| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 9480, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0010548523206751054, | |
| "grad_norm": 1.2911568880081177, | |
| "learning_rate": 0.00015822784810126583, | |
| "loss": 7.5122, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.002109704641350211, | |
| "grad_norm": 1.1530958414077759, | |
| "learning_rate": 0.00031645569620253165, | |
| "loss": 6.906, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0031645569620253164, | |
| "grad_norm": 0.8528544306755066, | |
| "learning_rate": 0.00047468354430379745, | |
| "loss": 6.2587, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.004219409282700422, | |
| "grad_norm": 0.9695652723312378, | |
| "learning_rate": 0.0006329113924050633, | |
| "loss": 5.7729, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.005274261603375527, | |
| "grad_norm": 0.8636988401412964, | |
| "learning_rate": 0.0007911392405063291, | |
| "loss": 5.3159, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.006329113924050633, | |
| "grad_norm": 0.9786996841430664, | |
| "learning_rate": 0.0009493670886075949, | |
| "loss": 4.8059, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.007383966244725738, | |
| "grad_norm": 1.5117980241775513, | |
| "learning_rate": 0.0011075949367088608, | |
| "loss": 4.3993, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.008438818565400843, | |
| "grad_norm": 0.8246951699256897, | |
| "learning_rate": 0.0012658227848101266, | |
| "loss": 4.1436, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.00949367088607595, | |
| "grad_norm": 1.2078777551651, | |
| "learning_rate": 0.0014240506329113926, | |
| "loss": 3.9317, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.010548523206751054, | |
| "grad_norm": 0.9894590377807617, | |
| "learning_rate": 0.0015, | |
| "loss": 3.7802, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.011603375527426161, | |
| "grad_norm": 0.8762866854667664, | |
| "learning_rate": 0.0015, | |
| "loss": 3.622, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.012658227848101266, | |
| "grad_norm": 0.9929517507553101, | |
| "learning_rate": 0.0015, | |
| "loss": 3.5117, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.013713080168776372, | |
| "grad_norm": 0.7622191905975342, | |
| "learning_rate": 0.0015, | |
| "loss": 3.4092, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.014767932489451477, | |
| "grad_norm": 0.7740503549575806, | |
| "learning_rate": 0.0015, | |
| "loss": 3.3058, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.015822784810126583, | |
| "grad_norm": 0.8201466202735901, | |
| "learning_rate": 0.0015, | |
| "loss": 3.2344, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.016877637130801686, | |
| "grad_norm": 0.829488217830658, | |
| "learning_rate": 0.0015, | |
| "loss": 3.175, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.017932489451476793, | |
| "grad_norm": 1.1429024934768677, | |
| "learning_rate": 0.0015, | |
| "loss": 3.0982, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.0189873417721519, | |
| "grad_norm": 0.7297139167785645, | |
| "learning_rate": 0.0015, | |
| "loss": 3.0552, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.020042194092827006, | |
| "grad_norm": 0.8383708000183105, | |
| "learning_rate": 0.0015, | |
| "loss": 2.9936, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.02109704641350211, | |
| "grad_norm": 0.7045958638191223, | |
| "learning_rate": 0.0015, | |
| "loss": 2.9414, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.022151898734177215, | |
| "grad_norm": 0.9524938464164734, | |
| "learning_rate": 0.0015, | |
| "loss": 2.9054, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.023206751054852322, | |
| "grad_norm": 0.8136796951293945, | |
| "learning_rate": 0.0015, | |
| "loss": 2.8717, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.024261603375527425, | |
| "grad_norm": 0.7210149765014648, | |
| "learning_rate": 0.0015, | |
| "loss": 2.8125, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.02531645569620253, | |
| "grad_norm": 1.000649333000183, | |
| "learning_rate": 0.0015, | |
| "loss": 2.7796, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.026371308016877638, | |
| "grad_norm": 0.9027130007743835, | |
| "learning_rate": 0.0015, | |
| "loss": 2.7438, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.027426160337552744, | |
| "grad_norm": 0.8297377824783325, | |
| "learning_rate": 0.0015, | |
| "loss": 2.7113, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.028481012658227847, | |
| "grad_norm": 0.9188632965087891, | |
| "learning_rate": 0.0015, | |
| "loss": 2.6782, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.029535864978902954, | |
| "grad_norm": 0.7633031606674194, | |
| "learning_rate": 0.0015, | |
| "loss": 2.6398, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.03059071729957806, | |
| "grad_norm": 0.9104907512664795, | |
| "learning_rate": 0.0015, | |
| "loss": 2.6154, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.03164556962025317, | |
| "grad_norm": 0.8751909732818604, | |
| "learning_rate": 0.0015, | |
| "loss": 2.5923, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03270042194092827, | |
| "grad_norm": 0.8580324053764343, | |
| "learning_rate": 0.0015, | |
| "loss": 2.5698, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.03375527426160337, | |
| "grad_norm": 0.9248947501182556, | |
| "learning_rate": 0.0015, | |
| "loss": 2.5412, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.03481012658227848, | |
| "grad_norm": 0.7492154836654663, | |
| "learning_rate": 0.0015, | |
| "loss": 2.5276, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.035864978902953586, | |
| "grad_norm": 0.8280085325241089, | |
| "learning_rate": 0.0015, | |
| "loss": 2.5025, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.03691983122362869, | |
| "grad_norm": 1.2397363185882568, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4612, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.0379746835443038, | |
| "grad_norm": 0.8337180614471436, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4443, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.039029535864978905, | |
| "grad_norm": 0.8829106092453003, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4333, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.04008438818565401, | |
| "grad_norm": 0.7179862856864929, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4126, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.04113924050632911, | |
| "grad_norm": 1.0350315570831299, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3976, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.04219409282700422, | |
| "grad_norm": 1.0989201068878174, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3746, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.043248945147679324, | |
| "grad_norm": 0.8930501937866211, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3621, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.04430379746835443, | |
| "grad_norm": 0.8837328553199768, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3439, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.04535864978902954, | |
| "grad_norm": 0.7845432162284851, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3147, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.046413502109704644, | |
| "grad_norm": 0.9344269037246704, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3019, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.04746835443037975, | |
| "grad_norm": 0.8818278908729553, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3035, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.04852320675105485, | |
| "grad_norm": 0.6915808916091919, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2788, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.049578059071729956, | |
| "grad_norm": 0.7943204045295715, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2654, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.05063291139240506, | |
| "grad_norm": 1.4367399215698242, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2561, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.05168776371308017, | |
| "grad_norm": 0.7440197467803955, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2299, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.052742616033755275, | |
| "grad_norm": 1.1008273363113403, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2256, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05379746835443038, | |
| "grad_norm": 1.3533852100372314, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2155, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.05485232067510549, | |
| "grad_norm": 0.7687584161758423, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1925, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.05590717299578059, | |
| "grad_norm": 0.7902509570121765, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1772, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.056962025316455694, | |
| "grad_norm": 1.0160045623779297, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1684, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.0580168776371308, | |
| "grad_norm": 0.7253097295761108, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1657, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.05907172995780591, | |
| "grad_norm": 0.730516791343689, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1272, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.060126582278481014, | |
| "grad_norm": 0.8411962389945984, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1316, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.06118143459915612, | |
| "grad_norm": 0.9713028073310852, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1361, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.06223628691983123, | |
| "grad_norm": 1.1778079271316528, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1091, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.06329113924050633, | |
| "grad_norm": 0.6379977464675903, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0929, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.06434599156118144, | |
| "grad_norm": 0.8067770004272461, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0945, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.06540084388185655, | |
| "grad_norm": 0.8583511114120483, | |
| "learning_rate": 0.0015, | |
| "loss": 2.082, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.06645569620253164, | |
| "grad_norm": 1.134140133857727, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0674, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.06751054852320675, | |
| "grad_norm": 0.8797304630279541, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0776, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.06856540084388185, | |
| "grad_norm": 1.2700276374816895, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0655, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.06962025316455696, | |
| "grad_norm": 0.8297256231307983, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0615, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.07067510548523206, | |
| "grad_norm": 0.8372782468795776, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0386, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.07172995780590717, | |
| "grad_norm": 0.7951764464378357, | |
| "learning_rate": 0.0015, | |
| "loss": 2.032, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.07278481012658228, | |
| "grad_norm": 0.766379177570343, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0374, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.07383966244725738, | |
| "grad_norm": 0.7371763586997986, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0164, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.07489451476793249, | |
| "grad_norm": 1.0791587829589844, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0091, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.0759493670886076, | |
| "grad_norm": 0.7908256649971008, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9936, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.0770042194092827, | |
| "grad_norm": 0.8527359962463379, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9844, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.07805907172995781, | |
| "grad_norm": 0.8309198021888733, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9956, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.07911392405063292, | |
| "grad_norm": 1.3767690658569336, | |
| "learning_rate": 0.0015, | |
| "loss": 1.978, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.08016877637130802, | |
| "grad_norm": 0.713165819644928, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9813, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.08122362869198312, | |
| "grad_norm": 0.8718229532241821, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9639, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.08227848101265822, | |
| "grad_norm": 0.8156098127365112, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9645, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.08333333333333333, | |
| "grad_norm": 0.9464802742004395, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9661, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.08438818565400844, | |
| "grad_norm": 0.7087975740432739, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9405, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.08544303797468354, | |
| "grad_norm": 0.6990103125572205, | |
| "learning_rate": 0.0015, | |
| "loss": 1.931, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.08649789029535865, | |
| "grad_norm": 1.6295113563537598, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9414, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.08755274261603375, | |
| "grad_norm": 0.7534579038619995, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9424, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.08860759493670886, | |
| "grad_norm": 0.9456313252449036, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9182, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.08966244725738397, | |
| "grad_norm": 0.770246148109436, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9216, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.09071729957805907, | |
| "grad_norm": 0.8989337682723999, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9174, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.09177215189873418, | |
| "grad_norm": 0.8580651879310608, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9165, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.09282700421940929, | |
| "grad_norm": 0.8543643951416016, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9084, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.0938818565400844, | |
| "grad_norm": 1.1457914113998413, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9017, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.0949367088607595, | |
| "grad_norm": 0.7464300394058228, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8976, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.09599156118143459, | |
| "grad_norm": 0.8035700917243958, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9015, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.0970464135021097, | |
| "grad_norm": 0.7010158896446228, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8904, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.0981012658227848, | |
| "grad_norm": 0.8178808093070984, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8712, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.09915611814345991, | |
| "grad_norm": 1.0780892372131348, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8812, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.10021097046413502, | |
| "grad_norm": 1.793686032295227, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8672, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.10126582278481013, | |
| "grad_norm": 1.1407757997512817, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8798, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.10232067510548523, | |
| "grad_norm": 1.027924656867981, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8681, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.10337552742616034, | |
| "grad_norm": 0.9612833261489868, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8474, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.10443037974683544, | |
| "grad_norm": 0.7732949256896973, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8529, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.10548523206751055, | |
| "grad_norm": 0.9289379119873047, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8605, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.10654008438818566, | |
| "grad_norm": 0.7720140814781189, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8526, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.10759493670886076, | |
| "grad_norm": 0.910430908203125, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8438, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.10864978902953587, | |
| "grad_norm": 0.7728487849235535, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8445, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.10970464135021098, | |
| "grad_norm": 0.8599476218223572, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8354, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.11075949367088607, | |
| "grad_norm": 1.2211686372756958, | |
| "learning_rate": 0.0015, | |
| "loss": 1.839, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.11181434599156118, | |
| "grad_norm": 0.7726693749427795, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8263, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.11286919831223628, | |
| "grad_norm": 0.8299482464790344, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8151, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.11392405063291139, | |
| "grad_norm": 0.6944489479064941, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8128, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.1149789029535865, | |
| "grad_norm": 0.671230673789978, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8235, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.1160337552742616, | |
| "grad_norm": 0.7142427563667297, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8081, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.11708860759493671, | |
| "grad_norm": 0.6354637145996094, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8143, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.11814345991561181, | |
| "grad_norm": 0.7774524092674255, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8084, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.11919831223628692, | |
| "grad_norm": 1.088733196258545, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7934, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.12025316455696203, | |
| "grad_norm": 0.82485032081604, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8074, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.12130801687763713, | |
| "grad_norm": 1.128315806388855, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7956, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.12236286919831224, | |
| "grad_norm": 0.7950155138969421, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8044, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.12341772151898735, | |
| "grad_norm": 0.7386237382888794, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7874, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.12447257383966245, | |
| "grad_norm": 0.7935805916786194, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7747, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.12552742616033755, | |
| "grad_norm": 0.7353824973106384, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7793, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.12658227848101267, | |
| "grad_norm": 1.8419713973999023, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7993, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.12763713080168776, | |
| "grad_norm": 0.7695147395133972, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7884, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.12869198312236288, | |
| "grad_norm": 0.6916271448135376, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7601, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.12974683544303797, | |
| "grad_norm": 0.6987491846084595, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7529, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.1308016877637131, | |
| "grad_norm": 0.7663790583610535, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7754, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.13185654008438819, | |
| "grad_norm": 0.7070717811584473, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7734, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.13291139240506328, | |
| "grad_norm": 0.7695779800415039, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7583, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.1339662447257384, | |
| "grad_norm": 0.8380961418151855, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7568, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.1350210970464135, | |
| "grad_norm": 0.741341233253479, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7734, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.1360759493670886, | |
| "grad_norm": 0.8968631625175476, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7575, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.1371308016877637, | |
| "grad_norm": 0.6770614385604858, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7471, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.13818565400843882, | |
| "grad_norm": 0.7884116172790527, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7506, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.13924050632911392, | |
| "grad_norm": 0.8187959790229797, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7519, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.14029535864978904, | |
| "grad_norm": 1.047227382659912, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7434, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.14135021097046413, | |
| "grad_norm": 0.729209303855896, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7457, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.14240506329113925, | |
| "grad_norm": 0.6964226961135864, | |
| "learning_rate": 0.0015, | |
| "loss": 1.749, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.14345991561181434, | |
| "grad_norm": 0.6814247965812683, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7365, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.14451476793248946, | |
| "grad_norm": 0.7576081156730652, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7334, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.14556962025316456, | |
| "grad_norm": 0.7002779841423035, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7256, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.14662447257383968, | |
| "grad_norm": 0.832123875617981, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7396, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.14767932489451477, | |
| "grad_norm": 0.670310378074646, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7194, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.14873417721518986, | |
| "grad_norm": 0.7912747263908386, | |
| "learning_rate": 0.0015, | |
| "loss": 1.724, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.14978902953586498, | |
| "grad_norm": 0.752036988735199, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7272, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.15084388185654007, | |
| "grad_norm": 0.8257012963294983, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7195, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.1518987341772152, | |
| "grad_norm": 0.6395012140274048, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7203, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.1529535864978903, | |
| "grad_norm": 1.1045129299163818, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7222, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.1540084388185654, | |
| "grad_norm": 0.6789548397064209, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7086, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.1550632911392405, | |
| "grad_norm": 0.8164582848548889, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7098, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.15611814345991562, | |
| "grad_norm": 0.6820360422134399, | |
| "learning_rate": 0.0015, | |
| "loss": 1.713, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.1571729957805907, | |
| "grad_norm": 0.6760832071304321, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7102, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.15822784810126583, | |
| "grad_norm": 0.7806017398834229, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7094, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.15928270042194093, | |
| "grad_norm": 0.8451957106590271, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7126, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.16033755274261605, | |
| "grad_norm": 0.7115667462348938, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7092, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.16139240506329114, | |
| "grad_norm": 0.6442527174949646, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6909, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.16244725738396623, | |
| "grad_norm": 0.6934787034988403, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6835, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.16350210970464135, | |
| "grad_norm": 0.8956286311149597, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6842, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.16455696202531644, | |
| "grad_norm": 0.6710405349731445, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6887, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.16561181434599156, | |
| "grad_norm": 0.6764975786209106, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6835, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.16666666666666666, | |
| "grad_norm": 0.6724684834480286, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6908, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.16772151898734178, | |
| "grad_norm": 0.859164834022522, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6862, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.16877637130801687, | |
| "grad_norm": 0.7298034429550171, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6919, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.169831223628692, | |
| "grad_norm": 0.6594526171684265, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6892, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.17088607594936708, | |
| "grad_norm": 0.6169994473457336, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6609, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.1719409282700422, | |
| "grad_norm": 0.7296561002731323, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6726, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.1729957805907173, | |
| "grad_norm": 0.6978626847267151, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6756, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.17405063291139242, | |
| "grad_norm": 1.0857014656066895, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6678, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.1751054852320675, | |
| "grad_norm": 0.6385810375213623, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6669, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.17616033755274263, | |
| "grad_norm": 0.9220226407051086, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6769, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.17721518987341772, | |
| "grad_norm": 1.4121848344802856, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6633, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.17827004219409281, | |
| "grad_norm": 0.7881584167480469, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6546, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.17932489451476794, | |
| "grad_norm": 0.6409121155738831, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6718, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.18037974683544303, | |
| "grad_norm": 0.8101775050163269, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6544, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.18143459915611815, | |
| "grad_norm": 0.6343249678611755, | |
| "learning_rate": 0.0015, | |
| "loss": 1.651, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.18248945147679324, | |
| "grad_norm": 0.7380338907241821, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6558, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.18354430379746836, | |
| "grad_norm": 0.7978950142860413, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6561, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.18459915611814345, | |
| "grad_norm": 0.9163487553596497, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6779, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.18565400843881857, | |
| "grad_norm": 0.6927811503410339, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6588, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.18670886075949367, | |
| "grad_norm": 0.6966419816017151, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6612, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.1877637130801688, | |
| "grad_norm": 0.6320984959602356, | |
| "learning_rate": 0.0015, | |
| "loss": 1.657, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.18881856540084388, | |
| "grad_norm": 0.6386429667472839, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6593, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.189873417721519, | |
| "grad_norm": 0.6478818655014038, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6558, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.1909282700421941, | |
| "grad_norm": 0.6771355271339417, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6437, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.19198312236286919, | |
| "grad_norm": 0.7388620376586914, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6406, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.1930379746835443, | |
| "grad_norm": 0.7418689131736755, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6384, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.1940928270042194, | |
| "grad_norm": 0.6168630123138428, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6368, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.19514767932489452, | |
| "grad_norm": 0.7633735537528992, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6434, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.1962025316455696, | |
| "grad_norm": 0.6480771899223328, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6476, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.19725738396624473, | |
| "grad_norm": 0.6744736433029175, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6479, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.19831223628691982, | |
| "grad_norm": 0.625220000743866, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6316, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.19936708860759494, | |
| "grad_norm": 0.636688768863678, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6384, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.20042194092827004, | |
| "grad_norm": 0.9341202974319458, | |
| "learning_rate": 0.0015, | |
| "loss": 1.637, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.20147679324894516, | |
| "grad_norm": 1.0301650762557983, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6372, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.20253164556962025, | |
| "grad_norm": 0.6672254800796509, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6312, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.20358649789029537, | |
| "grad_norm": 0.8981499075889587, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6329, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.20464135021097046, | |
| "grad_norm": 0.6990149021148682, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6252, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.20569620253164558, | |
| "grad_norm": 0.6157594919204712, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6239, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.20675105485232068, | |
| "grad_norm": 0.6597276329994202, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6248, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.20780590717299577, | |
| "grad_norm": 0.6546605825424194, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6354, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.2088607594936709, | |
| "grad_norm": 0.7146171927452087, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6207, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.20991561181434598, | |
| "grad_norm": 0.6842172741889954, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6175, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.2109704641350211, | |
| "grad_norm": 0.7051606774330139, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6161, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2120253164556962, | |
| "grad_norm": 1.0707893371582031, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6239, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.21308016877637131, | |
| "grad_norm": 0.7594018578529358, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6324, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.2141350210970464, | |
| "grad_norm": 0.6623409390449524, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6194, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.21518987341772153, | |
| "grad_norm": 0.6320188641548157, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6173, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.21624472573839662, | |
| "grad_norm": 0.756416380405426, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6166, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.21729957805907174, | |
| "grad_norm": 0.7982528805732727, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6063, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.21835443037974683, | |
| "grad_norm": 0.83033686876297, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6108, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.21940928270042195, | |
| "grad_norm": 0.7017702460289001, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6147, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.22046413502109705, | |
| "grad_norm": 0.8277395963668823, | |
| "learning_rate": 0.0015, | |
| "loss": 1.608, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.22151898734177214, | |
| "grad_norm": 0.8305484056472778, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6151, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.22257383966244726, | |
| "grad_norm": 0.8680278658866882, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6059, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.22362869198312235, | |
| "grad_norm": 0.6655151844024658, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6058, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.22468354430379747, | |
| "grad_norm": 0.7406314611434937, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6052, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.22573839662447256, | |
| "grad_norm": 0.590162992477417, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6005, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.22679324894514769, | |
| "grad_norm": 0.7729781866073608, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6058, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.22784810126582278, | |
| "grad_norm": 1.0980957746505737, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6208, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.2289029535864979, | |
| "grad_norm": 0.7007785439491272, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6039, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.229957805907173, | |
| "grad_norm": 0.7019352912902832, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5898, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.2310126582278481, | |
| "grad_norm": 0.6203840374946594, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6022, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.2320675105485232, | |
| "grad_norm": 0.6874725222587585, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6034, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.23312236286919832, | |
| "grad_norm": 0.5788426399230957, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5894, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.23417721518987342, | |
| "grad_norm": 0.7042095065116882, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6029, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.23523206751054854, | |
| "grad_norm": 0.9149144887924194, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5979, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.23628691983122363, | |
| "grad_norm": 0.8104167580604553, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5993, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.23734177215189872, | |
| "grad_norm": 1.0500192642211914, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5904, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.23839662447257384, | |
| "grad_norm": 1.0872136354446411, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5838, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.23945147679324894, | |
| "grad_norm": 0.855541467666626, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6158, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.24050632911392406, | |
| "grad_norm": 1.3494621515274048, | |
| "learning_rate": 0.0015, | |
| "loss": 1.6022, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.24156118143459915, | |
| "grad_norm": 0.9795144200325012, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5896, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.24261603375527427, | |
| "grad_norm": 0.6512905359268188, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5793, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.24367088607594936, | |
| "grad_norm": 0.6993908286094666, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5821, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.24472573839662448, | |
| "grad_norm": 0.6173799633979797, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5959, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.24578059071729957, | |
| "grad_norm": 0.7689816355705261, | |
| "learning_rate": 0.0015, | |
| "loss": 1.577, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.2468354430379747, | |
| "grad_norm": 0.8228141665458679, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5746, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.2478902953586498, | |
| "grad_norm": 0.6704361438751221, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5789, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.2489451476793249, | |
| "grad_norm": 0.8982261419296265, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5865, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 0.7105319499969482, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5854, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.2510548523206751, | |
| "grad_norm": 0.590654194355011, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5799, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.2521097046413502, | |
| "grad_norm": 0.60832679271698, | |
| "learning_rate": 0.0015, | |
| "loss": 1.582, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.25316455696202533, | |
| "grad_norm": 0.6442340612411499, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5801, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.2542194092827004, | |
| "grad_norm": 0.7528612017631531, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5829, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.2552742616033755, | |
| "grad_norm": 0.9050551652908325, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5761, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.2563291139240506, | |
| "grad_norm": 0.7395772933959961, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5739, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.25738396624472576, | |
| "grad_norm": 0.758362352848053, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5848, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.25843881856540085, | |
| "grad_norm": 0.5878888964653015, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5685, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.25949367088607594, | |
| "grad_norm": 0.6303932666778564, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5837, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.26054852320675104, | |
| "grad_norm": 1.1393781900405884, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5771, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.2616033755274262, | |
| "grad_norm": 0.7777864336967468, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5828, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.2626582278481013, | |
| "grad_norm": 0.5970431566238403, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5623, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.26371308016877637, | |
| "grad_norm": 0.8533217906951904, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5695, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.26476793248945146, | |
| "grad_norm": 0.5901464223861694, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5704, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 0.26582278481012656, | |
| "grad_norm": 0.7525666356086731, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5735, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.2668776371308017, | |
| "grad_norm": 0.6917023658752441, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5696, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 0.2679324894514768, | |
| "grad_norm": 0.7417186498641968, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5712, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.2689873417721519, | |
| "grad_norm": 0.6252340078353882, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5657, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.270042194092827, | |
| "grad_norm": 0.7001520395278931, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5735, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.27109704641350213, | |
| "grad_norm": 0.6428219676017761, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5655, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 0.2721518987341772, | |
| "grad_norm": 0.6188051104545593, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5654, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 0.2732067510548523, | |
| "grad_norm": 0.6137269735336304, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5704, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 0.2742616033755274, | |
| "grad_norm": 0.6634780764579773, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5707, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.27531645569620256, | |
| "grad_norm": 0.6164573431015015, | |
| "learning_rate": 0.0015, | |
| "loss": 1.567, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 0.27637130801687765, | |
| "grad_norm": 0.67591792345047, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5696, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 0.27742616033755274, | |
| "grad_norm": 0.9030578136444092, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5592, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 0.27848101265822783, | |
| "grad_norm": 0.6329770088195801, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5689, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.2795358649789029, | |
| "grad_norm": 0.7755404114723206, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5585, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.2805907172995781, | |
| "grad_norm": 0.6837880611419678, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5538, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 0.28164556962025317, | |
| "grad_norm": 0.6190932989120483, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5561, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 0.28270042194092826, | |
| "grad_norm": 0.6166747808456421, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5544, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 0.28375527426160335, | |
| "grad_norm": 0.68338543176651, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5552, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 0.2848101265822785, | |
| "grad_norm": 0.6845906376838684, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5486, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.2858649789029536, | |
| "grad_norm": 0.5682319402694702, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5566, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 0.2869198312236287, | |
| "grad_norm": 0.5855002999305725, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5574, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.2879746835443038, | |
| "grad_norm": 0.6031274199485779, | |
| "learning_rate": 0.0015, | |
| "loss": 1.54, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 0.2890295358649789, | |
| "grad_norm": 1.0569102764129639, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5523, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 0.290084388185654, | |
| "grad_norm": 0.6331525444984436, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5547, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.2911392405063291, | |
| "grad_norm": 0.7460474967956543, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5646, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.2921940928270042, | |
| "grad_norm": 0.6277437806129456, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5591, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 0.29324894514767935, | |
| "grad_norm": 0.6590365767478943, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5562, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 0.29430379746835444, | |
| "grad_norm": 0.6113288998603821, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5555, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 0.29535864978902954, | |
| "grad_norm": 0.5620890855789185, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5516, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.29641350210970463, | |
| "grad_norm": 0.6174807548522949, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5501, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 0.2974683544303797, | |
| "grad_norm": 0.6747241616249084, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5571, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 0.29852320675105487, | |
| "grad_norm": 0.9310271143913269, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5321, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 0.29957805907172996, | |
| "grad_norm": 1.3230280876159668, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5384, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 0.30063291139240506, | |
| "grad_norm": 1.2144172191619873, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5487, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.30168776371308015, | |
| "grad_norm": 0.6515137553215027, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5432, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 0.3027426160337553, | |
| "grad_norm": 0.7877812385559082, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5425, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 0.3037974683544304, | |
| "grad_norm": 0.6689189672470093, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5313, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.3048523206751055, | |
| "grad_norm": 0.6395917534828186, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5345, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 0.3059071729957806, | |
| "grad_norm": 0.7754723429679871, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5428, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.3069620253164557, | |
| "grad_norm": 0.7293116450309753, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5438, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 0.3080168776371308, | |
| "grad_norm": 1.249470829963684, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5401, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 0.3090717299578059, | |
| "grad_norm": 0.6346191763877869, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5335, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 0.310126582278481, | |
| "grad_norm": 0.6461648941040039, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5374, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.3111814345991561, | |
| "grad_norm": 0.7199176549911499, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5476, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.31223628691983124, | |
| "grad_norm": 0.7429535388946533, | |
| "learning_rate": 0.0015, | |
| "loss": 1.529, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 0.31329113924050633, | |
| "grad_norm": 0.9656321406364441, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5238, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 0.3143459915611814, | |
| "grad_norm": 1.1244347095489502, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5349, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 0.3154008438818565, | |
| "grad_norm": 0.7225420475006104, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5484, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 0.31645569620253167, | |
| "grad_norm": 0.6689170598983765, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5242, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.31751054852320676, | |
| "grad_norm": 0.7187021374702454, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5339, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 0.31856540084388185, | |
| "grad_norm": 0.6817131042480469, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5356, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 0.31962025316455694, | |
| "grad_norm": 0.7035455703735352, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5229, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 0.3206751054852321, | |
| "grad_norm": 0.7125651240348816, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5336, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 0.3217299578059072, | |
| "grad_norm": 0.6795884966850281, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5301, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.3227848101265823, | |
| "grad_norm": 0.6996033787727356, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5408, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 0.32383966244725737, | |
| "grad_norm": 0.5682607889175415, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5276, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 0.32489451476793246, | |
| "grad_norm": 0.5759983658790588, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5197, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 0.3259493670886076, | |
| "grad_norm": 0.582955539226532, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5363, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 0.3270042194092827, | |
| "grad_norm": 0.6314529180526733, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5232, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.3280590717299578, | |
| "grad_norm": 0.6088457703590393, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5263, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 0.3291139240506329, | |
| "grad_norm": 0.5606082677841187, | |
| "learning_rate": 0.0015, | |
| "loss": 1.526, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 0.33016877637130804, | |
| "grad_norm": 0.7216711044311523, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5379, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 0.33122362869198313, | |
| "grad_norm": 0.7543506622314453, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5192, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 0.3322784810126582, | |
| "grad_norm": 0.5943099856376648, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5272, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 0.60477215051651, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5282, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 0.33438818565400846, | |
| "grad_norm": 0.6148793697357178, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5293, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 0.33544303797468356, | |
| "grad_norm": 0.591863751411438, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5238, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 0.33649789029535865, | |
| "grad_norm": 0.6953649520874023, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5245, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 0.33755274261603374, | |
| "grad_norm": 0.74885493516922, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5195, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.33860759493670883, | |
| "grad_norm": 0.7037512063980103, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5322, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 0.339662447257384, | |
| "grad_norm": 0.725911021232605, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5245, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 0.3407172995780591, | |
| "grad_norm": 0.6004663705825806, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5144, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 0.34177215189873417, | |
| "grad_norm": 0.7108480930328369, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4998, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 0.34282700421940926, | |
| "grad_norm": 0.6768283843994141, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5249, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.3438818565400844, | |
| "grad_norm": 0.5562488436698914, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5131, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 0.3449367088607595, | |
| "grad_norm": 0.631421685218811, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5294, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 0.3459915611814346, | |
| "grad_norm": 0.7008868455886841, | |
| "learning_rate": 0.0015, | |
| "loss": 1.529, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 0.3470464135021097, | |
| "grad_norm": 0.775509238243103, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5112, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 0.34810126582278483, | |
| "grad_norm": 0.5601686835289001, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5219, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.3491561181434599, | |
| "grad_norm": 0.5643460750579834, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5197, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 0.350210970464135, | |
| "grad_norm": 0.5928805470466614, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5228, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 0.3512658227848101, | |
| "grad_norm": 0.5419443249702454, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5147, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 0.35232067510548526, | |
| "grad_norm": 0.7151007056236267, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5166, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 0.35337552742616035, | |
| "grad_norm": 0.5484126210212708, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5121, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.35443037974683544, | |
| "grad_norm": 0.6396426558494568, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5102, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 0.35548523206751054, | |
| "grad_norm": 0.6092063188552856, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5189, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 0.35654008438818563, | |
| "grad_norm": 0.6935619711875916, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5133, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 0.3575949367088608, | |
| "grad_norm": 0.551314115524292, | |
| "learning_rate": 0.0015, | |
| "loss": 1.505, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 0.35864978902953587, | |
| "grad_norm": 0.5668650269508362, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5166, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.35970464135021096, | |
| "grad_norm": 0.6797813177108765, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5137, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 0.36075949367088606, | |
| "grad_norm": 0.7667569518089294, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4977, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 0.3618143459915612, | |
| "grad_norm": 0.7580966353416443, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5138, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 0.3628691983122363, | |
| "grad_norm": 0.5931734442710876, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5105, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 0.3639240506329114, | |
| "grad_norm": 0.5610995292663574, | |
| "learning_rate": 0.0015, | |
| "loss": 1.505, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.3649789029535865, | |
| "grad_norm": 0.712026059627533, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4982, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 0.36603375527426163, | |
| "grad_norm": 0.9137971997261047, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5166, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 0.3670886075949367, | |
| "grad_norm": 0.764611005783081, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5218, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 0.3681434599156118, | |
| "grad_norm": 0.5769204497337341, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5108, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 0.3691983122362869, | |
| "grad_norm": 0.6009048223495483, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4939, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.370253164556962, | |
| "grad_norm": 0.7048670053482056, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5146, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 0.37130801687763715, | |
| "grad_norm": 0.6143278479576111, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5105, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 0.37236286919831224, | |
| "grad_norm": 0.5736284852027893, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5183, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 0.37341772151898733, | |
| "grad_norm": 0.741503894329071, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5123, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 0.3744725738396624, | |
| "grad_norm": 0.6641824245452881, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5028, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.3755274261603376, | |
| "grad_norm": 0.568838357925415, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5011, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 0.37658227848101267, | |
| "grad_norm": 0.7397016286849976, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4981, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 0.37763713080168776, | |
| "grad_norm": 0.5978692173957825, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5027, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 0.37869198312236285, | |
| "grad_norm": 0.5610095262527466, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5129, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 0.379746835443038, | |
| "grad_norm": 0.7330253720283508, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4975, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.3808016877637131, | |
| "grad_norm": 0.5927225351333618, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5004, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 0.3818565400843882, | |
| "grad_norm": 0.5715808868408203, | |
| "learning_rate": 0.0015, | |
| "loss": 1.503, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 0.3829113924050633, | |
| "grad_norm": 0.6043118834495544, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4984, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 0.38396624472573837, | |
| "grad_norm": 0.7284471392631531, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5008, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 0.3850210970464135, | |
| "grad_norm": 0.6941373944282532, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5129, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.3860759493670886, | |
| "grad_norm": 0.9151371717453003, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5016, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 0.3871308016877637, | |
| "grad_norm": 0.5350086092948914, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4956, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 0.3881856540084388, | |
| "grad_norm": 0.9535840749740601, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4972, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 0.38924050632911394, | |
| "grad_norm": 0.6288740038871765, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5085, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 0.39029535864978904, | |
| "grad_norm": 0.6044665575027466, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5114, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.39135021097046413, | |
| "grad_norm": 0.6169572472572327, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4977, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 0.3924050632911392, | |
| "grad_norm": 0.572669267654419, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4784, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 0.39345991561181437, | |
| "grad_norm": 0.5647141933441162, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4965, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 0.39451476793248946, | |
| "grad_norm": 0.604538083076477, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4967, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 0.39556962025316456, | |
| "grad_norm": 0.729308545589447, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5039, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.39662447257383965, | |
| "grad_norm": 0.9094111323356628, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4912, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 0.39767932489451474, | |
| "grad_norm": 0.6112377643585205, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5059, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 0.3987341772151899, | |
| "grad_norm": 0.5163501501083374, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4938, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 0.399789029535865, | |
| "grad_norm": 0.683885931968689, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4966, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 0.4008438818565401, | |
| "grad_norm": 0.5952377319335938, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5046, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.40189873417721517, | |
| "grad_norm": 0.6416290998458862, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4908, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 0.4029535864978903, | |
| "grad_norm": 0.6121982336044312, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4825, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 0.4040084388185654, | |
| "grad_norm": 0.6151083707809448, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4865, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 0.4050632911392405, | |
| "grad_norm": 0.7092777490615845, | |
| "learning_rate": 0.0015, | |
| "loss": 1.5029, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 0.4061181434599156, | |
| "grad_norm": 0.6845822334289551, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4841, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.40717299578059074, | |
| "grad_norm": 0.8743574619293213, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4983, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 0.40822784810126583, | |
| "grad_norm": 0.7992430925369263, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4876, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 0.4092827004219409, | |
| "grad_norm": 0.6235871315002441, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4798, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 0.410337552742616, | |
| "grad_norm": 0.6806468367576599, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4806, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 0.41139240506329117, | |
| "grad_norm": 1.147836446762085, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4805, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.41244725738396626, | |
| "grad_norm": 0.6263267993927002, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4856, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 0.41350210970464135, | |
| "grad_norm": 0.6188958883285522, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4809, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 0.41455696202531644, | |
| "grad_norm": 0.5184334516525269, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4862, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 0.41561181434599154, | |
| "grad_norm": 0.6880916357040405, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4813, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 0.4166666666666667, | |
| "grad_norm": 0.8529372811317444, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4879, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.4177215189873418, | |
| "grad_norm": 0.6398452520370483, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4903, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 0.41877637130801687, | |
| "grad_norm": 0.7702900171279907, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4788, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 0.41983122362869196, | |
| "grad_norm": 0.6239493489265442, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4908, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 0.4208860759493671, | |
| "grad_norm": 0.6391934752464294, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4797, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 0.4219409282700422, | |
| "grad_norm": 0.7288190722465515, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4941, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.4229957805907173, | |
| "grad_norm": 0.5728703737258911, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4841, | |
| "step": 4010 | |
| }, | |
| { | |
| "epoch": 0.4240506329113924, | |
| "grad_norm": 0.5602556467056274, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4835, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 0.42510548523206754, | |
| "grad_norm": 0.7178170084953308, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4751, | |
| "step": 4030 | |
| }, | |
| { | |
| "epoch": 0.42616033755274263, | |
| "grad_norm": 0.5581310391426086, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4853, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 0.4272151898734177, | |
| "grad_norm": 0.5198979377746582, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4871, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.4282700421940928, | |
| "grad_norm": 0.6657981276512146, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4923, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 0.4293248945147679, | |
| "grad_norm": 0.765648603439331, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4845, | |
| "step": 4070 | |
| }, | |
| { | |
| "epoch": 0.43037974683544306, | |
| "grad_norm": 0.6785349249839783, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4892, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 0.43143459915611815, | |
| "grad_norm": 0.5718680620193481, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4837, | |
| "step": 4090 | |
| }, | |
| { | |
| "epoch": 0.43248945147679324, | |
| "grad_norm": 0.553119421005249, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4814, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.43354430379746833, | |
| "grad_norm": 0.5533855557441711, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4721, | |
| "step": 4110 | |
| }, | |
| { | |
| "epoch": 0.4345991561181435, | |
| "grad_norm": 0.6695696115493774, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4782, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 0.4356540084388186, | |
| "grad_norm": 0.5840980410575867, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4717, | |
| "step": 4130 | |
| }, | |
| { | |
| "epoch": 0.43670886075949367, | |
| "grad_norm": 0.6174877285957336, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4761, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 0.43776371308016876, | |
| "grad_norm": 0.6185278296470642, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4792, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.4388185654008439, | |
| "grad_norm": 0.5936148762702942, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4735, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 0.439873417721519, | |
| "grad_norm": 0.6119903326034546, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4731, | |
| "step": 4170 | |
| }, | |
| { | |
| "epoch": 0.4409282700421941, | |
| "grad_norm": 0.5228918194770813, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4704, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 0.4419831223628692, | |
| "grad_norm": 0.6289933323860168, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4762, | |
| "step": 4190 | |
| }, | |
| { | |
| "epoch": 0.4430379746835443, | |
| "grad_norm": 0.6278483271598816, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4773, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.4440928270042194, | |
| "grad_norm": 0.7192302942276001, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4685, | |
| "step": 4210 | |
| }, | |
| { | |
| "epoch": 0.4451476793248945, | |
| "grad_norm": 0.7284325957298279, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4787, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 0.4462025316455696, | |
| "grad_norm": 0.8943430781364441, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4741, | |
| "step": 4230 | |
| }, | |
| { | |
| "epoch": 0.4472573839662447, | |
| "grad_norm": 0.6659002304077148, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4737, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 0.44831223628691985, | |
| "grad_norm": 0.642861008644104, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4774, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.44936708860759494, | |
| "grad_norm": 0.7793336510658264, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4825, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 0.45042194092827004, | |
| "grad_norm": 0.5228441953659058, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4727, | |
| "step": 4270 | |
| }, | |
| { | |
| "epoch": 0.45147679324894513, | |
| "grad_norm": 0.9414927959442139, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4664, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 0.4525316455696203, | |
| "grad_norm": 0.8323007822036743, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4782, | |
| "step": 4290 | |
| }, | |
| { | |
| "epoch": 0.45358649789029537, | |
| "grad_norm": 0.9512622356414795, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4791, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.45464135021097046, | |
| "grad_norm": 0.5568503141403198, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4762, | |
| "step": 4310 | |
| }, | |
| { | |
| "epoch": 0.45569620253164556, | |
| "grad_norm": 0.8442806005477905, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4698, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 0.45675105485232065, | |
| "grad_norm": 0.6293421387672424, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4765, | |
| "step": 4330 | |
| }, | |
| { | |
| "epoch": 0.4578059071729958, | |
| "grad_norm": 0.7461580634117126, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4684, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 0.4588607594936709, | |
| "grad_norm": 0.8717284798622131, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4717, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.459915611814346, | |
| "grad_norm": 0.5873362421989441, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4875, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 0.4609704641350211, | |
| "grad_norm": 0.5259530544281006, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4619, | |
| "step": 4370 | |
| }, | |
| { | |
| "epoch": 0.4620253164556962, | |
| "grad_norm": 0.5751098394393921, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4669, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 0.4630801687763713, | |
| "grad_norm": 0.5525999069213867, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4649, | |
| "step": 4390 | |
| }, | |
| { | |
| "epoch": 0.4641350210970464, | |
| "grad_norm": 0.579777717590332, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4694, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.4651898734177215, | |
| "grad_norm": 0.6130363941192627, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4834, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.46624472573839665, | |
| "grad_norm": 0.6964607834815979, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4687, | |
| "step": 4420 | |
| }, | |
| { | |
| "epoch": 0.46729957805907174, | |
| "grad_norm": 0.5204871892929077, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4713, | |
| "step": 4430 | |
| }, | |
| { | |
| "epoch": 0.46835443037974683, | |
| "grad_norm": 0.801157534122467, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4676, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 0.4694092827004219, | |
| "grad_norm": 0.7642245292663574, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4765, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.4704641350210971, | |
| "grad_norm": 0.5725940465927124, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4615, | |
| "step": 4460 | |
| }, | |
| { | |
| "epoch": 0.47151898734177217, | |
| "grad_norm": 0.5430575609207153, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4712, | |
| "step": 4470 | |
| }, | |
| { | |
| "epoch": 0.47257383966244726, | |
| "grad_norm": 0.6782746911048889, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4767, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 0.47362869198312235, | |
| "grad_norm": 0.9323136806488037, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4633, | |
| "step": 4490 | |
| }, | |
| { | |
| "epoch": 0.47468354430379744, | |
| "grad_norm": 0.6096838116645813, | |
| "learning_rate": 0.0015, | |
| "loss": 1.464, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.4757383966244726, | |
| "grad_norm": 0.5575011372566223, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4688, | |
| "step": 4510 | |
| }, | |
| { | |
| "epoch": 0.4767932489451477, | |
| "grad_norm": 0.6049365401268005, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4522, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 0.4778481012658228, | |
| "grad_norm": 0.6802952289581299, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4695, | |
| "step": 4530 | |
| }, | |
| { | |
| "epoch": 0.47890295358649787, | |
| "grad_norm": 0.8433916568756104, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4752, | |
| "step": 4540 | |
| }, | |
| { | |
| "epoch": 0.479957805907173, | |
| "grad_norm": 0.8293811082839966, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4695, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.4810126582278481, | |
| "grad_norm": 0.7346205115318298, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4581, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 0.4820675105485232, | |
| "grad_norm": 0.570183515548706, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4408, | |
| "step": 4570 | |
| }, | |
| { | |
| "epoch": 0.4831223628691983, | |
| "grad_norm": 0.7086605429649353, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4648, | |
| "step": 4580 | |
| }, | |
| { | |
| "epoch": 0.48417721518987344, | |
| "grad_norm": 0.619171142578125, | |
| "learning_rate": 0.0015, | |
| "loss": 1.452, | |
| "step": 4590 | |
| }, | |
| { | |
| "epoch": 0.48523206751054854, | |
| "grad_norm": 0.6127137541770935, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4723, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.48628691983122363, | |
| "grad_norm": 0.734324038028717, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4763, | |
| "step": 4610 | |
| }, | |
| { | |
| "epoch": 0.4873417721518987, | |
| "grad_norm": 0.7030718922615051, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4489, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 0.4883966244725738, | |
| "grad_norm": 0.5352341532707214, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4678, | |
| "step": 4630 | |
| }, | |
| { | |
| "epoch": 0.48945147679324896, | |
| "grad_norm": 0.6763170957565308, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4677, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 0.49050632911392406, | |
| "grad_norm": 0.6143730282783508, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4707, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.49156118143459915, | |
| "grad_norm": 0.5258755087852478, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4559, | |
| "step": 4660 | |
| }, | |
| { | |
| "epoch": 0.49261603375527424, | |
| "grad_norm": 0.67002272605896, | |
| "learning_rate": 0.0015, | |
| "loss": 1.456, | |
| "step": 4670 | |
| }, | |
| { | |
| "epoch": 0.4936708860759494, | |
| "grad_norm": 0.5338280200958252, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4567, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 0.4947257383966245, | |
| "grad_norm": 0.6864109635353088, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4524, | |
| "step": 4690 | |
| }, | |
| { | |
| "epoch": 0.4957805907172996, | |
| "grad_norm": 0.5267427563667297, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4605, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.49683544303797467, | |
| "grad_norm": 0.5892050862312317, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4631, | |
| "step": 4710 | |
| }, | |
| { | |
| "epoch": 0.4978902953586498, | |
| "grad_norm": 0.7090893983840942, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4585, | |
| "step": 4720 | |
| }, | |
| { | |
| "epoch": 0.4989451476793249, | |
| "grad_norm": 0.5960710048675537, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4561, | |
| "step": 4730 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.6754327416419983, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4544, | |
| "step": 4740 | |
| }, | |
| { | |
| "epoch": 0.5010548523206751, | |
| "grad_norm": 0.6936537623405457, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4756, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.5021097046413502, | |
| "grad_norm": 0.5348150730133057, | |
| "learning_rate": 0.0015, | |
| "loss": 1.459, | |
| "step": 4760 | |
| }, | |
| { | |
| "epoch": 0.5031645569620253, | |
| "grad_norm": 0.6094486117362976, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4567, | |
| "step": 4770 | |
| }, | |
| { | |
| "epoch": 0.5042194092827004, | |
| "grad_norm": 0.6765019297599792, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4678, | |
| "step": 4780 | |
| }, | |
| { | |
| "epoch": 0.5052742616033755, | |
| "grad_norm": 0.5961644053459167, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4503, | |
| "step": 4790 | |
| }, | |
| { | |
| "epoch": 0.5063291139240507, | |
| "grad_norm": 0.6130753755569458, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4536, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.5073839662447257, | |
| "grad_norm": 0.5550441145896912, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4551, | |
| "step": 4810 | |
| }, | |
| { | |
| "epoch": 0.5084388185654009, | |
| "grad_norm": 0.6500513553619385, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4735, | |
| "step": 4820 | |
| }, | |
| { | |
| "epoch": 0.509493670886076, | |
| "grad_norm": 0.7694984674453735, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4563, | |
| "step": 4830 | |
| }, | |
| { | |
| "epoch": 0.510548523206751, | |
| "grad_norm": 0.6021876335144043, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4505, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 0.5116033755274262, | |
| "grad_norm": 0.5139393210411072, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4624, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.5126582278481012, | |
| "grad_norm": 0.610668957233429, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4502, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 0.5137130801687764, | |
| "grad_norm": 0.8146686553955078, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4455, | |
| "step": 4870 | |
| }, | |
| { | |
| "epoch": 0.5147679324894515, | |
| "grad_norm": 0.706887423992157, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4652, | |
| "step": 4880 | |
| }, | |
| { | |
| "epoch": 0.5158227848101266, | |
| "grad_norm": 0.6309559941291809, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4552, | |
| "step": 4890 | |
| }, | |
| { | |
| "epoch": 0.5168776371308017, | |
| "grad_norm": 0.6033481955528259, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4637, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.5179324894514767, | |
| "grad_norm": 0.5887914896011353, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4594, | |
| "step": 4910 | |
| }, | |
| { | |
| "epoch": 0.5189873417721519, | |
| "grad_norm": 0.5958083271980286, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4467, | |
| "step": 4920 | |
| }, | |
| { | |
| "epoch": 0.520042194092827, | |
| "grad_norm": 0.6756103038787842, | |
| "learning_rate": 0.0015, | |
| "loss": 1.457, | |
| "step": 4930 | |
| }, | |
| { | |
| "epoch": 0.5210970464135021, | |
| "grad_norm": 0.5370129346847534, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4581, | |
| "step": 4940 | |
| }, | |
| { | |
| "epoch": 0.5221518987341772, | |
| "grad_norm": 0.6108761429786682, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4629, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.5232067510548524, | |
| "grad_norm": 0.8404567241668701, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4434, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 0.5242616033755274, | |
| "grad_norm": 0.5708929300308228, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4523, | |
| "step": 4970 | |
| }, | |
| { | |
| "epoch": 0.5253164556962026, | |
| "grad_norm": 0.5647048354148865, | |
| "learning_rate": 0.0015, | |
| "loss": 1.446, | |
| "step": 4980 | |
| }, | |
| { | |
| "epoch": 0.5263713080168776, | |
| "grad_norm": 0.5417590141296387, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4491, | |
| "step": 4990 | |
| }, | |
| { | |
| "epoch": 0.5274261603375527, | |
| "grad_norm": 0.7352495193481445, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4607, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.5284810126582279, | |
| "grad_norm": 0.7228708267211914, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4534, | |
| "step": 5010 | |
| }, | |
| { | |
| "epoch": 0.5295358649789029, | |
| "grad_norm": 0.6637216210365295, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4468, | |
| "step": 5020 | |
| }, | |
| { | |
| "epoch": 0.5305907172995781, | |
| "grad_norm": 0.5074453353881836, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4306, | |
| "step": 5030 | |
| }, | |
| { | |
| "epoch": 0.5316455696202531, | |
| "grad_norm": 0.5190367698669434, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4592, | |
| "step": 5040 | |
| }, | |
| { | |
| "epoch": 0.5327004219409283, | |
| "grad_norm": 0.5237929224967957, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4468, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 0.5337552742616034, | |
| "grad_norm": 0.5514045357704163, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4597, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 0.5348101265822784, | |
| "grad_norm": 0.7031023502349854, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4477, | |
| "step": 5070 | |
| }, | |
| { | |
| "epoch": 0.5358649789029536, | |
| "grad_norm": 0.5926287174224854, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4608, | |
| "step": 5080 | |
| }, | |
| { | |
| "epoch": 0.5369198312236287, | |
| "grad_norm": 0.5033077001571655, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4493, | |
| "step": 5090 | |
| }, | |
| { | |
| "epoch": 0.5379746835443038, | |
| "grad_norm": 0.5126532316207886, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4342, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.5390295358649789, | |
| "grad_norm": 0.6754027009010315, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4507, | |
| "step": 5110 | |
| }, | |
| { | |
| "epoch": 0.540084388185654, | |
| "grad_norm": 0.6079409718513489, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4433, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.5411392405063291, | |
| "grad_norm": 0.5625218749046326, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4438, | |
| "step": 5130 | |
| }, | |
| { | |
| "epoch": 0.5421940928270043, | |
| "grad_norm": 0.8679294586181641, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4604, | |
| "step": 5140 | |
| }, | |
| { | |
| "epoch": 0.5432489451476793, | |
| "grad_norm": 0.5662575364112854, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4583, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 0.5443037974683544, | |
| "grad_norm": 0.5976355671882629, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4506, | |
| "step": 5160 | |
| }, | |
| { | |
| "epoch": 0.5453586497890295, | |
| "grad_norm": 0.5095018148422241, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4506, | |
| "step": 5170 | |
| }, | |
| { | |
| "epoch": 0.5464135021097046, | |
| "grad_norm": 0.8036375045776367, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4592, | |
| "step": 5180 | |
| }, | |
| { | |
| "epoch": 0.5474683544303798, | |
| "grad_norm": 0.5669030547142029, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4434, | |
| "step": 5190 | |
| }, | |
| { | |
| "epoch": 0.5485232067510548, | |
| "grad_norm": 0.7651669979095459, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4566, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.54957805907173, | |
| "grad_norm": 0.5880611538887024, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4392, | |
| "step": 5210 | |
| }, | |
| { | |
| "epoch": 0.5506329113924051, | |
| "grad_norm": 0.6016876697540283, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4502, | |
| "step": 5220 | |
| }, | |
| { | |
| "epoch": 0.5516877637130801, | |
| "grad_norm": 0.5552685856819153, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4418, | |
| "step": 5230 | |
| }, | |
| { | |
| "epoch": 0.5527426160337553, | |
| "grad_norm": 0.68308025598526, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4491, | |
| "step": 5240 | |
| }, | |
| { | |
| "epoch": 0.5537974683544303, | |
| "grad_norm": 0.6933003067970276, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4413, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.5548523206751055, | |
| "grad_norm": 0.5555622577667236, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4483, | |
| "step": 5260 | |
| }, | |
| { | |
| "epoch": 0.5559071729957806, | |
| "grad_norm": 0.6394757032394409, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4559, | |
| "step": 5270 | |
| }, | |
| { | |
| "epoch": 0.5569620253164557, | |
| "grad_norm": 0.704420268535614, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4484, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 0.5580168776371308, | |
| "grad_norm": 0.5235179662704468, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4487, | |
| "step": 5290 | |
| }, | |
| { | |
| "epoch": 0.5590717299578059, | |
| "grad_norm": 0.610462486743927, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4505, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.560126582278481, | |
| "grad_norm": 0.7268403172492981, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4379, | |
| "step": 5310 | |
| }, | |
| { | |
| "epoch": 0.5611814345991561, | |
| "grad_norm": 0.6289941668510437, | |
| "learning_rate": 0.0015, | |
| "loss": 1.439, | |
| "step": 5320 | |
| }, | |
| { | |
| "epoch": 0.5622362869198312, | |
| "grad_norm": 0.6171261072158813, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4493, | |
| "step": 5330 | |
| }, | |
| { | |
| "epoch": 0.5632911392405063, | |
| "grad_norm": 0.5034967064857483, | |
| "learning_rate": 0.0015, | |
| "loss": 1.442, | |
| "step": 5340 | |
| }, | |
| { | |
| "epoch": 0.5643459915611815, | |
| "grad_norm": 0.5746512413024902, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4342, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 0.5654008438818565, | |
| "grad_norm": 0.8130653500556946, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4459, | |
| "step": 5360 | |
| }, | |
| { | |
| "epoch": 0.5664556962025317, | |
| "grad_norm": 0.7359591722488403, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4428, | |
| "step": 5370 | |
| }, | |
| { | |
| "epoch": 0.5675105485232067, | |
| "grad_norm": 0.543518602848053, | |
| "learning_rate": 0.0015, | |
| "loss": 1.438, | |
| "step": 5380 | |
| }, | |
| { | |
| "epoch": 0.5685654008438819, | |
| "grad_norm": 0.5612589716911316, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4356, | |
| "step": 5390 | |
| }, | |
| { | |
| "epoch": 0.569620253164557, | |
| "grad_norm": 0.678749144077301, | |
| "learning_rate": 0.0015, | |
| "loss": 1.437, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.570675105485232, | |
| "grad_norm": 0.5381031036376953, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4393, | |
| "step": 5410 | |
| }, | |
| { | |
| "epoch": 0.5717299578059072, | |
| "grad_norm": 0.5442678928375244, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4348, | |
| "step": 5420 | |
| }, | |
| { | |
| "epoch": 0.5727848101265823, | |
| "grad_norm": 0.643040657043457, | |
| "learning_rate": 0.0015, | |
| "loss": 1.43, | |
| "step": 5430 | |
| }, | |
| { | |
| "epoch": 0.5738396624472574, | |
| "grad_norm": 0.5506795048713684, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4304, | |
| "step": 5440 | |
| }, | |
| { | |
| "epoch": 0.5748945147679325, | |
| "grad_norm": 0.6098572015762329, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4348, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 0.5759493670886076, | |
| "grad_norm": 0.6758444905281067, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4392, | |
| "step": 5460 | |
| }, | |
| { | |
| "epoch": 0.5770042194092827, | |
| "grad_norm": 0.5184838771820068, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4396, | |
| "step": 5470 | |
| }, | |
| { | |
| "epoch": 0.5780590717299579, | |
| "grad_norm": 0.9065413475036621, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4399, | |
| "step": 5480 | |
| }, | |
| { | |
| "epoch": 0.5791139240506329, | |
| "grad_norm": 0.5394531488418579, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4473, | |
| "step": 5490 | |
| }, | |
| { | |
| "epoch": 0.580168776371308, | |
| "grad_norm": 0.6227542757987976, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4391, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.5812236286919831, | |
| "grad_norm": 0.8134665489196777, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4447, | |
| "step": 5510 | |
| }, | |
| { | |
| "epoch": 0.5822784810126582, | |
| "grad_norm": 0.5177081823348999, | |
| "learning_rate": 0.0015, | |
| "loss": 1.436, | |
| "step": 5520 | |
| }, | |
| { | |
| "epoch": 0.5833333333333334, | |
| "grad_norm": 0.6241568326950073, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4298, | |
| "step": 5530 | |
| }, | |
| { | |
| "epoch": 0.5843881856540084, | |
| "grad_norm": 0.5243352651596069, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4286, | |
| "step": 5540 | |
| }, | |
| { | |
| "epoch": 0.5854430379746836, | |
| "grad_norm": 0.5504492521286011, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4453, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 0.5864978902953587, | |
| "grad_norm": 0.5448554158210754, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4487, | |
| "step": 5560 | |
| }, | |
| { | |
| "epoch": 0.5875527426160337, | |
| "grad_norm": 0.5740591287612915, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4324, | |
| "step": 5570 | |
| }, | |
| { | |
| "epoch": 0.5886075949367089, | |
| "grad_norm": 0.7025555968284607, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4334, | |
| "step": 5580 | |
| }, | |
| { | |
| "epoch": 0.5896624472573839, | |
| "grad_norm": 0.5911626815795898, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4366, | |
| "step": 5590 | |
| }, | |
| { | |
| "epoch": 0.5907172995780591, | |
| "grad_norm": 0.503633439540863, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4313, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.5917721518987342, | |
| "grad_norm": 0.6805056929588318, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4384, | |
| "step": 5610 | |
| }, | |
| { | |
| "epoch": 0.5928270042194093, | |
| "grad_norm": 0.5583463907241821, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4463, | |
| "step": 5620 | |
| }, | |
| { | |
| "epoch": 0.5938818565400844, | |
| "grad_norm": 0.6833731532096863, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4413, | |
| "step": 5630 | |
| }, | |
| { | |
| "epoch": 0.5949367088607594, | |
| "grad_norm": 0.5771982073783875, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4511, | |
| "step": 5640 | |
| }, | |
| { | |
| "epoch": 0.5959915611814346, | |
| "grad_norm": 0.6087093353271484, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4457, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 0.5970464135021097, | |
| "grad_norm": 0.7789549231529236, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4377, | |
| "step": 5660 | |
| }, | |
| { | |
| "epoch": 0.5981012658227848, | |
| "grad_norm": 0.6058619022369385, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4469, | |
| "step": 5670 | |
| }, | |
| { | |
| "epoch": 0.5991561181434599, | |
| "grad_norm": 0.5375464558601379, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4455, | |
| "step": 5680 | |
| }, | |
| { | |
| "epoch": 0.6002109704641351, | |
| "grad_norm": 0.8218992352485657, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4239, | |
| "step": 5690 | |
| }, | |
| { | |
| "epoch": 0.6012658227848101, | |
| "grad_norm": 0.5948262810707092, | |
| "learning_rate": 0.0015, | |
| "loss": 1.43, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.6023206751054853, | |
| "grad_norm": 0.5964241623878479, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4337, | |
| "step": 5710 | |
| }, | |
| { | |
| "epoch": 0.6033755274261603, | |
| "grad_norm": 0.5453081727027893, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4377, | |
| "step": 5720 | |
| }, | |
| { | |
| "epoch": 0.6044303797468354, | |
| "grad_norm": 0.5510696768760681, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4355, | |
| "step": 5730 | |
| }, | |
| { | |
| "epoch": 0.6054852320675106, | |
| "grad_norm": 0.5511655211448669, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4331, | |
| "step": 5740 | |
| }, | |
| { | |
| "epoch": 0.6065400843881856, | |
| "grad_norm": 0.6737241744995117, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4107, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.6075949367088608, | |
| "grad_norm": 0.5429701805114746, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4199, | |
| "step": 5760 | |
| }, | |
| { | |
| "epoch": 0.6086497890295358, | |
| "grad_norm": 0.7445785403251648, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4221, | |
| "step": 5770 | |
| }, | |
| { | |
| "epoch": 0.609704641350211, | |
| "grad_norm": 0.5367471575737, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4298, | |
| "step": 5780 | |
| }, | |
| { | |
| "epoch": 0.6107594936708861, | |
| "grad_norm": 0.6427586674690247, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4255, | |
| "step": 5790 | |
| }, | |
| { | |
| "epoch": 0.6118143459915611, | |
| "grad_norm": 0.6622743606567383, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4321, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.6128691983122363, | |
| "grad_norm": 0.5611438751220703, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4336, | |
| "step": 5810 | |
| }, | |
| { | |
| "epoch": 0.6139240506329114, | |
| "grad_norm": 0.5816289782524109, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4464, | |
| "step": 5820 | |
| }, | |
| { | |
| "epoch": 0.6149789029535865, | |
| "grad_norm": 0.5909755825996399, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4297, | |
| "step": 5830 | |
| }, | |
| { | |
| "epoch": 0.6160337552742616, | |
| "grad_norm": 0.7232508659362793, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4431, | |
| "step": 5840 | |
| }, | |
| { | |
| "epoch": 0.6170886075949367, | |
| "grad_norm": 0.8321460485458374, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4451, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 0.6181434599156118, | |
| "grad_norm": 0.5028564929962158, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4327, | |
| "step": 5860 | |
| }, | |
| { | |
| "epoch": 0.619198312236287, | |
| "grad_norm": 0.7685743570327759, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4256, | |
| "step": 5870 | |
| }, | |
| { | |
| "epoch": 0.620253164556962, | |
| "grad_norm": 0.5092529058456421, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4391, | |
| "step": 5880 | |
| }, | |
| { | |
| "epoch": 0.6213080168776371, | |
| "grad_norm": 0.5273023247718811, | |
| "learning_rate": 0.0015, | |
| "loss": 1.428, | |
| "step": 5890 | |
| }, | |
| { | |
| "epoch": 0.6223628691983122, | |
| "grad_norm": 0.5593839883804321, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4306, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.6234177215189873, | |
| "grad_norm": 0.5306974053382874, | |
| "learning_rate": 0.0015, | |
| "loss": 1.425, | |
| "step": 5910 | |
| }, | |
| { | |
| "epoch": 0.6244725738396625, | |
| "grad_norm": 0.5432724356651306, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4189, | |
| "step": 5920 | |
| }, | |
| { | |
| "epoch": 0.6255274261603375, | |
| "grad_norm": 1.200668215751648, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4307, | |
| "step": 5930 | |
| }, | |
| { | |
| "epoch": 0.6265822784810127, | |
| "grad_norm": 0.6720802187919617, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4346, | |
| "step": 5940 | |
| }, | |
| { | |
| "epoch": 0.6276371308016878, | |
| "grad_norm": 0.5483372807502747, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4231, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 0.6286919831223629, | |
| "grad_norm": 0.8048787117004395, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4273, | |
| "step": 5960 | |
| }, | |
| { | |
| "epoch": 0.629746835443038, | |
| "grad_norm": 0.5510703325271606, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4342, | |
| "step": 5970 | |
| }, | |
| { | |
| "epoch": 0.630801687763713, | |
| "grad_norm": 0.5123088955879211, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4146, | |
| "step": 5980 | |
| }, | |
| { | |
| "epoch": 0.6318565400843882, | |
| "grad_norm": 0.6412525177001953, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4331, | |
| "step": 5990 | |
| }, | |
| { | |
| "epoch": 0.6329113924050633, | |
| "grad_norm": 0.5405699610710144, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4421, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.6339662447257384, | |
| "grad_norm": 0.5397794842720032, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4293, | |
| "step": 6010 | |
| }, | |
| { | |
| "epoch": 0.6350210970464135, | |
| "grad_norm": 0.6345342397689819, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4332, | |
| "step": 6020 | |
| }, | |
| { | |
| "epoch": 0.6360759493670886, | |
| "grad_norm": 0.5723614692687988, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4158, | |
| "step": 6030 | |
| }, | |
| { | |
| "epoch": 0.6371308016877637, | |
| "grad_norm": 0.5299696326255798, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4341, | |
| "step": 6040 | |
| }, | |
| { | |
| "epoch": 0.6381856540084389, | |
| "grad_norm": 0.7609160542488098, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4174, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 0.6392405063291139, | |
| "grad_norm": 0.5781378746032715, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4343, | |
| "step": 6060 | |
| }, | |
| { | |
| "epoch": 0.640295358649789, | |
| "grad_norm": 0.7221751809120178, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4247, | |
| "step": 6070 | |
| }, | |
| { | |
| "epoch": 0.6413502109704642, | |
| "grad_norm": 0.49339455366134644, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4251, | |
| "step": 6080 | |
| }, | |
| { | |
| "epoch": 0.6424050632911392, | |
| "grad_norm": 0.5464122295379639, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4235, | |
| "step": 6090 | |
| }, | |
| { | |
| "epoch": 0.6434599156118144, | |
| "grad_norm": 0.6804472804069519, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4191, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.6445147679324894, | |
| "grad_norm": 0.6100871562957764, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4308, | |
| "step": 6110 | |
| }, | |
| { | |
| "epoch": 0.6455696202531646, | |
| "grad_norm": 0.49249380826950073, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4239, | |
| "step": 6120 | |
| }, | |
| { | |
| "epoch": 0.6466244725738397, | |
| "grad_norm": 0.5322859883308411, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4243, | |
| "step": 6130 | |
| }, | |
| { | |
| "epoch": 0.6476793248945147, | |
| "grad_norm": 0.6617424488067627, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4211, | |
| "step": 6140 | |
| }, | |
| { | |
| "epoch": 0.6487341772151899, | |
| "grad_norm": 0.5216740369796753, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4342, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 0.6497890295358649, | |
| "grad_norm": 0.6193856596946716, | |
| "learning_rate": 0.0015, | |
| "loss": 1.4267, | |
| "step": 6160 | |
| }, | |
| { | |
| "epoch": 0.6508438818565401, | |
| "grad_norm": 0.5639455318450928, | |
| "learning_rate": 0.0014834368975312174, | |
| "loss": 1.4098, | |
| "step": 6170 | |
| }, | |
| { | |
| "epoch": 0.6518987341772152, | |
| "grad_norm": 0.5887807011604309, | |
| "learning_rate": 0.0014629899726345957, | |
| "loss": 1.4291, | |
| "step": 6180 | |
| }, | |
| { | |
| "epoch": 0.6529535864978903, | |
| "grad_norm": 0.5052474737167358, | |
| "learning_rate": 0.0014428248775471316, | |
| "loss": 1.43, | |
| "step": 6190 | |
| }, | |
| { | |
| "epoch": 0.6540084388185654, | |
| "grad_norm": 0.5447661280632019, | |
| "learning_rate": 0.00142293772767289, | |
| "loss": 1.4228, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.6550632911392406, | |
| "grad_norm": 0.5228587985038757, | |
| "learning_rate": 0.001403324691959192, | |
| "loss": 1.4163, | |
| "step": 6210 | |
| }, | |
| { | |
| "epoch": 0.6561181434599156, | |
| "grad_norm": 0.5889528393745422, | |
| "learning_rate": 0.0013839819921586025, | |
| "loss": 1.4231, | |
| "step": 6220 | |
| }, | |
| { | |
| "epoch": 0.6571729957805907, | |
| "grad_norm": 0.5708619952201843, | |
| "learning_rate": 0.0013649059021010894, | |
| "loss": 1.4109, | |
| "step": 6230 | |
| }, | |
| { | |
| "epoch": 0.6582278481012658, | |
| "grad_norm": 0.6014156937599182, | |
| "learning_rate": 0.0013460927469762154, | |
| "loss": 1.4135, | |
| "step": 6240 | |
| }, | |
| { | |
| "epoch": 0.6592827004219409, | |
| "grad_norm": 0.5244723558425903, | |
| "learning_rate": 0.0013275389026252255, | |
| "loss": 1.4207, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 0.6603375527426161, | |
| "grad_norm": 0.5485060811042786, | |
| "learning_rate": 0.0013092407948428887, | |
| "loss": 1.4081, | |
| "step": 6260 | |
| }, | |
| { | |
| "epoch": 0.6613924050632911, | |
| "grad_norm": 0.7510374188423157, | |
| "learning_rate": 0.001291194898688966, | |
| "loss": 1.4185, | |
| "step": 6270 | |
| }, | |
| { | |
| "epoch": 0.6624472573839663, | |
| "grad_norm": 0.49165019392967224, | |
| "learning_rate": 0.001273397737809166, | |
| "loss": 1.4151, | |
| "step": 6280 | |
| }, | |
| { | |
| "epoch": 0.6635021097046413, | |
| "grad_norm": 0.5156455039978027, | |
| "learning_rate": 0.001255845883765463, | |
| "loss": 1.4084, | |
| "step": 6290 | |
| }, | |
| { | |
| "epoch": 0.6645569620253164, | |
| "grad_norm": 0.6343047618865967, | |
| "learning_rate": 0.001238535955375642, | |
| "loss": 1.4036, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.6656118143459916, | |
| "grad_norm": 0.6259722709655762, | |
| "learning_rate": 0.0012214646180619506, | |
| "loss": 1.3995, | |
| "step": 6310 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 0.47512054443359375, | |
| "learning_rate": 0.001204628583208727, | |
| "loss": 1.3935, | |
| "step": 6320 | |
| }, | |
| { | |
| "epoch": 0.6677215189873418, | |
| "grad_norm": 0.6797612309455872, | |
| "learning_rate": 0.0011880246075288827, | |
| "loss": 1.4032, | |
| "step": 6330 | |
| }, | |
| { | |
| "epoch": 0.6687763713080169, | |
| "grad_norm": 0.6804598569869995, | |
| "learning_rate": 0.001171649492439115, | |
| "loss": 1.3936, | |
| "step": 6340 | |
| }, | |
| { | |
| "epoch": 0.669831223628692, | |
| "grad_norm": 0.5329886674880981, | |
| "learning_rate": 0.0011555000834437364, | |
| "loss": 1.4002, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 0.6708860759493671, | |
| "grad_norm": 0.6307658553123474, | |
| "learning_rate": 0.0011395732695269908, | |
| "loss": 1.3984, | |
| "step": 6360 | |
| }, | |
| { | |
| "epoch": 0.6719409282700421, | |
| "grad_norm": 0.5046476125717163, | |
| "learning_rate": 0.0011238659825537505, | |
| "loss": 1.3772, | |
| "step": 6370 | |
| }, | |
| { | |
| "epoch": 0.6729957805907173, | |
| "grad_norm": 0.5106565356254578, | |
| "learning_rate": 0.0011083751966784717, | |
| "loss": 1.381, | |
| "step": 6380 | |
| }, | |
| { | |
| "epoch": 0.6740506329113924, | |
| "grad_norm": 0.9580277800559998, | |
| "learning_rate": 0.0010930979277622953, | |
| "loss": 1.3967, | |
| "step": 6390 | |
| }, | |
| { | |
| "epoch": 0.6751054852320675, | |
| "grad_norm": 0.5464169979095459, | |
| "learning_rate": 0.0010780312327981854, | |
| "loss": 1.395, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.6761603375527426, | |
| "grad_norm": 0.503548800945282, | |
| "learning_rate": 0.0010631722093439888, | |
| "loss": 1.3895, | |
| "step": 6410 | |
| }, | |
| { | |
| "epoch": 0.6772151898734177, | |
| "grad_norm": 0.5025346279144287, | |
| "learning_rate": 0.00104851799496331, | |
| "loss": 1.3778, | |
| "step": 6420 | |
| }, | |
| { | |
| "epoch": 0.6782700421940928, | |
| "grad_norm": 0.4954620599746704, | |
| "learning_rate": 0.0010340657666740914, | |
| "loss": 1.3883, | |
| "step": 6430 | |
| }, | |
| { | |
| "epoch": 0.679324894514768, | |
| "grad_norm": 0.56938636302948, | |
| "learning_rate": 0.0010198127404047975, | |
| "loss": 1.372, | |
| "step": 6440 | |
| }, | |
| { | |
| "epoch": 0.680379746835443, | |
| "grad_norm": 0.5315554738044739, | |
| "learning_rate": 0.0010057561704580897, | |
| "loss": 1.3757, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 0.6814345991561181, | |
| "grad_norm": 0.5042809844017029, | |
| "learning_rate": 0.0009918933489818985, | |
| "loss": 1.3863, | |
| "step": 6460 | |
| }, | |
| { | |
| "epoch": 0.6824894514767933, | |
| "grad_norm": 0.5185731649398804, | |
| "learning_rate": 0.0009782216054477827, | |
| "loss": 1.3798, | |
| "step": 6470 | |
| }, | |
| { | |
| "epoch": 0.6835443037974683, | |
| "grad_norm": 0.6783595681190491, | |
| "learning_rate": 0.0009647383061364801, | |
| "loss": 1.386, | |
| "step": 6480 | |
| }, | |
| { | |
| "epoch": 0.6845991561181435, | |
| "grad_norm": 0.5089995265007019, | |
| "learning_rate": 0.0009514408536305495, | |
| "loss": 1.3809, | |
| "step": 6490 | |
| }, | |
| { | |
| "epoch": 0.6856540084388185, | |
| "grad_norm": 0.480932354927063, | |
| "learning_rate": 0.0009383266863140042, | |
| "loss": 1.3939, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.6867088607594937, | |
| "grad_norm": 0.4996776878833771, | |
| "learning_rate": 0.000925393277878844, | |
| "loss": 1.3939, | |
| "step": 6510 | |
| }, | |
| { | |
| "epoch": 0.6877637130801688, | |
| "grad_norm": 0.4868309497833252, | |
| "learning_rate": 0.0009126381368383879, | |
| "loss": 1.3754, | |
| "step": 6520 | |
| }, | |
| { | |
| "epoch": 0.6888185654008439, | |
| "grad_norm": 0.5652695894241333, | |
| "learning_rate": 0.0009000588060473156, | |
| "loss": 1.3678, | |
| "step": 6530 | |
| }, | |
| { | |
| "epoch": 0.689873417721519, | |
| "grad_norm": 0.5811905264854431, | |
| "learning_rate": 0.0008876528622283235, | |
| "loss": 1.3757, | |
| "step": 6540 | |
| }, | |
| { | |
| "epoch": 0.6909282700421941, | |
| "grad_norm": 0.5382415056228638, | |
| "learning_rate": 0.0008754179155053053, | |
| "loss": 1.3714, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 0.6919831223628692, | |
| "grad_norm": 0.7246244549751282, | |
| "learning_rate": 0.0008633516089429683, | |
| "loss": 1.3727, | |
| "step": 6560 | |
| }, | |
| { | |
| "epoch": 0.6930379746835443, | |
| "grad_norm": 0.49916544556617737, | |
| "learning_rate": 0.0008514516180927928, | |
| "loss": 1.3682, | |
| "step": 6570 | |
| }, | |
| { | |
| "epoch": 0.6940928270042194, | |
| "grad_norm": 0.5470711588859558, | |
| "learning_rate": 0.0008397156505452524, | |
| "loss": 1.3623, | |
| "step": 6580 | |
| }, | |
| { | |
| "epoch": 0.6951476793248945, | |
| "grad_norm": 0.6230381727218628, | |
| "learning_rate": 0.0008281414454882051, | |
| "loss": 1.3648, | |
| "step": 6590 | |
| }, | |
| { | |
| "epoch": 0.6962025316455697, | |
| "grad_norm": 0.5245243906974792, | |
| "learning_rate": 0.0008167267732713704, | |
| "loss": 1.3725, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.6972573839662447, | |
| "grad_norm": 0.505135178565979, | |
| "learning_rate": 0.0008054694349768117, | |
| "loss": 1.3576, | |
| "step": 6610 | |
| }, | |
| { | |
| "epoch": 0.6983122362869199, | |
| "grad_norm": 0.4960167109966278, | |
| "learning_rate": 0.0007943672619953359, | |
| "loss": 1.3668, | |
| "step": 6620 | |
| }, | |
| { | |
| "epoch": 0.6993670886075949, | |
| "grad_norm": 0.5264673829078674, | |
| "learning_rate": 0.0007834181156087356, | |
| "loss": 1.3599, | |
| "step": 6630 | |
| }, | |
| { | |
| "epoch": 0.70042194092827, | |
| "grad_norm": 0.6166502833366394, | |
| "learning_rate": 0.0007726198865777852, | |
| "loss": 1.3636, | |
| "step": 6640 | |
| }, | |
| { | |
| "epoch": 0.7014767932489452, | |
| "grad_norm": 0.536520779132843, | |
| "learning_rate": 0.0007619704947359191, | |
| "loss": 1.3552, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 0.7025316455696202, | |
| "grad_norm": 0.495047390460968, | |
| "learning_rate": 0.0007514678885885087, | |
| "loss": 1.3559, | |
| "step": 6660 | |
| }, | |
| { | |
| "epoch": 0.7035864978902954, | |
| "grad_norm": 0.47605499625205994, | |
| "learning_rate": 0.0007411100449176633, | |
| "loss": 1.3585, | |
| "step": 6670 | |
| }, | |
| { | |
| "epoch": 0.7046413502109705, | |
| "grad_norm": 0.4973078668117523, | |
| "learning_rate": 0.0007308949683924791, | |
| "loss": 1.3604, | |
| "step": 6680 | |
| }, | |
| { | |
| "epoch": 0.7056962025316456, | |
| "grad_norm": 0.5651484131813049, | |
| "learning_rate": 0.000720820691184658, | |
| "loss": 1.3511, | |
| "step": 6690 | |
| }, | |
| { | |
| "epoch": 0.7067510548523207, | |
| "grad_norm": 0.46959301829338074, | |
| "learning_rate": 0.0007108852725894269, | |
| "loss": 1.3531, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.7078059071729957, | |
| "grad_norm": 0.5081770420074463, | |
| "learning_rate": 0.000701086798651681, | |
| "loss": 1.3544, | |
| "step": 6710 | |
| }, | |
| { | |
| "epoch": 0.7088607594936709, | |
| "grad_norm": 0.5139282941818237, | |
| "learning_rate": 0.0006914233817972798, | |
| "loss": 1.3427, | |
| "step": 6720 | |
| }, | |
| { | |
| "epoch": 0.709915611814346, | |
| "grad_norm": 0.49106651544570923, | |
| "learning_rate": 0.0006818931604694261, | |
| "loss": 1.3531, | |
| "step": 6730 | |
| }, | |
| { | |
| "epoch": 0.7109704641350211, | |
| "grad_norm": 0.46146702766418457, | |
| "learning_rate": 0.0006724942987700563, | |
| "loss": 1.3528, | |
| "step": 6740 | |
| }, | |
| { | |
| "epoch": 0.7120253164556962, | |
| "grad_norm": 0.4885563552379608, | |
| "learning_rate": 0.0006632249861061732, | |
| "loss": 1.3578, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 0.7130801687763713, | |
| "grad_norm": 0.6136794090270996, | |
| "learning_rate": 0.0006540834368410549, | |
| "loss": 1.3502, | |
| "step": 6760 | |
| }, | |
| { | |
| "epoch": 0.7141350210970464, | |
| "grad_norm": 0.5482812523841858, | |
| "learning_rate": 0.0006450678899502701, | |
| "loss": 1.3528, | |
| "step": 6770 | |
| }, | |
| { | |
| "epoch": 0.7151898734177216, | |
| "grad_norm": 0.5637302398681641, | |
| "learning_rate": 0.0006361766086824345, | |
| "loss": 1.3499, | |
| "step": 6780 | |
| }, | |
| { | |
| "epoch": 0.7162447257383966, | |
| "grad_norm": 0.46481969952583313, | |
| "learning_rate": 0.000627407880224645, | |
| "loss": 1.3556, | |
| "step": 6790 | |
| }, | |
| { | |
| "epoch": 0.7172995780590717, | |
| "grad_norm": 0.4623727798461914, | |
| "learning_rate": 0.0006187600153725225, | |
| "loss": 1.3395, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.7183544303797469, | |
| "grad_norm": 0.8997368216514587, | |
| "learning_rate": 0.0006102313482048055, | |
| "loss": 1.3446, | |
| "step": 6810 | |
| }, | |
| { | |
| "epoch": 0.7194092827004219, | |
| "grad_norm": 0.5417001247406006, | |
| "learning_rate": 0.0006018202357624274, | |
| "loss": 1.3472, | |
| "step": 6820 | |
| }, | |
| { | |
| "epoch": 0.7204641350210971, | |
| "grad_norm": 0.6152294874191284, | |
| "learning_rate": 0.0005935250577320168, | |
| "loss": 1.3418, | |
| "step": 6830 | |
| }, | |
| { | |
| "epoch": 0.7215189873417721, | |
| "grad_norm": 0.672227144241333, | |
| "learning_rate": 0.0005853442161337618, | |
| "loss": 1.3325, | |
| "step": 6840 | |
| }, | |
| { | |
| "epoch": 0.7225738396624473, | |
| "grad_norm": 0.4651714265346527, | |
| "learning_rate": 0.0005772761350135759, | |
| "loss": 1.3408, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 0.7236286919831224, | |
| "grad_norm": 0.5990096926689148, | |
| "learning_rate": 0.0005693192601395058, | |
| "loss": 1.3335, | |
| "step": 6860 | |
| }, | |
| { | |
| "epoch": 0.7246835443037974, | |
| "grad_norm": 0.4861851632595062, | |
| "learning_rate": 0.000561472058702326, | |
| "loss": 1.331, | |
| "step": 6870 | |
| }, | |
| { | |
| "epoch": 0.7257383966244726, | |
| "grad_norm": 0.5941367745399475, | |
| "learning_rate": 0.000553733019020258, | |
| "loss": 1.337, | |
| "step": 6880 | |
| }, | |
| { | |
| "epoch": 0.7267932489451476, | |
| "grad_norm": 0.5600623488426208, | |
| "learning_rate": 0.0005461006502477612, | |
| "loss": 1.3305, | |
| "step": 6890 | |
| }, | |
| { | |
| "epoch": 0.7278481012658228, | |
| "grad_norm": 0.5032252669334412, | |
| "learning_rate": 0.0005385734820883369, | |
| "loss": 1.3289, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.7289029535864979, | |
| "grad_norm": 0.5026607513427734, | |
| "learning_rate": 0.0005311500645112907, | |
| "loss": 1.355, | |
| "step": 6910 | |
| }, | |
| { | |
| "epoch": 0.729957805907173, | |
| "grad_norm": 0.5264232158660889, | |
| "learning_rate": 0.0005238289674723993, | |
| "loss": 1.3353, | |
| "step": 6920 | |
| }, | |
| { | |
| "epoch": 0.7310126582278481, | |
| "grad_norm": 0.47649210691452026, | |
| "learning_rate": 0.0005166087806384274, | |
| "loss": 1.3377, | |
| "step": 6930 | |
| }, | |
| { | |
| "epoch": 0.7320675105485233, | |
| "grad_norm": 0.4772520363330841, | |
| "learning_rate": 0.0005094881131154418, | |
| "loss": 1.3419, | |
| "step": 6940 | |
| }, | |
| { | |
| "epoch": 0.7331223628691983, | |
| "grad_norm": 0.5586071610450745, | |
| "learning_rate": 0.0005024655931808696, | |
| "loss": 1.3374, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 0.7341772151898734, | |
| "grad_norm": 0.49496790766716003, | |
| "learning_rate": 0.0004955398680192508, | |
| "loss": 1.328, | |
| "step": 6960 | |
| }, | |
| { | |
| "epoch": 0.7352320675105485, | |
| "grad_norm": 0.5294111967086792, | |
| "learning_rate": 0.000488709603461632, | |
| "loss": 1.3248, | |
| "step": 6970 | |
| }, | |
| { | |
| "epoch": 0.7362869198312236, | |
| "grad_norm": 0.47923532128334045, | |
| "learning_rate": 0.000481973483728553, | |
| "loss": 1.3276, | |
| "step": 6980 | |
| }, | |
| { | |
| "epoch": 0.7373417721518988, | |
| "grad_norm": 0.4925960600376129, | |
| "learning_rate": 0.0004753302111765748, | |
| "loss": 1.3268, | |
| "step": 6990 | |
| }, | |
| { | |
| "epoch": 0.7383966244725738, | |
| "grad_norm": 0.4940263032913208, | |
| "learning_rate": 0.0004687785060483032, | |
| "loss": 1.3384, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.739451476793249, | |
| "grad_norm": 0.5015807151794434, | |
| "learning_rate": 0.0004623171062258558, | |
| "loss": 1.3094, | |
| "step": 7010 | |
| }, | |
| { | |
| "epoch": 0.740506329113924, | |
| "grad_norm": 0.7085549235343933, | |
| "learning_rate": 0.0004559447669877288, | |
| "loss": 1.3253, | |
| "step": 7020 | |
| }, | |
| { | |
| "epoch": 0.7415611814345991, | |
| "grad_norm": 0.4822874367237091, | |
| "learning_rate": 0.00044966026076901413, | |
| "loss": 1.3294, | |
| "step": 7030 | |
| }, | |
| { | |
| "epoch": 0.7426160337552743, | |
| "grad_norm": 0.6149103045463562, | |
| "learning_rate": 0.00044346237692492177, | |
| "loss": 1.334, | |
| "step": 7040 | |
| }, | |
| { | |
| "epoch": 0.7436708860759493, | |
| "grad_norm": 0.5246915221214294, | |
| "learning_rate": 0.0004373499214975615, | |
| "loss": 1.3194, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 0.7447257383966245, | |
| "grad_norm": 0.5026849508285522, | |
| "learning_rate": 0.0004313217169859396, | |
| "loss": 1.324, | |
| "step": 7060 | |
| }, | |
| { | |
| "epoch": 0.7457805907172996, | |
| "grad_norm": 0.585382878780365, | |
| "learning_rate": 0.0004253766021191256, | |
| "loss": 1.3323, | |
| "step": 7070 | |
| }, | |
| { | |
| "epoch": 0.7468354430379747, | |
| "grad_norm": 0.4607876241207123, | |
| "learning_rate": 0.00041951343163254497, | |
| "loss": 1.3266, | |
| "step": 7080 | |
| }, | |
| { | |
| "epoch": 0.7478902953586498, | |
| "grad_norm": 0.5416311621665955, | |
| "learning_rate": 0.00041373107604735626, | |
| "loss": 1.3255, | |
| "step": 7090 | |
| }, | |
| { | |
| "epoch": 0.7489451476793249, | |
| "grad_norm": 0.48268038034439087, | |
| "learning_rate": 0.0004080284214528687, | |
| "loss": 1.3225, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 0.5381031036376953, | |
| "learning_rate": 0.0004024043692919589, | |
| "loss": 1.3282, | |
| "step": 7110 | |
| }, | |
| { | |
| "epoch": 0.7510548523206751, | |
| "grad_norm": 0.5726626515388489, | |
| "learning_rate": 0.0003968578361494449, | |
| "loss": 1.325, | |
| "step": 7120 | |
| }, | |
| { | |
| "epoch": 0.7521097046413502, | |
| "grad_norm": 0.4642830789089203, | |
| "learning_rate": 0.000391387753543378, | |
| "loss": 1.3323, | |
| "step": 7130 | |
| }, | |
| { | |
| "epoch": 0.7531645569620253, | |
| "grad_norm": 0.4678974747657776, | |
| "learning_rate": 0.00038599306771921023, | |
| "loss": 1.3162, | |
| "step": 7140 | |
| }, | |
| { | |
| "epoch": 0.7542194092827004, | |
| "grad_norm": 0.5471410751342773, | |
| "learning_rate": 0.0003806727394468004, | |
| "loss": 1.3134, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 0.7552742616033755, | |
| "grad_norm": 0.46361368894577026, | |
| "learning_rate": 0.0003754257438202162, | |
| "loss": 1.3225, | |
| "step": 7160 | |
| }, | |
| { | |
| "epoch": 0.7563291139240507, | |
| "grad_norm": 0.6638835668563843, | |
| "learning_rate": 0.0003702510700602974, | |
| "loss": 1.3296, | |
| "step": 7170 | |
| }, | |
| { | |
| "epoch": 0.7573839662447257, | |
| "grad_norm": 0.5422773957252502, | |
| "learning_rate": 0.0003651477213199393, | |
| "loss": 1.3072, | |
| "step": 7180 | |
| }, | |
| { | |
| "epoch": 0.7584388185654009, | |
| "grad_norm": 0.49227139353752136, | |
| "learning_rate": 0.000360114714492061, | |
| "loss": 1.3101, | |
| "step": 7190 | |
| }, | |
| { | |
| "epoch": 0.759493670886076, | |
| "grad_norm": 0.49915429949760437, | |
| "learning_rate": 0.0003551510800202195, | |
| "loss": 1.315, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.760548523206751, | |
| "grad_norm": 0.5297178030014038, | |
| "learning_rate": 0.0003502558617118353, | |
| "loss": 1.319, | |
| "step": 7210 | |
| }, | |
| { | |
| "epoch": 0.7616033755274262, | |
| "grad_norm": 0.55345618724823, | |
| "learning_rate": 0.0003454281165539914, | |
| "loss": 1.3311, | |
| "step": 7220 | |
| }, | |
| { | |
| "epoch": 0.7626582278481012, | |
| "grad_norm": 0.5680271983146667, | |
| "learning_rate": 0.00034066691453177176, | |
| "loss": 1.3288, | |
| "step": 7230 | |
| }, | |
| { | |
| "epoch": 0.7637130801687764, | |
| "grad_norm": 0.47741663455963135, | |
| "learning_rate": 0.0003359713384491037, | |
| "loss": 1.3229, | |
| "step": 7240 | |
| }, | |
| { | |
| "epoch": 0.7647679324894515, | |
| "grad_norm": 0.5046437978744507, | |
| "learning_rate": 0.00033134048375206944, | |
| "loss": 1.3176, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 0.7658227848101266, | |
| "grad_norm": 0.457300066947937, | |
| "learning_rate": 0.0003267734583546536, | |
| "loss": 1.3159, | |
| "step": 7260 | |
| }, | |
| { | |
| "epoch": 0.7668776371308017, | |
| "grad_norm": 0.5015650987625122, | |
| "learning_rate": 0.00032226938246689157, | |
| "loss": 1.3147, | |
| "step": 7270 | |
| }, | |
| { | |
| "epoch": 0.7679324894514767, | |
| "grad_norm": 0.5395479202270508, | |
| "learning_rate": 0.0003178273884253874, | |
| "loss": 1.3158, | |
| "step": 7280 | |
| }, | |
| { | |
| "epoch": 0.7689873417721519, | |
| "grad_norm": 0.4769648015499115, | |
| "learning_rate": 0.0003134466205261674, | |
| "loss": 1.3257, | |
| "step": 7290 | |
| }, | |
| { | |
| "epoch": 0.770042194092827, | |
| "grad_norm": 0.5480552315711975, | |
| "learning_rate": 0.0003091262348598378, | |
| "loss": 1.3288, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.7710970464135021, | |
| "grad_norm": 0.47319650650024414, | |
| "learning_rate": 0.0003048653991490141, | |
| "loss": 1.3094, | |
| "step": 7310 | |
| }, | |
| { | |
| "epoch": 0.7721518987341772, | |
| "grad_norm": 0.5190073847770691, | |
| "learning_rate": 0.00030066329258799187, | |
| "loss": 1.3068, | |
| "step": 7320 | |
| }, | |
| { | |
| "epoch": 0.7732067510548524, | |
| "grad_norm": 0.5128898620605469, | |
| "learning_rate": 0.0002965191056846266, | |
| "loss": 1.3107, | |
| "step": 7330 | |
| }, | |
| { | |
| "epoch": 0.7742616033755274, | |
| "grad_norm": 0.5712609887123108, | |
| "learning_rate": 0.000292432040104394, | |
| "loss": 1.3076, | |
| "step": 7340 | |
| }, | |
| { | |
| "epoch": 0.7753164556962026, | |
| "grad_norm": 0.4749037027359009, | |
| "learning_rate": 0.00028840130851659853, | |
| "loss": 1.3045, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 0.7763713080168776, | |
| "grad_norm": 0.48418837785720825, | |
| "learning_rate": 0.0002844261344427028, | |
| "loss": 1.3137, | |
| "step": 7360 | |
| }, | |
| { | |
| "epoch": 0.7774261603375527, | |
| "grad_norm": 0.47260743379592896, | |
| "learning_rate": 0.0002805057521067471, | |
| "loss": 1.3122, | |
| "step": 7370 | |
| }, | |
| { | |
| "epoch": 0.7784810126582279, | |
| "grad_norm": 0.47214198112487793, | |
| "learning_rate": 0.00027663940628783017, | |
| "loss": 1.3006, | |
| "step": 7380 | |
| }, | |
| { | |
| "epoch": 0.7795358649789029, | |
| "grad_norm": 0.5273362398147583, | |
| "learning_rate": 0.00027282635217462393, | |
| "loss": 1.3119, | |
| "step": 7390 | |
| }, | |
| { | |
| "epoch": 0.7805907172995781, | |
| "grad_norm": 0.5290525555610657, | |
| "learning_rate": 0.0002690658552218937, | |
| "loss": 1.3165, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.7816455696202531, | |
| "grad_norm": 0.48313039541244507, | |
| "learning_rate": 0.00026535719100899516, | |
| "loss": 1.2975, | |
| "step": 7410 | |
| }, | |
| { | |
| "epoch": 0.7827004219409283, | |
| "grad_norm": 0.49278801679611206, | |
| "learning_rate": 0.00026169964510032245, | |
| "loss": 1.3032, | |
| "step": 7420 | |
| }, | |
| { | |
| "epoch": 0.7837552742616034, | |
| "grad_norm": 0.4592038691043854, | |
| "learning_rate": 0.00025809251290767984, | |
| "loss": 1.2909, | |
| "step": 7430 | |
| }, | |
| { | |
| "epoch": 0.7848101265822784, | |
| "grad_norm": 0.47983959317207336, | |
| "learning_rate": 0.00025453509955454957, | |
| "loss": 1.2932, | |
| "step": 7440 | |
| }, | |
| { | |
| "epoch": 0.7858649789029536, | |
| "grad_norm": 0.5009822845458984, | |
| "learning_rate": 0.00025102671974223175, | |
| "loss": 1.2955, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 0.7869198312236287, | |
| "grad_norm": 0.47341182827949524, | |
| "learning_rate": 0.00024756669761782815, | |
| "loss": 1.3073, | |
| "step": 7460 | |
| }, | |
| { | |
| "epoch": 0.7879746835443038, | |
| "grad_norm": 0.4908011555671692, | |
| "learning_rate": 0.0002441543666440464, | |
| "loss": 1.2989, | |
| "step": 7470 | |
| }, | |
| { | |
| "epoch": 0.7890295358649789, | |
| "grad_norm": 0.5158747434616089, | |
| "learning_rate": 0.00024078906947079878, | |
| "loss": 1.3039, | |
| "step": 7480 | |
| }, | |
| { | |
| "epoch": 0.790084388185654, | |
| "grad_norm": 0.469517320394516, | |
| "learning_rate": 0.00023747015780857005, | |
| "loss": 1.3101, | |
| "step": 7490 | |
| }, | |
| { | |
| "epoch": 0.7911392405063291, | |
| "grad_norm": 0.47917693853378296, | |
| "learning_rate": 0.00023419699230353144, | |
| "loss": 1.3061, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.7921940928270043, | |
| "grad_norm": 0.4800731837749481, | |
| "learning_rate": 0.00023096894241437586, | |
| "loss": 1.312, | |
| "step": 7510 | |
| }, | |
| { | |
| "epoch": 0.7932489451476793, | |
| "grad_norm": 0.48870769143104553, | |
| "learning_rate": 0.00022778538629085056, | |
| "loss": 1.2957, | |
| "step": 7520 | |
| }, | |
| { | |
| "epoch": 0.7943037974683544, | |
| "grad_norm": 0.48116040229797363, | |
| "learning_rate": 0.00022464571065396427, | |
| "loss": 1.2967, | |
| "step": 7530 | |
| }, | |
| { | |
| "epoch": 0.7953586497890295, | |
| "grad_norm": 0.5820254683494568, | |
| "learning_rate": 0.00022154931067784521, | |
| "loss": 1.2962, | |
| "step": 7540 | |
| }, | |
| { | |
| "epoch": 0.7964135021097046, | |
| "grad_norm": 0.4812745451927185, | |
| "learning_rate": 0.00021849558987322782, | |
| "loss": 1.2924, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 0.7974683544303798, | |
| "grad_norm": 0.5120149254798889, | |
| "learning_rate": 0.0002154839599725452, | |
| "loss": 1.2939, | |
| "step": 7560 | |
| }, | |
| { | |
| "epoch": 0.7985232067510548, | |
| "grad_norm": 0.4970279633998871, | |
| "learning_rate": 0.00021251384081660544, | |
| "loss": 1.307, | |
| "step": 7570 | |
| }, | |
| { | |
| "epoch": 0.79957805907173, | |
| "grad_norm": 0.4871574938297272, | |
| "learning_rate": 0.0002095846602428303, | |
| "loss": 1.3018, | |
| "step": 7580 | |
| }, | |
| { | |
| "epoch": 0.8006329113924051, | |
| "grad_norm": 0.4676864445209503, | |
| "learning_rate": 0.00020669585397503358, | |
| "loss": 1.293, | |
| "step": 7590 | |
| }, | |
| { | |
| "epoch": 0.8016877637130801, | |
| "grad_norm": 0.4470655620098114, | |
| "learning_rate": 0.0002038468655147195, | |
| "loss": 1.2965, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.8027426160337553, | |
| "grad_norm": 0.5174440741539001, | |
| "learning_rate": 0.00020103714603387894, | |
| "loss": 1.3112, | |
| "step": 7610 | |
| }, | |
| { | |
| "epoch": 0.8037974683544303, | |
| "grad_norm": 0.4520186483860016, | |
| "learning_rate": 0.00019826615426926338, | |
| "loss": 1.2835, | |
| "step": 7620 | |
| }, | |
| { | |
| "epoch": 0.8048523206751055, | |
| "grad_norm": 0.45952776074409485, | |
| "learning_rate": 0.00019553335641811625, | |
| "loss": 1.3055, | |
| "step": 7630 | |
| }, | |
| { | |
| "epoch": 0.8059071729957806, | |
| "grad_norm": 0.5378235578536987, | |
| "learning_rate": 0.0001928382260353415, | |
| "loss": 1.2962, | |
| "step": 7640 | |
| }, | |
| { | |
| "epoch": 0.8069620253164557, | |
| "grad_norm": 0.4733293056488037, | |
| "learning_rate": 0.00019018024393208902, | |
| "loss": 1.3121, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 0.8080168776371308, | |
| "grad_norm": 0.4595848321914673, | |
| "learning_rate": 0.00018755889807573872, | |
| "loss": 1.2913, | |
| "step": 7660 | |
| }, | |
| { | |
| "epoch": 0.8090717299578059, | |
| "grad_norm": 0.4874734580516815, | |
| "learning_rate": 0.00018497368349126262, | |
| "loss": 1.2967, | |
| "step": 7670 | |
| }, | |
| { | |
| "epoch": 0.810126582278481, | |
| "grad_norm": 0.5490835309028625, | |
| "learning_rate": 0.00018242410216394648, | |
| "loss": 1.3108, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.8111814345991561, | |
| "grad_norm": 0.44863075017929077, | |
| "learning_rate": 0.0001799096629434529, | |
| "loss": 1.2844, | |
| "step": 7690 | |
| }, | |
| { | |
| "epoch": 0.8122362869198312, | |
| "grad_norm": 0.4595239758491516, | |
| "learning_rate": 0.00017742988144920578, | |
| "loss": 1.2936, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.8132911392405063, | |
| "grad_norm": 0.6216907501220703, | |
| "learning_rate": 0.00017498427997707976, | |
| "loss": 1.2918, | |
| "step": 7710 | |
| }, | |
| { | |
| "epoch": 0.8143459915611815, | |
| "grad_norm": 0.4747745394706726, | |
| "learning_rate": 0.00017257238740737548, | |
| "loss": 1.2967, | |
| "step": 7720 | |
| }, | |
| { | |
| "epoch": 0.8154008438818565, | |
| "grad_norm": 0.49981772899627686, | |
| "learning_rate": 0.00017019373911406307, | |
| "loss": 1.3045, | |
| "step": 7730 | |
| }, | |
| { | |
| "epoch": 0.8164556962025317, | |
| "grad_norm": 0.46517640352249146, | |
| "learning_rate": 0.000167847876875277, | |
| "loss": 1.2998, | |
| "step": 7740 | |
| }, | |
| { | |
| "epoch": 0.8175105485232067, | |
| "grad_norm": 0.4871491491794586, | |
| "learning_rate": 0.00016553434878504428, | |
| "loss": 1.2859, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 0.8185654008438819, | |
| "grad_norm": 0.5478642582893372, | |
| "learning_rate": 0.00016325270916622947, | |
| "loss": 1.2875, | |
| "step": 7760 | |
| }, | |
| { | |
| "epoch": 0.819620253164557, | |
| "grad_norm": 0.4661363959312439, | |
| "learning_rate": 0.00016100251848467966, | |
| "loss": 1.2952, | |
| "step": 7770 | |
| }, | |
| { | |
| "epoch": 0.820675105485232, | |
| "grad_norm": 0.5517392158508301, | |
| "learning_rate": 0.0001587833432645528, | |
| "loss": 1.2882, | |
| "step": 7780 | |
| }, | |
| { | |
| "epoch": 0.8217299578059072, | |
| "grad_norm": 0.5190399289131165, | |
| "learning_rate": 0.00015659475600481292, | |
| "loss": 1.3051, | |
| "step": 7790 | |
| }, | |
| { | |
| "epoch": 0.8227848101265823, | |
| "grad_norm": 0.46546727418899536, | |
| "learning_rate": 0.00015443633509687688, | |
| "loss": 1.2928, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.8238396624472574, | |
| "grad_norm": 0.5459626913070679, | |
| "learning_rate": 0.00015230766474339536, | |
| "loss": 1.293, | |
| "step": 7810 | |
| }, | |
| { | |
| "epoch": 0.8248945147679325, | |
| "grad_norm": 0.4505077302455902, | |
| "learning_rate": 0.00015020833487815416, | |
| "loss": 1.2994, | |
| "step": 7820 | |
| }, | |
| { | |
| "epoch": 0.8259493670886076, | |
| "grad_norm": 0.47471731901168823, | |
| "learning_rate": 0.0001481379410870792, | |
| "loss": 1.2903, | |
| "step": 7830 | |
| }, | |
| { | |
| "epoch": 0.8270042194092827, | |
| "grad_norm": 0.4919489324092865, | |
| "learning_rate": 0.00014609608453033013, | |
| "loss": 1.2758, | |
| "step": 7840 | |
| }, | |
| { | |
| "epoch": 0.8280590717299579, | |
| "grad_norm": 0.48681142926216125, | |
| "learning_rate": 0.00014408237186546807, | |
| "loss": 1.2849, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 0.8291139240506329, | |
| "grad_norm": 0.4562098979949951, | |
| "learning_rate": 0.00014209641517168273, | |
| "loss": 1.2734, | |
| "step": 7860 | |
| }, | |
| { | |
| "epoch": 0.830168776371308, | |
| "grad_norm": 0.5935469269752502, | |
| "learning_rate": 0.00014013783187506265, | |
| "loss": 1.2873, | |
| "step": 7870 | |
| }, | |
| { | |
| "epoch": 0.8312236286919831, | |
| "grad_norm": 0.47083306312561035, | |
| "learning_rate": 0.00013820624467489697, | |
| "loss": 1.3074, | |
| "step": 7880 | |
| }, | |
| { | |
| "epoch": 0.8322784810126582, | |
| "grad_norm": 0.4802079498767853, | |
| "learning_rate": 0.00013630128147099213, | |
| "loss": 1.3035, | |
| "step": 7890 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "grad_norm": 0.4673043489456177, | |
| "learning_rate": 0.00013442257529199068, | |
| "loss": 1.2809, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.8343881856540084, | |
| "grad_norm": 0.521275520324707, | |
| "learning_rate": 0.00013256976422467803, | |
| "loss": 1.2926, | |
| "step": 7910 | |
| }, | |
| { | |
| "epoch": 0.8354430379746836, | |
| "grad_norm": 0.5621090531349182, | |
| "learning_rate": 0.00013074249134426366, | |
| "loss": 1.2899, | |
| "step": 7920 | |
| }, | |
| { | |
| "epoch": 0.8364978902953587, | |
| "grad_norm": 0.46371445059776306, | |
| "learning_rate": 0.0001289404046456233, | |
| "loss": 1.3009, | |
| "step": 7930 | |
| }, | |
| { | |
| "epoch": 0.8375527426160337, | |
| "grad_norm": 0.4880622625350952, | |
| "learning_rate": 0.0001271631569754887, | |
| "loss": 1.291, | |
| "step": 7940 | |
| }, | |
| { | |
| "epoch": 0.8386075949367089, | |
| "grad_norm": 0.49742773175239563, | |
| "learning_rate": 0.0001254104059655723, | |
| "loss": 1.2959, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 0.8396624472573839, | |
| "grad_norm": 0.45721614360809326, | |
| "learning_rate": 0.00012368181396661337, | |
| "loss": 1.2823, | |
| "step": 7960 | |
| }, | |
| { | |
| "epoch": 0.8407172995780591, | |
| "grad_norm": 0.46568551659584045, | |
| "learning_rate": 0.00012197704798333364, | |
| "loss": 1.2831, | |
| "step": 7970 | |
| }, | |
| { | |
| "epoch": 0.8417721518987342, | |
| "grad_norm": 0.47091469168663025, | |
| "learning_rate": 0.00012029577961028894, | |
| "loss": 1.282, | |
| "step": 7980 | |
| }, | |
| { | |
| "epoch": 0.8428270042194093, | |
| "grad_norm": 0.5390154719352722, | |
| "learning_rate": 0.00011863768496860542, | |
| "loss": 1.2927, | |
| "step": 7990 | |
| }, | |
| { | |
| "epoch": 0.8438818565400844, | |
| "grad_norm": 0.577823281288147, | |
| "learning_rate": 0.00011700244464358777, | |
| "loss": 1.2881, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.8449367088607594, | |
| "grad_norm": 0.47146788239479065, | |
| "learning_rate": 0.00011538974362318715, | |
| "loss": 1.2935, | |
| "step": 8010 | |
| }, | |
| { | |
| "epoch": 0.8459915611814346, | |
| "grad_norm": 0.49528032541275024, | |
| "learning_rate": 0.00011379927123731737, | |
| "loss": 1.2876, | |
| "step": 8020 | |
| }, | |
| { | |
| "epoch": 0.8470464135021097, | |
| "grad_norm": 0.48834657669067383, | |
| "learning_rate": 0.0001122307210980077, | |
| "loss": 1.2853, | |
| "step": 8030 | |
| }, | |
| { | |
| "epoch": 0.8481012658227848, | |
| "grad_norm": 0.45491474866867065, | |
| "learning_rate": 0.00011068379104038026, | |
| "loss": 1.2962, | |
| "step": 8040 | |
| }, | |
| { | |
| "epoch": 0.8491561181434599, | |
| "grad_norm": 0.45627567172050476, | |
| "learning_rate": 0.00010915818306444116, | |
| "loss": 1.2778, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 0.8502109704641351, | |
| "grad_norm": 0.4637642800807953, | |
| "learning_rate": 0.00010765360327767384, | |
| "loss": 1.2853, | |
| "step": 8060 | |
| }, | |
| { | |
| "epoch": 0.8512658227848101, | |
| "grad_norm": 0.4641653299331665, | |
| "learning_rate": 0.00010616976183842376, | |
| "loss": 1.291, | |
| "step": 8070 | |
| }, | |
| { | |
| "epoch": 0.8523206751054853, | |
| "grad_norm": 0.4908168911933899, | |
| "learning_rate": 0.00010470637290006365, | |
| "loss": 1.2926, | |
| "step": 8080 | |
| }, | |
| { | |
| "epoch": 0.8533755274261603, | |
| "grad_norm": 0.4617968797683716, | |
| "learning_rate": 0.00010326315455592764, | |
| "loss": 1.28, | |
| "step": 8090 | |
| }, | |
| { | |
| "epoch": 0.8544303797468354, | |
| "grad_norm": 0.47194594144821167, | |
| "learning_rate": 0.0001018398287850053, | |
| "loss": 1.2752, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.8554852320675106, | |
| "grad_norm": 0.5317877531051636, | |
| "learning_rate": 0.00010043612139838357, | |
| "loss": 1.2938, | |
| "step": 8110 | |
| }, | |
| { | |
| "epoch": 0.8565400843881856, | |
| "grad_norm": 0.46418246626853943, | |
| "learning_rate": 9.905176198642719e-05, | |
| "loss": 1.2834, | |
| "step": 8120 | |
| }, | |
| { | |
| "epoch": 0.8575949367088608, | |
| "grad_norm": 0.49755966663360596, | |
| "learning_rate": 9.76864838666871e-05, | |
| "loss": 1.2832, | |
| "step": 8130 | |
| }, | |
| { | |
| "epoch": 0.8586497890295358, | |
| "grad_norm": 0.46786341071128845, | |
| "learning_rate": 9.634002403252676e-05, | |
| "loss": 1.2836, | |
| "step": 8140 | |
| }, | |
| { | |
| "epoch": 0.859704641350211, | |
| "grad_norm": 0.4702818989753723, | |
| "learning_rate": 9.501212310245681e-05, | |
| "loss": 1.2783, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 0.8607594936708861, | |
| "grad_norm": 0.44937238097190857, | |
| "learning_rate": 9.370252527016777e-05, | |
| "loss": 1.2899, | |
| "step": 8160 | |
| }, | |
| { | |
| "epoch": 0.8618143459915611, | |
| "grad_norm": 0.49406906962394714, | |
| "learning_rate": 9.241097825525163e-05, | |
| "loss": 1.2807, | |
| "step": 8170 | |
| }, | |
| { | |
| "epoch": 0.8628691983122363, | |
| "grad_norm": 0.5151875019073486, | |
| "learning_rate": 9.113723325460276e-05, | |
| "loss": 1.288, | |
| "step": 8180 | |
| }, | |
| { | |
| "epoch": 0.8639240506329114, | |
| "grad_norm": 0.5420994162559509, | |
| "learning_rate": 8.988104489448849e-05, | |
| "loss": 1.2815, | |
| "step": 8190 | |
| }, | |
| { | |
| "epoch": 0.8649789029535865, | |
| "grad_norm": 0.4622182250022888, | |
| "learning_rate": 8.864217118328042e-05, | |
| "loss": 1.2922, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.8660337552742616, | |
| "grad_norm": 0.48794785141944885, | |
| "learning_rate": 8.742037346483729e-05, | |
| "loss": 1.2854, | |
| "step": 8210 | |
| }, | |
| { | |
| "epoch": 0.8670886075949367, | |
| "grad_norm": 0.4837190508842468, | |
| "learning_rate": 8.62154163725303e-05, | |
| "loss": 1.2934, | |
| "step": 8220 | |
| }, | |
| { | |
| "epoch": 0.8681434599156118, | |
| "grad_norm": 0.47257596254348755, | |
| "learning_rate": 8.502706778390219e-05, | |
| "loss": 1.29, | |
| "step": 8230 | |
| }, | |
| { | |
| "epoch": 0.869198312236287, | |
| "grad_norm": 0.46022605895996094, | |
| "learning_rate": 8.38550987759513e-05, | |
| "loss": 1.2858, | |
| "step": 8240 | |
| }, | |
| { | |
| "epoch": 0.870253164556962, | |
| "grad_norm": 0.4648395776748657, | |
| "learning_rate": 8.269928358103191e-05, | |
| "loss": 1.3001, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 0.8713080168776371, | |
| "grad_norm": 0.4802897870540619, | |
| "learning_rate": 8.155939954336243e-05, | |
| "loss": 1.2892, | |
| "step": 8260 | |
| }, | |
| { | |
| "epoch": 0.8723628691983122, | |
| "grad_norm": 0.49936237931251526, | |
| "learning_rate": 8.043522707613312e-05, | |
| "loss": 1.2858, | |
| "step": 8270 | |
| }, | |
| { | |
| "epoch": 0.8734177215189873, | |
| "grad_norm": 0.4365840256214142, | |
| "learning_rate": 7.932654961920486e-05, | |
| "loss": 1.2715, | |
| "step": 8280 | |
| }, | |
| { | |
| "epoch": 0.8744725738396625, | |
| "grad_norm": 0.4833032488822937, | |
| "learning_rate": 7.823315359739135e-05, | |
| "loss": 1.272, | |
| "step": 8290 | |
| }, | |
| { | |
| "epoch": 0.8755274261603375, | |
| "grad_norm": 0.48729777336120605, | |
| "learning_rate": 7.715482837931577e-05, | |
| "loss": 1.2958, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.8765822784810127, | |
| "grad_norm": 0.4626319110393524, | |
| "learning_rate": 7.6091366236835e-05, | |
| "loss": 1.2749, | |
| "step": 8310 | |
| }, | |
| { | |
| "epoch": 0.8776371308016878, | |
| "grad_norm": 0.46448102593421936, | |
| "learning_rate": 7.504256230502289e-05, | |
| "loss": 1.2919, | |
| "step": 8320 | |
| }, | |
| { | |
| "epoch": 0.8786919831223629, | |
| "grad_norm": 0.4604092538356781, | |
| "learning_rate": 7.400821454270524e-05, | |
| "loss": 1.2859, | |
| "step": 8330 | |
| }, | |
| { | |
| "epoch": 0.879746835443038, | |
| "grad_norm": 0.45476552844047546, | |
| "learning_rate": 7.29881236935386e-05, | |
| "loss": 1.274, | |
| "step": 8340 | |
| }, | |
| { | |
| "epoch": 0.880801687763713, | |
| "grad_norm": 0.46123555302619934, | |
| "learning_rate": 7.198209324762562e-05, | |
| "loss": 1.2777, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 0.8818565400843882, | |
| "grad_norm": 0.4466295838356018, | |
| "learning_rate": 7.098992940365946e-05, | |
| "loss": 1.2739, | |
| "step": 8360 | |
| }, | |
| { | |
| "epoch": 0.8829113924050633, | |
| "grad_norm": 0.4567776024341583, | |
| "learning_rate": 7.001144103159e-05, | |
| "loss": 1.2824, | |
| "step": 8370 | |
| }, | |
| { | |
| "epoch": 0.8839662447257384, | |
| "grad_norm": 0.47256141901016235, | |
| "learning_rate": 6.904643963580461e-05, | |
| "loss": 1.2882, | |
| "step": 8380 | |
| }, | |
| { | |
| "epoch": 0.8850210970464135, | |
| "grad_norm": 0.5056165456771851, | |
| "learning_rate": 6.809473931881644e-05, | |
| "loss": 1.2796, | |
| "step": 8390 | |
| }, | |
| { | |
| "epoch": 0.8860759493670886, | |
| "grad_norm": 0.4639674127101898, | |
| "learning_rate": 6.71561567454532e-05, | |
| "loss": 1.2782, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.8871308016877637, | |
| "grad_norm": 0.47839340567588806, | |
| "learning_rate": 6.623051110753948e-05, | |
| "loss": 1.2982, | |
| "step": 8410 | |
| }, | |
| { | |
| "epoch": 0.8881856540084389, | |
| "grad_norm": 0.46292224526405334, | |
| "learning_rate": 6.531762408906607e-05, | |
| "loss": 1.2858, | |
| "step": 8420 | |
| }, | |
| { | |
| "epoch": 0.8892405063291139, | |
| "grad_norm": 0.4787479639053345, | |
| "learning_rate": 6.441731983183912e-05, | |
| "loss": 1.2826, | |
| "step": 8430 | |
| }, | |
| { | |
| "epoch": 0.890295358649789, | |
| "grad_norm": 0.46926361322402954, | |
| "learning_rate": 6.352942490160292e-05, | |
| "loss": 1.2751, | |
| "step": 8440 | |
| }, | |
| { | |
| "epoch": 0.8913502109704642, | |
| "grad_norm": 0.5352609157562256, | |
| "learning_rate": 6.265376825462966e-05, | |
| "loss": 1.2856, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 0.8924050632911392, | |
| "grad_norm": 0.536734402179718, | |
| "learning_rate": 6.179018120476945e-05, | |
| "loss": 1.2794, | |
| "step": 8460 | |
| }, | |
| { | |
| "epoch": 0.8934599156118144, | |
| "grad_norm": 0.44269171357154846, | |
| "learning_rate": 6.0938497390954946e-05, | |
| "loss": 1.2805, | |
| "step": 8470 | |
| }, | |
| { | |
| "epoch": 0.8945147679324894, | |
| "grad_norm": 0.4609244763851166, | |
| "learning_rate": 6.009855274515339e-05, | |
| "loss": 1.2765, | |
| "step": 8480 | |
| }, | |
| { | |
| "epoch": 0.8955696202531646, | |
| "grad_norm": 0.48474302887916565, | |
| "learning_rate": 5.9270185460760735e-05, | |
| "loss": 1.281, | |
| "step": 8490 | |
| }, | |
| { | |
| "epoch": 0.8966244725738397, | |
| "grad_norm": 0.45089036226272583, | |
| "learning_rate": 5.8453235961431225e-05, | |
| "loss": 1.2712, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.8976793248945147, | |
| "grad_norm": 0.4711897671222687, | |
| "learning_rate": 5.764754687033678e-05, | |
| "loss": 1.2824, | |
| "step": 8510 | |
| }, | |
| { | |
| "epoch": 0.8987341772151899, | |
| "grad_norm": 0.47829535603523254, | |
| "learning_rate": 5.6852962979849836e-05, | |
| "loss": 1.2769, | |
| "step": 8520 | |
| }, | |
| { | |
| "epoch": 0.8997890295358649, | |
| "grad_norm": 0.4775094985961914, | |
| "learning_rate": 5.6069331221644284e-05, | |
| "loss": 1.2932, | |
| "step": 8530 | |
| }, | |
| { | |
| "epoch": 0.9008438818565401, | |
| "grad_norm": 0.4563244581222534, | |
| "learning_rate": 5.529650063720842e-05, | |
| "loss": 1.2872, | |
| "step": 8540 | |
| }, | |
| { | |
| "epoch": 0.9018987341772152, | |
| "grad_norm": 0.4800124764442444, | |
| "learning_rate": 5.453432234876445e-05, | |
| "loss": 1.2797, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 0.9029535864978903, | |
| "grad_norm": 0.46603870391845703, | |
| "learning_rate": 5.37826495305886e-05, | |
| "loss": 1.272, | |
| "step": 8560 | |
| }, | |
| { | |
| "epoch": 0.9040084388185654, | |
| "grad_norm": 0.5261629223823547, | |
| "learning_rate": 5.304133738072674e-05, | |
| "loss": 1.2907, | |
| "step": 8570 | |
| }, | |
| { | |
| "epoch": 0.9050632911392406, | |
| "grad_norm": 0.47402966022491455, | |
| "learning_rate": 5.2310243093099814e-05, | |
| "loss": 1.2805, | |
| "step": 8580 | |
| }, | |
| { | |
| "epoch": 0.9061181434599156, | |
| "grad_norm": 0.5603070259094238, | |
| "learning_rate": 5.158922582999368e-05, | |
| "loss": 1.2857, | |
| "step": 8590 | |
| }, | |
| { | |
| "epoch": 0.9071729957805907, | |
| "grad_norm": 0.4486725628376007, | |
| "learning_rate": 5.087814669492819e-05, | |
| "loss": 1.2796, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.9082278481012658, | |
| "grad_norm": 0.4944787919521332, | |
| "learning_rate": 5.017686870590028e-05, | |
| "loss": 1.2788, | |
| "step": 8610 | |
| }, | |
| { | |
| "epoch": 0.9092827004219409, | |
| "grad_norm": 0.4802660048007965, | |
| "learning_rate": 4.948525676899577e-05, | |
| "loss": 1.2749, | |
| "step": 8620 | |
| }, | |
| { | |
| "epoch": 0.9103375527426161, | |
| "grad_norm": 0.48038867115974426, | |
| "learning_rate": 4.880317765236493e-05, | |
| "loss": 1.2865, | |
| "step": 8630 | |
| }, | |
| { | |
| "epoch": 0.9113924050632911, | |
| "grad_norm": 0.5557894706726074, | |
| "learning_rate": 4.8130499960556755e-05, | |
| "loss": 1.2679, | |
| "step": 8640 | |
| }, | |
| { | |
| "epoch": 0.9124472573839663, | |
| "grad_norm": 0.4686918258666992, | |
| "learning_rate": 4.746709410920699e-05, | |
| "loss": 1.2718, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 0.9135021097046413, | |
| "grad_norm": 0.5099261999130249, | |
| "learning_rate": 4.681283230007507e-05, | |
| "loss": 1.2728, | |
| "step": 8660 | |
| }, | |
| { | |
| "epoch": 0.9145569620253164, | |
| "grad_norm": 0.508106529712677, | |
| "learning_rate": 4.616758849642509e-05, | |
| "loss": 1.2793, | |
| "step": 8670 | |
| }, | |
| { | |
| "epoch": 0.9156118143459916, | |
| "grad_norm": 0.4402479827404022, | |
| "learning_rate": 4.553123839874615e-05, | |
| "loss": 1.2907, | |
| "step": 8680 | |
| }, | |
| { | |
| "epoch": 0.9166666666666666, | |
| "grad_norm": 0.48903051018714905, | |
| "learning_rate": 4.490365942080736e-05, | |
| "loss": 1.2817, | |
| "step": 8690 | |
| }, | |
| { | |
| "epoch": 0.9177215189873418, | |
| "grad_norm": 0.45789381861686707, | |
| "learning_rate": 4.428473066604285e-05, | |
| "loss": 1.2809, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.9187763713080169, | |
| "grad_norm": 0.4754452407360077, | |
| "learning_rate": 4.367433290426233e-05, | |
| "loss": 1.2802, | |
| "step": 8710 | |
| }, | |
| { | |
| "epoch": 0.919831223628692, | |
| "grad_norm": 0.4568568468093872, | |
| "learning_rate": 4.3072348548682595e-05, | |
| "loss": 1.2731, | |
| "step": 8720 | |
| }, | |
| { | |
| "epoch": 0.9208860759493671, | |
| "grad_norm": 0.4637247622013092, | |
| "learning_rate": 4.247866163327575e-05, | |
| "loss": 1.2856, | |
| "step": 8730 | |
| }, | |
| { | |
| "epoch": 0.9219409282700421, | |
| "grad_norm": 0.4735919237136841, | |
| "learning_rate": 4.1893157790429404e-05, | |
| "loss": 1.273, | |
| "step": 8740 | |
| }, | |
| { | |
| "epoch": 0.9229957805907173, | |
| "grad_norm": 0.4588661789894104, | |
| "learning_rate": 4.1315724228915066e-05, | |
| "loss": 1.2726, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 0.9240506329113924, | |
| "grad_norm": 0.46978574991226196, | |
| "learning_rate": 4.074624971216005e-05, | |
| "loss": 1.2616, | |
| "step": 8760 | |
| }, | |
| { | |
| "epoch": 0.9251054852320675, | |
| "grad_norm": 0.4512912333011627, | |
| "learning_rate": 4.018462453681889e-05, | |
| "loss": 1.2716, | |
| "step": 8770 | |
| }, | |
| { | |
| "epoch": 0.9261603375527426, | |
| "grad_norm": 0.45371580123901367, | |
| "learning_rate": 3.963074051164014e-05, | |
| "loss": 1.2799, | |
| "step": 8780 | |
| }, | |
| { | |
| "epoch": 0.9272151898734177, | |
| "grad_norm": 0.4601062834262848, | |
| "learning_rate": 3.908449093662446e-05, | |
| "loss": 1.2732, | |
| "step": 8790 | |
| }, | |
| { | |
| "epoch": 0.9282700421940928, | |
| "grad_norm": 0.461967408657074, | |
| "learning_rate": 3.854577058246998e-05, | |
| "loss": 1.2708, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.929324894514768, | |
| "grad_norm": 0.45294222235679626, | |
| "learning_rate": 3.801447567030094e-05, | |
| "loss": 1.2897, | |
| "step": 8810 | |
| }, | |
| { | |
| "epoch": 0.930379746835443, | |
| "grad_norm": 0.4439496397972107, | |
| "learning_rate": 3.7490503851675777e-05, | |
| "loss": 1.2757, | |
| "step": 8820 | |
| }, | |
| { | |
| "epoch": 0.9314345991561181, | |
| "grad_norm": 0.4653451442718506, | |
| "learning_rate": 3.6973754188870806e-05, | |
| "loss": 1.2939, | |
| "step": 8830 | |
| }, | |
| { | |
| "epoch": 0.9324894514767933, | |
| "grad_norm": 0.4448198676109314, | |
| "learning_rate": 3.6464127135435536e-05, | |
| "loss": 1.2775, | |
| "step": 8840 | |
| }, | |
| { | |
| "epoch": 0.9335443037974683, | |
| "grad_norm": 0.46878382563591003, | |
| "learning_rate": 3.596152451701616e-05, | |
| "loss": 1.2777, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 0.9345991561181435, | |
| "grad_norm": 0.5037729144096375, | |
| "learning_rate": 3.5465849512443226e-05, | |
| "loss": 1.2766, | |
| "step": 8860 | |
| }, | |
| { | |
| "epoch": 0.9356540084388185, | |
| "grad_norm": 0.46913567185401917, | |
| "learning_rate": 3.4977006635080086e-05, | |
| "loss": 1.2801, | |
| "step": 8870 | |
| }, | |
| { | |
| "epoch": 0.9367088607594937, | |
| "grad_norm": 0.5404587388038635, | |
| "learning_rate": 3.449490171442838e-05, | |
| "loss": 1.282, | |
| "step": 8880 | |
| }, | |
| { | |
| "epoch": 0.9377637130801688, | |
| "grad_norm": 0.4438877999782562, | |
| "learning_rate": 3.401944187798702e-05, | |
| "loss": 1.283, | |
| "step": 8890 | |
| }, | |
| { | |
| "epoch": 0.9388185654008439, | |
| "grad_norm": 0.45989537239074707, | |
| "learning_rate": 3.355053553336137e-05, | |
| "loss": 1.2742, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.939873417721519, | |
| "grad_norm": 0.4791684150695801, | |
| "learning_rate": 3.308809235061882e-05, | |
| "loss": 1.272, | |
| "step": 8910 | |
| }, | |
| { | |
| "epoch": 0.9409282700421941, | |
| "grad_norm": 0.4399260878562927, | |
| "learning_rate": 3.263202324488772e-05, | |
| "loss": 1.2796, | |
| "step": 8920 | |
| }, | |
| { | |
| "epoch": 0.9419831223628692, | |
| "grad_norm": 0.4575617015361786, | |
| "learning_rate": 3.218224035919609e-05, | |
| "loss": 1.2766, | |
| "step": 8930 | |
| }, | |
| { | |
| "epoch": 0.9430379746835443, | |
| "grad_norm": 0.4506317377090454, | |
| "learning_rate": 3.173865704754688e-05, | |
| "loss": 1.2765, | |
| "step": 8940 | |
| }, | |
| { | |
| "epoch": 0.9440928270042194, | |
| "grad_norm": 0.45531153678894043, | |
| "learning_rate": 3.130118785822657e-05, | |
| "loss": 1.2748, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 0.9451476793248945, | |
| "grad_norm": 0.4633644223213196, | |
| "learning_rate": 3.08697485173437e-05, | |
| "loss": 1.2816, | |
| "step": 8960 | |
| }, | |
| { | |
| "epoch": 0.9462025316455697, | |
| "grad_norm": 0.5337445735931396, | |
| "learning_rate": 3.0444255912594442e-05, | |
| "loss": 1.2772, | |
| "step": 8970 | |
| }, | |
| { | |
| "epoch": 0.9472573839662447, | |
| "grad_norm": 0.5188563466072083, | |
| "learning_rate": 3.002462807725185e-05, | |
| "loss": 1.275, | |
| "step": 8980 | |
| }, | |
| { | |
| "epoch": 0.9483122362869199, | |
| "grad_norm": 0.48296037316322327, | |
| "learning_rate": 2.9610784174375868e-05, | |
| "loss": 1.2797, | |
| "step": 8990 | |
| }, | |
| { | |
| "epoch": 0.9493670886075949, | |
| "grad_norm": 0.46105465292930603, | |
| "learning_rate": 2.920264448124087e-05, | |
| "loss": 1.2758, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.95042194092827, | |
| "grad_norm": 0.45554500818252563, | |
| "learning_rate": 2.8800130373977936e-05, | |
| "loss": 1.2726, | |
| "step": 9010 | |
| }, | |
| { | |
| "epoch": 0.9514767932489452, | |
| "grad_norm": 0.4668406844139099, | |
| "learning_rate": 2.84031643124288e-05, | |
| "loss": 1.2734, | |
| "step": 9020 | |
| }, | |
| { | |
| "epoch": 0.9525316455696202, | |
| "grad_norm": 0.4780922532081604, | |
| "learning_rate": 2.8011669825208517e-05, | |
| "loss": 1.2923, | |
| "step": 9030 | |
| }, | |
| { | |
| "epoch": 0.9535864978902954, | |
| "grad_norm": 0.4693585932254791, | |
| "learning_rate": 2.762557149497405e-05, | |
| "loss": 1.2676, | |
| "step": 9040 | |
| }, | |
| { | |
| "epoch": 0.9546413502109705, | |
| "grad_norm": 0.4524311423301697, | |
| "learning_rate": 2.724479494389592e-05, | |
| "loss": 1.2813, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 0.9556962025316456, | |
| "grad_norm": 0.44265279173851013, | |
| "learning_rate": 2.6869266819330058e-05, | |
| "loss": 1.2797, | |
| "step": 9060 | |
| }, | |
| { | |
| "epoch": 0.9567510548523207, | |
| "grad_norm": 0.4572458863258362, | |
| "learning_rate": 2.6498914779687228e-05, | |
| "loss": 1.2814, | |
| "step": 9070 | |
| }, | |
| { | |
| "epoch": 0.9578059071729957, | |
| "grad_norm": 0.4525168836116791, | |
| "learning_rate": 2.6133667480497115e-05, | |
| "loss": 1.2802, | |
| "step": 9080 | |
| }, | |
| { | |
| "epoch": 0.9588607594936709, | |
| "grad_norm": 0.4815879464149475, | |
| "learning_rate": 2.5773454560664597e-05, | |
| "loss": 1.2822, | |
| "step": 9090 | |
| }, | |
| { | |
| "epoch": 0.959915611814346, | |
| "grad_norm": 0.44361522793769836, | |
| "learning_rate": 2.541820662891541e-05, | |
| "loss": 1.2632, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.9609704641350211, | |
| "grad_norm": 0.4554971754550934, | |
| "learning_rate": 2.5067855250428616e-05, | |
| "loss": 1.2706, | |
| "step": 9110 | |
| }, | |
| { | |
| "epoch": 0.9620253164556962, | |
| "grad_norm": 0.520605742931366, | |
| "learning_rate": 2.472233293365335e-05, | |
| "loss": 1.2838, | |
| "step": 9120 | |
| }, | |
| { | |
| "epoch": 0.9630801687763713, | |
| "grad_norm": 0.4688946604728699, | |
| "learning_rate": 2.4381573117307307e-05, | |
| "loss": 1.268, | |
| "step": 9130 | |
| }, | |
| { | |
| "epoch": 0.9641350210970464, | |
| "grad_norm": 0.5332078337669373, | |
| "learning_rate": 2.4045510157554362e-05, | |
| "loss": 1.2868, | |
| "step": 9140 | |
| }, | |
| { | |
| "epoch": 0.9651898734177216, | |
| "grad_norm": 0.47867295145988464, | |
| "learning_rate": 2.3714079315358985e-05, | |
| "loss": 1.2729, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 0.9662447257383966, | |
| "grad_norm": 0.46423396468162537, | |
| "learning_rate": 2.338721674401494e-05, | |
| "loss": 1.2813, | |
| "step": 9160 | |
| }, | |
| { | |
| "epoch": 0.9672995780590717, | |
| "grad_norm": 0.46052441000938416, | |
| "learning_rate": 2.30648594768459e-05, | |
| "loss": 1.2796, | |
| "step": 9170 | |
| }, | |
| { | |
| "epoch": 0.9683544303797469, | |
| "grad_norm": 0.4625004529953003, | |
| "learning_rate": 2.2746945415075523e-05, | |
| "loss": 1.276, | |
| "step": 9180 | |
| }, | |
| { | |
| "epoch": 0.9694092827004219, | |
| "grad_norm": 0.46688348054885864, | |
| "learning_rate": 2.2433413315864803e-05, | |
| "loss": 1.2731, | |
| "step": 9190 | |
| }, | |
| { | |
| "epoch": 0.9704641350210971, | |
| "grad_norm": 0.4896974563598633, | |
| "learning_rate": 2.2124202780514277e-05, | |
| "loss": 1.2757, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.9715189873417721, | |
| "grad_norm": 0.4933200180530548, | |
| "learning_rate": 2.1819254242828815e-05, | |
| "loss": 1.2764, | |
| "step": 9210 | |
| }, | |
| { | |
| "epoch": 0.9725738396624473, | |
| "grad_norm": 0.4900830388069153, | |
| "learning_rate": 2.151850895764285e-05, | |
| "loss": 1.2717, | |
| "step": 9220 | |
| }, | |
| { | |
| "epoch": 0.9736286919831224, | |
| "grad_norm": 0.45735400915145874, | |
| "learning_rate": 2.12219089895037e-05, | |
| "loss": 1.2646, | |
| "step": 9230 | |
| }, | |
| { | |
| "epoch": 0.9746835443037974, | |
| "grad_norm": 0.48359930515289307, | |
| "learning_rate": 2.092939720151092e-05, | |
| "loss": 1.2612, | |
| "step": 9240 | |
| }, | |
| { | |
| "epoch": 0.9757383966244726, | |
| "grad_norm": 0.4569590985774994, | |
| "learning_rate": 2.064091724430947e-05, | |
| "loss": 1.2719, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 0.9767932489451476, | |
| "grad_norm": 0.47614622116088867, | |
| "learning_rate": 2.0356413545234603e-05, | |
| "loss": 1.2671, | |
| "step": 9260 | |
| }, | |
| { | |
| "epoch": 0.9778481012658228, | |
| "grad_norm": 0.48315489292144775, | |
| "learning_rate": 2.0075831297606357e-05, | |
| "loss": 1.2791, | |
| "step": 9270 | |
| }, | |
| { | |
| "epoch": 0.9789029535864979, | |
| "grad_norm": 0.4540780484676361, | |
| "learning_rate": 1.9799116450171627e-05, | |
| "loss": 1.2692, | |
| "step": 9280 | |
| }, | |
| { | |
| "epoch": 0.979957805907173, | |
| "grad_norm": 0.465520977973938, | |
| "learning_rate": 1.952621569669175e-05, | |
| "loss": 1.2733, | |
| "step": 9290 | |
| }, | |
| { | |
| "epoch": 0.9810126582278481, | |
| "grad_norm": 0.4556669294834137, | |
| "learning_rate": 1.9257076465673605e-05, | |
| "loss": 1.2719, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.9820675105485233, | |
| "grad_norm": 0.46988987922668457, | |
| "learning_rate": 1.899164691024229e-05, | |
| "loss": 1.2741, | |
| "step": 9310 | |
| }, | |
| { | |
| "epoch": 0.9831223628691983, | |
| "grad_norm": 0.48423823714256287, | |
| "learning_rate": 1.872987589815331e-05, | |
| "loss": 1.2691, | |
| "step": 9320 | |
| }, | |
| { | |
| "epoch": 0.9841772151898734, | |
| "grad_norm": 0.4487670063972473, | |
| "learning_rate": 1.8471713001942538e-05, | |
| "loss": 1.2821, | |
| "step": 9330 | |
| }, | |
| { | |
| "epoch": 0.9852320675105485, | |
| "grad_norm": 0.4768301546573639, | |
| "learning_rate": 1.8217108489211845e-05, | |
| "loss": 1.2784, | |
| "step": 9340 | |
| }, | |
| { | |
| "epoch": 0.9862869198312236, | |
| "grad_norm": 0.462040513753891, | |
| "learning_rate": 1.7966013313048696e-05, | |
| "loss": 1.2883, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 0.9873417721518988, | |
| "grad_norm": 0.4433370530605316, | |
| "learning_rate": 1.7718379102577752e-05, | |
| "loss": 1.2808, | |
| "step": 9360 | |
| }, | |
| { | |
| "epoch": 0.9883966244725738, | |
| "grad_norm": 0.4689382314682007, | |
| "learning_rate": 1.7474158153642745e-05, | |
| "loss": 1.2792, | |
| "step": 9370 | |
| }, | |
| { | |
| "epoch": 0.989451476793249, | |
| "grad_norm": 0.4566853940486908, | |
| "learning_rate": 1.7233303419616745e-05, | |
| "loss": 1.2648, | |
| "step": 9380 | |
| }, | |
| { | |
| "epoch": 0.990506329113924, | |
| "grad_norm": 0.48965659737586975, | |
| "learning_rate": 1.699576850233916e-05, | |
| "loss": 1.27, | |
| "step": 9390 | |
| }, | |
| { | |
| "epoch": 0.9915611814345991, | |
| "grad_norm": 0.45181959867477417, | |
| "learning_rate": 1.6761507643177553e-05, | |
| "loss": 1.2827, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.9926160337552743, | |
| "grad_norm": 0.5077662467956543, | |
| "learning_rate": 1.6530475714212752e-05, | |
| "loss": 1.2588, | |
| "step": 9410 | |
| }, | |
| { | |
| "epoch": 0.9936708860759493, | |
| "grad_norm": 0.4643486440181732, | |
| "learning_rate": 1.6302628209545423e-05, | |
| "loss": 1.2787, | |
| "step": 9420 | |
| }, | |
| { | |
| "epoch": 0.9947257383966245, | |
| "grad_norm": 0.4604697525501251, | |
| "learning_rate": 1.6077921236722464e-05, | |
| "loss": 1.2783, | |
| "step": 9430 | |
| }, | |
| { | |
| "epoch": 0.9957805907172996, | |
| "grad_norm": 0.44910043478012085, | |
| "learning_rate": 1.5856311508281594e-05, | |
| "loss": 1.2722, | |
| "step": 9440 | |
| }, | |
| { | |
| "epoch": 0.9968354430379747, | |
| "grad_norm": 0.45268765091896057, | |
| "learning_rate": 1.5637756333412454e-05, | |
| "loss": 1.2814, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 0.9978902953586498, | |
| "grad_norm": 0.4484785199165344, | |
| "learning_rate": 1.542221360973268e-05, | |
| "loss": 1.2664, | |
| "step": 9460 | |
| }, | |
| { | |
| "epoch": 0.9989451476793249, | |
| "grad_norm": 0.4483862519264221, | |
| "learning_rate": 1.5209641815177312e-05, | |
| "loss": 1.2798, | |
| "step": 9470 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 1.4878731966018677, | |
| "learning_rate": 1.5e-05, | |
| "loss": 1.2718, | |
| "step": 9480 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 9480, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.832308198648013e+16, | |
| "train_batch_size": 1024, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |