| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.5708840656431975, | |
| "eval_steps": 337, | |
| "global_step": 1348, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00042350449973530967, | |
| "grad_norm": 0.027733758091926575, | |
| "learning_rate": 2e-05, | |
| "loss": 10.3743, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00042350449973530967, | |
| "eval_loss": 10.376607894897461, | |
| "eval_runtime": 3.5039, | |
| "eval_samples_per_second": 283.969, | |
| "eval_steps_per_second": 142.127, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0008470089994706193, | |
| "grad_norm": 0.02669823355972767, | |
| "learning_rate": 4e-05, | |
| "loss": 10.374, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.001270513499205929, | |
| "grad_norm": 0.021611209958791733, | |
| "learning_rate": 6e-05, | |
| "loss": 10.3801, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0016940179989412387, | |
| "grad_norm": 0.027095356956124306, | |
| "learning_rate": 8e-05, | |
| "loss": 10.3786, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0021175224986765486, | |
| "grad_norm": 0.030345361679792404, | |
| "learning_rate": 0.0001, | |
| "loss": 10.378, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.002541026998411858, | |
| "grad_norm": 0.025746231898665428, | |
| "learning_rate": 0.00012, | |
| "loss": 10.3767, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.002964531498147168, | |
| "grad_norm": 0.026296626776456833, | |
| "learning_rate": 0.00014, | |
| "loss": 10.375, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0033880359978824773, | |
| "grad_norm": 0.026994528248906136, | |
| "learning_rate": 0.00016, | |
| "loss": 10.3775, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0038115404976177872, | |
| "grad_norm": 0.02642114832997322, | |
| "learning_rate": 0.00018, | |
| "loss": 10.3785, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.004235044997353097, | |
| "grad_norm": 0.03136637434363365, | |
| "learning_rate": 0.0002, | |
| "loss": 10.3753, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.004658549497088407, | |
| "grad_norm": 0.022933412343263626, | |
| "learning_rate": 0.00019999972435042745, | |
| "loss": 10.3753, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.005082053996823716, | |
| "grad_norm": 0.02334180846810341, | |
| "learning_rate": 0.0001999988974032295, | |
| "loss": 10.3753, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.005505558496559026, | |
| "grad_norm": 0.03419239819049835, | |
| "learning_rate": 0.00019999751916296505, | |
| "loss": 10.3767, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.005929062996294336, | |
| "grad_norm": 0.022363845258951187, | |
| "learning_rate": 0.0001999955896372324, | |
| "loss": 10.3773, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.006352567496029645, | |
| "grad_norm": 0.025751987472176552, | |
| "learning_rate": 0.0001999931088366689, | |
| "loss": 10.3788, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.006776071995764955, | |
| "grad_norm": 0.02451767958700657, | |
| "learning_rate": 0.00019999007677495127, | |
| "loss": 10.3781, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.007199576495500265, | |
| "grad_norm": 0.023951657116413116, | |
| "learning_rate": 0.00019998649346879524, | |
| "loss": 10.3746, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0076230809952355745, | |
| "grad_norm": 0.02496548369526863, | |
| "learning_rate": 0.0001999823589379555, | |
| "loss": 10.3739, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.008046585494970884, | |
| "grad_norm": 0.02257522940635681, | |
| "learning_rate": 0.0001999776732052257, | |
| "loss": 10.3755, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.008470089994706194, | |
| "grad_norm": 0.027529660612344742, | |
| "learning_rate": 0.00019997243629643827, | |
| "loss": 10.3753, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.008893594494441503, | |
| "grad_norm": 0.025733161717653275, | |
| "learning_rate": 0.0001999666482404642, | |
| "loss": 10.376, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.009317098994176813, | |
| "grad_norm": 0.028723513707518578, | |
| "learning_rate": 0.00019996030906921302, | |
| "loss": 10.373, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.009740603493912123, | |
| "grad_norm": 0.03213539347052574, | |
| "learning_rate": 0.00019995341881763254, | |
| "loss": 10.3749, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.010164107993647432, | |
| "grad_norm": 0.033838726580142975, | |
| "learning_rate": 0.0001999459775237086, | |
| "loss": 10.374, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.010587612493382742, | |
| "grad_norm": 0.029485682025551796, | |
| "learning_rate": 0.00019993798522846508, | |
| "loss": 10.3769, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.011011116993118053, | |
| "grad_norm": 0.03559406101703644, | |
| "learning_rate": 0.00019992944197596337, | |
| "loss": 10.3748, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.011434621492853361, | |
| "grad_norm": 0.033679552376270294, | |
| "learning_rate": 0.00019992034781330235, | |
| "loss": 10.3733, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.011858125992588672, | |
| "grad_norm": 0.032387569546699524, | |
| "learning_rate": 0.00019991070279061808, | |
| "loss": 10.3711, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.01228163049232398, | |
| "grad_norm": 0.03400762379169464, | |
| "learning_rate": 0.0001999005069610835, | |
| "loss": 10.3726, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.01270513499205929, | |
| "grad_norm": 0.030853325501084328, | |
| "learning_rate": 0.0001998897603809081, | |
| "loss": 10.3756, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0131286394917946, | |
| "grad_norm": 0.039369914680719376, | |
| "learning_rate": 0.00019987846310933768, | |
| "loss": 10.373, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.01355214399152991, | |
| "grad_norm": 0.04837853088974953, | |
| "learning_rate": 0.00019986661520865405, | |
| "loss": 10.3751, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.01397564849126522, | |
| "grad_norm": 0.045920245349407196, | |
| "learning_rate": 0.00019985421674417452, | |
| "loss": 10.3693, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.01439915299100053, | |
| "grad_norm": 0.04273354262113571, | |
| "learning_rate": 0.00019984126778425178, | |
| "loss": 10.3702, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.014822657490735839, | |
| "grad_norm": 0.048800382763147354, | |
| "learning_rate": 0.0001998277684002733, | |
| "loss": 10.3745, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.015246161990471149, | |
| "grad_norm": 0.05085352063179016, | |
| "learning_rate": 0.00019981371866666109, | |
| "loss": 10.3745, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.015669666490206458, | |
| "grad_norm": 0.05351710319519043, | |
| "learning_rate": 0.0001997991186608712, | |
| "loss": 10.3718, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.016093170989941768, | |
| "grad_norm": 0.048667021095752716, | |
| "learning_rate": 0.0001997839684633933, | |
| "loss": 10.3713, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.016516675489677078, | |
| "grad_norm": 0.04797323793172836, | |
| "learning_rate": 0.0001997682681577504, | |
| "loss": 10.3696, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.01694017998941239, | |
| "grad_norm": 0.04910498112440109, | |
| "learning_rate": 0.00019975201783049805, | |
| "loss": 10.3753, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0173636844891477, | |
| "grad_norm": 0.04971903935074806, | |
| "learning_rate": 0.00019973521757122418, | |
| "loss": 10.3724, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.017787188988883006, | |
| "grad_norm": 0.06753654778003693, | |
| "learning_rate": 0.00019971786747254852, | |
| "loss": 10.3717, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.018210693488618316, | |
| "grad_norm": 0.04575066268444061, | |
| "learning_rate": 0.00019969996763012198, | |
| "loss": 10.3708, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.018634197988353626, | |
| "grad_norm": 0.0645640566945076, | |
| "learning_rate": 0.00019968151814262627, | |
| "loss": 10.37, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.019057702488088937, | |
| "grad_norm": 0.07102999091148376, | |
| "learning_rate": 0.00019966251911177323, | |
| "loss": 10.371, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.019481206987824247, | |
| "grad_norm": 0.07168291509151459, | |
| "learning_rate": 0.00019964297064230436, | |
| "loss": 10.3691, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.019904711487559554, | |
| "grad_norm": 0.06409385055303574, | |
| "learning_rate": 0.0001996228728419902, | |
| "loss": 10.3712, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.020328215987294864, | |
| "grad_norm": 0.06838654726743698, | |
| "learning_rate": 0.00019960222582162976, | |
| "loss": 10.3681, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.020751720487030174, | |
| "grad_norm": 0.10280771553516388, | |
| "learning_rate": 0.0001995810296950499, | |
| "loss": 10.3681, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.021175224986765485, | |
| "grad_norm": 0.08676479011774063, | |
| "learning_rate": 0.00019955928457910464, | |
| "loss": 10.3678, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.021598729486500795, | |
| "grad_norm": 0.0853012353181839, | |
| "learning_rate": 0.00019953699059367468, | |
| "loss": 10.3662, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.022022233986236105, | |
| "grad_norm": 0.09723170846700668, | |
| "learning_rate": 0.00019951414786166654, | |
| "loss": 10.3658, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.022445738485971412, | |
| "grad_norm": 0.08303584158420563, | |
| "learning_rate": 0.00019949075650901196, | |
| "loss": 10.3637, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.022869242985706723, | |
| "grad_norm": 0.09252041578292847, | |
| "learning_rate": 0.00019946681666466737, | |
| "loss": 10.3663, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.023292747485442033, | |
| "grad_norm": 0.07753727585077286, | |
| "learning_rate": 0.00019944232846061283, | |
| "loss": 10.363, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.023716251985177343, | |
| "grad_norm": 0.09255944937467575, | |
| "learning_rate": 0.00019941729203185165, | |
| "loss": 10.3598, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.024139756484912653, | |
| "grad_norm": 0.1016978919506073, | |
| "learning_rate": 0.0001993917075164095, | |
| "loss": 10.3632, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.02456326098464796, | |
| "grad_norm": 0.09657126665115356, | |
| "learning_rate": 0.00019936557505533344, | |
| "loss": 10.363, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.02498676548438327, | |
| "grad_norm": 0.08984223008155823, | |
| "learning_rate": 0.00019933889479269162, | |
| "loss": 10.366, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.02541026998411858, | |
| "grad_norm": 0.11775655299425125, | |
| "learning_rate": 0.0001993116668755721, | |
| "loss": 10.3619, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.02583377448385389, | |
| "grad_norm": 0.10796835273504257, | |
| "learning_rate": 0.00019928389145408213, | |
| "loss": 10.3609, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0262572789835892, | |
| "grad_norm": 0.10575428605079651, | |
| "learning_rate": 0.00019925556868134736, | |
| "loss": 10.3571, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.026680783483324512, | |
| "grad_norm": 0.12180919945240021, | |
| "learning_rate": 0.000199226698713511, | |
| "loss": 10.3553, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.02710428798305982, | |
| "grad_norm": 0.10506236553192139, | |
| "learning_rate": 0.00019919728170973296, | |
| "loss": 10.3593, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.02752779248279513, | |
| "grad_norm": 0.09984668344259262, | |
| "learning_rate": 0.00019916731783218888, | |
| "loss": 10.3593, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.02795129698253044, | |
| "grad_norm": 0.10555399954319, | |
| "learning_rate": 0.00019913680724606945, | |
| "loss": 10.3591, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.02837480148226575, | |
| "grad_norm": 0.08873631060123444, | |
| "learning_rate": 0.00019910575011957918, | |
| "loss": 10.3568, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.02879830598200106, | |
| "grad_norm": 0.09549526870250702, | |
| "learning_rate": 0.00019907414662393574, | |
| "loss": 10.3574, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.029221810481736367, | |
| "grad_norm": 0.09750920534133911, | |
| "learning_rate": 0.000199041996933369, | |
| "loss": 10.3536, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.029645314981471677, | |
| "grad_norm": 0.10485312342643738, | |
| "learning_rate": 0.00019900930122511993, | |
| "loss": 10.3566, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.030068819481206988, | |
| "grad_norm": 0.09972581267356873, | |
| "learning_rate": 0.00019897605967943963, | |
| "loss": 10.3531, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.030492323980942298, | |
| "grad_norm": 0.08210822939872742, | |
| "learning_rate": 0.00019894227247958845, | |
| "loss": 10.3534, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.030915828480677608, | |
| "grad_norm": 0.08600655943155289, | |
| "learning_rate": 0.00019890793981183503, | |
| "loss": 10.356, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.031339332980412915, | |
| "grad_norm": 0.08058468252420425, | |
| "learning_rate": 0.00019887306186545497, | |
| "loss": 10.3549, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.03176283748014823, | |
| "grad_norm": 0.0659925639629364, | |
| "learning_rate": 0.00019883763883273012, | |
| "loss": 10.3507, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.032186341979883536, | |
| "grad_norm": 0.06881393492221832, | |
| "learning_rate": 0.0001988016709089474, | |
| "loss": 10.3529, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.03260984647961884, | |
| "grad_norm": 0.0784982293844223, | |
| "learning_rate": 0.00019876515829239763, | |
| "loss": 10.3528, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.033033350979354156, | |
| "grad_norm": 0.06941844522953033, | |
| "learning_rate": 0.00019872810118437456, | |
| "loss": 10.351, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.03345685547908946, | |
| "grad_norm": 0.06965084373950958, | |
| "learning_rate": 0.00019869049978917368, | |
| "loss": 10.3507, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.03388035997882478, | |
| "grad_norm": 0.0600489042699337, | |
| "learning_rate": 0.00019865235431409123, | |
| "loss": 10.3514, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.034303864478560084, | |
| "grad_norm": 0.06106571480631828, | |
| "learning_rate": 0.00019861366496942283, | |
| "loss": 10.3501, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.0347273689782954, | |
| "grad_norm": 0.05668988823890686, | |
| "learning_rate": 0.0001985744319684625, | |
| "loss": 10.3479, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.035150873478030704, | |
| "grad_norm": 0.05988716706633568, | |
| "learning_rate": 0.00019853465552750147, | |
| "loss": 10.3472, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.03557437797776601, | |
| "grad_norm": 0.047210004180669785, | |
| "learning_rate": 0.00019849433586582692, | |
| "loss": 10.3522, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.035997882477501325, | |
| "grad_norm": 0.04648837819695473, | |
| "learning_rate": 0.00019845347320572078, | |
| "loss": 10.3489, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.03642138697723663, | |
| "grad_norm": 0.057975709438323975, | |
| "learning_rate": 0.00019841206777245857, | |
| "loss": 10.3482, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.036844891476971946, | |
| "grad_norm": 0.06715747714042664, | |
| "learning_rate": 0.00019837011979430806, | |
| "loss": 10.3486, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.03726839597670725, | |
| "grad_norm": 0.05633699893951416, | |
| "learning_rate": 0.00019832762950252813, | |
| "loss": 10.3506, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.03769190047644256, | |
| "grad_norm": 0.04711679369211197, | |
| "learning_rate": 0.00019828459713136737, | |
| "loss": 10.349, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.03811540497617787, | |
| "grad_norm": 0.050088070333004, | |
| "learning_rate": 0.0001982410229180629, | |
| "loss": 10.3457, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.03853890947591318, | |
| "grad_norm": 0.0481443926692009, | |
| "learning_rate": 0.00019819690710283893, | |
| "loss": 10.3488, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.038962413975648494, | |
| "grad_norm": 0.04781080409884453, | |
| "learning_rate": 0.0001981522499289056, | |
| "loss": 10.3476, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.0393859184753838, | |
| "grad_norm": 0.04098181053996086, | |
| "learning_rate": 0.00019810705164245756, | |
| "loss": 10.3486, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.03980942297511911, | |
| "grad_norm": 0.050709549337625504, | |
| "learning_rate": 0.00019806131249267255, | |
| "loss": 10.3465, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.04023292747485442, | |
| "grad_norm": 0.04031967371702194, | |
| "learning_rate": 0.00019801503273171012, | |
| "loss": 10.3497, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.04065643197458973, | |
| "grad_norm": 0.029422029852867126, | |
| "learning_rate": 0.00019796821261471018, | |
| "loss": 10.3476, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.04107993647432504, | |
| "grad_norm": 0.04418569803237915, | |
| "learning_rate": 0.00019792085239979162, | |
| "loss": 10.3488, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.04150344097406035, | |
| "grad_norm": 0.05277906730771065, | |
| "learning_rate": 0.00019787295234805096, | |
| "loss": 10.3495, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.041926945473795656, | |
| "grad_norm": 0.03719155862927437, | |
| "learning_rate": 0.00019782451272356075, | |
| "loss": 10.3493, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.04235044997353097, | |
| "grad_norm": 0.042001668363809586, | |
| "learning_rate": 0.0001977755337933682, | |
| "loss": 10.3474, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.042773954473266276, | |
| "grad_norm": 0.045230720192193985, | |
| "learning_rate": 0.00019772601582749376, | |
| "loss": 10.3498, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.04319745897300159, | |
| "grad_norm": 0.03940007835626602, | |
| "learning_rate": 0.00019767595909892953, | |
| "loss": 10.3499, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.0436209634727369, | |
| "grad_norm": 0.044866979122161865, | |
| "learning_rate": 0.00019762536388363784, | |
| "loss": 10.3464, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.04404446797247221, | |
| "grad_norm": 0.039521895349025726, | |
| "learning_rate": 0.00019757423046054968, | |
| "loss": 10.3491, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.04446797247220752, | |
| "grad_norm": 0.04928427189588547, | |
| "learning_rate": 0.00019752255911156317, | |
| "loss": 10.345, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.044891476971942824, | |
| "grad_norm": 0.04378641024231911, | |
| "learning_rate": 0.00019747035012154202, | |
| "loss": 10.3488, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.04531498147167814, | |
| "grad_norm": 0.048980504274368286, | |
| "learning_rate": 0.00019741760377831396, | |
| "loss": 10.3468, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.045738485971413445, | |
| "grad_norm": 0.04159266874194145, | |
| "learning_rate": 0.00019736432037266912, | |
| "loss": 10.3478, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.04616199047114876, | |
| "grad_norm": 0.0287900660187006, | |
| "learning_rate": 0.00019731050019835842, | |
| "loss": 10.3497, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.046585494970884066, | |
| "grad_norm": 0.03839430958032608, | |
| "learning_rate": 0.00019725614355209204, | |
| "loss": 10.35, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.04700899947061937, | |
| "grad_norm": 0.04616628587245941, | |
| "learning_rate": 0.00019720125073353776, | |
| "loss": 10.3471, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.047432503970354686, | |
| "grad_norm": 0.05492490157485008, | |
| "learning_rate": 0.00019714582204531918, | |
| "loss": 10.3503, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.04785600847008999, | |
| "grad_norm": 0.037890441715717316, | |
| "learning_rate": 0.00019708985779301417, | |
| "loss": 10.3488, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.04827951296982531, | |
| "grad_norm": 0.036491066217422485, | |
| "learning_rate": 0.00019703335828515322, | |
| "loss": 10.3476, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.048703017469560614, | |
| "grad_norm": 0.03580768033862114, | |
| "learning_rate": 0.00019697632383321756, | |
| "loss": 10.3509, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.04912652196929592, | |
| "grad_norm": 0.04286257550120354, | |
| "learning_rate": 0.0001969187547516377, | |
| "loss": 10.3475, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.049550026469031234, | |
| "grad_norm": 0.06037011742591858, | |
| "learning_rate": 0.00019686065135779144, | |
| "loss": 10.3534, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.04997353096876654, | |
| "grad_norm": 0.05510607734322548, | |
| "learning_rate": 0.00019680201397200236, | |
| "loss": 10.3529, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.050397035468501855, | |
| "grad_norm": 0.04488476365804672, | |
| "learning_rate": 0.00019674284291753785, | |
| "loss": 10.3482, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.05082053996823716, | |
| "grad_norm": 0.02746366150677204, | |
| "learning_rate": 0.00019668313852060735, | |
| "loss": 10.3507, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.05124404446797247, | |
| "grad_norm": 0.031117988750338554, | |
| "learning_rate": 0.00019662290111036078, | |
| "loss": 10.3472, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.05166754896770778, | |
| "grad_norm": 0.04041313752532005, | |
| "learning_rate": 0.00019656213101888645, | |
| "loss": 10.3468, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.05209105346744309, | |
| "grad_norm": 0.04518342763185501, | |
| "learning_rate": 0.00019650082858120932, | |
| "loss": 10.35, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.0525145579671784, | |
| "grad_norm": 0.034027136862277985, | |
| "learning_rate": 0.00019643899413528926, | |
| "loss": 10.3474, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.05293806246691371, | |
| "grad_norm": 0.0336722694337368, | |
| "learning_rate": 0.000196376628022019, | |
| "loss": 10.347, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.053361566966649024, | |
| "grad_norm": 0.03731876611709595, | |
| "learning_rate": 0.00019631373058522238, | |
| "loss": 10.3484, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.05378507146638433, | |
| "grad_norm": 0.038337815552949905, | |
| "learning_rate": 0.00019625030217165245, | |
| "loss": 10.3493, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.05420857596611964, | |
| "grad_norm": 0.036029715090990067, | |
| "learning_rate": 0.00019618634313098952, | |
| "loss": 10.346, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.05463208046585495, | |
| "grad_norm": 0.031205767765641212, | |
| "learning_rate": 0.00019612185381583924, | |
| "loss": 10.3502, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.05505558496559026, | |
| "grad_norm": 0.04413217306137085, | |
| "learning_rate": 0.0001960568345817306, | |
| "loss": 10.3507, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.05547908946532557, | |
| "grad_norm": 0.03828402981162071, | |
| "learning_rate": 0.00019599128578711415, | |
| "loss": 10.3485, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.05590259396506088, | |
| "grad_norm": 0.03328114375472069, | |
| "learning_rate": 0.0001959252077933598, | |
| "loss": 10.3481, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.056326098464796186, | |
| "grad_norm": 0.04720017686486244, | |
| "learning_rate": 0.000195858600964755, | |
| "loss": 10.3468, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.0567496029645315, | |
| "grad_norm": 0.03394393250346184, | |
| "learning_rate": 0.00019579146566850252, | |
| "loss": 10.3457, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.057173107464266806, | |
| "grad_norm": 0.03747075796127319, | |
| "learning_rate": 0.0001957238022747188, | |
| "loss": 10.3488, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.05759661196400212, | |
| "grad_norm": 0.03510262444615364, | |
| "learning_rate": 0.00019565561115643152, | |
| "loss": 10.3504, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.05802011646373743, | |
| "grad_norm": 0.03729300945997238, | |
| "learning_rate": 0.00019558689268957767, | |
| "loss": 10.3464, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.058443620963472734, | |
| "grad_norm": 0.029604580253362656, | |
| "learning_rate": 0.00019551764725300166, | |
| "loss": 10.3438, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.05886712546320805, | |
| "grad_norm": 0.039334215223789215, | |
| "learning_rate": 0.0001954478752284529, | |
| "loss": 10.3472, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.059290629962943354, | |
| "grad_norm": 0.04949035122990608, | |
| "learning_rate": 0.00019537757700058403, | |
| "loss": 10.3476, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.05971413446267867, | |
| "grad_norm": 0.034930143505334854, | |
| "learning_rate": 0.00019530675295694857, | |
| "loss": 10.3475, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.060137638962413975, | |
| "grad_norm": 0.02766244113445282, | |
| "learning_rate": 0.00019523540348799885, | |
| "loss": 10.3457, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.06056114346214928, | |
| "grad_norm": 0.02754233032464981, | |
| "learning_rate": 0.0001951635289870839, | |
| "loss": 10.3471, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.060984647961884596, | |
| "grad_norm": 0.05756423994898796, | |
| "learning_rate": 0.00019509112985044717, | |
| "loss": 10.348, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.0614081524616199, | |
| "grad_norm": 0.03342543542385101, | |
| "learning_rate": 0.00019501820647722457, | |
| "loss": 10.349, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.061831656961355216, | |
| "grad_norm": 0.04082402214407921, | |
| "learning_rate": 0.00019494475926944195, | |
| "loss": 10.3486, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.06225516146109052, | |
| "grad_norm": 0.03864405304193497, | |
| "learning_rate": 0.00019487078863201322, | |
| "loss": 10.351, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.06267866596082583, | |
| "grad_norm": 0.028355760499835014, | |
| "learning_rate": 0.00019479629497273781, | |
| "loss": 10.3474, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.06310217046056114, | |
| "grad_norm": 0.03946223482489586, | |
| "learning_rate": 0.00019472127870229867, | |
| "loss": 10.349, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.06352567496029646, | |
| "grad_norm": 0.04293173551559448, | |
| "learning_rate": 0.00019464574023425984, | |
| "loss": 10.3508, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.06394917946003176, | |
| "grad_norm": 0.04612809792160988, | |
| "learning_rate": 0.0001945696799850642, | |
| "loss": 10.3473, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.06437268395976707, | |
| "grad_norm": 0.04514515772461891, | |
| "learning_rate": 0.00019449309837403137, | |
| "loss": 10.3484, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.06479618845950239, | |
| "grad_norm": 0.03168589621782303, | |
| "learning_rate": 0.00019441599582335498, | |
| "loss": 10.3465, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.06521969295923769, | |
| "grad_norm": 0.04755236580967903, | |
| "learning_rate": 0.00019433837275810082, | |
| "loss": 10.3474, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.065643197458973, | |
| "grad_norm": 0.031274329870939255, | |
| "learning_rate": 0.00019426022960620417, | |
| "loss": 10.3451, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.06606670195870831, | |
| "grad_norm": 0.036476653069257736, | |
| "learning_rate": 0.00019418156679846754, | |
| "loss": 10.3483, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.06649020645844363, | |
| "grad_norm": 0.0386991873383522, | |
| "learning_rate": 0.0001941023847685583, | |
| "loss": 10.3474, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.06691371095817893, | |
| "grad_norm": 0.034830257296562195, | |
| "learning_rate": 0.00019402268395300637, | |
| "loss": 10.3493, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.06733721545791424, | |
| "grad_norm": 0.03715137764811516, | |
| "learning_rate": 0.00019394246479120163, | |
| "loss": 10.3529, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.06776071995764955, | |
| "grad_norm": 0.03950640186667442, | |
| "learning_rate": 0.00019386172772539162, | |
| "loss": 10.3479, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.06818422445738485, | |
| "grad_norm": 0.04391263425350189, | |
| "learning_rate": 0.0001937804732006791, | |
| "loss": 10.3456, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.06860772895712017, | |
| "grad_norm": 0.047799207270145416, | |
| "learning_rate": 0.00019369870166501959, | |
| "loss": 10.3451, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.06903123345685548, | |
| "grad_norm": 0.031426433473825455, | |
| "learning_rate": 0.00019361641356921883, | |
| "loss": 10.3499, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.0694547379565908, | |
| "grad_norm": 0.0359007902443409, | |
| "learning_rate": 0.00019353360936693041, | |
| "loss": 10.3433, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.0698782424563261, | |
| "grad_norm": 0.02672567404806614, | |
| "learning_rate": 0.00019345028951465318, | |
| "loss": 10.343, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.07030174695606141, | |
| "grad_norm": 0.04336037486791611, | |
| "learning_rate": 0.0001933664544717288, | |
| "loss": 10.3488, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.07072525145579672, | |
| "grad_norm": 0.030480332672595978, | |
| "learning_rate": 0.0001932821047003391, | |
| "loss": 10.3464, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.07114875595553202, | |
| "grad_norm": 0.03520766645669937, | |
| "learning_rate": 0.00019319724066550373, | |
| "loss": 10.3475, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.07157226045526734, | |
| "grad_norm": 0.020646894350647926, | |
| "learning_rate": 0.0001931118628350773, | |
| "loss": 10.3476, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.07199576495500265, | |
| "grad_norm": 0.04156513512134552, | |
| "learning_rate": 0.00019302597167974707, | |
| "loss": 10.3485, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.07241926945473795, | |
| "grad_norm": 0.02938881516456604, | |
| "learning_rate": 0.0001929395676730303, | |
| "loss": 10.3464, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.07284277395447326, | |
| "grad_norm": 0.03371270000934601, | |
| "learning_rate": 0.00019285265129127151, | |
| "loss": 10.3443, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.07326627845420858, | |
| "grad_norm": 0.045955732464790344, | |
| "learning_rate": 0.00019276522301363996, | |
| "loss": 10.346, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.07368978295394389, | |
| "grad_norm": 0.022017156705260277, | |
| "learning_rate": 0.000192677283322127, | |
| "loss": 10.3461, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.07411328745367919, | |
| "grad_norm": 0.045463208109140396, | |
| "learning_rate": 0.0001925888327015434, | |
| "loss": 10.3462, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.0745367919534145, | |
| "grad_norm": 0.041146885603666306, | |
| "learning_rate": 0.00019249987163951667, | |
| "loss": 10.3453, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.07496029645314982, | |
| "grad_norm": 0.04077988117933273, | |
| "learning_rate": 0.0001924104006264884, | |
| "loss": 10.3472, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.07538380095288512, | |
| "grad_norm": 0.033624131232500076, | |
| "learning_rate": 0.00019232042015571152, | |
| "loss": 10.3493, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.07580730545262043, | |
| "grad_norm": 0.04149757698178291, | |
| "learning_rate": 0.00019222993072324758, | |
| "loss": 10.347, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.07623080995235575, | |
| "grad_norm": 0.04390670359134674, | |
| "learning_rate": 0.00019213893282796405, | |
| "loss": 10.3499, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.07665431445209106, | |
| "grad_norm": 0.04238109290599823, | |
| "learning_rate": 0.00019204742697153155, | |
| "loss": 10.3482, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.07707781895182636, | |
| "grad_norm": 0.0415191613137722, | |
| "learning_rate": 0.0001919554136584211, | |
| "loss": 10.3485, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.07750132345156167, | |
| "grad_norm": 0.04313662648200989, | |
| "learning_rate": 0.0001918628933959013, | |
| "loss": 10.3447, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.07792482795129699, | |
| "grad_norm": 0.0481775663793087, | |
| "learning_rate": 0.00019176986669403555, | |
| "loss": 10.3456, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.07834833245103229, | |
| "grad_norm": 0.031192272901535034, | |
| "learning_rate": 0.0001916763340656793, | |
| "loss": 10.3488, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.0787718369507676, | |
| "grad_norm": 0.04172395542263985, | |
| "learning_rate": 0.00019158229602647708, | |
| "loss": 10.3442, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.07919534145050292, | |
| "grad_norm": 0.03788716346025467, | |
| "learning_rate": 0.00019148775309485983, | |
| "loss": 10.3443, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.07961884595023822, | |
| "grad_norm": 0.0322580486536026, | |
| "learning_rate": 0.00019139270579204194, | |
| "loss": 10.3478, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.08004235044997353, | |
| "grad_norm": 0.035218119621276855, | |
| "learning_rate": 0.00019129715464201832, | |
| "loss": 10.3475, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.08046585494970884, | |
| "grad_norm": 0.0283061470836401, | |
| "learning_rate": 0.0001912011001715617, | |
| "loss": 10.3469, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.08088935944944416, | |
| "grad_norm": 0.03684883192181587, | |
| "learning_rate": 0.00019110454291021954, | |
| "loss": 10.3483, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.08131286394917946, | |
| "grad_norm": 0.028339441865682602, | |
| "learning_rate": 0.00019100748339031113, | |
| "loss": 10.3484, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.08173636844891477, | |
| "grad_norm": 0.03159940615296364, | |
| "learning_rate": 0.00019090992214692488, | |
| "loss": 10.346, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.08215987294865008, | |
| "grad_norm": 0.029895318672060966, | |
| "learning_rate": 0.00019081185971791504, | |
| "loss": 10.3481, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.08258337744838538, | |
| "grad_norm": 0.04218447580933571, | |
| "learning_rate": 0.0001907132966438989, | |
| "loss": 10.3453, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.0830068819481207, | |
| "grad_norm": 0.042808372527360916, | |
| "learning_rate": 0.00019061423346825395, | |
| "loss": 10.3466, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.08343038644785601, | |
| "grad_norm": 0.03805699571967125, | |
| "learning_rate": 0.00019051467073711456, | |
| "loss": 10.3466, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.08385389094759131, | |
| "grad_norm": 0.04781021177768707, | |
| "learning_rate": 0.00019041460899936921, | |
| "loss": 10.3436, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.08427739544732663, | |
| "grad_norm": 0.025532910600304604, | |
| "learning_rate": 0.00019031404880665739, | |
| "loss": 10.3478, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.08470089994706194, | |
| "grad_norm": 0.030978702008724213, | |
| "learning_rate": 0.00019021299071336664, | |
| "loss": 10.3455, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.08512440444679725, | |
| "grad_norm": 0.03757680207490921, | |
| "learning_rate": 0.00019011143527662935, | |
| "loss": 10.3481, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.08554790894653255, | |
| "grad_norm": 0.04030987620353699, | |
| "learning_rate": 0.00019000938305631975, | |
| "loss": 10.3465, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.08597141344626787, | |
| "grad_norm": 0.04490538313984871, | |
| "learning_rate": 0.00018990683461505087, | |
| "loss": 10.3444, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.08639491794600318, | |
| "grad_norm": 0.03259282931685448, | |
| "learning_rate": 0.00018980379051817138, | |
| "loss": 10.3471, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.08681842244573848, | |
| "grad_norm": 0.04348522052168846, | |
| "learning_rate": 0.00018970025133376253, | |
| "loss": 10.3488, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.0872419269454738, | |
| "grad_norm": 0.0327000729739666, | |
| "learning_rate": 0.00018959621763263494, | |
| "loss": 10.347, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.08766543144520911, | |
| "grad_norm": 0.043357010930776596, | |
| "learning_rate": 0.0001894916899883255, | |
| "loss": 10.3514, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.08808893594494442, | |
| "grad_norm": 0.03237845376133919, | |
| "learning_rate": 0.00018938666897709425, | |
| "loss": 10.3454, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.08851244044467972, | |
| "grad_norm": 0.040286242961883545, | |
| "learning_rate": 0.0001892811551779211, | |
| "loss": 10.3446, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.08893594494441504, | |
| "grad_norm": 0.03772817552089691, | |
| "learning_rate": 0.00018917514917250275, | |
| "loss": 10.3458, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.08935944944415035, | |
| "grad_norm": 0.0332292765378952, | |
| "learning_rate": 0.00018906865154524942, | |
| "loss": 10.3453, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.08978295394388565, | |
| "grad_norm": 0.038554366677999496, | |
| "learning_rate": 0.00018896166288328155, | |
| "loss": 10.3463, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.09020645844362096, | |
| "grad_norm": 0.05035999044775963, | |
| "learning_rate": 0.00018885418377642674, | |
| "loss": 10.346, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.09062996294335628, | |
| "grad_norm": 0.03604661673307419, | |
| "learning_rate": 0.00018874621481721645, | |
| "loss": 10.3474, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.09105346744309158, | |
| "grad_norm": 0.0357426181435585, | |
| "learning_rate": 0.00018863775660088258, | |
| "loss": 10.347, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.09147697194282689, | |
| "grad_norm": 0.031967032700777054, | |
| "learning_rate": 0.00018852880972535432, | |
| "loss": 10.3471, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.0919004764425622, | |
| "grad_norm": 0.031692031770944595, | |
| "learning_rate": 0.0001884193747912549, | |
| "loss": 10.3457, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.09232398094229752, | |
| "grad_norm": 0.04586448892951012, | |
| "learning_rate": 0.00018830945240189817, | |
| "loss": 10.3457, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.09274748544203282, | |
| "grad_norm": 0.039624523371458054, | |
| "learning_rate": 0.00018819904316328532, | |
| "loss": 10.3455, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.09317098994176813, | |
| "grad_norm": 0.042145851999521255, | |
| "learning_rate": 0.00018808814768410157, | |
| "loss": 10.3445, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.09359449444150345, | |
| "grad_norm": 0.0323745459318161, | |
| "learning_rate": 0.0001879767665757127, | |
| "loss": 10.3408, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.09401799894123875, | |
| "grad_norm": 0.03138961270451546, | |
| "learning_rate": 0.00018786490045216182, | |
| "loss": 10.3448, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.09444150344097406, | |
| "grad_norm": 0.0241545382887125, | |
| "learning_rate": 0.00018775254993016595, | |
| "loss": 10.3481, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.09486500794070937, | |
| "grad_norm": 0.03655562922358513, | |
| "learning_rate": 0.0001876397156291125, | |
| "loss": 10.3438, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.09528851244044469, | |
| "grad_norm": 0.042878881096839905, | |
| "learning_rate": 0.00018752639817105606, | |
| "loss": 10.345, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.09571201694017999, | |
| "grad_norm": 0.03456420823931694, | |
| "learning_rate": 0.0001874125981807148, | |
| "loss": 10.3447, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.0961355214399153, | |
| "grad_norm": 0.033554937690496445, | |
| "learning_rate": 0.00018729831628546702, | |
| "loss": 10.3467, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.09655902593965061, | |
| "grad_norm": 0.028273334726691246, | |
| "learning_rate": 0.00018718355311534793, | |
| "loss": 10.348, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.09698253043938591, | |
| "grad_norm": 0.03176790103316307, | |
| "learning_rate": 0.00018706830930304585, | |
| "loss": 10.3438, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.09740603493912123, | |
| "grad_norm": 0.04405641928315163, | |
| "learning_rate": 0.000186952585483899, | |
| "loss": 10.3432, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.09782953943885654, | |
| "grad_norm": 0.039611659944057465, | |
| "learning_rate": 0.00018683638229589168, | |
| "loss": 10.3477, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.09825304393859184, | |
| "grad_norm": 0.03426756337285042, | |
| "learning_rate": 0.00018671970037965118, | |
| "loss": 10.3482, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.09867654843832716, | |
| "grad_norm": 0.0545201450586319, | |
| "learning_rate": 0.00018660254037844388, | |
| "loss": 10.3437, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.09910005293806247, | |
| "grad_norm": 0.041258279234170914, | |
| "learning_rate": 0.00018648490293817185, | |
| "loss": 10.3463, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.09952355743779778, | |
| "grad_norm": 0.025181951001286507, | |
| "learning_rate": 0.00018636678870736928, | |
| "loss": 10.3454, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.09994706193753308, | |
| "grad_norm": 0.02877328358590603, | |
| "learning_rate": 0.00018624819833719896, | |
| "loss": 10.3448, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.1003705664372684, | |
| "grad_norm": 0.049800995737314224, | |
| "learning_rate": 0.00018612913248144852, | |
| "loss": 10.3473, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.10079407093700371, | |
| "grad_norm": 0.02784821018576622, | |
| "learning_rate": 0.0001860095917965271, | |
| "loss": 10.3458, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.10121757543673901, | |
| "grad_norm": 0.0490335077047348, | |
| "learning_rate": 0.00018588957694146138, | |
| "loss": 10.3444, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.10164107993647432, | |
| "grad_norm": 0.03331589698791504, | |
| "learning_rate": 0.0001857690885778923, | |
| "loss": 10.3478, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.10206458443620964, | |
| "grad_norm": 0.03565964475274086, | |
| "learning_rate": 0.00018564812737007112, | |
| "loss": 10.3445, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.10248808893594494, | |
| "grad_norm": 0.036929886788129807, | |
| "learning_rate": 0.00018552669398485598, | |
| "loss": 10.3427, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.10291159343568025, | |
| "grad_norm": 0.03922504186630249, | |
| "learning_rate": 0.0001854047890917081, | |
| "loss": 10.3466, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.10333509793541557, | |
| "grad_norm": 0.03552815318107605, | |
| "learning_rate": 0.0001852824133626881, | |
| "loss": 10.3414, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.10375860243515088, | |
| "grad_norm": 0.051186930388212204, | |
| "learning_rate": 0.0001851595674724523, | |
| "loss": 10.3479, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.10418210693488618, | |
| "grad_norm": 0.03887473791837692, | |
| "learning_rate": 0.00018503625209824906, | |
| "loss": 10.3456, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.10460561143462149, | |
| "grad_norm": 0.03032403625547886, | |
| "learning_rate": 0.00018491246791991502, | |
| "loss": 10.3421, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.1050291159343568, | |
| "grad_norm": 0.03820972517132759, | |
| "learning_rate": 0.0001847882156198713, | |
| "loss": 10.3479, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.1054526204340921, | |
| "grad_norm": 0.04590925946831703, | |
| "learning_rate": 0.0001846634958831197, | |
| "loss": 10.3442, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.10587612493382742, | |
| "grad_norm": 0.033261772245168686, | |
| "learning_rate": 0.00018453830939723913, | |
| "loss": 10.3457, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.10629962943356273, | |
| "grad_norm": 0.037462469190359116, | |
| "learning_rate": 0.00018441265685238158, | |
| "loss": 10.3421, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.10672313393329805, | |
| "grad_norm": 0.0378030389547348, | |
| "learning_rate": 0.00018428653894126846, | |
| "loss": 10.345, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.10714663843303335, | |
| "grad_norm": 0.030371299013495445, | |
| "learning_rate": 0.00018415995635918676, | |
| "loss": 10.3488, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.10757014293276866, | |
| "grad_norm": 0.028029056265950203, | |
| "learning_rate": 0.00018403290980398512, | |
| "loss": 10.3436, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.10799364743250398, | |
| "grad_norm": 0.03999907523393631, | |
| "learning_rate": 0.00018390539997607014, | |
| "loss": 10.3432, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.10841715193223928, | |
| "grad_norm": 0.046100400388240814, | |
| "learning_rate": 0.00018377742757840244, | |
| "loss": 10.3444, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.10884065643197459, | |
| "grad_norm": 0.03245000168681145, | |
| "learning_rate": 0.0001836489933164927, | |
| "loss": 10.3434, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.1092641609317099, | |
| "grad_norm": 0.028921889141201973, | |
| "learning_rate": 0.000183520097898398, | |
| "loss": 10.3453, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.1096876654314452, | |
| "grad_norm": 0.033345550298690796, | |
| "learning_rate": 0.00018339074203471757, | |
| "loss": 10.3431, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.11011116993118052, | |
| "grad_norm": 0.036411840468645096, | |
| "learning_rate": 0.00018326092643858923, | |
| "loss": 10.3433, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.11053467443091583, | |
| "grad_norm": 0.039705585688352585, | |
| "learning_rate": 0.00018313065182568527, | |
| "loss": 10.3447, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.11095817893065114, | |
| "grad_norm": 0.0351327620446682, | |
| "learning_rate": 0.00018299991891420847, | |
| "loss": 10.3451, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.11138168343038644, | |
| "grad_norm": 0.04147129878401756, | |
| "learning_rate": 0.00018286872842488832, | |
| "loss": 10.3408, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.11180518793012176, | |
| "grad_norm": 0.036880653351545334, | |
| "learning_rate": 0.00018273708108097677, | |
| "loss": 10.3433, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.11222869242985707, | |
| "grad_norm": 0.05368657410144806, | |
| "learning_rate": 0.00018260497760824458, | |
| "loss": 10.3491, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.11265219692959237, | |
| "grad_norm": 0.03917551413178444, | |
| "learning_rate": 0.00018247241873497707, | |
| "loss": 10.3421, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.11307570142932769, | |
| "grad_norm": 0.03874152526259422, | |
| "learning_rate": 0.0001823394051919701, | |
| "loss": 10.3435, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.113499205929063, | |
| "grad_norm": 0.04026317596435547, | |
| "learning_rate": 0.0001822059377125263, | |
| "loss": 10.3456, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.11392271042879831, | |
| "grad_norm": 0.06843981891870499, | |
| "learning_rate": 0.00018207201703245062, | |
| "loss": 10.3463, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.11434621492853361, | |
| "grad_norm": 0.05057435482740402, | |
| "learning_rate": 0.00018193764389004674, | |
| "loss": 10.3409, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.11476971942826893, | |
| "grad_norm": 0.03488187864422798, | |
| "learning_rate": 0.0001818028190261126, | |
| "loss": 10.3471, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.11519322392800424, | |
| "grad_norm": 0.05958685651421547, | |
| "learning_rate": 0.0001816675431839365, | |
| "loss": 10.3482, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.11561672842773954, | |
| "grad_norm": 0.03940315917134285, | |
| "learning_rate": 0.000181531817109293, | |
| "loss": 10.3487, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.11604023292747485, | |
| "grad_norm": 0.044273462146520615, | |
| "learning_rate": 0.00018139564155043885, | |
| "loss": 10.3443, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.11646373742721017, | |
| "grad_norm": 0.028659025207161903, | |
| "learning_rate": 0.00018125901725810865, | |
| "loss": 10.346, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.11688724192694547, | |
| "grad_norm": 0.054137472063302994, | |
| "learning_rate": 0.00018112194498551106, | |
| "loss": 10.3445, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.11731074642668078, | |
| "grad_norm": 0.028421467170119286, | |
| "learning_rate": 0.00018098442548832426, | |
| "loss": 10.3408, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.1177342509264161, | |
| "grad_norm": 0.036345433443784714, | |
| "learning_rate": 0.0001808464595246921, | |
| "loss": 10.3425, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.11815775542615141, | |
| "grad_norm": 0.034704405814409256, | |
| "learning_rate": 0.00018070804785521975, | |
| "loss": 10.3469, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.11858125992588671, | |
| "grad_norm": 0.04459795728325844, | |
| "learning_rate": 0.0001805691912429696, | |
| "loss": 10.3433, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.11900476442562202, | |
| "grad_norm": 0.03889624401926994, | |
| "learning_rate": 0.0001804298904534569, | |
| "loss": 10.3421, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.11942826892535734, | |
| "grad_norm": 0.03947869688272476, | |
| "learning_rate": 0.0001802901462546457, | |
| "loss": 10.3403, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.11985177342509264, | |
| "grad_norm": 0.055074214935302734, | |
| "learning_rate": 0.00018014995941694468, | |
| "loss": 10.344, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.12027527792482795, | |
| "grad_norm": 0.029949650168418884, | |
| "learning_rate": 0.00018000933071320258, | |
| "loss": 10.3431, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.12069878242456326, | |
| "grad_norm": 0.041310764849185944, | |
| "learning_rate": 0.0001798682609187043, | |
| "loss": 10.3423, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.12112228692429856, | |
| "grad_norm": 0.0433335155248642, | |
| "learning_rate": 0.00017972675081116637, | |
| "loss": 10.3431, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.12154579142403388, | |
| "grad_norm": 0.03031458891928196, | |
| "learning_rate": 0.0001795848011707328, | |
| "loss": 10.3417, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.12196929592376919, | |
| "grad_norm": 0.03494204208254814, | |
| "learning_rate": 0.00017944241277997077, | |
| "loss": 10.345, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.1223928004235045, | |
| "grad_norm": 0.027185741811990738, | |
| "learning_rate": 0.0001792995864238663, | |
| "loss": 10.3429, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.1228163049232398, | |
| "grad_norm": 0.056868940591812134, | |
| "learning_rate": 0.00017915632288981978, | |
| "loss": 10.3404, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.12323980942297512, | |
| "grad_norm": 0.03859318792819977, | |
| "learning_rate": 0.0001790126229676419, | |
| "loss": 10.3404, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.12366331392271043, | |
| "grad_norm": 0.029093610122799873, | |
| "learning_rate": 0.0001788684874495491, | |
| "loss": 10.3451, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.12408681842244573, | |
| "grad_norm": 0.0350499302148819, | |
| "learning_rate": 0.00017872391713015924, | |
| "loss": 10.3397, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.12451032292218105, | |
| "grad_norm": 0.03968917950987816, | |
| "learning_rate": 0.00017857891280648728, | |
| "loss": 10.3428, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.12493382742191636, | |
| "grad_norm": 0.038241248577833176, | |
| "learning_rate": 0.00017843347527794081, | |
| "loss": 10.3415, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.12535733192165166, | |
| "grad_norm": 0.03694219887256622, | |
| "learning_rate": 0.00017828760534631565, | |
| "loss": 10.341, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.125780836421387, | |
| "grad_norm": 0.03630373254418373, | |
| "learning_rate": 0.00017814130381579155, | |
| "loss": 10.3388, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.1262043409211223, | |
| "grad_norm": 0.031546298414468765, | |
| "learning_rate": 0.00017799457149292753, | |
| "loss": 10.3418, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.1266278454208576, | |
| "grad_norm": 0.05403247848153114, | |
| "learning_rate": 0.00017784740918665767, | |
| "loss": 10.3451, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.12705134992059292, | |
| "grad_norm": 0.03717590495944023, | |
| "learning_rate": 0.00017769981770828652, | |
| "loss": 10.3419, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.12747485442032822, | |
| "grad_norm": 0.05108652263879776, | |
| "learning_rate": 0.0001775517978714846, | |
| "loss": 10.3414, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.12789835892006352, | |
| "grad_norm": 0.04460764676332474, | |
| "learning_rate": 0.000177403350492284, | |
| "loss": 10.3443, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.12832186341979884, | |
| "grad_norm": 0.049197494983673096, | |
| "learning_rate": 0.00017725447638907392, | |
| "loss": 10.3426, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.12874536791953414, | |
| "grad_norm": 0.029134899377822876, | |
| "learning_rate": 0.0001771051763825959, | |
| "loss": 10.3409, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.12916887241926944, | |
| "grad_norm": 0.03543701395392418, | |
| "learning_rate": 0.00017695545129593973, | |
| "loss": 10.3442, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.12959237691900477, | |
| "grad_norm": 0.03356650099158287, | |
| "learning_rate": 0.00017680530195453845, | |
| "loss": 10.3429, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.13001588141874007, | |
| "grad_norm": 0.0437590628862381, | |
| "learning_rate": 0.00017665472918616413, | |
| "loss": 10.3449, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.13043938591847537, | |
| "grad_norm": 0.03688879683613777, | |
| "learning_rate": 0.00017650373382092314, | |
| "loss": 10.3398, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.1308628904182107, | |
| "grad_norm": 0.02486329711973667, | |
| "learning_rate": 0.00017635231669125165, | |
| "loss": 10.3408, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.131286394917946, | |
| "grad_norm": 0.03931661695241928, | |
| "learning_rate": 0.000176200478631911, | |
| "loss": 10.3393, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.13170989941768133, | |
| "grad_norm": 0.03328394889831543, | |
| "learning_rate": 0.00017604822047998306, | |
| "loss": 10.3418, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.13213340391741663, | |
| "grad_norm": 0.04236508905887604, | |
| "learning_rate": 0.0001758955430748658, | |
| "loss": 10.3432, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.13255690841715193, | |
| "grad_norm": 0.026039429008960724, | |
| "learning_rate": 0.0001757424472582684, | |
| "loss": 10.3464, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.13298041291688725, | |
| "grad_norm": 0.0246927160769701, | |
| "learning_rate": 0.00017558893387420682, | |
| "loss": 10.3451, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.13340391741662255, | |
| "grad_norm": 0.030340131372213364, | |
| "learning_rate": 0.00017543500376899902, | |
| "loss": 10.3401, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.13382742191635785, | |
| "grad_norm": 0.04260968044400215, | |
| "learning_rate": 0.00017528065779126033, | |
| "loss": 10.3414, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.13425092641609318, | |
| "grad_norm": 0.03421700373291969, | |
| "learning_rate": 0.00017512589679189887, | |
| "loss": 10.3402, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.13467443091582848, | |
| "grad_norm": 0.03428565710783005, | |
| "learning_rate": 0.0001749707216241106, | |
| "loss": 10.3406, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.13509793541556378, | |
| "grad_norm": 0.042442288249731064, | |
| "learning_rate": 0.000174815133143375, | |
| "loss": 10.3381, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.1355214399152991, | |
| "grad_norm": 0.0397978350520134, | |
| "learning_rate": 0.00017465913220744998, | |
| "loss": 10.3427, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.1359449444150344, | |
| "grad_norm": 0.03605269640684128, | |
| "learning_rate": 0.00017450271967636737, | |
| "loss": 10.3397, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.1363684489147697, | |
| "grad_norm": 0.034129634499549866, | |
| "learning_rate": 0.00017434589641242813, | |
| "loss": 10.3463, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.13679195341450504, | |
| "grad_norm": 0.03590450435876846, | |
| "learning_rate": 0.0001741886632801976, | |
| "loss": 10.3416, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.13721545791424034, | |
| "grad_norm": 0.040069352835416794, | |
| "learning_rate": 0.0001740310211465006, | |
| "loss": 10.3427, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.13763896241397564, | |
| "grad_norm": 0.03840317204594612, | |
| "learning_rate": 0.00017387297088041693, | |
| "loss": 10.3431, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.13806246691371096, | |
| "grad_norm": 0.04085763916373253, | |
| "learning_rate": 0.0001737145133532764, | |
| "loss": 10.3379, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.13848597141344626, | |
| "grad_norm": 0.03601207211613655, | |
| "learning_rate": 0.0001735556494386539, | |
| "loss": 10.3407, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.1389094759131816, | |
| "grad_norm": 0.03058718331158161, | |
| "learning_rate": 0.00017339638001236492, | |
| "loss": 10.3411, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.1393329804129169, | |
| "grad_norm": 0.03896321728825569, | |
| "learning_rate": 0.0001732367059524604, | |
| "loss": 10.3426, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.1397564849126522, | |
| "grad_norm": 0.040502067655324936, | |
| "learning_rate": 0.0001730766281392221, | |
| "loss": 10.3411, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.14017998941238752, | |
| "grad_norm": 0.032813332974910736, | |
| "learning_rate": 0.0001729161474551576, | |
| "loss": 10.343, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.14060349391212282, | |
| "grad_norm": 0.032831039279699326, | |
| "learning_rate": 0.00017275526478499555, | |
| "loss": 10.3403, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.14102699841185812, | |
| "grad_norm": 0.033066242933273315, | |
| "learning_rate": 0.00017259398101568076, | |
| "loss": 10.3439, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.14145050291159345, | |
| "grad_norm": 0.032812707126140594, | |
| "learning_rate": 0.00017243229703636922, | |
| "loss": 10.3396, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.14187400741132875, | |
| "grad_norm": 0.03849957883358002, | |
| "learning_rate": 0.0001722702137384234, | |
| "loss": 10.3437, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.14229751191106405, | |
| "grad_norm": 0.047831226140260696, | |
| "learning_rate": 0.00017210773201540707, | |
| "loss": 10.3375, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.14272101641079937, | |
| "grad_norm": 0.04042219743132591, | |
| "learning_rate": 0.0001719448527630806, | |
| "loss": 10.3405, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.14272101641079937, | |
| "eval_loss": 10.340270042419434, | |
| "eval_runtime": 3.4931, | |
| "eval_samples_per_second": 284.85, | |
| "eval_steps_per_second": 142.568, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.14314452091053467, | |
| "grad_norm": 0.030297674238681793, | |
| "learning_rate": 0.00017178157687939592, | |
| "loss": 10.3392, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.14356802541026997, | |
| "grad_norm": 0.030716104432940483, | |
| "learning_rate": 0.00017161790526449156, | |
| "loss": 10.3387, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.1439915299100053, | |
| "grad_norm": 0.034860242158174515, | |
| "learning_rate": 0.00017145383882068778, | |
| "loss": 10.3383, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.1444150344097406, | |
| "grad_norm": 0.04767249897122383, | |
| "learning_rate": 0.00017128937845248146, | |
| "loss": 10.3434, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.1448385389094759, | |
| "grad_norm": 0.02438390627503395, | |
| "learning_rate": 0.00017112452506654117, | |
| "loss": 10.3438, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.14526204340921123, | |
| "grad_norm": 0.04478878155350685, | |
| "learning_rate": 0.00017095927957170228, | |
| "loss": 10.3411, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.14568554790894653, | |
| "grad_norm": 0.03832190856337547, | |
| "learning_rate": 0.00017079364287896174, | |
| "loss": 10.3427, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.14610905240868186, | |
| "grad_norm": 0.03669346123933792, | |
| "learning_rate": 0.00017062761590147323, | |
| "loss": 10.3416, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.14653255690841716, | |
| "grad_norm": 0.03234705701470375, | |
| "learning_rate": 0.00017046119955454206, | |
| "loss": 10.3382, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.14695606140815246, | |
| "grad_norm": 0.028199292719364166, | |
| "learning_rate": 0.00017029439475562015, | |
| "loss": 10.3395, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.14737956590788778, | |
| "grad_norm": 0.03456572815775871, | |
| "learning_rate": 0.0001701272024243009, | |
| "loss": 10.3412, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.14780307040762308, | |
| "grad_norm": 0.041843071579933167, | |
| "learning_rate": 0.00016995962348231424, | |
| "loss": 10.3384, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.14822657490735838, | |
| "grad_norm": 0.0353543683886528, | |
| "learning_rate": 0.0001697916588535214, | |
| "loss": 10.3402, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.1486500794070937, | |
| "grad_norm": 0.03074280545115471, | |
| "learning_rate": 0.00016962330946391, | |
| "loss": 10.3408, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.149073583906829, | |
| "grad_norm": 0.029107527807354927, | |
| "learning_rate": 0.00016945457624158871, | |
| "loss": 10.3404, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.1494970884065643, | |
| "grad_norm": 0.028990836814045906, | |
| "learning_rate": 0.00016928546011678238, | |
| "loss": 10.3366, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.14992059290629964, | |
| "grad_norm": 0.026332266628742218, | |
| "learning_rate": 0.00016911596202182677, | |
| "loss": 10.3423, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.15034409740603494, | |
| "grad_norm": 0.044704243540763855, | |
| "learning_rate": 0.00016894608289116342, | |
| "loss": 10.3407, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.15076760190577024, | |
| "grad_norm": 0.036491744220256805, | |
| "learning_rate": 0.00016877582366133455, | |
| "loss": 10.3393, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.15119110640550557, | |
| "grad_norm": 0.02925538271665573, | |
| "learning_rate": 0.0001686051852709778, | |
| "loss": 10.3394, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.15161461090524087, | |
| "grad_norm": 0.03139074891805649, | |
| "learning_rate": 0.00016843416866082117, | |
| "loss": 10.3381, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.15203811540497617, | |
| "grad_norm": 0.03710184246301651, | |
| "learning_rate": 0.00016826277477367775, | |
| "loss": 10.3378, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.1524616199047115, | |
| "grad_norm": 0.0361437126994133, | |
| "learning_rate": 0.0001680910045544406, | |
| "loss": 10.3408, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.1528851244044468, | |
| "grad_norm": 0.04383867606520653, | |
| "learning_rate": 0.0001679188589500775, | |
| "loss": 10.3415, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.15330862890418212, | |
| "grad_norm": 0.03228599205613136, | |
| "learning_rate": 0.0001677463389096256, | |
| "loss": 10.3413, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.15373213340391742, | |
| "grad_norm": 0.03311069682240486, | |
| "learning_rate": 0.00016757344538418653, | |
| "loss": 10.3409, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.15415563790365272, | |
| "grad_norm": 0.037153564393520355, | |
| "learning_rate": 0.00016740017932692075, | |
| "loss": 10.338, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.15457914240338805, | |
| "grad_norm": 0.03567847982048988, | |
| "learning_rate": 0.00016722654169304253, | |
| "loss": 10.3395, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.15500264690312335, | |
| "grad_norm": 0.026938440278172493, | |
| "learning_rate": 0.0001670525334398147, | |
| "loss": 10.3397, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.15542615140285865, | |
| "grad_norm": 0.02322826161980629, | |
| "learning_rate": 0.00016687815552654327, | |
| "loss": 10.3386, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.15584965590259398, | |
| "grad_norm": 0.03586160019040108, | |
| "learning_rate": 0.00016670340891457216, | |
| "loss": 10.3396, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.15627316040232928, | |
| "grad_norm": 0.03536440059542656, | |
| "learning_rate": 0.00016652829456727797, | |
| "loss": 10.3412, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.15669666490206458, | |
| "grad_norm": 0.025009091943502426, | |
| "learning_rate": 0.00016635281345006461, | |
| "loss": 10.34, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.1571201694017999, | |
| "grad_norm": 0.02612980827689171, | |
| "learning_rate": 0.00016617696653035795, | |
| "loss": 10.3401, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.1575436739015352, | |
| "grad_norm": 0.04117359593510628, | |
| "learning_rate": 0.00016600075477760058, | |
| "loss": 10.3393, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.1579671784012705, | |
| "grad_norm": 0.0326978899538517, | |
| "learning_rate": 0.00016582417916324635, | |
| "loss": 10.3384, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.15839068290100583, | |
| "grad_norm": 0.044377487152814865, | |
| "learning_rate": 0.00016564724066075515, | |
| "loss": 10.3382, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.15881418740074113, | |
| "grad_norm": 0.050321295857429504, | |
| "learning_rate": 0.00016546994024558743, | |
| "loss": 10.3387, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.15923769190047643, | |
| "grad_norm": 0.022547969594597816, | |
| "learning_rate": 0.00016529227889519886, | |
| "loss": 10.3385, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.15966119640021176, | |
| "grad_norm": 0.034384775906801224, | |
| "learning_rate": 0.00016511425758903493, | |
| "loss": 10.3391, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.16008470089994706, | |
| "grad_norm": 0.02677147649228573, | |
| "learning_rate": 0.00016493587730852558, | |
| "loss": 10.3399, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.16050820539968239, | |
| "grad_norm": 0.03600003570318222, | |
| "learning_rate": 0.00016475713903707978, | |
| "loss": 10.3418, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.16093170989941769, | |
| "grad_norm": 0.032675545662641525, | |
| "learning_rate": 0.00016457804376008008, | |
| "loss": 10.3388, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.16135521439915299, | |
| "grad_norm": 0.03568057715892792, | |
| "learning_rate": 0.00016439859246487724, | |
| "loss": 10.3362, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.1617787188988883, | |
| "grad_norm": 0.035958852618932724, | |
| "learning_rate": 0.00016421878614078468, | |
| "loss": 10.3396, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.1622022233986236, | |
| "grad_norm": 0.03066675178706646, | |
| "learning_rate": 0.00016403862577907315, | |
| "loss": 10.3426, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.1626257278983589, | |
| "grad_norm": 0.042271457612514496, | |
| "learning_rate": 0.0001638581123729652, | |
| "loss": 10.3404, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.16304923239809424, | |
| "grad_norm": 0.034284938126802444, | |
| "learning_rate": 0.00016367724691762967, | |
| "loss": 10.3381, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.16347273689782954, | |
| "grad_norm": 0.023705342784523964, | |
| "learning_rate": 0.00016349603041017626, | |
| "loss": 10.3375, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.16389624139756484, | |
| "grad_norm": 0.031792912632226944, | |
| "learning_rate": 0.00016331446384965003, | |
| "loss": 10.3383, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.16431974589730017, | |
| "grad_norm": 0.035305608063936234, | |
| "learning_rate": 0.0001631325482370259, | |
| "loss": 10.3434, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.16474325039703547, | |
| "grad_norm": 0.03486499562859535, | |
| "learning_rate": 0.00016295028457520306, | |
| "loss": 10.3428, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.16516675489677077, | |
| "grad_norm": 0.03409821167588234, | |
| "learning_rate": 0.00016276767386899955, | |
| "loss": 10.3386, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.1655902593965061, | |
| "grad_norm": 0.02966834418475628, | |
| "learning_rate": 0.0001625847171251466, | |
| "loss": 10.3393, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.1660137638962414, | |
| "grad_norm": 0.02835707552731037, | |
| "learning_rate": 0.00016240141535228323, | |
| "loss": 10.3388, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.1664372683959767, | |
| "grad_norm": 0.03911609947681427, | |
| "learning_rate": 0.00016221776956095046, | |
| "loss": 10.3423, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.16686077289571202, | |
| "grad_norm": 0.02803829312324524, | |
| "learning_rate": 0.00016203378076358598, | |
| "loss": 10.3427, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.16728427739544732, | |
| "grad_norm": 0.03135819733142853, | |
| "learning_rate": 0.00016184944997451854, | |
| "loss": 10.3364, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.16770778189518262, | |
| "grad_norm": 0.03102540783584118, | |
| "learning_rate": 0.00016166477820996216, | |
| "loss": 10.3403, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.16813128639491795, | |
| "grad_norm": 0.026423562318086624, | |
| "learning_rate": 0.0001614797664880107, | |
| "loss": 10.3372, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.16855479089465325, | |
| "grad_norm": 0.03439650684595108, | |
| "learning_rate": 0.00016129441582863217, | |
| "loss": 10.342, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.16897829539438858, | |
| "grad_norm": 0.03104579448699951, | |
| "learning_rate": 0.00016110872725366316, | |
| "loss": 10.3377, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.16940179989412388, | |
| "grad_norm": 0.03529537469148636, | |
| "learning_rate": 0.0001609227017868033, | |
| "loss": 10.3395, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.16982530439385918, | |
| "grad_norm": 0.03002871572971344, | |
| "learning_rate": 0.00016073634045360932, | |
| "loss": 10.3411, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.1702488088935945, | |
| "grad_norm": 0.03288958594202995, | |
| "learning_rate": 0.00016054964428148963, | |
| "loss": 10.3367, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.1706723133933298, | |
| "grad_norm": 0.026648705825209618, | |
| "learning_rate": 0.00016036261429969867, | |
| "loss": 10.3367, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.1710958178930651, | |
| "grad_norm": 0.035582203418016434, | |
| "learning_rate": 0.00016017525153933114, | |
| "loss": 10.3375, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.17151932239280043, | |
| "grad_norm": 0.024190323427319527, | |
| "learning_rate": 0.00015998755703331634, | |
| "loss": 10.3416, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.17194282689253573, | |
| "grad_norm": 0.03089403547346592, | |
| "learning_rate": 0.00015979953181641246, | |
| "loss": 10.3418, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.17236633139227103, | |
| "grad_norm": 0.026094770058989525, | |
| "learning_rate": 0.00015961117692520088, | |
| "loss": 10.3357, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.17278983589200636, | |
| "grad_norm": 0.04286188259720802, | |
| "learning_rate": 0.00015942249339808058, | |
| "loss": 10.3408, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.17321334039174166, | |
| "grad_norm": 0.0313500352203846, | |
| "learning_rate": 0.00015923348227526218, | |
| "loss": 10.3354, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.17363684489147696, | |
| "grad_norm": 0.03544219583272934, | |
| "learning_rate": 0.00015904414459876238, | |
| "loss": 10.3367, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.1740603493912123, | |
| "grad_norm": 0.03017052263021469, | |
| "learning_rate": 0.00015885448141239822, | |
| "loss": 10.3418, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.1744838538909476, | |
| "grad_norm": 0.030451800674200058, | |
| "learning_rate": 0.00015866449376178117, | |
| "loss": 10.3386, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.1749073583906829, | |
| "grad_norm": 0.035226162523031235, | |
| "learning_rate": 0.00015847418269431153, | |
| "loss": 10.3412, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.17533086289041822, | |
| "grad_norm": 0.02857392653822899, | |
| "learning_rate": 0.00015828354925917262, | |
| "loss": 10.3414, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.17575436739015352, | |
| "grad_norm": 0.050622567534446716, | |
| "learning_rate": 0.00015809259450732494, | |
| "loss": 10.3392, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.17617787188988884, | |
| "grad_norm": 0.0338461808860302, | |
| "learning_rate": 0.00015790131949150035, | |
| "loss": 10.3419, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.17660137638962414, | |
| "grad_norm": 0.027923308312892914, | |
| "learning_rate": 0.00015770972526619646, | |
| "loss": 10.3385, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.17702488088935944, | |
| "grad_norm": 0.03212830424308777, | |
| "learning_rate": 0.0001575178128876705, | |
| "loss": 10.339, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.17744838538909477, | |
| "grad_norm": 0.020661218091845512, | |
| "learning_rate": 0.00015732558341393385, | |
| "loss": 10.338, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.17787188988883007, | |
| "grad_norm": 0.02785920538008213, | |
| "learning_rate": 0.00015713303790474594, | |
| "loss": 10.3392, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.17829539438856537, | |
| "grad_norm": 0.018963869661092758, | |
| "learning_rate": 0.00015694017742160846, | |
| "loss": 10.3381, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.1787188988883007, | |
| "grad_norm": 0.02660539373755455, | |
| "learning_rate": 0.0001567470030277597, | |
| "loss": 10.3389, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.179142403388036, | |
| "grad_norm": 0.03342144191265106, | |
| "learning_rate": 0.00015655351578816834, | |
| "loss": 10.3395, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.1795659078877713, | |
| "grad_norm": 0.03541478142142296, | |
| "learning_rate": 0.00015635971676952797, | |
| "loss": 10.3356, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.17998941238750663, | |
| "grad_norm": 0.04339861124753952, | |
| "learning_rate": 0.00015616560704025088, | |
| "loss": 10.34, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.18041291688724193, | |
| "grad_norm": 0.030052557587623596, | |
| "learning_rate": 0.00015597118767046232, | |
| "loss": 10.3366, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.18083642138697723, | |
| "grad_norm": 0.03362065181136131, | |
| "learning_rate": 0.00015577645973199465, | |
| "loss": 10.3446, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.18125992588671255, | |
| "grad_norm": 0.033407680690288544, | |
| "learning_rate": 0.00015558142429838133, | |
| "loss": 10.3382, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.18168343038644785, | |
| "grad_norm": 0.03306809812784195, | |
| "learning_rate": 0.00015538608244485103, | |
| "loss": 10.3391, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.18210693488618315, | |
| "grad_norm": 0.035972122102975845, | |
| "learning_rate": 0.0001551904352483217, | |
| "loss": 10.3378, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.18253043938591848, | |
| "grad_norm": 0.02942793443799019, | |
| "learning_rate": 0.0001549944837873947, | |
| "loss": 10.341, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.18295394388565378, | |
| "grad_norm": 0.0311295036226511, | |
| "learning_rate": 0.00015479822914234875, | |
| "loss": 10.3427, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.1833774483853891, | |
| "grad_norm": 0.03349452093243599, | |
| "learning_rate": 0.00015460167239513396, | |
| "loss": 10.3335, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.1838009528851244, | |
| "grad_norm": 0.024683522060513496, | |
| "learning_rate": 0.00015440481462936613, | |
| "loss": 10.3403, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.1842244573848597, | |
| "grad_norm": 0.02533009834587574, | |
| "learning_rate": 0.00015420765693032035, | |
| "loss": 10.3352, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.18464796188459504, | |
| "grad_norm": 0.02682666666805744, | |
| "learning_rate": 0.0001540102003849253, | |
| "loss": 10.3351, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.18507146638433034, | |
| "grad_norm": 0.026133093982934952, | |
| "learning_rate": 0.0001538124460817573, | |
| "loss": 10.3377, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.18549497088406564, | |
| "grad_norm": 0.04049040377140045, | |
| "learning_rate": 0.00015361439511103414, | |
| "loss": 10.3402, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.18591847538380096, | |
| "grad_norm": 0.02733178623020649, | |
| "learning_rate": 0.00015341604856460904, | |
| "loss": 10.3352, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.18634197988353626, | |
| "grad_norm": 0.02330494113266468, | |
| "learning_rate": 0.0001532174075359649, | |
| "loss": 10.341, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.18676548438327156, | |
| "grad_norm": 0.03259949013590813, | |
| "learning_rate": 0.00015301847312020796, | |
| "loss": 10.3403, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.1871889888830069, | |
| "grad_norm": 0.05194835364818573, | |
| "learning_rate": 0.000152819246414062, | |
| "loss": 10.3413, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.1876124933827422, | |
| "grad_norm": 0.0325242318212986, | |
| "learning_rate": 0.0001526197285158621, | |
| "loss": 10.3396, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.1880359978824775, | |
| "grad_norm": 0.02710815891623497, | |
| "learning_rate": 0.00015241992052554876, | |
| "loss": 10.34, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.18845950238221282, | |
| "grad_norm": 0.024676240980625153, | |
| "learning_rate": 0.0001522198235446617, | |
| "loss": 10.3362, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.18888300688194812, | |
| "grad_norm": 0.02788936160504818, | |
| "learning_rate": 0.0001520194386763339, | |
| "loss": 10.3376, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.18930651138168342, | |
| "grad_norm": 0.03856251761317253, | |
| "learning_rate": 0.00015181876702528537, | |
| "loss": 10.3352, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.18973001588141875, | |
| "grad_norm": 0.03264036402106285, | |
| "learning_rate": 0.00015161780969781728, | |
| "loss": 10.338, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.19015352038115405, | |
| "grad_norm": 0.027694035321474075, | |
| "learning_rate": 0.00015141656780180558, | |
| "loss": 10.3354, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.19057702488088937, | |
| "grad_norm": 0.030413135886192322, | |
| "learning_rate": 0.00015121504244669515, | |
| "loss": 10.3383, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.19100052938062467, | |
| "grad_norm": 0.03150556609034538, | |
| "learning_rate": 0.0001510132347434936, | |
| "loss": 10.3389, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.19142403388035997, | |
| "grad_norm": 0.029888266697525978, | |
| "learning_rate": 0.000150811145804765, | |
| "loss": 10.3394, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.1918475383800953, | |
| "grad_norm": 0.03171524032950401, | |
| "learning_rate": 0.000150608776744624, | |
| "loss": 10.334, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.1922710428798306, | |
| "grad_norm": 0.032492250204086304, | |
| "learning_rate": 0.00015040612867872947, | |
| "loss": 10.3366, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.1926945473795659, | |
| "grad_norm": 0.030303264036774635, | |
| "learning_rate": 0.00015020320272427843, | |
| "loss": 10.3366, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.19311805187930123, | |
| "grad_norm": 0.03860599547624588, | |
| "learning_rate": 0.00015000000000000001, | |
| "loss": 10.3379, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.19354155637903653, | |
| "grad_norm": 0.03272419795393944, | |
| "learning_rate": 0.00014979652162614904, | |
| "loss": 10.3352, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.19396506087877183, | |
| "grad_norm": 0.038201820105314255, | |
| "learning_rate": 0.00014959276872450006, | |
| "loss": 10.3362, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.19438856537850716, | |
| "grad_norm": 0.025923024863004684, | |
| "learning_rate": 0.00014938874241834108, | |
| "loss": 10.3403, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.19481206987824246, | |
| "grad_norm": 0.03889621049165726, | |
| "learning_rate": 0.00014918444383246737, | |
| "loss": 10.3385, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.19523557437797776, | |
| "grad_norm": 0.031947895884513855, | |
| "learning_rate": 0.00014897987409317532, | |
| "loss": 10.3385, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.19565907887771308, | |
| "grad_norm": 0.03579488396644592, | |
| "learning_rate": 0.00014877503432825614, | |
| "loss": 10.3339, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.19608258337744838, | |
| "grad_norm": 0.033163949847221375, | |
| "learning_rate": 0.00014856992566698965, | |
| "loss": 10.3402, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.19650608787718368, | |
| "grad_norm": 0.03128167986869812, | |
| "learning_rate": 0.00014836454924013824, | |
| "loss": 10.3408, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.196929592376919, | |
| "grad_norm": 0.04108097031712532, | |
| "learning_rate": 0.00014815890617994034, | |
| "loss": 10.3394, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.1973530968766543, | |
| "grad_norm": 0.04260754585266113, | |
| "learning_rate": 0.0001479529976201044, | |
| "loss": 10.3428, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.19777660137638964, | |
| "grad_norm": 0.027531959116458893, | |
| "learning_rate": 0.00014774682469580248, | |
| "loss": 10.3395, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.19820010587612494, | |
| "grad_norm": 0.028333760797977448, | |
| "learning_rate": 0.00014754038854366424, | |
| "loss": 10.3374, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.19862361037586024, | |
| "grad_norm": 0.029396837577223778, | |
| "learning_rate": 0.00014733369030177042, | |
| "loss": 10.3363, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.19904711487559557, | |
| "grad_norm": 0.029380813241004944, | |
| "learning_rate": 0.00014712673110964665, | |
| "loss": 10.3372, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.19947061937533087, | |
| "grad_norm": 0.02283712849020958, | |
| "learning_rate": 0.0001469195121082571, | |
| "loss": 10.3408, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.19989412387506617, | |
| "grad_norm": 0.025367606431245804, | |
| "learning_rate": 0.00014671203443999845, | |
| "loss": 10.3383, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.2003176283748015, | |
| "grad_norm": 0.034685924649238586, | |
| "learning_rate": 0.0001465042992486933, | |
| "loss": 10.3373, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.2007411328745368, | |
| "grad_norm": 0.0398382693529129, | |
| "learning_rate": 0.00014629630767958396, | |
| "loss": 10.3374, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.2011646373742721, | |
| "grad_norm": 0.03815117105841637, | |
| "learning_rate": 0.00014608806087932619, | |
| "loss": 10.3382, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.20158814187400742, | |
| "grad_norm": 0.028847893700003624, | |
| "learning_rate": 0.0001458795599959828, | |
| "loss": 10.3355, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.20201164637374272, | |
| "grad_norm": 0.033290982246398926, | |
| "learning_rate": 0.00014567080617901735, | |
| "loss": 10.3353, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.20243515087347802, | |
| "grad_norm": 0.03120148368179798, | |
| "learning_rate": 0.00014546180057928792, | |
| "loss": 10.3365, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.20285865537321335, | |
| "grad_norm": 0.03227855637669563, | |
| "learning_rate": 0.00014525254434904055, | |
| "loss": 10.3373, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.20328215987294865, | |
| "grad_norm": 0.02253713831305504, | |
| "learning_rate": 0.00014504303864190307, | |
| "loss": 10.3379, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.20370566437268395, | |
| "grad_norm": 0.027942582964897156, | |
| "learning_rate": 0.00014483328461287862, | |
| "loss": 10.3387, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.20412916887241928, | |
| "grad_norm": 0.028897034004330635, | |
| "learning_rate": 0.0001446232834183394, | |
| "loss": 10.3406, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.20455267337215458, | |
| "grad_norm": 0.03516876697540283, | |
| "learning_rate": 0.00014441303621602017, | |
| "loss": 10.3317, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.20497617787188988, | |
| "grad_norm": 0.030100248754024506, | |
| "learning_rate": 0.00014420254416501197, | |
| "loss": 10.3365, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.2053996823716252, | |
| "grad_norm": 0.020048066973686218, | |
| "learning_rate": 0.00014399180842575575, | |
| "loss": 10.3426, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.2058231868713605, | |
| "grad_norm": 0.031375959515571594, | |
| "learning_rate": 0.00014378083016003572, | |
| "loss": 10.3376, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.20624669137109583, | |
| "grad_norm": 0.034831635653972626, | |
| "learning_rate": 0.00014356961053097332, | |
| "loss": 10.3354, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.20667019587083113, | |
| "grad_norm": 0.030198190361261368, | |
| "learning_rate": 0.00014335815070302054, | |
| "loss": 10.3361, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.20709370037056643, | |
| "grad_norm": 0.031040605157613754, | |
| "learning_rate": 0.00014314645184195364, | |
| "loss": 10.3412, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.20751720487030176, | |
| "grad_norm": 0.05391615629196167, | |
| "learning_rate": 0.00014293451511486658, | |
| "loss": 10.3402, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.20794070937003706, | |
| "grad_norm": 0.030534790828824043, | |
| "learning_rate": 0.00014272234169016474, | |
| "loss": 10.3402, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.20836421386977236, | |
| "grad_norm": 0.03578052297234535, | |
| "learning_rate": 0.00014250993273755844, | |
| "loss": 10.3348, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.20878771836950769, | |
| "grad_norm": 0.03920895233750343, | |
| "learning_rate": 0.00014229728942805636, | |
| "loss": 10.3417, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.20921122286924299, | |
| "grad_norm": 0.030715953558683395, | |
| "learning_rate": 0.00014208441293395925, | |
| "loss": 10.3379, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.20963472736897829, | |
| "grad_norm": 0.036160390824079514, | |
| "learning_rate": 0.00014187130442885345, | |
| "loss": 10.3368, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.2100582318687136, | |
| "grad_norm": 0.032142747193574905, | |
| "learning_rate": 0.0001416579650876043, | |
| "loss": 10.3404, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.2104817363684489, | |
| "grad_norm": 0.02567223645746708, | |
| "learning_rate": 0.00014144439608634976, | |
| "loss": 10.3387, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.2109052408681842, | |
| "grad_norm": 0.03470413014292717, | |
| "learning_rate": 0.0001412305986024939, | |
| "loss": 10.3419, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.21132874536791954, | |
| "grad_norm": 0.036063164472579956, | |
| "learning_rate": 0.00014101657381470045, | |
| "loss": 10.3335, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.21175224986765484, | |
| "grad_norm": 0.02859325334429741, | |
| "learning_rate": 0.00014080232290288622, | |
| "loss": 10.3385, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.21217575436739014, | |
| "grad_norm": 0.03691897913813591, | |
| "learning_rate": 0.00014058784704821465, | |
| "loss": 10.3371, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.21259925886712547, | |
| "grad_norm": 0.02370496280491352, | |
| "learning_rate": 0.0001403731474330893, | |
| "loss": 10.3373, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.21302276336686077, | |
| "grad_norm": 0.02717514894902706, | |
| "learning_rate": 0.0001401582252411473, | |
| "loss": 10.3362, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.2134462678665961, | |
| "grad_norm": 0.027684593573212624, | |
| "learning_rate": 0.00013994308165725288, | |
| "loss": 10.3407, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.2138697723663314, | |
| "grad_norm": 0.027036601677536964, | |
| "learning_rate": 0.00013972771786749074, | |
| "loss": 10.3387, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.2142932768660667, | |
| "grad_norm": 0.03559018298983574, | |
| "learning_rate": 0.00013951213505915969, | |
| "loss": 10.3398, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.21471678136580202, | |
| "grad_norm": 0.04133779555559158, | |
| "learning_rate": 0.0001392963344207658, | |
| "loss": 10.3355, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.21514028586553732, | |
| "grad_norm": 0.03785044327378273, | |
| "learning_rate": 0.0001390803171420162, | |
| "loss": 10.3344, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.21556379036527262, | |
| "grad_norm": 0.023411711677908897, | |
| "learning_rate": 0.00013886408441381233, | |
| "loss": 10.3362, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.21598729486500795, | |
| "grad_norm": 0.0443277508020401, | |
| "learning_rate": 0.00013864763742824334, | |
| "loss": 10.339, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.21641079936474325, | |
| "grad_norm": 0.036806512624025345, | |
| "learning_rate": 0.0001384309773785796, | |
| "loss": 10.338, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.21683430386447855, | |
| "grad_norm": 0.02885564975440502, | |
| "learning_rate": 0.00013821410545926613, | |
| "loss": 10.3333, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.21725780836421388, | |
| "grad_norm": 0.03067517653107643, | |
| "learning_rate": 0.00013799702286591598, | |
| "loss": 10.3356, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.21768131286394918, | |
| "grad_norm": 0.03321646526455879, | |
| "learning_rate": 0.00013777973079530362, | |
| "loss": 10.3388, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.21810481736368448, | |
| "grad_norm": 0.03147870674729347, | |
| "learning_rate": 0.00013756223044535833, | |
| "loss": 10.3391, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.2185283218634198, | |
| "grad_norm": 0.02573389932513237, | |
| "learning_rate": 0.00013734452301515776, | |
| "loss": 10.3377, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.2189518263631551, | |
| "grad_norm": 0.026358777657151222, | |
| "learning_rate": 0.00013712660970492107, | |
| "loss": 10.3371, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.2193753308628904, | |
| "grad_norm": 0.02714933454990387, | |
| "learning_rate": 0.00013690849171600245, | |
| "loss": 10.3378, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.21979883536262573, | |
| "grad_norm": 0.02859034389257431, | |
| "learning_rate": 0.00013669017025088456, | |
| "loss": 10.3365, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.22022233986236103, | |
| "grad_norm": 0.044585928320884705, | |
| "learning_rate": 0.00013647164651317176, | |
| "loss": 10.3362, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.22064584436209636, | |
| "grad_norm": 0.053858619183301926, | |
| "learning_rate": 0.00013625292170758356, | |
| "loss": 10.3373, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.22106934886183166, | |
| "grad_norm": 0.03403494879603386, | |
| "learning_rate": 0.00013603399703994787, | |
| "loss": 10.3309, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.22149285336156696, | |
| "grad_norm": 0.028249001130461693, | |
| "learning_rate": 0.00013581487371719457, | |
| "loss": 10.3379, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.2219163578613023, | |
| "grad_norm": 0.028280075639486313, | |
| "learning_rate": 0.00013559555294734868, | |
| "loss": 10.3388, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.2223398623610376, | |
| "grad_norm": 0.04397103190422058, | |
| "learning_rate": 0.00013537603593952367, | |
| "loss": 10.3335, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.2227633668607729, | |
| "grad_norm": 0.035089749842882156, | |
| "learning_rate": 0.000135156323903915, | |
| "loss": 10.34, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.22318687136050822, | |
| "grad_norm": 0.03598684072494507, | |
| "learning_rate": 0.00013493641805179319, | |
| "loss": 10.3348, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.22361037586024352, | |
| "grad_norm": 0.03583105653524399, | |
| "learning_rate": 0.0001347163195954973, | |
| "loss": 10.3383, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.22403388035997882, | |
| "grad_norm": 0.03622949495911598, | |
| "learning_rate": 0.0001344960297484283, | |
| "loss": 10.3378, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.22445738485971414, | |
| "grad_norm": 0.027924714609980583, | |
| "learning_rate": 0.00013427554972504226, | |
| "loss": 10.3372, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.22488088935944944, | |
| "grad_norm": 0.047317858785390854, | |
| "learning_rate": 0.00013405488074084358, | |
| "loss": 10.3375, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.22530439385918474, | |
| "grad_norm": 0.031993038952350616, | |
| "learning_rate": 0.0001338340240123785, | |
| "loss": 10.3371, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.22572789835892007, | |
| "grad_norm": 0.03276574984192848, | |
| "learning_rate": 0.00013361298075722833, | |
| "loss": 10.3376, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.22615140285865537, | |
| "grad_norm": 0.024694286286830902, | |
| "learning_rate": 0.00013339175219400257, | |
| "loss": 10.34, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.22657490735839067, | |
| "grad_norm": 0.031688570976257324, | |
| "learning_rate": 0.00013317033954233246, | |
| "loss": 10.3411, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.226998411858126, | |
| "grad_norm": 0.03652056306600571, | |
| "learning_rate": 0.00013294874402286402, | |
| "loss": 10.3329, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.2274219163578613, | |
| "grad_norm": 0.03224468603730202, | |
| "learning_rate": 0.0001327269668572515, | |
| "loss": 10.3386, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.22784542085759663, | |
| "grad_norm": 0.034342508763074875, | |
| "learning_rate": 0.00013250500926815045, | |
| "loss": 10.3371, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.22826892535733193, | |
| "grad_norm": 0.030163973569869995, | |
| "learning_rate": 0.0001322828724792112, | |
| "loss": 10.336, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.22869242985706723, | |
| "grad_norm": 0.030578266829252243, | |
| "learning_rate": 0.00013206055771507197, | |
| "loss": 10.3391, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.22911593435680255, | |
| "grad_norm": 0.035477470606565475, | |
| "learning_rate": 0.00013183806620135216, | |
| "loss": 10.3384, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.22953943885653785, | |
| "grad_norm": 0.026009559631347656, | |
| "learning_rate": 0.00013161539916464558, | |
| "loss": 10.3369, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.22996294335627315, | |
| "grad_norm": 0.033704426139593124, | |
| "learning_rate": 0.00013139255783251367, | |
| "loss": 10.3369, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.23038644785600848, | |
| "grad_norm": 0.03469805791974068, | |
| "learning_rate": 0.00013116954343347882, | |
| "loss": 10.3359, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.23080995235574378, | |
| "grad_norm": 0.029503265395760536, | |
| "learning_rate": 0.0001309463571970175, | |
| "loss": 10.3337, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.23123345685547908, | |
| "grad_norm": 0.027178343385457993, | |
| "learning_rate": 0.0001307230003535535, | |
| "loss": 10.3383, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.2316569613552144, | |
| "grad_norm": 0.026484569534659386, | |
| "learning_rate": 0.00013049947413445125, | |
| "loss": 10.3411, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.2320804658549497, | |
| "grad_norm": 0.03568257763981819, | |
| "learning_rate": 0.00013027577977200883, | |
| "loss": 10.3351, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.232503970354685, | |
| "grad_norm": 0.044057317078113556, | |
| "learning_rate": 0.0001300519184994513, | |
| "loss": 10.3367, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.23292747485442034, | |
| "grad_norm": 0.03619583323597908, | |
| "learning_rate": 0.00012982789155092407, | |
| "loss": 10.3385, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.23335097935415564, | |
| "grad_norm": 0.042276639491319656, | |
| "learning_rate": 0.00012960370016148567, | |
| "loss": 10.337, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.23377448385389094, | |
| "grad_norm": 0.03055988810956478, | |
| "learning_rate": 0.00012937934556710143, | |
| "loss": 10.3385, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.23419798835362626, | |
| "grad_norm": 0.02854546532034874, | |
| "learning_rate": 0.00012915482900463624, | |
| "loss": 10.3393, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.23462149285336156, | |
| "grad_norm": 0.029309969395399094, | |
| "learning_rate": 0.00012893015171184797, | |
| "loss": 10.3319, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.23504499735309686, | |
| "grad_norm": 0.0332510843873024, | |
| "learning_rate": 0.00012870531492738065, | |
| "loss": 10.3338, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.2354685018528322, | |
| "grad_norm": 0.03669944778084755, | |
| "learning_rate": 0.00012848031989075754, | |
| "loss": 10.3325, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.2358920063525675, | |
| "grad_norm": 0.027661770582199097, | |
| "learning_rate": 0.00012825516784237436, | |
| "loss": 10.3382, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.23631551085230282, | |
| "grad_norm": 0.029674025252461433, | |
| "learning_rate": 0.0001280298600234924, | |
| "loss": 10.3387, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.23673901535203812, | |
| "grad_norm": 0.03104621358215809, | |
| "learning_rate": 0.00012780439767623181, | |
| "loss": 10.3354, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.23716251985177342, | |
| "grad_norm": 0.0300068948417902, | |
| "learning_rate": 0.0001275787820435645, | |
| "loss": 10.3396, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.23758602435150875, | |
| "grad_norm": 0.03742906451225281, | |
| "learning_rate": 0.00012735301436930758, | |
| "loss": 10.3364, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.23800952885124405, | |
| "grad_norm": 0.029214419424533844, | |
| "learning_rate": 0.0001271270958981163, | |
| "loss": 10.3368, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.23843303335097935, | |
| "grad_norm": 0.034154172986745834, | |
| "learning_rate": 0.00012690102787547722, | |
| "loss": 10.3364, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.23885653785071467, | |
| "grad_norm": 0.024321483448147774, | |
| "learning_rate": 0.00012667481154770148, | |
| "loss": 10.3348, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.23928004235044997, | |
| "grad_norm": 0.030538305640220642, | |
| "learning_rate": 0.0001264484481619177, | |
| "loss": 10.3374, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.23970354685018527, | |
| "grad_norm": 0.028275547549128532, | |
| "learning_rate": 0.00012622193896606528, | |
| "loss": 10.3343, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.2401270513499206, | |
| "grad_norm": 0.024137398228049278, | |
| "learning_rate": 0.00012599528520888757, | |
| "loss": 10.3363, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.2405505558496559, | |
| "grad_norm": 0.0387752428650856, | |
| "learning_rate": 0.00012576848813992475, | |
| "loss": 10.3355, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.2409740603493912, | |
| "grad_norm": 0.02671218290925026, | |
| "learning_rate": 0.00012554154900950708, | |
| "loss": 10.339, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.24139756484912653, | |
| "grad_norm": 0.031162571161985397, | |
| "learning_rate": 0.00012531446906874808, | |
| "loss": 10.3402, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.24182106934886183, | |
| "grad_norm": 0.03754870593547821, | |
| "learning_rate": 0.00012508724956953755, | |
| "loss": 10.3392, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.24224457384859713, | |
| "grad_norm": 0.030516209080815315, | |
| "learning_rate": 0.00012485989176453462, | |
| "loss": 10.3373, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.24266807834833246, | |
| "grad_norm": 0.04033865034580231, | |
| "learning_rate": 0.0001246323969071609, | |
| "loss": 10.3358, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.24309158284806776, | |
| "grad_norm": 0.0301966555416584, | |
| "learning_rate": 0.00012440476625159364, | |
| "loss": 10.335, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.24351508734780308, | |
| "grad_norm": 0.036701519042253494, | |
| "learning_rate": 0.00012417700105275866, | |
| "loss": 10.3382, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.24393859184753838, | |
| "grad_norm": 0.02948085404932499, | |
| "learning_rate": 0.00012394910256632356, | |
| "loss": 10.3342, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.24436209634727368, | |
| "grad_norm": 0.0245877243578434, | |
| "learning_rate": 0.00012372107204869077, | |
| "loss": 10.3364, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.244785600847009, | |
| "grad_norm": 0.023439116775989532, | |
| "learning_rate": 0.00012349291075699058, | |
| "loss": 10.3361, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.2452091053467443, | |
| "grad_norm": 0.026123927906155586, | |
| "learning_rate": 0.00012326461994907424, | |
| "loss": 10.3398, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.2456326098464796, | |
| "grad_norm": 0.03437687084078789, | |
| "learning_rate": 0.000123036200883507, | |
| "loss": 10.3373, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.24605611434621494, | |
| "grad_norm": 0.03299521282315254, | |
| "learning_rate": 0.00012280765481956124, | |
| "loss": 10.3344, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.24647961884595024, | |
| "grad_norm": 0.03710121661424637, | |
| "learning_rate": 0.0001225789830172094, | |
| "loss": 10.3354, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.24690312334568554, | |
| "grad_norm": 0.032498303800821304, | |
| "learning_rate": 0.0001223501867371173, | |
| "loss": 10.3344, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.24732662784542087, | |
| "grad_norm": 0.03610834851861, | |
| "learning_rate": 0.00012212126724063676, | |
| "loss": 10.3359, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.24775013234515617, | |
| "grad_norm": 0.03149677813053131, | |
| "learning_rate": 0.00012189222578979903, | |
| "loss": 10.3376, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.24817363684489147, | |
| "grad_norm": 0.031013086438179016, | |
| "learning_rate": 0.00012166306364730766, | |
| "loss": 10.3333, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.2485971413446268, | |
| "grad_norm": 0.030261732637882233, | |
| "learning_rate": 0.00012143378207653164, | |
| "loss": 10.3327, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.2490206458443621, | |
| "grad_norm": 0.030076345428824425, | |
| "learning_rate": 0.00012120438234149827, | |
| "loss": 10.3393, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.2494441503440974, | |
| "grad_norm": 0.027937186881899834, | |
| "learning_rate": 0.00012097486570688634, | |
| "loss": 10.3386, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.24986765484383272, | |
| "grad_norm": 0.037603769451379776, | |
| "learning_rate": 0.00012074523343801906, | |
| "loss": 10.3306, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.25029115934356805, | |
| "grad_norm": 0.027752617374062538, | |
| "learning_rate": 0.0001205154868008572, | |
| "loss": 10.3352, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.2507146638433033, | |
| "grad_norm": 0.030105147510766983, | |
| "learning_rate": 0.000120285627061992, | |
| "loss": 10.3306, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.25113816834303865, | |
| "grad_norm": 0.026609288528561592, | |
| "learning_rate": 0.00012005565548863822, | |
| "loss": 10.3347, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.251561672842774, | |
| "grad_norm": 0.04250922426581383, | |
| "learning_rate": 0.00011982557334862723, | |
| "loss": 10.3303, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.25198517734250925, | |
| "grad_norm": 0.03030312806367874, | |
| "learning_rate": 0.00011959538191039985, | |
| "loss": 10.3389, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.2524086818422446, | |
| "grad_norm": 0.03314143419265747, | |
| "learning_rate": 0.00011936508244299948, | |
| "loss": 10.336, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.2528321863419799, | |
| "grad_norm": 0.03237884119153023, | |
| "learning_rate": 0.0001191346762160652, | |
| "loss": 10.3406, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.2532556908417152, | |
| "grad_norm": 0.02621031180024147, | |
| "learning_rate": 0.00011890416449982451, | |
| "loss": 10.3367, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.2536791953414505, | |
| "grad_norm": 0.023484721779823303, | |
| "learning_rate": 0.00011867354856508656, | |
| "loss": 10.3327, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.25410269984118583, | |
| "grad_norm": 0.02962653897702694, | |
| "learning_rate": 0.00011844282968323501, | |
| "loss": 10.3359, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.2545262043409211, | |
| "grad_norm": 0.028051255270838737, | |
| "learning_rate": 0.0001182120091262211, | |
| "loss": 10.3356, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.25494970884065643, | |
| "grad_norm": 0.029255535453557968, | |
| "learning_rate": 0.00011798108816655657, | |
| "loss": 10.3365, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.25537321334039176, | |
| "grad_norm": 0.029102357104420662, | |
| "learning_rate": 0.00011775006807730667, | |
| "loss": 10.3347, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.25579671784012703, | |
| "grad_norm": 0.02935311570763588, | |
| "learning_rate": 0.00011751895013208325, | |
| "loss": 10.3369, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.25622022233986236, | |
| "grad_norm": 0.033430956304073334, | |
| "learning_rate": 0.00011728773560503751, | |
| "loss": 10.3381, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.2566437268395977, | |
| "grad_norm": 0.03818434476852417, | |
| "learning_rate": 0.00011705642577085316, | |
| "loss": 10.3354, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.25706723133933296, | |
| "grad_norm": 0.029122449457645416, | |
| "learning_rate": 0.00011682502190473938, | |
| "loss": 10.3382, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.2574907358390683, | |
| "grad_norm": 0.030079467222094536, | |
| "learning_rate": 0.00011659352528242366, | |
| "loss": 10.3413, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.2579142403388036, | |
| "grad_norm": 0.02247581258416176, | |
| "learning_rate": 0.00011636193718014494, | |
| "loss": 10.3364, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.2583377448385389, | |
| "grad_norm": 0.032431941479444504, | |
| "learning_rate": 0.00011613025887464641, | |
| "loss": 10.3323, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.2587612493382742, | |
| "grad_norm": 0.032824281603097916, | |
| "learning_rate": 0.00011589849164316862, | |
| "loss": 10.3351, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.25918475383800954, | |
| "grad_norm": 0.036410853266716, | |
| "learning_rate": 0.00011566663676344232, | |
| "loss": 10.3414, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.2596082583377448, | |
| "grad_norm": 0.03686416149139404, | |
| "learning_rate": 0.00011543469551368144, | |
| "loss": 10.3375, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.26003176283748014, | |
| "grad_norm": 0.04031093418598175, | |
| "learning_rate": 0.00011520266917257618, | |
| "loss": 10.3361, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.26045526733721547, | |
| "grad_norm": 0.027354370802640915, | |
| "learning_rate": 0.00011497055901928577, | |
| "loss": 10.3334, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.26087877183695074, | |
| "grad_norm": 0.029079321771860123, | |
| "learning_rate": 0.00011473836633343144, | |
| "loss": 10.3376, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.26130227633668607, | |
| "grad_norm": 0.027393948286771774, | |
| "learning_rate": 0.00011450609239508951, | |
| "loss": 10.3359, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.2617257808364214, | |
| "grad_norm": 0.037023283541202545, | |
| "learning_rate": 0.00011427373848478422, | |
| "loss": 10.336, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.2621492853361567, | |
| "grad_norm": 0.04202662780880928, | |
| "learning_rate": 0.00011404130588348072, | |
| "loss": 10.3383, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.262572789835892, | |
| "grad_norm": 0.031701017171144485, | |
| "learning_rate": 0.00011380879587257792, | |
| "loss": 10.3356, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.2629962943356273, | |
| "grad_norm": 0.03459370136260986, | |
| "learning_rate": 0.00011357620973390151, | |
| "loss": 10.3337, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.26341979883536265, | |
| "grad_norm": 0.03404482826590538, | |
| "learning_rate": 0.0001133435487496969, | |
| "loss": 10.3373, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.2638433033350979, | |
| "grad_norm": 0.03435559198260307, | |
| "learning_rate": 0.0001131108142026221, | |
| "loss": 10.3394, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.26426680783483325, | |
| "grad_norm": 0.04172271490097046, | |
| "learning_rate": 0.00011287800737574072, | |
| "loss": 10.3312, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.2646903123345686, | |
| "grad_norm": 0.024423452094197273, | |
| "learning_rate": 0.00011264512955251478, | |
| "loss": 10.3384, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.26511381683430385, | |
| "grad_norm": 0.036313965916633606, | |
| "learning_rate": 0.00011241218201679773, | |
| "loss": 10.3343, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.2655373213340392, | |
| "grad_norm": 0.03670899197459221, | |
| "learning_rate": 0.00011217916605282728, | |
| "loss": 10.3421, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.2659608258337745, | |
| "grad_norm": 0.04206259921193123, | |
| "learning_rate": 0.00011194608294521854, | |
| "loss": 10.3304, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.2663843303335098, | |
| "grad_norm": 0.029241429641842842, | |
| "learning_rate": 0.00011171293397895665, | |
| "loss": 10.3403, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.2668078348332451, | |
| "grad_norm": 0.029772555455565453, | |
| "learning_rate": 0.00011147972043938988, | |
| "loss": 10.3356, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.26723133933298043, | |
| "grad_norm": 0.038933563977479935, | |
| "learning_rate": 0.00011124644361222245, | |
| "loss": 10.3396, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.2676548438327157, | |
| "grad_norm": 0.03326569125056267, | |
| "learning_rate": 0.00011101310478350754, | |
| "loss": 10.337, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.26807834833245103, | |
| "grad_norm": 0.03632461279630661, | |
| "learning_rate": 0.00011077970523964011, | |
| "loss": 10.337, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.26850185283218636, | |
| "grad_norm": 0.03578447178006172, | |
| "learning_rate": 0.00011054624626734984, | |
| "loss": 10.3358, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.26892535733192163, | |
| "grad_norm": 0.032311227172613144, | |
| "learning_rate": 0.0001103127291536941, | |
| "loss": 10.3417, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.26934886183165696, | |
| "grad_norm": 0.03721488639712334, | |
| "learning_rate": 0.00011007915518605067, | |
| "loss": 10.3341, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.2697723663313923, | |
| "grad_norm": 0.026686688885092735, | |
| "learning_rate": 0.00010984552565211089, | |
| "loss": 10.3337, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.27019587083112756, | |
| "grad_norm": 0.03955764323472977, | |
| "learning_rate": 0.00010961184183987233, | |
| "loss": 10.3331, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.2706193753308629, | |
| "grad_norm": 0.024867044761776924, | |
| "learning_rate": 0.00010937810503763191, | |
| "loss": 10.3319, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.2710428798305982, | |
| "grad_norm": 0.026639580726623535, | |
| "learning_rate": 0.00010914431653397856, | |
| "loss": 10.3394, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.2714663843303335, | |
| "grad_norm": 0.04265257716178894, | |
| "learning_rate": 0.00010891047761778637, | |
| "loss": 10.3355, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.2718898888300688, | |
| "grad_norm": 0.03401639685034752, | |
| "learning_rate": 0.00010867658957820723, | |
| "loss": 10.3362, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.27231339332980414, | |
| "grad_norm": 0.03278350457549095, | |
| "learning_rate": 0.00010844265370466393, | |
| "loss": 10.3369, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.2727368978295394, | |
| "grad_norm": 0.03625522553920746, | |
| "learning_rate": 0.00010820867128684292, | |
| "loss": 10.3386, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.27316040232927474, | |
| "grad_norm": 0.028470052406191826, | |
| "learning_rate": 0.0001079746436146873, | |
| "loss": 10.3359, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.27358390682901007, | |
| "grad_norm": 0.03894231840968132, | |
| "learning_rate": 0.00010774057197838963, | |
| "loss": 10.3363, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.27400741132874534, | |
| "grad_norm": 0.04798604175448418, | |
| "learning_rate": 0.00010750645766838477, | |
| "loss": 10.3351, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.27443091582848067, | |
| "grad_norm": 0.038566704839468, | |
| "learning_rate": 0.00010727230197534299, | |
| "loss": 10.3386, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.274854420328216, | |
| "grad_norm": 0.038909364491701126, | |
| "learning_rate": 0.0001070381061901626, | |
| "loss": 10.3376, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.27527792482795127, | |
| "grad_norm": 0.029502833262085915, | |
| "learning_rate": 0.00010680387160396293, | |
| "loss": 10.3356, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.2757014293276866, | |
| "grad_norm": 0.028758224099874496, | |
| "learning_rate": 0.00010656959950807728, | |
| "loss": 10.3313, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.2761249338274219, | |
| "grad_norm": 0.024828476831316948, | |
| "learning_rate": 0.0001063352911940457, | |
| "loss": 10.3318, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.27654843832715725, | |
| "grad_norm": 0.02429981529712677, | |
| "learning_rate": 0.00010610094795360795, | |
| "loss": 10.333, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.2769719428268925, | |
| "grad_norm": 0.028827672824263573, | |
| "learning_rate": 0.00010586657107869626, | |
| "loss": 10.3318, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.27739544732662785, | |
| "grad_norm": 0.04222332313656807, | |
| "learning_rate": 0.00010563216186142839, | |
| "loss": 10.3354, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.2778189518263632, | |
| "grad_norm": 0.04045010358095169, | |
| "learning_rate": 0.00010539772159410036, | |
| "loss": 10.3356, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.27824245632609845, | |
| "grad_norm": 0.02479146048426628, | |
| "learning_rate": 0.00010516325156917926, | |
| "loss": 10.3395, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.2786659608258338, | |
| "grad_norm": 0.036765843629837036, | |
| "learning_rate": 0.00010492875307929644, | |
| "loss": 10.3334, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.2790894653255691, | |
| "grad_norm": 0.02949843928217888, | |
| "learning_rate": 0.00010469422741724003, | |
| "loss": 10.3405, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.2795129698253044, | |
| "grad_norm": 0.02545243129134178, | |
| "learning_rate": 0.000104459675875948, | |
| "loss": 10.3339, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.2799364743250397, | |
| "grad_norm": 0.032835401594638824, | |
| "learning_rate": 0.00010422509974850099, | |
| "loss": 10.3426, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.28035997882477504, | |
| "grad_norm": 0.029005464166402817, | |
| "learning_rate": 0.00010399050032811519, | |
| "loss": 10.3353, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.2807834833245103, | |
| "grad_norm": 0.02459227293729782, | |
| "learning_rate": 0.00010375587890813518, | |
| "loss": 10.3345, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.28120698782424564, | |
| "grad_norm": 0.04449470341205597, | |
| "learning_rate": 0.00010352123678202685, | |
| "loss": 10.3358, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.28163049232398096, | |
| "grad_norm": 0.025347614660859108, | |
| "learning_rate": 0.00010328657524337029, | |
| "loss": 10.3357, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.28205399682371624, | |
| "grad_norm": 0.028995616361498833, | |
| "learning_rate": 0.00010305189558585248, | |
| "loss": 10.3386, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.28247750132345156, | |
| "grad_norm": 0.029563058167696, | |
| "learning_rate": 0.00010281719910326042, | |
| "loss": 10.3369, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.2829010058231869, | |
| "grad_norm": 0.03033272735774517, | |
| "learning_rate": 0.00010258248708947375, | |
| "loss": 10.337, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.28332451032292216, | |
| "grad_norm": 0.03558272868394852, | |
| "learning_rate": 0.00010234776083845787, | |
| "loss": 10.3345, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.2837480148226575, | |
| "grad_norm": 0.023746639490127563, | |
| "learning_rate": 0.00010211302164425655, | |
| "loss": 10.3326, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.2841715193223928, | |
| "grad_norm": 0.02846304513514042, | |
| "learning_rate": 0.00010187827080098498, | |
| "loss": 10.3353, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.2845950238221281, | |
| "grad_norm": 0.035858284682035446, | |
| "learning_rate": 0.00010164350960282252, | |
| "loss": 10.336, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.2850185283218634, | |
| "grad_norm": 0.026505351066589355, | |
| "learning_rate": 0.00010140873934400567, | |
| "loss": 10.3382, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.28544203282159875, | |
| "grad_norm": 0.02379724755883217, | |
| "learning_rate": 0.00010117396131882087, | |
| "loss": 10.3372, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.28544203282159875, | |
| "eval_loss": 10.334465026855469, | |
| "eval_runtime": 3.4817, | |
| "eval_samples_per_second": 285.783, | |
| "eval_steps_per_second": 143.035, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.285865537321334, | |
| "grad_norm": 0.030138272792100906, | |
| "learning_rate": 0.00010093917682159735, | |
| "loss": 10.3361, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.28628904182106935, | |
| "grad_norm": 0.023656543344259262, | |
| "learning_rate": 0.00010070438714670002, | |
| "loss": 10.3345, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.2867125463208047, | |
| "grad_norm": 0.035104621201753616, | |
| "learning_rate": 0.00010046959358852244, | |
| "loss": 10.3347, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.28713605082053995, | |
| "grad_norm": 0.030601153150200844, | |
| "learning_rate": 0.00010023479744147936, | |
| "loss": 10.3325, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.2875595553202753, | |
| "grad_norm": 0.030649134889245033, | |
| "learning_rate": 0.0001, | |
| "loss": 10.3351, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.2879830598200106, | |
| "grad_norm": 0.04906442388892174, | |
| "learning_rate": 9.976520255852065e-05, | |
| "loss": 10.3382, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.2884065643197459, | |
| "grad_norm": 0.036667853593826294, | |
| "learning_rate": 9.953040641147761e-05, | |
| "loss": 10.3336, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.2888300688194812, | |
| "grad_norm": 0.032969675958156586, | |
| "learning_rate": 9.929561285329999e-05, | |
| "loss": 10.3347, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.28925357331921653, | |
| "grad_norm": 0.04049724340438843, | |
| "learning_rate": 9.906082317840266e-05, | |
| "loss": 10.337, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.2896770778189518, | |
| "grad_norm": 0.03217809647321701, | |
| "learning_rate": 9.882603868117917e-05, | |
| "loss": 10.332, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.29010058231868713, | |
| "grad_norm": 0.03514156490564346, | |
| "learning_rate": 9.859126065599434e-05, | |
| "loss": 10.3331, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.29052408681842246, | |
| "grad_norm": 0.02941136807203293, | |
| "learning_rate": 9.83564903971775e-05, | |
| "loss": 10.3325, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.29094759131815773, | |
| "grad_norm": 0.026193542405962944, | |
| "learning_rate": 9.812172919901506e-05, | |
| "loss": 10.3382, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.29137109581789306, | |
| "grad_norm": 0.031999170780181885, | |
| "learning_rate": 9.788697835574347e-05, | |
| "loss": 10.3378, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.2917946003176284, | |
| "grad_norm": 0.0316544771194458, | |
| "learning_rate": 9.765223916154217e-05, | |
| "loss": 10.3369, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.2922181048173637, | |
| "grad_norm": 0.030304009094834328, | |
| "learning_rate": 9.741751291052626e-05, | |
| "loss": 10.3381, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.292641609317099, | |
| "grad_norm": 0.035043906420469284, | |
| "learning_rate": 9.718280089673959e-05, | |
| "loss": 10.3327, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.2930651138168343, | |
| "grad_norm": 0.031086809933185577, | |
| "learning_rate": 9.694810441414754e-05, | |
| "loss": 10.3331, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.29348861831656964, | |
| "grad_norm": 0.03664236515760422, | |
| "learning_rate": 9.671342475662975e-05, | |
| "loss": 10.3384, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.2939121228163049, | |
| "grad_norm": 0.036936696618795395, | |
| "learning_rate": 9.647876321797314e-05, | |
| "loss": 10.3379, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.29433562731604024, | |
| "grad_norm": 0.03095340169966221, | |
| "learning_rate": 9.624412109186484e-05, | |
| "loss": 10.3351, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.29475913181577557, | |
| "grad_norm": 0.026670867577195168, | |
| "learning_rate": 9.600949967188484e-05, | |
| "loss": 10.3324, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.29518263631551084, | |
| "grad_norm": 0.03176816925406456, | |
| "learning_rate": 9.577490025149903e-05, | |
| "loss": 10.336, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.29560614081524617, | |
| "grad_norm": 0.041850414127111435, | |
| "learning_rate": 9.554032412405204e-05, | |
| "loss": 10.3335, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.2960296453149815, | |
| "grad_norm": 0.02709740586578846, | |
| "learning_rate": 9.530577258275998e-05, | |
| "loss": 10.335, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.29645314981471677, | |
| "grad_norm": 0.03338076174259186, | |
| "learning_rate": 9.507124692070355e-05, | |
| "loss": 10.3393, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.2968766543144521, | |
| "grad_norm": 0.03312176465988159, | |
| "learning_rate": 9.483674843082075e-05, | |
| "loss": 10.336, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.2973001588141874, | |
| "grad_norm": 0.026730258017778397, | |
| "learning_rate": 9.460227840589967e-05, | |
| "loss": 10.3366, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.2977236633139227, | |
| "grad_norm": 0.04017185419797897, | |
| "learning_rate": 9.436783813857161e-05, | |
| "loss": 10.3349, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.298147167813658, | |
| "grad_norm": 0.025352856144309044, | |
| "learning_rate": 9.413342892130376e-05, | |
| "loss": 10.331, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.29857067231339335, | |
| "grad_norm": 0.04028523713350296, | |
| "learning_rate": 9.389905204639206e-05, | |
| "loss": 10.3326, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.2989941768131286, | |
| "grad_norm": 0.034634605050086975, | |
| "learning_rate": 9.366470880595434e-05, | |
| "loss": 10.3326, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.29941768131286395, | |
| "grad_norm": 0.037610601633787155, | |
| "learning_rate": 9.343040049192274e-05, | |
| "loss": 10.3342, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.2998411858125993, | |
| "grad_norm": 0.0313008613884449, | |
| "learning_rate": 9.31961283960371e-05, | |
| "loss": 10.3337, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.30026469031233455, | |
| "grad_norm": 0.03718707337975502, | |
| "learning_rate": 9.296189380983747e-05, | |
| "loss": 10.3325, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.3006881948120699, | |
| "grad_norm": 0.03456999734044075, | |
| "learning_rate": 9.272769802465705e-05, | |
| "loss": 10.3325, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.3011116993118052, | |
| "grad_norm": 0.03181077539920807, | |
| "learning_rate": 9.249354233161523e-05, | |
| "loss": 10.3338, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.3015352038115405, | |
| "grad_norm": 0.0410895049571991, | |
| "learning_rate": 9.225942802161042e-05, | |
| "loss": 10.3376, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.3019587083112758, | |
| "grad_norm": 0.05550311505794525, | |
| "learning_rate": 9.202535638531273e-05, | |
| "loss": 10.3373, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.30238221281101113, | |
| "grad_norm": 0.03022390976548195, | |
| "learning_rate": 9.179132871315708e-05, | |
| "loss": 10.3323, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.3028057173107464, | |
| "grad_norm": 0.058899421244859695, | |
| "learning_rate": 9.155734629533611e-05, | |
| "loss": 10.3373, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.30322922181048173, | |
| "grad_norm": 0.0289511289447546, | |
| "learning_rate": 9.132341042179279e-05, | |
| "loss": 10.3365, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.30365272631021706, | |
| "grad_norm": 0.024074682965874672, | |
| "learning_rate": 9.108952238221365e-05, | |
| "loss": 10.3343, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.30407623080995233, | |
| "grad_norm": 0.03383636474609375, | |
| "learning_rate": 9.085568346602145e-05, | |
| "loss": 10.3376, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.30449973530968766, | |
| "grad_norm": 0.03680823743343353, | |
| "learning_rate": 9.062189496236813e-05, | |
| "loss": 10.332, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.304923239809423, | |
| "grad_norm": 0.034177515655756, | |
| "learning_rate": 9.038815816012767e-05, | |
| "loss": 10.3365, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.30534674430915826, | |
| "grad_norm": 0.04184051603078842, | |
| "learning_rate": 9.015447434788915e-05, | |
| "loss": 10.3308, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.3057702488088936, | |
| "grad_norm": 0.031081423163414, | |
| "learning_rate": 8.992084481394934e-05, | |
| "loss": 10.332, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.3061937533086289, | |
| "grad_norm": 0.04926011338829994, | |
| "learning_rate": 8.968727084630594e-05, | |
| "loss": 10.3388, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.30661725780836424, | |
| "grad_norm": 0.03448108211159706, | |
| "learning_rate": 8.945375373265017e-05, | |
| "loss": 10.3371, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.3070407623080995, | |
| "grad_norm": 0.030851799994707108, | |
| "learning_rate": 8.92202947603599e-05, | |
| "loss": 10.3402, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.30746426680783484, | |
| "grad_norm": 0.03434957191348076, | |
| "learning_rate": 8.898689521649251e-05, | |
| "loss": 10.3371, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.30788777130757017, | |
| "grad_norm": 0.034013282507658005, | |
| "learning_rate": 8.875355638777757e-05, | |
| "loss": 10.3344, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.30831127580730544, | |
| "grad_norm": 0.03570681810379028, | |
| "learning_rate": 8.852027956061015e-05, | |
| "loss": 10.3333, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.30873478030704077, | |
| "grad_norm": 0.04296912997961044, | |
| "learning_rate": 8.828706602104337e-05, | |
| "loss": 10.3388, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.3091582848067761, | |
| "grad_norm": 0.037189483642578125, | |
| "learning_rate": 8.805391705478147e-05, | |
| "loss": 10.335, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.30958178930651137, | |
| "grad_norm": 0.02627628669142723, | |
| "learning_rate": 8.782083394717272e-05, | |
| "loss": 10.3354, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.3100052938062467, | |
| "grad_norm": 0.026290280744433403, | |
| "learning_rate": 8.758781798320233e-05, | |
| "loss": 10.3344, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.310428798305982, | |
| "grad_norm": 0.033993784338235855, | |
| "learning_rate": 8.735487044748523e-05, | |
| "loss": 10.3324, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.3108523028057173, | |
| "grad_norm": 0.02894951030611992, | |
| "learning_rate": 8.712199262425927e-05, | |
| "loss": 10.3343, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.3112758073054526, | |
| "grad_norm": 0.02918967790901661, | |
| "learning_rate": 8.68891857973779e-05, | |
| "loss": 10.3364, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.31169931180518795, | |
| "grad_norm": 0.04133673012256622, | |
| "learning_rate": 8.665645125030311e-05, | |
| "loss": 10.3339, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.3121228163049232, | |
| "grad_norm": 0.03206159546971321, | |
| "learning_rate": 8.642379026609849e-05, | |
| "loss": 10.3422, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.31254632080465855, | |
| "grad_norm": 0.03564688563346863, | |
| "learning_rate": 8.619120412742212e-05, | |
| "loss": 10.3388, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.3129698253043939, | |
| "grad_norm": 0.033441901206970215, | |
| "learning_rate": 8.595869411651931e-05, | |
| "loss": 10.3375, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.31339332980412915, | |
| "grad_norm": 0.0351875014603138, | |
| "learning_rate": 8.572626151521581e-05, | |
| "loss": 10.3327, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.3138168343038645, | |
| "grad_norm": 0.046769220381975174, | |
| "learning_rate": 8.549390760491051e-05, | |
| "loss": 10.3333, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.3142403388035998, | |
| "grad_norm": 0.02873465232551098, | |
| "learning_rate": 8.526163366656858e-05, | |
| "loss": 10.3342, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.3146638433033351, | |
| "grad_norm": 0.03012407198548317, | |
| "learning_rate": 8.502944098071427e-05, | |
| "loss": 10.334, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.3150873478030704, | |
| "grad_norm": 0.03743249177932739, | |
| "learning_rate": 8.479733082742384e-05, | |
| "loss": 10.3344, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.31551085230280573, | |
| "grad_norm": 0.02463219314813614, | |
| "learning_rate": 8.456530448631855e-05, | |
| "loss": 10.3322, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.315934356802541, | |
| "grad_norm": 0.035319242626428604, | |
| "learning_rate": 8.433336323655774e-05, | |
| "loss": 10.3363, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.31635786130227633, | |
| "grad_norm": 0.03892083838582039, | |
| "learning_rate": 8.41015083568314e-05, | |
| "loss": 10.3344, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.31678136580201166, | |
| "grad_norm": 0.04276084899902344, | |
| "learning_rate": 8.386974112535358e-05, | |
| "loss": 10.3367, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.31720487030174693, | |
| "grad_norm": 0.03648482635617256, | |
| "learning_rate": 8.363806281985509e-05, | |
| "loss": 10.333, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.31762837480148226, | |
| "grad_norm": 0.03600320592522621, | |
| "learning_rate": 8.340647471757636e-05, | |
| "loss": 10.3314, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.3180518793012176, | |
| "grad_norm": 0.0343911312520504, | |
| "learning_rate": 8.317497809526063e-05, | |
| "loss": 10.3391, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.31847538380095286, | |
| "grad_norm": 0.028392106294631958, | |
| "learning_rate": 8.294357422914685e-05, | |
| "loss": 10.3343, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.3188988883006882, | |
| "grad_norm": 0.03276420384645462, | |
| "learning_rate": 8.27122643949625e-05, | |
| "loss": 10.3329, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.3193223928004235, | |
| "grad_norm": 0.030692044645547867, | |
| "learning_rate": 8.248104986791676e-05, | |
| "loss": 10.3287, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.3197458973001588, | |
| "grad_norm": 0.037886835634708405, | |
| "learning_rate": 8.224993192269334e-05, | |
| "loss": 10.3316, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.3201694017998941, | |
| "grad_norm": 0.029941901564598083, | |
| "learning_rate": 8.201891183344345e-05, | |
| "loss": 10.3293, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.32059290629962944, | |
| "grad_norm": 0.0404081791639328, | |
| "learning_rate": 8.178799087377894e-05, | |
| "loss": 10.3364, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.32101641079936477, | |
| "grad_norm": 0.03296668082475662, | |
| "learning_rate": 8.1557170316765e-05, | |
| "loss": 10.3363, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.32143991529910004, | |
| "grad_norm": 0.03453279659152031, | |
| "learning_rate": 8.132645143491346e-05, | |
| "loss": 10.3369, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.32186341979883537, | |
| "grad_norm": 0.042309049516916275, | |
| "learning_rate": 8.10958355001755e-05, | |
| "loss": 10.3325, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.3222869242985707, | |
| "grad_norm": 0.03626590222120285, | |
| "learning_rate": 8.086532378393482e-05, | |
| "loss": 10.3374, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.32271042879830597, | |
| "grad_norm": 0.029558565467596054, | |
| "learning_rate": 8.063491755700051e-05, | |
| "loss": 10.3367, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.3231339332980413, | |
| "grad_norm": 0.031136656180024147, | |
| "learning_rate": 8.04046180896002e-05, | |
| "loss": 10.3312, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.3235574377977766, | |
| "grad_norm": 0.03206343576312065, | |
| "learning_rate": 8.017442665137278e-05, | |
| "loss": 10.3357, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.3239809422975119, | |
| "grad_norm": 0.04191575571894646, | |
| "learning_rate": 7.994434451136177e-05, | |
| "loss": 10.3358, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.3244044467972472, | |
| "grad_norm": 0.03315071761608124, | |
| "learning_rate": 7.971437293800803e-05, | |
| "loss": 10.3338, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.32482795129698255, | |
| "grad_norm": 0.03882451355457306, | |
| "learning_rate": 7.948451319914282e-05, | |
| "loss": 10.3311, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.3252514557967178, | |
| "grad_norm": 0.046539660543203354, | |
| "learning_rate": 7.925476656198095e-05, | |
| "loss": 10.3364, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.32567496029645315, | |
| "grad_norm": 0.035186078399419785, | |
| "learning_rate": 7.90251342931137e-05, | |
| "loss": 10.3322, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.3260984647961885, | |
| "grad_norm": 0.02894584834575653, | |
| "learning_rate": 7.879561765850176e-05, | |
| "loss": 10.335, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.32652196929592375, | |
| "grad_norm": 0.05743710324168205, | |
| "learning_rate": 7.856621792346837e-05, | |
| "loss": 10.3358, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.3269454737956591, | |
| "grad_norm": 0.03184637799859047, | |
| "learning_rate": 7.833693635269235e-05, | |
| "loss": 10.3323, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.3273689782953944, | |
| "grad_norm": 0.028403330594301224, | |
| "learning_rate": 7.8107774210201e-05, | |
| "loss": 10.3368, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.3277924827951297, | |
| "grad_norm": 0.03520669415593147, | |
| "learning_rate": 7.78787327593633e-05, | |
| "loss": 10.3329, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.328215987294865, | |
| "grad_norm": 0.033936478197574615, | |
| "learning_rate": 7.764981326288273e-05, | |
| "loss": 10.3354, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.32863949179460034, | |
| "grad_norm": 0.04284480959177017, | |
| "learning_rate": 7.74210169827906e-05, | |
| "loss": 10.3353, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.3290629962943356, | |
| "grad_norm": 0.035401035100221634, | |
| "learning_rate": 7.719234518043881e-05, | |
| "loss": 10.3383, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 0.32948650079407094, | |
| "grad_norm": 0.02768767438828945, | |
| "learning_rate": 7.696379911649303e-05, | |
| "loss": 10.333, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.32991000529380626, | |
| "grad_norm": 0.03562779724597931, | |
| "learning_rate": 7.673538005092578e-05, | |
| "loss": 10.3365, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 0.33033350979354154, | |
| "grad_norm": 0.03725546598434448, | |
| "learning_rate": 7.650708924300944e-05, | |
| "loss": 10.3284, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.33075701429327686, | |
| "grad_norm": 0.032989148050546646, | |
| "learning_rate": 7.627892795130925e-05, | |
| "loss": 10.3375, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 0.3311805187930122, | |
| "grad_norm": 0.02557358518242836, | |
| "learning_rate": 7.605089743367644e-05, | |
| "loss": 10.3355, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.33160402329274746, | |
| "grad_norm": 0.03748362138867378, | |
| "learning_rate": 7.582299894724138e-05, | |
| "loss": 10.3362, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 0.3320275277924828, | |
| "grad_norm": 0.04423379525542259, | |
| "learning_rate": 7.55952337484064e-05, | |
| "loss": 10.3372, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.3324510322922181, | |
| "grad_norm": 0.03931692987680435, | |
| "learning_rate": 7.536760309283912e-05, | |
| "loss": 10.3319, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.3328745367919534, | |
| "grad_norm": 0.028346918523311615, | |
| "learning_rate": 7.514010823546543e-05, | |
| "loss": 10.3355, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.3332980412916887, | |
| "grad_norm": 0.041941095143556595, | |
| "learning_rate": 7.491275043046246e-05, | |
| "loss": 10.3351, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 0.33372154579142405, | |
| "grad_norm": 0.03487636148929596, | |
| "learning_rate": 7.46855309312519e-05, | |
| "loss": 10.3333, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.3341450502911593, | |
| "grad_norm": 0.032287437468767166, | |
| "learning_rate": 7.445845099049294e-05, | |
| "loss": 10.3308, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 0.33456855479089465, | |
| "grad_norm": 0.03427242115139961, | |
| "learning_rate": 7.423151186007527e-05, | |
| "loss": 10.3318, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.33499205929063, | |
| "grad_norm": 0.03202645853161812, | |
| "learning_rate": 7.400471479111247e-05, | |
| "loss": 10.3365, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 0.33541556379036525, | |
| "grad_norm": 0.036663834005594254, | |
| "learning_rate": 7.377806103393473e-05, | |
| "loss": 10.3315, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.3358390682901006, | |
| "grad_norm": 0.037792034447193146, | |
| "learning_rate": 7.355155183808234e-05, | |
| "loss": 10.3371, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 0.3362625727898359, | |
| "grad_norm": 0.03239692375063896, | |
| "learning_rate": 7.332518845229859e-05, | |
| "loss": 10.3333, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.33668607728957123, | |
| "grad_norm": 0.028021618723869324, | |
| "learning_rate": 7.309897212452279e-05, | |
| "loss": 10.3329, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.3371095817893065, | |
| "grad_norm": 0.03356965258717537, | |
| "learning_rate": 7.287290410188373e-05, | |
| "loss": 10.3318, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.33753308628904183, | |
| "grad_norm": 0.030086075887084007, | |
| "learning_rate": 7.264698563069246e-05, | |
| "loss": 10.3378, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 0.33795659078877716, | |
| "grad_norm": 0.03997505083680153, | |
| "learning_rate": 7.242121795643552e-05, | |
| "loss": 10.3386, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.33838009528851243, | |
| "grad_norm": 0.03527563437819481, | |
| "learning_rate": 7.219560232376821e-05, | |
| "loss": 10.338, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 0.33880359978824776, | |
| "grad_norm": 0.032303664833307266, | |
| "learning_rate": 7.197013997650762e-05, | |
| "loss": 10.3403, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.3392271042879831, | |
| "grad_norm": 0.03437490016222, | |
| "learning_rate": 7.174483215762568e-05, | |
| "loss": 10.3319, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 0.33965060878771836, | |
| "grad_norm": 0.03714921697974205, | |
| "learning_rate": 7.151968010924249e-05, | |
| "loss": 10.3357, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.3400741132874537, | |
| "grad_norm": 0.03493595868349075, | |
| "learning_rate": 7.12946850726194e-05, | |
| "loss": 10.333, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 0.340497617787189, | |
| "grad_norm": 0.027758195996284485, | |
| "learning_rate": 7.106984828815206e-05, | |
| "loss": 10.3392, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.3409211222869243, | |
| "grad_norm": 0.03370975703001022, | |
| "learning_rate": 7.084517099536377e-05, | |
| "loss": 10.3326, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.3413446267866596, | |
| "grad_norm": 0.03559848666191101, | |
| "learning_rate": 7.062065443289859e-05, | |
| "loss": 10.3339, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.34176813128639494, | |
| "grad_norm": 0.03199275955557823, | |
| "learning_rate": 7.039629983851432e-05, | |
| "loss": 10.3325, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 0.3421916357861302, | |
| "grad_norm": 0.0587792843580246, | |
| "learning_rate": 7.017210844907598e-05, | |
| "loss": 10.3334, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.34261514028586554, | |
| "grad_norm": 0.04129471629858017, | |
| "learning_rate": 6.994808150054872e-05, | |
| "loss": 10.3343, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 0.34303864478560087, | |
| "grad_norm": 0.03535553812980652, | |
| "learning_rate": 6.972422022799121e-05, | |
| "loss": 10.3325, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.34346214928533614, | |
| "grad_norm": 0.03517236188054085, | |
| "learning_rate": 6.95005258655488e-05, | |
| "loss": 10.3332, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 0.34388565378507147, | |
| "grad_norm": 0.03364865854382515, | |
| "learning_rate": 6.927699964644652e-05, | |
| "loss": 10.3341, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.3443091582848068, | |
| "grad_norm": 0.04013295844197273, | |
| "learning_rate": 6.905364280298252e-05, | |
| "loss": 10.3285, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 0.34473266278454207, | |
| "grad_norm": 0.03748088330030441, | |
| "learning_rate": 6.883045656652122e-05, | |
| "loss": 10.3321, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.3451561672842774, | |
| "grad_norm": 0.032113492488861084, | |
| "learning_rate": 6.860744216748634e-05, | |
| "loss": 10.3329, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.3455796717840127, | |
| "grad_norm": 0.031743258237838745, | |
| "learning_rate": 6.838460083535445e-05, | |
| "loss": 10.3353, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.346003176283748, | |
| "grad_norm": 0.0373242124915123, | |
| "learning_rate": 6.816193379864786e-05, | |
| "loss": 10.3358, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 0.3464266807834833, | |
| "grad_norm": 0.03148363158106804, | |
| "learning_rate": 6.793944228492803e-05, | |
| "loss": 10.3392, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.34685018528321865, | |
| "grad_norm": 0.02826239913702011, | |
| "learning_rate": 6.77171275207888e-05, | |
| "loss": 10.3354, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 0.3472736897829539, | |
| "grad_norm": 0.02493412233889103, | |
| "learning_rate": 6.749499073184957e-05, | |
| "loss": 10.3331, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.34769719428268925, | |
| "grad_norm": 0.030130870640277863, | |
| "learning_rate": 6.727303314274852e-05, | |
| "loss": 10.3335, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 0.3481206987824246, | |
| "grad_norm": 0.03829304128885269, | |
| "learning_rate": 6.705125597713598e-05, | |
| "loss": 10.337, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.34854420328215985, | |
| "grad_norm": 0.036645397543907166, | |
| "learning_rate": 6.682966045766758e-05, | |
| "loss": 10.3323, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 0.3489677077818952, | |
| "grad_norm": 0.03612329065799713, | |
| "learning_rate": 6.660824780599744e-05, | |
| "loss": 10.3288, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.3493912122816305, | |
| "grad_norm": 0.04094702750444412, | |
| "learning_rate": 6.638701924277174e-05, | |
| "loss": 10.3292, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.3498147167813658, | |
| "grad_norm": 0.031782373785972595, | |
| "learning_rate": 6.61659759876215e-05, | |
| "loss": 10.333, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.3502382212811011, | |
| "grad_norm": 0.03760769963264465, | |
| "learning_rate": 6.594511925915646e-05, | |
| "loss": 10.3337, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 0.35066172578083643, | |
| "grad_norm": 0.033052217215299606, | |
| "learning_rate": 6.572445027495779e-05, | |
| "loss": 10.3336, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.35108523028057176, | |
| "grad_norm": 0.0381859727203846, | |
| "learning_rate": 6.550397025157169e-05, | |
| "loss": 10.3385, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 0.35150873478030703, | |
| "grad_norm": 0.033314503729343414, | |
| "learning_rate": 6.528368040450268e-05, | |
| "loss": 10.3333, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.35193223928004236, | |
| "grad_norm": 0.029306232929229736, | |
| "learning_rate": 6.506358194820685e-05, | |
| "loss": 10.3326, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 0.3523557437797777, | |
| "grad_norm": 0.035479675978422165, | |
| "learning_rate": 6.484367609608503e-05, | |
| "loss": 10.3346, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.35277924827951296, | |
| "grad_norm": 0.028150904923677444, | |
| "learning_rate": 6.462396406047634e-05, | |
| "loss": 10.336, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 0.3532027527792483, | |
| "grad_norm": 0.029760006815195084, | |
| "learning_rate": 6.440444705265136e-05, | |
| "loss": 10.3317, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.3536262572789836, | |
| "grad_norm": 0.039765894412994385, | |
| "learning_rate": 6.418512628280544e-05, | |
| "loss": 10.3309, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.3540497617787189, | |
| "grad_norm": 0.02911820076406002, | |
| "learning_rate": 6.396600296005213e-05, | |
| "loss": 10.3351, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.3544732662784542, | |
| "grad_norm": 0.0354015938937664, | |
| "learning_rate": 6.374707829241648e-05, | |
| "loss": 10.3336, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 0.35489677077818954, | |
| "grad_norm": 0.030309785157442093, | |
| "learning_rate": 6.352835348682823e-05, | |
| "loss": 10.3339, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.3553202752779248, | |
| "grad_norm": 0.03831500932574272, | |
| "learning_rate": 6.330982974911542e-05, | |
| "loss": 10.3343, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 0.35574377977766014, | |
| "grad_norm": 0.02785351127386093, | |
| "learning_rate": 6.309150828399754e-05, | |
| "loss": 10.3333, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.35616728427739547, | |
| "grad_norm": 0.033175136893987656, | |
| "learning_rate": 6.287339029507894e-05, | |
| "loss": 10.3336, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 0.35659078877713074, | |
| "grad_norm": 0.03539146110415459, | |
| "learning_rate": 6.265547698484226e-05, | |
| "loss": 10.3291, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.35701429327686607, | |
| "grad_norm": 0.033553339540958405, | |
| "learning_rate": 6.243776955464169e-05, | |
| "loss": 10.3332, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 0.3574377977766014, | |
| "grad_norm": 0.02890482172369957, | |
| "learning_rate": 6.22202692046964e-05, | |
| "loss": 10.3323, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.35786130227633667, | |
| "grad_norm": 0.035188328474760056, | |
| "learning_rate": 6.200297713408405e-05, | |
| "loss": 10.3333, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.358284806776072, | |
| "grad_norm": 0.025319932028651237, | |
| "learning_rate": 6.178589454073386e-05, | |
| "loss": 10.3335, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.3587083112758073, | |
| "grad_norm": 0.03855932876467705, | |
| "learning_rate": 6.156902262142041e-05, | |
| "loss": 10.3339, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 0.3591318157755426, | |
| "grad_norm": 0.03646783158183098, | |
| "learning_rate": 6.135236257175668e-05, | |
| "loss": 10.3318, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.3595553202752779, | |
| "grad_norm": 0.030003085732460022, | |
| "learning_rate": 6.11359155861877e-05, | |
| "loss": 10.3356, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 0.35997882477501325, | |
| "grad_norm": 0.0374993160367012, | |
| "learning_rate": 6.091968285798379e-05, | |
| "loss": 10.3337, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.3604023292747485, | |
| "grad_norm": 0.030076105147600174, | |
| "learning_rate": 6.0703665579234235e-05, | |
| "loss": 10.3314, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 0.36082583377448385, | |
| "grad_norm": 0.03631633147597313, | |
| "learning_rate": 6.048786494084036e-05, | |
| "loss": 10.3328, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.3612493382742192, | |
| "grad_norm": 0.040016159415245056, | |
| "learning_rate": 6.027228213250926e-05, | |
| "loss": 10.3283, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 0.36167284277395445, | |
| "grad_norm": 0.0370076447725296, | |
| "learning_rate": 6.005691834274716e-05, | |
| "loss": 10.332, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.3620963472736898, | |
| "grad_norm": 0.030154719948768616, | |
| "learning_rate": 5.984177475885272e-05, | |
| "loss": 10.3317, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.3625198517734251, | |
| "grad_norm": 0.03587184473872185, | |
| "learning_rate": 5.962685256691071e-05, | |
| "loss": 10.3323, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.3629433562731604, | |
| "grad_norm": 0.03230219706892967, | |
| "learning_rate": 5.941215295178537e-05, | |
| "loss": 10.3358, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 0.3633668607728957, | |
| "grad_norm": 0.03372441977262497, | |
| "learning_rate": 5.919767709711381e-05, | |
| "loss": 10.3354, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.36379036527263103, | |
| "grad_norm": 0.03200405836105347, | |
| "learning_rate": 5.898342618529955e-05, | |
| "loss": 10.3328, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 0.3642138697723663, | |
| "grad_norm": 0.029012855142354965, | |
| "learning_rate": 5.876940139750612e-05, | |
| "loss": 10.332, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.36463737427210163, | |
| "grad_norm": 0.03257838636636734, | |
| "learning_rate": 5.8555603913650246e-05, | |
| "loss": 10.3345, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 0.36506087877183696, | |
| "grad_norm": 0.03526037186384201, | |
| "learning_rate": 5.834203491239574e-05, | |
| "loss": 10.3391, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.36548438327157223, | |
| "grad_norm": 0.02882193773984909, | |
| "learning_rate": 5.812869557114658e-05, | |
| "loss": 10.3312, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 0.36590788777130756, | |
| "grad_norm": 0.036499861627817154, | |
| "learning_rate": 5.791558706604074e-05, | |
| "loss": 10.3337, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.3663313922710429, | |
| "grad_norm": 0.029512615874409676, | |
| "learning_rate": 5.7702710571943696e-05, | |
| "loss": 10.3326, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.3667548967707782, | |
| "grad_norm": 0.029680045321583748, | |
| "learning_rate": 5.7490067262441615e-05, | |
| "loss": 10.3327, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.3671784012705135, | |
| "grad_norm": 0.03450106456875801, | |
| "learning_rate": 5.727765830983525e-05, | |
| "loss": 10.3335, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 0.3676019057702488, | |
| "grad_norm": 0.04042774438858032, | |
| "learning_rate": 5.7065484885133466e-05, | |
| "loss": 10.3325, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.36802541026998414, | |
| "grad_norm": 0.026330476626753807, | |
| "learning_rate": 5.685354815804638e-05, | |
| "loss": 10.3357, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 0.3684489147697194, | |
| "grad_norm": 0.04447445273399353, | |
| "learning_rate": 5.664184929697945e-05, | |
| "loss": 10.3358, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.36887241926945474, | |
| "grad_norm": 0.03038848005235195, | |
| "learning_rate": 5.643038946902668e-05, | |
| "loss": 10.3287, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 0.36929592376919007, | |
| "grad_norm": 0.03465115278959274, | |
| "learning_rate": 5.621916983996429e-05, | |
| "loss": 10.3332, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.36971942826892534, | |
| "grad_norm": 0.03414342552423477, | |
| "learning_rate": 5.600819157424427e-05, | |
| "loss": 10.3313, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 0.37014293276866067, | |
| "grad_norm": 0.03694528341293335, | |
| "learning_rate": 5.579745583498801e-05, | |
| "loss": 10.3327, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.370566437268396, | |
| "grad_norm": 0.03394203633069992, | |
| "learning_rate": 5.558696378397983e-05, | |
| "loss": 10.3348, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.37098994176813127, | |
| "grad_norm": 0.03003198839724064, | |
| "learning_rate": 5.537671658166063e-05, | |
| "loss": 10.3372, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.3714134462678666, | |
| "grad_norm": 0.03449974209070206, | |
| "learning_rate": 5.51667153871214e-05, | |
| "loss": 10.3282, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 0.3718369507676019, | |
| "grad_norm": 0.03920021653175354, | |
| "learning_rate": 5.495696135809696e-05, | |
| "loss": 10.3374, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.3722604552673372, | |
| "grad_norm": 0.02695903740823269, | |
| "learning_rate": 5.4747455650959464e-05, | |
| "loss": 10.3328, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 0.3726839597670725, | |
| "grad_norm": 0.02427126094698906, | |
| "learning_rate": 5.453819942071211e-05, | |
| "loss": 10.3338, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.37310746426680785, | |
| "grad_norm": 0.04181993380188942, | |
| "learning_rate": 5.432919382098267e-05, | |
| "loss": 10.3335, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 0.3735309687665431, | |
| "grad_norm": 0.03968915343284607, | |
| "learning_rate": 5.412044000401726e-05, | |
| "loss": 10.3298, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.37395447326627845, | |
| "grad_norm": 0.04720834270119667, | |
| "learning_rate": 5.391193912067386e-05, | |
| "loss": 10.3303, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 0.3743779777660138, | |
| "grad_norm": 0.03866041824221611, | |
| "learning_rate": 5.3703692320416034e-05, | |
| "loss": 10.3328, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.37480148226574905, | |
| "grad_norm": 0.031206615269184113, | |
| "learning_rate": 5.3495700751306735e-05, | |
| "loss": 10.3295, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.3752249867654844, | |
| "grad_norm": 0.02863953448832035, | |
| "learning_rate": 5.328796556000153e-05, | |
| "loss": 10.3352, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.3756484912652197, | |
| "grad_norm": 0.027401378378272057, | |
| "learning_rate": 5.308048789174289e-05, | |
| "loss": 10.3319, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 0.376071995764955, | |
| "grad_norm": 0.039630185812711716, | |
| "learning_rate": 5.2873268890353424e-05, | |
| "loss": 10.3303, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 0.3764955002646903, | |
| "grad_norm": 0.037442997097969055, | |
| "learning_rate": 5.266630969822958e-05, | |
| "loss": 10.3304, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 0.37691900476442564, | |
| "grad_norm": 0.03644905984401703, | |
| "learning_rate": 5.2459611456335746e-05, | |
| "loss": 10.3322, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.3773425092641609, | |
| "grad_norm": 0.03049355559051037, | |
| "learning_rate": 5.225317530419751e-05, | |
| "loss": 10.3303, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 0.37776601376389624, | |
| "grad_norm": 0.0509609617292881, | |
| "learning_rate": 5.2047002379895636e-05, | |
| "loss": 10.3276, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 0.37818951826363156, | |
| "grad_norm": 0.039859797805547714, | |
| "learning_rate": 5.1841093820059686e-05, | |
| "loss": 10.3278, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 0.37861302276336684, | |
| "grad_norm": 0.040018096566200256, | |
| "learning_rate": 5.163545075986178e-05, | |
| "loss": 10.3321, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 0.37903652726310216, | |
| "grad_norm": 0.030337341129779816, | |
| "learning_rate": 5.143007433301035e-05, | |
| "loss": 10.3373, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.3794600317628375, | |
| "grad_norm": 0.0383714959025383, | |
| "learning_rate": 5.12249656717439e-05, | |
| "loss": 10.3338, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.37988353626257276, | |
| "grad_norm": 0.03546814247965813, | |
| "learning_rate": 5.10201259068247e-05, | |
| "loss": 10.3348, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 0.3803070407623081, | |
| "grad_norm": 0.025767376646399498, | |
| "learning_rate": 5.081555616753264e-05, | |
| "loss": 10.336, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 0.3807305452620434, | |
| "grad_norm": 0.03003775328397751, | |
| "learning_rate": 5.061125758165896e-05, | |
| "loss": 10.3323, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 0.38115404976177875, | |
| "grad_norm": 0.04286766052246094, | |
| "learning_rate": 5.040723127549998e-05, | |
| "loss": 10.3369, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.381577554261514, | |
| "grad_norm": 0.03410165011882782, | |
| "learning_rate": 5.0203478373850955e-05, | |
| "loss": 10.3316, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 0.38200105876124935, | |
| "grad_norm": 0.03914531320333481, | |
| "learning_rate": 5.000000000000002e-05, | |
| "loss": 10.3333, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 0.3824245632609847, | |
| "grad_norm": 0.02953316643834114, | |
| "learning_rate": 4.979679727572159e-05, | |
| "loss": 10.3354, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 0.38284806776071995, | |
| "grad_norm": 0.034125540405511856, | |
| "learning_rate": 4.959387132127054e-05, | |
| "loss": 10.3298, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 0.3832715722604553, | |
| "grad_norm": 0.035765476524829865, | |
| "learning_rate": 4.939122325537604e-05, | |
| "loss": 10.3343, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.3836950767601906, | |
| "grad_norm": 0.04484931752085686, | |
| "learning_rate": 4.918885419523499e-05, | |
| "loss": 10.3357, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 0.3841185812599259, | |
| "grad_norm": 0.033943966031074524, | |
| "learning_rate": 4.898676525650639e-05, | |
| "loss": 10.3321, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 0.3845420857596612, | |
| "grad_norm": 0.03354055806994438, | |
| "learning_rate": 4.8784957553304876e-05, | |
| "loss": 10.3308, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 0.38496559025939653, | |
| "grad_norm": 0.029053689911961555, | |
| "learning_rate": 4.858343219819442e-05, | |
| "loss": 10.3289, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 0.3853890947591318, | |
| "grad_norm": 0.04069928824901581, | |
| "learning_rate": 4.838219030218274e-05, | |
| "loss": 10.3315, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.38581259925886713, | |
| "grad_norm": 0.035262517631053925, | |
| "learning_rate": 4.818123297471463e-05, | |
| "loss": 10.3373, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 0.38623610375860246, | |
| "grad_norm": 0.034540239721536636, | |
| "learning_rate": 4.7980561323666115e-05, | |
| "loss": 10.323, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 0.38665960825833773, | |
| "grad_norm": 0.031878579407930374, | |
| "learning_rate": 4.77801764553383e-05, | |
| "loss": 10.3362, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 0.38708311275807306, | |
| "grad_norm": 0.029519235715270042, | |
| "learning_rate": 4.758007947445125e-05, | |
| "loss": 10.3275, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 0.3875066172578084, | |
| "grad_norm": 0.03876268118619919, | |
| "learning_rate": 4.7380271484137915e-05, | |
| "loss": 10.3288, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.38793012175754366, | |
| "grad_norm": 0.029615303501486778, | |
| "learning_rate": 4.718075358593802e-05, | |
| "loss": 10.3347, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 0.388353626257279, | |
| "grad_norm": 0.030300240963697433, | |
| "learning_rate": 4.698152687979205e-05, | |
| "loss": 10.3329, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 0.3887771307570143, | |
| "grad_norm": 0.05134044587612152, | |
| "learning_rate": 4.678259246403512e-05, | |
| "loss": 10.3394, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 0.3892006352567496, | |
| "grad_norm": 0.04115286096930504, | |
| "learning_rate": 4.6583951435390973e-05, | |
| "loss": 10.3301, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 0.3896241397564849, | |
| "grad_norm": 0.033398233354091644, | |
| "learning_rate": 4.638560488896589e-05, | |
| "loss": 10.3336, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.39004764425622024, | |
| "grad_norm": 0.0268535315990448, | |
| "learning_rate": 4.618755391824268e-05, | |
| "loss": 10.3314, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 0.3904711487559555, | |
| "grad_norm": 0.042327847331762314, | |
| "learning_rate": 4.598979961507471e-05, | |
| "loss": 10.3317, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 0.39089465325569084, | |
| "grad_norm": 0.033836785703897476, | |
| "learning_rate": 4.57923430696797e-05, | |
| "loss": 10.3344, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 0.39131815775542617, | |
| "grad_norm": 0.03876091167330742, | |
| "learning_rate": 4.5595185370633875e-05, | |
| "loss": 10.3312, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.39174166225516144, | |
| "grad_norm": 0.04275533929467201, | |
| "learning_rate": 4.5398327604866054e-05, | |
| "loss": 10.3328, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.39216516675489677, | |
| "grad_norm": 0.038993559777736664, | |
| "learning_rate": 4.5201770857651274e-05, | |
| "loss": 10.3345, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 0.3925886712546321, | |
| "grad_norm": 0.028194980695843697, | |
| "learning_rate": 4.50055162126053e-05, | |
| "loss": 10.3356, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 0.39301217575436737, | |
| "grad_norm": 0.038792677223682404, | |
| "learning_rate": 4.48095647516783e-05, | |
| "loss": 10.3328, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.3934356802541027, | |
| "grad_norm": 0.031241275370121002, | |
| "learning_rate": 4.461391755514899e-05, | |
| "loss": 10.3274, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 0.393859184753838, | |
| "grad_norm": 0.04370317608118057, | |
| "learning_rate": 4.4418575701618715e-05, | |
| "loss": 10.3334, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.3942826892535733, | |
| "grad_norm": 0.032410670071840286, | |
| "learning_rate": 4.422354026800536e-05, | |
| "loss": 10.3373, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 0.3947061937533086, | |
| "grad_norm": 0.02156672440469265, | |
| "learning_rate": 4.4028812329537694e-05, | |
| "loss": 10.3344, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 0.39512969825304395, | |
| "grad_norm": 0.042322322726249695, | |
| "learning_rate": 4.3834392959749146e-05, | |
| "loss": 10.3309, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 0.3955532027527793, | |
| "grad_norm": 0.027538040652871132, | |
| "learning_rate": 4.3640283230472044e-05, | |
| "loss": 10.3305, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 0.39597670725251455, | |
| "grad_norm": 0.026913011446595192, | |
| "learning_rate": 4.344648421183166e-05, | |
| "loss": 10.3326, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.3964002117522499, | |
| "grad_norm": 0.03797266632318497, | |
| "learning_rate": 4.3252996972240324e-05, | |
| "loss": 10.3286, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 0.3968237162519852, | |
| "grad_norm": 0.03437899798154831, | |
| "learning_rate": 4.305982257839154e-05, | |
| "loss": 10.3333, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 0.3972472207517205, | |
| "grad_norm": 0.032235968858003616, | |
| "learning_rate": 4.286696209525409e-05, | |
| "loss": 10.3373, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 0.3976707252514558, | |
| "grad_norm": 0.03257599472999573, | |
| "learning_rate": 4.2674416586066165e-05, | |
| "loss": 10.3336, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 0.39809422975119113, | |
| "grad_norm": 0.03536880016326904, | |
| "learning_rate": 4.248218711232952e-05, | |
| "loss": 10.3347, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.3985177342509264, | |
| "grad_norm": 0.03932619467377663, | |
| "learning_rate": 4.229027473380355e-05, | |
| "loss": 10.3343, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 0.39894123875066173, | |
| "grad_norm": 0.03219004347920418, | |
| "learning_rate": 4.2098680508499665e-05, | |
| "loss": 10.3355, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 0.39936474325039706, | |
| "grad_norm": 0.03659631311893463, | |
| "learning_rate": 4.1907405492675065e-05, | |
| "loss": 10.3342, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 0.39978824775013233, | |
| "grad_norm": 0.02803085185587406, | |
| "learning_rate": 4.171645074082737e-05, | |
| "loss": 10.3313, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 0.40021175224986766, | |
| "grad_norm": 0.024601435288786888, | |
| "learning_rate": 4.15258173056885e-05, | |
| "loss": 10.3333, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.400635256749603, | |
| "grad_norm": 0.036193400621414185, | |
| "learning_rate": 4.133550623821885e-05, | |
| "loss": 10.3359, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 0.40105876124933826, | |
| "grad_norm": 0.03234044834971428, | |
| "learning_rate": 4.114551858760183e-05, | |
| "loss": 10.3351, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 0.4014822657490736, | |
| "grad_norm": 0.03343448042869568, | |
| "learning_rate": 4.095585540123762e-05, | |
| "loss": 10.3276, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 0.4019057702488089, | |
| "grad_norm": 0.030924122780561447, | |
| "learning_rate": 4.076651772473783e-05, | |
| "loss": 10.3379, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 0.4023292747485442, | |
| "grad_norm": 0.044260427355766296, | |
| "learning_rate": 4.0577506601919467e-05, | |
| "loss": 10.3332, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.4027527792482795, | |
| "grad_norm": 0.027923841029405594, | |
| "learning_rate": 4.038882307479912e-05, | |
| "loss": 10.3391, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 0.40317628374801484, | |
| "grad_norm": 0.0312094334512949, | |
| "learning_rate": 4.0200468183587556e-05, | |
| "loss": 10.3327, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 0.4035997882477501, | |
| "grad_norm": 0.03471310809254646, | |
| "learning_rate": 4.0012442966683674e-05, | |
| "loss": 10.3367, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 0.40402329274748544, | |
| "grad_norm": 0.03101627714931965, | |
| "learning_rate": 3.982474846066886e-05, | |
| "loss": 10.3284, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 0.40444679724722077, | |
| "grad_norm": 0.03931306675076485, | |
| "learning_rate": 3.963738570030134e-05, | |
| "loss": 10.3312, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.40487030174695604, | |
| "grad_norm": 0.024929361417889595, | |
| "learning_rate": 3.94503557185104e-05, | |
| "loss": 10.3326, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 0.40529380624669137, | |
| "grad_norm": 0.043676454573869705, | |
| "learning_rate": 3.926365954639073e-05, | |
| "loss": 10.3289, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 0.4057173107464267, | |
| "grad_norm": 0.03379151597619057, | |
| "learning_rate": 3.90772982131967e-05, | |
| "loss": 10.3342, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 0.40614081524616197, | |
| "grad_norm": 0.03445500135421753, | |
| "learning_rate": 3.8891272746336845e-05, | |
| "loss": 10.337, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 0.4065643197458973, | |
| "grad_norm": 0.03671969100832939, | |
| "learning_rate": 3.8705584171367885e-05, | |
| "loss": 10.3389, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.4069878242456326, | |
| "grad_norm": 0.03856462240219116, | |
| "learning_rate": 3.8520233511989324e-05, | |
| "loss": 10.3318, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 0.4074113287453679, | |
| "grad_norm": 0.037579286843538284, | |
| "learning_rate": 3.833522179003788e-05, | |
| "loss": 10.3312, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 0.4078348332451032, | |
| "grad_norm": 0.0324142761528492, | |
| "learning_rate": 3.8150550025481445e-05, | |
| "loss": 10.3357, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 0.40825833774483855, | |
| "grad_norm": 0.035630084574222565, | |
| "learning_rate": 3.796621923641404e-05, | |
| "loss": 10.3304, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 0.4086818422445738, | |
| "grad_norm": 0.029326455667614937, | |
| "learning_rate": 3.77822304390496e-05, | |
| "loss": 10.3306, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.40910534674430915, | |
| "grad_norm": 0.03198442980647087, | |
| "learning_rate": 3.7598584647716804e-05, | |
| "loss": 10.3319, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 0.4095288512440445, | |
| "grad_norm": 0.035467833280563354, | |
| "learning_rate": 3.7415282874853444e-05, | |
| "loss": 10.3316, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 0.40995235574377975, | |
| "grad_norm": 0.047377362847328186, | |
| "learning_rate": 3.723232613100046e-05, | |
| "loss": 10.3287, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 0.4103758602435151, | |
| "grad_norm": 0.036050811409950256, | |
| "learning_rate": 3.704971542479695e-05, | |
| "loss": 10.3347, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 0.4107993647432504, | |
| "grad_norm": 0.037851523607969284, | |
| "learning_rate": 3.6867451762974114e-05, | |
| "loss": 10.3334, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.41122286924298573, | |
| "grad_norm": 0.030836213380098343, | |
| "learning_rate": 3.6685536150349986e-05, | |
| "loss": 10.3328, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 0.411646373742721, | |
| "grad_norm": 0.026154899969697, | |
| "learning_rate": 3.650396958982377e-05, | |
| "loss": 10.3323, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.41206987824245633, | |
| "grad_norm": 0.036884456872940063, | |
| "learning_rate": 3.6322753082370365e-05, | |
| "loss": 10.33, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 0.41249338274219166, | |
| "grad_norm": 0.041880205273628235, | |
| "learning_rate": 3.614188762703482e-05, | |
| "loss": 10.3294, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 0.41291688724192693, | |
| "grad_norm": 0.04928620532155037, | |
| "learning_rate": 3.596137422092686e-05, | |
| "loss": 10.3351, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.41334039174166226, | |
| "grad_norm": 0.027833838015794754, | |
| "learning_rate": 3.578121385921533e-05, | |
| "loss": 10.3309, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 0.4137638962413976, | |
| "grad_norm": 0.03103015385568142, | |
| "learning_rate": 3.560140753512279e-05, | |
| "loss": 10.3359, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 0.41418740074113286, | |
| "grad_norm": 0.03528593108057976, | |
| "learning_rate": 3.542195623991991e-05, | |
| "loss": 10.3282, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.4146109052408682, | |
| "grad_norm": 0.03291507437825203, | |
| "learning_rate": 3.524286096292025e-05, | |
| "loss": 10.3309, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 0.4150344097406035, | |
| "grad_norm": 0.04097427800297737, | |
| "learning_rate": 3.5064122691474454e-05, | |
| "loss": 10.3362, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.4154579142403388, | |
| "grad_norm": 0.04069104790687561, | |
| "learning_rate": 3.4885742410965104e-05, | |
| "loss": 10.3347, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 0.4158814187400741, | |
| "grad_norm": 0.03851715475320816, | |
| "learning_rate": 3.4707721104801175e-05, | |
| "loss": 10.334, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 0.41630492323980944, | |
| "grad_norm": 0.03845444321632385, | |
| "learning_rate": 3.4530059754412555e-05, | |
| "loss": 10.3324, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 0.4167284277395447, | |
| "grad_norm": 0.027850644662976265, | |
| "learning_rate": 3.435275933924487e-05, | |
| "loss": 10.3309, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 0.41715193223928004, | |
| "grad_norm": 0.03326322138309479, | |
| "learning_rate": 3.417582083675365e-05, | |
| "loss": 10.3325, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.41757543673901537, | |
| "grad_norm": 0.027192946523427963, | |
| "learning_rate": 3.399924522239943e-05, | |
| "loss": 10.332, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 0.41799894123875064, | |
| "grad_norm": 0.035279251635074615, | |
| "learning_rate": 3.382303346964209e-05, | |
| "loss": 10.3317, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 0.41842244573848597, | |
| "grad_norm": 0.03443683683872223, | |
| "learning_rate": 3.36471865499354e-05, | |
| "loss": 10.3326, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 0.4188459502382213, | |
| "grad_norm": 0.030605580657720566, | |
| "learning_rate": 3.3471705432722035e-05, | |
| "loss": 10.3345, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 0.41926945473795657, | |
| "grad_norm": 0.032888561487197876, | |
| "learning_rate": 3.329659108542785e-05, | |
| "loss": 10.3265, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.4196929592376919, | |
| "grad_norm": 0.02829040214419365, | |
| "learning_rate": 3.3121844473456756e-05, | |
| "loss": 10.3325, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 0.4201164637374272, | |
| "grad_norm": 0.030971676111221313, | |
| "learning_rate": 3.294746656018532e-05, | |
| "loss": 10.3281, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.4205399682371625, | |
| "grad_norm": 0.03257730230689049, | |
| "learning_rate": 3.2773458306957495e-05, | |
| "loss": 10.3281, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 0.4209634727368978, | |
| "grad_norm": 0.03114408068358898, | |
| "learning_rate": 3.259982067307928e-05, | |
| "loss": 10.3343, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 0.42138697723663315, | |
| "grad_norm": 0.03386252745985985, | |
| "learning_rate": 3.2426554615813484e-05, | |
| "loss": 10.3316, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.4218104817363684, | |
| "grad_norm": 0.03416220098733902, | |
| "learning_rate": 3.2253661090374396e-05, | |
| "loss": 10.329, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 0.42223398623610375, | |
| "grad_norm": 0.031190721318125725, | |
| "learning_rate": 3.2081141049922535e-05, | |
| "loss": 10.3331, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 0.4226574907358391, | |
| "grad_norm": 0.03264687955379486, | |
| "learning_rate": 3.190899544555941e-05, | |
| "loss": 10.3313, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 0.42308099523557435, | |
| "grad_norm": 0.03346019983291626, | |
| "learning_rate": 3.173722522632228e-05, | |
| "loss": 10.3353, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 0.4235044997353097, | |
| "grad_norm": 0.03909333422780037, | |
| "learning_rate": 3.156583133917884e-05, | |
| "loss": 10.3316, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.423928004235045, | |
| "grad_norm": 0.030095241963863373, | |
| "learning_rate": 3.1394814729022235e-05, | |
| "loss": 10.3369, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 0.4243515087347803, | |
| "grad_norm": 0.02995004691183567, | |
| "learning_rate": 3.1224176338665476e-05, | |
| "loss": 10.3329, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 0.4247750132345156, | |
| "grad_norm": 0.039438553154468536, | |
| "learning_rate": 3.105391710883656e-05, | |
| "loss": 10.3305, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 0.42519851773425094, | |
| "grad_norm": 0.04090533405542374, | |
| "learning_rate": 3.088403797817325e-05, | |
| "loss": 10.3314, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 0.42562202223398626, | |
| "grad_norm": 0.0377449207007885, | |
| "learning_rate": 3.071453988321762e-05, | |
| "loss": 10.3298, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.42604552673372154, | |
| "grad_norm": 0.06536738574504852, | |
| "learning_rate": 3.0545423758411295e-05, | |
| "loss": 10.3276, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 0.42646903123345686, | |
| "grad_norm": 0.0357985682785511, | |
| "learning_rate": 3.037669053609006e-05, | |
| "loss": 10.3334, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 0.4268925357331922, | |
| "grad_norm": 0.03490246832370758, | |
| "learning_rate": 3.0208341146478602e-05, | |
| "loss": 10.3342, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 0.42731604023292746, | |
| "grad_norm": 0.03769504651427269, | |
| "learning_rate": 3.0040376517685764e-05, | |
| "loss": 10.3334, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 0.4277395447326628, | |
| "grad_norm": 0.027772339060902596, | |
| "learning_rate": 2.9872797575699097e-05, | |
| "loss": 10.3321, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.4281630492323981, | |
| "grad_norm": 0.034188639372587204, | |
| "learning_rate": 2.9705605244379853e-05, | |
| "loss": 10.3324, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 0.4281630492323981, | |
| "eval_loss": 10.331077575683594, | |
| "eval_runtime": 3.4933, | |
| "eval_samples_per_second": 284.832, | |
| "eval_steps_per_second": 142.559, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 0.4285865537321334, | |
| "grad_norm": 0.03415974974632263, | |
| "learning_rate": 2.9538800445457946e-05, | |
| "loss": 10.3323, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 0.4290100582318687, | |
| "grad_norm": 0.039172153919935226, | |
| "learning_rate": 2.9372384098526784e-05, | |
| "loss": 10.3347, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 0.42943356273160405, | |
| "grad_norm": 0.031853485852479935, | |
| "learning_rate": 2.9206357121038285e-05, | |
| "loss": 10.3338, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 0.4298570672313393, | |
| "grad_norm": 0.04923943430185318, | |
| "learning_rate": 2.904072042829775e-05, | |
| "loss": 10.3323, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.43028057173107465, | |
| "grad_norm": 0.03674182668328285, | |
| "learning_rate": 2.8875474933458847e-05, | |
| "loss": 10.3334, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 0.43070407623081, | |
| "grad_norm": 0.030546877533197403, | |
| "learning_rate": 2.871062154751858e-05, | |
| "loss": 10.3296, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 0.43112758073054525, | |
| "grad_norm": 0.030613403767347336, | |
| "learning_rate": 2.8546161179312248e-05, | |
| "loss": 10.3354, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 0.4315510852302806, | |
| "grad_norm": 0.030776720494031906, | |
| "learning_rate": 2.8382094735508457e-05, | |
| "loss": 10.3303, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 0.4319745897300159, | |
| "grad_norm": 0.03810757398605347, | |
| "learning_rate": 2.821842312060409e-05, | |
| "loss": 10.3334, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.4323980942297512, | |
| "grad_norm": 0.030035821720957756, | |
| "learning_rate": 2.8055147236919442e-05, | |
| "loss": 10.3345, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 0.4328215987294865, | |
| "grad_norm": 0.03650267794728279, | |
| "learning_rate": 2.789226798459298e-05, | |
| "loss": 10.3299, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 0.43324510322922183, | |
| "grad_norm": 0.030346672981977463, | |
| "learning_rate": 2.7729786261576617e-05, | |
| "loss": 10.334, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 0.4336686077289571, | |
| "grad_norm": 0.0330539271235466, | |
| "learning_rate": 2.7567702963630803e-05, | |
| "loss": 10.3316, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.43409211222869243, | |
| "grad_norm": 0.03174733370542526, | |
| "learning_rate": 2.740601898431925e-05, | |
| "loss": 10.3278, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.43451561672842776, | |
| "grad_norm": 0.03628386929631233, | |
| "learning_rate": 2.7244735215004446e-05, | |
| "loss": 10.3274, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 0.43493912122816303, | |
| "grad_norm": 0.024906015023589134, | |
| "learning_rate": 2.7083852544842436e-05, | |
| "loss": 10.3332, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 0.43536262572789836, | |
| "grad_norm": 0.043956976383924484, | |
| "learning_rate": 2.692337186077791e-05, | |
| "loss": 10.3266, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 0.4357861302276337, | |
| "grad_norm": 0.032996706664562225, | |
| "learning_rate": 2.67632940475396e-05, | |
| "loss": 10.3346, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 0.43620963472736896, | |
| "grad_norm": 0.044276829808950424, | |
| "learning_rate": 2.6603619987635086e-05, | |
| "loss": 10.3274, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.4366331392271043, | |
| "grad_norm": 0.038449618965387344, | |
| "learning_rate": 2.64443505613461e-05, | |
| "loss": 10.3341, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 0.4370566437268396, | |
| "grad_norm": 0.03220584616065025, | |
| "learning_rate": 2.6285486646723634e-05, | |
| "loss": 10.3324, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 0.4374801482265749, | |
| "grad_norm": 0.03746611624956131, | |
| "learning_rate": 2.612702911958308e-05, | |
| "loss": 10.3354, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 0.4379036527263102, | |
| "grad_norm": 0.04333876073360443, | |
| "learning_rate": 2.5968978853499425e-05, | |
| "loss": 10.329, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 0.43832715722604554, | |
| "grad_norm": 0.03539913892745972, | |
| "learning_rate": 2.581133671980246e-05, | |
| "loss": 10.3324, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.4387506617257808, | |
| "grad_norm": 0.04690808430314064, | |
| "learning_rate": 2.565410358757189e-05, | |
| "loss": 10.3316, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 0.43917416622551614, | |
| "grad_norm": 0.038458049297332764, | |
| "learning_rate": 2.5497280323632654e-05, | |
| "loss": 10.3431, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 0.43959767072525147, | |
| "grad_norm": 0.03451355919241905, | |
| "learning_rate": 2.534086779255005e-05, | |
| "loss": 10.3296, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 0.44002117522498674, | |
| "grad_norm": 0.03873763233423233, | |
| "learning_rate": 2.5184866856625023e-05, | |
| "loss": 10.3273, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 0.44044467972472207, | |
| "grad_norm": 0.044388849288225174, | |
| "learning_rate": 2.5029278375889387e-05, | |
| "loss": 10.3324, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.4408681842244574, | |
| "grad_norm": 0.03534289821982384, | |
| "learning_rate": 2.4874103208101183e-05, | |
| "loss": 10.3343, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 0.4412916887241927, | |
| "grad_norm": 0.0375693254172802, | |
| "learning_rate": 2.4719342208739693e-05, | |
| "loss": 10.3323, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 0.441715193223928, | |
| "grad_norm": 0.03341260179877281, | |
| "learning_rate": 2.456499623100098e-05, | |
| "loss": 10.3318, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 0.4421386977236633, | |
| "grad_norm": 0.04234972223639488, | |
| "learning_rate": 2.4411066125793203e-05, | |
| "loss": 10.3319, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 0.44256220222339865, | |
| "grad_norm": 0.031914252787828445, | |
| "learning_rate": 2.4257552741731592e-05, | |
| "loss": 10.3361, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.4429857067231339, | |
| "grad_norm": 0.05003447085618973, | |
| "learning_rate": 2.41044569251342e-05, | |
| "loss": 10.3313, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 0.44340921122286925, | |
| "grad_norm": 0.03364928439259529, | |
| "learning_rate": 2.3951779520016937e-05, | |
| "loss": 10.33, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 0.4438327157226046, | |
| "grad_norm": 0.028291532769799232, | |
| "learning_rate": 2.379952136808903e-05, | |
| "loss": 10.3336, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 0.44425622022233985, | |
| "grad_norm": 0.042799290269613266, | |
| "learning_rate": 2.3647683308748392e-05, | |
| "loss": 10.3348, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 0.4446797247220752, | |
| "grad_norm": 0.042522724717855453, | |
| "learning_rate": 2.3496266179076864e-05, | |
| "loss": 10.3288, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.4451032292218105, | |
| "grad_norm": 0.02918383479118347, | |
| "learning_rate": 2.3345270813835886e-05, | |
| "loss": 10.3361, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 0.4455267337215458, | |
| "grad_norm": 0.046009406447410583, | |
| "learning_rate": 2.319469804546156e-05, | |
| "loss": 10.3349, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 0.4459502382212811, | |
| "grad_norm": 0.03431849181652069, | |
| "learning_rate": 2.3044548704060288e-05, | |
| "loss": 10.3283, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 0.44637374272101643, | |
| "grad_norm": 0.03582574054598808, | |
| "learning_rate": 2.2894823617404104e-05, | |
| "loss": 10.3314, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 0.4467972472207517, | |
| "grad_norm": 0.02972414344549179, | |
| "learning_rate": 2.2745523610926122e-05, | |
| "loss": 10.3289, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.44722075172048703, | |
| "grad_norm": 0.03548819199204445, | |
| "learning_rate": 2.2596649507716018e-05, | |
| "loss": 10.3299, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 0.44764425622022236, | |
| "grad_norm": 0.04241335019469261, | |
| "learning_rate": 2.244820212851544e-05, | |
| "loss": 10.3308, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 0.44806776071995763, | |
| "grad_norm": 0.033176884055137634, | |
| "learning_rate": 2.2300182291713513e-05, | |
| "loss": 10.3351, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 0.44849126521969296, | |
| "grad_norm": 0.032935190945863724, | |
| "learning_rate": 2.2152590813342345e-05, | |
| "loss": 10.3356, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 0.4489147697194283, | |
| "grad_norm": 0.030969172716140747, | |
| "learning_rate": 2.2005428507072467e-05, | |
| "loss": 10.3307, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.44933827421916356, | |
| "grad_norm": 0.036834247410297394, | |
| "learning_rate": 2.1858696184208484e-05, | |
| "loss": 10.3324, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 0.4497617787188989, | |
| "grad_norm": 0.038617976009845734, | |
| "learning_rate": 2.1712394653684344e-05, | |
| "loss": 10.3371, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 0.4501852832186342, | |
| "grad_norm": 0.026445934548974037, | |
| "learning_rate": 2.15665247220592e-05, | |
| "loss": 10.3334, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 0.4506087877183695, | |
| "grad_norm": 0.04230870306491852, | |
| "learning_rate": 2.1421087193512756e-05, | |
| "loss": 10.3261, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 0.4510322922181048, | |
| "grad_norm": 0.03189300373196602, | |
| "learning_rate": 2.1276082869840765e-05, | |
| "loss": 10.3297, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.45145579671784014, | |
| "grad_norm": 0.03367699310183525, | |
| "learning_rate": 2.113151255045095e-05, | |
| "loss": 10.3308, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 0.4518793012175754, | |
| "grad_norm": 0.032475464046001434, | |
| "learning_rate": 2.0987377032358114e-05, | |
| "loss": 10.339, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 0.45230280571731074, | |
| "grad_norm": 0.04436371102929115, | |
| "learning_rate": 2.084367711018024e-05, | |
| "loss": 10.3301, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 0.45272631021704607, | |
| "grad_norm": 0.037988126277923584, | |
| "learning_rate": 2.070041357613376e-05, | |
| "loss": 10.3309, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 0.45314981471678134, | |
| "grad_norm": 0.03870435804128647, | |
| "learning_rate": 2.0557587220029228e-05, | |
| "loss": 10.3353, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.45357331921651667, | |
| "grad_norm": 0.03660368546843529, | |
| "learning_rate": 2.0415198829267212e-05, | |
| "loss": 10.3317, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 0.453996823716252, | |
| "grad_norm": 0.03593965247273445, | |
| "learning_rate": 2.0273249188833654e-05, | |
| "loss": 10.3343, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 0.45442032821598727, | |
| "grad_norm": 0.03798775374889374, | |
| "learning_rate": 2.013173908129573e-05, | |
| "loss": 10.329, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 0.4548438327157226, | |
| "grad_norm": 0.030165789648890495, | |
| "learning_rate": 1.9990669286797438e-05, | |
| "loss": 10.3325, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 0.4552673372154579, | |
| "grad_norm": 0.029242129996418953, | |
| "learning_rate": 1.985004058305535e-05, | |
| "loss": 10.3337, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.45569084171519325, | |
| "grad_norm": 0.029076050966978073, | |
| "learning_rate": 1.9709853745354313e-05, | |
| "loss": 10.3347, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 0.4561143462149285, | |
| "grad_norm": 0.039899520576000214, | |
| "learning_rate": 1.9570109546543126e-05, | |
| "loss": 10.3334, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 0.45653785071466385, | |
| "grad_norm": 0.03501451388001442, | |
| "learning_rate": 1.943080875703045e-05, | |
| "loss": 10.325, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 0.4569613552143992, | |
| "grad_norm": 0.029382554814219475, | |
| "learning_rate": 1.929195214478028e-05, | |
| "loss": 10.336, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 0.45738485971413445, | |
| "grad_norm": 0.03819538280367851, | |
| "learning_rate": 1.915354047530791e-05, | |
| "loss": 10.3329, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.4578083642138698, | |
| "grad_norm": 0.03543626144528389, | |
| "learning_rate": 1.901557451167578e-05, | |
| "loss": 10.3326, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 0.4582318687136051, | |
| "grad_norm": 0.04363977536559105, | |
| "learning_rate": 1.887805501448896e-05, | |
| "loss": 10.3289, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 0.4586553732133404, | |
| "grad_norm": 0.03918329253792763, | |
| "learning_rate": 1.8740982741891377e-05, | |
| "loss": 10.3276, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 0.4590788777130757, | |
| "grad_norm": 0.029666945338249207, | |
| "learning_rate": 1.860435844956121e-05, | |
| "loss": 10.3307, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 0.45950238221281103, | |
| "grad_norm": 0.035329993814229965, | |
| "learning_rate": 1.8468182890707007e-05, | |
| "loss": 10.3336, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.4599258867125463, | |
| "grad_norm": 0.040378130972385406, | |
| "learning_rate": 1.833245681606356e-05, | |
| "loss": 10.3296, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 0.46034939121228163, | |
| "grad_norm": 0.04233788326382637, | |
| "learning_rate": 1.8197180973887428e-05, | |
| "loss": 10.3312, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 0.46077289571201696, | |
| "grad_norm": 0.03670990467071533, | |
| "learning_rate": 1.806235610995327e-05, | |
| "loss": 10.3303, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 0.46119640021175223, | |
| "grad_norm": 0.03234660625457764, | |
| "learning_rate": 1.7927982967549384e-05, | |
| "loss": 10.3355, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 0.46161990471148756, | |
| "grad_norm": 0.042892660945653915, | |
| "learning_rate": 1.7794062287473735e-05, | |
| "loss": 10.331, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.4620434092112229, | |
| "grad_norm": 0.04852224513888359, | |
| "learning_rate": 1.7660594808029908e-05, | |
| "loss": 10.3361, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 0.46246691371095816, | |
| "grad_norm": 0.036822058260440826, | |
| "learning_rate": 1.7527581265022965e-05, | |
| "loss": 10.3364, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 0.4628904182106935, | |
| "grad_norm": 0.03043217770755291, | |
| "learning_rate": 1.7395022391755434e-05, | |
| "loss": 10.335, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 0.4633139227104288, | |
| "grad_norm": 0.027736082673072815, | |
| "learning_rate": 1.7262918919023243e-05, | |
| "loss": 10.3335, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 0.4637374272101641, | |
| "grad_norm": 0.03186174854636192, | |
| "learning_rate": 1.713127157511172e-05, | |
| "loss": 10.3365, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.4641609317098994, | |
| "grad_norm": 0.03788574039936066, | |
| "learning_rate": 1.700008108579154e-05, | |
| "loss": 10.3317, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 0.46458443620963474, | |
| "grad_norm": 0.047464434057474136, | |
| "learning_rate": 1.6869348174314738e-05, | |
| "loss": 10.3307, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 0.46500794070937, | |
| "grad_norm": 0.03223862871527672, | |
| "learning_rate": 1.673907356141079e-05, | |
| "loss": 10.3337, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 0.46543144520910534, | |
| "grad_norm": 0.02775878831744194, | |
| "learning_rate": 1.6609257965282453e-05, | |
| "loss": 10.3376, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 0.46585494970884067, | |
| "grad_norm": 0.0346621610224247, | |
| "learning_rate": 1.647990210160204e-05, | |
| "loss": 10.334, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.46627845420857594, | |
| "grad_norm": 0.03867461159825325, | |
| "learning_rate": 1.6351006683507297e-05, | |
| "loss": 10.3321, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 0.46670195870831127, | |
| "grad_norm": 0.033736009150743484, | |
| "learning_rate": 1.622257242159756e-05, | |
| "loss": 10.329, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 0.4671254632080466, | |
| "grad_norm": 0.03446945920586586, | |
| "learning_rate": 1.6094600023929884e-05, | |
| "loss": 10.3281, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 0.46754896770778187, | |
| "grad_norm": 0.03439204394817352, | |
| "learning_rate": 1.59670901960149e-05, | |
| "loss": 10.3339, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 0.4679724722075172, | |
| "grad_norm": 0.03250345215201378, | |
| "learning_rate": 1.5840043640813274e-05, | |
| "loss": 10.3308, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.4683959767072525, | |
| "grad_norm": 0.030219173058867455, | |
| "learning_rate": 1.5713461058731572e-05, | |
| "loss": 10.333, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 0.4688194812069878, | |
| "grad_norm": 0.031828220933675766, | |
| "learning_rate": 1.558734314761844e-05, | |
| "loss": 10.3353, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 0.4692429857067231, | |
| "grad_norm": 0.047410812228918076, | |
| "learning_rate": 1.546169060276088e-05, | |
| "loss": 10.3289, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 0.46966649020645845, | |
| "grad_norm": 0.036803584545850754, | |
| "learning_rate": 1.53365041168803e-05, | |
| "loss": 10.3358, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 0.4700899947061937, | |
| "grad_norm": 0.03534479811787605, | |
| "learning_rate": 1.5211784380128714e-05, | |
| "loss": 10.33, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.47051349920592905, | |
| "grad_norm": 0.036183904856443405, | |
| "learning_rate": 1.5087532080084976e-05, | |
| "loss": 10.3289, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 0.4709370037056644, | |
| "grad_norm": 0.033738043159246445, | |
| "learning_rate": 1.4963747901750936e-05, | |
| "loss": 10.3303, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 0.4713605082053997, | |
| "grad_norm": 0.03870893269777298, | |
| "learning_rate": 1.4840432527547732e-05, | |
| "loss": 10.3364, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 0.471784012705135, | |
| "grad_norm": 0.04043989256024361, | |
| "learning_rate": 1.4717586637311943e-05, | |
| "loss": 10.3316, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 0.4722075172048703, | |
| "grad_norm": 0.03024929389357567, | |
| "learning_rate": 1.4595210908291935e-05, | |
| "loss": 10.3364, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.47263102170460564, | |
| "grad_norm": 0.04411826282739639, | |
| "learning_rate": 1.447330601514405e-05, | |
| "loss": 10.3331, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 0.4730545262043409, | |
| "grad_norm": 0.03368929401040077, | |
| "learning_rate": 1.4351872629928908e-05, | |
| "loss": 10.3323, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 0.47347803070407624, | |
| "grad_norm": 0.038087401539087296, | |
| "learning_rate": 1.423091142210774e-05, | |
| "loss": 10.3295, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 0.47390153520381156, | |
| "grad_norm": 0.03507355973124504, | |
| "learning_rate": 1.4110423058538624e-05, | |
| "loss": 10.3273, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 0.47432503970354684, | |
| "grad_norm": 0.03440206125378609, | |
| "learning_rate": 1.3990408203472938e-05, | |
| "loss": 10.3336, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.47474854420328216, | |
| "grad_norm": 0.03201809525489807, | |
| "learning_rate": 1.387086751855149e-05, | |
| "loss": 10.3323, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 0.4751720487030175, | |
| "grad_norm": 0.02803219109773636, | |
| "learning_rate": 1.3751801662801056e-05, | |
| "loss": 10.3343, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 0.47559555320275276, | |
| "grad_norm": 0.03642897307872772, | |
| "learning_rate": 1.3633211292630742e-05, | |
| "loss": 10.3309, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 0.4760190577024881, | |
| "grad_norm": 0.04547721892595291, | |
| "learning_rate": 1.3515097061828164e-05, | |
| "loss": 10.3248, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 0.4764425622022234, | |
| "grad_norm": 0.03152972459793091, | |
| "learning_rate": 1.339745962155613e-05, | |
| "loss": 10.3396, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.4768660667019587, | |
| "grad_norm": 0.028171587735414505, | |
| "learning_rate": 1.3280299620348846e-05, | |
| "loss": 10.33, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 0.477289571201694, | |
| "grad_norm": 0.03410959243774414, | |
| "learning_rate": 1.3163617704108321e-05, | |
| "loss": 10.3344, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 0.47771307570142935, | |
| "grad_norm": 0.030304502695798874, | |
| "learning_rate": 1.304741451610103e-05, | |
| "loss": 10.3309, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 0.4781365802011646, | |
| "grad_norm": 0.03257643058896065, | |
| "learning_rate": 1.2931690696954135e-05, | |
| "loss": 10.3346, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 0.47856008470089995, | |
| "grad_norm": 0.04555933550000191, | |
| "learning_rate": 1.2816446884652066e-05, | |
| "loss": 10.3302, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.4789835892006353, | |
| "grad_norm": 0.0384778194129467, | |
| "learning_rate": 1.2701683714532975e-05, | |
| "loss": 10.3317, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 0.47940709370037055, | |
| "grad_norm": 0.03637570142745972, | |
| "learning_rate": 1.2587401819285239e-05, | |
| "loss": 10.3295, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 0.4798305982001059, | |
| "grad_norm": 0.04053565487265587, | |
| "learning_rate": 1.2473601828943949e-05, | |
| "loss": 10.3293, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 0.4802541026998412, | |
| "grad_norm": 0.042270079255104065, | |
| "learning_rate": 1.236028437088751e-05, | |
| "loss": 10.3271, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 0.4806776071995765, | |
| "grad_norm": 0.04081670939922333, | |
| "learning_rate": 1.2247450069834076e-05, | |
| "loss": 10.3365, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.4811011116993118, | |
| "grad_norm": 0.03796311840415001, | |
| "learning_rate": 1.2135099547838192e-05, | |
| "loss": 10.333, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 0.48152461619904713, | |
| "grad_norm": 0.02851458452641964, | |
| "learning_rate": 1.2023233424287328e-05, | |
| "loss": 10.3304, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 0.4819481206987824, | |
| "grad_norm": 0.03447718173265457, | |
| "learning_rate": 1.1911852315898463e-05, | |
| "loss": 10.3316, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 0.48237162519851773, | |
| "grad_norm": 0.037812747061252594, | |
| "learning_rate": 1.1800956836714682e-05, | |
| "loss": 10.3288, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 0.48279512969825306, | |
| "grad_norm": 0.03977108374238014, | |
| "learning_rate": 1.1690547598101864e-05, | |
| "loss": 10.3303, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.48321863419798833, | |
| "grad_norm": 0.031228644773364067, | |
| "learning_rate": 1.1580625208745145e-05, | |
| "loss": 10.3294, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 0.48364213869772366, | |
| "grad_norm": 0.0270911306142807, | |
| "learning_rate": 1.1471190274645704e-05, | |
| "loss": 10.3322, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 0.484065643197459, | |
| "grad_norm": 0.03246387094259262, | |
| "learning_rate": 1.1362243399117478e-05, | |
| "loss": 10.3306, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 0.48448914769719426, | |
| "grad_norm": 0.03161618486046791, | |
| "learning_rate": 1.1253785182783572e-05, | |
| "loss": 10.335, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 0.4849126521969296, | |
| "grad_norm": 0.03287721052765846, | |
| "learning_rate": 1.1145816223573259e-05, | |
| "loss": 10.3312, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.4853361566966649, | |
| "grad_norm": 0.029835056513547897, | |
| "learning_rate": 1.1038337116718467e-05, | |
| "loss": 10.3309, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 0.48575966119640024, | |
| "grad_norm": 0.03465202450752258, | |
| "learning_rate": 1.0931348454750601e-05, | |
| "loss": 10.3336, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 0.4861831656961355, | |
| "grad_norm": 0.03778757527470589, | |
| "learning_rate": 1.0824850827497246e-05, | |
| "loss": 10.3342, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 0.48660667019587084, | |
| "grad_norm": 0.03788898512721062, | |
| "learning_rate": 1.07188448220789e-05, | |
| "loss": 10.3338, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 0.48703017469560617, | |
| "grad_norm": 0.03392605856060982, | |
| "learning_rate": 1.061333102290576e-05, | |
| "loss": 10.3314, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.48745367919534144, | |
| "grad_norm": 0.03181210905313492, | |
| "learning_rate": 1.0508310011674516e-05, | |
| "loss": 10.3347, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 0.48787718369507677, | |
| "grad_norm": 0.03807486966252327, | |
| "learning_rate": 1.0403782367365088e-05, | |
| "loss": 10.3334, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 0.4883006881948121, | |
| "grad_norm": 0.04221343249082565, | |
| "learning_rate": 1.0299748666237485e-05, | |
| "loss": 10.33, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 0.48872419269454737, | |
| "grad_norm": 0.03662874549627304, | |
| "learning_rate": 1.0196209481828633e-05, | |
| "loss": 10.3337, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 0.4891476971942827, | |
| "grad_norm": 0.03761863335967064, | |
| "learning_rate": 1.0093165384949155e-05, | |
| "loss": 10.3363, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.489571201694018, | |
| "grad_norm": 0.03691156208515167, | |
| "learning_rate": 9.990616943680265e-06, | |
| "loss": 10.3355, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 0.4899947061937533, | |
| "grad_norm": 0.03406470641493797, | |
| "learning_rate": 9.888564723370664e-06, | |
| "loss": 10.3348, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 0.4904182106934886, | |
| "grad_norm": 0.03452722728252411, | |
| "learning_rate": 9.787009286633363e-06, | |
| "loss": 10.3332, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 0.49084171519322395, | |
| "grad_norm": 0.03500404581427574, | |
| "learning_rate": 9.685951193342602e-06, | |
| "loss": 10.3328, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 0.4912652196929592, | |
| "grad_norm": 0.034697335213422775, | |
| "learning_rate": 9.585391000630828e-06, | |
| "loss": 10.3292, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.49168872419269455, | |
| "grad_norm": 0.028287572786211967, | |
| "learning_rate": 9.485329262885457e-06, | |
| "loss": 10.3337, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 0.4921122286924299, | |
| "grad_norm": 0.0407349169254303, | |
| "learning_rate": 9.385766531746054e-06, | |
| "loss": 10.3314, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 0.49253573319216515, | |
| "grad_norm": 0.03521955758333206, | |
| "learning_rate": 9.28670335610109e-06, | |
| "loss": 10.3313, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 0.4929592376919005, | |
| "grad_norm": 0.038377124816179276, | |
| "learning_rate": 9.188140282084967e-06, | |
| "loss": 10.3295, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 0.4933827421916358, | |
| "grad_norm": 0.037929970771074295, | |
| "learning_rate": 9.090077853075118e-06, | |
| "loss": 10.331, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.4938062466913711, | |
| "grad_norm": 0.03767012432217598, | |
| "learning_rate": 8.992516609688862e-06, | |
| "loss": 10.3305, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 0.4942297511911064, | |
| "grad_norm": 0.04114054888486862, | |
| "learning_rate": 8.89545708978049e-06, | |
| "loss": 10.3327, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 0.49465325569084173, | |
| "grad_norm": 0.03139737620949745, | |
| "learning_rate": 8.798899828438333e-06, | |
| "loss": 10.3342, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 0.495076760190577, | |
| "grad_norm": 0.0350373312830925, | |
| "learning_rate": 8.70284535798168e-06, | |
| "loss": 10.3335, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 0.49550026469031233, | |
| "grad_norm": 0.03645787015557289, | |
| "learning_rate": 8.607294207958073e-06, | |
| "loss": 10.3285, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.49592376919004766, | |
| "grad_norm": 0.04092005640268326, | |
| "learning_rate": 8.512246905140165e-06, | |
| "loss": 10.332, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 0.49634727368978293, | |
| "grad_norm": 0.03972132131457329, | |
| "learning_rate": 8.417703973522917e-06, | |
| "loss": 10.3336, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 0.49677077818951826, | |
| "grad_norm": 0.02949652262032032, | |
| "learning_rate": 8.323665934320713e-06, | |
| "loss": 10.3329, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 0.4971942826892536, | |
| "grad_norm": 0.04814364016056061, | |
| "learning_rate": 8.23013330596445e-06, | |
| "loss": 10.3317, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 0.49761778718898886, | |
| "grad_norm": 0.0334940031170845, | |
| "learning_rate": 8.13710660409871e-06, | |
| "loss": 10.3367, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.4980412916887242, | |
| "grad_norm": 0.03809863701462746, | |
| "learning_rate": 8.044586341578886e-06, | |
| "loss": 10.3347, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 0.4984647961884595, | |
| "grad_norm": 0.03746895492076874, | |
| "learning_rate": 7.952573028468457e-06, | |
| "loss": 10.3362, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 0.4988883006881948, | |
| "grad_norm": 0.024187074974179268, | |
| "learning_rate": 7.861067172035962e-06, | |
| "loss": 10.3327, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 0.4993118051879301, | |
| "grad_norm": 0.03394331783056259, | |
| "learning_rate": 7.770069276752422e-06, | |
| "loss": 10.3268, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 0.49973530968766544, | |
| "grad_norm": 0.0327443964779377, | |
| "learning_rate": 7.679579844288509e-06, | |
| "loss": 10.332, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.5001588141874007, | |
| "grad_norm": 0.027774417772889137, | |
| "learning_rate": 7.589599373511602e-06, | |
| "loss": 10.329, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 0.5005823186871361, | |
| "grad_norm": 0.03464759886264801, | |
| "learning_rate": 7.500128360483338e-06, | |
| "loss": 10.3334, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 0.5010058231868714, | |
| "grad_norm": 0.03733719512820244, | |
| "learning_rate": 7.411167298456634e-06, | |
| "loss": 10.3307, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 0.5014293276866066, | |
| "grad_norm": 0.033785175532102585, | |
| "learning_rate": 7.32271667787302e-06, | |
| "loss": 10.3362, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 0.501852832186342, | |
| "grad_norm": 0.038209252059459686, | |
| "learning_rate": 7.234776986360059e-06, | |
| "loss": 10.3309, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.5022763366860773, | |
| "grad_norm": 0.03651139885187149, | |
| "learning_rate": 7.147348708728507e-06, | |
| "loss": 10.335, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 0.5026998411858126, | |
| "grad_norm": 0.03249209746718407, | |
| "learning_rate": 7.060432326969713e-06, | |
| "loss": 10.3326, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 0.503123345685548, | |
| "grad_norm": 0.049712520092725754, | |
| "learning_rate": 6.974028320252934e-06, | |
| "loss": 10.3269, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 0.5035468501852832, | |
| "grad_norm": 0.03345096856355667, | |
| "learning_rate": 6.888137164922725e-06, | |
| "loss": 10.3273, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 0.5039703546850185, | |
| "grad_norm": 0.028842521831393242, | |
| "learning_rate": 6.802759334496289e-06, | |
| "loss": 10.3299, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.5043938591847539, | |
| "grad_norm": 0.02980581857264042, | |
| "learning_rate": 6.717895299660892e-06, | |
| "loss": 10.3337, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 0.5048173636844892, | |
| "grad_norm": 0.032008688896894455, | |
| "learning_rate": 6.633545528271212e-06, | |
| "loss": 10.3275, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 0.5052408681842244, | |
| "grad_norm": 0.03007701225578785, | |
| "learning_rate": 6.549710485346827e-06, | |
| "loss": 10.3319, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 0.5056643726839598, | |
| "grad_norm": 0.03393697366118431, | |
| "learning_rate": 6.466390633069608e-06, | |
| "loss": 10.3292, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 0.5060878771836951, | |
| "grad_norm": 0.04486103355884552, | |
| "learning_rate": 6.383586430781197e-06, | |
| "loss": 10.3289, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.5065113816834304, | |
| "grad_norm": 0.03052888996899128, | |
| "learning_rate": 6.301298334980421e-06, | |
| "loss": 10.3374, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 0.5069348861831657, | |
| "grad_norm": 0.030694812536239624, | |
| "learning_rate": 6.219526799320919e-06, | |
| "loss": 10.3308, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 0.507358390682901, | |
| "grad_norm": 0.03446760028600693, | |
| "learning_rate": 6.138272274608403e-06, | |
| "loss": 10.3346, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 0.5077818951826363, | |
| "grad_norm": 0.033587660640478134, | |
| "learning_rate": 6.057535208798371e-06, | |
| "loss": 10.3337, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 0.5082053996823717, | |
| "grad_norm": 0.03484556823968887, | |
| "learning_rate": 5.977316046993642e-06, | |
| "loss": 10.3311, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.5086289041821069, | |
| "grad_norm": 0.03142661601305008, | |
| "learning_rate": 5.897615231441689e-06, | |
| "loss": 10.3335, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 0.5090524086818422, | |
| "grad_norm": 0.03492956608533859, | |
| "learning_rate": 5.81843320153248e-06, | |
| "loss": 10.3298, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 0.5094759131815776, | |
| "grad_norm": 0.035875819623470306, | |
| "learning_rate": 5.739770393795851e-06, | |
| "loss": 10.3339, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 0.5098994176813129, | |
| "grad_norm": 0.028575167059898376, | |
| "learning_rate": 5.6616272418991926e-06, | |
| "loss": 10.3306, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 0.5103229221810481, | |
| "grad_norm": 0.034280769526958466, | |
| "learning_rate": 5.584004176645052e-06, | |
| "loss": 10.3339, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.5107464266807835, | |
| "grad_norm": 0.03369034081697464, | |
| "learning_rate": 5.5069016259686635e-06, | |
| "loss": 10.3293, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 0.5111699311805188, | |
| "grad_norm": 0.03932506591081619, | |
| "learning_rate": 5.430320014935797e-06, | |
| "loss": 10.3339, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 0.5115934356802541, | |
| "grad_norm": 0.04464678466320038, | |
| "learning_rate": 5.354259765740177e-06, | |
| "loss": 10.3316, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 0.5120169401799894, | |
| "grad_norm": 0.033909354358911514, | |
| "learning_rate": 5.278721297701339e-06, | |
| "loss": 10.3317, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 0.5124404446797247, | |
| "grad_norm": 0.02771197073161602, | |
| "learning_rate": 5.203705027262184e-06, | |
| "loss": 10.3337, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.51286394917946, | |
| "grad_norm": 0.03711957111954689, | |
| "learning_rate": 5.129211367986786e-06, | |
| "loss": 10.3374, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 0.5132874536791954, | |
| "grad_norm": 0.04035378247499466, | |
| "learning_rate": 5.055240730558042e-06, | |
| "loss": 10.3278, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 0.5137109581789306, | |
| "grad_norm": 0.037376079708337784, | |
| "learning_rate": 4.981793522775457e-06, | |
| "loss": 10.3354, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 0.5141344626786659, | |
| "grad_norm": 0.033283621072769165, | |
| "learning_rate": 4.908870149552835e-06, | |
| "loss": 10.3304, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 0.5145579671784013, | |
| "grad_norm": 0.04279647022485733, | |
| "learning_rate": 4.836471012916144e-06, | |
| "loss": 10.3317, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.5149814716781366, | |
| "grad_norm": 0.026392200961709023, | |
| "learning_rate": 4.764596512001162e-06, | |
| "loss": 10.3338, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 0.5154049761778718, | |
| "grad_norm": 0.038188233971595764, | |
| "learning_rate": 4.693247043051441e-06, | |
| "loss": 10.3363, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 0.5158284806776072, | |
| "grad_norm": 0.03593307361006737, | |
| "learning_rate": 4.622422999415965e-06, | |
| "loss": 10.3302, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 0.5162519851773425, | |
| "grad_norm": 0.03967192396521568, | |
| "learning_rate": 4.5521247715470945e-06, | |
| "loss": 10.33, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 0.5166754896770778, | |
| "grad_norm": 0.0491623692214489, | |
| "learning_rate": 4.482352746998364e-06, | |
| "loss": 10.3386, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.5170989941768132, | |
| "grad_norm": 0.0371236614882946, | |
| "learning_rate": 4.413107310422326e-06, | |
| "loss": 10.3336, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 0.5175224986765484, | |
| "grad_norm": 0.027762679383158684, | |
| "learning_rate": 4.344388843568503e-06, | |
| "loss": 10.3282, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 0.5179460031762837, | |
| "grad_norm": 0.03931552171707153, | |
| "learning_rate": 4.2761977252811945e-06, | |
| "loss": 10.3331, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 0.5183695076760191, | |
| "grad_norm": 0.047121018171310425, | |
| "learning_rate": 4.2085343314974715e-06, | |
| "loss": 10.3297, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 0.5187930121757544, | |
| "grad_norm": 0.042633168399333954, | |
| "learning_rate": 4.141399035245052e-06, | |
| "loss": 10.3337, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.5192165166754896, | |
| "grad_norm": 0.03988894075155258, | |
| "learning_rate": 4.07479220664021e-06, | |
| "loss": 10.3262, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 0.519640021175225, | |
| "grad_norm": 0.030842246487736702, | |
| "learning_rate": 4.008714212885856e-06, | |
| "loss": 10.3322, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 0.5200635256749603, | |
| "grad_norm": 0.04261520504951477, | |
| "learning_rate": 3.943165418269401e-06, | |
| "loss": 10.328, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 0.5204870301746956, | |
| "grad_norm": 0.030063187703490257, | |
| "learning_rate": 3.87814618416078e-06, | |
| "loss": 10.3345, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 0.5209105346744309, | |
| "grad_norm": 0.030118783935904503, | |
| "learning_rate": 3.8136568690104957e-06, | |
| "loss": 10.3325, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.5213340391741662, | |
| "grad_norm": 0.03795788437128067, | |
| "learning_rate": 3.7496978283475648e-06, | |
| "loss": 10.3327, | |
| "step": 1231 | |
| }, | |
| { | |
| "epoch": 0.5217575436739015, | |
| "grad_norm": 0.036961231380701065, | |
| "learning_rate": 3.686269414777643e-06, | |
| "loss": 10.3344, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 0.5221810481736369, | |
| "grad_norm": 0.0403430350124836, | |
| "learning_rate": 3.623371977981027e-06, | |
| "loss": 10.3324, | |
| "step": 1233 | |
| }, | |
| { | |
| "epoch": 0.5226045526733721, | |
| "grad_norm": 0.03135257214307785, | |
| "learning_rate": 3.5610058647107538e-06, | |
| "loss": 10.3319, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 0.5230280571731075, | |
| "grad_norm": 0.0364365391433239, | |
| "learning_rate": 3.499171418790681e-06, | |
| "loss": 10.3343, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.5234515616728428, | |
| "grad_norm": 0.025732390582561493, | |
| "learning_rate": 3.437868981113557e-06, | |
| "loss": 10.3338, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 0.5238750661725781, | |
| "grad_norm": 0.03495744988322258, | |
| "learning_rate": 3.37709888963923e-06, | |
| "loss": 10.3302, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 0.5242985706723134, | |
| "grad_norm": 0.032097022980451584, | |
| "learning_rate": 3.3168614793926524e-06, | |
| "loss": 10.3356, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 0.5247220751720487, | |
| "grad_norm": 0.029357150197029114, | |
| "learning_rate": 3.2571570824621923e-06, | |
| "loss": 10.3304, | |
| "step": 1239 | |
| }, | |
| { | |
| "epoch": 0.525145579671784, | |
| "grad_norm": 0.03179454430937767, | |
| "learning_rate": 3.197986027997657e-06, | |
| "loss": 10.3311, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.5255690841715194, | |
| "grad_norm": 0.038864728063344955, | |
| "learning_rate": 3.1393486422085618e-06, | |
| "loss": 10.3308, | |
| "step": 1241 | |
| }, | |
| { | |
| "epoch": 0.5259925886712546, | |
| "grad_norm": 0.027193231508135796, | |
| "learning_rate": 3.08124524836233e-06, | |
| "loss": 10.3314, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 0.5264160931709899, | |
| "grad_norm": 0.035837847739458084, | |
| "learning_rate": 3.023676166782452e-06, | |
| "loss": 10.3327, | |
| "step": 1243 | |
| }, | |
| { | |
| "epoch": 0.5268395976707253, | |
| "grad_norm": 0.02682778798043728, | |
| "learning_rate": 2.9666417148468072e-06, | |
| "loss": 10.3325, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 0.5272631021704606, | |
| "grad_norm": 0.04898487776517868, | |
| "learning_rate": 2.910142206985833e-06, | |
| "loss": 10.3317, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.5276866066701958, | |
| "grad_norm": 0.030211864039301872, | |
| "learning_rate": 2.8541779546808256e-06, | |
| "loss": 10.3292, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 0.5281101111699312, | |
| "grad_norm": 0.03472064808011055, | |
| "learning_rate": 2.7987492664622307e-06, | |
| "loss": 10.3324, | |
| "step": 1247 | |
| }, | |
| { | |
| "epoch": 0.5285336156696665, | |
| "grad_norm": 0.03139955550432205, | |
| "learning_rate": 2.743856447907944e-06, | |
| "loss": 10.3309, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 0.5289571201694018, | |
| "grad_norm": 0.02904195711016655, | |
| "learning_rate": 2.689499801641593e-06, | |
| "loss": 10.332, | |
| "step": 1249 | |
| }, | |
| { | |
| "epoch": 0.5293806246691372, | |
| "grad_norm": 0.045261383056640625, | |
| "learning_rate": 2.6356796273309116e-06, | |
| "loss": 10.33, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.5298041291688724, | |
| "grad_norm": 0.03183293342590332, | |
| "learning_rate": 2.5823962216860562e-06, | |
| "loss": 10.3297, | |
| "step": 1251 | |
| }, | |
| { | |
| "epoch": 0.5302276336686077, | |
| "grad_norm": 0.04214952513575554, | |
| "learning_rate": 2.5296498784579846e-06, | |
| "loss": 10.3309, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 0.5306511381683431, | |
| "grad_norm": 0.03488962724804878, | |
| "learning_rate": 2.4774408884368215e-06, | |
| "loss": 10.3333, | |
| "step": 1253 | |
| }, | |
| { | |
| "epoch": 0.5310746426680784, | |
| "grad_norm": 0.03279737010598183, | |
| "learning_rate": 2.4257695394503287e-06, | |
| "loss": 10.3278, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 0.5314981471678136, | |
| "grad_norm": 0.03219415992498398, | |
| "learning_rate": 2.374636116362172e-06, | |
| "loss": 10.3334, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.531921651667549, | |
| "grad_norm": 0.05066683888435364, | |
| "learning_rate": 2.32404090107049e-06, | |
| "loss": 10.3306, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 0.5323451561672843, | |
| "grad_norm": 0.028979485854506493, | |
| "learning_rate": 2.2739841725062715e-06, | |
| "loss": 10.3319, | |
| "step": 1257 | |
| }, | |
| { | |
| "epoch": 0.5327686606670196, | |
| "grad_norm": 0.03191670775413513, | |
| "learning_rate": 2.2244662066318146e-06, | |
| "loss": 10.333, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 0.5331921651667549, | |
| "grad_norm": 0.04911280795931816, | |
| "learning_rate": 2.1754872764392698e-06, | |
| "loss": 10.3313, | |
| "step": 1259 | |
| }, | |
| { | |
| "epoch": 0.5336156696664902, | |
| "grad_norm": 0.039490871131420135, | |
| "learning_rate": 2.1270476519490435e-06, | |
| "loss": 10.3244, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.5340391741662255, | |
| "grad_norm": 0.03646280616521835, | |
| "learning_rate": 2.079147600208364e-06, | |
| "loss": 10.3303, | |
| "step": 1261 | |
| }, | |
| { | |
| "epoch": 0.5344626786659609, | |
| "grad_norm": 0.039123885333538055, | |
| "learning_rate": 2.0317873852898518e-06, | |
| "loss": 10.332, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 0.5348861831656961, | |
| "grad_norm": 0.04183242470026016, | |
| "learning_rate": 1.9849672682898944e-06, | |
| "loss": 10.3297, | |
| "step": 1263 | |
| }, | |
| { | |
| "epoch": 0.5353096876654314, | |
| "grad_norm": 0.03520303592085838, | |
| "learning_rate": 1.9386875073274636e-06, | |
| "loss": 10.3265, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 0.5357331921651668, | |
| "grad_norm": 0.0325089730322361, | |
| "learning_rate": 1.8929483575424455e-06, | |
| "loss": 10.3345, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.5361566966649021, | |
| "grad_norm": 0.029976682737469673, | |
| "learning_rate": 1.8477500710944007e-06, | |
| "loss": 10.3292, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 0.5365802011646373, | |
| "grad_norm": 0.034131329506635666, | |
| "learning_rate": 1.803092897161096e-06, | |
| "loss": 10.3276, | |
| "step": 1267 | |
| }, | |
| { | |
| "epoch": 0.5370037056643727, | |
| "grad_norm": 0.03793232887983322, | |
| "learning_rate": 1.75897708193713e-06, | |
| "loss": 10.3349, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 0.537427210164108, | |
| "grad_norm": 0.025969160720705986, | |
| "learning_rate": 1.715402868632643e-06, | |
| "loss": 10.3325, | |
| "step": 1269 | |
| }, | |
| { | |
| "epoch": 0.5378507146638433, | |
| "grad_norm": 0.04372668266296387, | |
| "learning_rate": 1.6723704974718756e-06, | |
| "loss": 10.33, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.5382742191635786, | |
| "grad_norm": 0.03358982875943184, | |
| "learning_rate": 1.629880205691936e-06, | |
| "loss": 10.3321, | |
| "step": 1271 | |
| }, | |
| { | |
| "epoch": 0.5386977236633139, | |
| "grad_norm": 0.045495398342609406, | |
| "learning_rate": 1.5879322275414332e-06, | |
| "loss": 10.3334, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 0.5391212281630492, | |
| "grad_norm": 0.02813423052430153, | |
| "learning_rate": 1.5465267942792127e-06, | |
| "loss": 10.332, | |
| "step": 1273 | |
| }, | |
| { | |
| "epoch": 0.5395447326627846, | |
| "grad_norm": 0.02770121954381466, | |
| "learning_rate": 1.5056641341730903e-06, | |
| "loss": 10.3296, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 0.5399682371625198, | |
| "grad_norm": 0.04436861723661423, | |
| "learning_rate": 1.465344472498531e-06, | |
| "loss": 10.3286, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.5403917416622551, | |
| "grad_norm": 0.043747782707214355, | |
| "learning_rate": 1.4255680315375164e-06, | |
| "loss": 10.3332, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 0.5408152461619905, | |
| "grad_norm": 0.028111323714256287, | |
| "learning_rate": 1.3863350305772017e-06, | |
| "loss": 10.3319, | |
| "step": 1277 | |
| }, | |
| { | |
| "epoch": 0.5412387506617258, | |
| "grad_norm": 0.03884616121649742, | |
| "learning_rate": 1.3476456859087828e-06, | |
| "loss": 10.3317, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 0.541662255161461, | |
| "grad_norm": 0.04214450716972351, | |
| "learning_rate": 1.3095002108263199e-06, | |
| "loss": 10.3336, | |
| "step": 1279 | |
| }, | |
| { | |
| "epoch": 0.5420857596611964, | |
| "grad_norm": 0.0312722884118557, | |
| "learning_rate": 1.2718988156254607e-06, | |
| "loss": 10.3357, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.5425092641609317, | |
| "grad_norm": 0.09322332590818405, | |
| "learning_rate": 1.2348417076023745e-06, | |
| "loss": 10.3333, | |
| "step": 1281 | |
| }, | |
| { | |
| "epoch": 0.542932768660667, | |
| "grad_norm": 0.04540476202964783, | |
| "learning_rate": 1.198329091052608e-06, | |
| "loss": 10.3309, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 0.5433562731604024, | |
| "grad_norm": 0.029997704550623894, | |
| "learning_rate": 1.1623611672698765e-06, | |
| "loss": 10.3358, | |
| "step": 1283 | |
| }, | |
| { | |
| "epoch": 0.5437797776601376, | |
| "grad_norm": 0.0350346714258194, | |
| "learning_rate": 1.1269381345450526e-06, | |
| "loss": 10.3306, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 0.5442032821598729, | |
| "grad_norm": 0.04271746799349785, | |
| "learning_rate": 1.0920601881650006e-06, | |
| "loss": 10.3313, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.5446267866596083, | |
| "grad_norm": 0.03767610713839531, | |
| "learning_rate": 1.0577275204115444e-06, | |
| "loss": 10.3275, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 0.5450502911593436, | |
| "grad_norm": 0.02964678965508938, | |
| "learning_rate": 1.0239403205604014e-06, | |
| "loss": 10.3296, | |
| "step": 1287 | |
| }, | |
| { | |
| "epoch": 0.5454737956590788, | |
| "grad_norm": 0.03278511017560959, | |
| "learning_rate": 9.906987748800944e-07, | |
| "loss": 10.3329, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 0.5458973001588142, | |
| "grad_norm": 0.05790937691926956, | |
| "learning_rate": 9.580030666309969e-07, | |
| "loss": 10.3372, | |
| "step": 1289 | |
| }, | |
| { | |
| "epoch": 0.5463208046585495, | |
| "grad_norm": 0.03746120631694794, | |
| "learning_rate": 9.258533760642563e-07, | |
| "loss": 10.3302, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.5467443091582848, | |
| "grad_norm": 0.03203713148832321, | |
| "learning_rate": 8.942498804208498e-07, | |
| "loss": 10.3328, | |
| "step": 1291 | |
| }, | |
| { | |
| "epoch": 0.5471678136580201, | |
| "grad_norm": 0.032408781349658966, | |
| "learning_rate": 8.631927539305862e-07, | |
| "loss": 10.3328, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 0.5475913181577554, | |
| "grad_norm": 0.038404081016778946, | |
| "learning_rate": 8.326821678111163e-07, | |
| "loss": 10.3357, | |
| "step": 1293 | |
| }, | |
| { | |
| "epoch": 0.5480148226574907, | |
| "grad_norm": 0.03704221174120903, | |
| "learning_rate": 8.027182902670571e-07, | |
| "loss": 10.3267, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 0.5484383271572261, | |
| "grad_norm": 0.02777581661939621, | |
| "learning_rate": 7.733012864890032e-07, | |
| "loss": 10.3331, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.5488618316569613, | |
| "grad_norm": 0.0339139886200428, | |
| "learning_rate": 7.444313186526608e-07, | |
| "loss": 10.3355, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 0.5492853361566966, | |
| "grad_norm": 0.027996981516480446, | |
| "learning_rate": 7.161085459178929e-07, | |
| "loss": 10.3301, | |
| "step": 1297 | |
| }, | |
| { | |
| "epoch": 0.549708840656432, | |
| "grad_norm": 0.04270913451910019, | |
| "learning_rate": 6.88333124427909e-07, | |
| "loss": 10.3269, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 0.5501323451561673, | |
| "grad_norm": 0.0351426862180233, | |
| "learning_rate": 6.611052073083768e-07, | |
| "loss": 10.3306, | |
| "step": 1299 | |
| }, | |
| { | |
| "epoch": 0.5505558496559025, | |
| "grad_norm": 0.0378975048661232, | |
| "learning_rate": 6.344249446665674e-07, | |
| "loss": 10.3283, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.5509793541556379, | |
| "grad_norm": 0.028754916042089462, | |
| "learning_rate": 6.082924835905446e-07, | |
| "loss": 10.3287, | |
| "step": 1301 | |
| }, | |
| { | |
| "epoch": 0.5514028586553732, | |
| "grad_norm": 0.0465865433216095, | |
| "learning_rate": 5.827079681483438e-07, | |
| "loss": 10.3325, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 0.5518263631551085, | |
| "grad_norm": 0.037231337279081345, | |
| "learning_rate": 5.576715393871613e-07, | |
| "loss": 10.3278, | |
| "step": 1303 | |
| }, | |
| { | |
| "epoch": 0.5522498676548439, | |
| "grad_norm": 0.03710845485329628, | |
| "learning_rate": 5.331833353326432e-07, | |
| "loss": 10.3344, | |
| "step": 1304 | |
| }, | |
| { | |
| "epoch": 0.5526733721545791, | |
| "grad_norm": 0.02809790149331093, | |
| "learning_rate": 5.092434909880317e-07, | |
| "loss": 10.3321, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 0.5530968766543145, | |
| "grad_norm": 0.045991264283657074, | |
| "learning_rate": 4.858521383334868e-07, | |
| "loss": 10.3345, | |
| "step": 1306 | |
| }, | |
| { | |
| "epoch": 0.5535203811540498, | |
| "grad_norm": 0.03640573099255562, | |
| "learning_rate": 4.630094063253321e-07, | |
| "loss": 10.3294, | |
| "step": 1307 | |
| }, | |
| { | |
| "epoch": 0.553943885653785, | |
| "grad_norm": 0.029001332819461823, | |
| "learning_rate": 4.4071542089535454e-07, | |
| "loss": 10.3318, | |
| "step": 1308 | |
| }, | |
| { | |
| "epoch": 0.5543673901535204, | |
| "grad_norm": 0.02934233844280243, | |
| "learning_rate": 4.18970304950117e-07, | |
| "loss": 10.3299, | |
| "step": 1309 | |
| }, | |
| { | |
| "epoch": 0.5547908946532557, | |
| "grad_norm": 0.03224503621459007, | |
| "learning_rate": 3.977741783702471e-07, | |
| "loss": 10.3285, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.555214399152991, | |
| "grad_norm": 0.03147895634174347, | |
| "learning_rate": 3.771271580098157e-07, | |
| "loss": 10.3325, | |
| "step": 1311 | |
| }, | |
| { | |
| "epoch": 0.5556379036527264, | |
| "grad_norm": 0.03843318298459053, | |
| "learning_rate": 3.570293576956596e-07, | |
| "loss": 10.3301, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 0.5560614081524616, | |
| "grad_norm": 0.0349433533847332, | |
| "learning_rate": 3.3748088822679325e-07, | |
| "loss": 10.332, | |
| "step": 1313 | |
| }, | |
| { | |
| "epoch": 0.5564849126521969, | |
| "grad_norm": 0.03259619325399399, | |
| "learning_rate": 3.184818573737425e-07, | |
| "loss": 10.3296, | |
| "step": 1314 | |
| }, | |
| { | |
| "epoch": 0.5569084171519323, | |
| "grad_norm": 0.03497344255447388, | |
| "learning_rate": 3.0003236987802274e-07, | |
| "loss": 10.3314, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 0.5573319216516676, | |
| "grad_norm": 0.03283681720495224, | |
| "learning_rate": 2.821325274514952e-07, | |
| "loss": 10.3307, | |
| "step": 1316 | |
| }, | |
| { | |
| "epoch": 0.5577554261514028, | |
| "grad_norm": 0.03914149850606918, | |
| "learning_rate": 2.6478242877583383e-07, | |
| "loss": 10.3321, | |
| "step": 1317 | |
| }, | |
| { | |
| "epoch": 0.5581789306511382, | |
| "grad_norm": 0.028979448601603508, | |
| "learning_rate": 2.4798216950198127e-07, | |
| "loss": 10.3295, | |
| "step": 1318 | |
| }, | |
| { | |
| "epoch": 0.5586024351508735, | |
| "grad_norm": 0.0339006632566452, | |
| "learning_rate": 2.317318422496273e-07, | |
| "loss": 10.3326, | |
| "step": 1319 | |
| }, | |
| { | |
| "epoch": 0.5590259396506088, | |
| "grad_norm": 0.027926115319132805, | |
| "learning_rate": 2.1603153660668674e-07, | |
| "loss": 10.3305, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.5594494441503441, | |
| "grad_norm": 0.031478822231292725, | |
| "learning_rate": 2.0088133912881113e-07, | |
| "loss": 10.3288, | |
| "step": 1321 | |
| }, | |
| { | |
| "epoch": 0.5598729486500794, | |
| "grad_norm": 0.03274491801857948, | |
| "learning_rate": 1.862813333389113e-07, | |
| "loss": 10.3361, | |
| "step": 1322 | |
| }, | |
| { | |
| "epoch": 0.5602964531498147, | |
| "grad_norm": 0.0399165078997612, | |
| "learning_rate": 1.722315997267021e-07, | |
| "loss": 10.3344, | |
| "step": 1323 | |
| }, | |
| { | |
| "epoch": 0.5607199576495501, | |
| "grad_norm": 0.030695218592882156, | |
| "learning_rate": 1.5873221574822516e-07, | |
| "loss": 10.3298, | |
| "step": 1324 | |
| }, | |
| { | |
| "epoch": 0.5611434621492853, | |
| "grad_norm": 0.03967565670609474, | |
| "learning_rate": 1.4578325582548237e-07, | |
| "loss": 10.3305, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.5615669666490206, | |
| "grad_norm": 0.03664049133658409, | |
| "learning_rate": 1.3338479134596958e-07, | |
| "loss": 10.3293, | |
| "step": 1326 | |
| }, | |
| { | |
| "epoch": 0.561990471148756, | |
| "grad_norm": 0.03802071511745453, | |
| "learning_rate": 1.2153689066233266e-07, | |
| "loss": 10.3305, | |
| "step": 1327 | |
| }, | |
| { | |
| "epoch": 0.5624139756484913, | |
| "grad_norm": 0.036713242530822754, | |
| "learning_rate": 1.1023961909192304e-07, | |
| "loss": 10.3287, | |
| "step": 1328 | |
| }, | |
| { | |
| "epoch": 0.5628374801482265, | |
| "grad_norm": 0.04824815317988396, | |
| "learning_rate": 9.949303891653161e-08, | |
| "loss": 10.3353, | |
| "step": 1329 | |
| }, | |
| { | |
| "epoch": 0.5632609846479619, | |
| "grad_norm": 0.03399055823683739, | |
| "learning_rate": 8.929720938193331e-08, | |
| "loss": 10.3302, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.5636844891476972, | |
| "grad_norm": 0.030519891530275345, | |
| "learning_rate": 7.965218669766516e-08, | |
| "loss": 10.3277, | |
| "step": 1331 | |
| }, | |
| { | |
| "epoch": 0.5641079936474325, | |
| "grad_norm": 0.03647278994321823, | |
| "learning_rate": 7.05580240366488e-08, | |
| "loss": 10.3276, | |
| "step": 1332 | |
| }, | |
| { | |
| "epoch": 0.5645314981471679, | |
| "grad_norm": 0.0370662622153759, | |
| "learning_rate": 6.201477153493506e-08, | |
| "loss": 10.3344, | |
| "step": 1333 | |
| }, | |
| { | |
| "epoch": 0.5649550026469031, | |
| "grad_norm": 0.038933202624320984, | |
| "learning_rate": 5.402247629139323e-08, | |
| "loss": 10.3313, | |
| "step": 1334 | |
| }, | |
| { | |
| "epoch": 0.5653785071466384, | |
| "grad_norm": 0.030461156740784645, | |
| "learning_rate": 4.658118236747777e-08, | |
| "loss": 10.3292, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 0.5658020116463738, | |
| "grad_norm": 0.030602607876062393, | |
| "learning_rate": 3.9690930786995264e-08, | |
| "loss": 10.3294, | |
| "step": 1336 | |
| }, | |
| { | |
| "epoch": 0.566225516146109, | |
| "grad_norm": 0.03394312039017677, | |
| "learning_rate": 3.335175953581571e-08, | |
| "loss": 10.3342, | |
| "step": 1337 | |
| }, | |
| { | |
| "epoch": 0.5666490206458443, | |
| "grad_norm": 0.051167815923690796, | |
| "learning_rate": 2.756370356175042e-08, | |
| "loss": 10.3349, | |
| "step": 1338 | |
| }, | |
| { | |
| "epoch": 0.5670725251455797, | |
| "grad_norm": 0.03260042518377304, | |
| "learning_rate": 2.232679477430777e-08, | |
| "loss": 10.3333, | |
| "step": 1339 | |
| }, | |
| { | |
| "epoch": 0.567496029645315, | |
| "grad_norm": 0.040678899735212326, | |
| "learning_rate": 1.7641062044515544e-08, | |
| "loss": 10.3287, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.5679195341450503, | |
| "grad_norm": 0.04332433268427849, | |
| "learning_rate": 1.350653120477663e-08, | |
| "loss": 10.3336, | |
| "step": 1341 | |
| }, | |
| { | |
| "epoch": 0.5683430386447856, | |
| "grad_norm": 0.03431249037384987, | |
| "learning_rate": 9.923225048724671e-09, | |
| "loss": 10.3331, | |
| "step": 1342 | |
| }, | |
| { | |
| "epoch": 0.5687665431445209, | |
| "grad_norm": 0.0347750224173069, | |
| "learning_rate": 6.891163331101957e-09, | |
| "loss": 10.3338, | |
| "step": 1343 | |
| }, | |
| { | |
| "epoch": 0.5691900476442562, | |
| "grad_norm": 0.031236495822668076, | |
| "learning_rate": 4.410362767626186e-09, | |
| "loss": 10.3311, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 0.5696135521439916, | |
| "grad_norm": 0.036892782896757126, | |
| "learning_rate": 2.4808370349460596e-09, | |
| "loss": 10.332, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 0.5700370566437268, | |
| "grad_norm": 0.02656089887022972, | |
| "learning_rate": 1.1025967705080576e-09, | |
| "loss": 10.3323, | |
| "step": 1346 | |
| }, | |
| { | |
| "epoch": 0.5704605611434621, | |
| "grad_norm": 0.03345981240272522, | |
| "learning_rate": 2.756495725342312e-10, | |
| "loss": 10.3286, | |
| "step": 1347 | |
| }, | |
| { | |
| "epoch": 0.5708840656431975, | |
| "grad_norm": 0.03345588967204094, | |
| "learning_rate": 0.0, | |
| "loss": 10.3287, | |
| "step": 1348 | |
| }, | |
| { | |
| "epoch": 0.5708840656431975, | |
| "eval_loss": 10.330697059631348, | |
| "eval_runtime": 3.473, | |
| "eval_samples_per_second": 286.494, | |
| "eval_steps_per_second": 143.391, | |
| "step": 1348 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1348, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 337, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 28945837916160.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |