| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.9996064541519087, |
| "eval_steps": 500, |
| "global_step": 1270, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0007870916961826052, |
| "grad_norm": 4.882866791973475, |
| "learning_rate": 7.8125e-08, |
| "loss": 0.357, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.0015741833923652105, |
| "grad_norm": 4.89981767485179, |
| "learning_rate": 1.5625e-07, |
| "loss": 0.3398, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.0023612750885478157, |
| "grad_norm": 4.908915346462736, |
| "learning_rate": 2.3437500000000003e-07, |
| "loss": 0.3326, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.003148366784730421, |
| "grad_norm": 4.731649267914947, |
| "learning_rate": 3.125e-07, |
| "loss": 0.342, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.003935458480913027, |
| "grad_norm": 4.769633474207938, |
| "learning_rate": 3.90625e-07, |
| "loss": 0.3431, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.004722550177095631, |
| "grad_norm": 4.754569879633701, |
| "learning_rate": 4.6875000000000006e-07, |
| "loss": 0.3369, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.005509641873278237, |
| "grad_norm": 4.354074850343827, |
| "learning_rate": 5.468750000000001e-07, |
| "loss": 0.3416, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.006296733569460842, |
| "grad_norm": 4.51384309102365, |
| "learning_rate": 6.25e-07, |
| "loss": 0.3577, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.0070838252656434475, |
| "grad_norm": 4.127868399899779, |
| "learning_rate": 7.03125e-07, |
| "loss": 0.3447, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.007870916961826053, |
| "grad_norm": 4.268670776824985, |
| "learning_rate": 7.8125e-07, |
| "loss": 0.318, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.008658008658008658, |
| "grad_norm": 3.7083716156028674, |
| "learning_rate": 8.59375e-07, |
| "loss": 0.3106, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.009445100354191263, |
| "grad_norm": 3.1545864445099263, |
| "learning_rate": 9.375000000000001e-07, |
| "loss": 0.3131, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.01023219205037387, |
| "grad_norm": 3.020045714405798, |
| "learning_rate": 1.0156250000000001e-06, |
| "loss": 0.3215, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.011019283746556474, |
| "grad_norm": 2.748876831681126, |
| "learning_rate": 1.0937500000000001e-06, |
| "loss": 0.3072, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.011806375442739079, |
| "grad_norm": 2.2307366833759485, |
| "learning_rate": 1.1718750000000001e-06, |
| "loss": 0.2922, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.012593467138921684, |
| "grad_norm": 2.219422516987874, |
| "learning_rate": 1.25e-06, |
| "loss": 0.2842, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.01338055883510429, |
| "grad_norm": 2.7073031779339973, |
| "learning_rate": 1.328125e-06, |
| "loss": 0.2674, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.014167650531286895, |
| "grad_norm": 2.873035911017537, |
| "learning_rate": 1.40625e-06, |
| "loss": 0.294, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.0149547422274695, |
| "grad_norm": 2.119880363778339, |
| "learning_rate": 1.484375e-06, |
| "loss": 0.2693, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.015741833923652106, |
| "grad_norm": 1.7740660860958901, |
| "learning_rate": 1.5625e-06, |
| "loss": 0.2607, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.01652892561983471, |
| "grad_norm": 1.654838099179133, |
| "learning_rate": 1.640625e-06, |
| "loss": 0.2539, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.017316017316017316, |
| "grad_norm": 2.1067372096520884, |
| "learning_rate": 1.71875e-06, |
| "loss": 0.274, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.01810310901219992, |
| "grad_norm": 2.227492365997846, |
| "learning_rate": 1.796875e-06, |
| "loss": 0.2667, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.018890200708382526, |
| "grad_norm": 1.9818482942437547, |
| "learning_rate": 1.8750000000000003e-06, |
| "loss": 0.2507, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.01967729240456513, |
| "grad_norm": 1.9916019664977938, |
| "learning_rate": 1.953125e-06, |
| "loss": 0.2302, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.02046438410074774, |
| "grad_norm": 2.0987871479467533, |
| "learning_rate": 2.0312500000000002e-06, |
| "loss": 0.2563, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.021251475796930343, |
| "grad_norm": 1.7851505967742112, |
| "learning_rate": 2.109375e-06, |
| "loss": 0.2432, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.02203856749311295, |
| "grad_norm": 1.6067598902195293, |
| "learning_rate": 2.1875000000000002e-06, |
| "loss": 0.2466, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.022825659189295553, |
| "grad_norm": 1.436243142469347, |
| "learning_rate": 2.265625e-06, |
| "loss": 0.2486, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.023612750885478158, |
| "grad_norm": 1.631080710695958, |
| "learning_rate": 2.3437500000000002e-06, |
| "loss": 0.2692, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.024399842581660763, |
| "grad_norm": 1.42554302342302, |
| "learning_rate": 2.421875e-06, |
| "loss": 0.2374, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.025186934277843367, |
| "grad_norm": 1.479794666013743, |
| "learning_rate": 2.5e-06, |
| "loss": 0.238, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.025974025974025976, |
| "grad_norm": 1.3857185652178832, |
| "learning_rate": 2.5781250000000004e-06, |
| "loss": 0.2366, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.02676111767020858, |
| "grad_norm": 1.335993998237778, |
| "learning_rate": 2.65625e-06, |
| "loss": 0.2251, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.027548209366391185, |
| "grad_norm": 1.5950255189913525, |
| "learning_rate": 2.7343750000000004e-06, |
| "loss": 0.2506, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.02833530106257379, |
| "grad_norm": 1.3773024411686232, |
| "learning_rate": 2.8125e-06, |
| "loss": 0.2107, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.029122392758756395, |
| "grad_norm": 1.391558709917223, |
| "learning_rate": 2.8906250000000004e-06, |
| "loss": 0.236, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.029909484454939, |
| "grad_norm": 1.4317153691023394, |
| "learning_rate": 2.96875e-06, |
| "loss": 0.2369, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.030696576151121605, |
| "grad_norm": 1.5982850208202695, |
| "learning_rate": 3.0468750000000004e-06, |
| "loss": 0.2135, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.03148366784730421, |
| "grad_norm": 1.2463748947443163, |
| "learning_rate": 3.125e-06, |
| "loss": 0.2109, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.032270759543486814, |
| "grad_norm": 1.489195263138514, |
| "learning_rate": 3.2031250000000004e-06, |
| "loss": 0.2254, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.03305785123966942, |
| "grad_norm": 1.1952984228039816, |
| "learning_rate": 3.28125e-06, |
| "loss": 0.2124, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.033844942935852024, |
| "grad_norm": 1.3331698750786545, |
| "learning_rate": 3.3593750000000003e-06, |
| "loss": 0.2192, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.03463203463203463, |
| "grad_norm": 1.3944936006633961, |
| "learning_rate": 3.4375e-06, |
| "loss": 0.2024, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.03541912632821724, |
| "grad_norm": 1.3992213437238004, |
| "learning_rate": 3.5156250000000003e-06, |
| "loss": 0.2274, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.03620621802439984, |
| "grad_norm": 1.3664016160053327, |
| "learning_rate": 3.59375e-06, |
| "loss": 0.2152, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.03699330972058245, |
| "grad_norm": 1.4891884814728509, |
| "learning_rate": 3.6718750000000003e-06, |
| "loss": 0.2292, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.03778040141676505, |
| "grad_norm": 1.3270512221194979, |
| "learning_rate": 3.7500000000000005e-06, |
| "loss": 0.1997, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.03856749311294766, |
| "grad_norm": 1.4002885202427642, |
| "learning_rate": 3.828125000000001e-06, |
| "loss": 0.2049, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.03935458480913026, |
| "grad_norm": 1.3129491965923514, |
| "learning_rate": 3.90625e-06, |
| "loss": 0.2108, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.04014167650531287, |
| "grad_norm": 1.5308223134960726, |
| "learning_rate": 3.984375e-06, |
| "loss": 0.2185, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.04092876820149548, |
| "grad_norm": 1.4637864330743848, |
| "learning_rate": 4.0625000000000005e-06, |
| "loss": 0.1959, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.04171585989767808, |
| "grad_norm": 1.3727993193826824, |
| "learning_rate": 4.140625000000001e-06, |
| "loss": 0.2079, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.04250295159386069, |
| "grad_norm": 1.324719262777532, |
| "learning_rate": 4.21875e-06, |
| "loss": 0.1986, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.04329004329004329, |
| "grad_norm": 1.5169702547195179, |
| "learning_rate": 4.296875e-06, |
| "loss": 0.2233, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.0440771349862259, |
| "grad_norm": 1.2762861570952524, |
| "learning_rate": 4.3750000000000005e-06, |
| "loss": 0.1887, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.0448642266824085, |
| "grad_norm": 1.3276694806352698, |
| "learning_rate": 4.453125000000001e-06, |
| "loss": 0.2039, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.045651318378591106, |
| "grad_norm": 1.3146250055299598, |
| "learning_rate": 4.53125e-06, |
| "loss": 0.2142, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.046438410074773714, |
| "grad_norm": 1.333854988794211, |
| "learning_rate": 4.609375e-06, |
| "loss": 0.1999, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.047225501770956316, |
| "grad_norm": 1.3581177171903593, |
| "learning_rate": 4.6875000000000004e-06, |
| "loss": 0.1957, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.048012593467138924, |
| "grad_norm": 1.3130565355707189, |
| "learning_rate": 4.765625000000001e-06, |
| "loss": 0.2122, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.048799685163321525, |
| "grad_norm": 1.4142816132337854, |
| "learning_rate": 4.84375e-06, |
| "loss": 0.2147, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.049586776859504134, |
| "grad_norm": 1.357898232243354, |
| "learning_rate": 4.921875e-06, |
| "loss": 0.1877, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.050373868555686735, |
| "grad_norm": 1.4153342064426397, |
| "learning_rate": 5e-06, |
| "loss": 0.1945, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.05116096025186934, |
| "grad_norm": 1.4495342715013748, |
| "learning_rate": 4.999991517675219e-06, |
| "loss": 0.1939, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.05194805194805195, |
| "grad_norm": 1.1539274129121713, |
| "learning_rate": 4.999966070758437e-06, |
| "loss": 0.2003, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.05273514364423455, |
| "grad_norm": 1.3379283904444008, |
| "learning_rate": 4.999923659422332e-06, |
| "loss": 0.2007, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.05352223534041716, |
| "grad_norm": 1.3492954613335875, |
| "learning_rate": 4.999864283954702e-06, |
| "loss": 0.1989, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.05430932703659976, |
| "grad_norm": 1.1801853129144864, |
| "learning_rate": 4.99978794475846e-06, |
| "loss": 0.2114, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.05509641873278237, |
| "grad_norm": 1.2068999367428581, |
| "learning_rate": 4.999694642351633e-06, |
| "loss": 0.2033, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.05588351042896497, |
| "grad_norm": 1.2287271472480104, |
| "learning_rate": 4.999584377367359e-06, |
| "loss": 0.1895, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.05667060212514758, |
| "grad_norm": 1.3129837217534652, |
| "learning_rate": 4.99945715055388e-06, |
| "loss": 0.1905, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.05745769382133018, |
| "grad_norm": 1.1734967025843308, |
| "learning_rate": 4.99931296277454e-06, |
| "loss": 0.213, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.05824478551751279, |
| "grad_norm": 1.3738466570011791, |
| "learning_rate": 4.999151815007776e-06, |
| "loss": 0.2214, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.0590318772136954, |
| "grad_norm": 1.273179655688277, |
| "learning_rate": 4.9989737083471165e-06, |
| "loss": 0.1894, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.059818968909878, |
| "grad_norm": 1.0843431120214646, |
| "learning_rate": 4.998778644001165e-06, |
| "loss": 0.1967, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.06060606060606061, |
| "grad_norm": 1.4896402431576707, |
| "learning_rate": 4.998566623293603e-06, |
| "loss": 0.1752, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.06139315230224321, |
| "grad_norm": 1.3405458603738243, |
| "learning_rate": 4.9983376476631725e-06, |
| "loss": 0.1998, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.06218024399842582, |
| "grad_norm": 1.3641086369593634, |
| "learning_rate": 4.998091718663671e-06, |
| "loss": 0.2047, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.06296733569460843, |
| "grad_norm": 1.3391162585136267, |
| "learning_rate": 4.997828837963937e-06, |
| "loss": 0.181, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.06375442739079103, |
| "grad_norm": 1.1899411991269295, |
| "learning_rate": 4.997549007347843e-06, |
| "loss": 0.1946, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.06454151908697363, |
| "grad_norm": 1.3917818646896112, |
| "learning_rate": 4.997252228714279e-06, |
| "loss": 0.1919, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.06532861078315624, |
| "grad_norm": 1.2543099071691322, |
| "learning_rate": 4.996938504077145e-06, |
| "loss": 0.1948, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.06611570247933884, |
| "grad_norm": 1.3941008619735185, |
| "learning_rate": 4.99660783556533e-06, |
| "loss": 0.1861, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.06690279417552145, |
| "grad_norm": 1.1765528133487257, |
| "learning_rate": 4.9962602254227075e-06, |
| "loss": 0.1817, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.06768988587170405, |
| "grad_norm": 1.223066746932356, |
| "learning_rate": 4.995895676008109e-06, |
| "loss": 0.1934, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.06847697756788666, |
| "grad_norm": 1.3140944559909808, |
| "learning_rate": 4.995514189795316e-06, |
| "loss": 0.197, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.06926406926406926, |
| "grad_norm": 1.1819977914205286, |
| "learning_rate": 4.99511576937304e-06, |
| "loss": 0.1972, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.07005116096025187, |
| "grad_norm": 1.3152579578345207, |
| "learning_rate": 4.994700417444907e-06, |
| "loss": 0.207, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.07083825265643448, |
| "grad_norm": 1.2064669225701854, |
| "learning_rate": 4.994268136829438e-06, |
| "loss": 0.1953, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.07162534435261708, |
| "grad_norm": 1.1619755664518439, |
| "learning_rate": 4.993818930460026e-06, |
| "loss": 0.1982, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.07241243604879968, |
| "grad_norm": 1.1792837872493809, |
| "learning_rate": 4.993352801384924e-06, |
| "loss": 0.1886, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.07319952774498228, |
| "grad_norm": 1.097328306217708, |
| "learning_rate": 4.992869752767218e-06, |
| "loss": 0.1673, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.0739866194411649, |
| "grad_norm": 1.2788239338552108, |
| "learning_rate": 4.992369787884809e-06, |
| "loss": 0.1972, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.0747737111373475, |
| "grad_norm": 1.1905278770669998, |
| "learning_rate": 4.991852910130388e-06, |
| "loss": 0.1872, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.0755608028335301, |
| "grad_norm": 1.2133270115400816, |
| "learning_rate": 4.9913191230114154e-06, |
| "loss": 0.1748, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.07634789452971272, |
| "grad_norm": 1.2840440499091732, |
| "learning_rate": 4.990768430150096e-06, |
| "loss": 0.1942, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.07713498622589532, |
| "grad_norm": 1.5346248945491554, |
| "learning_rate": 4.990200835283353e-06, |
| "loss": 0.1861, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.07792207792207792, |
| "grad_norm": 1.1936205681426777, |
| "learning_rate": 4.989616342262807e-06, |
| "loss": 0.1975, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.07870916961826052, |
| "grad_norm": 1.2662437794316659, |
| "learning_rate": 4.989014955054746e-06, |
| "loss": 0.1853, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.07949626131444314, |
| "grad_norm": 1.116915072535967, |
| "learning_rate": 4.988396677740097e-06, |
| "loss": 0.1738, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.08028335301062574, |
| "grad_norm": 1.1577366023558335, |
| "learning_rate": 4.9877615145144055e-06, |
| "loss": 0.2045, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.08107044470680834, |
| "grad_norm": 1.1022178093801993, |
| "learning_rate": 4.9871094696878e-06, |
| "loss": 0.1814, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.08185753640299095, |
| "grad_norm": 1.218327314143879, |
| "learning_rate": 4.986440547684963e-06, |
| "loss": 0.1822, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.08264462809917356, |
| "grad_norm": 1.0747362510591434, |
| "learning_rate": 4.985754753045108e-06, |
| "loss": 0.1639, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.08343171979535616, |
| "grad_norm": 1.236686976609853, |
| "learning_rate": 4.9850520904219406e-06, |
| "loss": 0.1773, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.08421881149153876, |
| "grad_norm": 1.2843110878866029, |
| "learning_rate": 4.98433256458363e-06, |
| "loss": 0.1931, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.08500590318772137, |
| "grad_norm": 1.2556201190754803, |
| "learning_rate": 4.983596180412779e-06, |
| "loss": 0.1891, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.08579299488390398, |
| "grad_norm": 1.1736861180333642, |
| "learning_rate": 4.982842942906386e-06, |
| "loss": 0.1932, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.08658008658008658, |
| "grad_norm": 1.250703274500956, |
| "learning_rate": 4.982072857175816e-06, |
| "loss": 0.1979, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.08736717827626919, |
| "grad_norm": 1.111910462348759, |
| "learning_rate": 4.981285928446762e-06, |
| "loss": 0.1729, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.0881542699724518, |
| "grad_norm": 1.2267625409230847, |
| "learning_rate": 4.980482162059214e-06, |
| "loss": 0.1993, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.0889413616686344, |
| "grad_norm": 1.350342930816002, |
| "learning_rate": 4.979661563467415e-06, |
| "loss": 0.1914, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.089728453364817, |
| "grad_norm": 0.9837790085016399, |
| "learning_rate": 4.978824138239835e-06, |
| "loss": 0.1852, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.09051554506099961, |
| "grad_norm": 1.1412715811918805, |
| "learning_rate": 4.977969892059123e-06, |
| "loss": 0.1791, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.09130263675718221, |
| "grad_norm": 1.091735318231847, |
| "learning_rate": 4.977098830722074e-06, |
| "loss": 0.1879, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.09208972845336481, |
| "grad_norm": 1.1356995797773966, |
| "learning_rate": 4.976210960139587e-06, |
| "loss": 0.1942, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.09287682014954743, |
| "grad_norm": 1.197221158258512, |
| "learning_rate": 4.975306286336628e-06, |
| "loss": 0.1822, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.09366391184573003, |
| "grad_norm": 1.1622435205009634, |
| "learning_rate": 4.974384815452187e-06, |
| "loss": 0.1938, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.09445100354191263, |
| "grad_norm": 1.1980574826361372, |
| "learning_rate": 4.9734465537392365e-06, |
| "loss": 0.1703, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.09523809523809523, |
| "grad_norm": 1.090793092501407, |
| "learning_rate": 4.972491507564688e-06, |
| "loss": 0.1681, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.09602518693427785, |
| "grad_norm": 1.2120296842604672, |
| "learning_rate": 4.9715196834093525e-06, |
| "loss": 0.1562, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.09681227863046045, |
| "grad_norm": 1.1420618659168036, |
| "learning_rate": 4.97053108786789e-06, |
| "loss": 0.1812, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.09759937032664305, |
| "grad_norm": 1.145370838994205, |
| "learning_rate": 4.969525727648774e-06, |
| "loss": 0.1873, |
| "step": 124 |
| }, |
| { |
| "epoch": 0.09838646202282567, |
| "grad_norm": 1.1676600414602372, |
| "learning_rate": 4.9685036095742365e-06, |
| "loss": 0.1972, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.09917355371900827, |
| "grad_norm": 1.204479600477317, |
| "learning_rate": 4.967464740580228e-06, |
| "loss": 0.1904, |
| "step": 126 |
| }, |
| { |
| "epoch": 0.09996064541519087, |
| "grad_norm": 1.119994012971968, |
| "learning_rate": 4.9664091277163664e-06, |
| "loss": 0.1851, |
| "step": 127 |
| }, |
| { |
| "epoch": 0.10074773711137347, |
| "grad_norm": 1.1043684264734095, |
| "learning_rate": 4.9653367781458946e-06, |
| "loss": 0.1926, |
| "step": 128 |
| }, |
| { |
| "epoch": 0.10153482880755609, |
| "grad_norm": 1.127680976136701, |
| "learning_rate": 4.964247699145626e-06, |
| "loss": 0.1886, |
| "step": 129 |
| }, |
| { |
| "epoch": 0.10232192050373869, |
| "grad_norm": 1.1813875803533898, |
| "learning_rate": 4.963141898105898e-06, |
| "loss": 0.1858, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.10310901219992129, |
| "grad_norm": 1.1497128287458092, |
| "learning_rate": 4.962019382530521e-06, |
| "loss": 0.1724, |
| "step": 131 |
| }, |
| { |
| "epoch": 0.1038961038961039, |
| "grad_norm": 1.223498886081565, |
| "learning_rate": 4.960880160036728e-06, |
| "loss": 0.194, |
| "step": 132 |
| }, |
| { |
| "epoch": 0.1046831955922865, |
| "grad_norm": 1.1861652899170938, |
| "learning_rate": 4.959724238355124e-06, |
| "loss": 0.1841, |
| "step": 133 |
| }, |
| { |
| "epoch": 0.1054702872884691, |
| "grad_norm": 1.0805114288365025, |
| "learning_rate": 4.958551625329631e-06, |
| "loss": 0.1646, |
| "step": 134 |
| }, |
| { |
| "epoch": 0.10625737898465171, |
| "grad_norm": 1.29808710670669, |
| "learning_rate": 4.957362328917437e-06, |
| "loss": 0.1833, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.10704447068083432, |
| "grad_norm": 1.2660501691777906, |
| "learning_rate": 4.95615635718894e-06, |
| "loss": 0.1753, |
| "step": 136 |
| }, |
| { |
| "epoch": 0.10783156237701692, |
| "grad_norm": 1.1429230314494303, |
| "learning_rate": 4.954933718327697e-06, |
| "loss": 0.1734, |
| "step": 137 |
| }, |
| { |
| "epoch": 0.10861865407319952, |
| "grad_norm": 1.114357361335831, |
| "learning_rate": 4.953694420630361e-06, |
| "loss": 0.1925, |
| "step": 138 |
| }, |
| { |
| "epoch": 0.10940574576938213, |
| "grad_norm": 1.1238119767186239, |
| "learning_rate": 4.952438472506636e-06, |
| "loss": 0.1805, |
| "step": 139 |
| }, |
| { |
| "epoch": 0.11019283746556474, |
| "grad_norm": 1.1524735878912507, |
| "learning_rate": 4.951165882479206e-06, |
| "loss": 0.1783, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.11097992916174734, |
| "grad_norm": 1.0546198047284017, |
| "learning_rate": 4.949876659183693e-06, |
| "loss": 0.1745, |
| "step": 141 |
| }, |
| { |
| "epoch": 0.11176702085792994, |
| "grad_norm": 1.0925714956018635, |
| "learning_rate": 4.94857081136858e-06, |
| "loss": 0.1763, |
| "step": 142 |
| }, |
| { |
| "epoch": 0.11255411255411256, |
| "grad_norm": 1.1039385653204372, |
| "learning_rate": 4.947248347895172e-06, |
| "loss": 0.1777, |
| "step": 143 |
| }, |
| { |
| "epoch": 0.11334120425029516, |
| "grad_norm": 1.145622347104172, |
| "learning_rate": 4.945909277737519e-06, |
| "loss": 0.1804, |
| "step": 144 |
| }, |
| { |
| "epoch": 0.11412829594647776, |
| "grad_norm": 1.0810330697861197, |
| "learning_rate": 4.944553609982363e-06, |
| "loss": 0.18, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.11491538764266036, |
| "grad_norm": 1.079722871077113, |
| "learning_rate": 4.943181353829077e-06, |
| "loss": 0.1805, |
| "step": 146 |
| }, |
| { |
| "epoch": 0.11570247933884298, |
| "grad_norm": 1.2122723148500483, |
| "learning_rate": 4.941792518589596e-06, |
| "loss": 0.2113, |
| "step": 147 |
| }, |
| { |
| "epoch": 0.11648957103502558, |
| "grad_norm": 1.1619622709214918, |
| "learning_rate": 4.940387113688364e-06, |
| "loss": 0.1714, |
| "step": 148 |
| }, |
| { |
| "epoch": 0.11727666273120818, |
| "grad_norm": 1.0508760593348456, |
| "learning_rate": 4.93896514866226e-06, |
| "loss": 0.1625, |
| "step": 149 |
| }, |
| { |
| "epoch": 0.1180637544273908, |
| "grad_norm": 1.0710088382142664, |
| "learning_rate": 4.93752663316054e-06, |
| "loss": 0.1778, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.1188508461235734, |
| "grad_norm": 1.0503531295721205, |
| "learning_rate": 4.936071576944769e-06, |
| "loss": 0.1726, |
| "step": 151 |
| }, |
| { |
| "epoch": 0.119637937819756, |
| "grad_norm": 1.0686610020146463, |
| "learning_rate": 4.934599989888753e-06, |
| "loss": 0.1769, |
| "step": 152 |
| }, |
| { |
| "epoch": 0.1204250295159386, |
| "grad_norm": 1.072378297090023, |
| "learning_rate": 4.933111881978478e-06, |
| "loss": 0.1866, |
| "step": 153 |
| }, |
| { |
| "epoch": 0.12121212121212122, |
| "grad_norm": 1.2495883030259693, |
| "learning_rate": 4.931607263312033e-06, |
| "loss": 0.1998, |
| "step": 154 |
| }, |
| { |
| "epoch": 0.12199921290830382, |
| "grad_norm": 1.109893027407933, |
| "learning_rate": 4.93008614409955e-06, |
| "loss": 0.1805, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.12278630460448642, |
| "grad_norm": 1.1570851370725408, |
| "learning_rate": 4.928548534663133e-06, |
| "loss": 0.1725, |
| "step": 156 |
| }, |
| { |
| "epoch": 0.12357339630066903, |
| "grad_norm": 1.1758781032456742, |
| "learning_rate": 4.9269944454367815e-06, |
| "loss": 0.176, |
| "step": 157 |
| }, |
| { |
| "epoch": 0.12436048799685163, |
| "grad_norm": 1.1408455648753233, |
| "learning_rate": 4.925423886966328e-06, |
| "loss": 0.1848, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.12514757969303425, |
| "grad_norm": 1.1318514267380126, |
| "learning_rate": 4.923836869909363e-06, |
| "loss": 0.1764, |
| "step": 159 |
| }, |
| { |
| "epoch": 0.12593467138921685, |
| "grad_norm": 1.1451300788977063, |
| "learning_rate": 4.9222334050351595e-06, |
| "loss": 0.1756, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.12672176308539945, |
| "grad_norm": 1.1117305593028235, |
| "learning_rate": 4.920613503224608e-06, |
| "loss": 0.1797, |
| "step": 161 |
| }, |
| { |
| "epoch": 0.12750885478158205, |
| "grad_norm": 1.1301581138966732, |
| "learning_rate": 4.9189771754701335e-06, |
| "loss": 0.1675, |
| "step": 162 |
| }, |
| { |
| "epoch": 0.12829594647776466, |
| "grad_norm": 1.0326917294149387, |
| "learning_rate": 4.917324432875627e-06, |
| "loss": 0.1784, |
| "step": 163 |
| }, |
| { |
| "epoch": 0.12908303817394726, |
| "grad_norm": 1.1983588521884831, |
| "learning_rate": 4.915655286656368e-06, |
| "loss": 0.1966, |
| "step": 164 |
| }, |
| { |
| "epoch": 0.12987012987012986, |
| "grad_norm": 1.0140424703790007, |
| "learning_rate": 4.9139697481389505e-06, |
| "loss": 0.1744, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.1306572215663125, |
| "grad_norm": 1.223539092779737, |
| "learning_rate": 4.9122678287612e-06, |
| "loss": 0.1831, |
| "step": 166 |
| }, |
| { |
| "epoch": 0.1314443132624951, |
| "grad_norm": 1.0918972348910556, |
| "learning_rate": 4.910549540072104e-06, |
| "loss": 0.1843, |
| "step": 167 |
| }, |
| { |
| "epoch": 0.1322314049586777, |
| "grad_norm": 1.1292739304249166, |
| "learning_rate": 4.908814893731728e-06, |
| "loss": 0.1552, |
| "step": 168 |
| }, |
| { |
| "epoch": 0.1330184966548603, |
| "grad_norm": 1.1923518362383727, |
| "learning_rate": 4.9070639015111406e-06, |
| "loss": 0.1895, |
| "step": 169 |
| }, |
| { |
| "epoch": 0.1338055883510429, |
| "grad_norm": 1.083542335588892, |
| "learning_rate": 4.905296575292329e-06, |
| "loss": 0.1745, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.1345926800472255, |
| "grad_norm": 1.2673623015109376, |
| "learning_rate": 4.90351292706812e-06, |
| "loss": 0.1726, |
| "step": 171 |
| }, |
| { |
| "epoch": 0.1353797717434081, |
| "grad_norm": 1.1129476624507257, |
| "learning_rate": 4.901712968942101e-06, |
| "loss": 0.1706, |
| "step": 172 |
| }, |
| { |
| "epoch": 0.13616686343959072, |
| "grad_norm": 1.1735922432656085, |
| "learning_rate": 4.899896713128536e-06, |
| "loss": 0.1741, |
| "step": 173 |
| }, |
| { |
| "epoch": 0.13695395513577333, |
| "grad_norm": 1.2331570034422519, |
| "learning_rate": 4.898064171952281e-06, |
| "loss": 0.1946, |
| "step": 174 |
| }, |
| { |
| "epoch": 0.13774104683195593, |
| "grad_norm": 1.2376618802061816, |
| "learning_rate": 4.896215357848706e-06, |
| "loss": 0.1715, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.13852813852813853, |
| "grad_norm": 1.0860947256302276, |
| "learning_rate": 4.894350283363603e-06, |
| "loss": 0.1664, |
| "step": 176 |
| }, |
| { |
| "epoch": 0.13931523022432113, |
| "grad_norm": 1.1284792933006988, |
| "learning_rate": 4.892468961153105e-06, |
| "loss": 0.1721, |
| "step": 177 |
| }, |
| { |
| "epoch": 0.14010232192050373, |
| "grad_norm": 1.1811695933066144, |
| "learning_rate": 4.8905714039836026e-06, |
| "loss": 0.1768, |
| "step": 178 |
| }, |
| { |
| "epoch": 0.14088941361668633, |
| "grad_norm": 1.1690172197627666, |
| "learning_rate": 4.888657624731652e-06, |
| "loss": 0.1784, |
| "step": 179 |
| }, |
| { |
| "epoch": 0.14167650531286896, |
| "grad_norm": 1.2215187765329307, |
| "learning_rate": 4.88672763638389e-06, |
| "loss": 0.1762, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.14246359700905156, |
| "grad_norm": 1.1657625904368065, |
| "learning_rate": 4.884781452036948e-06, |
| "loss": 0.1754, |
| "step": 181 |
| }, |
| { |
| "epoch": 0.14325068870523416, |
| "grad_norm": 1.0812019740421663, |
| "learning_rate": 4.88281908489736e-06, |
| "loss": 0.1745, |
| "step": 182 |
| }, |
| { |
| "epoch": 0.14403778040141677, |
| "grad_norm": 1.1662972444477193, |
| "learning_rate": 4.880840548281475e-06, |
| "loss": 0.1844, |
| "step": 183 |
| }, |
| { |
| "epoch": 0.14482487209759937, |
| "grad_norm": 1.1318261660435303, |
| "learning_rate": 4.878845855615364e-06, |
| "loss": 0.177, |
| "step": 184 |
| }, |
| { |
| "epoch": 0.14561196379378197, |
| "grad_norm": 1.0454173174935852, |
| "learning_rate": 4.876835020434733e-06, |
| "loss": 0.1726, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.14639905548996457, |
| "grad_norm": 1.1649728572384528, |
| "learning_rate": 4.874808056384826e-06, |
| "loss": 0.1829, |
| "step": 186 |
| }, |
| { |
| "epoch": 0.1471861471861472, |
| "grad_norm": 0.9809711097737751, |
| "learning_rate": 4.8727649772203375e-06, |
| "loss": 0.1626, |
| "step": 187 |
| }, |
| { |
| "epoch": 0.1479732388823298, |
| "grad_norm": 1.0024677570018588, |
| "learning_rate": 4.8707057968053175e-06, |
| "loss": 0.1564, |
| "step": 188 |
| }, |
| { |
| "epoch": 0.1487603305785124, |
| "grad_norm": 1.0801740218719516, |
| "learning_rate": 4.868630529113075e-06, |
| "loss": 0.1571, |
| "step": 189 |
| }, |
| { |
| "epoch": 0.149547422274695, |
| "grad_norm": 1.0633734918657578, |
| "learning_rate": 4.866539188226086e-06, |
| "loss": 0.1558, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.1503345139708776, |
| "grad_norm": 1.1110942685300096, |
| "learning_rate": 4.864431788335895e-06, |
| "loss": 0.1739, |
| "step": 191 |
| }, |
| { |
| "epoch": 0.1511216056670602, |
| "grad_norm": 1.088865739839623, |
| "learning_rate": 4.862308343743024e-06, |
| "loss": 0.1705, |
| "step": 192 |
| }, |
| { |
| "epoch": 0.1519086973632428, |
| "grad_norm": 1.158763785538179, |
| "learning_rate": 4.86016886885687e-06, |
| "loss": 0.1754, |
| "step": 193 |
| }, |
| { |
| "epoch": 0.15269578905942544, |
| "grad_norm": 1.0665033787081621, |
| "learning_rate": 4.858013378195609e-06, |
| "loss": 0.1814, |
| "step": 194 |
| }, |
| { |
| "epoch": 0.15348288075560804, |
| "grad_norm": 1.0347506383595513, |
| "learning_rate": 4.855841886386099e-06, |
| "loss": 0.1659, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.15426997245179064, |
| "grad_norm": 1.3652087096932124, |
| "learning_rate": 4.8536544081637785e-06, |
| "loss": 0.1693, |
| "step": 196 |
| }, |
| { |
| "epoch": 0.15505706414797324, |
| "grad_norm": 1.1571980267809596, |
| "learning_rate": 4.8514509583725685e-06, |
| "loss": 0.1735, |
| "step": 197 |
| }, |
| { |
| "epoch": 0.15584415584415584, |
| "grad_norm": 1.1126611625924816, |
| "learning_rate": 4.849231551964771e-06, |
| "loss": 0.1878, |
| "step": 198 |
| }, |
| { |
| "epoch": 0.15663124754033844, |
| "grad_norm": 1.0666827506948415, |
| "learning_rate": 4.846996204000967e-06, |
| "loss": 0.1686, |
| "step": 199 |
| }, |
| { |
| "epoch": 0.15741833923652104, |
| "grad_norm": 1.1408187983192677, |
| "learning_rate": 4.844744929649912e-06, |
| "loss": 0.1785, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.15820543093270367, |
| "grad_norm": 1.1050850982745672, |
| "learning_rate": 4.842477744188441e-06, |
| "loss": 0.1663, |
| "step": 201 |
| }, |
| { |
| "epoch": 0.15899252262888627, |
| "grad_norm": 1.0153624350350885, |
| "learning_rate": 4.840194663001354e-06, |
| "loss": 0.1755, |
| "step": 202 |
| }, |
| { |
| "epoch": 0.15977961432506887, |
| "grad_norm": 1.0251264155888737, |
| "learning_rate": 4.837895701581322e-06, |
| "loss": 0.1537, |
| "step": 203 |
| }, |
| { |
| "epoch": 0.16056670602125148, |
| "grad_norm": 1.0673153393456505, |
| "learning_rate": 4.835580875528776e-06, |
| "loss": 0.1633, |
| "step": 204 |
| }, |
| { |
| "epoch": 0.16135379771743408, |
| "grad_norm": 1.0273828987011315, |
| "learning_rate": 4.833250200551798e-06, |
| "loss": 0.1746, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.16214088941361668, |
| "grad_norm": 1.0964068866663357, |
| "learning_rate": 4.830903692466023e-06, |
| "loss": 0.1674, |
| "step": 206 |
| }, |
| { |
| "epoch": 0.16292798110979928, |
| "grad_norm": 1.1142080493277295, |
| "learning_rate": 4.828541367194527e-06, |
| "loss": 0.1828, |
| "step": 207 |
| }, |
| { |
| "epoch": 0.1637150728059819, |
| "grad_norm": 1.0617790409690397, |
| "learning_rate": 4.826163240767717e-06, |
| "loss": 0.1676, |
| "step": 208 |
| }, |
| { |
| "epoch": 0.1645021645021645, |
| "grad_norm": 1.2859855971245049, |
| "learning_rate": 4.8237693293232256e-06, |
| "loss": 0.1942, |
| "step": 209 |
| }, |
| { |
| "epoch": 0.1652892561983471, |
| "grad_norm": 1.000840540957111, |
| "learning_rate": 4.821359649105801e-06, |
| "loss": 0.1686, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.1660763478945297, |
| "grad_norm": 1.049595380158752, |
| "learning_rate": 4.818934216467195e-06, |
| "loss": 0.1696, |
| "step": 211 |
| }, |
| { |
| "epoch": 0.16686343959071231, |
| "grad_norm": 1.0218031530162965, |
| "learning_rate": 4.816493047866053e-06, |
| "loss": 0.1653, |
| "step": 212 |
| }, |
| { |
| "epoch": 0.16765053128689492, |
| "grad_norm": 1.0715206508098112, |
| "learning_rate": 4.8140361598678034e-06, |
| "loss": 0.1735, |
| "step": 213 |
| }, |
| { |
| "epoch": 0.16843762298307752, |
| "grad_norm": 1.093161202120212, |
| "learning_rate": 4.811563569144544e-06, |
| "loss": 0.1698, |
| "step": 214 |
| }, |
| { |
| "epoch": 0.16922471467926015, |
| "grad_norm": 1.078958887147992, |
| "learning_rate": 4.809075292474929e-06, |
| "loss": 0.1671, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.17001180637544275, |
| "grad_norm": 1.1213364259804648, |
| "learning_rate": 4.806571346744053e-06, |
| "loss": 0.1798, |
| "step": 216 |
| }, |
| { |
| "epoch": 0.17079889807162535, |
| "grad_norm": 1.102076724202232, |
| "learning_rate": 4.804051748943343e-06, |
| "loss": 0.1845, |
| "step": 217 |
| }, |
| { |
| "epoch": 0.17158598976780795, |
| "grad_norm": 1.1103430873095865, |
| "learning_rate": 4.801516516170437e-06, |
| "loss": 0.177, |
| "step": 218 |
| }, |
| { |
| "epoch": 0.17237308146399055, |
| "grad_norm": 1.228711789290585, |
| "learning_rate": 4.798965665629068e-06, |
| "loss": 0.1636, |
| "step": 219 |
| }, |
| { |
| "epoch": 0.17316017316017315, |
| "grad_norm": 1.1219855198900837, |
| "learning_rate": 4.796399214628949e-06, |
| "loss": 0.1802, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.17394726485635575, |
| "grad_norm": 1.1846418832749555, |
| "learning_rate": 4.7938171805856596e-06, |
| "loss": 0.1717, |
| "step": 221 |
| }, |
| { |
| "epoch": 0.17473435655253838, |
| "grad_norm": 1.0672386815907553, |
| "learning_rate": 4.791219581020518e-06, |
| "loss": 0.1663, |
| "step": 222 |
| }, |
| { |
| "epoch": 0.17552144824872098, |
| "grad_norm": 1.0398388591323704, |
| "learning_rate": 4.788606433560473e-06, |
| "loss": 0.1593, |
| "step": 223 |
| }, |
| { |
| "epoch": 0.1763085399449036, |
| "grad_norm": 1.1402534682960337, |
| "learning_rate": 4.785977755937977e-06, |
| "loss": 0.1876, |
| "step": 224 |
| }, |
| { |
| "epoch": 0.1770956316410862, |
| "grad_norm": 1.1260603683997887, |
| "learning_rate": 4.783333565990865e-06, |
| "loss": 0.172, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.1778827233372688, |
| "grad_norm": 1.062290554096683, |
| "learning_rate": 4.780673881662242e-06, |
| "loss": 0.1709, |
| "step": 226 |
| }, |
| { |
| "epoch": 0.1786698150334514, |
| "grad_norm": 1.0650729387286197, |
| "learning_rate": 4.777998721000353e-06, |
| "loss": 0.1614, |
| "step": 227 |
| }, |
| { |
| "epoch": 0.179456906729634, |
| "grad_norm": 1.0365419204779498, |
| "learning_rate": 4.775308102158461e-06, |
| "loss": 0.1605, |
| "step": 228 |
| }, |
| { |
| "epoch": 0.18024399842581662, |
| "grad_norm": 1.1444494636007958, |
| "learning_rate": 4.772602043394731e-06, |
| "loss": 0.1867, |
| "step": 229 |
| }, |
| { |
| "epoch": 0.18103109012199922, |
| "grad_norm": 1.1053808430839196, |
| "learning_rate": 4.769880563072097e-06, |
| "loss": 0.1627, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.18181818181818182, |
| "grad_norm": 1.0763207393373317, |
| "learning_rate": 4.767143679658143e-06, |
| "loss": 0.1703, |
| "step": 231 |
| }, |
| { |
| "epoch": 0.18260527351436442, |
| "grad_norm": 1.1302336936081483, |
| "learning_rate": 4.764391411724977e-06, |
| "loss": 0.1697, |
| "step": 232 |
| }, |
| { |
| "epoch": 0.18339236521054703, |
| "grad_norm": 1.059980991296742, |
| "learning_rate": 4.7616237779491026e-06, |
| "loss": 0.1658, |
| "step": 233 |
| }, |
| { |
| "epoch": 0.18417945690672963, |
| "grad_norm": 1.0952807461742509, |
| "learning_rate": 4.758840797111295e-06, |
| "loss": 0.1833, |
| "step": 234 |
| }, |
| { |
| "epoch": 0.18496654860291223, |
| "grad_norm": 1.0263555674269131, |
| "learning_rate": 4.756042488096472e-06, |
| "loss": 0.1732, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.18575364029909486, |
| "grad_norm": 1.088261327233659, |
| "learning_rate": 4.753228869893566e-06, |
| "loss": 0.1646, |
| "step": 236 |
| }, |
| { |
| "epoch": 0.18654073199527746, |
| "grad_norm": 1.0644325115229099, |
| "learning_rate": 4.750399961595395e-06, |
| "loss": 0.1576, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.18732782369146006, |
| "grad_norm": 0.9952967090049917, |
| "learning_rate": 4.747555782398537e-06, |
| "loss": 0.1598, |
| "step": 238 |
| }, |
| { |
| "epoch": 0.18811491538764266, |
| "grad_norm": 1.0300249714418026, |
| "learning_rate": 4.7446963516031904e-06, |
| "loss": 0.1883, |
| "step": 239 |
| }, |
| { |
| "epoch": 0.18890200708382526, |
| "grad_norm": 1.0275382678304879, |
| "learning_rate": 4.741821688613054e-06, |
| "loss": 0.1704, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.18968909878000786, |
| "grad_norm": 1.0616733952682182, |
| "learning_rate": 4.738931812935186e-06, |
| "loss": 0.1907, |
| "step": 241 |
| }, |
| { |
| "epoch": 0.19047619047619047, |
| "grad_norm": 1.0103628221724312, |
| "learning_rate": 4.736026744179878e-06, |
| "loss": 0.1556, |
| "step": 242 |
| }, |
| { |
| "epoch": 0.1912632821723731, |
| "grad_norm": 1.0535669337117792, |
| "learning_rate": 4.73310650206052e-06, |
| "loss": 0.1809, |
| "step": 243 |
| }, |
| { |
| "epoch": 0.1920503738685557, |
| "grad_norm": 1.0554553563643476, |
| "learning_rate": 4.730171106393466e-06, |
| "loss": 0.1675, |
| "step": 244 |
| }, |
| { |
| "epoch": 0.1928374655647383, |
| "grad_norm": 0.9417424551436594, |
| "learning_rate": 4.7272205770979e-06, |
| "loss": 0.1438, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.1936245572609209, |
| "grad_norm": 1.1154888244817747, |
| "learning_rate": 4.724254934195698e-06, |
| "loss": 0.1765, |
| "step": 246 |
| }, |
| { |
| "epoch": 0.1944116489571035, |
| "grad_norm": 1.1742188581521773, |
| "learning_rate": 4.721274197811298e-06, |
| "loss": 0.1711, |
| "step": 247 |
| }, |
| { |
| "epoch": 0.1951987406532861, |
| "grad_norm": 1.057640390538921, |
| "learning_rate": 4.71827838817156e-06, |
| "loss": 0.1678, |
| "step": 248 |
| }, |
| { |
| "epoch": 0.1959858323494687, |
| "grad_norm": 1.022336905613029, |
| "learning_rate": 4.715267525605627e-06, |
| "loss": 0.1552, |
| "step": 249 |
| }, |
| { |
| "epoch": 0.19677292404565133, |
| "grad_norm": 1.181830506383501, |
| "learning_rate": 4.712241630544792e-06, |
| "loss": 0.1765, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.19756001574183393, |
| "grad_norm": 1.1571296526874602, |
| "learning_rate": 4.709200723522353e-06, |
| "loss": 0.1758, |
| "step": 251 |
| }, |
| { |
| "epoch": 0.19834710743801653, |
| "grad_norm": 1.082056647389628, |
| "learning_rate": 4.706144825173481e-06, |
| "loss": 0.1638, |
| "step": 252 |
| }, |
| { |
| "epoch": 0.19913419913419914, |
| "grad_norm": 1.0648327864294944, |
| "learning_rate": 4.703073956235071e-06, |
| "loss": 0.1747, |
| "step": 253 |
| }, |
| { |
| "epoch": 0.19992129083038174, |
| "grad_norm": 1.1273460773870558, |
| "learning_rate": 4.6999881375456116e-06, |
| "loss": 0.1767, |
| "step": 254 |
| }, |
| { |
| "epoch": 0.20070838252656434, |
| "grad_norm": 1.0782376126285664, |
| "learning_rate": 4.696887390045035e-06, |
| "loss": 0.169, |
| "step": 255 |
| }, |
| { |
| "epoch": 0.20149547422274694, |
| "grad_norm": 1.043398805036875, |
| "learning_rate": 4.693771734774578e-06, |
| "loss": 0.1774, |
| "step": 256 |
| }, |
| { |
| "epoch": 0.20228256591892957, |
| "grad_norm": 1.067320862475683, |
| "learning_rate": 4.690641192876643e-06, |
| "loss": 0.1607, |
| "step": 257 |
| }, |
| { |
| "epoch": 0.20306965761511217, |
| "grad_norm": 1.1843944163744937, |
| "learning_rate": 4.687495785594646e-06, |
| "loss": 0.1633, |
| "step": 258 |
| }, |
| { |
| "epoch": 0.20385674931129477, |
| "grad_norm": 1.0931562611646284, |
| "learning_rate": 4.684335534272881e-06, |
| "loss": 0.1687, |
| "step": 259 |
| }, |
| { |
| "epoch": 0.20464384100747737, |
| "grad_norm": 1.1204870400497637, |
| "learning_rate": 4.68116046035637e-06, |
| "loss": 0.1639, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.20543093270365997, |
| "grad_norm": 1.2082791443480092, |
| "learning_rate": 4.6779705853907205e-06, |
| "loss": 0.1683, |
| "step": 261 |
| }, |
| { |
| "epoch": 0.20621802439984258, |
| "grad_norm": 1.0646518318192153, |
| "learning_rate": 4.674765931021976e-06, |
| "loss": 0.1611, |
| "step": 262 |
| }, |
| { |
| "epoch": 0.20700511609602518, |
| "grad_norm": 1.1268791395123645, |
| "learning_rate": 4.671546518996473e-06, |
| "loss": 0.1553, |
| "step": 263 |
| }, |
| { |
| "epoch": 0.2077922077922078, |
| "grad_norm": 1.0048534045343525, |
| "learning_rate": 4.668312371160688e-06, |
| "loss": 0.1571, |
| "step": 264 |
| }, |
| { |
| "epoch": 0.2085792994883904, |
| "grad_norm": 1.0052893495164037, |
| "learning_rate": 4.665063509461098e-06, |
| "loss": 0.1679, |
| "step": 265 |
| }, |
| { |
| "epoch": 0.209366391184573, |
| "grad_norm": 0.9679422598052939, |
| "learning_rate": 4.661799955944019e-06, |
| "loss": 0.1556, |
| "step": 266 |
| }, |
| { |
| "epoch": 0.2101534828807556, |
| "grad_norm": 1.0487292157874373, |
| "learning_rate": 4.658521732755471e-06, |
| "loss": 0.183, |
| "step": 267 |
| }, |
| { |
| "epoch": 0.2109405745769382, |
| "grad_norm": 1.0878511570789495, |
| "learning_rate": 4.655228862141017e-06, |
| "loss": 0.1762, |
| "step": 268 |
| }, |
| { |
| "epoch": 0.2117276662731208, |
| "grad_norm": 0.9275216638767947, |
| "learning_rate": 4.651921366445613e-06, |
| "loss": 0.1483, |
| "step": 269 |
| }, |
| { |
| "epoch": 0.21251475796930341, |
| "grad_norm": 1.0291173856009612, |
| "learning_rate": 4.648599268113464e-06, |
| "loss": 0.1657, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.21330184966548604, |
| "grad_norm": 0.9814951923963836, |
| "learning_rate": 4.645262589687861e-06, |
| "loss": 0.1737, |
| "step": 271 |
| }, |
| { |
| "epoch": 0.21408894136166864, |
| "grad_norm": 0.9574503772544043, |
| "learning_rate": 4.641911353811038e-06, |
| "loss": 0.1638, |
| "step": 272 |
| }, |
| { |
| "epoch": 0.21487603305785125, |
| "grad_norm": 0.9684496500051328, |
| "learning_rate": 4.638545583224011e-06, |
| "loss": 0.1649, |
| "step": 273 |
| }, |
| { |
| "epoch": 0.21566312475403385, |
| "grad_norm": 1.0314787067828541, |
| "learning_rate": 4.635165300766428e-06, |
| "loss": 0.1699, |
| "step": 274 |
| }, |
| { |
| "epoch": 0.21645021645021645, |
| "grad_norm": 1.0287264097080684, |
| "learning_rate": 4.63177052937641e-06, |
| "loss": 0.1602, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.21723730814639905, |
| "grad_norm": 1.1114659065296888, |
| "learning_rate": 4.628361292090403e-06, |
| "loss": 0.1783, |
| "step": 276 |
| }, |
| { |
| "epoch": 0.21802439984258165, |
| "grad_norm": 1.0298788844790752, |
| "learning_rate": 4.6249376120430115e-06, |
| "loss": 0.1678, |
| "step": 277 |
| }, |
| { |
| "epoch": 0.21881149153876425, |
| "grad_norm": 1.0099420287081406, |
| "learning_rate": 4.621499512466847e-06, |
| "loss": 0.1672, |
| "step": 278 |
| }, |
| { |
| "epoch": 0.21959858323494688, |
| "grad_norm": 0.9892117727941296, |
| "learning_rate": 4.618047016692374e-06, |
| "loss": 0.1663, |
| "step": 279 |
| }, |
| { |
| "epoch": 0.22038567493112948, |
| "grad_norm": 0.9289360238552057, |
| "learning_rate": 4.614580148147744e-06, |
| "loss": 0.1563, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.22117276662731208, |
| "grad_norm": 0.9603340451855991, |
| "learning_rate": 4.61109893035864e-06, |
| "loss": 0.1561, |
| "step": 281 |
| }, |
| { |
| "epoch": 0.22195985832349469, |
| "grad_norm": 1.0449269347565262, |
| "learning_rate": 4.607603386948119e-06, |
| "loss": 0.165, |
| "step": 282 |
| }, |
| { |
| "epoch": 0.2227469500196773, |
| "grad_norm": 0.990226128298578, |
| "learning_rate": 4.604093541636448e-06, |
| "loss": 0.1704, |
| "step": 283 |
| }, |
| { |
| "epoch": 0.2235340417158599, |
| "grad_norm": 1.031797952555019, |
| "learning_rate": 4.600569418240946e-06, |
| "loss": 0.1677, |
| "step": 284 |
| }, |
| { |
| "epoch": 0.2243211334120425, |
| "grad_norm": 1.0506428763431659, |
| "learning_rate": 4.597031040675819e-06, |
| "loss": 0.1802, |
| "step": 285 |
| }, |
| { |
| "epoch": 0.22510822510822512, |
| "grad_norm": 0.980146123693525, |
| "learning_rate": 4.593478432952002e-06, |
| "loss": 0.1656, |
| "step": 286 |
| }, |
| { |
| "epoch": 0.22589531680440772, |
| "grad_norm": 1.0058178922055618, |
| "learning_rate": 4.589911619176993e-06, |
| "loss": 0.1601, |
| "step": 287 |
| }, |
| { |
| "epoch": 0.22668240850059032, |
| "grad_norm": 1.1532752501338874, |
| "learning_rate": 4.586330623554691e-06, |
| "loss": 0.1707, |
| "step": 288 |
| }, |
| { |
| "epoch": 0.22746950019677292, |
| "grad_norm": 0.9925104519486038, |
| "learning_rate": 4.582735470385229e-06, |
| "loss": 0.1712, |
| "step": 289 |
| }, |
| { |
| "epoch": 0.22825659189295552, |
| "grad_norm": 1.1312813134045174, |
| "learning_rate": 4.579126184064814e-06, |
| "loss": 0.1607, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.22904368358913813, |
| "grad_norm": 1.2454875330122912, |
| "learning_rate": 4.575502789085555e-06, |
| "loss": 0.1656, |
| "step": 291 |
| }, |
| { |
| "epoch": 0.22983077528532073, |
| "grad_norm": 0.9825183210915687, |
| "learning_rate": 4.571865310035304e-06, |
| "loss": 0.1589, |
| "step": 292 |
| }, |
| { |
| "epoch": 0.23061786698150336, |
| "grad_norm": 1.0887371255437703, |
| "learning_rate": 4.568213771597484e-06, |
| "loss": 0.1585, |
| "step": 293 |
| }, |
| { |
| "epoch": 0.23140495867768596, |
| "grad_norm": 1.0975434488519114, |
| "learning_rate": 4.564548198550922e-06, |
| "loss": 0.1435, |
| "step": 294 |
| }, |
| { |
| "epoch": 0.23219205037386856, |
| "grad_norm": 1.0593259383463134, |
| "learning_rate": 4.5608686157696844e-06, |
| "loss": 0.167, |
| "step": 295 |
| }, |
| { |
| "epoch": 0.23297914207005116, |
| "grad_norm": 1.1536948102561841, |
| "learning_rate": 4.557175048222901e-06, |
| "loss": 0.1621, |
| "step": 296 |
| }, |
| { |
| "epoch": 0.23376623376623376, |
| "grad_norm": 1.1369019291567328, |
| "learning_rate": 4.5534675209746076e-06, |
| "loss": 0.1654, |
| "step": 297 |
| }, |
| { |
| "epoch": 0.23455332546241636, |
| "grad_norm": 0.9585590140764199, |
| "learning_rate": 4.5497460591835615e-06, |
| "loss": 0.148, |
| "step": 298 |
| }, |
| { |
| "epoch": 0.23534041715859896, |
| "grad_norm": 1.2337420030262027, |
| "learning_rate": 4.546010688103082e-06, |
| "loss": 0.1599, |
| "step": 299 |
| }, |
| { |
| "epoch": 0.2361275088547816, |
| "grad_norm": 1.1641848426244756, |
| "learning_rate": 4.542261433080874e-06, |
| "loss": 0.1641, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.2369146005509642, |
| "grad_norm": 0.9715264597638171, |
| "learning_rate": 4.538498319558854e-06, |
| "loss": 0.1604, |
| "step": 301 |
| }, |
| { |
| "epoch": 0.2377016922471468, |
| "grad_norm": 1.2043568904283137, |
| "learning_rate": 4.534721373072986e-06, |
| "loss": 0.1561, |
| "step": 302 |
| }, |
| { |
| "epoch": 0.2384887839433294, |
| "grad_norm": 1.087701432883666, |
| "learning_rate": 4.530930619253097e-06, |
| "loss": 0.1573, |
| "step": 303 |
| }, |
| { |
| "epoch": 0.239275875639512, |
| "grad_norm": 1.0432095830081018, |
| "learning_rate": 4.527126083822713e-06, |
| "loss": 0.1576, |
| "step": 304 |
| }, |
| { |
| "epoch": 0.2400629673356946, |
| "grad_norm": 1.1515388977241858, |
| "learning_rate": 4.523307792598877e-06, |
| "loss": 0.1836, |
| "step": 305 |
| }, |
| { |
| "epoch": 0.2408500590318772, |
| "grad_norm": 1.1236907370811289, |
| "learning_rate": 4.519475771491978e-06, |
| "loss": 0.1654, |
| "step": 306 |
| }, |
| { |
| "epoch": 0.24163715072805983, |
| "grad_norm": 1.0492490872684002, |
| "learning_rate": 4.515630046505575e-06, |
| "loss": 0.1604, |
| "step": 307 |
| }, |
| { |
| "epoch": 0.24242424242424243, |
| "grad_norm": 1.0414505694174347, |
| "learning_rate": 4.511770643736217e-06, |
| "loss": 0.1587, |
| "step": 308 |
| }, |
| { |
| "epoch": 0.24321133412042503, |
| "grad_norm": 0.9963463131455829, |
| "learning_rate": 4.507897589373272e-06, |
| "loss": 0.1536, |
| "step": 309 |
| }, |
| { |
| "epoch": 0.24399842581660763, |
| "grad_norm": 0.9437267739253786, |
| "learning_rate": 4.504010909698744e-06, |
| "loss": 0.1573, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.24478551751279023, |
| "grad_norm": 0.9915304289222059, |
| "learning_rate": 4.500110631087095e-06, |
| "loss": 0.1519, |
| "step": 311 |
| }, |
| { |
| "epoch": 0.24557260920897284, |
| "grad_norm": 0.9782358310573961, |
| "learning_rate": 4.496196780005069e-06, |
| "loss": 0.1629, |
| "step": 312 |
| }, |
| { |
| "epoch": 0.24635970090515544, |
| "grad_norm": 1.0770165377269398, |
| "learning_rate": 4.492269383011512e-06, |
| "loss": 0.1623, |
| "step": 313 |
| }, |
| { |
| "epoch": 0.24714679260133807, |
| "grad_norm": 1.052396599909024, |
| "learning_rate": 4.4883284667571894e-06, |
| "loss": 0.1533, |
| "step": 314 |
| }, |
| { |
| "epoch": 0.24793388429752067, |
| "grad_norm": 1.0084809840218907, |
| "learning_rate": 4.4843740579846055e-06, |
| "loss": 0.1512, |
| "step": 315 |
| }, |
| { |
| "epoch": 0.24872097599370327, |
| "grad_norm": 1.0756395659672484, |
| "learning_rate": 4.480406183527823e-06, |
| "loss": 0.1682, |
| "step": 316 |
| }, |
| { |
| "epoch": 0.24950806768988587, |
| "grad_norm": 1.095604151904482, |
| "learning_rate": 4.476424870312286e-06, |
| "loss": 0.1588, |
| "step": 317 |
| }, |
| { |
| "epoch": 0.2502951593860685, |
| "grad_norm": 1.073871876794014, |
| "learning_rate": 4.472430145354622e-06, |
| "loss": 0.1663, |
| "step": 318 |
| }, |
| { |
| "epoch": 0.2510822510822511, |
| "grad_norm": 1.00181438336178, |
| "learning_rate": 4.46842203576248e-06, |
| "loss": 0.1668, |
| "step": 319 |
| }, |
| { |
| "epoch": 0.2518693427784337, |
| "grad_norm": 1.0179064844212398, |
| "learning_rate": 4.464400568734327e-06, |
| "loss": 0.1618, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.2526564344746163, |
| "grad_norm": 1.1266566093245078, |
| "learning_rate": 4.460365771559275e-06, |
| "loss": 0.1726, |
| "step": 321 |
| }, |
| { |
| "epoch": 0.2534435261707989, |
| "grad_norm": 1.0831980755033608, |
| "learning_rate": 4.456317671616892e-06, |
| "loss": 0.1674, |
| "step": 322 |
| }, |
| { |
| "epoch": 0.2542306178669815, |
| "grad_norm": 0.9991360442603613, |
| "learning_rate": 4.452256296377017e-06, |
| "loss": 0.1534, |
| "step": 323 |
| }, |
| { |
| "epoch": 0.2550177095631641, |
| "grad_norm": 0.9497710360440503, |
| "learning_rate": 4.448181673399573e-06, |
| "loss": 0.1562, |
| "step": 324 |
| }, |
| { |
| "epoch": 0.25580480125934674, |
| "grad_norm": 1.1113260986403124, |
| "learning_rate": 4.444093830334381e-06, |
| "loss": 0.1639, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.2565918929555293, |
| "grad_norm": 1.1452949830587935, |
| "learning_rate": 4.4399927949209685e-06, |
| "loss": 0.1633, |
| "step": 326 |
| }, |
| { |
| "epoch": 0.25737898465171194, |
| "grad_norm": 1.0842379105419755, |
| "learning_rate": 4.43587859498839e-06, |
| "loss": 0.1754, |
| "step": 327 |
| }, |
| { |
| "epoch": 0.2581660763478945, |
| "grad_norm": 1.0361570331888057, |
| "learning_rate": 4.431751258455029e-06, |
| "loss": 0.1629, |
| "step": 328 |
| }, |
| { |
| "epoch": 0.25895316804407714, |
| "grad_norm": 0.9514704452172565, |
| "learning_rate": 4.4276108133284115e-06, |
| "loss": 0.1615, |
| "step": 329 |
| }, |
| { |
| "epoch": 0.2597402597402597, |
| "grad_norm": 1.0051943736689641, |
| "learning_rate": 4.4234572877050175e-06, |
| "loss": 0.1635, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.26052735143644234, |
| "grad_norm": 1.061826511574687, |
| "learning_rate": 4.419290709770091e-06, |
| "loss": 0.1572, |
| "step": 331 |
| }, |
| { |
| "epoch": 0.261314443132625, |
| "grad_norm": 1.0098180333606226, |
| "learning_rate": 4.415111107797445e-06, |
| "loss": 0.1625, |
| "step": 332 |
| }, |
| { |
| "epoch": 0.26210153482880755, |
| "grad_norm": 0.9258158779374888, |
| "learning_rate": 4.4109185101492735e-06, |
| "loss": 0.163, |
| "step": 333 |
| }, |
| { |
| "epoch": 0.2628886265249902, |
| "grad_norm": 1.031959410480149, |
| "learning_rate": 4.406712945275955e-06, |
| "loss": 0.1601, |
| "step": 334 |
| }, |
| { |
| "epoch": 0.26367571822117275, |
| "grad_norm": 1.098174422684468, |
| "learning_rate": 4.402494441715864e-06, |
| "loss": 0.1632, |
| "step": 335 |
| }, |
| { |
| "epoch": 0.2644628099173554, |
| "grad_norm": 0.9325275936138202, |
| "learning_rate": 4.398263028095175e-06, |
| "loss": 0.1568, |
| "step": 336 |
| }, |
| { |
| "epoch": 0.26524990161353795, |
| "grad_norm": 0.9452361980478395, |
| "learning_rate": 4.394018733127667e-06, |
| "loss": 0.1514, |
| "step": 337 |
| }, |
| { |
| "epoch": 0.2660369933097206, |
| "grad_norm": 0.9440560796701104, |
| "learning_rate": 4.389761585614531e-06, |
| "loss": 0.1568, |
| "step": 338 |
| }, |
| { |
| "epoch": 0.2668240850059032, |
| "grad_norm": 0.9825093172685871, |
| "learning_rate": 4.3854916144441714e-06, |
| "loss": 0.1513, |
| "step": 339 |
| }, |
| { |
| "epoch": 0.2676111767020858, |
| "grad_norm": 0.9909422001877334, |
| "learning_rate": 4.381208848592017e-06, |
| "loss": 0.1607, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.2683982683982684, |
| "grad_norm": 1.026772957857381, |
| "learning_rate": 4.3769133171203146e-06, |
| "loss": 0.1579, |
| "step": 341 |
| }, |
| { |
| "epoch": 0.269185360094451, |
| "grad_norm": 0.9727634660522837, |
| "learning_rate": 4.372605049177939e-06, |
| "loss": 0.1611, |
| "step": 342 |
| }, |
| { |
| "epoch": 0.2699724517906336, |
| "grad_norm": 0.9991705382361779, |
| "learning_rate": 4.368284074000193e-06, |
| "loss": 0.1423, |
| "step": 343 |
| }, |
| { |
| "epoch": 0.2707595434868162, |
| "grad_norm": 1.0413166825567135, |
| "learning_rate": 4.363950420908608e-06, |
| "loss": 0.1531, |
| "step": 344 |
| }, |
| { |
| "epoch": 0.2715466351829988, |
| "grad_norm": 1.051399331371367, |
| "learning_rate": 4.3596041193107475e-06, |
| "loss": 0.1537, |
| "step": 345 |
| }, |
| { |
| "epoch": 0.27233372687918145, |
| "grad_norm": 1.1268002118202416, |
| "learning_rate": 4.355245198700003e-06, |
| "loss": 0.1687, |
| "step": 346 |
| }, |
| { |
| "epoch": 0.273120818575364, |
| "grad_norm": 1.0579162910588005, |
| "learning_rate": 4.3508736886554e-06, |
| "loss": 0.1545, |
| "step": 347 |
| }, |
| { |
| "epoch": 0.27390791027154665, |
| "grad_norm": 1.0780531804812832, |
| "learning_rate": 4.346489618841393e-06, |
| "loss": 0.1478, |
| "step": 348 |
| }, |
| { |
| "epoch": 0.2746950019677292, |
| "grad_norm": 1.1629336261073622, |
| "learning_rate": 4.342093019007664e-06, |
| "loss": 0.1507, |
| "step": 349 |
| }, |
| { |
| "epoch": 0.27548209366391185, |
| "grad_norm": 0.9806357134318359, |
| "learning_rate": 4.337683918988924e-06, |
| "loss": 0.1605, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.2762691853600944, |
| "grad_norm": 1.0271547256327147, |
| "learning_rate": 4.333262348704708e-06, |
| "loss": 0.1544, |
| "step": 351 |
| }, |
| { |
| "epoch": 0.27705627705627706, |
| "grad_norm": 1.040963108089893, |
| "learning_rate": 4.328828338159173e-06, |
| "loss": 0.1505, |
| "step": 352 |
| }, |
| { |
| "epoch": 0.2778433687524597, |
| "grad_norm": 1.036202462349552, |
| "learning_rate": 4.324381917440891e-06, |
| "loss": 0.1558, |
| "step": 353 |
| }, |
| { |
| "epoch": 0.27863046044864226, |
| "grad_norm": 0.975994343559266, |
| "learning_rate": 4.319923116722651e-06, |
| "loss": 0.1641, |
| "step": 354 |
| }, |
| { |
| "epoch": 0.2794175521448249, |
| "grad_norm": 1.039409188253541, |
| "learning_rate": 4.315451966261248e-06, |
| "loss": 0.1549, |
| "step": 355 |
| }, |
| { |
| "epoch": 0.28020464384100746, |
| "grad_norm": 1.047725080130562, |
| "learning_rate": 4.310968496397284e-06, |
| "loss": 0.165, |
| "step": 356 |
| }, |
| { |
| "epoch": 0.2809917355371901, |
| "grad_norm": 1.0011313336241248, |
| "learning_rate": 4.306472737554957e-06, |
| "loss": 0.1456, |
| "step": 357 |
| }, |
| { |
| "epoch": 0.28177882723337266, |
| "grad_norm": 0.9015679935576075, |
| "learning_rate": 4.301964720241857e-06, |
| "loss": 0.1369, |
| "step": 358 |
| }, |
| { |
| "epoch": 0.2825659189295553, |
| "grad_norm": 1.049381444767021, |
| "learning_rate": 4.297444475048755e-06, |
| "loss": 0.1563, |
| "step": 359 |
| }, |
| { |
| "epoch": 0.2833530106257379, |
| "grad_norm": 1.0194195709152667, |
| "learning_rate": 4.292912032649403e-06, |
| "loss": 0.1649, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.2841401023219205, |
| "grad_norm": 0.957368492301693, |
| "learning_rate": 4.2883674238003195e-06, |
| "loss": 0.1515, |
| "step": 361 |
| }, |
| { |
| "epoch": 0.2849271940181031, |
| "grad_norm": 1.1143901057936236, |
| "learning_rate": 4.2838106793405825e-06, |
| "loss": 0.1625, |
| "step": 362 |
| }, |
| { |
| "epoch": 0.2857142857142857, |
| "grad_norm": 1.0882799366794436, |
| "learning_rate": 4.2792418301916225e-06, |
| "loss": 0.153, |
| "step": 363 |
| }, |
| { |
| "epoch": 0.2865013774104683, |
| "grad_norm": 1.0192177035415801, |
| "learning_rate": 4.274660907357009e-06, |
| "loss": 0.1645, |
| "step": 364 |
| }, |
| { |
| "epoch": 0.2872884691066509, |
| "grad_norm": 1.0205240256871457, |
| "learning_rate": 4.2700679419222415e-06, |
| "loss": 0.1459, |
| "step": 365 |
| }, |
| { |
| "epoch": 0.28807556080283353, |
| "grad_norm": 1.2141057030738465, |
| "learning_rate": 4.265462965054539e-06, |
| "loss": 0.1597, |
| "step": 366 |
| }, |
| { |
| "epoch": 0.28886265249901616, |
| "grad_norm": 1.017121088131926, |
| "learning_rate": 4.260846008002631e-06, |
| "loss": 0.1619, |
| "step": 367 |
| }, |
| { |
| "epoch": 0.28964974419519873, |
| "grad_norm": 1.0877328731947116, |
| "learning_rate": 4.25621710209654e-06, |
| "loss": 0.1716, |
| "step": 368 |
| }, |
| { |
| "epoch": 0.29043683589138136, |
| "grad_norm": 1.1099554764936985, |
| "learning_rate": 4.251576278747372e-06, |
| "loss": 0.1599, |
| "step": 369 |
| }, |
| { |
| "epoch": 0.29122392758756394, |
| "grad_norm": 0.962048395782395, |
| "learning_rate": 4.246923569447105e-06, |
| "loss": 0.1465, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.29201101928374656, |
| "grad_norm": 1.0782102946203345, |
| "learning_rate": 4.24225900576837e-06, |
| "loss": 0.1584, |
| "step": 371 |
| }, |
| { |
| "epoch": 0.29279811097992914, |
| "grad_norm": 1.0600722154446367, |
| "learning_rate": 4.237582619364244e-06, |
| "loss": 0.1518, |
| "step": 372 |
| }, |
| { |
| "epoch": 0.29358520267611177, |
| "grad_norm": 1.0154082912245785, |
| "learning_rate": 4.23289444196803e-06, |
| "loss": 0.1455, |
| "step": 373 |
| }, |
| { |
| "epoch": 0.2943722943722944, |
| "grad_norm": 1.1254176051245297, |
| "learning_rate": 4.228194505393041e-06, |
| "loss": 0.1544, |
| "step": 374 |
| }, |
| { |
| "epoch": 0.29515938606847697, |
| "grad_norm": 1.1003313998342341, |
| "learning_rate": 4.22348284153239e-06, |
| "loss": 0.1611, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.2959464777646596, |
| "grad_norm": 0.9110264218620379, |
| "learning_rate": 4.218759482358765e-06, |
| "loss": 0.1479, |
| "step": 376 |
| }, |
| { |
| "epoch": 0.2967335694608422, |
| "grad_norm": 1.0433752096490876, |
| "learning_rate": 4.214024459924221e-06, |
| "loss": 0.1561, |
| "step": 377 |
| }, |
| { |
| "epoch": 0.2975206611570248, |
| "grad_norm": 0.9985242964251728, |
| "learning_rate": 4.209277806359956e-06, |
| "loss": 0.1486, |
| "step": 378 |
| }, |
| { |
| "epoch": 0.2983077528532074, |
| "grad_norm": 0.9830203630270159, |
| "learning_rate": 4.204519553876095e-06, |
| "loss": 0.153, |
| "step": 379 |
| }, |
| { |
| "epoch": 0.29909484454939, |
| "grad_norm": 1.0623334389041004, |
| "learning_rate": 4.199749734761473e-06, |
| "loss": 0.1584, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.29988193624557263, |
| "grad_norm": 1.007050119697646, |
| "learning_rate": 4.194968381383414e-06, |
| "loss": 0.162, |
| "step": 381 |
| }, |
| { |
| "epoch": 0.3006690279417552, |
| "grad_norm": 0.9212276043601202, |
| "learning_rate": 4.1901755261875116e-06, |
| "loss": 0.1417, |
| "step": 382 |
| }, |
| { |
| "epoch": 0.30145611963793784, |
| "grad_norm": 1.0195210503773229, |
| "learning_rate": 4.18537120169741e-06, |
| "loss": 0.1631, |
| "step": 383 |
| }, |
| { |
| "epoch": 0.3022432113341204, |
| "grad_norm": 0.9791393783618954, |
| "learning_rate": 4.1805554405145805e-06, |
| "loss": 0.151, |
| "step": 384 |
| }, |
| { |
| "epoch": 0.30303030303030304, |
| "grad_norm": 0.9560471554995319, |
| "learning_rate": 4.175728275318105e-06, |
| "loss": 0.1537, |
| "step": 385 |
| }, |
| { |
| "epoch": 0.3038173947264856, |
| "grad_norm": 0.9732207377472094, |
| "learning_rate": 4.170889738864448e-06, |
| "loss": 0.1541, |
| "step": 386 |
| }, |
| { |
| "epoch": 0.30460448642266824, |
| "grad_norm": 1.0273971232086052, |
| "learning_rate": 4.166039863987241e-06, |
| "loss": 0.1623, |
| "step": 387 |
| }, |
| { |
| "epoch": 0.30539157811885087, |
| "grad_norm": 1.0066781633766182, |
| "learning_rate": 4.161178683597055e-06, |
| "loss": 0.1623, |
| "step": 388 |
| }, |
| { |
| "epoch": 0.30617866981503344, |
| "grad_norm": 0.9519906303887405, |
| "learning_rate": 4.156306230681178e-06, |
| "loss": 0.1606, |
| "step": 389 |
| }, |
| { |
| "epoch": 0.3069657615112161, |
| "grad_norm": 1.0274010773396909, |
| "learning_rate": 4.151422538303393e-06, |
| "loss": 0.1588, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.30775285320739865, |
| "grad_norm": 1.0092975668609663, |
| "learning_rate": 4.1465276396037516e-06, |
| "loss": 0.1549, |
| "step": 391 |
| }, |
| { |
| "epoch": 0.3085399449035813, |
| "grad_norm": 0.9952464194936945, |
| "learning_rate": 4.141621567798351e-06, |
| "loss": 0.1468, |
| "step": 392 |
| }, |
| { |
| "epoch": 0.30932703659976385, |
| "grad_norm": 0.9809046515355889, |
| "learning_rate": 4.136704356179105e-06, |
| "loss": 0.1509, |
| "step": 393 |
| }, |
| { |
| "epoch": 0.3101141282959465, |
| "grad_norm": 1.0624718823483572, |
| "learning_rate": 4.131776038113524e-06, |
| "loss": 0.1629, |
| "step": 394 |
| }, |
| { |
| "epoch": 0.3109012199921291, |
| "grad_norm": 0.9361042861540975, |
| "learning_rate": 4.126836647044484e-06, |
| "loss": 0.1453, |
| "step": 395 |
| }, |
| { |
| "epoch": 0.3116883116883117, |
| "grad_norm": 1.0675702598561039, |
| "learning_rate": 4.121886216489999e-06, |
| "loss": 0.1657, |
| "step": 396 |
| }, |
| { |
| "epoch": 0.3124754033844943, |
| "grad_norm": 1.0221190108601212, |
| "learning_rate": 4.116924780042997e-06, |
| "loss": 0.1609, |
| "step": 397 |
| }, |
| { |
| "epoch": 0.3132624950806769, |
| "grad_norm": 0.98812521742716, |
| "learning_rate": 4.111952371371091e-06, |
| "loss": 0.1488, |
| "step": 398 |
| }, |
| { |
| "epoch": 0.3140495867768595, |
| "grad_norm": 0.9689235987787954, |
| "learning_rate": 4.106969024216348e-06, |
| "loss": 0.1547, |
| "step": 399 |
| }, |
| { |
| "epoch": 0.3148366784730421, |
| "grad_norm": 1.0046393279094348, |
| "learning_rate": 4.101974772395066e-06, |
| "loss": 0.1467, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.3156237701692247, |
| "grad_norm": 0.968527185963086, |
| "learning_rate": 4.096969649797534e-06, |
| "loss": 0.1432, |
| "step": 401 |
| }, |
| { |
| "epoch": 0.31641086186540734, |
| "grad_norm": 1.0188815460176754, |
| "learning_rate": 4.091953690387815e-06, |
| "loss": 0.1521, |
| "step": 402 |
| }, |
| { |
| "epoch": 0.3171979535615899, |
| "grad_norm": 1.035965071382904, |
| "learning_rate": 4.086926928203506e-06, |
| "loss": 0.1575, |
| "step": 403 |
| }, |
| { |
| "epoch": 0.31798504525777255, |
| "grad_norm": 1.0400881738148544, |
| "learning_rate": 4.081889397355509e-06, |
| "loss": 0.1646, |
| "step": 404 |
| }, |
| { |
| "epoch": 0.3187721369539551, |
| "grad_norm": 1.0353365656909388, |
| "learning_rate": 4.076841132027805e-06, |
| "loss": 0.1578, |
| "step": 405 |
| }, |
| { |
| "epoch": 0.31955922865013775, |
| "grad_norm": 0.9785090873779988, |
| "learning_rate": 4.071782166477213e-06, |
| "loss": 0.1485, |
| "step": 406 |
| }, |
| { |
| "epoch": 0.3203463203463203, |
| "grad_norm": 1.0365440161437718, |
| "learning_rate": 4.066712535033164e-06, |
| "loss": 0.1644, |
| "step": 407 |
| }, |
| { |
| "epoch": 0.32113341204250295, |
| "grad_norm": 0.9337858697268638, |
| "learning_rate": 4.061632272097467e-06, |
| "loss": 0.1396, |
| "step": 408 |
| }, |
| { |
| "epoch": 0.3219205037386856, |
| "grad_norm": 0.9930564105014524, |
| "learning_rate": 4.056541412144073e-06, |
| "loss": 0.1466, |
| "step": 409 |
| }, |
| { |
| "epoch": 0.32270759543486816, |
| "grad_norm": 1.0123860857315623, |
| "learning_rate": 4.051439989718845e-06, |
| "loss": 0.1718, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.3234946871310508, |
| "grad_norm": 0.9886983463565112, |
| "learning_rate": 4.0463280394393216e-06, |
| "loss": 0.1465, |
| "step": 411 |
| }, |
| { |
| "epoch": 0.32428177882723336, |
| "grad_norm": 0.9489896550219313, |
| "learning_rate": 4.041205595994478e-06, |
| "loss": 0.1553, |
| "step": 412 |
| }, |
| { |
| "epoch": 0.325068870523416, |
| "grad_norm": 0.935055903913981, |
| "learning_rate": 4.036072694144501e-06, |
| "loss": 0.1486, |
| "step": 413 |
| }, |
| { |
| "epoch": 0.32585596221959856, |
| "grad_norm": 1.0109287737515016, |
| "learning_rate": 4.030929368720539e-06, |
| "loss": 0.1563, |
| "step": 414 |
| }, |
| { |
| "epoch": 0.3266430539157812, |
| "grad_norm": 0.9682667210224672, |
| "learning_rate": 4.025775654624481e-06, |
| "loss": 0.154, |
| "step": 415 |
| }, |
| { |
| "epoch": 0.3274301456119638, |
| "grad_norm": 0.9195115794003238, |
| "learning_rate": 4.020611586828705e-06, |
| "loss": 0.1433, |
| "step": 416 |
| }, |
| { |
| "epoch": 0.3282172373081464, |
| "grad_norm": 0.886911970381121, |
| "learning_rate": 4.015437200375855e-06, |
| "loss": 0.1374, |
| "step": 417 |
| }, |
| { |
| "epoch": 0.329004329004329, |
| "grad_norm": 1.021240159520919, |
| "learning_rate": 4.01025253037859e-06, |
| "loss": 0.1567, |
| "step": 418 |
| }, |
| { |
| "epoch": 0.3297914207005116, |
| "grad_norm": 0.9462875663398478, |
| "learning_rate": 4.005057612019353e-06, |
| "loss": 0.1516, |
| "step": 419 |
| }, |
| { |
| "epoch": 0.3305785123966942, |
| "grad_norm": 0.9850150964188347, |
| "learning_rate": 3.9998524805501335e-06, |
| "loss": 0.149, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.3313656040928768, |
| "grad_norm": 1.0466582919958791, |
| "learning_rate": 3.994637171292223e-06, |
| "loss": 0.1504, |
| "step": 421 |
| }, |
| { |
| "epoch": 0.3321526957890594, |
| "grad_norm": 0.9892923577104711, |
| "learning_rate": 3.989411719635979e-06, |
| "loss": 0.1465, |
| "step": 422 |
| }, |
| { |
| "epoch": 0.33293978748524206, |
| "grad_norm": 1.0797299646481497, |
| "learning_rate": 3.984176161040585e-06, |
| "loss": 0.1655, |
| "step": 423 |
| }, |
| { |
| "epoch": 0.33372687918142463, |
| "grad_norm": 1.0280855278386611, |
| "learning_rate": 3.978930531033807e-06, |
| "loss": 0.1614, |
| "step": 424 |
| }, |
| { |
| "epoch": 0.33451397087760726, |
| "grad_norm": 1.0013220452351206, |
| "learning_rate": 3.973674865211754e-06, |
| "loss": 0.1529, |
| "step": 425 |
| }, |
| { |
| "epoch": 0.33530106257378983, |
| "grad_norm": 1.0185754306460224, |
| "learning_rate": 3.968409199238639e-06, |
| "loss": 0.1535, |
| "step": 426 |
| }, |
| { |
| "epoch": 0.33608815426997246, |
| "grad_norm": 0.9680713968649773, |
| "learning_rate": 3.963133568846533e-06, |
| "loss": 0.1532, |
| "step": 427 |
| }, |
| { |
| "epoch": 0.33687524596615503, |
| "grad_norm": 1.0465737054070945, |
| "learning_rate": 3.957848009835125e-06, |
| "loss": 0.1557, |
| "step": 428 |
| }, |
| { |
| "epoch": 0.33766233766233766, |
| "grad_norm": 1.0072097384887637, |
| "learning_rate": 3.952552558071475e-06, |
| "loss": 0.1686, |
| "step": 429 |
| }, |
| { |
| "epoch": 0.3384494293585203, |
| "grad_norm": 1.0495298691679416, |
| "learning_rate": 3.947247249489779e-06, |
| "loss": 0.1487, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.33923652105470287, |
| "grad_norm": 1.0214586461562896, |
| "learning_rate": 3.941932120091116e-06, |
| "loss": 0.1621, |
| "step": 431 |
| }, |
| { |
| "epoch": 0.3400236127508855, |
| "grad_norm": 1.0494096714602847, |
| "learning_rate": 3.93660720594321e-06, |
| "loss": 0.1598, |
| "step": 432 |
| }, |
| { |
| "epoch": 0.34081070444706807, |
| "grad_norm": 1.0334818385570048, |
| "learning_rate": 3.93127254318018e-06, |
| "loss": 0.1577, |
| "step": 433 |
| }, |
| { |
| "epoch": 0.3415977961432507, |
| "grad_norm": 0.9700994625756835, |
| "learning_rate": 3.925928168002302e-06, |
| "loss": 0.1526, |
| "step": 434 |
| }, |
| { |
| "epoch": 0.34238488783943327, |
| "grad_norm": 1.047736033995709, |
| "learning_rate": 3.920574116675756e-06, |
| "loss": 0.1581, |
| "step": 435 |
| }, |
| { |
| "epoch": 0.3431719795356159, |
| "grad_norm": 1.0493869403649712, |
| "learning_rate": 3.915210425532383e-06, |
| "loss": 0.1495, |
| "step": 436 |
| }, |
| { |
| "epoch": 0.34395907123179853, |
| "grad_norm": 1.010254528268069, |
| "learning_rate": 3.90983713096944e-06, |
| "loss": 0.1539, |
| "step": 437 |
| }, |
| { |
| "epoch": 0.3447461629279811, |
| "grad_norm": 0.9846398029609658, |
| "learning_rate": 3.9044542694493515e-06, |
| "loss": 0.1463, |
| "step": 438 |
| }, |
| { |
| "epoch": 0.34553325462416373, |
| "grad_norm": 1.2083136674514858, |
| "learning_rate": 3.899061877499461e-06, |
| "loss": 0.1601, |
| "step": 439 |
| }, |
| { |
| "epoch": 0.3463203463203463, |
| "grad_norm": 0.97978554217786, |
| "learning_rate": 3.893659991711782e-06, |
| "loss": 0.139, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.34710743801652894, |
| "grad_norm": 1.1022405018344112, |
| "learning_rate": 3.888248648742756e-06, |
| "loss": 0.1617, |
| "step": 441 |
| }, |
| { |
| "epoch": 0.3478945297127115, |
| "grad_norm": 1.0077367730076683, |
| "learning_rate": 3.882827885312999e-06, |
| "loss": 0.1488, |
| "step": 442 |
| }, |
| { |
| "epoch": 0.34868162140889414, |
| "grad_norm": 1.0119669193080498, |
| "learning_rate": 3.877397738207051e-06, |
| "loss": 0.1433, |
| "step": 443 |
| }, |
| { |
| "epoch": 0.34946871310507677, |
| "grad_norm": 0.9336119872704435, |
| "learning_rate": 3.8719582442731276e-06, |
| "loss": 0.1393, |
| "step": 444 |
| }, |
| { |
| "epoch": 0.35025580480125934, |
| "grad_norm": 1.0144372790745282, |
| "learning_rate": 3.866509440422873e-06, |
| "loss": 0.1515, |
| "step": 445 |
| }, |
| { |
| "epoch": 0.35104289649744197, |
| "grad_norm": 1.0618851735205919, |
| "learning_rate": 3.861051363631107e-06, |
| "loss": 0.1403, |
| "step": 446 |
| }, |
| { |
| "epoch": 0.35182998819362454, |
| "grad_norm": 1.0256940692518137, |
| "learning_rate": 3.855584050935574e-06, |
| "loss": 0.1533, |
| "step": 447 |
| }, |
| { |
| "epoch": 0.3526170798898072, |
| "grad_norm": 1.004262449427633, |
| "learning_rate": 3.85010753943669e-06, |
| "loss": 0.1437, |
| "step": 448 |
| }, |
| { |
| "epoch": 0.35340417158598975, |
| "grad_norm": 0.9608822952661715, |
| "learning_rate": 3.844621866297295e-06, |
| "loss": 0.1374, |
| "step": 449 |
| }, |
| { |
| "epoch": 0.3541912632821724, |
| "grad_norm": 1.032805552257636, |
| "learning_rate": 3.839127068742399e-06, |
| "loss": 0.1612, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.354978354978355, |
| "grad_norm": 1.089158815357864, |
| "learning_rate": 3.833623184058926e-06, |
| "loss": 0.1564, |
| "step": 451 |
| }, |
| { |
| "epoch": 0.3557654466745376, |
| "grad_norm": 1.0527347217082683, |
| "learning_rate": 3.8281102495954684e-06, |
| "loss": 0.1475, |
| "step": 452 |
| }, |
| { |
| "epoch": 0.3565525383707202, |
| "grad_norm": 1.012969201356965, |
| "learning_rate": 3.8225883027620245e-06, |
| "loss": 0.1443, |
| "step": 453 |
| }, |
| { |
| "epoch": 0.3573396300669028, |
| "grad_norm": 0.9952397622221426, |
| "learning_rate": 3.817057381029752e-06, |
| "loss": 0.1488, |
| "step": 454 |
| }, |
| { |
| "epoch": 0.3581267217630854, |
| "grad_norm": 0.9773911500192811, |
| "learning_rate": 3.811517521930711e-06, |
| "loss": 0.1419, |
| "step": 455 |
| }, |
| { |
| "epoch": 0.358913813459268, |
| "grad_norm": 1.04344141144674, |
| "learning_rate": 3.805968763057609e-06, |
| "loss": 0.1335, |
| "step": 456 |
| }, |
| { |
| "epoch": 0.3597009051554506, |
| "grad_norm": 0.9127224357677829, |
| "learning_rate": 3.8004111420635453e-06, |
| "loss": 0.1421, |
| "step": 457 |
| }, |
| { |
| "epoch": 0.36048799685163324, |
| "grad_norm": 0.948335811441799, |
| "learning_rate": 3.7948446966617568e-06, |
| "loss": 0.1545, |
| "step": 458 |
| }, |
| { |
| "epoch": 0.3612750885478158, |
| "grad_norm": 1.054156015531643, |
| "learning_rate": 3.7892694646253624e-06, |
| "loss": 0.1462, |
| "step": 459 |
| }, |
| { |
| "epoch": 0.36206218024399844, |
| "grad_norm": 1.0883694334017704, |
| "learning_rate": 3.783685483787105e-06, |
| "loss": 0.1469, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.362849271940181, |
| "grad_norm": 1.0265972829923478, |
| "learning_rate": 3.7780927920390965e-06, |
| "loss": 0.1572, |
| "step": 461 |
| }, |
| { |
| "epoch": 0.36363636363636365, |
| "grad_norm": 0.9351090223515385, |
| "learning_rate": 3.772491427332557e-06, |
| "loss": 0.1317, |
| "step": 462 |
| }, |
| { |
| "epoch": 0.3644234553325462, |
| "grad_norm": 0.96672130032329, |
| "learning_rate": 3.766881427677563e-06, |
| "loss": 0.1474, |
| "step": 463 |
| }, |
| { |
| "epoch": 0.36521054702872885, |
| "grad_norm": 0.9284227954997755, |
| "learning_rate": 3.761262831142788e-06, |
| "loss": 0.144, |
| "step": 464 |
| }, |
| { |
| "epoch": 0.3659976387249115, |
| "grad_norm": 1.02329013434613, |
| "learning_rate": 3.755635675855238e-06, |
| "loss": 0.1459, |
| "step": 465 |
| }, |
| { |
| "epoch": 0.36678473042109405, |
| "grad_norm": 0.9548918394606087, |
| "learning_rate": 3.7500000000000005e-06, |
| "loss": 0.1431, |
| "step": 466 |
| }, |
| { |
| "epoch": 0.3675718221172767, |
| "grad_norm": 1.0029018534160843, |
| "learning_rate": 3.744355841819983e-06, |
| "loss": 0.1551, |
| "step": 467 |
| }, |
| { |
| "epoch": 0.36835891381345925, |
| "grad_norm": 1.0170466682076178, |
| "learning_rate": 3.7387032396156497e-06, |
| "loss": 0.1574, |
| "step": 468 |
| }, |
| { |
| "epoch": 0.3691460055096419, |
| "grad_norm": 0.950504547373793, |
| "learning_rate": 3.7330422317447686e-06, |
| "loss": 0.1413, |
| "step": 469 |
| }, |
| { |
| "epoch": 0.36993309720582446, |
| "grad_norm": 1.00490564394254, |
| "learning_rate": 3.7273728566221447e-06, |
| "loss": 0.1539, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.3707201889020071, |
| "grad_norm": 1.0241073155182219, |
| "learning_rate": 3.721695152719364e-06, |
| "loss": 0.1505, |
| "step": 471 |
| }, |
| { |
| "epoch": 0.3715072805981897, |
| "grad_norm": 1.0650129974030413, |
| "learning_rate": 3.716009158564528e-06, |
| "loss": 0.1517, |
| "step": 472 |
| }, |
| { |
| "epoch": 0.3722943722943723, |
| "grad_norm": 1.0412500508709561, |
| "learning_rate": 3.710314912741997e-06, |
| "loss": 0.1447, |
| "step": 473 |
| }, |
| { |
| "epoch": 0.3730814639905549, |
| "grad_norm": 1.0273490151395026, |
| "learning_rate": 3.7046124538921237e-06, |
| "loss": 0.1429, |
| "step": 474 |
| }, |
| { |
| "epoch": 0.3738685556867375, |
| "grad_norm": 0.9952543111661871, |
| "learning_rate": 3.698901820710995e-06, |
| "loss": 0.1418, |
| "step": 475 |
| }, |
| { |
| "epoch": 0.3746556473829201, |
| "grad_norm": 1.0824700054534682, |
| "learning_rate": 3.693183051950168e-06, |
| "loss": 0.1437, |
| "step": 476 |
| }, |
| { |
| "epoch": 0.3754427390791027, |
| "grad_norm": 1.0142752196453109, |
| "learning_rate": 3.6874561864164056e-06, |
| "loss": 0.1435, |
| "step": 477 |
| }, |
| { |
| "epoch": 0.3762298307752853, |
| "grad_norm": 0.9888106276082754, |
| "learning_rate": 3.6817212629714135e-06, |
| "loss": 0.1395, |
| "step": 478 |
| }, |
| { |
| "epoch": 0.37701692247146795, |
| "grad_norm": 0.9673698851206235, |
| "learning_rate": 3.675978320531579e-06, |
| "loss": 0.1425, |
| "step": 479 |
| }, |
| { |
| "epoch": 0.3778040141676505, |
| "grad_norm": 1.096283920214562, |
| "learning_rate": 3.670227398067705e-06, |
| "loss": 0.1515, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.37859110586383315, |
| "grad_norm": 1.0303413811027284, |
| "learning_rate": 3.664468534604745e-06, |
| "loss": 0.1462, |
| "step": 481 |
| }, |
| { |
| "epoch": 0.37937819756001573, |
| "grad_norm": 0.9550517801003708, |
| "learning_rate": 3.6587017692215387e-06, |
| "loss": 0.1483, |
| "step": 482 |
| }, |
| { |
| "epoch": 0.38016528925619836, |
| "grad_norm": 1.0499765951347195, |
| "learning_rate": 3.6529271410505483e-06, |
| "loss": 0.1516, |
| "step": 483 |
| }, |
| { |
| "epoch": 0.38095238095238093, |
| "grad_norm": 1.0612285971687154, |
| "learning_rate": 3.6471446892775896e-06, |
| "loss": 0.145, |
| "step": 484 |
| }, |
| { |
| "epoch": 0.38173947264856356, |
| "grad_norm": 0.9976574649139153, |
| "learning_rate": 3.6413544531415712e-06, |
| "loss": 0.1493, |
| "step": 485 |
| }, |
| { |
| "epoch": 0.3825265643447462, |
| "grad_norm": 1.011974051278155, |
| "learning_rate": 3.635556471934224e-06, |
| "loss": 0.1557, |
| "step": 486 |
| }, |
| { |
| "epoch": 0.38331365604092876, |
| "grad_norm": 1.015959048224715, |
| "learning_rate": 3.629750784999835e-06, |
| "loss": 0.152, |
| "step": 487 |
| }, |
| { |
| "epoch": 0.3841007477371114, |
| "grad_norm": 0.9638439392236781, |
| "learning_rate": 3.623937431734982e-06, |
| "loss": 0.1464, |
| "step": 488 |
| }, |
| { |
| "epoch": 0.38488783943329397, |
| "grad_norm": 0.9820530085625633, |
| "learning_rate": 3.6181164515882663e-06, |
| "loss": 0.1468, |
| "step": 489 |
| }, |
| { |
| "epoch": 0.3856749311294766, |
| "grad_norm": 0.9281524539517508, |
| "learning_rate": 3.6122878840600417e-06, |
| "loss": 0.1451, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.38646202282565917, |
| "grad_norm": 1.039305922376239, |
| "learning_rate": 3.606451768702151e-06, |
| "loss": 0.1486, |
| "step": 491 |
| }, |
| { |
| "epoch": 0.3872491145218418, |
| "grad_norm": 1.026987888426606, |
| "learning_rate": 3.600608145117656e-06, |
| "loss": 0.1381, |
| "step": 492 |
| }, |
| { |
| "epoch": 0.3880362062180244, |
| "grad_norm": 1.058827889093346, |
| "learning_rate": 3.594757052960566e-06, |
| "loss": 0.1555, |
| "step": 493 |
| }, |
| { |
| "epoch": 0.388823297914207, |
| "grad_norm": 1.0027016575115129, |
| "learning_rate": 3.588898531935573e-06, |
| "loss": 0.1413, |
| "step": 494 |
| }, |
| { |
| "epoch": 0.38961038961038963, |
| "grad_norm": 1.0766471714794614, |
| "learning_rate": 3.583032621797778e-06, |
| "loss": 0.1418, |
| "step": 495 |
| }, |
| { |
| "epoch": 0.3903974813065722, |
| "grad_norm": 1.0326313481110534, |
| "learning_rate": 3.5771593623524263e-06, |
| "loss": 0.1345, |
| "step": 496 |
| }, |
| { |
| "epoch": 0.39118457300275483, |
| "grad_norm": 0.9649958546178075, |
| "learning_rate": 3.5712787934546336e-06, |
| "loss": 0.1397, |
| "step": 497 |
| }, |
| { |
| "epoch": 0.3919716646989374, |
| "grad_norm": 1.0461258832079, |
| "learning_rate": 3.5653909550091138e-06, |
| "loss": 0.16, |
| "step": 498 |
| }, |
| { |
| "epoch": 0.39275875639512003, |
| "grad_norm": 0.9741702004779168, |
| "learning_rate": 3.559495886969916e-06, |
| "loss": 0.1366, |
| "step": 499 |
| }, |
| { |
| "epoch": 0.39354584809130266, |
| "grad_norm": 0.9875184829637668, |
| "learning_rate": 3.553593629340144e-06, |
| "loss": 0.1391, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.39354584809130266, |
| "eval_loss": 0.14773064851760864, |
| "eval_runtime": 18.0322, |
| "eval_samples_per_second": 45.585, |
| "eval_steps_per_second": 5.712, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.39433293978748524, |
| "grad_norm": 0.9563741831859393, |
| "learning_rate": 3.5476842221716915e-06, |
| "loss": 0.1453, |
| "step": 501 |
| }, |
| { |
| "epoch": 0.39512003148366787, |
| "grad_norm": 0.9839000041167648, |
| "learning_rate": 3.541767705564967e-06, |
| "loss": 0.1509, |
| "step": 502 |
| }, |
| { |
| "epoch": 0.39590712317985044, |
| "grad_norm": 0.9666175985112762, |
| "learning_rate": 3.535844119668622e-06, |
| "loss": 0.1436, |
| "step": 503 |
| }, |
| { |
| "epoch": 0.39669421487603307, |
| "grad_norm": 1.0513295542177603, |
| "learning_rate": 3.5299135046792816e-06, |
| "loss": 0.1371, |
| "step": 504 |
| }, |
| { |
| "epoch": 0.39748130657221564, |
| "grad_norm": 1.0136623338528887, |
| "learning_rate": 3.5239759008412666e-06, |
| "loss": 0.1498, |
| "step": 505 |
| }, |
| { |
| "epoch": 0.39826839826839827, |
| "grad_norm": 0.9764920494655156, |
| "learning_rate": 3.518031348446324e-06, |
| "loss": 0.1371, |
| "step": 506 |
| }, |
| { |
| "epoch": 0.3990554899645809, |
| "grad_norm": 1.0113031849627157, |
| "learning_rate": 3.5120798878333544e-06, |
| "loss": 0.1453, |
| "step": 507 |
| }, |
| { |
| "epoch": 0.3998425816607635, |
| "grad_norm": 0.9947509560502654, |
| "learning_rate": 3.506121559388135e-06, |
| "loss": 0.1233, |
| "step": 508 |
| }, |
| { |
| "epoch": 0.4006296733569461, |
| "grad_norm": 1.1135464243984814, |
| "learning_rate": 3.500156403543046e-06, |
| "loss": 0.151, |
| "step": 509 |
| }, |
| { |
| "epoch": 0.4014167650531287, |
| "grad_norm": 1.0687025563863246, |
| "learning_rate": 3.4941844607768007e-06, |
| "loss": 0.1384, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.4022038567493113, |
| "grad_norm": 0.9654525860741724, |
| "learning_rate": 3.488205771614164e-06, |
| "loss": 0.1348, |
| "step": 511 |
| }, |
| { |
| "epoch": 0.4029909484454939, |
| "grad_norm": 1.07357744190682, |
| "learning_rate": 3.4822203766256834e-06, |
| "loss": 0.1412, |
| "step": 512 |
| }, |
| { |
| "epoch": 0.4037780401416765, |
| "grad_norm": 1.2491546536330014, |
| "learning_rate": 3.4762283164274104e-06, |
| "loss": 0.1523, |
| "step": 513 |
| }, |
| { |
| "epoch": 0.40456513183785914, |
| "grad_norm": 1.0398955239354635, |
| "learning_rate": 3.4702296316806243e-06, |
| "loss": 0.1507, |
| "step": 514 |
| }, |
| { |
| "epoch": 0.4053522235340417, |
| "grad_norm": 0.947562520308943, |
| "learning_rate": 3.4642243630915606e-06, |
| "loss": 0.1486, |
| "step": 515 |
| }, |
| { |
| "epoch": 0.40613931523022434, |
| "grad_norm": 0.9405204018759319, |
| "learning_rate": 3.45821255141113e-06, |
| "loss": 0.1287, |
| "step": 516 |
| }, |
| { |
| "epoch": 0.4069264069264069, |
| "grad_norm": 1.01025400774114, |
| "learning_rate": 3.452194237434642e-06, |
| "loss": 0.1349, |
| "step": 517 |
| }, |
| { |
| "epoch": 0.40771349862258954, |
| "grad_norm": 1.0404932578099988, |
| "learning_rate": 3.446169462001534e-06, |
| "loss": 0.1508, |
| "step": 518 |
| }, |
| { |
| "epoch": 0.4085005903187721, |
| "grad_norm": 1.029425420995215, |
| "learning_rate": 3.4401382659950868e-06, |
| "loss": 0.1362, |
| "step": 519 |
| }, |
| { |
| "epoch": 0.40928768201495475, |
| "grad_norm": 1.025768159905711, |
| "learning_rate": 3.4341006903421493e-06, |
| "loss": 0.1437, |
| "step": 520 |
| }, |
| { |
| "epoch": 0.4100747737111374, |
| "grad_norm": 0.9507044448226175, |
| "learning_rate": 3.4280567760128658e-06, |
| "loss": 0.1393, |
| "step": 521 |
| }, |
| { |
| "epoch": 0.41086186540731995, |
| "grad_norm": 1.0374082813027519, |
| "learning_rate": 3.4220065640203916e-06, |
| "loss": 0.16, |
| "step": 522 |
| }, |
| { |
| "epoch": 0.4116489571035026, |
| "grad_norm": 0.9378353888939086, |
| "learning_rate": 3.415950095420616e-06, |
| "loss": 0.1355, |
| "step": 523 |
| }, |
| { |
| "epoch": 0.41243604879968515, |
| "grad_norm": 0.924561930587711, |
| "learning_rate": 3.4098874113118863e-06, |
| "loss": 0.1452, |
| "step": 524 |
| }, |
| { |
| "epoch": 0.4132231404958678, |
| "grad_norm": 0.9505049489489825, |
| "learning_rate": 3.403818552834727e-06, |
| "loss": 0.1448, |
| "step": 525 |
| }, |
| { |
| "epoch": 0.41401023219205035, |
| "grad_norm": 0.9701870488491394, |
| "learning_rate": 3.397743561171562e-06, |
| "loss": 0.1341, |
| "step": 526 |
| }, |
| { |
| "epoch": 0.414797323888233, |
| "grad_norm": 0.9122288876708122, |
| "learning_rate": 3.3916624775464318e-06, |
| "loss": 0.1291, |
| "step": 527 |
| }, |
| { |
| "epoch": 0.4155844155844156, |
| "grad_norm": 0.985150804267496, |
| "learning_rate": 3.385575343224718e-06, |
| "loss": 0.141, |
| "step": 528 |
| }, |
| { |
| "epoch": 0.4163715072805982, |
| "grad_norm": 0.9910844190276262, |
| "learning_rate": 3.3794821995128606e-06, |
| "loss": 0.1473, |
| "step": 529 |
| }, |
| { |
| "epoch": 0.4171585989767808, |
| "grad_norm": 0.9925292173111532, |
| "learning_rate": 3.3733830877580796e-06, |
| "loss": 0.1492, |
| "step": 530 |
| }, |
| { |
| "epoch": 0.4179456906729634, |
| "grad_norm": 0.9483537804421872, |
| "learning_rate": 3.3672780493480927e-06, |
| "loss": 0.1476, |
| "step": 531 |
| }, |
| { |
| "epoch": 0.418732782369146, |
| "grad_norm": 0.9716970355806354, |
| "learning_rate": 3.3611671257108323e-06, |
| "loss": 0.1288, |
| "step": 532 |
| }, |
| { |
| "epoch": 0.4195198740653286, |
| "grad_norm": 1.0525983321400059, |
| "learning_rate": 3.3550503583141726e-06, |
| "loss": 0.1541, |
| "step": 533 |
| }, |
| { |
| "epoch": 0.4203069657615112, |
| "grad_norm": 0.9476841095185634, |
| "learning_rate": 3.3489277886656373e-06, |
| "loss": 0.1395, |
| "step": 534 |
| }, |
| { |
| "epoch": 0.42109405745769385, |
| "grad_norm": 0.8883884320293254, |
| "learning_rate": 3.342799458312127e-06, |
| "loss": 0.1374, |
| "step": 535 |
| }, |
| { |
| "epoch": 0.4218811491538764, |
| "grad_norm": 1.026818858865084, |
| "learning_rate": 3.336665408839633e-06, |
| "loss": 0.1413, |
| "step": 536 |
| }, |
| { |
| "epoch": 0.42266824085005905, |
| "grad_norm": 0.9146805645048051, |
| "learning_rate": 3.330525681872954e-06, |
| "loss": 0.1352, |
| "step": 537 |
| }, |
| { |
| "epoch": 0.4234553325462416, |
| "grad_norm": 1.0439955820386841, |
| "learning_rate": 3.3243803190754166e-06, |
| "loss": 0.1482, |
| "step": 538 |
| }, |
| { |
| "epoch": 0.42424242424242425, |
| "grad_norm": 0.9964413472110166, |
| "learning_rate": 3.3182293621485923e-06, |
| "loss": 0.1524, |
| "step": 539 |
| }, |
| { |
| "epoch": 0.42502951593860683, |
| "grad_norm": 0.9626977177442709, |
| "learning_rate": 3.312072852832012e-06, |
| "loss": 0.1427, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.42581660763478946, |
| "grad_norm": 0.8939878261316884, |
| "learning_rate": 3.3059108329028845e-06, |
| "loss": 0.1283, |
| "step": 541 |
| }, |
| { |
| "epoch": 0.4266036993309721, |
| "grad_norm": 1.0155176108909485, |
| "learning_rate": 3.299743344175814e-06, |
| "loss": 0.1434, |
| "step": 542 |
| }, |
| { |
| "epoch": 0.42739079102715466, |
| "grad_norm": 0.983969699589635, |
| "learning_rate": 3.293570428502515e-06, |
| "loss": 0.1479, |
| "step": 543 |
| }, |
| { |
| "epoch": 0.4281778827233373, |
| "grad_norm": 1.006021089515589, |
| "learning_rate": 3.287392127771526e-06, |
| "loss": 0.1386, |
| "step": 544 |
| }, |
| { |
| "epoch": 0.42896497441951986, |
| "grad_norm": 0.9594215523929834, |
| "learning_rate": 3.2812084839079316e-06, |
| "loss": 0.1326, |
| "step": 545 |
| }, |
| { |
| "epoch": 0.4297520661157025, |
| "grad_norm": 0.958170693889915, |
| "learning_rate": 3.275019538873071e-06, |
| "loss": 0.1418, |
| "step": 546 |
| }, |
| { |
| "epoch": 0.43053915781188506, |
| "grad_norm": 1.0256772435691563, |
| "learning_rate": 3.268825334664259e-06, |
| "loss": 0.1526, |
| "step": 547 |
| }, |
| { |
| "epoch": 0.4313262495080677, |
| "grad_norm": 1.0552921930006323, |
| "learning_rate": 3.2626259133144955e-06, |
| "loss": 0.1441, |
| "step": 548 |
| }, |
| { |
| "epoch": 0.43211334120425027, |
| "grad_norm": 1.0605556822333475, |
| "learning_rate": 3.2564213168921867e-06, |
| "loss": 0.1431, |
| "step": 549 |
| }, |
| { |
| "epoch": 0.4329004329004329, |
| "grad_norm": 0.982864648139079, |
| "learning_rate": 3.2502115875008523e-06, |
| "loss": 0.149, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.4336875245966155, |
| "grad_norm": 1.0049025924744737, |
| "learning_rate": 3.2439967672788462e-06, |
| "loss": 0.1334, |
| "step": 551 |
| }, |
| { |
| "epoch": 0.4344746162927981, |
| "grad_norm": 0.9263761254409442, |
| "learning_rate": 3.2377768983990677e-06, |
| "loss": 0.1401, |
| "step": 552 |
| }, |
| { |
| "epoch": 0.43526170798898073, |
| "grad_norm": 1.0390883988019344, |
| "learning_rate": 3.2315520230686747e-06, |
| "loss": 0.1493, |
| "step": 553 |
| }, |
| { |
| "epoch": 0.4360487996851633, |
| "grad_norm": 0.9913895430005143, |
| "learning_rate": 3.2253221835287984e-06, |
| "loss": 0.1406, |
| "step": 554 |
| }, |
| { |
| "epoch": 0.43683589138134593, |
| "grad_norm": 0.9801664753977715, |
| "learning_rate": 3.2190874220542577e-06, |
| "loss": 0.1341, |
| "step": 555 |
| }, |
| { |
| "epoch": 0.4376229830775285, |
| "grad_norm": 0.9519041413125566, |
| "learning_rate": 3.2128477809532687e-06, |
| "loss": 0.1469, |
| "step": 556 |
| }, |
| { |
| "epoch": 0.43841007477371113, |
| "grad_norm": 1.0289764585305627, |
| "learning_rate": 3.2066033025671612e-06, |
| "loss": 0.1473, |
| "step": 557 |
| }, |
| { |
| "epoch": 0.43919716646989376, |
| "grad_norm": 1.0005689404595521, |
| "learning_rate": 3.200354029270091e-06, |
| "loss": 0.1477, |
| "step": 558 |
| }, |
| { |
| "epoch": 0.43998425816607634, |
| "grad_norm": 1.056463362355424, |
| "learning_rate": 3.1941000034687516e-06, |
| "loss": 0.1488, |
| "step": 559 |
| }, |
| { |
| "epoch": 0.44077134986225897, |
| "grad_norm": 0.9574144607007496, |
| "learning_rate": 3.187841267602084e-06, |
| "loss": 0.1445, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.44155844155844154, |
| "grad_norm": 0.9562953543913302, |
| "learning_rate": 3.1815778641409924e-06, |
| "loss": 0.1414, |
| "step": 561 |
| }, |
| { |
| "epoch": 0.44234553325462417, |
| "grad_norm": 0.9444651486667015, |
| "learning_rate": 3.1753098355880557e-06, |
| "loss": 0.138, |
| "step": 562 |
| }, |
| { |
| "epoch": 0.44313262495080674, |
| "grad_norm": 0.9465351053953429, |
| "learning_rate": 3.169037224477236e-06, |
| "loss": 0.1437, |
| "step": 563 |
| }, |
| { |
| "epoch": 0.44391971664698937, |
| "grad_norm": 1.0206836940486426, |
| "learning_rate": 3.162760073373594e-06, |
| "loss": 0.1411, |
| "step": 564 |
| }, |
| { |
| "epoch": 0.444706808343172, |
| "grad_norm": 1.0878905236564318, |
| "learning_rate": 3.1564784248729965e-06, |
| "loss": 0.1408, |
| "step": 565 |
| }, |
| { |
| "epoch": 0.4454939000393546, |
| "grad_norm": 1.0130102955883906, |
| "learning_rate": 3.15019232160183e-06, |
| "loss": 0.1428, |
| "step": 566 |
| }, |
| { |
| "epoch": 0.4462809917355372, |
| "grad_norm": 0.980086016231054, |
| "learning_rate": 3.1439018062167092e-06, |
| "loss": 0.143, |
| "step": 567 |
| }, |
| { |
| "epoch": 0.4470680834317198, |
| "grad_norm": 1.0249915137559014, |
| "learning_rate": 3.1376069214041917e-06, |
| "loss": 0.1471, |
| "step": 568 |
| }, |
| { |
| "epoch": 0.4478551751279024, |
| "grad_norm": 1.1016327132095007, |
| "learning_rate": 3.1313077098804817e-06, |
| "loss": 0.1606, |
| "step": 569 |
| }, |
| { |
| "epoch": 0.448642266824085, |
| "grad_norm": 1.0411771801722989, |
| "learning_rate": 3.1250042143911462e-06, |
| "loss": 0.1499, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.4494293585202676, |
| "grad_norm": 1.0122030093902548, |
| "learning_rate": 3.118696477710822e-06, |
| "loss": 0.141, |
| "step": 571 |
| }, |
| { |
| "epoch": 0.45021645021645024, |
| "grad_norm": 1.0708872672849516, |
| "learning_rate": 3.1123845426429265e-06, |
| "loss": 0.128, |
| "step": 572 |
| }, |
| { |
| "epoch": 0.4510035419126328, |
| "grad_norm": 1.029737403462412, |
| "learning_rate": 3.106068452019365e-06, |
| "loss": 0.1383, |
| "step": 573 |
| }, |
| { |
| "epoch": 0.45179063360881544, |
| "grad_norm": 0.9988296671107193, |
| "learning_rate": 3.099748248700245e-06, |
| "loss": 0.1376, |
| "step": 574 |
| }, |
| { |
| "epoch": 0.452577725304998, |
| "grad_norm": 1.0475513726672416, |
| "learning_rate": 3.0934239755735782e-06, |
| "loss": 0.1355, |
| "step": 575 |
| }, |
| { |
| "epoch": 0.45336481700118064, |
| "grad_norm": 1.0654745191838768, |
| "learning_rate": 3.0870956755549973e-06, |
| "loss": 0.143, |
| "step": 576 |
| }, |
| { |
| "epoch": 0.4541519086973632, |
| "grad_norm": 0.9397526290083124, |
| "learning_rate": 3.0807633915874585e-06, |
| "loss": 0.1406, |
| "step": 577 |
| }, |
| { |
| "epoch": 0.45493900039354584, |
| "grad_norm": 1.052837564760308, |
| "learning_rate": 3.0744271666409526e-06, |
| "loss": 0.1454, |
| "step": 578 |
| }, |
| { |
| "epoch": 0.4557260920897285, |
| "grad_norm": 1.1289865006459998, |
| "learning_rate": 3.0680870437122145e-06, |
| "loss": 0.1554, |
| "step": 579 |
| }, |
| { |
| "epoch": 0.45651318378591105, |
| "grad_norm": 0.9614320131595296, |
| "learning_rate": 3.0617430658244295e-06, |
| "loss": 0.1368, |
| "step": 580 |
| }, |
| { |
| "epoch": 0.4573002754820937, |
| "grad_norm": 0.9849943444472453, |
| "learning_rate": 3.0553952760269427e-06, |
| "loss": 0.1372, |
| "step": 581 |
| }, |
| { |
| "epoch": 0.45808736717827625, |
| "grad_norm": 0.9938446057985301, |
| "learning_rate": 3.0490437173949656e-06, |
| "loss": 0.1397, |
| "step": 582 |
| }, |
| { |
| "epoch": 0.4588744588744589, |
| "grad_norm": 0.9430590894578916, |
| "learning_rate": 3.0426884330292844e-06, |
| "loss": 0.1404, |
| "step": 583 |
| }, |
| { |
| "epoch": 0.45966155057064145, |
| "grad_norm": 0.8998337127762756, |
| "learning_rate": 3.0363294660559685e-06, |
| "loss": 0.133, |
| "step": 584 |
| }, |
| { |
| "epoch": 0.4604486422668241, |
| "grad_norm": 0.9469777276964015, |
| "learning_rate": 3.0299668596260755e-06, |
| "loss": 0.1429, |
| "step": 585 |
| }, |
| { |
| "epoch": 0.4612357339630067, |
| "grad_norm": 0.9961208676961326, |
| "learning_rate": 3.023600656915362e-06, |
| "loss": 0.1381, |
| "step": 586 |
| }, |
| { |
| "epoch": 0.4620228256591893, |
| "grad_norm": 0.9726679702119774, |
| "learning_rate": 3.017230901123985e-06, |
| "loss": 0.1391, |
| "step": 587 |
| }, |
| { |
| "epoch": 0.4628099173553719, |
| "grad_norm": 1.016233754336966, |
| "learning_rate": 3.0108576354762176e-06, |
| "loss": 0.1464, |
| "step": 588 |
| }, |
| { |
| "epoch": 0.4635970090515545, |
| "grad_norm": 0.891890572894692, |
| "learning_rate": 3.0044809032201448e-06, |
| "loss": 0.1312, |
| "step": 589 |
| }, |
| { |
| "epoch": 0.4643841007477371, |
| "grad_norm": 0.9300922465018149, |
| "learning_rate": 2.9981007476273787e-06, |
| "loss": 0.1272, |
| "step": 590 |
| }, |
| { |
| "epoch": 0.4651711924439197, |
| "grad_norm": 1.0381540629264334, |
| "learning_rate": 2.9917172119927607e-06, |
| "loss": 0.1479, |
| "step": 591 |
| }, |
| { |
| "epoch": 0.4659582841401023, |
| "grad_norm": 1.0642195977009175, |
| "learning_rate": 2.9853303396340695e-06, |
| "loss": 0.1364, |
| "step": 592 |
| }, |
| { |
| "epoch": 0.46674537583628495, |
| "grad_norm": 0.9295272897205104, |
| "learning_rate": 2.9789401738917244e-06, |
| "loss": 0.1249, |
| "step": 593 |
| }, |
| { |
| "epoch": 0.4675324675324675, |
| "grad_norm": 1.0180029223750298, |
| "learning_rate": 2.9725467581284944e-06, |
| "loss": 0.1407, |
| "step": 594 |
| }, |
| { |
| "epoch": 0.46831955922865015, |
| "grad_norm": 1.1385262618991847, |
| "learning_rate": 2.966150135729203e-06, |
| "loss": 0.1502, |
| "step": 595 |
| }, |
| { |
| "epoch": 0.4691066509248327, |
| "grad_norm": 1.0067715931565462, |
| "learning_rate": 2.9597503501004345e-06, |
| "loss": 0.1286, |
| "step": 596 |
| }, |
| { |
| "epoch": 0.46989374262101535, |
| "grad_norm": 0.9465710841629198, |
| "learning_rate": 2.9533474446702346e-06, |
| "loss": 0.1358, |
| "step": 597 |
| }, |
| { |
| "epoch": 0.4706808343171979, |
| "grad_norm": 1.04804051578767, |
| "learning_rate": 2.946941462887824e-06, |
| "loss": 0.1333, |
| "step": 598 |
| }, |
| { |
| "epoch": 0.47146792601338056, |
| "grad_norm": 1.0917713383450702, |
| "learning_rate": 2.940532448223296e-06, |
| "loss": 0.1462, |
| "step": 599 |
| }, |
| { |
| "epoch": 0.4722550177095632, |
| "grad_norm": 0.9580513732250364, |
| "learning_rate": 2.9341204441673267e-06, |
| "loss": 0.1321, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.47304210940574576, |
| "grad_norm": 0.9439921102070582, |
| "learning_rate": 2.927705494230875e-06, |
| "loss": 0.1441, |
| "step": 601 |
| }, |
| { |
| "epoch": 0.4738292011019284, |
| "grad_norm": 1.0178216949448748, |
| "learning_rate": 2.9212876419448943e-06, |
| "loss": 0.1405, |
| "step": 602 |
| }, |
| { |
| "epoch": 0.47461629279811096, |
| "grad_norm": 1.0297426762245179, |
| "learning_rate": 2.9148669308600298e-06, |
| "loss": 0.1392, |
| "step": 603 |
| }, |
| { |
| "epoch": 0.4754033844942936, |
| "grad_norm": 0.9415986568330708, |
| "learning_rate": 2.9084434045463255e-06, |
| "loss": 0.1282, |
| "step": 604 |
| }, |
| { |
| "epoch": 0.47619047619047616, |
| "grad_norm": 1.0337230890115443, |
| "learning_rate": 2.9020171065929327e-06, |
| "loss": 0.1394, |
| "step": 605 |
| }, |
| { |
| "epoch": 0.4769775678866588, |
| "grad_norm": 1.0540052550471415, |
| "learning_rate": 2.895588080607807e-06, |
| "loss": 0.1472, |
| "step": 606 |
| }, |
| { |
| "epoch": 0.4777646595828414, |
| "grad_norm": 1.0081872244466563, |
| "learning_rate": 2.8891563702174174e-06, |
| "loss": 0.1372, |
| "step": 607 |
| }, |
| { |
| "epoch": 0.478551751279024, |
| "grad_norm": 1.0145019904402564, |
| "learning_rate": 2.8827220190664505e-06, |
| "loss": 0.1399, |
| "step": 608 |
| }, |
| { |
| "epoch": 0.4793388429752066, |
| "grad_norm": 1.0258604105718838, |
| "learning_rate": 2.8762850708175098e-06, |
| "loss": 0.1499, |
| "step": 609 |
| }, |
| { |
| "epoch": 0.4801259346713892, |
| "grad_norm": 1.0836484331180423, |
| "learning_rate": 2.869845569150825e-06, |
| "loss": 0.1388, |
| "step": 610 |
| }, |
| { |
| "epoch": 0.4809130263675718, |
| "grad_norm": 0.9946389106293178, |
| "learning_rate": 2.863403557763951e-06, |
| "loss": 0.1323, |
| "step": 611 |
| }, |
| { |
| "epoch": 0.4817001180637544, |
| "grad_norm": 0.9968164583365795, |
| "learning_rate": 2.856959080371474e-06, |
| "loss": 0.1402, |
| "step": 612 |
| }, |
| { |
| "epoch": 0.48248720975993703, |
| "grad_norm": 1.0526146596249044, |
| "learning_rate": 2.8505121807047155e-06, |
| "loss": 0.1342, |
| "step": 613 |
| }, |
| { |
| "epoch": 0.48327430145611966, |
| "grad_norm": 0.9881771003275511, |
| "learning_rate": 2.8440629025114308e-06, |
| "loss": 0.1414, |
| "step": 614 |
| }, |
| { |
| "epoch": 0.48406139315230223, |
| "grad_norm": 1.0170639400089367, |
| "learning_rate": 2.8376112895555184e-06, |
| "loss": 0.1415, |
| "step": 615 |
| }, |
| { |
| "epoch": 0.48484848484848486, |
| "grad_norm": 0.9618458339894986, |
| "learning_rate": 2.83115738561672e-06, |
| "loss": 0.125, |
| "step": 616 |
| }, |
| { |
| "epoch": 0.48563557654466744, |
| "grad_norm": 1.166675709546666, |
| "learning_rate": 2.8247012344903235e-06, |
| "loss": 0.1537, |
| "step": 617 |
| }, |
| { |
| "epoch": 0.48642266824085006, |
| "grad_norm": 1.0308351089525765, |
| "learning_rate": 2.8182428799868643e-06, |
| "loss": 0.1435, |
| "step": 618 |
| }, |
| { |
| "epoch": 0.48720975993703264, |
| "grad_norm": 0.9008466844444718, |
| "learning_rate": 2.811782365931832e-06, |
| "loss": 0.1255, |
| "step": 619 |
| }, |
| { |
| "epoch": 0.48799685163321527, |
| "grad_norm": 1.0328591551300574, |
| "learning_rate": 2.8053197361653684e-06, |
| "loss": 0.1431, |
| "step": 620 |
| }, |
| { |
| "epoch": 0.4887839433293979, |
| "grad_norm": 1.0223227370370647, |
| "learning_rate": 2.7988550345419733e-06, |
| "loss": 0.1302, |
| "step": 621 |
| }, |
| { |
| "epoch": 0.48957103502558047, |
| "grad_norm": 1.0130656273790444, |
| "learning_rate": 2.792388304930207e-06, |
| "loss": 0.1413, |
| "step": 622 |
| }, |
| { |
| "epoch": 0.4903581267217631, |
| "grad_norm": 0.9678629630338841, |
| "learning_rate": 2.7859195912123875e-06, |
| "loss": 0.1411, |
| "step": 623 |
| }, |
| { |
| "epoch": 0.4911452184179457, |
| "grad_norm": 1.0630235458290422, |
| "learning_rate": 2.779448937284302e-06, |
| "loss": 0.144, |
| "step": 624 |
| }, |
| { |
| "epoch": 0.4919323101141283, |
| "grad_norm": 1.1368466359085148, |
| "learning_rate": 2.772976387054899e-06, |
| "loss": 0.1603, |
| "step": 625 |
| }, |
| { |
| "epoch": 0.4927194018103109, |
| "grad_norm": 1.0638972206646764, |
| "learning_rate": 2.766501984445999e-06, |
| "loss": 0.1469, |
| "step": 626 |
| }, |
| { |
| "epoch": 0.4935064935064935, |
| "grad_norm": 0.9878723437777639, |
| "learning_rate": 2.7600257733919887e-06, |
| "loss": 0.1347, |
| "step": 627 |
| }, |
| { |
| "epoch": 0.49429358520267613, |
| "grad_norm": 0.9482438523704221, |
| "learning_rate": 2.7535477978395295e-06, |
| "loss": 0.1301, |
| "step": 628 |
| }, |
| { |
| "epoch": 0.4950806768988587, |
| "grad_norm": 1.0213978192147322, |
| "learning_rate": 2.7470681017472556e-06, |
| "loss": 0.1442, |
| "step": 629 |
| }, |
| { |
| "epoch": 0.49586776859504134, |
| "grad_norm": 1.0113916573838844, |
| "learning_rate": 2.740586729085476e-06, |
| "loss": 0.1477, |
| "step": 630 |
| }, |
| { |
| "epoch": 0.4966548602912239, |
| "grad_norm": 1.0353820062718653, |
| "learning_rate": 2.7341037238358774e-06, |
| "loss": 0.1483, |
| "step": 631 |
| }, |
| { |
| "epoch": 0.49744195198740654, |
| "grad_norm": 1.0552352024187672, |
| "learning_rate": 2.727619129991224e-06, |
| "loss": 0.1328, |
| "step": 632 |
| }, |
| { |
| "epoch": 0.4982290436835891, |
| "grad_norm": 0.9937705442973395, |
| "learning_rate": 2.7211329915550615e-06, |
| "loss": 0.1409, |
| "step": 633 |
| }, |
| { |
| "epoch": 0.49901613537977174, |
| "grad_norm": 1.0486309341654392, |
| "learning_rate": 2.714645352541415e-06, |
| "loss": 0.15, |
| "step": 634 |
| }, |
| { |
| "epoch": 0.49980322707595437, |
| "grad_norm": 1.015369060592149, |
| "learning_rate": 2.7081562569744948e-06, |
| "loss": 0.1298, |
| "step": 635 |
| }, |
| { |
| "epoch": 0.500590318772137, |
| "grad_norm": 1.014091287328762, |
| "learning_rate": 2.701665748888393e-06, |
| "loss": 0.139, |
| "step": 636 |
| }, |
| { |
| "epoch": 0.5013774104683195, |
| "grad_norm": 1.010797057516188, |
| "learning_rate": 2.695173872326788e-06, |
| "loss": 0.1306, |
| "step": 637 |
| }, |
| { |
| "epoch": 0.5021645021645021, |
| "grad_norm": 0.9886264059190445, |
| "learning_rate": 2.6886806713426435e-06, |
| "loss": 0.1493, |
| "step": 638 |
| }, |
| { |
| "epoch": 0.5029515938606848, |
| "grad_norm": 0.9006497838538798, |
| "learning_rate": 2.6821861899979116e-06, |
| "loss": 0.127, |
| "step": 639 |
| }, |
| { |
| "epoch": 0.5037386855568674, |
| "grad_norm": 1.0409028373992908, |
| "learning_rate": 2.6756904723632325e-06, |
| "loss": 0.1453, |
| "step": 640 |
| }, |
| { |
| "epoch": 0.50452577725305, |
| "grad_norm": 0.9741943151013064, |
| "learning_rate": 2.6691935625176357e-06, |
| "loss": 0.1353, |
| "step": 641 |
| }, |
| { |
| "epoch": 0.5053128689492326, |
| "grad_norm": 0.949636504358609, |
| "learning_rate": 2.6626955045482405e-06, |
| "loss": 0.1335, |
| "step": 642 |
| }, |
| { |
| "epoch": 0.5060999606454152, |
| "grad_norm": 0.9249297082390363, |
| "learning_rate": 2.6561963425499575e-06, |
| "loss": 0.1338, |
| "step": 643 |
| }, |
| { |
| "epoch": 0.5068870523415978, |
| "grad_norm": 1.0151555535359889, |
| "learning_rate": 2.649696120625188e-06, |
| "loss": 0.1515, |
| "step": 644 |
| }, |
| { |
| "epoch": 0.5076741440377804, |
| "grad_norm": 1.039472398997662, |
| "learning_rate": 2.643194882883528e-06, |
| "loss": 0.1474, |
| "step": 645 |
| }, |
| { |
| "epoch": 0.508461235733963, |
| "grad_norm": 0.9434610266773801, |
| "learning_rate": 2.6366926734414648e-06, |
| "loss": 0.1304, |
| "step": 646 |
| }, |
| { |
| "epoch": 0.5092483274301456, |
| "grad_norm": 0.8865198426440791, |
| "learning_rate": 2.6301895364220816e-06, |
| "loss": 0.1202, |
| "step": 647 |
| }, |
| { |
| "epoch": 0.5100354191263282, |
| "grad_norm": 0.9546278944005607, |
| "learning_rate": 2.6236855159547527e-06, |
| "loss": 0.1291, |
| "step": 648 |
| }, |
| { |
| "epoch": 0.5108225108225108, |
| "grad_norm": 0.9693013564144493, |
| "learning_rate": 2.6171806561748503e-06, |
| "loss": 0.1339, |
| "step": 649 |
| }, |
| { |
| "epoch": 0.5116096025186935, |
| "grad_norm": 1.0027100891356027, |
| "learning_rate": 2.610675001223441e-06, |
| "loss": 0.1407, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.512396694214876, |
| "grad_norm": 0.8560729540932264, |
| "learning_rate": 2.6041685952469877e-06, |
| "loss": 0.116, |
| "step": 651 |
| }, |
| { |
| "epoch": 0.5131837859110586, |
| "grad_norm": 0.9257606438562741, |
| "learning_rate": 2.597661482397049e-06, |
| "loss": 0.1262, |
| "step": 652 |
| }, |
| { |
| "epoch": 0.5139708776072412, |
| "grad_norm": 1.0514657045725575, |
| "learning_rate": 2.5911537068299803e-06, |
| "loss": 0.1469, |
| "step": 653 |
| }, |
| { |
| "epoch": 0.5147579693034239, |
| "grad_norm": 0.9545058570137028, |
| "learning_rate": 2.584645312706634e-06, |
| "loss": 0.1302, |
| "step": 654 |
| }, |
| { |
| "epoch": 0.5155450609996065, |
| "grad_norm": 0.9392962167917809, |
| "learning_rate": 2.5781363441920614e-06, |
| "loss": 0.1335, |
| "step": 655 |
| }, |
| { |
| "epoch": 0.516332152695789, |
| "grad_norm": 0.9496925045032614, |
| "learning_rate": 2.5716268454552094e-06, |
| "loss": 0.135, |
| "step": 656 |
| }, |
| { |
| "epoch": 0.5171192443919717, |
| "grad_norm": 1.002838327785164, |
| "learning_rate": 2.565116860668625e-06, |
| "loss": 0.1316, |
| "step": 657 |
| }, |
| { |
| "epoch": 0.5179063360881543, |
| "grad_norm": 1.053225279922735, |
| "learning_rate": 2.5586064340081516e-06, |
| "loss": 0.1512, |
| "step": 658 |
| }, |
| { |
| "epoch": 0.5186934277843369, |
| "grad_norm": 0.9441130740117648, |
| "learning_rate": 2.5520956096526323e-06, |
| "loss": 0.131, |
| "step": 659 |
| }, |
| { |
| "epoch": 0.5194805194805194, |
| "grad_norm": 0.9889334534500898, |
| "learning_rate": 2.5455844317836077e-06, |
| "loss": 0.1331, |
| "step": 660 |
| }, |
| { |
| "epoch": 0.5202676111767021, |
| "grad_norm": 0.9646380257679634, |
| "learning_rate": 2.53907294458502e-06, |
| "loss": 0.1291, |
| "step": 661 |
| }, |
| { |
| "epoch": 0.5210547028728847, |
| "grad_norm": 1.0529655580058879, |
| "learning_rate": 2.5325611922429074e-06, |
| "loss": 0.1491, |
| "step": 662 |
| }, |
| { |
| "epoch": 0.5218417945690673, |
| "grad_norm": 0.9926818678117324, |
| "learning_rate": 2.5260492189451076e-06, |
| "loss": 0.1443, |
| "step": 663 |
| }, |
| { |
| "epoch": 0.52262888626525, |
| "grad_norm": 0.9144135454846201, |
| "learning_rate": 2.51953706888096e-06, |
| "loss": 0.1217, |
| "step": 664 |
| }, |
| { |
| "epoch": 0.5234159779614325, |
| "grad_norm": 0.921592430215234, |
| "learning_rate": 2.513024786241001e-06, |
| "loss": 0.1248, |
| "step": 665 |
| }, |
| { |
| "epoch": 0.5242030696576151, |
| "grad_norm": 0.9514782593826102, |
| "learning_rate": 2.5065124152166692e-06, |
| "loss": 0.1297, |
| "step": 666 |
| }, |
| { |
| "epoch": 0.5249901613537977, |
| "grad_norm": 1.0117305817250293, |
| "learning_rate": 2.5e-06, |
| "loss": 0.1497, |
| "step": 667 |
| }, |
| { |
| "epoch": 0.5257772530499804, |
| "grad_norm": 1.013668842181626, |
| "learning_rate": 2.4934875847833308e-06, |
| "loss": 0.1224, |
| "step": 668 |
| }, |
| { |
| "epoch": 0.526564344746163, |
| "grad_norm": 1.000211936689413, |
| "learning_rate": 2.4869752137589994e-06, |
| "loss": 0.1419, |
| "step": 669 |
| }, |
| { |
| "epoch": 0.5273514364423455, |
| "grad_norm": 0.9733370358487723, |
| "learning_rate": 2.48046293111904e-06, |
| "loss": 0.1245, |
| "step": 670 |
| }, |
| { |
| "epoch": 0.5281385281385281, |
| "grad_norm": 1.0646062724041805, |
| "learning_rate": 2.473950781054893e-06, |
| "loss": 0.1383, |
| "step": 671 |
| }, |
| { |
| "epoch": 0.5289256198347108, |
| "grad_norm": 0.930785726380819, |
| "learning_rate": 2.467438807757094e-06, |
| "loss": 0.1295, |
| "step": 672 |
| }, |
| { |
| "epoch": 0.5297127115308934, |
| "grad_norm": 0.9786127857256359, |
| "learning_rate": 2.460927055414981e-06, |
| "loss": 0.146, |
| "step": 673 |
| }, |
| { |
| "epoch": 0.5304998032270759, |
| "grad_norm": 0.9744140929407867, |
| "learning_rate": 2.4544155682163922e-06, |
| "loss": 0.1298, |
| "step": 674 |
| }, |
| { |
| "epoch": 0.5312868949232585, |
| "grad_norm": 0.9484782784554407, |
| "learning_rate": 2.447904390347369e-06, |
| "loss": 0.1278, |
| "step": 675 |
| }, |
| { |
| "epoch": 0.5320739866194412, |
| "grad_norm": 1.1172109264151044, |
| "learning_rate": 2.441393565991849e-06, |
| "loss": 0.146, |
| "step": 676 |
| }, |
| { |
| "epoch": 0.5328610783156238, |
| "grad_norm": 0.9832334733375834, |
| "learning_rate": 2.4348831393313763e-06, |
| "loss": 0.1341, |
| "step": 677 |
| }, |
| { |
| "epoch": 0.5336481700118064, |
| "grad_norm": 0.9993612072993626, |
| "learning_rate": 2.428373154544791e-06, |
| "loss": 0.1348, |
| "step": 678 |
| }, |
| { |
| "epoch": 0.5344352617079889, |
| "grad_norm": 0.8743067419696096, |
| "learning_rate": 2.42186365580794e-06, |
| "loss": 0.1127, |
| "step": 679 |
| }, |
| { |
| "epoch": 0.5352223534041716, |
| "grad_norm": 0.9878917132746777, |
| "learning_rate": 2.4153546872933667e-06, |
| "loss": 0.1289, |
| "step": 680 |
| }, |
| { |
| "epoch": 0.5360094451003542, |
| "grad_norm": 0.9872701224310093, |
| "learning_rate": 2.4088462931700214e-06, |
| "loss": 0.1382, |
| "step": 681 |
| }, |
| { |
| "epoch": 0.5367965367965368, |
| "grad_norm": 1.0291331541759994, |
| "learning_rate": 2.4023385176029516e-06, |
| "loss": 0.1398, |
| "step": 682 |
| }, |
| { |
| "epoch": 0.5375836284927195, |
| "grad_norm": 1.0314844938730774, |
| "learning_rate": 2.3958314047530127e-06, |
| "loss": 0.1407, |
| "step": 683 |
| }, |
| { |
| "epoch": 0.538370720188902, |
| "grad_norm": 0.9922009235690711, |
| "learning_rate": 2.3893249987765598e-06, |
| "loss": 0.1375, |
| "step": 684 |
| }, |
| { |
| "epoch": 0.5391578118850846, |
| "grad_norm": 1.0407160996339295, |
| "learning_rate": 2.3828193438251497e-06, |
| "loss": 0.1356, |
| "step": 685 |
| }, |
| { |
| "epoch": 0.5399449035812672, |
| "grad_norm": 0.9860703004700557, |
| "learning_rate": 2.376314484045248e-06, |
| "loss": 0.132, |
| "step": 686 |
| }, |
| { |
| "epoch": 0.5407319952774499, |
| "grad_norm": 1.0540933767364977, |
| "learning_rate": 2.369810463577919e-06, |
| "loss": 0.1467, |
| "step": 687 |
| }, |
| { |
| "epoch": 0.5415190869736324, |
| "grad_norm": 1.0135356185084303, |
| "learning_rate": 2.3633073265585356e-06, |
| "loss": 0.1381, |
| "step": 688 |
| }, |
| { |
| "epoch": 0.542306178669815, |
| "grad_norm": 0.9743937278639236, |
| "learning_rate": 2.3568051171164724e-06, |
| "loss": 0.1324, |
| "step": 689 |
| }, |
| { |
| "epoch": 0.5430932703659976, |
| "grad_norm": 1.0422560526589146, |
| "learning_rate": 2.350303879374813e-06, |
| "loss": 0.136, |
| "step": 690 |
| }, |
| { |
| "epoch": 0.5438803620621803, |
| "grad_norm": 1.0503391352080245, |
| "learning_rate": 2.3438036574500434e-06, |
| "loss": 0.147, |
| "step": 691 |
| }, |
| { |
| "epoch": 0.5446674537583629, |
| "grad_norm": 0.9557517793781123, |
| "learning_rate": 2.3373044954517603e-06, |
| "loss": 0.1216, |
| "step": 692 |
| }, |
| { |
| "epoch": 0.5454545454545454, |
| "grad_norm": 0.9898057468780994, |
| "learning_rate": 2.330806437482365e-06, |
| "loss": 0.1342, |
| "step": 693 |
| }, |
| { |
| "epoch": 0.546241637150728, |
| "grad_norm": 0.9685373418602369, |
| "learning_rate": 2.3243095276367687e-06, |
| "loss": 0.1294, |
| "step": 694 |
| }, |
| { |
| "epoch": 0.5470287288469107, |
| "grad_norm": 1.0187901801029866, |
| "learning_rate": 2.317813810002089e-06, |
| "loss": 0.1366, |
| "step": 695 |
| }, |
| { |
| "epoch": 0.5478158205430933, |
| "grad_norm": 1.036393473441657, |
| "learning_rate": 2.3113193286573577e-06, |
| "loss": 0.1384, |
| "step": 696 |
| }, |
| { |
| "epoch": 0.5486029122392759, |
| "grad_norm": 0.9735402694275894, |
| "learning_rate": 2.3048261276732133e-06, |
| "loss": 0.1325, |
| "step": 697 |
| }, |
| { |
| "epoch": 0.5493900039354584, |
| "grad_norm": 0.9435211562075637, |
| "learning_rate": 2.298334251111607e-06, |
| "loss": 0.1272, |
| "step": 698 |
| }, |
| { |
| "epoch": 0.5501770956316411, |
| "grad_norm": 0.9238771765346788, |
| "learning_rate": 2.2918437430255056e-06, |
| "loss": 0.1329, |
| "step": 699 |
| }, |
| { |
| "epoch": 0.5509641873278237, |
| "grad_norm": 0.9732329075427437, |
| "learning_rate": 2.285354647458585e-06, |
| "loss": 0.1316, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.5517512790240063, |
| "grad_norm": 1.032698839528823, |
| "learning_rate": 2.2788670084449393e-06, |
| "loss": 0.1438, |
| "step": 701 |
| }, |
| { |
| "epoch": 0.5525383707201889, |
| "grad_norm": 0.9453000245373157, |
| "learning_rate": 2.2723808700087764e-06, |
| "loss": 0.1349, |
| "step": 702 |
| }, |
| { |
| "epoch": 0.5533254624163715, |
| "grad_norm": 1.017447417352295, |
| "learning_rate": 2.2658962761641235e-06, |
| "loss": 0.1346, |
| "step": 703 |
| }, |
| { |
| "epoch": 0.5541125541125541, |
| "grad_norm": 1.0593240948345142, |
| "learning_rate": 2.2594132709145245e-06, |
| "loss": 0.1391, |
| "step": 704 |
| }, |
| { |
| "epoch": 0.5548996458087367, |
| "grad_norm": 1.0524825811903469, |
| "learning_rate": 2.2529318982527453e-06, |
| "loss": 0.1397, |
| "step": 705 |
| }, |
| { |
| "epoch": 0.5556867375049194, |
| "grad_norm": 0.9994684504324172, |
| "learning_rate": 2.246452202160471e-06, |
| "loss": 0.1416, |
| "step": 706 |
| }, |
| { |
| "epoch": 0.5564738292011019, |
| "grad_norm": 1.033406524556106, |
| "learning_rate": 2.2399742266080126e-06, |
| "loss": 0.1269, |
| "step": 707 |
| }, |
| { |
| "epoch": 0.5572609208972845, |
| "grad_norm": 0.9246218098662428, |
| "learning_rate": 2.233498015554002e-06, |
| "loss": 0.1242, |
| "step": 708 |
| }, |
| { |
| "epoch": 0.5580480125934671, |
| "grad_norm": 0.904021008692359, |
| "learning_rate": 2.227023612945102e-06, |
| "loss": 0.1217, |
| "step": 709 |
| }, |
| { |
| "epoch": 0.5588351042896498, |
| "grad_norm": 0.9921447266760961, |
| "learning_rate": 2.220551062715699e-06, |
| "loss": 0.1241, |
| "step": 710 |
| }, |
| { |
| "epoch": 0.5596221959858324, |
| "grad_norm": 1.0273952935358304, |
| "learning_rate": 2.2140804087876134e-06, |
| "loss": 0.1244, |
| "step": 711 |
| }, |
| { |
| "epoch": 0.5604092876820149, |
| "grad_norm": 0.9958421204937957, |
| "learning_rate": 2.207611695069794e-06, |
| "loss": 0.1277, |
| "step": 712 |
| }, |
| { |
| "epoch": 0.5611963793781976, |
| "grad_norm": 1.0226304738126037, |
| "learning_rate": 2.2011449654580266e-06, |
| "loss": 0.1319, |
| "step": 713 |
| }, |
| { |
| "epoch": 0.5619834710743802, |
| "grad_norm": 0.9427241568832295, |
| "learning_rate": 2.1946802638346324e-06, |
| "loss": 0.1208, |
| "step": 714 |
| }, |
| { |
| "epoch": 0.5627705627705628, |
| "grad_norm": 0.9526667511261941, |
| "learning_rate": 2.1882176340681682e-06, |
| "loss": 0.1234, |
| "step": 715 |
| }, |
| { |
| "epoch": 0.5635576544667453, |
| "grad_norm": 0.9726636294262463, |
| "learning_rate": 2.181757120013136e-06, |
| "loss": 0.1241, |
| "step": 716 |
| }, |
| { |
| "epoch": 0.564344746162928, |
| "grad_norm": 0.9577642489452165, |
| "learning_rate": 2.1752987655096765e-06, |
| "loss": 0.1286, |
| "step": 717 |
| }, |
| { |
| "epoch": 0.5651318378591106, |
| "grad_norm": 0.9119267395234483, |
| "learning_rate": 2.1688426143832804e-06, |
| "loss": 0.132, |
| "step": 718 |
| }, |
| { |
| "epoch": 0.5659189295552932, |
| "grad_norm": 0.944139995902989, |
| "learning_rate": 2.162388710444482e-06, |
| "loss": 0.1234, |
| "step": 719 |
| }, |
| { |
| "epoch": 0.5667060212514758, |
| "grad_norm": 0.9645692490749199, |
| "learning_rate": 2.155937097488571e-06, |
| "loss": 0.1251, |
| "step": 720 |
| }, |
| { |
| "epoch": 0.5674931129476584, |
| "grad_norm": 0.9720255939912888, |
| "learning_rate": 2.1494878192952857e-06, |
| "loss": 0.1319, |
| "step": 721 |
| }, |
| { |
| "epoch": 0.568280204643841, |
| "grad_norm": 0.9511775624645177, |
| "learning_rate": 2.1430409196285268e-06, |
| "loss": 0.1327, |
| "step": 722 |
| }, |
| { |
| "epoch": 0.5690672963400236, |
| "grad_norm": 0.9008868958605895, |
| "learning_rate": 2.1365964422360495e-06, |
| "loss": 0.1257, |
| "step": 723 |
| }, |
| { |
| "epoch": 0.5698543880362062, |
| "grad_norm": 1.0087528087899673, |
| "learning_rate": 2.1301544308491755e-06, |
| "loss": 0.1404, |
| "step": 724 |
| }, |
| { |
| "epoch": 0.5706414797323889, |
| "grad_norm": 0.9236247008656706, |
| "learning_rate": 2.1237149291824906e-06, |
| "loss": 0.1122, |
| "step": 725 |
| }, |
| { |
| "epoch": 0.5714285714285714, |
| "grad_norm": 0.9656011748968637, |
| "learning_rate": 2.11727798093355e-06, |
| "loss": 0.1238, |
| "step": 726 |
| }, |
| { |
| "epoch": 0.572215663124754, |
| "grad_norm": 0.9574761410065884, |
| "learning_rate": 2.110843629782583e-06, |
| "loss": 0.1205, |
| "step": 727 |
| }, |
| { |
| "epoch": 0.5730027548209367, |
| "grad_norm": 1.0073901156504852, |
| "learning_rate": 2.1044119193921935e-06, |
| "loss": 0.141, |
| "step": 728 |
| }, |
| { |
| "epoch": 0.5737898465171193, |
| "grad_norm": 1.0135184994615516, |
| "learning_rate": 2.097982893407068e-06, |
| "loss": 0.1391, |
| "step": 729 |
| }, |
| { |
| "epoch": 0.5745769382133018, |
| "grad_norm": 0.9943855979768463, |
| "learning_rate": 2.0915565954536745e-06, |
| "loss": 0.1261, |
| "step": 730 |
| }, |
| { |
| "epoch": 0.5753640299094844, |
| "grad_norm": 0.9263403288426786, |
| "learning_rate": 2.085133069139971e-06, |
| "loss": 0.1199, |
| "step": 731 |
| }, |
| { |
| "epoch": 0.5761511216056671, |
| "grad_norm": 0.9250174978891127, |
| "learning_rate": 2.078712358055106e-06, |
| "loss": 0.1292, |
| "step": 732 |
| }, |
| { |
| "epoch": 0.5769382133018497, |
| "grad_norm": 0.9594966083023022, |
| "learning_rate": 2.0722945057691253e-06, |
| "loss": 0.13, |
| "step": 733 |
| }, |
| { |
| "epoch": 0.5777253049980323, |
| "grad_norm": 0.9996158846425939, |
| "learning_rate": 2.0658795558326745e-06, |
| "loss": 0.1346, |
| "step": 734 |
| }, |
| { |
| "epoch": 0.5785123966942148, |
| "grad_norm": 0.9512575411801811, |
| "learning_rate": 2.059467551776705e-06, |
| "loss": 0.1277, |
| "step": 735 |
| }, |
| { |
| "epoch": 0.5792994883903975, |
| "grad_norm": 0.9956048932258434, |
| "learning_rate": 2.053058537112177e-06, |
| "loss": 0.1364, |
| "step": 736 |
| }, |
| { |
| "epoch": 0.5800865800865801, |
| "grad_norm": 0.9822563309180476, |
| "learning_rate": 2.0466525553297666e-06, |
| "loss": 0.124, |
| "step": 737 |
| }, |
| { |
| "epoch": 0.5808736717827627, |
| "grad_norm": 0.9983781840041562, |
| "learning_rate": 2.0402496498995667e-06, |
| "loss": 0.1347, |
| "step": 738 |
| }, |
| { |
| "epoch": 0.5816607634789452, |
| "grad_norm": 0.9827604253780587, |
| "learning_rate": 2.0338498642707977e-06, |
| "loss": 0.1369, |
| "step": 739 |
| }, |
| { |
| "epoch": 0.5824478551751279, |
| "grad_norm": 1.0756801683767687, |
| "learning_rate": 2.027453241871506e-06, |
| "loss": 0.1323, |
| "step": 740 |
| }, |
| { |
| "epoch": 0.5832349468713105, |
| "grad_norm": 1.0081971409472221, |
| "learning_rate": 2.0210598261082764e-06, |
| "loss": 0.1356, |
| "step": 741 |
| }, |
| { |
| "epoch": 0.5840220385674931, |
| "grad_norm": 0.9898906943423369, |
| "learning_rate": 2.014669660365931e-06, |
| "loss": 0.1368, |
| "step": 742 |
| }, |
| { |
| "epoch": 0.5848091302636758, |
| "grad_norm": 0.9690524566063999, |
| "learning_rate": 2.0082827880072393e-06, |
| "loss": 0.135, |
| "step": 743 |
| }, |
| { |
| "epoch": 0.5855962219598583, |
| "grad_norm": 0.9708412001010785, |
| "learning_rate": 2.0018992523726217e-06, |
| "loss": 0.1252, |
| "step": 744 |
| }, |
| { |
| "epoch": 0.5863833136560409, |
| "grad_norm": 0.9322317029959182, |
| "learning_rate": 1.995519096779855e-06, |
| "loss": 0.1205, |
| "step": 745 |
| }, |
| { |
| "epoch": 0.5871704053522235, |
| "grad_norm": 0.934365579766912, |
| "learning_rate": 1.9891423645237832e-06, |
| "loss": 0.1194, |
| "step": 746 |
| }, |
| { |
| "epoch": 0.5879574970484062, |
| "grad_norm": 0.9421279165900748, |
| "learning_rate": 1.982769098876015e-06, |
| "loss": 0.1319, |
| "step": 747 |
| }, |
| { |
| "epoch": 0.5887445887445888, |
| "grad_norm": 0.9954570155310445, |
| "learning_rate": 1.9763993430846394e-06, |
| "loss": 0.1369, |
| "step": 748 |
| }, |
| { |
| "epoch": 0.5895316804407713, |
| "grad_norm": 0.931791564112829, |
| "learning_rate": 1.970033140373925e-06, |
| "loss": 0.1315, |
| "step": 749 |
| }, |
| { |
| "epoch": 0.5903187721369539, |
| "grad_norm": 0.9679723780616554, |
| "learning_rate": 1.9636705339440327e-06, |
| "loss": 0.1377, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.5911058638331366, |
| "grad_norm": 0.9863750681505877, |
| "learning_rate": 1.957311566970716e-06, |
| "loss": 0.1293, |
| "step": 751 |
| }, |
| { |
| "epoch": 0.5918929555293192, |
| "grad_norm": 0.9598202963903522, |
| "learning_rate": 1.9509562826050353e-06, |
| "loss": 0.1273, |
| "step": 752 |
| }, |
| { |
| "epoch": 0.5926800472255017, |
| "grad_norm": 0.9800756257622318, |
| "learning_rate": 1.944604723973058e-06, |
| "loss": 0.1284, |
| "step": 753 |
| }, |
| { |
| "epoch": 0.5934671389216843, |
| "grad_norm": 1.0001757874575956, |
| "learning_rate": 1.938256934175571e-06, |
| "loss": 0.1303, |
| "step": 754 |
| }, |
| { |
| "epoch": 0.594254230617867, |
| "grad_norm": 0.9299561635899479, |
| "learning_rate": 1.9319129562877863e-06, |
| "loss": 0.1239, |
| "step": 755 |
| }, |
| { |
| "epoch": 0.5950413223140496, |
| "grad_norm": 1.0345095738407815, |
| "learning_rate": 1.925572833359048e-06, |
| "loss": 0.1305, |
| "step": 756 |
| }, |
| { |
| "epoch": 0.5958284140102322, |
| "grad_norm": 1.0520535233317054, |
| "learning_rate": 1.9192366084125423e-06, |
| "loss": 0.1373, |
| "step": 757 |
| }, |
| { |
| "epoch": 0.5966155057064148, |
| "grad_norm": 1.029018918955376, |
| "learning_rate": 1.9129043244450027e-06, |
| "loss": 0.1382, |
| "step": 758 |
| }, |
| { |
| "epoch": 0.5974025974025974, |
| "grad_norm": 1.0294584813791954, |
| "learning_rate": 1.906576024426422e-06, |
| "loss": 0.1368, |
| "step": 759 |
| }, |
| { |
| "epoch": 0.59818968909878, |
| "grad_norm": 0.9330122675132353, |
| "learning_rate": 1.9002517512997555e-06, |
| "loss": 0.1145, |
| "step": 760 |
| }, |
| { |
| "epoch": 0.5989767807949626, |
| "grad_norm": 0.9623676868988281, |
| "learning_rate": 1.8939315479806352e-06, |
| "loss": 0.1335, |
| "step": 761 |
| }, |
| { |
| "epoch": 0.5997638724911453, |
| "grad_norm": 0.9245436138689049, |
| "learning_rate": 1.8876154573570744e-06, |
| "loss": 0.1307, |
| "step": 762 |
| }, |
| { |
| "epoch": 0.6005509641873278, |
| "grad_norm": 0.8942520983106202, |
| "learning_rate": 1.8813035222891785e-06, |
| "loss": 0.1272, |
| "step": 763 |
| }, |
| { |
| "epoch": 0.6013380558835104, |
| "grad_norm": 0.9343347217079488, |
| "learning_rate": 1.8749957856088546e-06, |
| "loss": 0.1317, |
| "step": 764 |
| }, |
| { |
| "epoch": 0.602125147579693, |
| "grad_norm": 0.9393324326245188, |
| "learning_rate": 1.8686922901195197e-06, |
| "loss": 0.1313, |
| "step": 765 |
| }, |
| { |
| "epoch": 0.6029122392758757, |
| "grad_norm": 0.9333264421793994, |
| "learning_rate": 1.8623930785958092e-06, |
| "loss": 0.1226, |
| "step": 766 |
| }, |
| { |
| "epoch": 0.6036993309720582, |
| "grad_norm": 0.9718728327996774, |
| "learning_rate": 1.8560981937832916e-06, |
| "loss": 0.1314, |
| "step": 767 |
| }, |
| { |
| "epoch": 0.6044864226682408, |
| "grad_norm": 0.9437466891844623, |
| "learning_rate": 1.849807678398171e-06, |
| "loss": 0.1271, |
| "step": 768 |
| }, |
| { |
| "epoch": 0.6052735143644234, |
| "grad_norm": 0.9433172532376, |
| "learning_rate": 1.8435215751270048e-06, |
| "loss": 0.1083, |
| "step": 769 |
| }, |
| { |
| "epoch": 0.6060606060606061, |
| "grad_norm": 0.9486485907428178, |
| "learning_rate": 1.8372399266264069e-06, |
| "loss": 0.1245, |
| "step": 770 |
| }, |
| { |
| "epoch": 0.6068476977567887, |
| "grad_norm": 0.9345585045873044, |
| "learning_rate": 1.8309627755227643e-06, |
| "loss": 0.1205, |
| "step": 771 |
| }, |
| { |
| "epoch": 0.6076347894529712, |
| "grad_norm": 1.0082946745736912, |
| "learning_rate": 1.8246901644119447e-06, |
| "loss": 0.1337, |
| "step": 772 |
| }, |
| { |
| "epoch": 0.6084218811491539, |
| "grad_norm": 0.9640602588467792, |
| "learning_rate": 1.8184221358590078e-06, |
| "loss": 0.123, |
| "step": 773 |
| }, |
| { |
| "epoch": 0.6092089728453365, |
| "grad_norm": 1.0256774883323883, |
| "learning_rate": 1.812158732397917e-06, |
| "loss": 0.1331, |
| "step": 774 |
| }, |
| { |
| "epoch": 0.6099960645415191, |
| "grad_norm": 0.9485492161002549, |
| "learning_rate": 1.8058999965312484e-06, |
| "loss": 0.1328, |
| "step": 775 |
| }, |
| { |
| "epoch": 0.6107831562377017, |
| "grad_norm": 0.9763406590147844, |
| "learning_rate": 1.799645970729909e-06, |
| "loss": 0.1309, |
| "step": 776 |
| }, |
| { |
| "epoch": 0.6115702479338843, |
| "grad_norm": 0.9917227985654803, |
| "learning_rate": 1.793396697432839e-06, |
| "loss": 0.1349, |
| "step": 777 |
| }, |
| { |
| "epoch": 0.6123573396300669, |
| "grad_norm": 0.9926597353156553, |
| "learning_rate": 1.7871522190467327e-06, |
| "loss": 0.1303, |
| "step": 778 |
| }, |
| { |
| "epoch": 0.6131444313262495, |
| "grad_norm": 0.9259479446299848, |
| "learning_rate": 1.7809125779457432e-06, |
| "loss": 0.1145, |
| "step": 779 |
| }, |
| { |
| "epoch": 0.6139315230224321, |
| "grad_norm": 0.9291022839595524, |
| "learning_rate": 1.7746778164712024e-06, |
| "loss": 0.119, |
| "step": 780 |
| }, |
| { |
| "epoch": 0.6147186147186147, |
| "grad_norm": 0.951539094911597, |
| "learning_rate": 1.768447976931326e-06, |
| "loss": 0.1261, |
| "step": 781 |
| }, |
| { |
| "epoch": 0.6155057064147973, |
| "grad_norm": 0.993556323836548, |
| "learning_rate": 1.7622231016009333e-06, |
| "loss": 0.1297, |
| "step": 782 |
| }, |
| { |
| "epoch": 0.6162927981109799, |
| "grad_norm": 0.9391833540663885, |
| "learning_rate": 1.7560032327211546e-06, |
| "loss": 0.124, |
| "step": 783 |
| }, |
| { |
| "epoch": 0.6170798898071626, |
| "grad_norm": 1.031878772377542, |
| "learning_rate": 1.7497884124991487e-06, |
| "loss": 0.1308, |
| "step": 784 |
| }, |
| { |
| "epoch": 0.6178669815033452, |
| "grad_norm": 0.9834671756142636, |
| "learning_rate": 1.7435786831078144e-06, |
| "loss": 0.1303, |
| "step": 785 |
| }, |
| { |
| "epoch": 0.6186540731995277, |
| "grad_norm": 0.9859388240495401, |
| "learning_rate": 1.7373740866855043e-06, |
| "loss": 0.1326, |
| "step": 786 |
| }, |
| { |
| "epoch": 0.6194411648957103, |
| "grad_norm": 1.0156315373671152, |
| "learning_rate": 1.731174665335742e-06, |
| "loss": 0.1333, |
| "step": 787 |
| }, |
| { |
| "epoch": 0.620228256591893, |
| "grad_norm": 0.8457875340285443, |
| "learning_rate": 1.724980461126929e-06, |
| "loss": 0.1149, |
| "step": 788 |
| }, |
| { |
| "epoch": 0.6210153482880756, |
| "grad_norm": 0.9812167735229308, |
| "learning_rate": 1.7187915160920692e-06, |
| "loss": 0.1341, |
| "step": 789 |
| }, |
| { |
| "epoch": 0.6218024399842582, |
| "grad_norm": 0.9479256338770862, |
| "learning_rate": 1.7126078722284739e-06, |
| "loss": 0.1171, |
| "step": 790 |
| }, |
| { |
| "epoch": 0.6225895316804407, |
| "grad_norm": 0.9626147311559159, |
| "learning_rate": 1.706429571497486e-06, |
| "loss": 0.1195, |
| "step": 791 |
| }, |
| { |
| "epoch": 0.6233766233766234, |
| "grad_norm": 0.996537388600602, |
| "learning_rate": 1.7002566558241862e-06, |
| "loss": 0.1347, |
| "step": 792 |
| }, |
| { |
| "epoch": 0.624163715072806, |
| "grad_norm": 1.0979789094333103, |
| "learning_rate": 1.694089167097116e-06, |
| "loss": 0.1442, |
| "step": 793 |
| }, |
| { |
| "epoch": 0.6249508067689886, |
| "grad_norm": 0.9903394042224888, |
| "learning_rate": 1.6879271471679887e-06, |
| "loss": 0.1275, |
| "step": 794 |
| }, |
| { |
| "epoch": 0.6257378984651711, |
| "grad_norm": 0.9904940377814807, |
| "learning_rate": 1.681770637851409e-06, |
| "loss": 0.139, |
| "step": 795 |
| }, |
| { |
| "epoch": 0.6265249901613538, |
| "grad_norm": 0.9969630818236452, |
| "learning_rate": 1.675619680924584e-06, |
| "loss": 0.1325, |
| "step": 796 |
| }, |
| { |
| "epoch": 0.6273120818575364, |
| "grad_norm": 1.0558109930918702, |
| "learning_rate": 1.6694743181270474e-06, |
| "loss": 0.1448, |
| "step": 797 |
| }, |
| { |
| "epoch": 0.628099173553719, |
| "grad_norm": 0.956496508451797, |
| "learning_rate": 1.663334591160368e-06, |
| "loss": 0.1217, |
| "step": 798 |
| }, |
| { |
| "epoch": 0.6288862652499017, |
| "grad_norm": 0.9677341961932617, |
| "learning_rate": 1.657200541687874e-06, |
| "loss": 0.136, |
| "step": 799 |
| }, |
| { |
| "epoch": 0.6296733569460842, |
| "grad_norm": 1.0046971327809577, |
| "learning_rate": 1.6510722113343633e-06, |
| "loss": 0.1322, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.6304604486422668, |
| "grad_norm": 0.9254874616921521, |
| "learning_rate": 1.6449496416858285e-06, |
| "loss": 0.1227, |
| "step": 801 |
| }, |
| { |
| "epoch": 0.6312475403384494, |
| "grad_norm": 1.128513971443689, |
| "learning_rate": 1.6388328742891679e-06, |
| "loss": 0.1357, |
| "step": 802 |
| }, |
| { |
| "epoch": 0.6320346320346321, |
| "grad_norm": 1.0186779597478501, |
| "learning_rate": 1.6327219506519082e-06, |
| "loss": 0.1369, |
| "step": 803 |
| }, |
| { |
| "epoch": 0.6328217237308147, |
| "grad_norm": 0.9605839825909683, |
| "learning_rate": 1.6266169122419208e-06, |
| "loss": 0.1222, |
| "step": 804 |
| }, |
| { |
| "epoch": 0.6336088154269972, |
| "grad_norm": 1.0294396317293524, |
| "learning_rate": 1.6205178004871392e-06, |
| "loss": 0.1265, |
| "step": 805 |
| }, |
| { |
| "epoch": 0.6343959071231798, |
| "grad_norm": 0.9797067002404048, |
| "learning_rate": 1.6144246567752831e-06, |
| "loss": 0.1298, |
| "step": 806 |
| }, |
| { |
| "epoch": 0.6351829988193625, |
| "grad_norm": 0.9531255383226177, |
| "learning_rate": 1.6083375224535689e-06, |
| "loss": 0.1204, |
| "step": 807 |
| }, |
| { |
| "epoch": 0.6359700905155451, |
| "grad_norm": 0.9228494475342526, |
| "learning_rate": 1.6022564388284391e-06, |
| "loss": 0.1122, |
| "step": 808 |
| }, |
| { |
| "epoch": 0.6367571822117276, |
| "grad_norm": 0.9854787445128979, |
| "learning_rate": 1.596181447165273e-06, |
| "loss": 0.1287, |
| "step": 809 |
| }, |
| { |
| "epoch": 0.6375442739079102, |
| "grad_norm": 0.9205768495534565, |
| "learning_rate": 1.5901125886881147e-06, |
| "loss": 0.1206, |
| "step": 810 |
| }, |
| { |
| "epoch": 0.6383313656040929, |
| "grad_norm": 0.9798975631304712, |
| "learning_rate": 1.5840499045793845e-06, |
| "loss": 0.1231, |
| "step": 811 |
| }, |
| { |
| "epoch": 0.6391184573002755, |
| "grad_norm": 0.9296415144186752, |
| "learning_rate": 1.5779934359796095e-06, |
| "loss": 0.1202, |
| "step": 812 |
| }, |
| { |
| "epoch": 0.6399055489964581, |
| "grad_norm": 0.9814672778856722, |
| "learning_rate": 1.5719432239871347e-06, |
| "loss": 0.1211, |
| "step": 813 |
| }, |
| { |
| "epoch": 0.6406926406926406, |
| "grad_norm": 0.9650667010737961, |
| "learning_rate": 1.5658993096578512e-06, |
| "loss": 0.123, |
| "step": 814 |
| }, |
| { |
| "epoch": 0.6414797323888233, |
| "grad_norm": 0.9606428490346777, |
| "learning_rate": 1.5598617340049145e-06, |
| "loss": 0.1196, |
| "step": 815 |
| }, |
| { |
| "epoch": 0.6422668240850059, |
| "grad_norm": 0.9865842262641049, |
| "learning_rate": 1.5538305379984661e-06, |
| "loss": 0.1414, |
| "step": 816 |
| }, |
| { |
| "epoch": 0.6430539157811885, |
| "grad_norm": 0.9734870716677574, |
| "learning_rate": 1.547805762565358e-06, |
| "loss": 0.1286, |
| "step": 817 |
| }, |
| { |
| "epoch": 0.6438410074773712, |
| "grad_norm": 0.9439567875437019, |
| "learning_rate": 1.5417874485888706e-06, |
| "loss": 0.1109, |
| "step": 818 |
| }, |
| { |
| "epoch": 0.6446280991735537, |
| "grad_norm": 1.0235998100882107, |
| "learning_rate": 1.5357756369084398e-06, |
| "loss": 0.123, |
| "step": 819 |
| }, |
| { |
| "epoch": 0.6454151908697363, |
| "grad_norm": 1.0269524388931728, |
| "learning_rate": 1.5297703683193755e-06, |
| "loss": 0.1324, |
| "step": 820 |
| }, |
| { |
| "epoch": 0.6462022825659189, |
| "grad_norm": 0.9493642789135233, |
| "learning_rate": 1.5237716835725907e-06, |
| "loss": 0.1125, |
| "step": 821 |
| }, |
| { |
| "epoch": 0.6469893742621016, |
| "grad_norm": 1.162601877497598, |
| "learning_rate": 1.5177796233743174e-06, |
| "loss": 0.1249, |
| "step": 822 |
| }, |
| { |
| "epoch": 0.6477764659582841, |
| "grad_norm": 1.034038388079516, |
| "learning_rate": 1.511794228385837e-06, |
| "loss": 0.1217, |
| "step": 823 |
| }, |
| { |
| "epoch": 0.6485635576544667, |
| "grad_norm": 0.9672639615152381, |
| "learning_rate": 1.5058155392232004e-06, |
| "loss": 0.1208, |
| "step": 824 |
| }, |
| { |
| "epoch": 0.6493506493506493, |
| "grad_norm": 1.080256521732267, |
| "learning_rate": 1.4998435964569552e-06, |
| "loss": 0.1279, |
| "step": 825 |
| }, |
| { |
| "epoch": 0.650137741046832, |
| "grad_norm": 0.9417495839242918, |
| "learning_rate": 1.4938784406118663e-06, |
| "loss": 0.1249, |
| "step": 826 |
| }, |
| { |
| "epoch": 0.6509248327430146, |
| "grad_norm": 1.006350286001005, |
| "learning_rate": 1.4879201121666466e-06, |
| "loss": 0.1251, |
| "step": 827 |
| }, |
| { |
| "epoch": 0.6517119244391971, |
| "grad_norm": 0.968507626389286, |
| "learning_rate": 1.4819686515536763e-06, |
| "loss": 0.1203, |
| "step": 828 |
| }, |
| { |
| "epoch": 0.6524990161353798, |
| "grad_norm": 0.979256644659201, |
| "learning_rate": 1.4760240991587338e-06, |
| "loss": 0.1309, |
| "step": 829 |
| }, |
| { |
| "epoch": 0.6532861078315624, |
| "grad_norm": 1.041802414674734, |
| "learning_rate": 1.4700864953207192e-06, |
| "loss": 0.124, |
| "step": 830 |
| }, |
| { |
| "epoch": 0.654073199527745, |
| "grad_norm": 0.9623673717149763, |
| "learning_rate": 1.4641558803313783e-06, |
| "loss": 0.1153, |
| "step": 831 |
| }, |
| { |
| "epoch": 0.6548602912239276, |
| "grad_norm": 1.0049463554640272, |
| "learning_rate": 1.4582322944350335e-06, |
| "loss": 0.123, |
| "step": 832 |
| }, |
| { |
| "epoch": 0.6556473829201102, |
| "grad_norm": 0.9822560730942449, |
| "learning_rate": 1.4523157778283082e-06, |
| "loss": 0.1253, |
| "step": 833 |
| }, |
| { |
| "epoch": 0.6564344746162928, |
| "grad_norm": 1.0300014906979744, |
| "learning_rate": 1.4464063706598563e-06, |
| "loss": 0.121, |
| "step": 834 |
| }, |
| { |
| "epoch": 0.6572215663124754, |
| "grad_norm": 0.9605069437184749, |
| "learning_rate": 1.440504113030084e-06, |
| "loss": 0.1303, |
| "step": 835 |
| }, |
| { |
| "epoch": 0.658008658008658, |
| "grad_norm": 1.0062748427154549, |
| "learning_rate": 1.4346090449908862e-06, |
| "loss": 0.1254, |
| "step": 836 |
| }, |
| { |
| "epoch": 0.6587957497048406, |
| "grad_norm": 1.003505120930448, |
| "learning_rate": 1.4287212065453681e-06, |
| "loss": 0.1293, |
| "step": 837 |
| }, |
| { |
| "epoch": 0.6595828414010232, |
| "grad_norm": 0.9215109848797975, |
| "learning_rate": 1.4228406376475741e-06, |
| "loss": 0.1156, |
| "step": 838 |
| }, |
| { |
| "epoch": 0.6603699330972058, |
| "grad_norm": 1.0375359512611602, |
| "learning_rate": 1.4169673782022232e-06, |
| "loss": 0.1251, |
| "step": 839 |
| }, |
| { |
| "epoch": 0.6611570247933884, |
| "grad_norm": 1.0075633482471045, |
| "learning_rate": 1.411101468064429e-06, |
| "loss": 0.1273, |
| "step": 840 |
| }, |
| { |
| "epoch": 0.6619441164895711, |
| "grad_norm": 1.0079245494150497, |
| "learning_rate": 1.4052429470394353e-06, |
| "loss": 0.1302, |
| "step": 841 |
| }, |
| { |
| "epoch": 0.6627312081857536, |
| "grad_norm": 0.9589739631373009, |
| "learning_rate": 1.3993918548823453e-06, |
| "loss": 0.1219, |
| "step": 842 |
| }, |
| { |
| "epoch": 0.6635182998819362, |
| "grad_norm": 0.9854619269672102, |
| "learning_rate": 1.3935482312978494e-06, |
| "loss": 0.1264, |
| "step": 843 |
| }, |
| { |
| "epoch": 0.6643053915781189, |
| "grad_norm": 1.0139593156707545, |
| "learning_rate": 1.3877121159399587e-06, |
| "loss": 0.1352, |
| "step": 844 |
| }, |
| { |
| "epoch": 0.6650924832743015, |
| "grad_norm": 0.9879913850797528, |
| "learning_rate": 1.381883548411735e-06, |
| "loss": 0.1252, |
| "step": 845 |
| }, |
| { |
| "epoch": 0.6658795749704841, |
| "grad_norm": 0.9828821822604814, |
| "learning_rate": 1.376062568265018e-06, |
| "loss": 0.1262, |
| "step": 846 |
| }, |
| { |
| "epoch": 0.6666666666666666, |
| "grad_norm": 0.9902383754663022, |
| "learning_rate": 1.370249215000166e-06, |
| "loss": 0.1339, |
| "step": 847 |
| }, |
| { |
| "epoch": 0.6674537583628493, |
| "grad_norm": 1.0169925787410046, |
| "learning_rate": 1.3644435280657765e-06, |
| "loss": 0.1325, |
| "step": 848 |
| }, |
| { |
| "epoch": 0.6682408500590319, |
| "grad_norm": 0.9802382914836032, |
| "learning_rate": 1.3586455468584292e-06, |
| "loss": 0.1294, |
| "step": 849 |
| }, |
| { |
| "epoch": 0.6690279417552145, |
| "grad_norm": 0.9103087080426163, |
| "learning_rate": 1.3528553107224108e-06, |
| "loss": 0.1132, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.669815033451397, |
| "grad_norm": 1.0322697690605673, |
| "learning_rate": 1.347072858949453e-06, |
| "loss": 0.1326, |
| "step": 851 |
| }, |
| { |
| "epoch": 0.6706021251475797, |
| "grad_norm": 0.940497609406273, |
| "learning_rate": 1.3412982307784617e-06, |
| "loss": 0.1142, |
| "step": 852 |
| }, |
| { |
| "epoch": 0.6713892168437623, |
| "grad_norm": 0.9651333506256994, |
| "learning_rate": 1.3355314653952555e-06, |
| "loss": 0.12, |
| "step": 853 |
| }, |
| { |
| "epoch": 0.6721763085399449, |
| "grad_norm": 0.8974492403550183, |
| "learning_rate": 1.3297726019322948e-06, |
| "loss": 0.1252, |
| "step": 854 |
| }, |
| { |
| "epoch": 0.6729634002361276, |
| "grad_norm": 0.9779192150286001, |
| "learning_rate": 1.3240216794684212e-06, |
| "loss": 0.1265, |
| "step": 855 |
| }, |
| { |
| "epoch": 0.6737504919323101, |
| "grad_norm": 1.0060169889058102, |
| "learning_rate": 1.3182787370285865e-06, |
| "loss": 0.1305, |
| "step": 856 |
| }, |
| { |
| "epoch": 0.6745375836284927, |
| "grad_norm": 0.9623311050243877, |
| "learning_rate": 1.3125438135835955e-06, |
| "loss": 0.114, |
| "step": 857 |
| }, |
| { |
| "epoch": 0.6753246753246753, |
| "grad_norm": 1.005880860747008, |
| "learning_rate": 1.3068169480498333e-06, |
| "loss": 0.1237, |
| "step": 858 |
| }, |
| { |
| "epoch": 0.676111767020858, |
| "grad_norm": 1.0295442665880505, |
| "learning_rate": 1.3010981792890053e-06, |
| "loss": 0.141, |
| "step": 859 |
| }, |
| { |
| "epoch": 0.6768988587170406, |
| "grad_norm": 0.9746775819035803, |
| "learning_rate": 1.2953875461078777e-06, |
| "loss": 0.1174, |
| "step": 860 |
| }, |
| { |
| "epoch": 0.6776859504132231, |
| "grad_norm": 0.9651023742880912, |
| "learning_rate": 1.289685087258004e-06, |
| "loss": 0.1179, |
| "step": 861 |
| }, |
| { |
| "epoch": 0.6784730421094057, |
| "grad_norm": 0.9778504990448126, |
| "learning_rate": 1.283990841435473e-06, |
| "loss": 0.1232, |
| "step": 862 |
| }, |
| { |
| "epoch": 0.6792601338055884, |
| "grad_norm": 0.9823411560425596, |
| "learning_rate": 1.2783048472806364e-06, |
| "loss": 0.1214, |
| "step": 863 |
| }, |
| { |
| "epoch": 0.680047225501771, |
| "grad_norm": 0.9509119170509043, |
| "learning_rate": 1.2726271433778559e-06, |
| "loss": 0.1331, |
| "step": 864 |
| }, |
| { |
| "epoch": 0.6808343171979535, |
| "grad_norm": 0.9637465369074552, |
| "learning_rate": 1.266957768255232e-06, |
| "loss": 0.1221, |
| "step": 865 |
| }, |
| { |
| "epoch": 0.6816214088941361, |
| "grad_norm": 1.0309739334485784, |
| "learning_rate": 1.2612967603843512e-06, |
| "loss": 0.1337, |
| "step": 866 |
| }, |
| { |
| "epoch": 0.6824085005903188, |
| "grad_norm": 0.9227141127754309, |
| "learning_rate": 1.2556441581800182e-06, |
| "loss": 0.1118, |
| "step": 867 |
| }, |
| { |
| "epoch": 0.6831955922865014, |
| "grad_norm": 0.983027599423059, |
| "learning_rate": 1.2500000000000007e-06, |
| "loss": 0.1201, |
| "step": 868 |
| }, |
| { |
| "epoch": 0.683982683982684, |
| "grad_norm": 0.969869074022873, |
| "learning_rate": 1.2443643241447629e-06, |
| "loss": 0.1205, |
| "step": 869 |
| }, |
| { |
| "epoch": 0.6847697756788665, |
| "grad_norm": 0.9626068462653994, |
| "learning_rate": 1.2387371688572133e-06, |
| "loss": 0.1294, |
| "step": 870 |
| }, |
| { |
| "epoch": 0.6855568673750492, |
| "grad_norm": 0.9924688128052054, |
| "learning_rate": 1.233118572322437e-06, |
| "loss": 0.1193, |
| "step": 871 |
| }, |
| { |
| "epoch": 0.6863439590712318, |
| "grad_norm": 0.9409212105627156, |
| "learning_rate": 1.2275085726674442e-06, |
| "loss": 0.1186, |
| "step": 872 |
| }, |
| { |
| "epoch": 0.6871310507674144, |
| "grad_norm": 0.9321864217317675, |
| "learning_rate": 1.2219072079609046e-06, |
| "loss": 0.118, |
| "step": 873 |
| }, |
| { |
| "epoch": 0.6879181424635971, |
| "grad_norm": 0.8802354237634122, |
| "learning_rate": 1.2163145162128948e-06, |
| "loss": 0.1092, |
| "step": 874 |
| }, |
| { |
| "epoch": 0.6887052341597796, |
| "grad_norm": 0.9820858832906886, |
| "learning_rate": 1.2107305353746376e-06, |
| "loss": 0.1261, |
| "step": 875 |
| }, |
| { |
| "epoch": 0.6894923258559622, |
| "grad_norm": 1.0214787998802317, |
| "learning_rate": 1.2051553033382426e-06, |
| "loss": 0.121, |
| "step": 876 |
| }, |
| { |
| "epoch": 0.6902794175521448, |
| "grad_norm": 0.9157258726824631, |
| "learning_rate": 1.1995888579364551e-06, |
| "loss": 0.1189, |
| "step": 877 |
| }, |
| { |
| "epoch": 0.6910665092483275, |
| "grad_norm": 0.9531462191249618, |
| "learning_rate": 1.1940312369423919e-06, |
| "loss": 0.1184, |
| "step": 878 |
| }, |
| { |
| "epoch": 0.69185360094451, |
| "grad_norm": 0.9748879770068989, |
| "learning_rate": 1.18848247806929e-06, |
| "loss": 0.1201, |
| "step": 879 |
| }, |
| { |
| "epoch": 0.6926406926406926, |
| "grad_norm": 0.9952760658770881, |
| "learning_rate": 1.1829426189702487e-06, |
| "loss": 0.1211, |
| "step": 880 |
| }, |
| { |
| "epoch": 0.6934277843368752, |
| "grad_norm": 0.9561514586133496, |
| "learning_rate": 1.177411697237977e-06, |
| "loss": 0.1208, |
| "step": 881 |
| }, |
| { |
| "epoch": 0.6942148760330579, |
| "grad_norm": 1.0289787958991654, |
| "learning_rate": 1.1718897504045328e-06, |
| "loss": 0.1329, |
| "step": 882 |
| }, |
| { |
| "epoch": 0.6950019677292405, |
| "grad_norm": 1.012367533381528, |
| "learning_rate": 1.1663768159410748e-06, |
| "loss": 0.1286, |
| "step": 883 |
| }, |
| { |
| "epoch": 0.695789059425423, |
| "grad_norm": 0.9932326189371155, |
| "learning_rate": 1.160872931257602e-06, |
| "loss": 0.1207, |
| "step": 884 |
| }, |
| { |
| "epoch": 0.6965761511216056, |
| "grad_norm": 0.9375475650331836, |
| "learning_rate": 1.1553781337027061e-06, |
| "loss": 0.1162, |
| "step": 885 |
| }, |
| { |
| "epoch": 0.6973632428177883, |
| "grad_norm": 1.0035582921316957, |
| "learning_rate": 1.149892460563311e-06, |
| "loss": 0.1272, |
| "step": 886 |
| }, |
| { |
| "epoch": 0.6981503345139709, |
| "grad_norm": 0.969216495536807, |
| "learning_rate": 1.1444159490644278e-06, |
| "loss": 0.1322, |
| "step": 887 |
| }, |
| { |
| "epoch": 0.6989374262101535, |
| "grad_norm": 0.9727140149487835, |
| "learning_rate": 1.1389486363688935e-06, |
| "loss": 0.1109, |
| "step": 888 |
| }, |
| { |
| "epoch": 0.699724517906336, |
| "grad_norm": 1.035921852021017, |
| "learning_rate": 1.1334905595771274e-06, |
| "loss": 0.125, |
| "step": 889 |
| }, |
| { |
| "epoch": 0.7005116096025187, |
| "grad_norm": 0.9817389469807767, |
| "learning_rate": 1.1280417557268735e-06, |
| "loss": 0.1263, |
| "step": 890 |
| }, |
| { |
| "epoch": 0.7012987012987013, |
| "grad_norm": 0.941993125359632, |
| "learning_rate": 1.12260226179295e-06, |
| "loss": 0.1204, |
| "step": 891 |
| }, |
| { |
| "epoch": 0.7020857929948839, |
| "grad_norm": 0.9776393790876531, |
| "learning_rate": 1.1171721146870015e-06, |
| "loss": 0.1351, |
| "step": 892 |
| }, |
| { |
| "epoch": 0.7028728846910665, |
| "grad_norm": 1.0272253940679958, |
| "learning_rate": 1.1117513512572436e-06, |
| "loss": 0.1297, |
| "step": 893 |
| }, |
| { |
| "epoch": 0.7036599763872491, |
| "grad_norm": 0.9825257474446853, |
| "learning_rate": 1.1063400082882188e-06, |
| "loss": 0.1089, |
| "step": 894 |
| }, |
| { |
| "epoch": 0.7044470680834317, |
| "grad_norm": 0.9941185616779367, |
| "learning_rate": 1.10093812250054e-06, |
| "loss": 0.1182, |
| "step": 895 |
| }, |
| { |
| "epoch": 0.7052341597796143, |
| "grad_norm": 0.9809962090348159, |
| "learning_rate": 1.095545730550649e-06, |
| "loss": 0.1221, |
| "step": 896 |
| }, |
| { |
| "epoch": 0.706021251475797, |
| "grad_norm": 0.9359419759382669, |
| "learning_rate": 1.0901628690305593e-06, |
| "loss": 0.1175, |
| "step": 897 |
| }, |
| { |
| "epoch": 0.7068083431719795, |
| "grad_norm": 1.0254153011332428, |
| "learning_rate": 1.0847895744676173e-06, |
| "loss": 0.1364, |
| "step": 898 |
| }, |
| { |
| "epoch": 0.7075954348681621, |
| "grad_norm": 1.0451822058149052, |
| "learning_rate": 1.0794258833242452e-06, |
| "loss": 0.1341, |
| "step": 899 |
| }, |
| { |
| "epoch": 0.7083825265643447, |
| "grad_norm": 1.0267091614696302, |
| "learning_rate": 1.0740718319976992e-06, |
| "loss": 0.1284, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.7091696182605274, |
| "grad_norm": 0.8928053655240218, |
| "learning_rate": 1.0687274568198208e-06, |
| "loss": 0.1009, |
| "step": 901 |
| }, |
| { |
| "epoch": 0.70995670995671, |
| "grad_norm": 1.057234091878292, |
| "learning_rate": 1.063392794056792e-06, |
| "loss": 0.1346, |
| "step": 902 |
| }, |
| { |
| "epoch": 0.7107438016528925, |
| "grad_norm": 0.9612239375437197, |
| "learning_rate": 1.0580678799088847e-06, |
| "loss": 0.1158, |
| "step": 903 |
| }, |
| { |
| "epoch": 0.7115308933490752, |
| "grad_norm": 0.9876590104136502, |
| "learning_rate": 1.0527527505102213e-06, |
| "loss": 0.1193, |
| "step": 904 |
| }, |
| { |
| "epoch": 0.7123179850452578, |
| "grad_norm": 1.0155629806285287, |
| "learning_rate": 1.0474474419285255e-06, |
| "loss": 0.1206, |
| "step": 905 |
| }, |
| { |
| "epoch": 0.7131050767414404, |
| "grad_norm": 1.108914897353474, |
| "learning_rate": 1.0421519901648759e-06, |
| "loss": 0.1244, |
| "step": 906 |
| }, |
| { |
| "epoch": 0.7138921684376229, |
| "grad_norm": 0.9624208122062576, |
| "learning_rate": 1.0368664311534674e-06, |
| "loss": 0.122, |
| "step": 907 |
| }, |
| { |
| "epoch": 0.7146792601338056, |
| "grad_norm": 0.9051835119610858, |
| "learning_rate": 1.031590800761361e-06, |
| "loss": 0.1115, |
| "step": 908 |
| }, |
| { |
| "epoch": 0.7154663518299882, |
| "grad_norm": 1.001641822545354, |
| "learning_rate": 1.0263251347882467e-06, |
| "loss": 0.1205, |
| "step": 909 |
| }, |
| { |
| "epoch": 0.7162534435261708, |
| "grad_norm": 0.9587134043689033, |
| "learning_rate": 1.021069468966194e-06, |
| "loss": 0.114, |
| "step": 910 |
| }, |
| { |
| "epoch": 0.7170405352223534, |
| "grad_norm": 0.9734138315261187, |
| "learning_rate": 1.0158238389594164e-06, |
| "loss": 0.1237, |
| "step": 911 |
| }, |
| { |
| "epoch": 0.717827626918536, |
| "grad_norm": 0.9654730718585164, |
| "learning_rate": 1.0105882803640215e-06, |
| "loss": 0.1241, |
| "step": 912 |
| }, |
| { |
| "epoch": 0.7186147186147186, |
| "grad_norm": 1.0069324283880368, |
| "learning_rate": 1.0053628287077782e-06, |
| "loss": 0.129, |
| "step": 913 |
| }, |
| { |
| "epoch": 0.7194018103109012, |
| "grad_norm": 0.9724112904663149, |
| "learning_rate": 1.000147519449867e-06, |
| "loss": 0.1217, |
| "step": 914 |
| }, |
| { |
| "epoch": 0.7201889020070839, |
| "grad_norm": 0.9009157337976961, |
| "learning_rate": 9.94942387980648e-07, |
| "loss": 0.1215, |
| "step": 915 |
| }, |
| { |
| "epoch": 0.7209759937032665, |
| "grad_norm": 0.9903710831464596, |
| "learning_rate": 9.89747469621411e-07, |
| "loss": 0.1247, |
| "step": 916 |
| }, |
| { |
| "epoch": 0.721763085399449, |
| "grad_norm": 0.9627081908787005, |
| "learning_rate": 9.845627996241459e-07, |
| "loss": 0.1235, |
| "step": 917 |
| }, |
| { |
| "epoch": 0.7225501770956316, |
| "grad_norm": 0.9666668030573422, |
| "learning_rate": 9.793884131712943e-07, |
| "loss": 0.123, |
| "step": 918 |
| }, |
| { |
| "epoch": 0.7233372687918143, |
| "grad_norm": 0.9644911943474369, |
| "learning_rate": 9.742243453755202e-07, |
| "loss": 0.1142, |
| "step": 919 |
| }, |
| { |
| "epoch": 0.7241243604879969, |
| "grad_norm": 0.9984844200263358, |
| "learning_rate": 9.690706312794618e-07, |
| "loss": 0.1251, |
| "step": 920 |
| }, |
| { |
| "epoch": 0.7249114521841794, |
| "grad_norm": 0.988262708447867, |
| "learning_rate": 9.639273058555004e-07, |
| "loss": 0.1233, |
| "step": 921 |
| }, |
| { |
| "epoch": 0.725698543880362, |
| "grad_norm": 0.9062607929130434, |
| "learning_rate": 9.587944040055225e-07, |
| "loss": 0.1116, |
| "step": 922 |
| }, |
| { |
| "epoch": 0.7264856355765447, |
| "grad_norm": 1.0132516720132552, |
| "learning_rate": 9.536719605606795e-07, |
| "loss": 0.1314, |
| "step": 923 |
| }, |
| { |
| "epoch": 0.7272727272727273, |
| "grad_norm": 0.9210291352044477, |
| "learning_rate": 9.485600102811556e-07, |
| "loss": 0.108, |
| "step": 924 |
| }, |
| { |
| "epoch": 0.7280598189689099, |
| "grad_norm": 1.0099118734494892, |
| "learning_rate": 9.434585878559277e-07, |
| "loss": 0.1172, |
| "step": 925 |
| }, |
| { |
| "epoch": 0.7288469106650924, |
| "grad_norm": 1.0237482529235973, |
| "learning_rate": 9.383677279025347e-07, |
| "loss": 0.1186, |
| "step": 926 |
| }, |
| { |
| "epoch": 0.7296340023612751, |
| "grad_norm": 0.9855331385764105, |
| "learning_rate": 9.332874649668369e-07, |
| "loss": 0.1185, |
| "step": 927 |
| }, |
| { |
| "epoch": 0.7304210940574577, |
| "grad_norm": 0.9369233888911801, |
| "learning_rate": 9.282178335227885e-07, |
| "loss": 0.1067, |
| "step": 928 |
| }, |
| { |
| "epoch": 0.7312081857536403, |
| "grad_norm": 1.025834900254658, |
| "learning_rate": 9.231588679721956e-07, |
| "loss": 0.1256, |
| "step": 929 |
| }, |
| { |
| "epoch": 0.731995277449823, |
| "grad_norm": 1.0004815551544541, |
| "learning_rate": 9.181106026444913e-07, |
| "loss": 0.1171, |
| "step": 930 |
| }, |
| { |
| "epoch": 0.7327823691460055, |
| "grad_norm": 0.9247417584553485, |
| "learning_rate": 9.130730717964948e-07, |
| "loss": 0.1132, |
| "step": 931 |
| }, |
| { |
| "epoch": 0.7335694608421881, |
| "grad_norm": 0.9769073592720867, |
| "learning_rate": 9.08046309612185e-07, |
| "loss": 0.1242, |
| "step": 932 |
| }, |
| { |
| "epoch": 0.7343565525383707, |
| "grad_norm": 0.96681906386633, |
| "learning_rate": 9.030303502024662e-07, |
| "loss": 0.1179, |
| "step": 933 |
| }, |
| { |
| "epoch": 0.7351436442345534, |
| "grad_norm": 1.021595769957744, |
| "learning_rate": 8.980252276049345e-07, |
| "loss": 0.1161, |
| "step": 934 |
| }, |
| { |
| "epoch": 0.7359307359307359, |
| "grad_norm": 0.9231065432942811, |
| "learning_rate": 8.930309757836517e-07, |
| "loss": 0.1149, |
| "step": 935 |
| }, |
| { |
| "epoch": 0.7367178276269185, |
| "grad_norm": 1.026367432921577, |
| "learning_rate": 8.880476286289091e-07, |
| "loss": 0.1284, |
| "step": 936 |
| }, |
| { |
| "epoch": 0.7375049193231011, |
| "grad_norm": 0.981215601065822, |
| "learning_rate": 8.830752199570033e-07, |
| "loss": 0.1133, |
| "step": 937 |
| }, |
| { |
| "epoch": 0.7382920110192838, |
| "grad_norm": 0.9212608185738064, |
| "learning_rate": 8.781137835100021e-07, |
| "loss": 0.1077, |
| "step": 938 |
| }, |
| { |
| "epoch": 0.7390791027154664, |
| "grad_norm": 0.9833427367903659, |
| "learning_rate": 8.731633529555167e-07, |
| "loss": 0.1164, |
| "step": 939 |
| }, |
| { |
| "epoch": 0.7398661944116489, |
| "grad_norm": 0.9854894539977124, |
| "learning_rate": 8.682239618864763e-07, |
| "loss": 0.1155, |
| "step": 940 |
| }, |
| { |
| "epoch": 0.7406532861078315, |
| "grad_norm": 0.9551803394241506, |
| "learning_rate": 8.632956438208962e-07, |
| "loss": 0.1162, |
| "step": 941 |
| }, |
| { |
| "epoch": 0.7414403778040142, |
| "grad_norm": 0.9042419017178762, |
| "learning_rate": 8.583784322016503e-07, |
| "loss": 0.109, |
| "step": 942 |
| }, |
| { |
| "epoch": 0.7422274695001968, |
| "grad_norm": 0.9609816099291726, |
| "learning_rate": 8.534723603962497e-07, |
| "loss": 0.1191, |
| "step": 943 |
| }, |
| { |
| "epoch": 0.7430145611963794, |
| "grad_norm": 1.0149972325544658, |
| "learning_rate": 8.48577461696608e-07, |
| "loss": 0.1192, |
| "step": 944 |
| }, |
| { |
| "epoch": 0.743801652892562, |
| "grad_norm": 0.9584184891745349, |
| "learning_rate": 8.436937693188232e-07, |
| "loss": 0.1267, |
| "step": 945 |
| }, |
| { |
| "epoch": 0.7445887445887446, |
| "grad_norm": 0.9986011121611049, |
| "learning_rate": 8.38821316402946e-07, |
| "loss": 0.1177, |
| "step": 946 |
| }, |
| { |
| "epoch": 0.7453758362849272, |
| "grad_norm": 0.962256278467975, |
| "learning_rate": 8.339601360127592e-07, |
| "loss": 0.1131, |
| "step": 947 |
| }, |
| { |
| "epoch": 0.7461629279811098, |
| "grad_norm": 0.9419406227649391, |
| "learning_rate": 8.291102611355526e-07, |
| "loss": 0.1123, |
| "step": 948 |
| }, |
| { |
| "epoch": 0.7469500196772924, |
| "grad_norm": 0.9728190857016107, |
| "learning_rate": 8.242717246818957e-07, |
| "loss": 0.1197, |
| "step": 949 |
| }, |
| { |
| "epoch": 0.747737111373475, |
| "grad_norm": 1.0169044023539633, |
| "learning_rate": 8.1944455948542e-07, |
| "loss": 0.1219, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.7485242030696576, |
| "grad_norm": 0.9972018368498321, |
| "learning_rate": 8.146287983025902e-07, |
| "loss": 0.1241, |
| "step": 951 |
| }, |
| { |
| "epoch": 0.7493112947658402, |
| "grad_norm": 1.040910663691627, |
| "learning_rate": 8.098244738124888e-07, |
| "loss": 0.1138, |
| "step": 952 |
| }, |
| { |
| "epoch": 0.7500983864620229, |
| "grad_norm": 1.0438538265069202, |
| "learning_rate": 8.050316186165862e-07, |
| "loss": 0.134, |
| "step": 953 |
| }, |
| { |
| "epoch": 0.7508854781582054, |
| "grad_norm": 0.9793759854817412, |
| "learning_rate": 8.002502652385278e-07, |
| "loss": 0.1241, |
| "step": 954 |
| }, |
| { |
| "epoch": 0.751672569854388, |
| "grad_norm": 0.9636283038275181, |
| "learning_rate": 7.954804461239054e-07, |
| "loss": 0.1171, |
| "step": 955 |
| }, |
| { |
| "epoch": 0.7524596615505706, |
| "grad_norm": 0.9416057200961391, |
| "learning_rate": 7.907221936400452e-07, |
| "loss": 0.1194, |
| "step": 956 |
| }, |
| { |
| "epoch": 0.7532467532467533, |
| "grad_norm": 0.9258555685816136, |
| "learning_rate": 7.859755400757793e-07, |
| "loss": 0.1199, |
| "step": 957 |
| }, |
| { |
| "epoch": 0.7540338449429359, |
| "grad_norm": 0.9697084160189383, |
| "learning_rate": 7.812405176412354e-07, |
| "loss": 0.1206, |
| "step": 958 |
| }, |
| { |
| "epoch": 0.7548209366391184, |
| "grad_norm": 1.008937777573116, |
| "learning_rate": 7.76517158467611e-07, |
| "loss": 0.1238, |
| "step": 959 |
| }, |
| { |
| "epoch": 0.755608028335301, |
| "grad_norm": 0.987888023607684, |
| "learning_rate": 7.718054946069589e-07, |
| "loss": 0.1246, |
| "step": 960 |
| }, |
| { |
| "epoch": 0.7563951200314837, |
| "grad_norm": 0.9699505992391279, |
| "learning_rate": 7.671055580319706e-07, |
| "loss": 0.1203, |
| "step": 961 |
| }, |
| { |
| "epoch": 0.7571822117276663, |
| "grad_norm": 0.9382257403962697, |
| "learning_rate": 7.62417380635756e-07, |
| "loss": 0.1151, |
| "step": 962 |
| }, |
| { |
| "epoch": 0.7579693034238488, |
| "grad_norm": 0.9437354430265479, |
| "learning_rate": 7.577409942316305e-07, |
| "loss": 0.1163, |
| "step": 963 |
| }, |
| { |
| "epoch": 0.7587563951200315, |
| "grad_norm": 0.9503047394426882, |
| "learning_rate": 7.530764305528959e-07, |
| "loss": 0.1211, |
| "step": 964 |
| }, |
| { |
| "epoch": 0.7595434868162141, |
| "grad_norm": 1.0356071156065598, |
| "learning_rate": 7.484237212526288e-07, |
| "loss": 0.1273, |
| "step": 965 |
| }, |
| { |
| "epoch": 0.7603305785123967, |
| "grad_norm": 0.9856511020736725, |
| "learning_rate": 7.437828979034606e-07, |
| "loss": 0.1315, |
| "step": 966 |
| }, |
| { |
| "epoch": 0.7611176702085793, |
| "grad_norm": 0.9629717208752256, |
| "learning_rate": 7.391539919973698e-07, |
| "loss": 0.1062, |
| "step": 967 |
| }, |
| { |
| "epoch": 0.7619047619047619, |
| "grad_norm": 0.9607163301231785, |
| "learning_rate": 7.345370349454611e-07, |
| "loss": 0.1189, |
| "step": 968 |
| }, |
| { |
| "epoch": 0.7626918536009445, |
| "grad_norm": 0.9578086152431808, |
| "learning_rate": 7.2993205807776e-07, |
| "loss": 0.1183, |
| "step": 969 |
| }, |
| { |
| "epoch": 0.7634789452971271, |
| "grad_norm": 0.9162347277154375, |
| "learning_rate": 7.253390926429918e-07, |
| "loss": 0.1104, |
| "step": 970 |
| }, |
| { |
| "epoch": 0.7642660369933097, |
| "grad_norm": 0.9822027407988481, |
| "learning_rate": 7.207581698083782e-07, |
| "loss": 0.1304, |
| "step": 971 |
| }, |
| { |
| "epoch": 0.7650531286894924, |
| "grad_norm": 0.940452384125095, |
| "learning_rate": 7.161893206594175e-07, |
| "loss": 0.1168, |
| "step": 972 |
| }, |
| { |
| "epoch": 0.7658402203856749, |
| "grad_norm": 0.9588362662800347, |
| "learning_rate": 7.116325761996818e-07, |
| "loss": 0.1206, |
| "step": 973 |
| }, |
| { |
| "epoch": 0.7666273120818575, |
| "grad_norm": 1.011535036970359, |
| "learning_rate": 7.070879673505976e-07, |
| "loss": 0.1141, |
| "step": 974 |
| }, |
| { |
| "epoch": 0.7674144037780402, |
| "grad_norm": 0.9688050929102817, |
| "learning_rate": 7.025555249512461e-07, |
| "loss": 0.1134, |
| "step": 975 |
| }, |
| { |
| "epoch": 0.7682014954742228, |
| "grad_norm": 0.9177610587932681, |
| "learning_rate": 6.980352797581438e-07, |
| "loss": 0.1089, |
| "step": 976 |
| }, |
| { |
| "epoch": 0.7689885871704053, |
| "grad_norm": 1.0472410615763514, |
| "learning_rate": 6.935272624450432e-07, |
| "loss": 0.1249, |
| "step": 977 |
| }, |
| { |
| "epoch": 0.7697756788665879, |
| "grad_norm": 0.9636506719030409, |
| "learning_rate": 6.890315036027156e-07, |
| "loss": 0.1166, |
| "step": 978 |
| }, |
| { |
| "epoch": 0.7705627705627706, |
| "grad_norm": 0.9412495575321557, |
| "learning_rate": 6.845480337387525e-07, |
| "loss": 0.1195, |
| "step": 979 |
| }, |
| { |
| "epoch": 0.7713498622589532, |
| "grad_norm": 0.9930839323289444, |
| "learning_rate": 6.800768832773505e-07, |
| "loss": 0.1267, |
| "step": 980 |
| }, |
| { |
| "epoch": 0.7721369539551358, |
| "grad_norm": 0.9929049734904327, |
| "learning_rate": 6.756180825591099e-07, |
| "loss": 0.1199, |
| "step": 981 |
| }, |
| { |
| "epoch": 0.7729240456513183, |
| "grad_norm": 0.9842816070024283, |
| "learning_rate": 6.711716618408282e-07, |
| "loss": 0.1179, |
| "step": 982 |
| }, |
| { |
| "epoch": 0.773711137347501, |
| "grad_norm": 0.9944580491304532, |
| "learning_rate": 6.66737651295292e-07, |
| "loss": 0.1198, |
| "step": 983 |
| }, |
| { |
| "epoch": 0.7744982290436836, |
| "grad_norm": 0.9839708888147434, |
| "learning_rate": 6.623160810110765e-07, |
| "loss": 0.1193, |
| "step": 984 |
| }, |
| { |
| "epoch": 0.7752853207398662, |
| "grad_norm": 0.9098159637038072, |
| "learning_rate": 6.579069809923367e-07, |
| "loss": 0.1123, |
| "step": 985 |
| }, |
| { |
| "epoch": 0.7760724124360489, |
| "grad_norm": 0.9606605502093222, |
| "learning_rate": 6.535103811586085e-07, |
| "loss": 0.1174, |
| "step": 986 |
| }, |
| { |
| "epoch": 0.7768595041322314, |
| "grad_norm": 1.0270282467875798, |
| "learning_rate": 6.491263113446005e-07, |
| "loss": 0.1287, |
| "step": 987 |
| }, |
| { |
| "epoch": 0.777646595828414, |
| "grad_norm": 0.9366844684145114, |
| "learning_rate": 6.44754801299998e-07, |
| "loss": 0.1158, |
| "step": 988 |
| }, |
| { |
| "epoch": 0.7784336875245966, |
| "grad_norm": 0.9037352215899601, |
| "learning_rate": 6.403958806892535e-07, |
| "loss": 0.1053, |
| "step": 989 |
| }, |
| { |
| "epoch": 0.7792207792207793, |
| "grad_norm": 0.9440225724017625, |
| "learning_rate": 6.360495790913926e-07, |
| "loss": 0.114, |
| "step": 990 |
| }, |
| { |
| "epoch": 0.7800078709169618, |
| "grad_norm": 0.9351662026461205, |
| "learning_rate": 6.317159259998074e-07, |
| "loss": 0.113, |
| "step": 991 |
| }, |
| { |
| "epoch": 0.7807949626131444, |
| "grad_norm": 0.9872272779542443, |
| "learning_rate": 6.273949508220612e-07, |
| "loss": 0.1217, |
| "step": 992 |
| }, |
| { |
| "epoch": 0.781582054309327, |
| "grad_norm": 1.0021043961378415, |
| "learning_rate": 6.23086682879686e-07, |
| "loss": 0.1194, |
| "step": 993 |
| }, |
| { |
| "epoch": 0.7823691460055097, |
| "grad_norm": 0.9797645648660196, |
| "learning_rate": 6.187911514079834e-07, |
| "loss": 0.1294, |
| "step": 994 |
| }, |
| { |
| "epoch": 0.7831562377016923, |
| "grad_norm": 1.0054784443943467, |
| "learning_rate": 6.14508385555829e-07, |
| "loss": 0.1236, |
| "step": 995 |
| }, |
| { |
| "epoch": 0.7839433293978748, |
| "grad_norm": 0.9433076242026539, |
| "learning_rate": 6.102384143854698e-07, |
| "loss": 0.1147, |
| "step": 996 |
| }, |
| { |
| "epoch": 0.7847304210940574, |
| "grad_norm": 0.9383907844400864, |
| "learning_rate": 6.059812668723336e-07, |
| "loss": 0.115, |
| "step": 997 |
| }, |
| { |
| "epoch": 0.7855175127902401, |
| "grad_norm": 0.9452315722932242, |
| "learning_rate": 6.017369719048255e-07, |
| "loss": 0.1154, |
| "step": 998 |
| }, |
| { |
| "epoch": 0.7863046044864227, |
| "grad_norm": 0.9247930090252802, |
| "learning_rate": 5.975055582841358e-07, |
| "loss": 0.1127, |
| "step": 999 |
| }, |
| { |
| "epoch": 0.7870916961826053, |
| "grad_norm": 1.0061872579787852, |
| "learning_rate": 5.932870547240455e-07, |
| "loss": 0.1183, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.7870916961826053, |
| "eval_loss": 0.11849173903465271, |
| "eval_runtime": 18.0453, |
| "eval_samples_per_second": 45.552, |
| "eval_steps_per_second": 5.708, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.7878787878787878, |
| "grad_norm": 0.9623831636196449, |
| "learning_rate": 5.890814898507277e-07, |
| "loss": 0.1201, |
| "step": 1001 |
| }, |
| { |
| "epoch": 0.7886658795749705, |
| "grad_norm": 1.0229456536544794, |
| "learning_rate": 5.848888922025553e-07, |
| "loss": 0.1223, |
| "step": 1002 |
| }, |
| { |
| "epoch": 0.7894529712711531, |
| "grad_norm": 1.0277580850565635, |
| "learning_rate": 5.8070929022991e-07, |
| "loss": 0.1178, |
| "step": 1003 |
| }, |
| { |
| "epoch": 0.7902400629673357, |
| "grad_norm": 1.0160977169162413, |
| "learning_rate": 5.76542712294983e-07, |
| "loss": 0.1216, |
| "step": 1004 |
| }, |
| { |
| "epoch": 0.7910271546635183, |
| "grad_norm": 0.9181674404236817, |
| "learning_rate": 5.723891866715899e-07, |
| "loss": 0.1118, |
| "step": 1005 |
| }, |
| { |
| "epoch": 0.7918142463597009, |
| "grad_norm": 0.9799718055295829, |
| "learning_rate": 5.682487415449719e-07, |
| "loss": 0.1217, |
| "step": 1006 |
| }, |
| { |
| "epoch": 0.7926013380558835, |
| "grad_norm": 0.980465660075739, |
| "learning_rate": 5.641214050116098e-07, |
| "loss": 0.1252, |
| "step": 1007 |
| }, |
| { |
| "epoch": 0.7933884297520661, |
| "grad_norm": 0.9234542096536653, |
| "learning_rate": 5.600072050790317e-07, |
| "loss": 0.1096, |
| "step": 1008 |
| }, |
| { |
| "epoch": 0.7941755214482488, |
| "grad_norm": 0.8947896913580902, |
| "learning_rate": 5.559061696656199e-07, |
| "loss": 0.1075, |
| "step": 1009 |
| }, |
| { |
| "epoch": 0.7949626131444313, |
| "grad_norm": 0.9790572754851533, |
| "learning_rate": 5.518183266004276e-07, |
| "loss": 0.1171, |
| "step": 1010 |
| }, |
| { |
| "epoch": 0.7957497048406139, |
| "grad_norm": 0.9667752845159187, |
| "learning_rate": 5.477437036229832e-07, |
| "loss": 0.1098, |
| "step": 1011 |
| }, |
| { |
| "epoch": 0.7965367965367965, |
| "grad_norm": 1.1238301396219903, |
| "learning_rate": 5.436823283831083e-07, |
| "loss": 0.1373, |
| "step": 1012 |
| }, |
| { |
| "epoch": 0.7973238882329792, |
| "grad_norm": 0.9927017197297952, |
| "learning_rate": 5.396342284407252e-07, |
| "loss": 0.1188, |
| "step": 1013 |
| }, |
| { |
| "epoch": 0.7981109799291618, |
| "grad_norm": 0.9934845633471078, |
| "learning_rate": 5.355994312656734e-07, |
| "loss": 0.1142, |
| "step": 1014 |
| }, |
| { |
| "epoch": 0.7988980716253443, |
| "grad_norm": 0.9432843712008361, |
| "learning_rate": 5.315779642375199e-07, |
| "loss": 0.1158, |
| "step": 1015 |
| }, |
| { |
| "epoch": 0.799685163321527, |
| "grad_norm": 1.0251992909650254, |
| "learning_rate": 5.275698546453775e-07, |
| "loss": 0.1175, |
| "step": 1016 |
| }, |
| { |
| "epoch": 0.8004722550177096, |
| "grad_norm": 1.010003247709555, |
| "learning_rate": 5.235751296877148e-07, |
| "loss": 0.1223, |
| "step": 1017 |
| }, |
| { |
| "epoch": 0.8012593467138922, |
| "grad_norm": 1.0290265946769084, |
| "learning_rate": 5.195938164721767e-07, |
| "loss": 0.1213, |
| "step": 1018 |
| }, |
| { |
| "epoch": 0.8020464384100747, |
| "grad_norm": 1.0065180201235937, |
| "learning_rate": 5.156259420153962e-07, |
| "loss": 0.1238, |
| "step": 1019 |
| }, |
| { |
| "epoch": 0.8028335301062574, |
| "grad_norm": 0.9673994358176322, |
| "learning_rate": 5.116715332428118e-07, |
| "loss": 0.1106, |
| "step": 1020 |
| }, |
| { |
| "epoch": 0.80362062180244, |
| "grad_norm": 0.9855226309577549, |
| "learning_rate": 5.077306169884888e-07, |
| "loss": 0.1107, |
| "step": 1021 |
| }, |
| { |
| "epoch": 0.8044077134986226, |
| "grad_norm": 0.9719491378095487, |
| "learning_rate": 5.038032199949313e-07, |
| "loss": 0.1241, |
| "step": 1022 |
| }, |
| { |
| "epoch": 0.8051948051948052, |
| "grad_norm": 0.9767976318784359, |
| "learning_rate": 4.998893689129061e-07, |
| "loss": 0.1165, |
| "step": 1023 |
| }, |
| { |
| "epoch": 0.8059818968909878, |
| "grad_norm": 0.9563874747183178, |
| "learning_rate": 4.959890903012568e-07, |
| "loss": 0.1084, |
| "step": 1024 |
| }, |
| { |
| "epoch": 0.8067689885871704, |
| "grad_norm": 0.9724824504692731, |
| "learning_rate": 4.921024106267283e-07, |
| "loss": 0.1199, |
| "step": 1025 |
| }, |
| { |
| "epoch": 0.807556080283353, |
| "grad_norm": 0.9939320935755256, |
| "learning_rate": 4.882293562637827e-07, |
| "loss": 0.1221, |
| "step": 1026 |
| }, |
| { |
| "epoch": 0.8083431719795356, |
| "grad_norm": 0.9407936084194475, |
| "learning_rate": 4.843699534944258e-07, |
| "loss": 0.114, |
| "step": 1027 |
| }, |
| { |
| "epoch": 0.8091302636757183, |
| "grad_norm": 1.00154528232707, |
| "learning_rate": 4.805242285080222e-07, |
| "loss": 0.116, |
| "step": 1028 |
| }, |
| { |
| "epoch": 0.8099173553719008, |
| "grad_norm": 0.9808095624927836, |
| "learning_rate": 4.7669220740112376e-07, |
| "loss": 0.1166, |
| "step": 1029 |
| }, |
| { |
| "epoch": 0.8107044470680834, |
| "grad_norm": 0.9094800308143445, |
| "learning_rate": 4.728739161772874e-07, |
| "loss": 0.1091, |
| "step": 1030 |
| }, |
| { |
| "epoch": 0.811491538764266, |
| "grad_norm": 0.8938814168806941, |
| "learning_rate": 4.690693807469035e-07, |
| "loss": 0.1032, |
| "step": 1031 |
| }, |
| { |
| "epoch": 0.8122786304604487, |
| "grad_norm": 0.9946966466940443, |
| "learning_rate": 4.6527862692701487e-07, |
| "loss": 0.1158, |
| "step": 1032 |
| }, |
| { |
| "epoch": 0.8130657221566312, |
| "grad_norm": 0.9504344369313477, |
| "learning_rate": 4.615016804411465e-07, |
| "loss": 0.1114, |
| "step": 1033 |
| }, |
| { |
| "epoch": 0.8138528138528138, |
| "grad_norm": 0.9453878426287087, |
| "learning_rate": 4.5773856691912726e-07, |
| "loss": 0.1208, |
| "step": 1034 |
| }, |
| { |
| "epoch": 0.8146399055489965, |
| "grad_norm": 1.0239509589587361, |
| "learning_rate": 4.53989311896918e-07, |
| "loss": 0.1186, |
| "step": 1035 |
| }, |
| { |
| "epoch": 0.8154269972451791, |
| "grad_norm": 0.915087834897205, |
| "learning_rate": 4.502539408164386e-07, |
| "loss": 0.106, |
| "step": 1036 |
| }, |
| { |
| "epoch": 0.8162140889413617, |
| "grad_norm": 0.936045456280838, |
| "learning_rate": 4.465324790253922e-07, |
| "loss": 0.1104, |
| "step": 1037 |
| }, |
| { |
| "epoch": 0.8170011806375442, |
| "grad_norm": 0.986165452237028, |
| "learning_rate": 4.428249517770986e-07, |
| "loss": 0.1137, |
| "step": 1038 |
| }, |
| { |
| "epoch": 0.8177882723337269, |
| "grad_norm": 0.9641211589952483, |
| "learning_rate": 4.391313842303166e-07, |
| "loss": 0.1164, |
| "step": 1039 |
| }, |
| { |
| "epoch": 0.8185753640299095, |
| "grad_norm": 0.9334697738707801, |
| "learning_rate": 4.3545180144907857e-07, |
| "loss": 0.113, |
| "step": 1040 |
| }, |
| { |
| "epoch": 0.8193624557260921, |
| "grad_norm": 0.9871068120631671, |
| "learning_rate": 4.3178622840251647e-07, |
| "loss": 0.1241, |
| "step": 1041 |
| }, |
| { |
| "epoch": 0.8201495474222747, |
| "grad_norm": 0.9529229208831895, |
| "learning_rate": 4.2813468996469654e-07, |
| "loss": 0.121, |
| "step": 1042 |
| }, |
| { |
| "epoch": 0.8209366391184573, |
| "grad_norm": 0.9336358260352323, |
| "learning_rate": 4.2449721091444545e-07, |
| "loss": 0.1134, |
| "step": 1043 |
| }, |
| { |
| "epoch": 0.8217237308146399, |
| "grad_norm": 1.0307131336178375, |
| "learning_rate": 4.2087381593518716e-07, |
| "loss": 0.1274, |
| "step": 1044 |
| }, |
| { |
| "epoch": 0.8225108225108225, |
| "grad_norm": 0.9433490812621332, |
| "learning_rate": 4.1726452961477147e-07, |
| "loss": 0.116, |
| "step": 1045 |
| }, |
| { |
| "epoch": 0.8232979142070052, |
| "grad_norm": 0.9751019494649681, |
| "learning_rate": 4.136693764453101e-07, |
| "loss": 0.1129, |
| "step": 1046 |
| }, |
| { |
| "epoch": 0.8240850059031877, |
| "grad_norm": 1.056328093801445, |
| "learning_rate": 4.1008838082300743e-07, |
| "loss": 0.1168, |
| "step": 1047 |
| }, |
| { |
| "epoch": 0.8248720975993703, |
| "grad_norm": 1.0079092402082175, |
| "learning_rate": 4.065215670479991e-07, |
| "loss": 0.1258, |
| "step": 1048 |
| }, |
| { |
| "epoch": 0.8256591892955529, |
| "grad_norm": 0.9852819205932097, |
| "learning_rate": 4.02968959324182e-07, |
| "loss": 0.1161, |
| "step": 1049 |
| }, |
| { |
| "epoch": 0.8264462809917356, |
| "grad_norm": 0.9840178810234324, |
| "learning_rate": 3.9943058175905493e-07, |
| "loss": 0.1184, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.8272333726879182, |
| "grad_norm": 1.0203308364665442, |
| "learning_rate": 3.9590645836355275e-07, |
| "loss": 0.1232, |
| "step": 1051 |
| }, |
| { |
| "epoch": 0.8280204643841007, |
| "grad_norm": 0.967005062015959, |
| "learning_rate": 3.923966130518814e-07, |
| "loss": 0.1209, |
| "step": 1052 |
| }, |
| { |
| "epoch": 0.8288075560802833, |
| "grad_norm": 0.9434419280443521, |
| "learning_rate": 3.889010696413606e-07, |
| "loss": 0.1211, |
| "step": 1053 |
| }, |
| { |
| "epoch": 0.829594647776466, |
| "grad_norm": 0.9442684717641329, |
| "learning_rate": 3.8541985185225645e-07, |
| "loss": 0.1078, |
| "step": 1054 |
| }, |
| { |
| "epoch": 0.8303817394726486, |
| "grad_norm": 0.9927726007062886, |
| "learning_rate": 3.819529833076263e-07, |
| "loss": 0.1214, |
| "step": 1055 |
| }, |
| { |
| "epoch": 0.8311688311688312, |
| "grad_norm": 0.9486732370194126, |
| "learning_rate": 3.7850048753315274e-07, |
| "loss": 0.1087, |
| "step": 1056 |
| }, |
| { |
| "epoch": 0.8319559228650137, |
| "grad_norm": 0.9803070634979109, |
| "learning_rate": 3.750623879569895e-07, |
| "loss": 0.1181, |
| "step": 1057 |
| }, |
| { |
| "epoch": 0.8327430145611964, |
| "grad_norm": 1.0010555667628969, |
| "learning_rate": 3.716387079095973e-07, |
| "loss": 0.1172, |
| "step": 1058 |
| }, |
| { |
| "epoch": 0.833530106257379, |
| "grad_norm": 0.9548728126471466, |
| "learning_rate": 3.6822947062359004e-07, |
| "loss": 0.1125, |
| "step": 1059 |
| }, |
| { |
| "epoch": 0.8343171979535616, |
| "grad_norm": 1.0333251380116057, |
| "learning_rate": 3.6483469923357327e-07, |
| "loss": 0.1119, |
| "step": 1060 |
| }, |
| { |
| "epoch": 0.8351042896497441, |
| "grad_norm": 0.9473784570136893, |
| "learning_rate": 3.614544167759901e-07, |
| "loss": 0.1136, |
| "step": 1061 |
| }, |
| { |
| "epoch": 0.8358913813459268, |
| "grad_norm": 0.9283902827668026, |
| "learning_rate": 3.5808864618896295e-07, |
| "loss": 0.1004, |
| "step": 1062 |
| }, |
| { |
| "epoch": 0.8366784730421094, |
| "grad_norm": 0.9856546672764643, |
| "learning_rate": 3.5473741031213983e-07, |
| "loss": 0.1136, |
| "step": 1063 |
| }, |
| { |
| "epoch": 0.837465564738292, |
| "grad_norm": 0.955024488651013, |
| "learning_rate": 3.51400731886537e-07, |
| "loss": 0.1199, |
| "step": 1064 |
| }, |
| { |
| "epoch": 0.8382526564344747, |
| "grad_norm": 0.9527578800496054, |
| "learning_rate": 3.4807863355438703e-07, |
| "loss": 0.1178, |
| "step": 1065 |
| }, |
| { |
| "epoch": 0.8390397481306572, |
| "grad_norm": 0.9748866277343534, |
| "learning_rate": 3.447711378589841e-07, |
| "loss": 0.1126, |
| "step": 1066 |
| }, |
| { |
| "epoch": 0.8398268398268398, |
| "grad_norm": 0.9722315190803439, |
| "learning_rate": 3.414782672445291e-07, |
| "loss": 0.1143, |
| "step": 1067 |
| }, |
| { |
| "epoch": 0.8406139315230224, |
| "grad_norm": 1.0223007557494088, |
| "learning_rate": 3.3820004405598157e-07, |
| "loss": 0.1141, |
| "step": 1068 |
| }, |
| { |
| "epoch": 0.8414010232192051, |
| "grad_norm": 0.969999076611352, |
| "learning_rate": 3.3493649053890325e-07, |
| "loss": 0.1161, |
| "step": 1069 |
| }, |
| { |
| "epoch": 0.8421881149153877, |
| "grad_norm": 0.9878840165050939, |
| "learning_rate": 3.3168762883931256e-07, |
| "loss": 0.1164, |
| "step": 1070 |
| }, |
| { |
| "epoch": 0.8429752066115702, |
| "grad_norm": 1.030579753477139, |
| "learning_rate": 3.284534810035278e-07, |
| "loss": 0.1258, |
| "step": 1071 |
| }, |
| { |
| "epoch": 0.8437622983077528, |
| "grad_norm": 0.987395029549749, |
| "learning_rate": 3.252340689780245e-07, |
| "loss": 0.1219, |
| "step": 1072 |
| }, |
| { |
| "epoch": 0.8445493900039355, |
| "grad_norm": 0.9750378310364627, |
| "learning_rate": 3.2202941460927977e-07, |
| "loss": 0.1275, |
| "step": 1073 |
| }, |
| { |
| "epoch": 0.8453364817001181, |
| "grad_norm": 0.9446071297273908, |
| "learning_rate": 3.1883953964363057e-07, |
| "loss": 0.1177, |
| "step": 1074 |
| }, |
| { |
| "epoch": 0.8461235733963006, |
| "grad_norm": 0.9469617356782836, |
| "learning_rate": 3.156644657271196e-07, |
| "loss": 0.1128, |
| "step": 1075 |
| }, |
| { |
| "epoch": 0.8469106650924833, |
| "grad_norm": 0.951509014508041, |
| "learning_rate": 3.12504214405355e-07, |
| "loss": 0.108, |
| "step": 1076 |
| }, |
| { |
| "epoch": 0.8476977567886659, |
| "grad_norm": 0.9472516964934904, |
| "learning_rate": 3.093588071233578e-07, |
| "loss": 0.1141, |
| "step": 1077 |
| }, |
| { |
| "epoch": 0.8484848484848485, |
| "grad_norm": 0.9084748554063148, |
| "learning_rate": 3.06228265225422e-07, |
| "loss": 0.1116, |
| "step": 1078 |
| }, |
| { |
| "epoch": 0.8492719401810311, |
| "grad_norm": 0.9337280963011981, |
| "learning_rate": 3.031126099549653e-07, |
| "loss": 0.1119, |
| "step": 1079 |
| }, |
| { |
| "epoch": 0.8500590318772137, |
| "grad_norm": 0.9417103223229273, |
| "learning_rate": 3.000118624543888e-07, |
| "loss": 0.1117, |
| "step": 1080 |
| }, |
| { |
| "epoch": 0.8508461235733963, |
| "grad_norm": 0.9259525236801444, |
| "learning_rate": 2.9692604376492935e-07, |
| "loss": 0.1067, |
| "step": 1081 |
| }, |
| { |
| "epoch": 0.8516332152695789, |
| "grad_norm": 0.98856338356212, |
| "learning_rate": 2.9385517482651974e-07, |
| "loss": 0.1218, |
| "step": 1082 |
| }, |
| { |
| "epoch": 0.8524203069657615, |
| "grad_norm": 0.9208981942587281, |
| "learning_rate": 2.907992764776471e-07, |
| "loss": 0.1078, |
| "step": 1083 |
| }, |
| { |
| "epoch": 0.8532073986619442, |
| "grad_norm": 1.0070559943311361, |
| "learning_rate": 2.877583694552083e-07, |
| "loss": 0.1236, |
| "step": 1084 |
| }, |
| { |
| "epoch": 0.8539944903581267, |
| "grad_norm": 1.0056266955151931, |
| "learning_rate": 2.847324743943733e-07, |
| "loss": 0.1168, |
| "step": 1085 |
| }, |
| { |
| "epoch": 0.8547815820543093, |
| "grad_norm": 0.9843297655404193, |
| "learning_rate": 2.8172161182844076e-07, |
| "loss": 0.1179, |
| "step": 1086 |
| }, |
| { |
| "epoch": 0.855568673750492, |
| "grad_norm": 0.9794092387444499, |
| "learning_rate": 2.7872580218870293e-07, |
| "loss": 0.1143, |
| "step": 1087 |
| }, |
| { |
| "epoch": 0.8563557654466746, |
| "grad_norm": 0.9193678895589191, |
| "learning_rate": 2.757450658043029e-07, |
| "loss": 0.1033, |
| "step": 1088 |
| }, |
| { |
| "epoch": 0.8571428571428571, |
| "grad_norm": 0.943695053791707, |
| "learning_rate": 2.7277942290210105e-07, |
| "loss": 0.1197, |
| "step": 1089 |
| }, |
| { |
| "epoch": 0.8579299488390397, |
| "grad_norm": 0.9996493471626344, |
| "learning_rate": 2.698288936065338e-07, |
| "loss": 0.1121, |
| "step": 1090 |
| }, |
| { |
| "epoch": 0.8587170405352224, |
| "grad_norm": 0.9457549280655658, |
| "learning_rate": 2.6689349793947993e-07, |
| "loss": 0.1135, |
| "step": 1091 |
| }, |
| { |
| "epoch": 0.859504132231405, |
| "grad_norm": 0.9370356637509019, |
| "learning_rate": 2.639732558201219e-07, |
| "loss": 0.1142, |
| "step": 1092 |
| }, |
| { |
| "epoch": 0.8602912239275876, |
| "grad_norm": 0.9355392394517238, |
| "learning_rate": 2.610681870648149e-07, |
| "loss": 0.1101, |
| "step": 1093 |
| }, |
| { |
| "epoch": 0.8610783156237701, |
| "grad_norm": 0.9395826896807081, |
| "learning_rate": 2.5817831138694685e-07, |
| "loss": 0.1143, |
| "step": 1094 |
| }, |
| { |
| "epoch": 0.8618654073199528, |
| "grad_norm": 0.9169613408255519, |
| "learning_rate": 2.553036483968094e-07, |
| "loss": 0.1125, |
| "step": 1095 |
| }, |
| { |
| "epoch": 0.8626524990161354, |
| "grad_norm": 0.9214498686393413, |
| "learning_rate": 2.5244421760146354e-07, |
| "loss": 0.1061, |
| "step": 1096 |
| }, |
| { |
| "epoch": 0.863439590712318, |
| "grad_norm": 0.9759019789855492, |
| "learning_rate": 2.496000384046046e-07, |
| "loss": 0.1134, |
| "step": 1097 |
| }, |
| { |
| "epoch": 0.8642266824085005, |
| "grad_norm": 1.002942267675624, |
| "learning_rate": 2.467711301064349e-07, |
| "loss": 0.1249, |
| "step": 1098 |
| }, |
| { |
| "epoch": 0.8650137741046832, |
| "grad_norm": 0.9774180094397517, |
| "learning_rate": 2.4395751190352924e-07, |
| "loss": 0.1192, |
| "step": 1099 |
| }, |
| { |
| "epoch": 0.8658008658008658, |
| "grad_norm": 0.9977028529540362, |
| "learning_rate": 2.411592028887058e-07, |
| "loss": 0.1189, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.8665879574970484, |
| "grad_norm": 0.9904872465257003, |
| "learning_rate": 2.383762220508984e-07, |
| "loss": 0.1183, |
| "step": 1101 |
| }, |
| { |
| "epoch": 0.867375049193231, |
| "grad_norm": 0.9148412123541501, |
| "learning_rate": 2.356085882750242e-07, |
| "loss": 0.1078, |
| "step": 1102 |
| }, |
| { |
| "epoch": 0.8681621408894136, |
| "grad_norm": 1.0117176122067204, |
| "learning_rate": 2.328563203418574e-07, |
| "loss": 0.1217, |
| "step": 1103 |
| }, |
| { |
| "epoch": 0.8689492325855962, |
| "grad_norm": 0.9872081797420905, |
| "learning_rate": 2.3011943692790389e-07, |
| "loss": 0.117, |
| "step": 1104 |
| }, |
| { |
| "epoch": 0.8697363242817788, |
| "grad_norm": 0.9476791655485511, |
| "learning_rate": 2.2739795660526948e-07, |
| "loss": 0.1157, |
| "step": 1105 |
| }, |
| { |
| "epoch": 0.8705234159779615, |
| "grad_norm": 0.9183530530163464, |
| "learning_rate": 2.246918978415394e-07, |
| "loss": 0.1108, |
| "step": 1106 |
| }, |
| { |
| "epoch": 0.8713105076741441, |
| "grad_norm": 0.9622583408924335, |
| "learning_rate": 2.2200127899964786e-07, |
| "loss": 0.1188, |
| "step": 1107 |
| }, |
| { |
| "epoch": 0.8720975993703266, |
| "grad_norm": 0.9915067004751748, |
| "learning_rate": 2.1932611833775846e-07, |
| "loss": 0.1151, |
| "step": 1108 |
| }, |
| { |
| "epoch": 0.8728846910665092, |
| "grad_norm": 0.9404810815181894, |
| "learning_rate": 2.1666643400913512e-07, |
| "loss": 0.1133, |
| "step": 1109 |
| }, |
| { |
| "epoch": 0.8736717827626919, |
| "grad_norm": 0.9750904254975121, |
| "learning_rate": 2.1402224406202377e-07, |
| "loss": 0.1187, |
| "step": 1110 |
| }, |
| { |
| "epoch": 0.8744588744588745, |
| "grad_norm": 0.942666742797311, |
| "learning_rate": 2.1139356643952667e-07, |
| "loss": 0.1133, |
| "step": 1111 |
| }, |
| { |
| "epoch": 0.875245966155057, |
| "grad_norm": 0.9261400322366565, |
| "learning_rate": 2.0878041897948121e-07, |
| "loss": 0.1095, |
| "step": 1112 |
| }, |
| { |
| "epoch": 0.8760330578512396, |
| "grad_norm": 1.0714254257987408, |
| "learning_rate": 2.0618281941434058e-07, |
| "loss": 0.1197, |
| "step": 1113 |
| }, |
| { |
| "epoch": 0.8768201495474223, |
| "grad_norm": 0.9322738584358286, |
| "learning_rate": 2.036007853710503e-07, |
| "loss": 0.114, |
| "step": 1114 |
| }, |
| { |
| "epoch": 0.8776072412436049, |
| "grad_norm": 0.9346649367642453, |
| "learning_rate": 2.0103433437093256e-07, |
| "loss": 0.1027, |
| "step": 1115 |
| }, |
| { |
| "epoch": 0.8783943329397875, |
| "grad_norm": 0.9499461297298013, |
| "learning_rate": 1.9848348382956294e-07, |
| "loss": 0.1228, |
| "step": 1116 |
| }, |
| { |
| "epoch": 0.87918142463597, |
| "grad_norm": 0.9811562591520676, |
| "learning_rate": 1.9594825105665654e-07, |
| "loss": 0.1168, |
| "step": 1117 |
| }, |
| { |
| "epoch": 0.8799685163321527, |
| "grad_norm": 0.901833893893408, |
| "learning_rate": 1.934286532559468e-07, |
| "loss": 0.0992, |
| "step": 1118 |
| }, |
| { |
| "epoch": 0.8807556080283353, |
| "grad_norm": 0.9566664879297264, |
| "learning_rate": 1.9092470752507225e-07, |
| "loss": 0.1114, |
| "step": 1119 |
| }, |
| { |
| "epoch": 0.8815426997245179, |
| "grad_norm": 0.8992036132523128, |
| "learning_rate": 1.8843643085545677e-07, |
| "loss": 0.1113, |
| "step": 1120 |
| }, |
| { |
| "epoch": 0.8823297914207006, |
| "grad_norm": 1.031563578089281, |
| "learning_rate": 1.8596384013219726e-07, |
| "loss": 0.1168, |
| "step": 1121 |
| }, |
| { |
| "epoch": 0.8831168831168831, |
| "grad_norm": 0.942026348911581, |
| "learning_rate": 1.8350695213394777e-07, |
| "loss": 0.1206, |
| "step": 1122 |
| }, |
| { |
| "epoch": 0.8839039748130657, |
| "grad_norm": 0.9439190891698341, |
| "learning_rate": 1.8106578353280585e-07, |
| "loss": 0.1138, |
| "step": 1123 |
| }, |
| { |
| "epoch": 0.8846910665092483, |
| "grad_norm": 0.9261746330624396, |
| "learning_rate": 1.7864035089419973e-07, |
| "loss": 0.1118, |
| "step": 1124 |
| }, |
| { |
| "epoch": 0.885478158205431, |
| "grad_norm": 0.9958265328548447, |
| "learning_rate": 1.7623067067677467e-07, |
| "loss": 0.1235, |
| "step": 1125 |
| }, |
| { |
| "epoch": 0.8862652499016135, |
| "grad_norm": 1.0264812939307284, |
| "learning_rate": 1.7383675923228372e-07, |
| "loss": 0.1221, |
| "step": 1126 |
| }, |
| { |
| "epoch": 0.8870523415977961, |
| "grad_norm": 1.0457105898882355, |
| "learning_rate": 1.7145863280547348e-07, |
| "loss": 0.1183, |
| "step": 1127 |
| }, |
| { |
| "epoch": 0.8878394332939787, |
| "grad_norm": 0.9616200935637597, |
| "learning_rate": 1.6909630753397716e-07, |
| "loss": 0.1055, |
| "step": 1128 |
| }, |
| { |
| "epoch": 0.8886265249901614, |
| "grad_norm": 1.0114965743550393, |
| "learning_rate": 1.6674979944820258e-07, |
| "loss": 0.1247, |
| "step": 1129 |
| }, |
| { |
| "epoch": 0.889413616686344, |
| "grad_norm": 1.0002918205099012, |
| "learning_rate": 1.644191244712251e-07, |
| "loss": 0.1245, |
| "step": 1130 |
| }, |
| { |
| "epoch": 0.8902007083825265, |
| "grad_norm": 1.0106011414793612, |
| "learning_rate": 1.621042984186777e-07, |
| "loss": 0.1222, |
| "step": 1131 |
| }, |
| { |
| "epoch": 0.8909878000787091, |
| "grad_norm": 0.9663301334412446, |
| "learning_rate": 1.598053369986463e-07, |
| "loss": 0.1194, |
| "step": 1132 |
| }, |
| { |
| "epoch": 0.8917748917748918, |
| "grad_norm": 0.9465191160189073, |
| "learning_rate": 1.5752225581155995e-07, |
| "loss": 0.1085, |
| "step": 1133 |
| }, |
| { |
| "epoch": 0.8925619834710744, |
| "grad_norm": 1.0466368501224192, |
| "learning_rate": 1.5525507035008852e-07, |
| "loss": 0.1306, |
| "step": 1134 |
| }, |
| { |
| "epoch": 0.893349075167257, |
| "grad_norm": 0.9494535695924311, |
| "learning_rate": 1.5300379599903408e-07, |
| "loss": 0.1172, |
| "step": 1135 |
| }, |
| { |
| "epoch": 0.8941361668634396, |
| "grad_norm": 0.9062893885912572, |
| "learning_rate": 1.507684480352292e-07, |
| "loss": 0.109, |
| "step": 1136 |
| }, |
| { |
| "epoch": 0.8949232585596222, |
| "grad_norm": 0.9371709714503786, |
| "learning_rate": 1.4854904162743127e-07, |
| "loss": 0.1047, |
| "step": 1137 |
| }, |
| { |
| "epoch": 0.8957103502558048, |
| "grad_norm": 1.0110155118454318, |
| "learning_rate": 1.4634559183622193e-07, |
| "loss": 0.126, |
| "step": 1138 |
| }, |
| { |
| "epoch": 0.8964974419519874, |
| "grad_norm": 1.0367843431150414, |
| "learning_rate": 1.4415811361390142e-07, |
| "loss": 0.1251, |
| "step": 1139 |
| }, |
| { |
| "epoch": 0.89728453364817, |
| "grad_norm": 0.9836901063346848, |
| "learning_rate": 1.4198662180439166e-07, |
| "loss": 0.1225, |
| "step": 1140 |
| }, |
| { |
| "epoch": 0.8980716253443526, |
| "grad_norm": 1.0729047805019176, |
| "learning_rate": 1.3983113114313078e-07, |
| "loss": 0.1321, |
| "step": 1141 |
| }, |
| { |
| "epoch": 0.8988587170405352, |
| "grad_norm": 0.9172308694693998, |
| "learning_rate": 1.3769165625697633e-07, |
| "loss": 0.1094, |
| "step": 1142 |
| }, |
| { |
| "epoch": 0.8996458087367178, |
| "grad_norm": 0.9246235655173746, |
| "learning_rate": 1.355682116641052e-07, |
| "loss": 0.1098, |
| "step": 1143 |
| }, |
| { |
| "epoch": 0.9004329004329005, |
| "grad_norm": 0.9671203173685715, |
| "learning_rate": 1.3346081177391474e-07, |
| "loss": 0.1084, |
| "step": 1144 |
| }, |
| { |
| "epoch": 0.901219992129083, |
| "grad_norm": 0.9683781341379243, |
| "learning_rate": 1.3136947088692537e-07, |
| "loss": 0.119, |
| "step": 1145 |
| }, |
| { |
| "epoch": 0.9020070838252656, |
| "grad_norm": 0.9395787949515245, |
| "learning_rate": 1.2929420319468254e-07, |
| "loss": 0.1135, |
| "step": 1146 |
| }, |
| { |
| "epoch": 0.9027941755214482, |
| "grad_norm": 0.9442679409050593, |
| "learning_rate": 1.272350227796626e-07, |
| "loss": 0.1215, |
| "step": 1147 |
| }, |
| { |
| "epoch": 0.9035812672176309, |
| "grad_norm": 0.9549776744266577, |
| "learning_rate": 1.2519194361517468e-07, |
| "loss": 0.1065, |
| "step": 1148 |
| }, |
| { |
| "epoch": 0.9043683589138135, |
| "grad_norm": 0.867426121426782, |
| "learning_rate": 1.231649795652684e-07, |
| "loss": 0.0945, |
| "step": 1149 |
| }, |
| { |
| "epoch": 0.905155450609996, |
| "grad_norm": 0.9448412697453261, |
| "learning_rate": 1.2115414438463646e-07, |
| "loss": 0.1101, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.9059425423061787, |
| "grad_norm": 0.9917169805460462, |
| "learning_rate": 1.1915945171852572e-07, |
| "loss": 0.1163, |
| "step": 1151 |
| }, |
| { |
| "epoch": 0.9067296340023613, |
| "grad_norm": 0.943816804931634, |
| "learning_rate": 1.171809151026404e-07, |
| "loss": 0.1079, |
| "step": 1152 |
| }, |
| { |
| "epoch": 0.9075167256985439, |
| "grad_norm": 0.9275143855618218, |
| "learning_rate": 1.1521854796305243e-07, |
| "loss": 0.1075, |
| "step": 1153 |
| }, |
| { |
| "epoch": 0.9083038173947264, |
| "grad_norm": 1.0135210788160407, |
| "learning_rate": 1.1327236361611066e-07, |
| "loss": 0.1267, |
| "step": 1154 |
| }, |
| { |
| "epoch": 0.9090909090909091, |
| "grad_norm": 0.918433915733085, |
| "learning_rate": 1.1134237526834901e-07, |
| "loss": 0.1134, |
| "step": 1155 |
| }, |
| { |
| "epoch": 0.9098780007870917, |
| "grad_norm": 1.0485642131970048, |
| "learning_rate": 1.0942859601639793e-07, |
| "loss": 0.1181, |
| "step": 1156 |
| }, |
| { |
| "epoch": 0.9106650924832743, |
| "grad_norm": 0.9666449795771067, |
| "learning_rate": 1.0753103884689503e-07, |
| "loss": 0.1142, |
| "step": 1157 |
| }, |
| { |
| "epoch": 0.911452184179457, |
| "grad_norm": 0.9155262976479461, |
| "learning_rate": 1.0564971663639761e-07, |
| "loss": 0.1079, |
| "step": 1158 |
| }, |
| { |
| "epoch": 0.9122392758756395, |
| "grad_norm": 0.9136247354670238, |
| "learning_rate": 1.0378464215129419e-07, |
| "loss": 0.1131, |
| "step": 1159 |
| }, |
| { |
| "epoch": 0.9130263675718221, |
| "grad_norm": 0.9434235785738653, |
| "learning_rate": 1.0193582804771868e-07, |
| "loss": 0.1126, |
| "step": 1160 |
| }, |
| { |
| "epoch": 0.9138134592680047, |
| "grad_norm": 0.9320439540121583, |
| "learning_rate": 1.0010328687146464e-07, |
| "loss": 0.1128, |
| "step": 1161 |
| }, |
| { |
| "epoch": 0.9146005509641874, |
| "grad_norm": 0.9883313764295104, |
| "learning_rate": 9.828703105789983e-08, |
| "loss": 0.1189, |
| "step": 1162 |
| }, |
| { |
| "epoch": 0.91538764266037, |
| "grad_norm": 0.962978283773575, |
| "learning_rate": 9.648707293188092e-08, |
| "loss": 0.1181, |
| "step": 1163 |
| }, |
| { |
| "epoch": 0.9161747343565525, |
| "grad_norm": 0.9154295867986278, |
| "learning_rate": 9.470342470767197e-08, |
| "loss": 0.1077, |
| "step": 1164 |
| }, |
| { |
| "epoch": 0.9169618260527351, |
| "grad_norm": 0.9190343113758656, |
| "learning_rate": 9.293609848885971e-08, |
| "loss": 0.1101, |
| "step": 1165 |
| }, |
| { |
| "epoch": 0.9177489177489178, |
| "grad_norm": 0.8935174011429071, |
| "learning_rate": 9.118510626827198e-08, |
| "loss": 0.1112, |
| "step": 1166 |
| }, |
| { |
| "epoch": 0.9185360094451004, |
| "grad_norm": 0.9304289493526803, |
| "learning_rate": 8.945045992789669e-08, |
| "loss": 0.1037, |
| "step": 1167 |
| }, |
| { |
| "epoch": 0.9193231011412829, |
| "grad_norm": 0.9708158046423621, |
| "learning_rate": 8.773217123880074e-08, |
| "loss": 0.1255, |
| "step": 1168 |
| }, |
| { |
| "epoch": 0.9201101928374655, |
| "grad_norm": 0.95525649253936, |
| "learning_rate": 8.603025186105064e-08, |
| "loss": 0.1119, |
| "step": 1169 |
| }, |
| { |
| "epoch": 0.9208972845336482, |
| "grad_norm": 1.0094573892269945, |
| "learning_rate": 8.434471334363204e-08, |
| "loss": 0.1201, |
| "step": 1170 |
| }, |
| { |
| "epoch": 0.9216843762298308, |
| "grad_norm": 0.9367157782024292, |
| "learning_rate": 8.267556712437342e-08, |
| "loss": 0.1057, |
| "step": 1171 |
| }, |
| { |
| "epoch": 0.9224714679260134, |
| "grad_norm": 0.9747763894177717, |
| "learning_rate": 8.102282452986693e-08, |
| "loss": 0.1098, |
| "step": 1172 |
| }, |
| { |
| "epoch": 0.9232585596221959, |
| "grad_norm": 0.9372662016679384, |
| "learning_rate": 7.938649677539268e-08, |
| "loss": 0.1081, |
| "step": 1173 |
| }, |
| { |
| "epoch": 0.9240456513183786, |
| "grad_norm": 0.9887803764792047, |
| "learning_rate": 7.77665949648404e-08, |
| "loss": 0.1199, |
| "step": 1174 |
| }, |
| { |
| "epoch": 0.9248327430145612, |
| "grad_norm": 0.948252366615409, |
| "learning_rate": 7.616313009063791e-08, |
| "loss": 0.1064, |
| "step": 1175 |
| }, |
| { |
| "epoch": 0.9256198347107438, |
| "grad_norm": 0.948272373380358, |
| "learning_rate": 7.457611303367196e-08, |
| "loss": 0.1153, |
| "step": 1176 |
| }, |
| { |
| "epoch": 0.9264069264069265, |
| "grad_norm": 0.951666041729817, |
| "learning_rate": 7.300555456321884e-08, |
| "loss": 0.1175, |
| "step": 1177 |
| }, |
| { |
| "epoch": 0.927194018103109, |
| "grad_norm": 0.982566934759444, |
| "learning_rate": 7.145146533686725e-08, |
| "loss": 0.1164, |
| "step": 1178 |
| }, |
| { |
| "epoch": 0.9279811097992916, |
| "grad_norm": 0.9553483812186222, |
| "learning_rate": 6.991385590044947e-08, |
| "loss": 0.1169, |
| "step": 1179 |
| }, |
| { |
| "epoch": 0.9287682014954742, |
| "grad_norm": 0.9724385774768447, |
| "learning_rate": 6.839273668796747e-08, |
| "loss": 0.1078, |
| "step": 1180 |
| }, |
| { |
| "epoch": 0.9295552931916569, |
| "grad_norm": 0.9324047746069145, |
| "learning_rate": 6.688811802152279e-08, |
| "loss": 0.1162, |
| "step": 1181 |
| }, |
| { |
| "epoch": 0.9303423848878394, |
| "grad_norm": 0.9711188611376046, |
| "learning_rate": 6.540001011124703e-08, |
| "loss": 0.1089, |
| "step": 1182 |
| }, |
| { |
| "epoch": 0.931129476584022, |
| "grad_norm": 1.0007682860058293, |
| "learning_rate": 6.392842305523172e-08, |
| "loss": 0.1225, |
| "step": 1183 |
| }, |
| { |
| "epoch": 0.9319165682802046, |
| "grad_norm": 0.9074164360304593, |
| "learning_rate": 6.247336683946031e-08, |
| "loss": 0.1086, |
| "step": 1184 |
| }, |
| { |
| "epoch": 0.9327036599763873, |
| "grad_norm": 0.9132051814101239, |
| "learning_rate": 6.103485133774039e-08, |
| "loss": 0.1168, |
| "step": 1185 |
| }, |
| { |
| "epoch": 0.9334907516725699, |
| "grad_norm": 0.9362633318018305, |
| "learning_rate": 5.961288631163687e-08, |
| "loss": 0.1162, |
| "step": 1186 |
| }, |
| { |
| "epoch": 0.9342778433687524, |
| "grad_norm": 1.0037784789548483, |
| "learning_rate": 5.820748141040444e-08, |
| "loss": 0.1246, |
| "step": 1187 |
| }, |
| { |
| "epoch": 0.935064935064935, |
| "grad_norm": 0.9109713869964553, |
| "learning_rate": 5.681864617092414e-08, |
| "loss": 0.1062, |
| "step": 1188 |
| }, |
| { |
| "epoch": 0.9358520267611177, |
| "grad_norm": 0.9353212505070928, |
| "learning_rate": 5.544639001763719e-08, |
| "loss": 0.1116, |
| "step": 1189 |
| }, |
| { |
| "epoch": 0.9366391184573003, |
| "grad_norm": 0.9763975248080838, |
| "learning_rate": 5.4090722262481463e-08, |
| "loss": 0.1183, |
| "step": 1190 |
| }, |
| { |
| "epoch": 0.9374262101534829, |
| "grad_norm": 0.9389308712439575, |
| "learning_rate": 5.2751652104828245e-08, |
| "loss": 0.1125, |
| "step": 1191 |
| }, |
| { |
| "epoch": 0.9382133018496654, |
| "grad_norm": 0.9219409357377748, |
| "learning_rate": 5.142918863141999e-08, |
| "loss": 0.1045, |
| "step": 1192 |
| }, |
| { |
| "epoch": 0.9390003935458481, |
| "grad_norm": 0.9112274236581333, |
| "learning_rate": 5.012334081630821e-08, |
| "loss": 0.1154, |
| "step": 1193 |
| }, |
| { |
| "epoch": 0.9397874852420307, |
| "grad_norm": 1.0413638466805462, |
| "learning_rate": 4.8834117520793754e-08, |
| "loss": 0.1235, |
| "step": 1194 |
| }, |
| { |
| "epoch": 0.9405745769382133, |
| "grad_norm": 0.929473983088696, |
| "learning_rate": 4.756152749336468e-08, |
| "loss": 0.1216, |
| "step": 1195 |
| }, |
| { |
| "epoch": 0.9413616686343959, |
| "grad_norm": 0.9499785152690334, |
| "learning_rate": 4.6305579369638474e-08, |
| "loss": 0.119, |
| "step": 1196 |
| }, |
| { |
| "epoch": 0.9421487603305785, |
| "grad_norm": 0.9512339302739883, |
| "learning_rate": 4.506628167230326e-08, |
| "loss": 0.1128, |
| "step": 1197 |
| }, |
| { |
| "epoch": 0.9429358520267611, |
| "grad_norm": 0.9789643064479855, |
| "learning_rate": 4.384364281105974e-08, |
| "loss": 0.1156, |
| "step": 1198 |
| }, |
| { |
| "epoch": 0.9437229437229437, |
| "grad_norm": 0.9338458352272411, |
| "learning_rate": 4.2637671082563225e-08, |
| "loss": 0.1097, |
| "step": 1199 |
| }, |
| { |
| "epoch": 0.9445100354191264, |
| "grad_norm": 0.9477658752462017, |
| "learning_rate": 4.144837467036922e-08, |
| "loss": 0.1062, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.9452971271153089, |
| "grad_norm": 0.9108658408708349, |
| "learning_rate": 4.0275761644876785e-08, |
| "loss": 0.113, |
| "step": 1201 |
| }, |
| { |
| "epoch": 0.9460842188114915, |
| "grad_norm": 0.9670323890311822, |
| "learning_rate": 3.911983996327251e-08, |
| "loss": 0.1159, |
| "step": 1202 |
| }, |
| { |
| "epoch": 0.9468713105076741, |
| "grad_norm": 0.9508320183305409, |
| "learning_rate": 3.798061746947995e-08, |
| "loss": 0.1183, |
| "step": 1203 |
| }, |
| { |
| "epoch": 0.9476584022038568, |
| "grad_norm": 0.9446280692518585, |
| "learning_rate": 3.6858101894102774e-08, |
| "loss": 0.1039, |
| "step": 1204 |
| }, |
| { |
| "epoch": 0.9484454939000394, |
| "grad_norm": 0.8915603321077954, |
| "learning_rate": 3.575230085437448e-08, |
| "loss": 0.1131, |
| "step": 1205 |
| }, |
| { |
| "epoch": 0.9492325855962219, |
| "grad_norm": 1.0049928450920236, |
| "learning_rate": 3.466322185410542e-08, |
| "loss": 0.1075, |
| "step": 1206 |
| }, |
| { |
| "epoch": 0.9500196772924046, |
| "grad_norm": 0.9377285133019151, |
| "learning_rate": 3.3590872283633944e-08, |
| "loss": 0.1047, |
| "step": 1207 |
| }, |
| { |
| "epoch": 0.9508067689885872, |
| "grad_norm": 1.0283573568735918, |
| "learning_rate": 3.253525941977309e-08, |
| "loss": 0.1207, |
| "step": 1208 |
| }, |
| { |
| "epoch": 0.9515938606847698, |
| "grad_norm": 0.8716127646526632, |
| "learning_rate": 3.1496390425764246e-08, |
| "loss": 0.1034, |
| "step": 1209 |
| }, |
| { |
| "epoch": 0.9523809523809523, |
| "grad_norm": 0.9413176172911034, |
| "learning_rate": 3.047427235122663e-08, |
| "loss": 0.1094, |
| "step": 1210 |
| }, |
| { |
| "epoch": 0.953168044077135, |
| "grad_norm": 0.9235158713031617, |
| "learning_rate": 2.9468912132110117e-08, |
| "loss": 0.1112, |
| "step": 1211 |
| }, |
| { |
| "epoch": 0.9539551357733176, |
| "grad_norm": 1.0172961607730988, |
| "learning_rate": 2.8480316590648315e-08, |
| "loss": 0.1174, |
| "step": 1212 |
| }, |
| { |
| "epoch": 0.9547422274695002, |
| "grad_norm": 0.9633734296197954, |
| "learning_rate": 2.750849243531223e-08, |
| "loss": 0.119, |
| "step": 1213 |
| }, |
| { |
| "epoch": 0.9555293191656828, |
| "grad_norm": 0.8730009976362983, |
| "learning_rate": 2.655344626076417e-08, |
| "loss": 0.1012, |
| "step": 1214 |
| }, |
| { |
| "epoch": 0.9563164108618654, |
| "grad_norm": 0.9472387317654097, |
| "learning_rate": 2.5615184547813364e-08, |
| "loss": 0.1169, |
| "step": 1215 |
| }, |
| { |
| "epoch": 0.957103502558048, |
| "grad_norm": 1.009282211648514, |
| "learning_rate": 2.4693713663372643e-08, |
| "loss": 0.1193, |
| "step": 1216 |
| }, |
| { |
| "epoch": 0.9578905942542306, |
| "grad_norm": 0.9320555090575322, |
| "learning_rate": 2.378903986041403e-08, |
| "loss": 0.1134, |
| "step": 1217 |
| }, |
| { |
| "epoch": 0.9586776859504132, |
| "grad_norm": 0.9964824551488268, |
| "learning_rate": 2.2901169277927126e-08, |
| "loss": 0.123, |
| "step": 1218 |
| }, |
| { |
| "epoch": 0.9594647776465959, |
| "grad_norm": 0.9252548740156445, |
| "learning_rate": 2.2030107940877733e-08, |
| "loss": 0.1145, |
| "step": 1219 |
| }, |
| { |
| "epoch": 0.9602518693427784, |
| "grad_norm": 1.0057993388132023, |
| "learning_rate": 2.117586176016512e-08, |
| "loss": 0.1246, |
| "step": 1220 |
| }, |
| { |
| "epoch": 0.961038961038961, |
| "grad_norm": 0.953596307625609, |
| "learning_rate": 2.0338436532584826e-08, |
| "loss": 0.1113, |
| "step": 1221 |
| }, |
| { |
| "epoch": 0.9618260527351437, |
| "grad_norm": 0.8951377954636907, |
| "learning_rate": 1.9517837940786767e-08, |
| "loss": 0.1033, |
| "step": 1222 |
| }, |
| { |
| "epoch": 0.9626131444313263, |
| "grad_norm": 1.0221216536124687, |
| "learning_rate": 1.8714071553238012e-08, |
| "loss": 0.125, |
| "step": 1223 |
| }, |
| { |
| "epoch": 0.9634002361275088, |
| "grad_norm": 0.9387870706305922, |
| "learning_rate": 1.7927142824184784e-08, |
| "loss": 0.1144, |
| "step": 1224 |
| }, |
| { |
| "epoch": 0.9641873278236914, |
| "grad_norm": 0.9321486745496109, |
| "learning_rate": 1.7157057093614704e-08, |
| "loss": 0.1094, |
| "step": 1225 |
| }, |
| { |
| "epoch": 0.9649744195198741, |
| "grad_norm": 1.0128840378759991, |
| "learning_rate": 1.6403819587221814e-08, |
| "loss": 0.1217, |
| "step": 1226 |
| }, |
| { |
| "epoch": 0.9657615112160567, |
| "grad_norm": 0.924640460993744, |
| "learning_rate": 1.5667435416370226e-08, |
| "loss": 0.1115, |
| "step": 1227 |
| }, |
| { |
| "epoch": 0.9665486029122393, |
| "grad_norm": 0.9932790569946806, |
| "learning_rate": 1.494790957805997e-08, |
| "loss": 0.115, |
| "step": 1228 |
| }, |
| { |
| "epoch": 0.9673356946084218, |
| "grad_norm": 0.9381951620042324, |
| "learning_rate": 1.4245246954892323e-08, |
| "loss": 0.1096, |
| "step": 1229 |
| }, |
| { |
| "epoch": 0.9681227863046045, |
| "grad_norm": 0.9070754098607924, |
| "learning_rate": 1.3559452315037025e-08, |
| "loss": 0.107, |
| "step": 1230 |
| }, |
| { |
| "epoch": 0.9689098780007871, |
| "grad_norm": 0.9138545833001099, |
| "learning_rate": 1.2890530312200944e-08, |
| "loss": 0.1027, |
| "step": 1231 |
| }, |
| { |
| "epoch": 0.9696969696969697, |
| "grad_norm": 0.959458803400461, |
| "learning_rate": 1.2238485485594753e-08, |
| "loss": 0.1163, |
| "step": 1232 |
| }, |
| { |
| "epoch": 0.9704840613931524, |
| "grad_norm": 0.9577614004721761, |
| "learning_rate": 1.160332225990296e-08, |
| "loss": 0.1148, |
| "step": 1233 |
| }, |
| { |
| "epoch": 0.9712711530893349, |
| "grad_norm": 0.9713054396060389, |
| "learning_rate": 1.0985044945254763e-08, |
| "loss": 0.1249, |
| "step": 1234 |
| }, |
| { |
| "epoch": 0.9720582447855175, |
| "grad_norm": 0.9727071823355634, |
| "learning_rate": 1.0383657737192964e-08, |
| "loss": 0.1188, |
| "step": 1235 |
| }, |
| { |
| "epoch": 0.9728453364817001, |
| "grad_norm": 0.9525541993565831, |
| "learning_rate": 9.79916471664677e-09, |
| "loss": 0.1129, |
| "step": 1236 |
| }, |
| { |
| "epoch": 0.9736324281778828, |
| "grad_norm": 0.9792192000262094, |
| "learning_rate": 9.231569849904309e-09, |
| "loss": 0.124, |
| "step": 1237 |
| }, |
| { |
| "epoch": 0.9744195198740653, |
| "grad_norm": 0.9711922853486228, |
| "learning_rate": 8.680876988584607e-09, |
| "loss": 0.1114, |
| "step": 1238 |
| }, |
| { |
| "epoch": 0.9752066115702479, |
| "grad_norm": 0.9623623423156289, |
| "learning_rate": 8.147089869612045e-09, |
| "loss": 0.1064, |
| "step": 1239 |
| }, |
| { |
| "epoch": 0.9759937032664305, |
| "grad_norm": 0.9583105118384698, |
| "learning_rate": 7.630212115191381e-09, |
| "loss": 0.1128, |
| "step": 1240 |
| }, |
| { |
| "epoch": 0.9767807949626132, |
| "grad_norm": 0.958416495157565, |
| "learning_rate": 7.130247232782217e-09, |
| "loss": 0.1124, |
| "step": 1241 |
| }, |
| { |
| "epoch": 0.9775678866587958, |
| "grad_norm": 0.9739671638616083, |
| "learning_rate": 6.647198615076789e-09, |
| "loss": 0.1239, |
| "step": 1242 |
| }, |
| { |
| "epoch": 0.9783549783549783, |
| "grad_norm": 0.9831503405039121, |
| "learning_rate": 6.181069539974716e-09, |
| "loss": 0.1152, |
| "step": 1243 |
| }, |
| { |
| "epoch": 0.9791420700511609, |
| "grad_norm": 0.9837400272891019, |
| "learning_rate": 5.7318631705630126e-09, |
| "loss": 0.1196, |
| "step": 1244 |
| }, |
| { |
| "epoch": 0.9799291617473436, |
| "grad_norm": 0.9663876709057511, |
| "learning_rate": 5.299582555093052e-09, |
| "loss": 0.1174, |
| "step": 1245 |
| }, |
| { |
| "epoch": 0.9807162534435262, |
| "grad_norm": 0.9077428753737183, |
| "learning_rate": 4.884230626960307e-09, |
| "loss": 0.1046, |
| "step": 1246 |
| }, |
| { |
| "epoch": 0.9815033451397088, |
| "grad_norm": 0.9624717672564354, |
| "learning_rate": 4.485810204684638e-09, |
| "loss": 0.1143, |
| "step": 1247 |
| }, |
| { |
| "epoch": 0.9822904368358913, |
| "grad_norm": 0.9768731662603329, |
| "learning_rate": 4.104323991891424e-09, |
| "loss": 0.1111, |
| "step": 1248 |
| }, |
| { |
| "epoch": 0.983077528532074, |
| "grad_norm": 0.9918364976987204, |
| "learning_rate": 3.739774577292688e-09, |
| "loss": 0.1146, |
| "step": 1249 |
| }, |
| { |
| "epoch": 0.9838646202282566, |
| "grad_norm": 0.9235361594154657, |
| "learning_rate": 3.392164434669609e-09, |
| "loss": 0.115, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.9846517119244392, |
| "grad_norm": 0.9407569711042593, |
| "learning_rate": 3.0614959228558728e-09, |
| "loss": 0.1048, |
| "step": 1251 |
| }, |
| { |
| "epoch": 0.9854388036206218, |
| "grad_norm": 0.9503971642912823, |
| "learning_rate": 2.7477712857215676e-09, |
| "loss": 0.1118, |
| "step": 1252 |
| }, |
| { |
| "epoch": 0.9862258953168044, |
| "grad_norm": 0.9423931683522886, |
| "learning_rate": 2.450992652157924e-09, |
| "loss": 0.1052, |
| "step": 1253 |
| }, |
| { |
| "epoch": 0.987012987012987, |
| "grad_norm": 0.9692730394836648, |
| "learning_rate": 2.1711620360634344e-09, |
| "loss": 0.1134, |
| "step": 1254 |
| }, |
| { |
| "epoch": 0.9878000787091696, |
| "grad_norm": 0.8578557269739953, |
| "learning_rate": 1.9082813363294205e-09, |
| "loss": 0.1062, |
| "step": 1255 |
| }, |
| { |
| "epoch": 0.9885871704053523, |
| "grad_norm": 0.9417059944579995, |
| "learning_rate": 1.662352336827544e-09, |
| "loss": 0.1198, |
| "step": 1256 |
| }, |
| { |
| "epoch": 0.9893742621015348, |
| "grad_norm": 1.0199292836158018, |
| "learning_rate": 1.4333767063973159e-09, |
| "loss": 0.1085, |
| "step": 1257 |
| }, |
| { |
| "epoch": 0.9901613537977174, |
| "grad_norm": 0.9283942300248242, |
| "learning_rate": 1.221355998835272e-09, |
| "loss": 0.1069, |
| "step": 1258 |
| }, |
| { |
| "epoch": 0.9909484454939, |
| "grad_norm": 0.9097063653552931, |
| "learning_rate": 1.0262916528841483e-09, |
| "loss": 0.1074, |
| "step": 1259 |
| }, |
| { |
| "epoch": 0.9917355371900827, |
| "grad_norm": 0.9806238353833303, |
| "learning_rate": 8.481849922237217e-10, |
| "loss": 0.1136, |
| "step": 1260 |
| }, |
| { |
| "epoch": 0.9925226288862653, |
| "grad_norm": 0.9354892517698297, |
| "learning_rate": 6.870372254602631e-10, |
| "loss": 0.1048, |
| "step": 1261 |
| }, |
| { |
| "epoch": 0.9933097205824478, |
| "grad_norm": 0.9045069809298675, |
| "learning_rate": 5.428494461201527e-10, |
| "loss": 0.109, |
| "step": 1262 |
| }, |
| { |
| "epoch": 0.9940968122786304, |
| "grad_norm": 0.9462573748322036, |
| "learning_rate": 4.156226326415547e-10, |
| "loss": 0.1156, |
| "step": 1263 |
| }, |
| { |
| "epoch": 0.9948839039748131, |
| "grad_norm": 0.9123815541723352, |
| "learning_rate": 3.0535764836747696e-10, |
| "loss": 0.1078, |
| "step": 1264 |
| }, |
| { |
| "epoch": 0.9956709956709957, |
| "grad_norm": 0.960544516493779, |
| "learning_rate": 2.1205524154105372e-10, |
| "loss": 0.1088, |
| "step": 1265 |
| }, |
| { |
| "epoch": 0.9964580873671782, |
| "grad_norm": 0.8887242195384208, |
| "learning_rate": 1.357160452988837e-10, |
| "loss": 0.1031, |
| "step": 1266 |
| }, |
| { |
| "epoch": 0.9972451790633609, |
| "grad_norm": 0.9502744942086568, |
| "learning_rate": 7.63405776685322e-11, |
| "loss": 0.1124, |
| "step": 1267 |
| }, |
| { |
| "epoch": 0.9980322707595435, |
| "grad_norm": 0.8901953170723566, |
| "learning_rate": 3.3929241563535056e-11, |
| "loss": 0.1001, |
| "step": 1268 |
| }, |
| { |
| "epoch": 0.9988193624557261, |
| "grad_norm": 0.9602428624879035, |
| "learning_rate": 8.482324780900718e-12, |
| "loss": 0.1116, |
| "step": 1269 |
| }, |
| { |
| "epoch": 0.9996064541519087, |
| "grad_norm": 0.9779040145681426, |
| "learning_rate": 0.0, |
| "loss": 0.119, |
| "step": 1270 |
| }, |
| { |
| "epoch": 0.9996064541519087, |
| "step": 1270, |
| "total_flos": 223330201436160.0, |
| "train_loss": 0.14531472616308316, |
| "train_runtime": 7100.3282, |
| "train_samples_per_second": 11.452, |
| "train_steps_per_second": 0.179 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 1270, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 1, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": false, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 223330201436160.0, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
