308666706085, "reward/raw_mean": 0.234375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": -5.587935447692871e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2953782379627228, "ppo/advantage_mean": 5.587935447692871e-09, "loss/infonce": 1.0054278373718262, "step": 104, "loss": 0.30162835121154785, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.5} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3040950894355774, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 1.0787874460220337, "step": 105, "loss": 0.3236362338066101, "reward/raw_mean": 0.1015625, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.75} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2960994839668274, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 0.9714431762695312, "step": 106, "loss": 0.2914329767227173, "reward/raw_mean": 0.3046875, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.298149049282074, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.6305728554725647, "step": 107, "loss": 0.18917185068130493, "reward/raw_mean": 0.203125, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 1.0} {"loss/policy": 1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29616114497184753, "ppo/advantage_mean": -1.862645149230957e-09, "loss/infonce": 1.0093042850494385, "step": 108, "loss": 0.3027912974357605, "reward/raw_mean": 0.203125, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.75} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30815988779067993, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 1.0795462131500244, "step": 109, "loss": 0.3238638639450073, "reward/raw_mean": 0.34375, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3051372170448303, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.043616533279419, "step": 110, "loss": 0.3130849599838257, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 1.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28520190715789795, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.2688426971435547, "step": 111, "loss": 0.3806528151035309, "reward/raw_mean": 0.0234375, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2987751364707947, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.9450063109397888, "step": 112, "loss": 0.28350189328193665, "reward/raw_mean": 0.296875, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2876006066799164, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.7678290009498596, "step": 113, "loss": 0.23034870624542236, "reward/raw_mean": 0.1875, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.299569308757782, "ppo/advantage_mean": 0.0, "loss/infonce": 0.92820143699646, "step": 114, "loss": 0.27846044301986694, "reward/raw_mean": 0.1015625, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30355507135391235, "ppo/advantage_mean": 0.0, "loss/infonce": 0.5220562219619751, "step": 115, "loss": 0.15661686658859253, "reward/raw_mean": 0.296875, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29189908504486084, "ppo/advantage_mean": 0.0, "loss/infonce": 0.7255768179893494, "step": 116, "loss": 0.21767304837703705, "reward/raw_mean": 0.09375, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 1.0} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2973690629005432, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 1.2260003089904785, "step": 117, "loss": 0.36780011653900146, "reward/raw_mean": 0.4140625, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3119138777256012, "ppo/advantage_mean": 0.0, "loss/infonce": 1.1674306392669678, "step": 118, "loss": 0.3502292037010193, "reward/raw_mean": 0.078125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2955324947834015, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.6158958673477173, "step": 119, "loss": 0.18476876616477966, "reward/raw_mean": 0.5546875, "reward/gold_mean": 0.9375, "retrieval/gold_in_topN": 1.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28881198167800903, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8755487203598022, "step": 120, "loss": 0.2626646161079407, "reward/raw_mean": 0.171875, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3015792965888977, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 0.868010401725769, "step": 121, "loss": 0.2604031264781952, "reward/raw_mean": 0.15625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": 1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.320588082075119, "ppo/advantage_mean": -1.862645149230957e-09, "loss/infonce": 0.9000284671783447, "step": 122, "loss": 0.27000856399536133, "reward/raw_mean": 0.1328125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2916778028011322, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.067695140838623, "step": 123, "loss": 0.3203085660934448, "reward/raw_mean": 0.078125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29571324586868286, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.0525281429290771, "step": 124, "loss": 0.31575846672058105, "reward/raw_mean": 0.046875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2922792136669159, "ppo/advantage_mean": 0.0, "loss/infonce": 0.8899449706077576, "step": 125, "loss": 0.2669835090637207, "reward/raw_mean": 0.1953125, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.75} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2902148365974426, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 0.8820682764053345, "step": 126, "loss": 0.26462048292160034, "reward/raw_mean": 0.1640625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30022817850112915, "ppo/advantage_mean": 0.0, "loss/infonce": 1.222020149230957, "step": 127, "loss": 0.36660605669021606, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.0} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3148430585861206, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 0.8019125461578369, "step": 128, "loss": 0.24057379364967346, "reward/raw_mean": 0.5234375, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30429917573928833, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 0.9892771244049072, "step": 129, "loss": 0.2967831492424011, "reward/raw_mean": 0.2734375, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29582732915878296, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 1.0920305252075195, "step": 130, "loss": 0.32760918140411377, "reward/raw_mean": 0.2265625, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29013165831565857, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.464030921459198, "step": 131, "loss": 0.13920928537845612, "reward/raw_mean": 0.2109375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 1.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28905123472213745, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.9151110649108887, "step": 132, "loss": 0.27453333139419556, "reward/raw_mean": 0.0078125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3033716678619385, "ppo/advantage_mean": 0.0, "loss/infonce": 1.0338654518127441, "step": 133, "loss": 0.3101596534252167, "reward/raw_mean": 0.15625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2970007061958313, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.9695098996162415, "step": 134, "loss": 0.29085299372673035, "reward/raw_mean": 0.421875, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29051175713539124, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 0.5134996771812439, "step": 135, "loss": 0.15404990315437317, "reward/raw_mean": 0.3046875, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2889440953731537, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.11637001484632492, "step": 136, "loss": 0.034911010414361954, "reward/raw_mean": 0.546875, "reward/gold_mean": 0.75, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3092021942138672, "ppo/advantage_mean": 0.0, "loss/infonce": 1.0128648281097412, "step": 137, "loss": 0.3038594722747803, "reward/raw_mean": 0.0703125, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3065745234489441, "ppo/advantage_mean": 0.0, "loss/infonce": 1.1030097007751465, "step": 138, "loss": 0.33090293407440186, "reward/raw_mean": 0.28125, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.5} {"loss/policy": 5.587935447692871e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.296448677778244, "ppo/advantage_mean": -5.587935447692871e-09, "loss/infonce": 1.2568113803863525, "step": 139, "loss": 0.3770434260368347, "reward/raw_mean": 0.234375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3056856393814087, "ppo/advantage_mean": 0.0, "loss/infonce": 0.7309000492095947, "step": 140, "loss": 0.2192700207233429, "reward/raw_mean": 0.21875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30359333753585815, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.5571833252906799, "step": 141, "loss": 0.16715499758720398, "reward/raw_mean": 0.2578125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29662948846817017, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 0.8569775223731995, "step": 142, "loss": 0.25709328055381775, "reward/raw_mean": 0.25, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3100774586200714, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.8958662748336792, "step": 143, "loss": 0.26875990629196167, "reward/raw_mean": 0.171875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2976346015930176, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9712724685668945, "step": 144, "loss": 0.29138174653053284, "reward/raw_mean": 0.140625, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29479050636291504, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.3418309688568115, "step": 145, "loss": 0.40254929661750793, "reward/raw_mean": 0.0078125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3157541751861572, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9511723518371582, "step": 146, "loss": 0.2853517234325409, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3046107292175293, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.103074312210083, "step": 147, "loss": 0.33092230558395386, "reward/raw_mean": 0.2890625, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3012419044971466, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.7944234609603882, "step": 148, "loss": 0.2383270412683487, "reward/raw_mean": 0.3359375, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2944275438785553, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.1397631168365479, "step": 149, "loss": 0.34192895889282227, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.300972044467926, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.6659022569656372, "step": 150, "loss": 0.19977068901062012, "reward/raw_mean": 0.0703125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3031638562679291, "ppo/advantage_mean": 0.0, "loss/infonce": 0.921462893486023, "step": 151, "loss": 0.2764388918876648, "reward/raw_mean": 0.3125, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3043043613433838, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.8194708824157715, "step": 152, "loss": 0.24584126472473145, "reward/raw_mean": 0.109375, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29999712109565735, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9297597408294678, "step": 153, "loss": 0.27892792224884033, "reward/raw_mean": 0.265625, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2934001386165619, "ppo/advantage_mean": 0.0, "loss/infonce": 0.6924792528152466, "step": 154, "loss": 0.2077437788248062, "reward/raw_mean": 0.078125, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30432504415512085, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.7928175330162048, "step": 155, "loss": 0.2378452718257904, "reward/raw_mean": 0.171875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2876494526863098, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.4752767086029053, "step": 156, "loss": 0.14258301258087158, "reward/raw_mean": 0.4296875, "reward/gold_mean": 0.625, "retrieval/gold_in_topN": 0.75} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2817780673503876, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 0.7557218670845032, "step": 157, "loss": 0.22671657800674438, "reward/raw_mean": 0.296875, "reward/gold_mean": 0.75, "retrieval/gold_in_topN": 1.0} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2848365902900696, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.43372225761413574, "step": 158, "loss": 0.13011668622493744, "reward/raw_mean": 0.0234375, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3123375177383423, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.0091400146484375, "step": 159, "loss": 0.30274200439453125, "reward/raw_mean": 0.234375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.294228732585907, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9130889773368835, "step": 160, "loss": 0.273926705121994, "reward/raw_mean": 0.0859375, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29864701628685, "ppo/advantage_mean": 0.0, "loss/infonce": 0.8597190380096436, "step": 161, "loss": 0.257915735244751, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29380035400390625, "ppo/advantage_mean": 0.0, "loss/infonce": 1.0665607452392578, "step": 162, "loss": 0.31996822357177734, "reward/raw_mean": 0.1171875, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.25} {"loss/policy": -5.587935447692871e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2940565347671509, "ppo/advantage_mean": 5.587935447692871e-09, "loss/infonce": 0.45891073346138, "step": 163, "loss": 0.13767322897911072, "reward/raw_mean": 0.3046875, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": 9.313225746154785e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3039538562297821, "ppo/advantage_mean": -9.313225746154785e-09, "loss/infonce": 1.140573263168335, "step": 164, "loss": 0.3421719968318939, "reward/raw_mean": 0.109375, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3073284924030304, "ppo/advantage_mean": 0.0, "loss/infonce": 1.1770529747009277, "step": 165, "loss": 0.35311591625213623, "reward/raw_mean": 0.109375, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30704864859580994, "ppo/advantage_mean": 0.0, "loss/infonce": 0.7998511791229248, "step": 166, "loss": 0.2399553656578064, "reward/raw_mean": 0.0625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30019521713256836, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.7803844213485718, "step": 167, "loss": 0.234115332365036, "reward/raw_mean": 0.125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30158141255378723, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.9070427417755127, "step": 168, "loss": 0.2721128463745117, "reward/raw_mean": 0.0703125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29076409339904785, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.9019904732704163, "step": 169, "loss": 0.2705971598625183, "reward/raw_mean": 0.046875, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.5} {"loss/policy": -1.4901161193847656e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29365262389183044, "ppo/advantage_mean": 1.4901161193847656e-08, "loss/infonce": 0.6546698808670044, "step": 170, "loss": 0.1964009553194046, "reward/raw_mean": 0.21875, "reward/gold_mean": 0.625, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2921338379383087, "ppo/advantage_mean": 0.0, "loss/infonce": 0.7149900794029236, "step": 171, "loss": 0.21449702978134155, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3078240156173706, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.8433718085289001, "step": 172, "loss": 0.253011554479599, "reward/raw_mean": 0.3046875, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.75} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2972494959831238, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 0.7911451458930969, "step": 173, "loss": 0.23734353482723236, "reward/raw_mean": 0.1171875, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": -1.4901161193847656e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2961505651473999, "ppo/advantage_mean": 1.4901161193847656e-08, "loss/infonce": 1.1002089977264404, "step": 174, "loss": 0.3300626873970032, "reward/raw_mean": 0.1953125, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3032396137714386, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.5239497423171997, "step": 175, "loss": 0.1571849286556244, "reward/raw_mean": 0.1953125, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2963773012161255, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8084653615951538, "step": 176, "loss": 0.24253961443901062, "reward/raw_mean": 0.0234375, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2987214922904968, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.0165023803710938, "step": 177, "loss": 0.3049507141113281, "reward/raw_mean": 0.21875, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3032291829586029, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.7848706841468811, "step": 178, "loss": 0.23546122014522552, "reward/raw_mean": 0.4609375, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29532140493392944, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.6097485423088074, "step": 179, "loss": 0.1829245686531067, "reward/raw_mean": 0.1328125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.308763712644577, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9499924778938293, "step": 180, "loss": 0.28499776124954224, "reward/raw_mean": 0.0859375, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.75} {"loss/policy": -1.4901161193847656e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30349016189575195, "ppo/advantage_mean": 1.4901161193847656e-08, "loss/infonce": 0.8806874752044678, "step": 181, "loss": 0.2642062306404114, "reward/raw_mean": 0.2109375, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29955625534057617, "ppo/advantage_mean": 0.0, "loss/infonce": 1.6235631704330444, "step": 182, "loss": 0.4870689809322357, "reward/raw_mean": 0.078125, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2998431921005249, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.4202280044555664, "step": 183, "loss": 0.42606842517852783, "reward/raw_mean": 0.3046875, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2975272834300995, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8772879242897034, "step": 184, "loss": 0.26318639516830444, "reward/raw_mean": 0.1328125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2973927855491638, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.7500694990158081, "step": 185, "loss": 0.2250208556652069, "reward/raw_mean": 0.1796875, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29593825340270996, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.27553391456604, "step": 186, "loss": 0.3826601803302765, "reward/raw_mean": 0.0859375, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.25} {"loss/policy": -9.313225746154785e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3007546067237854, "ppo/advantage_mean": 9.313225746154785e-09, "loss/infonce": 1.0639344453811646, "step": 187, "loss": 0.31918033957481384, "reward/raw_mean": 0.1796875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3040534555912018, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.2703824043273926, "step": 188, "loss": 0.38111475110054016, "reward/raw_mean": 0.0234375, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2912273406982422, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.6159539222717285, "step": 189, "loss": 0.18478618562221527, "reward/raw_mean": 0.0703125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3043210506439209, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.3321776390075684, "step": 190, "loss": 0.3996533155441284, "reward/raw_mean": 0.0078125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30053550004959106, "ppo/advantage_mean": 0.0, "loss/infonce": 0.903349757194519, "step": 191, "loss": 0.27100494503974915, "reward/raw_mean": 0.125, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.5} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2927333414554596, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.8726470470428467, "step": 192, "loss": 0.2617941200733185, "reward/raw_mean": 0.2890625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.294335275888443, "ppo/advantage_mean": 0.0, "loss/infonce": 0.8108165264129639, "step": 193, "loss": 0.2432449609041214, "reward/raw_mean": 0.078125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.284781813621521, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8246567249298096, "step": 194, "loss": 0.2473970204591751, "reward/raw_mean": 0.0078125, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30049771070480347, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9259413480758667, "step": 195, "loss": 0.2777824103832245, "reward/raw_mean": 0.3125, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": 5.587935447692871e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2975355386734009, "ppo/advantage_mean": -5.587935447692871e-09, "loss/infonce": 0.4013309180736542, "step": 196, "loss": 0.12039928883314133, "reward/raw_mean": 0.3984375, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29964667558670044, "ppo/advantage_mean": 0.0, "loss/infonce": 1.0622332096099854, "step": 197, "loss": 0.31866997480392456, "reward/raw_mean": 0.015625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3009151518344879, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.9512442946434021, "step": 198, "loss": 0.2853733003139496, "reward/raw_mean": 0.0078125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30242353677749634, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.5916481018066406, "step": 199, "loss": 0.17749443650245667, "reward/raw_mean": 0.0546875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2867797911167145, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.7063463926315308, "step": 200, "loss": 0.21190392971038818, "reward/raw_mean": 0.171875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29281240701675415, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.6260277032852173, "step": 201, "loss": 0.1878083199262619, "reward/raw_mean": 0.1640625, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.75} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3003455698490143, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 0.6878172159194946, "step": 202, "loss": 0.20634517073631287, "reward/raw_mean": 0.0859375, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30529120564460754, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9370609521865845, "step": 203, "loss": 0.2811183035373688, "reward/raw_mean": 0.25, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28734874725341797, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8686985969543457, "step": 204, "loss": 0.26060959696769714, "reward/raw_mean": 0.1796875, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.5} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2953088879585266, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.8476064205169678, "step": 205, "loss": 0.2542819380760193, "reward/raw_mean": 0.1953125, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.25} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3013022840023041, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 1.0226013660430908, "step": 206, "loss": 0.3067804276943207, "reward/raw_mean": 0.1640625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2982737421989441, "ppo/advantage_mean": 0.0, "loss/infonce": 1.2861933708190918, "step": 207, "loss": 0.38585802912712097, "reward/raw_mean": 0.265625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30488383769989014, "ppo/advantage_mean": 0.0, "loss/infonce": 1.0261633396148682, "step": 208, "loss": 0.3078490197658539, "reward/raw_mean": 0.1796875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -5.587935447692871e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30122536420822144, "ppo/advantage_mean": 5.587935447692871e-09, "loss/infonce": 0.922061562538147, "step": 209, "loss": 0.27661848068237305, "reward/raw_mean": 0.203125, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.5} {"loss/policy": 1.862645149230957e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30028462409973145, "ppo/advantage_mean": -1.862645149230957e-08, "loss/infonce": 0.47487422823905945, "step": 210, "loss": 0.14246228337287903, "reward/raw_mean": 0.3046875, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 1.0} {"loss/policy": 1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31001996994018555, "ppo/advantage_mean": -1.862645149230957e-09, "loss/infonce": 1.1074063777923584, "step": 211, "loss": 0.3322219252586365, "reward/raw_mean": 0.1953125, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30752429366111755, "ppo/advantage_mean": 0.0, "loss/infonce": 0.7800717353820801, "step": 212, "loss": 0.23402152955532074, "reward/raw_mean": 0.3046875, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.75} {"loss/policy": 5.587935447692871e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30084243416786194, "ppo/advantage_mean": -5.587935447692871e-09, "loss/infonce": 0.6492710113525391, "step": 213, "loss": 0.1947813183069229, "reward/raw_mean": 0.1640625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3027920424938202, "ppo/advantage_mean": 0.0, "loss/infonce": 0.8126015663146973, "step": 214, "loss": 0.2437804788351059, "reward/raw_mean": 0.2734375, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.75} {"loss/policy": -1.30385160446167e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29185110330581665, "ppo/advantage_mean": 1.30385160446167e-08, "loss/infonce": 0.8461946845054626, "step": 215, "loss": 0.25385841727256775, "reward/raw_mean": 0.3125, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29252249002456665, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.7178804874420166, "step": 216, "loss": 0.21536415815353394, "reward/raw_mean": 0.09375, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2931036651134491, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.6034841537475586, "step": 217, "loss": 0.18104524910449982, "reward/raw_mean": 0.34375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2827812135219574, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.5694001913070679, "step": 218, "loss": 0.17082005739212036, "reward/raw_mean": 0.609375, "reward/gold_mean": 0.875, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30237865447998047, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.4716808795928955, "step": 219, "loss": 0.14150425791740417, "reward/raw_mean": 0.2890625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29558390378952026, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.4155283570289612, "step": 220, "loss": 0.1246585100889206, "reward/raw_mean": 0.40625, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 1.0} {"loss/policy": 1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2922905683517456, "ppo/advantage_mean": -1.1175870895385742e-08, "loss/infonce": 0.514312744140625, "step": 221, "loss": 0.15429385006427765, "reward/raw_mean": 0.4765625, "reward/gold_mean": 0.75, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30545878410339355, "ppo/advantage_mean": 0.0, "loss/infonce": 0.8489686250686646, "step": 222, "loss": 0.25469058752059937, "reward/raw_mean": 0.4609375, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 1.0} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29272937774658203, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 0.8733744025230408, "step": 223, "loss": 0.2620123326778412, "reward/raw_mean": 0.0625, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3005495071411133, "ppo/advantage_mean": 0.0, "loss/infonce": 0.345365047454834, "step": 224, "loss": 0.10360951721668243, "reward/raw_mean": 0.328125, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29653772711753845, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.6548607349395752, "step": 225, "loss": 0.19645823538303375, "reward/raw_mean": 0.0703125, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.25} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29269149899482727, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 1.113673448562622, "step": 226, "loss": 0.3341020345687866, "reward/raw_mean": 0.4296875, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30656248331069946, "ppo/advantage_mean": 0.0, "loss/infonce": 0.7145819664001465, "step": 227, "loss": 0.2143746018409729, "reward/raw_mean": 0.2421875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 1.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29773348569869995, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.6348782777786255, "step": 228, "loss": 0.19046349823474884, "reward/raw_mean": 0.1328125, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28494569659233093, "ppo/advantage_mean": 0.0, "loss/infonce": 0.5952578783035278, "step": 229, "loss": 0.17857736349105835, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31056007742881775, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.0470178127288818, "step": 230, "loss": 0.314105361700058, "reward/raw_mean": 0.2421875, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.5} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29059678316116333, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.770958662033081, "step": 231, "loss": 0.2312876135110855, "reward/raw_mean": 0.046875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29752856492996216, "ppo/advantage_mean": 0.0, "loss/infonce": 1.2005727291107178, "step": 232, "loss": 0.3601718246936798, "reward/raw_mean": 0.234375, "reward/gold_mean": 0.5625, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3064115643501282, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9238795042037964, "step": 233, "loss": 0.27716386318206787, "reward/raw_mean": 0.1875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2908758521080017, "ppo/advantage_mean": 0.0, "loss/infonce": 0.5848391652107239, "step": 234, "loss": 0.17545175552368164, "reward/raw_mean": 0.3125, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2867802381515503, "ppo/advantage_mean": 0.0, "loss/infonce": 0.45560982823371887, "step": 235, "loss": 0.13668295741081238, "reward/raw_mean": 0.0625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2911965250968933, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.801969587802887, "step": 236, "loss": 0.2405908703804016, "reward/raw_mean": 0.171875, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.295573890209198, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9717341661453247, "step": 237, "loss": 0.29152026772499084, "reward/raw_mean": 0.40625, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2954380512237549, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.7946485877037048, "step": 238, "loss": 0.2383945882320404, "reward/raw_mean": 0.40625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30577659606933594, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.41120821237564087, "step": 239, "loss": 0.1233624666929245, "reward/raw_mean": 0.0234375, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.5} {"loss/policy": 1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3031918704509735, "ppo/advantage_mean": -1.862645149230957e-09, "loss/infonce": 0.779310941696167, "step": 240, "loss": 0.23379328846931458, "reward/raw_mean": 0.2265625, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.5} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3062828779220581, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 0.48003554344177246, "step": 241, "loss": 0.14401064813137054, "reward/raw_mean": 0.28125, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.32155388593673706, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 1.0686242580413818, "step": 242, "loss": 0.32058727741241455, "reward/raw_mean": 0.0625, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30365467071533203, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.4197145104408264, "step": 243, "loss": 0.12591436505317688, "reward/raw_mean": 0.484375, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3006548285484314, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.9988036751747131, "step": 244, "loss": 0.29964110255241394, "reward/raw_mean": 0.328125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.308640718460083, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 1.0669044256210327, "step": 245, "loss": 0.32007133960723877, "reward/raw_mean": 0.0625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2969999611377716, "ppo/advantage_mean": 0.0, "loss/infonce": 0.6577760577201843, "step": 246, "loss": 0.19733282923698425, "reward/raw_mean": 0.1875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2958356440067291, "ppo/advantage_mean": 0.0, "loss/infonce": 0.8628095388412476, "step": 247, "loss": 0.2588428854942322, "reward/raw_mean": 0.25, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29175910353660583, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 0.9052473306655884, "step": 248, "loss": 0.2715741991996765, "reward/raw_mean": 0.3125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -5.587935447692871e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30440613627433777, "ppo/advantage_mean": 5.587935447692871e-09, "loss/infonce": 0.8963605761528015, "step": 249, "loss": 0.26890817284584045, "reward/raw_mean": 0.1171875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30530092120170593, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.0302900075912476, "step": 250, "loss": 0.30908700823783875, "reward/raw_mean": 0.15625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3039848506450653, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.5787566900253296, "step": 251, "loss": 0.17362701892852783, "reward/raw_mean": 0.1171875, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 1.0} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2943435311317444, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.7718687653541565, "step": 252, "loss": 0.231560617685318, "reward/raw_mean": 0.0546875, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2978045344352722, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.0454657077789307, "step": 253, "loss": 0.31363973021507263, "reward/raw_mean": 0.1640625, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.75} {"loss/policy": 1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2915816307067871, "ppo/advantage_mean": -1.1175870895385742e-08, "loss/infonce": 0.6626891493797302, "step": 254, "loss": 0.1988067626953125, "reward/raw_mean": 0.3984375, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": 1.4901161193847656e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3048207461833954, "ppo/advantage_mean": -1.4901161193847656e-08, "loss/infonce": 0.6381198167800903, "step": 255, "loss": 0.19143596291542053, "reward/raw_mean": 0.0859375, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2895163297653198, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.0620040893554688, "step": 256, "loss": 0.31860125064849854, "reward/raw_mean": 0.2578125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -1.862645149230957e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.294687420129776, "ppo/advantage_mean": 1.862645149230957e-08, "loss/infonce": 0.8371142148971558, "step": 257, "loss": 0.2511342465877533, "reward/raw_mean": 0.1640625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 1.0} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2988932430744171, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.41787391901016235, "step": 258, "loss": 0.12536218762397766, "reward/raw_mean": 0.1796875, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 1.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2932094633579254, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.5810458064079285, "step": 259, "loss": 0.1743137538433075, "reward/raw_mean": 0.25, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 1.0} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2893695831298828, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 0.8617303371429443, "step": 260, "loss": 0.25851911306381226, "reward/raw_mean": 0.0546875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3108627498149872, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.7367111444473267, "step": 261, "loss": 0.22101335227489471, "reward/raw_mean": 0.078125, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29559120535850525, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8237748146057129, "step": 262, "loss": 0.24713245034217834, "reward/raw_mean": 0.2109375, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2907174527645111, "ppo/advantage_mean": 0.0, "loss/infonce": 1.0063592195510864, "step": 263, "loss": 0.3019077777862549, "reward/raw_mean": 0.078125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29460376501083374, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.5730853080749512, "step": 264, "loss": 0.1719256043434143, "reward/raw_mean": 0.265625, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29274463653564453, "ppo/advantage_mean": 0.0, "loss/infonce": 0.7609459161758423, "step": 265, "loss": 0.22828377783298492, "reward/raw_mean": 0.0859375, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30287250876426697, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.212874412536621, "step": 266, "loss": 0.3638623356819153, "reward/raw_mean": 0.3828125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2946320176124573, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 1.1359899044036865, "step": 267, "loss": 0.34079697728157043, "reward/raw_mean": 0.0546875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29265058040618896, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.5518744587898254, "step": 268, "loss": 0.16556234657764435, "reward/raw_mean": 0.046875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -1.4901161193847656e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3014029562473297, "ppo/advantage_mean": 1.4901161193847656e-08, "loss/infonce": 0.44606196880340576, "step": 269, "loss": 0.133818581700325, "reward/raw_mean": 0.2578125, "reward/gold_mean": 0.75, "retrieval/gold_in_topN": 1.0} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2830459475517273, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.08852775394916534, "step": 270, "loss": 0.026558324694633484, "reward/raw_mean": 0.2265625, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30123621225357056, "ppo/advantage_mean": 0.0, "loss/infonce": 0.8015556335449219, "step": 271, "loss": 0.24046669900417328, "reward/raw_mean": 0.203125, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3025047183036804, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9839944839477539, "step": 272, "loss": 0.29519835114479065, "reward/raw_mean": 0.0859375, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28962981700897217, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8226736783981323, "step": 273, "loss": 0.24680210649967194, "reward/raw_mean": 0.2578125, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3018006682395935, "ppo/advantage_mean": 0.0, "loss/infonce": 0.559166431427002, "step": 274, "loss": 0.16774994134902954, "reward/raw_mean": 0.3984375, "reward/gold_mean": 0.625, "retrieval/gold_in_topN": 1.0} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30336642265319824, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 1.2075681686401367, "step": 275, "loss": 0.3622704744338989, "reward/raw_mean": 0.234375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2899383306503296, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.6228379011154175, "step": 276, "loss": 0.1868513822555542, "reward/raw_mean": 0.46875, "reward/gold_mean": 0.75, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29841911792755127, "ppo/advantage_mean": 0.0, "loss/infonce": 0.550959587097168, "step": 277, "loss": 0.16528788208961487, "reward/raw_mean": 0.2265625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29558467864990234, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 1.3840103149414062, "step": 278, "loss": 0.41520312428474426, "reward/raw_mean": 0.3359375, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29517194628715515, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8944748640060425, "step": 279, "loss": 0.2683424651622772, "reward/raw_mean": 0.140625, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29821351170539856, "ppo/advantage_mean": 0.0, "loss/infonce": 1.0104860067367554, "step": 280, "loss": 0.3031458258628845, "reward/raw_mean": 0.15625, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3038651645183563, "ppo/advantage_mean": 0.0, "loss/infonce": 1.3468239307403564, "step": 281, "loss": 0.4040471911430359, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": 1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30688199400901794, "ppo/advantage_mean": -1.862645149230957e-09, "loss/infonce": 0.6905206441879272, "step": 282, "loss": 0.2071561962366104, "reward/raw_mean": 0.0703125, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2942497432231903, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.47904184460639954, "step": 283, "loss": 0.14371256530284882, "reward/raw_mean": 0.0078125, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29536473751068115, "ppo/advantage_mean": 0.0, "loss/infonce": 1.0494811534881592, "step": 284, "loss": 0.31484436988830566, "reward/raw_mean": 0.015625, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29719656705856323, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.0424525737762451, "step": 285, "loss": 0.31273579597473145, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29527658224105835, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.7164738178253174, "step": 286, "loss": 0.21494215726852417, "reward/raw_mean": 0.2109375, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29647132754325867, "ppo/advantage_mean": 0.0, "loss/infonce": 0.8735998868942261, "step": 287, "loss": 0.26207998394966125, "reward/raw_mean": 0.1015625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28772395849227905, "ppo/advantage_mean": 0.0, "loss/infonce": 1.3837509155273438, "step": 288, "loss": 0.4151252806186676, "reward/raw_mean": 0.0859375, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.5} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29143214225769043, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.5786004662513733, "step": 289, "loss": 0.173580139875412, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2839794456958771, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.7364362478256226, "step": 290, "loss": 0.22093088924884796, "reward/raw_mean": 0.140625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3056837022304535, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.9824456572532654, "step": 291, "loss": 0.2947337031364441, "reward/raw_mean": 0.0078125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29279881715774536, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.5050860047340393, "step": 292, "loss": 0.1515258103609085, "reward/raw_mean": 0.09375, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28983885049819946, "ppo/advantage_mean": 0.0, "loss/infonce": 0.5503572225570679, "step": 293, "loss": 0.16510717570781708, "reward/raw_mean": 0.15625, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30706918239593506, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9229543209075928, "step": 294, "loss": 0.27688631415367126, "reward/raw_mean": 0.1015625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2994837164878845, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.650637149810791, "step": 295, "loss": 0.1951911598443985, "reward/raw_mean": 0.0859375, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31230446696281433, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.4065905213356018, "step": 296, "loss": 0.12197716534137726, "reward/raw_mean": 0.078125, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28415411710739136, "ppo/advantage_mean": 0.0, "loss/infonce": 0.7570014595985413, "step": 297, "loss": 0.2271004468202591, "reward/raw_mean": 0.75, "reward/gold_mean": 0.75, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2927643358707428, "ppo/advantage_mean": 0.0, "loss/infonce": 1.3938446044921875, "step": 298, "loss": 0.41815340518951416, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": 1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2924700379371643, "ppo/advantage_mean": -1.862645149230957e-09, "loss/infonce": 0.6404284238815308, "step": 299, "loss": 0.19212853908538818, "reward/raw_mean": 0.1640625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3034464716911316, "ppo/advantage_mean": 0.0, "loss/infonce": 0.20670288801193237, "step": 300, "loss": 0.06201086938381195, "reward/raw_mean": 0.4453125, "reward/gold_mean": 0.8125, "retrieval/gold_in_topN": 1.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29702895879745483, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.569452166557312, "step": 301, "loss": 0.17083565890789032, "reward/raw_mean": 0.2578125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30226483941078186, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.7611817121505737, "step": 302, "loss": 0.22835451364517212, "reward/raw_mean": 0.2734375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3004966676235199, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.4996473491191864, "step": 303, "loss": 0.14989420771598816, "reward/raw_mean": 0.2109375, "reward/gold_mean": 0.625, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2898225486278534, "ppo/advantage_mean": 0.0, "loss/infonce": 0.8803208470344543, "step": 304, "loss": 0.2640962600708008, "reward/raw_mean": 0.015625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.300454705953598, "ppo/advantage_mean": 0.0, "loss/infonce": 0.958702027797699, "step": 305, "loss": 0.28761062026023865, "reward/raw_mean": 0.015625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3189663887023926, "ppo/advantage_mean": 0.0, "loss/infonce": 0.4018874764442444, "step": 306, "loss": 0.12056624889373779, "reward/raw_mean": 0.2890625, "reward/gold_mean": 0.625, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29896706342697144, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9421898126602173, "step": 307, "loss": 0.2826569676399231, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2944190502166748, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.6044038534164429, "step": 308, "loss": 0.1813211441040039, "reward/raw_mean": 0.5703125, "reward/gold_mean": 0.625, "retrieval/gold_in_topN": 0.75} {"loss/policy": 9.313225746154785e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2951631546020508, "ppo/advantage_mean": -9.313225746154785e-09, "loss/infonce": 0.7734475135803223, "step": 309, "loss": 0.23203428089618683, "reward/raw_mean": 0.34375, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3002999424934387, "ppo/advantage_mean": 0.0, "loss/infonce": 0.938201904296875, "step": 310, "loss": 0.28146058320999146, "reward/raw_mean": 0.2265625, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.310097336769104, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.5821896195411682, "step": 311, "loss": 0.17465689778327942, "reward/raw_mean": 0.2734375, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.5} {"loss/policy": 9.313225746154785e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28770506381988525, "ppo/advantage_mean": -9.313225746154785e-09, "loss/infonce": 1.1010854244232178, "step": 312, "loss": 0.3303256332874298, "reward/raw_mean": 0.328125, "reward/gold_mean": 0.75, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30277660489082336, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.2507082223892212, "step": 313, "loss": 0.37521249055862427, "reward/raw_mean": 0.0234375, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2943955659866333, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.6959377527236938, "step": 314, "loss": 0.20878133177757263, "reward/raw_mean": 0.0546875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2952099144458771, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.18594110012054443, "step": 315, "loss": 0.05578232929110527, "reward/raw_mean": 0.2578125, "reward/gold_mean": 0.6875, "retrieval/gold_in_topN": 0.75} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3013545572757721, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 0.22398865222930908, "step": 316, "loss": 0.06719659268856049, "reward/raw_mean": 0.1640625, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 1.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2962402105331421, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.6209895014762878, "step": 317, "loss": 0.18629685044288635, "reward/raw_mean": 0.3046875, "reward/gold_mean": 0.625, "retrieval/gold_in_topN": 0.75} {"loss/policy": -9.313225746154785e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29392921924591064, "ppo/advantage_mean": 9.313225746154785e-09, "loss/infonce": 1.380077838897705, "step": 318, "loss": 0.41402336955070496, "reward/raw_mean": 0.09375, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3168911635875702, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.2954709529876709, "step": 319, "loss": 0.08864128589630127, "reward/raw_mean": 0.0234375, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.5} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2973628044128418, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.7506076097488403, "step": 320, "loss": 0.22518229484558105, "reward/raw_mean": 0.265625, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29381120204925537, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.0118321180343628, "step": 321, "loss": 0.3035496473312378, "reward/raw_mean": 0.21875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2889130115509033, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 0.26824840903282166, "step": 322, "loss": 0.08047451078891754, "reward/raw_mean": 0.53125, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 1.0} {"loss/policy": 2.1420419216156006e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30861344933509827, "ppo/advantage_mean": -2.1420419216156006e-08, "loss/infonce": 0.36651065945625305, "step": 323, "loss": 0.10995322465896606, "reward/raw_mean": 0.5234375, "reward/gold_mean": 0.75, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2975013256072998, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.21312247216701508, "step": 324, "loss": 0.0639367401599884, "reward/raw_mean": 0.2890625, "reward/gold_mean": 0.5625, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2952447831630707, "ppo/advantage_mean": 0.0, "loss/infonce": 0.37552565336227417, "step": 325, "loss": 0.11265770345926285, "reward/raw_mean": 0.109375, "reward/gold_mean": 0.75, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30699461698532104, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.04873912036418915, "step": 326, "loss": 0.014621732756495476, "reward/raw_mean": 0.1953125, "reward/gold_mean": 0.5625, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30334657430648804, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.5993919372558594, "step": 327, "loss": 0.1798175871372223, "reward/raw_mean": 0.0546875, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 1.0} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2973634600639343, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 0.8301936388015747, "step": 328, "loss": 0.2490580976009369, "reward/raw_mean": 0.28125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.303905189037323, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.6966147422790527, "step": 329, "loss": 0.20898443460464478, "reward/raw_mean": 0.09375, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2966226041316986, "ppo/advantage_mean": 0.0, "loss/infonce": 0.4575583338737488, "step": 330, "loss": 0.13726750016212463, "reward/raw_mean": 0.4609375, "reward/gold_mean": 0.875, "retrieval/gold_in_topN": 1.0} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30105865001678467, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.9138174057006836, "step": 331, "loss": 0.274145245552063, "reward/raw_mean": 0.0625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3063874840736389, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.6942148208618164, "step": 332, "loss": 0.20826444029808044, "reward/raw_mean": 0.390625, "reward/gold_mean": 0.75, "retrieval/gold_in_topN": 0.75} {"loss/policy": 1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2902024984359741, "ppo/advantage_mean": -1.862645149230957e-09, "loss/infonce": 0.7452137470245361, "step": 333, "loss": 0.22356413304805756, "reward/raw_mean": 0.2578125, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.5} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2907778024673462, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.7596837282180786, "step": 334, "loss": 0.22790512442588806, "reward/raw_mean": 0.1328125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29610365629196167, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.23419928550720215, "step": 335, "loss": 0.07025977969169617, "reward/raw_mean": 0.484375, "reward/gold_mean": 0.6875, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30537277460098267, "ppo/advantage_mean": 0.0, "loss/infonce": 0.40619829297065735, "step": 336, "loss": 0.12185949087142944, "reward/raw_mean": 0.2109375, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3009055256843567, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.6635556221008301, "step": 337, "loss": 0.19906669855117798, "reward/raw_mean": 0.421875, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29049479961395264, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.7193570733070374, "step": 338, "loss": 0.21580712497234344, "reward/raw_mean": 0.28125, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2874602675437927, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.6968574523925781, "step": 339, "loss": 0.20905724167823792, "reward/raw_mean": 0.2578125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3064127564430237, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.274397611618042, "step": 340, "loss": 0.38231930136680603, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3024643659591675, "ppo/advantage_mean": 0.0, "loss/infonce": 0.6161624789237976, "step": 341, "loss": 0.18484875559806824, "reward/raw_mean": 0.3046875, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30166298151016235, "ppo/advantage_mean": 0.0, "loss/infonce": 0.501425564289093, "step": 342, "loss": 0.1504276692867279, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3100256025791168, "ppo/advantage_mean": 0.0, "loss/infonce": 0.766867995262146, "step": 343, "loss": 0.230060413479805, "reward/raw_mean": 0.0703125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": 1.4901161193847656e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3046135902404785, "ppo/advantage_mean": -1.4901161193847656e-08, "loss/infonce": 0.7023473978042603, "step": 344, "loss": 0.2107042372226715, "reward/raw_mean": 0.5546875, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": 5.587935447692871e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3123428523540497, "ppo/advantage_mean": -5.587935447692871e-09, "loss/infonce": 0.9056224822998047, "step": 345, "loss": 0.27168676257133484, "reward/raw_mean": 0.125, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28836727142333984, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.7338013648986816, "step": 346, "loss": 0.22014041244983673, "reward/raw_mean": 0.1796875, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.299473375082016, "ppo/advantage_mean": 0.0, "loss/infonce": 0.8202524185180664, "step": 347, "loss": 0.24607573449611664, "reward/raw_mean": 0.1953125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2982337176799774, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 0.7797302007675171, "step": 348, "loss": 0.23391908407211304, "reward/raw_mean": 0.171875, "reward/gold_mean": 0.5625, "retrieval/gold_in_topN": 0.75} {"loss/policy": 1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2950572371482849, "ppo/advantage_mean": -1.1175870895385742e-08, "loss/infonce": 1.6077394485473633, "step": 349, "loss": 0.4823218584060669, "reward/raw_mean": 0.484375, "reward/gold_mean": 0.6875, "retrieval/gold_in_topN": 0.25} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2903764247894287, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.4788314700126648, "step": 350, "loss": 0.14364944398403168, "reward/raw_mean": 0.1640625, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2916521430015564, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.28547853231430054, "step": 351, "loss": 0.08564355969429016, "reward/raw_mean": 0.359375, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3018020689487457, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.6639653444290161, "step": 352, "loss": 0.19918961822986603, "reward/raw_mean": 0.3671875, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3065747022628784, "ppo/advantage_mean": 0.0, "loss/infonce": 1.1148988008499146, "step": 353, "loss": 0.33446964621543884, "reward/raw_mean": 0.453125, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3031105399131775, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 1.3551368713378906, "step": 354, "loss": 0.4065410792827606, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3091612458229065, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 1.080134391784668, "step": 355, "loss": 0.32404032349586487, "reward/raw_mean": 0.0234375, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.25} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3098771274089813, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.7596354484558105, "step": 356, "loss": 0.22789064049720764, "reward/raw_mean": 0.21875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30385372042655945, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.41919729113578796, "step": 357, "loss": 0.12575919926166534, "reward/raw_mean": 0.078125, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2966722249984741, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.6420717239379883, "step": 358, "loss": 0.19262152910232544, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2997112572193146, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 0.5168105959892273, "step": 359, "loss": 0.15504318475723267, "reward/raw_mean": 0.2109375, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.308418869972229, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.5705037117004395, "step": 360, "loss": 0.17115110158920288, "reward/raw_mean": 0.1484375, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 1.0} {"loss/policy": 1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3033602237701416, "ppo/advantage_mean": -1.1175870895385742e-08, "loss/infonce": 0.6121572852134705, "step": 361, "loss": 0.18364720046520233, "reward/raw_mean": 0.1171875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2904060184955597, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.9779129028320312, "step": 362, "loss": 0.29337388277053833, "reward/raw_mean": 0.1328125, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.75} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2847919762134552, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 0.6341083645820618, "step": 363, "loss": 0.190232515335083, "reward/raw_mean": 0.1796875, "reward/gold_mean": 0.75, "retrieval/gold_in_topN": 1.0} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2884429097175598, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 1.0551714897155762, "step": 364, "loss": 0.31655144691467285, "reward/raw_mean": 0.0859375, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29030805826187134, "ppo/advantage_mean": 0.0, "loss/infonce": 0.8091946840286255, "step": 365, "loss": 0.24275840818881989, "reward/raw_mean": 0.0859375, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.75} {"loss/policy": -1.4901161193847656e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28901761770248413, "ppo/advantage_mean": 1.4901161193847656e-08, "loss/infonce": 0.6133477091789246, "step": 366, "loss": 0.1840043067932129, "reward/raw_mean": 0.3203125, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28940320014953613, "ppo/advantage_mean": 0.0, "loss/infonce": 0.5596575736999512, "step": 367, "loss": 0.1678972840309143, "reward/raw_mean": 0.2734375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.293472021818161, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.5517607927322388, "step": 368, "loss": 0.16552823781967163, "reward/raw_mean": 0.2109375, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 1.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29945895075798035, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.773394763469696, "step": 369, "loss": 0.23201844096183777, "reward/raw_mean": 0.3046875, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30406874418258667, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.7015682458877563, "step": 370, "loss": 0.21047048270702362, "reward/raw_mean": 0.40625, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29941532015800476, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.8449860215187073, "step": 371, "loss": 0.25349581241607666, "reward/raw_mean": 0.0546875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3082295060157776, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8194268941879272, "step": 372, "loss": 0.2458280771970749, "reward/raw_mean": 0.0703125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": 1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28275251388549805, "ppo/advantage_mean": -1.862645149230957e-09, "loss/infonce": 0.5901689529418945, "step": 373, "loss": 0.17705069482326508, "reward/raw_mean": 0.1796875, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.32239413261413574, "ppo/advantage_mean": 0.0, "loss/infonce": 0.42797988653182983, "step": 374, "loss": 0.1283939778804779, "reward/raw_mean": 0.28125, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29902029037475586, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.9768764972686768, "step": 375, "loss": 0.2930629551410675, "reward/raw_mean": 0.1015625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3058512806892395, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 0.46846461296081543, "step": 376, "loss": 0.14053940773010254, "reward/raw_mean": 0.1796875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2982681393623352, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 1.470818281173706, "step": 377, "loss": 0.44124549627304077, "reward/raw_mean": 0.375, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2929673194885254, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.1550663709640503, "step": 378, "loss": 0.34651991724967957, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": 5.587935447692871e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3020040690898895, "ppo/advantage_mean": -5.587935447692871e-09, "loss/infonce": 0.6914860010147095, "step": 379, "loss": 0.20744581520557404, "reward/raw_mean": 0.1875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3003227710723877, "ppo/advantage_mean": 0.0, "loss/infonce": 0.8190566301345825, "step": 380, "loss": 0.24571700394153595, "reward/raw_mean": 0.1484375, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.5} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3009507358074188, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.3999890685081482, "step": 381, "loss": 0.11999672651290894, "reward/raw_mean": 0.7734375, "reward/gold_mean": 0.6875, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30986109375953674, "ppo/advantage_mean": 0.0, "loss/infonce": 0.723702073097229, "step": 382, "loss": 0.21711063385009766, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -1.862645149230957e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2931267023086548, "ppo/advantage_mean": 1.862645149230957e-08, "loss/infonce": 0.9823389053344727, "step": 383, "loss": 0.2947016656398773, "reward/raw_mean": 0.2890625, "reward/gold_mean": 0.9375, "retrieval/gold_in_topN": 0.75} {"loss/policy": 1.30385160446167e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2932332158088684, "ppo/advantage_mean": -1.30385160446167e-08, "loss/infonce": 0.1657494455575943, "step": 384, "loss": 0.049724847078323364, "reward/raw_mean": 0.2578125, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 1.0} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29714885354042053, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 0.6925234198570251, "step": 385, "loss": 0.20775702595710754, "reward/raw_mean": 0.2265625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29628127813339233, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.7384421825408936, "step": 386, "loss": 0.2215326577425003, "reward/raw_mean": 0.046875, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30694299936294556, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8964685797691345, "step": 387, "loss": 0.26894059777259827, "reward/raw_mean": 0.4296875, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29392969608306885, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.45945948362350464, "step": 388, "loss": 0.13783785700798035, "reward/raw_mean": 0.390625, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30677530169487, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.47805696725845337, "step": 389, "loss": 0.143417090177536, "reward/raw_mean": 0.1953125, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.5} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3029263913631439, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 1.106743335723877, "step": 390, "loss": 0.332023024559021, "reward/raw_mean": 0.0859375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29873502254486084, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.4842633605003357, "step": 391, "loss": 0.14527902007102966, "reward/raw_mean": 0.3984375, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30102604627609253, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.6140720844268799, "step": 392, "loss": 0.18422162532806396, "reward/raw_mean": 0.546875, "reward/gold_mean": 0.625, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3015706539154053, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9683427810668945, "step": 393, "loss": 0.2905028462409973, "reward/raw_mean": 0.25, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30152082443237305, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 1.1465463638305664, "step": 394, "loss": 0.3439639210700989, "reward/raw_mean": 0.296875, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28806787729263306, "ppo/advantage_mean": 0.0, "loss/infonce": 0.8560073375701904, "step": 395, "loss": 0.25680220127105713, "reward/raw_mean": 0.390625, "reward/gold_mean": 0.75, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2931174635887146, "ppo/advantage_mean": 0.0, "loss/infonce": 0.5554456114768982, "step": 396, "loss": 0.1666336953639984, "reward/raw_mean": 0.0625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2954755425453186, "ppo/advantage_mean": 0.0, "loss/infonce": 0.5832453966140747, "step": 397, "loss": 0.17497362196445465, "reward/raw_mean": 0.1796875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31632184982299805, "ppo/advantage_mean": 0.0, "loss/infonce": 1.2689971923828125, "step": 398, "loss": 0.38069918751716614, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30525997281074524, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.513322114944458, "step": 399, "loss": 0.15399664640426636, "reward/raw_mean": 0.09375, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30471673607826233, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.4131936728954315, "step": 400, "loss": 0.12395811080932617, "reward/raw_mean": 0.3671875, "reward/gold_mean": 0.75, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2977359890937805, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.6800991296768188, "step": 401, "loss": 0.20402975380420685, "reward/raw_mean": 0.3671875, "reward/gold_mean": 0.625, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.297838419675827, "ppo/advantage_mean": 0.0, "loss/infonce": 0.8048428893089294, "step": 402, "loss": 0.2414528727531433, "reward/raw_mean": 0.25, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3104149103164673, "ppo/advantage_mean": 0.0, "loss/infonce": 1.0674870014190674, "step": 403, "loss": 0.3202461004257202, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30395519733428955, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9400833249092102, "step": 404, "loss": 0.282025009393692, "reward/raw_mean": 0.25, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2990646958351135, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.9476559162139893, "step": 405, "loss": 0.28429678082466125, "reward/raw_mean": 0.2109375, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29966306686401367, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.49558496475219727, "step": 406, "loss": 0.14867550134658813, "reward/raw_mean": 0.28125, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3059403896331787, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.5200903415679932, "step": 407, "loss": 0.15602710843086243, "reward/raw_mean": 0.1015625, "reward/gold_mean": 0.125, "retrieval/gold_in_topN": 1.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3032912611961365, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.7754000425338745, "step": 408, "loss": 0.2326200157403946, "reward/raw_mean": 0.3671875, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": 1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28267019987106323, "ppo/advantage_mean": -1.862645149230957e-09, "loss/infonce": 0.37814271450042725, "step": 409, "loss": 0.1134428158402443, "reward/raw_mean": 0.1953125, "reward/gold_mean": 0.5625, "retrieval/gold_in_topN": 1.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2924497127532959, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.45979955792427063, "step": 410, "loss": 0.13793987035751343, "reward/raw_mean": 0.28125, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 1.0} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29448312520980835, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.8680078983306885, "step": 411, "loss": 0.2604023814201355, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3028748035430908, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.2039833068847656, "step": 412, "loss": 0.36119499802589417, "reward/raw_mean": 0.1171875, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2922949194908142, "ppo/advantage_mean": 0.0, "loss/infonce": 0.6302621364593506, "step": 413, "loss": 0.18907864391803741, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": 1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29607102274894714, "ppo/advantage_mean": -1.1175870895385742e-08, "loss/infonce": 0.9406014084815979, "step": 414, "loss": 0.28218042850494385, "reward/raw_mean": 0.1015625, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 1.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29258283972740173, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8925913572311401, "step": 415, "loss": 0.2677774131298065, "reward/raw_mean": 0.2734375, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28784236311912537, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.6511287093162537, "step": 416, "loss": 0.19533862173557281, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29333773255348206, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9075900316238403, "step": 417, "loss": 0.27227702736854553, "reward/raw_mean": 0.2421875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31043118238449097, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.6234749555587769, "step": 418, "loss": 0.1870424896478653, "reward/raw_mean": 0.2421875, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30399274826049805, "ppo/advantage_mean": 0.0, "loss/infonce": 0.7540640830993652, "step": 419, "loss": 0.22621923685073853, "reward/raw_mean": 0.1171875, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28862231969833374, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.6872894763946533, "step": 420, "loss": 0.20618683099746704, "reward/raw_mean": 0.28125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30498477816581726, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 0.9154314994812012, "step": 421, "loss": 0.27462947368621826, "reward/raw_mean": 0.4140625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": 7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3126543164253235, "ppo/advantage_mean": -7.450580596923828e-09, "loss/infonce": 0.5816271305084229, "step": 422, "loss": 0.1744881570339203, "reward/raw_mean": 0.4765625, "reward/gold_mean": 0.6875, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30330267548561096, "ppo/advantage_mean": 0.0, "loss/infonce": 1.199042797088623, "step": 423, "loss": 0.3597128391265869, "reward/raw_mean": 0.4609375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2902953624725342, "ppo/advantage_mean": 0.0, "loss/infonce": 0.7271599769592285, "step": 424, "loss": 0.21814800798892975, "reward/raw_mean": 0.3359375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30159446597099304, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.6851967573165894, "step": 425, "loss": 0.20555901527404785, "reward/raw_mean": 0.0390625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -4.6566128730773926e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29562458395957947, "ppo/advantage_mean": 4.6566128730773926e-09, "loss/infonce": 0.9413676857948303, "step": 426, "loss": 0.28241032361984253, "reward/raw_mean": 0.1171875, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 1.0} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31080710887908936, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.6384531259536743, "step": 427, "loss": 0.19153594970703125, "reward/raw_mean": 0.109375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.310690313577652, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8127123117446899, "step": 428, "loss": 0.24381370842456818, "reward/raw_mean": 0.2734375, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.302250474691391, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.6540656685829163, "step": 429, "loss": 0.19621971249580383, "reward/raw_mean": 0.0390625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30987900495529175, "ppo/advantage_mean": 0.0, "loss/infonce": 1.3868865966796875, "step": 430, "loss": 0.4160659909248352, "reward/raw_mean": 0.25, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2976919412612915, "ppo/advantage_mean": 0.0, "loss/infonce": 0.5221198797225952, "step": 431, "loss": 0.15663596987724304, "reward/raw_mean": 0.2578125, "reward/gold_mean": 0.75, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29820579290390015, "ppo/advantage_mean": 0.0, "loss/infonce": 1.2312387228012085, "step": 432, "loss": 0.369371622800827, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.0} {"loss/policy": -5.587935447692871e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2891470193862915, "ppo/advantage_mean": 5.587935447692871e-09, "loss/infonce": 0.8785070180892944, "step": 433, "loss": 0.26355212926864624, "reward/raw_mean": 0.234375, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30556029081344604, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.0901200771331787, "step": 434, "loss": 0.3270360231399536, "reward/raw_mean": 0.3515625, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.25} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.28954339027404785, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 0.35176175832748413, "step": 435, "loss": 0.10552851855754852, "reward/raw_mean": 0.2265625, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29781341552734375, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.622369647026062, "step": 436, "loss": 0.1867108941078186, "reward/raw_mean": 0.578125, "reward/gold_mean": 0.75, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31266099214553833, "ppo/advantage_mean": 0.0, "loss/infonce": 0.4826454520225525, "step": 437, "loss": 0.14479364454746246, "reward/raw_mean": 0.0546875, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.5} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29961830377578735, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.6963191032409668, "step": 438, "loss": 0.208895742893219, "reward/raw_mean": 0.21875, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2945599853992462, "ppo/advantage_mean": 0.0, "loss/infonce": 0.3876861333847046, "step": 439, "loss": 0.11630584299564362, "reward/raw_mean": 0.1796875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 1.0} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3025546967983246, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.9202755093574524, "step": 440, "loss": 0.2760826647281647, "reward/raw_mean": 0.0390625, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3106520473957062, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.5035804510116577, "step": 441, "loss": 0.1510741412639618, "reward/raw_mean": 0.09375, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 1.0} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30885255336761475, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.7375227808952332, "step": 442, "loss": 0.22125683724880219, "reward/raw_mean": 0.0625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -5.587935447692871e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30033883452415466, "ppo/advantage_mean": 5.587935447692871e-09, "loss/infonce": 0.6522237062454224, "step": 443, "loss": 0.1956671178340912, "reward/raw_mean": 0.1484375, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 1.0} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29651767015457153, "ppo/advantage_mean": 0.0, "loss/infonce": 0.6745226383209229, "step": 444, "loss": 0.20235680043697357, "reward/raw_mean": 0.1328125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29713332653045654, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.6880037188529968, "step": 445, "loss": 0.20640112459659576, "reward/raw_mean": 0.2265625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2876276969909668, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.7241493463516235, "step": 446, "loss": 0.21724480390548706, "reward/raw_mean": 0.078125, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31032678484916687, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 1.0983941555023193, "step": 447, "loss": 0.32951825857162476, "reward/raw_mean": 0.359375, "reward/gold_mean": 0.5625, "retrieval/gold_in_topN": 0.75} {"loss/policy": 1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2966705560684204, "ppo/advantage_mean": -1.862645149230957e-09, "loss/infonce": 0.7719413042068481, "step": 448, "loss": 0.2315824031829834, "reward/raw_mean": 0.3359375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2911745309829712, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.662398636341095, "step": 449, "loss": 0.19871960580348969, "reward/raw_mean": 0.2890625, "reward/gold_mean": 0.5625, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3213757872581482, "ppo/advantage_mean": 0.0, "loss/infonce": 1.0550475120544434, "step": 450, "loss": 0.316514253616333, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.301262229681015, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.4959173798561096, "step": 451, "loss": 0.14877521991729736, "reward/raw_mean": 0.2578125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30543768405914307, "ppo/advantage_mean": 0.0, "loss/infonce": 0.4786754548549652, "step": 452, "loss": 0.1436026394367218, "reward/raw_mean": 0.34375, "reward/gold_mean": 0.75, "retrieval/gold_in_topN": 1.0} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2954030930995941, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.7824715375900269, "step": 453, "loss": 0.2347414493560791, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3006252944469452, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.3698391914367676, "step": 454, "loss": 0.1109517514705658, "reward/raw_mean": 0.03125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.317288875579834, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.16749505698680878, "step": 455, "loss": 0.050248514860868454, "reward/raw_mean": 0.578125, "reward/gold_mean": 0.8125, "retrieval/gold_in_topN": 1.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3004906177520752, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.6764429807662964, "step": 456, "loss": 0.2029329091310501, "reward/raw_mean": 0.53125, "reward/gold_mean": 0.6875, "retrieval/gold_in_topN": 0.5} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3093879520893097, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 0.6210685968399048, "step": 457, "loss": 0.18632057309150696, "reward/raw_mean": 0.2109375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 1.0} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29330360889434814, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8305956125259399, "step": 458, "loss": 0.2491786926984787, "reward/raw_mean": 0.515625, "reward/gold_mean": 0.625, "retrieval/gold_in_topN": 0.25} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2896314859390259, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 1.0702505111694336, "step": 459, "loss": 0.3210751712322235, "reward/raw_mean": 0.3359375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3113095164299011, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.6890296936035156, "step": 460, "loss": 0.20670892298221588, "reward/raw_mean": 0.203125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30884477496147156, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 1.0180753469467163, "step": 461, "loss": 0.3054226040840149, "reward/raw_mean": 0.15625, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31261926889419556, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.903563916683197, "step": 462, "loss": 0.271069198846817, "reward/raw_mean": 0.1328125, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30400145053863525, "ppo/advantage_mean": 0.0, "loss/infonce": 0.7615422010421753, "step": 463, "loss": 0.22846266627311707, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29472166299819946, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 1.0354887247085571, "step": 464, "loss": 0.3106466233730316, "reward/raw_mean": 0.4609375, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2960866689682007, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 0.48535868525505066, "step": 465, "loss": 0.145607590675354, "reward/raw_mean": 0.1640625, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.307328462600708, "ppo/advantage_mean": 0.0, "loss/infonce": 1.2677186727523804, "step": 466, "loss": 0.3803156316280365, "reward/raw_mean": 0.3203125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2950345277786255, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9818552732467651, "step": 467, "loss": 0.294556587934494, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2973374128341675, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8895963430404663, "step": 468, "loss": 0.2668789029121399, "reward/raw_mean": 0.0234375, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31154173612594604, "ppo/advantage_mean": 0.0, "loss/infonce": 0.4596962630748749, "step": 469, "loss": 0.13790889084339142, "reward/raw_mean": 0.0625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3004310131072998, "ppo/advantage_mean": 0.0, "loss/infonce": 0.9126430749893188, "step": 470, "loss": 0.27379292249679565, "reward/raw_mean": 0.0625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.319036602973938, "ppo/advantage_mean": 0.0, "loss/infonce": 1.3021697998046875, "step": 471, "loss": 0.3906509578227997, "reward/raw_mean": 0.0, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.25} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30457764863967896, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 1.1954951286315918, "step": 472, "loss": 0.35864853858947754, "reward/raw_mean": 0.015625, "reward/gold_mean": 0.0, "retrieval/gold_in_topN": 0.75} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2965831756591797, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.613370954990387, "step": 473, "loss": 0.1840112954378128, "reward/raw_mean": 0.1328125, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31834620237350464, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.5680142045021057, "step": 474, "loss": 0.17040427029132843, "reward/raw_mean": 0.3125, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 1.0} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3062400817871094, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.8614996671676636, "step": 475, "loss": 0.258449912071228, "reward/raw_mean": 0.3828125, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.292795330286026, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.7357416152954102, "step": 476, "loss": 0.220722496509552, "reward/raw_mean": 0.3828125, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.298237681388855, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.6564226150512695, "step": 477, "loss": 0.1969267874956131, "reward/raw_mean": 0.3125, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.5} {"loss/policy": 1.4901161193847656e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.31232157349586487, "ppo/advantage_mean": -1.4901161193847656e-08, "loss/infonce": 0.6370147466659546, "step": 478, "loss": 0.1911044418811798, "reward/raw_mean": 0.1640625, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.299661785364151, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8436691761016846, "step": 479, "loss": 0.25310075283050537, "reward/raw_mean": 0.125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2946995496749878, "ppo/advantage_mean": 0.0, "loss/infonce": 1.4761295318603516, "step": 480, "loss": 0.4428388774394989, "reward/raw_mean": 0.3125, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3024906516075134, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.34105896949768066, "step": 481, "loss": 0.1023176908493042, "reward/raw_mean": 0.0234375, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.29697537422180176, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.5705403089523315, "step": 482, "loss": 0.17116209864616394, "reward/raw_mean": 0.171875, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.25} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3108978271484375, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8219825029373169, "step": 483, "loss": 0.24659475684165955, "reward/raw_mean": 0.109375, "reward/gold_mean": 0.1875, "retrieval/gold_in_topN": 0.5} {"loss/policy": -1.862645149230957e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30234527587890625, "ppo/advantage_mean": 1.862645149230957e-09, "loss/infonce": 0.9217197895050049, "step": 484, "loss": 0.2765159606933594, "reward/raw_mean": 0.1171875, "reward/gold_mean": 0.0625, "retrieval/gold_in_topN": 0.0} {"loss/policy": -1.1175870895385742e-08, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30964022874832153, "ppo/advantage_mean": 1.1175870895385742e-08, "loss/infonce": 1.5055609941482544, "step": 485, "loss": 0.45166832208633423, "reward/raw_mean": 0.1171875, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.5} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.296198308467865, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.6039305329322815, "step": 486, "loss": 0.18117916584014893, "reward/raw_mean": 0.0390625, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2992171049118042, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 1.0392887592315674, "step": 487, "loss": 0.3117866516113281, "reward/raw_mean": 0.1328125, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 1.0} {"loss/policy": 3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2929108738899231, "ppo/advantage_mean": -3.725290298461914e-09, "loss/infonce": 0.9596564769744873, "step": 488, "loss": 0.2878969609737396, "reward/raw_mean": 0.4609375, "reward/gold_mean": 0.625, "retrieval/gold_in_topN": 0.75} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2958095669746399, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.785751461982727, "step": 489, "loss": 0.23572543263435364, "reward/raw_mean": 0.0625, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2941775918006897, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.4932156801223755, "step": 490, "loss": 0.1479647159576416, "reward/raw_mean": 0.4296875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2898826599121094, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8304603099822998, "step": 491, "loss": 0.24913810193538666, "reward/raw_mean": 0.265625, "reward/gold_mean": 0.8125, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3142240643501282, "ppo/advantage_mean": 0.0, "loss/infonce": 1.0509095191955566, "step": 492, "loss": 0.31527286767959595, "reward/raw_mean": 0.125, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30095183849334717, "ppo/advantage_mean": 0.0, "loss/infonce": 1.239243745803833, "step": 493, "loss": 0.3717731237411499, "reward/raw_mean": 0.296875, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.5} {"loss/policy": -7.450580596923828e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2970273196697235, "ppo/advantage_mean": 7.450580596923828e-09, "loss/infonce": 0.7982999086380005, "step": 494, "loss": 0.23948997259140015, "reward/raw_mean": 0.3125, "reward/gold_mean": 0.5, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.30770644545555115, "ppo/advantage_mean": 0.0, "loss/infonce": 0.2364426851272583, "step": 495, "loss": 0.07093280553817749, "reward/raw_mean": 0.296875, "reward/gold_mean": 0.4375, "retrieval/gold_in_topN": 0.75} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3065272569656372, "ppo/advantage_mean": 0.0, "loss/infonce": 0.6207149028778076, "step": 496, "loss": 0.18621447682380676, "reward/raw_mean": 0.1796875, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5} {"loss/policy": 0.0, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.2939179539680481, "ppo/advantage_mean": 0.0, "loss/infonce": 0.6605318784713745, "step": 497, "loss": 0.1981595754623413, "reward/raw_mean": 0.2109375, "reward/gold_mean": 0.3125, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3109589219093323, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.9366530179977417, "step": 498, "loss": 0.2809959053993225, "reward/raw_mean": 0.25, "reward/gold_mean": 0.375, "retrieval/gold_in_topN": 0.75} {"loss/policy": -3.725290298461914e-09, "loss/value": 0.0, "loss/kl": 0.0, "ppo/ratio_mean": 1.0, "ppo/clipfrac": 0.0, "ppo/value_mean": 0.3022209703922272, "ppo/advantage_mean": 3.725290298461914e-09, "loss/infonce": 0.8122575283050537, "step": 499, "loss": 0.2436772733926773, "reward/raw_mean": 0.2734375, "reward/gold_mean": 0.25, "retrieval/gold_in_topN": 0.5}