| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.2766730070897458, |
| "eval_steps": 500, |
| "global_step": 400, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0006916825177243646, |
| "grad_norm": 0.2954893708229065, |
| "learning_rate": 0.0, |
| "loss": 2.3812, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.0013833650354487291, |
| "grad_norm": 0.3372870683670044, |
| "learning_rate": 1.3698630136986302e-06, |
| "loss": 2.6562, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.0020750475531730937, |
| "grad_norm": 0.41708794236183167, |
| "learning_rate": 2.7397260273972604e-06, |
| "loss": 2.779, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.0027667300708974583, |
| "grad_norm": 0.48284393548965454, |
| "learning_rate": 4.10958904109589e-06, |
| "loss": 2.83, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.0034584125886218224, |
| "grad_norm": 0.328521728515625, |
| "learning_rate": 5.479452054794521e-06, |
| "loss": 2.3607, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.004150095106346187, |
| "grad_norm": 0.3413754403591156, |
| "learning_rate": 6.849315068493151e-06, |
| "loss": 1.9744, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.0048417776240705515, |
| "grad_norm": 0.43425995111465454, |
| "learning_rate": 8.21917808219178e-06, |
| "loss": 1.915, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.0055334601417949165, |
| "grad_norm": 0.45184195041656494, |
| "learning_rate": 9.589041095890411e-06, |
| "loss": 2.0002, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.006225142659519281, |
| "grad_norm": 0.3451146185398102, |
| "learning_rate": 1.0958904109589042e-05, |
| "loss": 2.5739, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.006916825177243645, |
| "grad_norm": 0.39400503039360046, |
| "learning_rate": 1.2328767123287671e-05, |
| "loss": 2.6139, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.00760850769496801, |
| "grad_norm": 0.4555888772010803, |
| "learning_rate": 1.3698630136986302e-05, |
| "loss": 2.4346, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.008300190212692375, |
| "grad_norm": 0.5475619435310364, |
| "learning_rate": 1.5068493150684931e-05, |
| "loss": 1.485, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.008991872730416739, |
| "grad_norm": 1.219076156616211, |
| "learning_rate": 1.643835616438356e-05, |
| "loss": 2.0555, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.009683555248141103, |
| "grad_norm": 0.3271433711051941, |
| "learning_rate": 1.780821917808219e-05, |
| "loss": 2.8418, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.010375237765865467, |
| "grad_norm": 0.3261394500732422, |
| "learning_rate": 1.9178082191780822e-05, |
| "loss": 2.287, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.011066920283589833, |
| "grad_norm": 0.39713796973228455, |
| "learning_rate": 2.0547945205479453e-05, |
| "loss": 2.7331, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.011758602801314197, |
| "grad_norm": 0.3226444125175476, |
| "learning_rate": 2.1917808219178083e-05, |
| "loss": 2.7307, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.012450285319038561, |
| "grad_norm": 0.4153122007846832, |
| "learning_rate": 2.328767123287671e-05, |
| "loss": 2.3229, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.013141967836762926, |
| "grad_norm": 0.29302462935447693, |
| "learning_rate": 2.4657534246575342e-05, |
| "loss": 2.2093, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.01383365035448729, |
| "grad_norm": 0.42535701394081116, |
| "learning_rate": 2.6027397260273973e-05, |
| "loss": 2.3698, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.014525332872211656, |
| "grad_norm": 0.33567023277282715, |
| "learning_rate": 2.7397260273972603e-05, |
| "loss": 2.6456, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.01521701538993602, |
| "grad_norm": 0.37375620007514954, |
| "learning_rate": 2.8767123287671234e-05, |
| "loss": 2.4148, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.015908697907660384, |
| "grad_norm": 0.2704203128814697, |
| "learning_rate": 3.0136986301369862e-05, |
| "loss": 1.6722, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.01660038042538475, |
| "grad_norm": 0.3946782946586609, |
| "learning_rate": 3.1506849315068496e-05, |
| "loss": 2.6759, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.017292062943109112, |
| "grad_norm": 0.368335098028183, |
| "learning_rate": 3.287671232876712e-05, |
| "loss": 2.6882, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.017983745460833478, |
| "grad_norm": 0.38029783964157104, |
| "learning_rate": 3.424657534246575e-05, |
| "loss": 2.6915, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.01867542797855784, |
| "grad_norm": 0.36253222823143005, |
| "learning_rate": 3.561643835616438e-05, |
| "loss": 2.1426, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.019367110496282206, |
| "grad_norm": 0.35769322514533997, |
| "learning_rate": 3.698630136986301e-05, |
| "loss": 2.6634, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.020058793014006572, |
| "grad_norm": 0.44577229022979736, |
| "learning_rate": 3.8356164383561644e-05, |
| "loss": 2.5996, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.020750475531730934, |
| "grad_norm": 0.5225628614425659, |
| "learning_rate": 3.9726027397260274e-05, |
| "loss": 2.5993, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.0214421580494553, |
| "grad_norm": 0.45649707317352295, |
| "learning_rate": 4.1095890410958905e-05, |
| "loss": 2.3344, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.022133840567179666, |
| "grad_norm": 0.31408172845840454, |
| "learning_rate": 4.2465753424657536e-05, |
| "loss": 2.343, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.02282552308490403, |
| "grad_norm": 0.3498693108558655, |
| "learning_rate": 4.383561643835617e-05, |
| "loss": 1.7196, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.023517205602628394, |
| "grad_norm": 0.5255292654037476, |
| "learning_rate": 4.520547945205479e-05, |
| "loss": 1.5354, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.024208888120352757, |
| "grad_norm": 0.3460487425327301, |
| "learning_rate": 4.657534246575342e-05, |
| "loss": 2.1435, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.024900570638077123, |
| "grad_norm": 0.47466233372688293, |
| "learning_rate": 4.794520547945205e-05, |
| "loss": 2.7025, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.02559225315580149, |
| "grad_norm": 0.39667466282844543, |
| "learning_rate": 4.9315068493150684e-05, |
| "loss": 2.2736, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.02628393567352585, |
| "grad_norm": 0.40798133611679077, |
| "learning_rate": 5.068493150684932e-05, |
| "loss": 1.4074, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.026975618191250217, |
| "grad_norm": 0.39603471755981445, |
| "learning_rate": 5.2054794520547945e-05, |
| "loss": 2.4354, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.02766730070897458, |
| "grad_norm": 0.5650593042373657, |
| "learning_rate": 5.342465753424658e-05, |
| "loss": 1.7307, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.028358983226698945, |
| "grad_norm": 0.7711919546127319, |
| "learning_rate": 5.479452054794521e-05, |
| "loss": 1.2067, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.02905066574442331, |
| "grad_norm": 0.46603119373321533, |
| "learning_rate": 5.616438356164384e-05, |
| "loss": 1.0539, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.029742348262147673, |
| "grad_norm": 0.2997550666332245, |
| "learning_rate": 5.753424657534247e-05, |
| "loss": 1.9997, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.03043403077987204, |
| "grad_norm": 0.4058617651462555, |
| "learning_rate": 5.89041095890411e-05, |
| "loss": 2.3323, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.0311257132975964, |
| "grad_norm": 0.3763635754585266, |
| "learning_rate": 6.0273972602739724e-05, |
| "loss": 1.8749, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.03181739581532077, |
| "grad_norm": 0.3507993221282959, |
| "learning_rate": 6.164383561643835e-05, |
| "loss": 2.3828, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.03250907833304513, |
| "grad_norm": 0.34240859746932983, |
| "learning_rate": 6.301369863013699e-05, |
| "loss": 2.3809, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.0332007608507695, |
| "grad_norm": 0.4183844029903412, |
| "learning_rate": 6.438356164383562e-05, |
| "loss": 2.2065, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.03389244336849386, |
| "grad_norm": 0.5209120512008667, |
| "learning_rate": 6.575342465753424e-05, |
| "loss": 2.061, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.034584125886218224, |
| "grad_norm": 0.6568111181259155, |
| "learning_rate": 6.712328767123288e-05, |
| "loss": 2.1576, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.03527580840394259, |
| "grad_norm": 0.5327390432357788, |
| "learning_rate": 6.84931506849315e-05, |
| "loss": 1.5562, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.035967490921666956, |
| "grad_norm": 0.6389997005462646, |
| "learning_rate": 6.986301369863014e-05, |
| "loss": 2.3966, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.03665917343939132, |
| "grad_norm": 0.6374388933181763, |
| "learning_rate": 7.123287671232876e-05, |
| "loss": 2.4013, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.03735085595711568, |
| "grad_norm": 0.4383719563484192, |
| "learning_rate": 7.26027397260274e-05, |
| "loss": 2.061, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.03804253847484005, |
| "grad_norm": 0.4306814670562744, |
| "learning_rate": 7.397260273972603e-05, |
| "loss": 2.4931, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.03873422099256441, |
| "grad_norm": 0.5693032145500183, |
| "learning_rate": 7.534246575342466e-05, |
| "loss": 2.2842, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.039425903510288775, |
| "grad_norm": 0.7792916297912598, |
| "learning_rate": 7.671232876712329e-05, |
| "loss": 1.8452, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.040117586028013144, |
| "grad_norm": 0.4936494827270508, |
| "learning_rate": 7.808219178082192e-05, |
| "loss": 1.8668, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.040809268545737507, |
| "grad_norm": 0.49626269936561584, |
| "learning_rate": 7.945205479452055e-05, |
| "loss": 1.9564, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.04150095106346187, |
| "grad_norm": 0.4702766537666321, |
| "learning_rate": 8.082191780821919e-05, |
| "loss": 2.419, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.04219263358118624, |
| "grad_norm": 0.6959900259971619, |
| "learning_rate": 8.219178082191781e-05, |
| "loss": 1.5835, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.0428843160989106, |
| "grad_norm": 0.5963976979255676, |
| "learning_rate": 8.356164383561645e-05, |
| "loss": 2.2932, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.04357599861663496, |
| "grad_norm": 0.4605785310268402, |
| "learning_rate": 8.493150684931507e-05, |
| "loss": 2.397, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.04426768113435933, |
| "grad_norm": 0.4479796588420868, |
| "learning_rate": 8.630136986301371e-05, |
| "loss": 2.0257, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.044959363652083695, |
| "grad_norm": 0.4342099726200104, |
| "learning_rate": 8.767123287671233e-05, |
| "loss": 1.9476, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.04565104616980806, |
| "grad_norm": 0.44643279910087585, |
| "learning_rate": 8.904109589041096e-05, |
| "loss": 2.4669, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.04634272868753242, |
| "grad_norm": 0.513254702091217, |
| "learning_rate": 9.041095890410958e-05, |
| "loss": 2.2581, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.04703441120525679, |
| "grad_norm": 0.6621958613395691, |
| "learning_rate": 9.178082191780822e-05, |
| "loss": 1.3013, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.04772609372298115, |
| "grad_norm": 0.43388229608535767, |
| "learning_rate": 9.315068493150684e-05, |
| "loss": 2.5325, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.048417776240705514, |
| "grad_norm": 0.539757490158081, |
| "learning_rate": 9.452054794520548e-05, |
| "loss": 1.76, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.04910945875842988, |
| "grad_norm": 0.33598029613494873, |
| "learning_rate": 9.58904109589041e-05, |
| "loss": 1.6931, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.049801141276154245, |
| "grad_norm": 0.8661749958992004, |
| "learning_rate": 9.726027397260274e-05, |
| "loss": 2.1326, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.05049282379387861, |
| "grad_norm": 0.5956202149391174, |
| "learning_rate": 9.863013698630137e-05, |
| "loss": 2.0554, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.05118450631160298, |
| "grad_norm": 0.7723128199577332, |
| "learning_rate": 0.0001, |
| "loss": 1.4837, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.05187618882932734, |
| "grad_norm": 0.5357879996299744, |
| "learning_rate": 9.992716678805537e-05, |
| "loss": 2.1688, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.0525678713470517, |
| "grad_norm": 0.48623108863830566, |
| "learning_rate": 9.985433357611071e-05, |
| "loss": 2.1567, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.053259553864776064, |
| "grad_norm": 0.4244164228439331, |
| "learning_rate": 9.978150036416607e-05, |
| "loss": 1.8235, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.053951236382500434, |
| "grad_norm": 0.6056337356567383, |
| "learning_rate": 9.970866715222141e-05, |
| "loss": 2.0154, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.054642918900224796, |
| "grad_norm": 0.5280382037162781, |
| "learning_rate": 9.963583394027677e-05, |
| "loss": 1.9959, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.05533460141794916, |
| "grad_norm": 0.44745901226997375, |
| "learning_rate": 9.956300072833212e-05, |
| "loss": 1.2171, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.05602628393567353, |
| "grad_norm": 0.7180289030075073, |
| "learning_rate": 9.949016751638748e-05, |
| "loss": 1.7036, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.05671796645339789, |
| "grad_norm": 0.6465222239494324, |
| "learning_rate": 9.941733430444284e-05, |
| "loss": 1.8992, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.05740964897112225, |
| "grad_norm": 0.4792153239250183, |
| "learning_rate": 9.934450109249819e-05, |
| "loss": 2.0712, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.05810133148884662, |
| "grad_norm": 0.47839808464050293, |
| "learning_rate": 9.927166788055353e-05, |
| "loss": 1.7548, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.058793014006570984, |
| "grad_norm": 0.7118557095527649, |
| "learning_rate": 9.919883466860888e-05, |
| "loss": 1.5835, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.05948469652429535, |
| "grad_norm": 0.5179592370986938, |
| "learning_rate": 9.912600145666424e-05, |
| "loss": 1.621, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.060176379042019716, |
| "grad_norm": 0.9235703349113464, |
| "learning_rate": 9.905316824471959e-05, |
| "loss": 0.8581, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.06086806155974408, |
| "grad_norm": 0.8734163045883179, |
| "learning_rate": 9.898033503277495e-05, |
| "loss": 2.2878, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.06155974407746844, |
| "grad_norm": 0.45049625635147095, |
| "learning_rate": 9.890750182083031e-05, |
| "loss": 2.2186, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.0622514265951928, |
| "grad_norm": 0.6119429469108582, |
| "learning_rate": 9.883466860888566e-05, |
| "loss": 2.2461, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.06294310911291717, |
| "grad_norm": 0.5095340013504028, |
| "learning_rate": 9.876183539694101e-05, |
| "loss": 2.2954, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.06363479163064154, |
| "grad_norm": 0.6094918251037598, |
| "learning_rate": 9.868900218499635e-05, |
| "loss": 2.3478, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.0643264741483659, |
| "grad_norm": 1.8420301675796509, |
| "learning_rate": 9.861616897305172e-05, |
| "loss": 2.213, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.06501815666609026, |
| "grad_norm": 0.6151532530784607, |
| "learning_rate": 9.854333576110706e-05, |
| "loss": 1.8737, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.06570983918381462, |
| "grad_norm": 0.5210903286933899, |
| "learning_rate": 9.847050254916242e-05, |
| "loss": 2.019, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.066401521701539, |
| "grad_norm": 0.535746693611145, |
| "learning_rate": 9.839766933721779e-05, |
| "loss": 2.399, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.06709320421926336, |
| "grad_norm": 0.8154505491256714, |
| "learning_rate": 9.832483612527313e-05, |
| "loss": 1.9442, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.06778488673698772, |
| "grad_norm": 0.521247923374176, |
| "learning_rate": 9.825200291332848e-05, |
| "loss": 2.5264, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.06847656925471209, |
| "grad_norm": 0.41544124484062195, |
| "learning_rate": 9.817916970138383e-05, |
| "loss": 1.8267, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.06916825177243645, |
| "grad_norm": 0.538253128528595, |
| "learning_rate": 9.810633648943919e-05, |
| "loss": 1.6463, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.06985993429016081, |
| "grad_norm": 0.8509485125541687, |
| "learning_rate": 9.803350327749454e-05, |
| "loss": 1.5877, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.07055161680788519, |
| "grad_norm": 0.47714293003082275, |
| "learning_rate": 9.79606700655499e-05, |
| "loss": 1.5043, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.07124329932560955, |
| "grad_norm": 0.5433914065361023, |
| "learning_rate": 9.788783685360526e-05, |
| "loss": 1.6454, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.07193498184333391, |
| "grad_norm": 0.6400073766708374, |
| "learning_rate": 9.781500364166059e-05, |
| "loss": 1.6994, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.07262666436105827, |
| "grad_norm": 0.6944612860679626, |
| "learning_rate": 9.774217042971595e-05, |
| "loss": 2.2063, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.07331834687878264, |
| "grad_norm": 0.6701168417930603, |
| "learning_rate": 9.76693372177713e-05, |
| "loss": 1.2487, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.074010029396507, |
| "grad_norm": 0.7888538837432861, |
| "learning_rate": 9.759650400582666e-05, |
| "loss": 1.8784, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.07470171191423136, |
| "grad_norm": 0.7106865048408508, |
| "learning_rate": 9.752367079388201e-05, |
| "loss": 2.3673, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.07539339443195574, |
| "grad_norm": 0.49588626623153687, |
| "learning_rate": 9.745083758193737e-05, |
| "loss": 1.8638, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.0760850769496801, |
| "grad_norm": 0.48046427965164185, |
| "learning_rate": 9.737800436999273e-05, |
| "loss": 2.333, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.07677675946740446, |
| "grad_norm": 0.6239060163497925, |
| "learning_rate": 9.730517115804807e-05, |
| "loss": 1.8378, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.07746844198512882, |
| "grad_norm": 0.6083034873008728, |
| "learning_rate": 9.723233794610343e-05, |
| "loss": 1.9826, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.07816012450285319, |
| "grad_norm": 0.8679235577583313, |
| "learning_rate": 9.715950473415877e-05, |
| "loss": 1.3375, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.07885180702057755, |
| "grad_norm": 0.542757511138916, |
| "learning_rate": 9.708667152221414e-05, |
| "loss": 1.6349, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.07954348953830193, |
| "grad_norm": 0.5968992114067078, |
| "learning_rate": 9.701383831026948e-05, |
| "loss": 1.3598, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.08023517205602629, |
| "grad_norm": 0.5693714022636414, |
| "learning_rate": 9.694100509832484e-05, |
| "loss": 2.1527, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.08092685457375065, |
| "grad_norm": 0.6307690143585205, |
| "learning_rate": 9.68681718863802e-05, |
| "loss": 1.7178, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.08161853709147501, |
| "grad_norm": 0.5192516446113586, |
| "learning_rate": 9.679533867443554e-05, |
| "loss": 1.8115, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.08231021960919938, |
| "grad_norm": 1.036036729812622, |
| "learning_rate": 9.67225054624909e-05, |
| "loss": 2.2423, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.08300190212692374, |
| "grad_norm": 0.6450679302215576, |
| "learning_rate": 9.664967225054625e-05, |
| "loss": 1.9298, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.0836935846446481, |
| "grad_norm": 0.8683953881263733, |
| "learning_rate": 9.657683903860161e-05, |
| "loss": 2.0791, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.08438526716237248, |
| "grad_norm": 0.7013575434684753, |
| "learning_rate": 9.650400582665696e-05, |
| "loss": 2.0814, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.08507694968009684, |
| "grad_norm": 0.6849325299263, |
| "learning_rate": 9.643117261471232e-05, |
| "loss": 1.83, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.0857686321978212, |
| "grad_norm": 0.9820392727851868, |
| "learning_rate": 9.635833940276767e-05, |
| "loss": 1.5665, |
| "step": 124 |
| }, |
| { |
| "epoch": 0.08646031471554556, |
| "grad_norm": 0.5546866655349731, |
| "learning_rate": 9.628550619082301e-05, |
| "loss": 1.7215, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.08715199723326993, |
| "grad_norm": 0.6748363971710205, |
| "learning_rate": 9.621267297887837e-05, |
| "loss": 1.6532, |
| "step": 126 |
| }, |
| { |
| "epoch": 0.08784367975099429, |
| "grad_norm": 0.4760434627532959, |
| "learning_rate": 9.613983976693372e-05, |
| "loss": 1.3476, |
| "step": 127 |
| }, |
| { |
| "epoch": 0.08853536226871866, |
| "grad_norm": 0.5710906386375427, |
| "learning_rate": 9.606700655498908e-05, |
| "loss": 2.025, |
| "step": 128 |
| }, |
| { |
| "epoch": 0.08922704478644303, |
| "grad_norm": 0.5950302481651306, |
| "learning_rate": 9.599417334304443e-05, |
| "loss": 2.6324, |
| "step": 129 |
| }, |
| { |
| "epoch": 0.08991872730416739, |
| "grad_norm": 0.4994860887527466, |
| "learning_rate": 9.592134013109979e-05, |
| "loss": 2.0315, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.09061040982189175, |
| "grad_norm": 0.6400249004364014, |
| "learning_rate": 9.584850691915514e-05, |
| "loss": 2.3628, |
| "step": 131 |
| }, |
| { |
| "epoch": 0.09130209233961611, |
| "grad_norm": 1.0873314142227173, |
| "learning_rate": 9.577567370721049e-05, |
| "loss": 1.2664, |
| "step": 132 |
| }, |
| { |
| "epoch": 0.09199377485734048, |
| "grad_norm": 0.5430288314819336, |
| "learning_rate": 9.570284049526585e-05, |
| "loss": 1.7676, |
| "step": 133 |
| }, |
| { |
| "epoch": 0.09268545737506484, |
| "grad_norm": 0.5969283580780029, |
| "learning_rate": 9.56300072833212e-05, |
| "loss": 2.0824, |
| "step": 134 |
| }, |
| { |
| "epoch": 0.09337713989278922, |
| "grad_norm": 0.6228020191192627, |
| "learning_rate": 9.555717407137656e-05, |
| "loss": 2.0687, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.09406882241051358, |
| "grad_norm": 0.6905536651611328, |
| "learning_rate": 9.54843408594319e-05, |
| "loss": 2.2948, |
| "step": 136 |
| }, |
| { |
| "epoch": 0.09476050492823794, |
| "grad_norm": 0.6397948861122131, |
| "learning_rate": 9.541150764748726e-05, |
| "loss": 1.4147, |
| "step": 137 |
| }, |
| { |
| "epoch": 0.0954521874459623, |
| "grad_norm": 0.5489908456802368, |
| "learning_rate": 9.533867443554261e-05, |
| "loss": 2.0582, |
| "step": 138 |
| }, |
| { |
| "epoch": 0.09614386996368667, |
| "grad_norm": 0.5713220834732056, |
| "learning_rate": 9.526584122359796e-05, |
| "loss": 2.1708, |
| "step": 139 |
| }, |
| { |
| "epoch": 0.09683555248141103, |
| "grad_norm": 0.6736558079719543, |
| "learning_rate": 9.519300801165332e-05, |
| "loss": 1.6808, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.0975272349991354, |
| "grad_norm": 1.1222553253173828, |
| "learning_rate": 9.512017479970867e-05, |
| "loss": 1.563, |
| "step": 141 |
| }, |
| { |
| "epoch": 0.09821891751685977, |
| "grad_norm": 0.6231582760810852, |
| "learning_rate": 9.504734158776403e-05, |
| "loss": 1.6458, |
| "step": 142 |
| }, |
| { |
| "epoch": 0.09891060003458413, |
| "grad_norm": 0.716089129447937, |
| "learning_rate": 9.497450837581938e-05, |
| "loss": 1.7055, |
| "step": 143 |
| }, |
| { |
| "epoch": 0.09960228255230849, |
| "grad_norm": 0.49629953503608704, |
| "learning_rate": 9.490167516387472e-05, |
| "loss": 1.2985, |
| "step": 144 |
| }, |
| { |
| "epoch": 0.10029396507003285, |
| "grad_norm": 0.8354843258857727, |
| "learning_rate": 9.482884195193008e-05, |
| "loss": 1.8557, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.10098564758775722, |
| "grad_norm": 0.7507709860801697, |
| "learning_rate": 9.475600873998543e-05, |
| "loss": 1.1816, |
| "step": 146 |
| }, |
| { |
| "epoch": 0.10167733010548158, |
| "grad_norm": 0.8973211050033569, |
| "learning_rate": 9.46831755280408e-05, |
| "loss": 1.4495, |
| "step": 147 |
| }, |
| { |
| "epoch": 0.10236901262320595, |
| "grad_norm": 0.7720927000045776, |
| "learning_rate": 9.461034231609614e-05, |
| "loss": 2.1743, |
| "step": 148 |
| }, |
| { |
| "epoch": 0.10306069514093032, |
| "grad_norm": 0.511075496673584, |
| "learning_rate": 9.45375091041515e-05, |
| "loss": 1.8267, |
| "step": 149 |
| }, |
| { |
| "epoch": 0.10375237765865468, |
| "grad_norm": 0.6523205041885376, |
| "learning_rate": 9.446467589220685e-05, |
| "loss": 1.9682, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.10444406017637904, |
| "grad_norm": 0.703565239906311, |
| "learning_rate": 9.43918426802622e-05, |
| "loss": 2.3272, |
| "step": 151 |
| }, |
| { |
| "epoch": 0.1051357426941034, |
| "grad_norm": 0.5773376226425171, |
| "learning_rate": 9.431900946831756e-05, |
| "loss": 1.9753, |
| "step": 152 |
| }, |
| { |
| "epoch": 0.10582742521182777, |
| "grad_norm": 0.6234054565429688, |
| "learning_rate": 9.42461762563729e-05, |
| "loss": 2.5085, |
| "step": 153 |
| }, |
| { |
| "epoch": 0.10651910772955213, |
| "grad_norm": 0.7076170444488525, |
| "learning_rate": 9.417334304442827e-05, |
| "loss": 1.9139, |
| "step": 154 |
| }, |
| { |
| "epoch": 0.1072107902472765, |
| "grad_norm": 0.5035070776939392, |
| "learning_rate": 9.410050983248361e-05, |
| "loss": 2.0164, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.10790247276500087, |
| "grad_norm": 1.046075701713562, |
| "learning_rate": 9.402767662053898e-05, |
| "loss": 1.2478, |
| "step": 156 |
| }, |
| { |
| "epoch": 0.10859415528272523, |
| "grad_norm": 0.5620921850204468, |
| "learning_rate": 9.395484340859432e-05, |
| "loss": 1.565, |
| "step": 157 |
| }, |
| { |
| "epoch": 0.10928583780044959, |
| "grad_norm": 0.8284108638763428, |
| "learning_rate": 9.388201019664967e-05, |
| "loss": 2.0823, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.10997752031817395, |
| "grad_norm": 0.584373950958252, |
| "learning_rate": 9.380917698470503e-05, |
| "loss": 1.6352, |
| "step": 159 |
| }, |
| { |
| "epoch": 0.11066920283589832, |
| "grad_norm": 0.7779399752616882, |
| "learning_rate": 9.373634377276038e-05, |
| "loss": 1.83, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.1113608853536227, |
| "grad_norm": 0.6635437607765198, |
| "learning_rate": 9.366351056081574e-05, |
| "loss": 1.8935, |
| "step": 161 |
| }, |
| { |
| "epoch": 0.11205256787134706, |
| "grad_norm": 0.6441919207572937, |
| "learning_rate": 9.359067734887109e-05, |
| "loss": 1.9566, |
| "step": 162 |
| }, |
| { |
| "epoch": 0.11274425038907142, |
| "grad_norm": 1.20408296585083, |
| "learning_rate": 9.351784413692645e-05, |
| "loss": 1.2529, |
| "step": 163 |
| }, |
| { |
| "epoch": 0.11343593290679578, |
| "grad_norm": 0.6320855021476746, |
| "learning_rate": 9.34450109249818e-05, |
| "loss": 1.8071, |
| "step": 164 |
| }, |
| { |
| "epoch": 0.11412761542452014, |
| "grad_norm": 0.5197457075119019, |
| "learning_rate": 9.337217771303714e-05, |
| "loss": 1.7303, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.1148192979422445, |
| "grad_norm": 0.6230719089508057, |
| "learning_rate": 9.32993445010925e-05, |
| "loss": 2.0685, |
| "step": 166 |
| }, |
| { |
| "epoch": 0.11551098045996887, |
| "grad_norm": 0.6185227036476135, |
| "learning_rate": 9.322651128914785e-05, |
| "loss": 1.9485, |
| "step": 167 |
| }, |
| { |
| "epoch": 0.11620266297769324, |
| "grad_norm": 0.7290086150169373, |
| "learning_rate": 9.315367807720321e-05, |
| "loss": 1.6923, |
| "step": 168 |
| }, |
| { |
| "epoch": 0.1168943454954176, |
| "grad_norm": 0.9122477769851685, |
| "learning_rate": 9.308084486525856e-05, |
| "loss": 1.3395, |
| "step": 169 |
| }, |
| { |
| "epoch": 0.11758602801314197, |
| "grad_norm": 0.5755578875541687, |
| "learning_rate": 9.300801165331392e-05, |
| "loss": 1.0775, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.11827771053086633, |
| "grad_norm": 0.6516758799552917, |
| "learning_rate": 9.293517844136927e-05, |
| "loss": 1.6174, |
| "step": 171 |
| }, |
| { |
| "epoch": 0.1189693930485907, |
| "grad_norm": 0.5685858726501465, |
| "learning_rate": 9.286234522942462e-05, |
| "loss": 1.851, |
| "step": 172 |
| }, |
| { |
| "epoch": 0.11966107556631506, |
| "grad_norm": 0.5784198641777039, |
| "learning_rate": 9.278951201747998e-05, |
| "loss": 1.712, |
| "step": 173 |
| }, |
| { |
| "epoch": 0.12035275808403943, |
| "grad_norm": 0.7357056736946106, |
| "learning_rate": 9.271667880553533e-05, |
| "loss": 2.2286, |
| "step": 174 |
| }, |
| { |
| "epoch": 0.1210444406017638, |
| "grad_norm": 0.5122838616371155, |
| "learning_rate": 9.264384559359069e-05, |
| "loss": 1.5581, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.12173612311948816, |
| "grad_norm": 0.6864873170852661, |
| "learning_rate": 9.257101238164603e-05, |
| "loss": 1.2737, |
| "step": 176 |
| }, |
| { |
| "epoch": 0.12242780563721252, |
| "grad_norm": 0.441771000623703, |
| "learning_rate": 9.24981791697014e-05, |
| "loss": 1.2291, |
| "step": 177 |
| }, |
| { |
| "epoch": 0.12311948815493688, |
| "grad_norm": 0.634203314781189, |
| "learning_rate": 9.242534595775674e-05, |
| "loss": 2.1797, |
| "step": 178 |
| }, |
| { |
| "epoch": 0.12381117067266124, |
| "grad_norm": 0.6232409477233887, |
| "learning_rate": 9.235251274581209e-05, |
| "loss": 1.5954, |
| "step": 179 |
| }, |
| { |
| "epoch": 0.1245028531903856, |
| "grad_norm": 0.7475627660751343, |
| "learning_rate": 9.227967953386745e-05, |
| "loss": 2.3014, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.12519453570810998, |
| "grad_norm": 0.7436168789863586, |
| "learning_rate": 9.22068463219228e-05, |
| "loss": 2.1053, |
| "step": 181 |
| }, |
| { |
| "epoch": 0.12588621822583435, |
| "grad_norm": 0.754830002784729, |
| "learning_rate": 9.213401310997816e-05, |
| "loss": 2.2968, |
| "step": 182 |
| }, |
| { |
| "epoch": 0.1265779007435587, |
| "grad_norm": 0.7968791723251343, |
| "learning_rate": 9.206117989803351e-05, |
| "loss": 1.9433, |
| "step": 183 |
| }, |
| { |
| "epoch": 0.12726958326128307, |
| "grad_norm": 0.7031124234199524, |
| "learning_rate": 9.198834668608885e-05, |
| "loss": 1.6162, |
| "step": 184 |
| }, |
| { |
| "epoch": 0.12796126577900743, |
| "grad_norm": 1.5919851064682007, |
| "learning_rate": 9.191551347414422e-05, |
| "loss": 2.3995, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.1286529482967318, |
| "grad_norm": 0.8990558385848999, |
| "learning_rate": 9.184268026219956e-05, |
| "loss": 2.1597, |
| "step": 186 |
| }, |
| { |
| "epoch": 0.12934463081445616, |
| "grad_norm": 0.6514426469802856, |
| "learning_rate": 9.176984705025492e-05, |
| "loss": 2.2172, |
| "step": 187 |
| }, |
| { |
| "epoch": 0.13003631333218052, |
| "grad_norm": 0.5041804313659668, |
| "learning_rate": 9.169701383831027e-05, |
| "loss": 1.6619, |
| "step": 188 |
| }, |
| { |
| "epoch": 0.13072799584990488, |
| "grad_norm": 0.5292882323265076, |
| "learning_rate": 9.162418062636563e-05, |
| "loss": 1.3204, |
| "step": 189 |
| }, |
| { |
| "epoch": 0.13141967836762924, |
| "grad_norm": 0.9279568791389465, |
| "learning_rate": 9.155134741442098e-05, |
| "loss": 1.2023, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.13211136088535363, |
| "grad_norm": 0.8143125772476196, |
| "learning_rate": 9.147851420247633e-05, |
| "loss": 1.5768, |
| "step": 191 |
| }, |
| { |
| "epoch": 0.132803043403078, |
| "grad_norm": 0.9918676614761353, |
| "learning_rate": 9.140568099053169e-05, |
| "loss": 1.3813, |
| "step": 192 |
| }, |
| { |
| "epoch": 0.13349472592080236, |
| "grad_norm": 0.6452721357345581, |
| "learning_rate": 9.133284777858704e-05, |
| "loss": 2.4017, |
| "step": 193 |
| }, |
| { |
| "epoch": 0.13418640843852672, |
| "grad_norm": 0.7932286858558655, |
| "learning_rate": 9.12600145666424e-05, |
| "loss": 1.1203, |
| "step": 194 |
| }, |
| { |
| "epoch": 0.13487809095625108, |
| "grad_norm": 0.5378614068031311, |
| "learning_rate": 9.118718135469774e-05, |
| "loss": 1.9103, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.13556977347397545, |
| "grad_norm": 0.6193660497665405, |
| "learning_rate": 9.11143481427531e-05, |
| "loss": 2.0729, |
| "step": 196 |
| }, |
| { |
| "epoch": 0.1362614559916998, |
| "grad_norm": 0.8437771797180176, |
| "learning_rate": 9.104151493080845e-05, |
| "loss": 2.4214, |
| "step": 197 |
| }, |
| { |
| "epoch": 0.13695313850942417, |
| "grad_norm": 0.8080397248268127, |
| "learning_rate": 9.09686817188638e-05, |
| "loss": 1.2323, |
| "step": 198 |
| }, |
| { |
| "epoch": 0.13764482102714853, |
| "grad_norm": 0.5907486081123352, |
| "learning_rate": 9.089584850691916e-05, |
| "loss": 1.6866, |
| "step": 199 |
| }, |
| { |
| "epoch": 0.1383365035448729, |
| "grad_norm": 0.6559591293334961, |
| "learning_rate": 9.082301529497451e-05, |
| "loss": 2.1268, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.13902818606259726, |
| "grad_norm": 0.6248623728752136, |
| "learning_rate": 9.075018208302987e-05, |
| "loss": 1.2631, |
| "step": 201 |
| }, |
| { |
| "epoch": 0.13971986858032162, |
| "grad_norm": 0.5625057220458984, |
| "learning_rate": 9.067734887108522e-05, |
| "loss": 1.1348, |
| "step": 202 |
| }, |
| { |
| "epoch": 0.14041155109804598, |
| "grad_norm": 0.5977293252944946, |
| "learning_rate": 9.060451565914058e-05, |
| "loss": 1.136, |
| "step": 203 |
| }, |
| { |
| "epoch": 0.14110323361577037, |
| "grad_norm": 0.6973084211349487, |
| "learning_rate": 9.053168244719593e-05, |
| "loss": 2.0764, |
| "step": 204 |
| }, |
| { |
| "epoch": 0.14179491613349474, |
| "grad_norm": 0.761034369468689, |
| "learning_rate": 9.045884923525127e-05, |
| "loss": 2.4017, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.1424865986512191, |
| "grad_norm": 0.7098332047462463, |
| "learning_rate": 9.038601602330664e-05, |
| "loss": 1.6705, |
| "step": 206 |
| }, |
| { |
| "epoch": 0.14317828116894346, |
| "grad_norm": 0.6458803415298462, |
| "learning_rate": 9.031318281136198e-05, |
| "loss": 1.2114, |
| "step": 207 |
| }, |
| { |
| "epoch": 0.14386996368666782, |
| "grad_norm": 0.6271463632583618, |
| "learning_rate": 9.024034959941734e-05, |
| "loss": 1.6756, |
| "step": 208 |
| }, |
| { |
| "epoch": 0.14456164620439219, |
| "grad_norm": 0.6152657866477966, |
| "learning_rate": 9.016751638747269e-05, |
| "loss": 1.5943, |
| "step": 209 |
| }, |
| { |
| "epoch": 0.14525332872211655, |
| "grad_norm": 0.6465914845466614, |
| "learning_rate": 9.009468317552805e-05, |
| "loss": 1.8369, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.1459450112398409, |
| "grad_norm": 2.220533847808838, |
| "learning_rate": 9.00218499635834e-05, |
| "loss": 2.7842, |
| "step": 211 |
| }, |
| { |
| "epoch": 0.14663669375756527, |
| "grad_norm": 0.562271773815155, |
| "learning_rate": 8.994901675163875e-05, |
| "loss": 1.9363, |
| "step": 212 |
| }, |
| { |
| "epoch": 0.14732837627528964, |
| "grad_norm": 0.646512508392334, |
| "learning_rate": 8.987618353969411e-05, |
| "loss": 1.9429, |
| "step": 213 |
| }, |
| { |
| "epoch": 0.148020058793014, |
| "grad_norm": 0.8589336276054382, |
| "learning_rate": 8.980335032774946e-05, |
| "loss": 1.6978, |
| "step": 214 |
| }, |
| { |
| "epoch": 0.14871174131073836, |
| "grad_norm": 0.7746891975402832, |
| "learning_rate": 8.973051711580482e-05, |
| "loss": 1.1804, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.14940342382846272, |
| "grad_norm": 0.809123158454895, |
| "learning_rate": 8.965768390386016e-05, |
| "loss": 1.5176, |
| "step": 216 |
| }, |
| { |
| "epoch": 0.1500951063461871, |
| "grad_norm": 0.6705849170684814, |
| "learning_rate": 8.958485069191553e-05, |
| "loss": 2.1671, |
| "step": 217 |
| }, |
| { |
| "epoch": 0.15078678886391148, |
| "grad_norm": 0.5576826333999634, |
| "learning_rate": 8.951201747997087e-05, |
| "loss": 1.2096, |
| "step": 218 |
| }, |
| { |
| "epoch": 0.15147847138163584, |
| "grad_norm": 0.8005768060684204, |
| "learning_rate": 8.943918426802622e-05, |
| "loss": 2.3528, |
| "step": 219 |
| }, |
| { |
| "epoch": 0.1521701538993602, |
| "grad_norm": 0.7752550840377808, |
| "learning_rate": 8.936635105608158e-05, |
| "loss": 1.7417, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.15286183641708456, |
| "grad_norm": 0.819672703742981, |
| "learning_rate": 8.929351784413693e-05, |
| "loss": 2.0707, |
| "step": 221 |
| }, |
| { |
| "epoch": 0.15355351893480892, |
| "grad_norm": 1.488770842552185, |
| "learning_rate": 8.922068463219229e-05, |
| "loss": 1.7984, |
| "step": 222 |
| }, |
| { |
| "epoch": 0.1542452014525333, |
| "grad_norm": 0.7640134692192078, |
| "learning_rate": 8.914785142024764e-05, |
| "loss": 1.858, |
| "step": 223 |
| }, |
| { |
| "epoch": 0.15493688397025765, |
| "grad_norm": 0.8346788883209229, |
| "learning_rate": 8.907501820830299e-05, |
| "loss": 1.7544, |
| "step": 224 |
| }, |
| { |
| "epoch": 0.155628566487982, |
| "grad_norm": 0.5825446844100952, |
| "learning_rate": 8.900218499635835e-05, |
| "loss": 1.8188, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.15632024900570637, |
| "grad_norm": 0.7010117173194885, |
| "learning_rate": 8.89293517844137e-05, |
| "loss": 1.9475, |
| "step": 226 |
| }, |
| { |
| "epoch": 0.15701193152343074, |
| "grad_norm": 0.5123314261436462, |
| "learning_rate": 8.885651857246906e-05, |
| "loss": 2.1235, |
| "step": 227 |
| }, |
| { |
| "epoch": 0.1577036140411551, |
| "grad_norm": 0.62752366065979, |
| "learning_rate": 8.87836853605244e-05, |
| "loss": 1.872, |
| "step": 228 |
| }, |
| { |
| "epoch": 0.15839529655887946, |
| "grad_norm": 0.6518980264663696, |
| "learning_rate": 8.871085214857976e-05, |
| "loss": 1.4556, |
| "step": 229 |
| }, |
| { |
| "epoch": 0.15908697907660385, |
| "grad_norm": 0.7556332349777222, |
| "learning_rate": 8.863801893663511e-05, |
| "loss": 2.0122, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.15977866159432821, |
| "grad_norm": 0.6503921747207642, |
| "learning_rate": 8.856518572469046e-05, |
| "loss": 2.0475, |
| "step": 231 |
| }, |
| { |
| "epoch": 0.16047034411205258, |
| "grad_norm": 0.7834317088127136, |
| "learning_rate": 8.849235251274582e-05, |
| "loss": 1.8671, |
| "step": 232 |
| }, |
| { |
| "epoch": 0.16116202662977694, |
| "grad_norm": 2.123568534851074, |
| "learning_rate": 8.841951930080117e-05, |
| "loss": 1.4212, |
| "step": 233 |
| }, |
| { |
| "epoch": 0.1618537091475013, |
| "grad_norm": 0.5151258111000061, |
| "learning_rate": 8.834668608885653e-05, |
| "loss": 1.1815, |
| "step": 234 |
| }, |
| { |
| "epoch": 0.16254539166522566, |
| "grad_norm": 0.6445286273956299, |
| "learning_rate": 8.827385287691188e-05, |
| "loss": 2.1009, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.16323707418295003, |
| "grad_norm": 0.9389359951019287, |
| "learning_rate": 8.820101966496724e-05, |
| "loss": 2.3638, |
| "step": 236 |
| }, |
| { |
| "epoch": 0.1639287567006744, |
| "grad_norm": 0.7288535833358765, |
| "learning_rate": 8.812818645302258e-05, |
| "loss": 1.5602, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.16462043921839875, |
| "grad_norm": 0.5802240967750549, |
| "learning_rate": 8.805535324107793e-05, |
| "loss": 1.8588, |
| "step": 238 |
| }, |
| { |
| "epoch": 0.1653121217361231, |
| "grad_norm": 0.6751568913459778, |
| "learning_rate": 8.798252002913329e-05, |
| "loss": 2.0206, |
| "step": 239 |
| }, |
| { |
| "epoch": 0.16600380425384748, |
| "grad_norm": 0.6931467056274414, |
| "learning_rate": 8.790968681718864e-05, |
| "loss": 1.9663, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.16669548677157184, |
| "grad_norm": 0.6931905150413513, |
| "learning_rate": 8.7836853605244e-05, |
| "loss": 1.5422, |
| "step": 241 |
| }, |
| { |
| "epoch": 0.1673871692892962, |
| "grad_norm": 0.7654033899307251, |
| "learning_rate": 8.776402039329935e-05, |
| "loss": 2.2165, |
| "step": 242 |
| }, |
| { |
| "epoch": 0.1680788518070206, |
| "grad_norm": 0.8461145162582397, |
| "learning_rate": 8.769118718135471e-05, |
| "loss": 2.1441, |
| "step": 243 |
| }, |
| { |
| "epoch": 0.16877053432474495, |
| "grad_norm": 0.6745240092277527, |
| "learning_rate": 8.761835396941004e-05, |
| "loss": 2.0226, |
| "step": 244 |
| }, |
| { |
| "epoch": 0.16946221684246932, |
| "grad_norm": 0.9059496521949768, |
| "learning_rate": 8.75455207574654e-05, |
| "loss": 1.3178, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.17015389936019368, |
| "grad_norm": 0.7393327951431274, |
| "learning_rate": 8.747268754552075e-05, |
| "loss": 1.7787, |
| "step": 246 |
| }, |
| { |
| "epoch": 0.17084558187791804, |
| "grad_norm": 1.1192548274993896, |
| "learning_rate": 8.739985433357611e-05, |
| "loss": 1.2677, |
| "step": 247 |
| }, |
| { |
| "epoch": 0.1715372643956424, |
| "grad_norm": 0.500165581703186, |
| "learning_rate": 8.732702112163147e-05, |
| "loss": 1.6749, |
| "step": 248 |
| }, |
| { |
| "epoch": 0.17222894691336676, |
| "grad_norm": 0.7118367552757263, |
| "learning_rate": 8.725418790968682e-05, |
| "loss": 1.6768, |
| "step": 249 |
| }, |
| { |
| "epoch": 0.17292062943109113, |
| "grad_norm": 0.6697540283203125, |
| "learning_rate": 8.718135469774218e-05, |
| "loss": 1.7236, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.1736123119488155, |
| "grad_norm": 0.621541440486908, |
| "learning_rate": 8.710852148579752e-05, |
| "loss": 1.1761, |
| "step": 251 |
| }, |
| { |
| "epoch": 0.17430399446653985, |
| "grad_norm": 1.05433189868927, |
| "learning_rate": 8.703568827385288e-05, |
| "loss": 1.5783, |
| "step": 252 |
| }, |
| { |
| "epoch": 0.17499567698426421, |
| "grad_norm": 0.7111071944236755, |
| "learning_rate": 8.696285506190823e-05, |
| "loss": 1.9291, |
| "step": 253 |
| }, |
| { |
| "epoch": 0.17568735950198858, |
| "grad_norm": 0.7538565397262573, |
| "learning_rate": 8.689002184996359e-05, |
| "loss": 1.5598, |
| "step": 254 |
| }, |
| { |
| "epoch": 0.17637904201971294, |
| "grad_norm": 0.7115334868431091, |
| "learning_rate": 8.681718863801895e-05, |
| "loss": 1.7721, |
| "step": 255 |
| }, |
| { |
| "epoch": 0.17707072453743733, |
| "grad_norm": 0.8842628002166748, |
| "learning_rate": 8.67443554260743e-05, |
| "loss": 1.8147, |
| "step": 256 |
| }, |
| { |
| "epoch": 0.1777624070551617, |
| "grad_norm": 0.6481782793998718, |
| "learning_rate": 8.667152221412966e-05, |
| "loss": 1.9606, |
| "step": 257 |
| }, |
| { |
| "epoch": 0.17845408957288605, |
| "grad_norm": 0.719637393951416, |
| "learning_rate": 8.659868900218499e-05, |
| "loss": 1.2452, |
| "step": 258 |
| }, |
| { |
| "epoch": 0.17914577209061042, |
| "grad_norm": 0.6171206831932068, |
| "learning_rate": 8.652585579024035e-05, |
| "loss": 1.367, |
| "step": 259 |
| }, |
| { |
| "epoch": 0.17983745460833478, |
| "grad_norm": 0.618354320526123, |
| "learning_rate": 8.64530225782957e-05, |
| "loss": 1.9576, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.18052913712605914, |
| "grad_norm": 0.6816132664680481, |
| "learning_rate": 8.638018936635106e-05, |
| "loss": 1.1272, |
| "step": 261 |
| }, |
| { |
| "epoch": 0.1812208196437835, |
| "grad_norm": 0.6745633482933044, |
| "learning_rate": 8.630735615440642e-05, |
| "loss": 2.1625, |
| "step": 262 |
| }, |
| { |
| "epoch": 0.18191250216150787, |
| "grad_norm": 0.8869469165802002, |
| "learning_rate": 8.623452294246177e-05, |
| "loss": 2.0639, |
| "step": 263 |
| }, |
| { |
| "epoch": 0.18260418467923223, |
| "grad_norm": 0.9394845366477966, |
| "learning_rate": 8.616168973051712e-05, |
| "loss": 1.6026, |
| "step": 264 |
| }, |
| { |
| "epoch": 0.1832958671969566, |
| "grad_norm": 0.6131375432014465, |
| "learning_rate": 8.608885651857246e-05, |
| "loss": 2.0913, |
| "step": 265 |
| }, |
| { |
| "epoch": 0.18398754971468095, |
| "grad_norm": 0.6758245229721069, |
| "learning_rate": 8.601602330662782e-05, |
| "loss": 1.2526, |
| "step": 266 |
| }, |
| { |
| "epoch": 0.18467923223240532, |
| "grad_norm": 0.7595621943473816, |
| "learning_rate": 8.594319009468317e-05, |
| "loss": 2.1043, |
| "step": 267 |
| }, |
| { |
| "epoch": 0.18537091475012968, |
| "grad_norm": 0.8051510453224182, |
| "learning_rate": 8.587035688273853e-05, |
| "loss": 1.5743, |
| "step": 268 |
| }, |
| { |
| "epoch": 0.18606259726785407, |
| "grad_norm": 0.6870526671409607, |
| "learning_rate": 8.57975236707939e-05, |
| "loss": 1.475, |
| "step": 269 |
| }, |
| { |
| "epoch": 0.18675427978557843, |
| "grad_norm": 0.6781574487686157, |
| "learning_rate": 8.572469045884924e-05, |
| "loss": 2.1719, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.1874459623033028, |
| "grad_norm": 0.7067494988441467, |
| "learning_rate": 8.565185724690459e-05, |
| "loss": 2.3096, |
| "step": 271 |
| }, |
| { |
| "epoch": 0.18813764482102716, |
| "grad_norm": 0.8800467848777771, |
| "learning_rate": 8.557902403495994e-05, |
| "loss": 1.9574, |
| "step": 272 |
| }, |
| { |
| "epoch": 0.18882932733875152, |
| "grad_norm": 0.5349634885787964, |
| "learning_rate": 8.55061908230153e-05, |
| "loss": 1.4169, |
| "step": 273 |
| }, |
| { |
| "epoch": 0.18952100985647588, |
| "grad_norm": 0.6334154009819031, |
| "learning_rate": 8.543335761107065e-05, |
| "loss": 1.9563, |
| "step": 274 |
| }, |
| { |
| "epoch": 0.19021269237420024, |
| "grad_norm": 0.8906685709953308, |
| "learning_rate": 8.536052439912601e-05, |
| "loss": 2.1072, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.1909043748919246, |
| "grad_norm": 0.7288912534713745, |
| "learning_rate": 8.528769118718137e-05, |
| "loss": 1.9452, |
| "step": 276 |
| }, |
| { |
| "epoch": 0.19159605740964897, |
| "grad_norm": 1.1331645250320435, |
| "learning_rate": 8.521485797523672e-05, |
| "loss": 1.9652, |
| "step": 277 |
| }, |
| { |
| "epoch": 0.19228773992737333, |
| "grad_norm": 0.6032379269599915, |
| "learning_rate": 8.514202476329206e-05, |
| "loss": 1.9178, |
| "step": 278 |
| }, |
| { |
| "epoch": 0.1929794224450977, |
| "grad_norm": 0.6424663662910461, |
| "learning_rate": 8.506919155134741e-05, |
| "loss": 1.9931, |
| "step": 279 |
| }, |
| { |
| "epoch": 0.19367110496282205, |
| "grad_norm": 0.5931838750839233, |
| "learning_rate": 8.499635833940277e-05, |
| "loss": 1.2334, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.19436278748054642, |
| "grad_norm": 0.7451290488243103, |
| "learning_rate": 8.492352512745812e-05, |
| "loss": 1.9906, |
| "step": 281 |
| }, |
| { |
| "epoch": 0.1950544699982708, |
| "grad_norm": 0.7306643128395081, |
| "learning_rate": 8.485069191551348e-05, |
| "loss": 2.1793, |
| "step": 282 |
| }, |
| { |
| "epoch": 0.19574615251599517, |
| "grad_norm": 0.6340318918228149, |
| "learning_rate": 8.477785870356884e-05, |
| "loss": 1.3639, |
| "step": 283 |
| }, |
| { |
| "epoch": 0.19643783503371953, |
| "grad_norm": 0.9806922674179077, |
| "learning_rate": 8.470502549162418e-05, |
| "loss": 1.5363, |
| "step": 284 |
| }, |
| { |
| "epoch": 0.1971295175514439, |
| "grad_norm": 0.6847662925720215, |
| "learning_rate": 8.463219227967954e-05, |
| "loss": 1.4774, |
| "step": 285 |
| }, |
| { |
| "epoch": 0.19782120006916826, |
| "grad_norm": 0.5814263224601746, |
| "learning_rate": 8.455935906773488e-05, |
| "loss": 1.4903, |
| "step": 286 |
| }, |
| { |
| "epoch": 0.19851288258689262, |
| "grad_norm": 0.582711935043335, |
| "learning_rate": 8.448652585579024e-05, |
| "loss": 1.479, |
| "step": 287 |
| }, |
| { |
| "epoch": 0.19920456510461698, |
| "grad_norm": 0.8541072607040405, |
| "learning_rate": 8.441369264384559e-05, |
| "loss": 1.6464, |
| "step": 288 |
| }, |
| { |
| "epoch": 0.19989624762234134, |
| "grad_norm": 0.6419352889060974, |
| "learning_rate": 8.434085943190095e-05, |
| "loss": 1.9266, |
| "step": 289 |
| }, |
| { |
| "epoch": 0.2005879301400657, |
| "grad_norm": 0.694685161113739, |
| "learning_rate": 8.426802621995631e-05, |
| "loss": 1.5357, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.20127961265779007, |
| "grad_norm": 0.7046255469322205, |
| "learning_rate": 8.419519300801165e-05, |
| "loss": 1.4103, |
| "step": 291 |
| }, |
| { |
| "epoch": 0.20197129517551443, |
| "grad_norm": 1.0252553224563599, |
| "learning_rate": 8.412235979606701e-05, |
| "loss": 1.8172, |
| "step": 292 |
| }, |
| { |
| "epoch": 0.2026629776932388, |
| "grad_norm": 0.7682600021362305, |
| "learning_rate": 8.404952658412236e-05, |
| "loss": 1.455, |
| "step": 293 |
| }, |
| { |
| "epoch": 0.20335466021096316, |
| "grad_norm": 0.9454272389411926, |
| "learning_rate": 8.397669337217772e-05, |
| "loss": 1.5882, |
| "step": 294 |
| }, |
| { |
| "epoch": 0.20404634272868752, |
| "grad_norm": 1.333929419517517, |
| "learning_rate": 8.390386016023307e-05, |
| "loss": 1.6718, |
| "step": 295 |
| }, |
| { |
| "epoch": 0.2047380252464119, |
| "grad_norm": 0.9410814046859741, |
| "learning_rate": 8.383102694828843e-05, |
| "loss": 2.0226, |
| "step": 296 |
| }, |
| { |
| "epoch": 0.20542970776413627, |
| "grad_norm": 0.7768377065658569, |
| "learning_rate": 8.375819373634379e-05, |
| "loss": 2.0137, |
| "step": 297 |
| }, |
| { |
| "epoch": 0.20612139028186063, |
| "grad_norm": 1.2727937698364258, |
| "learning_rate": 8.368536052439912e-05, |
| "loss": 1.9412, |
| "step": 298 |
| }, |
| { |
| "epoch": 0.206813072799585, |
| "grad_norm": 0.9015520811080933, |
| "learning_rate": 8.361252731245448e-05, |
| "loss": 1.6939, |
| "step": 299 |
| }, |
| { |
| "epoch": 0.20750475531730936, |
| "grad_norm": 0.5938074588775635, |
| "learning_rate": 8.353969410050983e-05, |
| "loss": 1.3246, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.20819643783503372, |
| "grad_norm": 0.7636774182319641, |
| "learning_rate": 8.346686088856519e-05, |
| "loss": 1.687, |
| "step": 301 |
| }, |
| { |
| "epoch": 0.20888812035275808, |
| "grad_norm": 0.8219266533851624, |
| "learning_rate": 8.339402767662054e-05, |
| "loss": 1.9761, |
| "step": 302 |
| }, |
| { |
| "epoch": 0.20957980287048245, |
| "grad_norm": 0.8183760643005371, |
| "learning_rate": 8.33211944646759e-05, |
| "loss": 2.2887, |
| "step": 303 |
| }, |
| { |
| "epoch": 0.2102714853882068, |
| "grad_norm": 0.8675097227096558, |
| "learning_rate": 8.324836125273125e-05, |
| "loss": 2.0024, |
| "step": 304 |
| }, |
| { |
| "epoch": 0.21096316790593117, |
| "grad_norm": 0.5965743660926819, |
| "learning_rate": 8.31755280407866e-05, |
| "loss": 1.4569, |
| "step": 305 |
| }, |
| { |
| "epoch": 0.21165485042365553, |
| "grad_norm": 0.8379791378974915, |
| "learning_rate": 8.310269482884196e-05, |
| "loss": 1.9685, |
| "step": 306 |
| }, |
| { |
| "epoch": 0.2123465329413799, |
| "grad_norm": 0.5711637735366821, |
| "learning_rate": 8.30298616168973e-05, |
| "loss": 1.8149, |
| "step": 307 |
| }, |
| { |
| "epoch": 0.21303821545910426, |
| "grad_norm": 0.8288681507110596, |
| "learning_rate": 8.295702840495266e-05, |
| "loss": 2.0087, |
| "step": 308 |
| }, |
| { |
| "epoch": 0.21372989797682865, |
| "grad_norm": 0.7129418849945068, |
| "learning_rate": 8.288419519300801e-05, |
| "loss": 1.7503, |
| "step": 309 |
| }, |
| { |
| "epoch": 0.214421580494553, |
| "grad_norm": 0.7114739418029785, |
| "learning_rate": 8.281136198106337e-05, |
| "loss": 1.6055, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.21511326301227737, |
| "grad_norm": 0.821076512336731, |
| "learning_rate": 8.273852876911872e-05, |
| "loss": 1.6553, |
| "step": 311 |
| }, |
| { |
| "epoch": 0.21580494553000173, |
| "grad_norm": 0.7488536238670349, |
| "learning_rate": 8.266569555717407e-05, |
| "loss": 1.9126, |
| "step": 312 |
| }, |
| { |
| "epoch": 0.2164966280477261, |
| "grad_norm": 0.822170615196228, |
| "learning_rate": 8.259286234522943e-05, |
| "loss": 2.0362, |
| "step": 313 |
| }, |
| { |
| "epoch": 0.21718831056545046, |
| "grad_norm": 0.7749188542366028, |
| "learning_rate": 8.252002913328478e-05, |
| "loss": 1.6924, |
| "step": 314 |
| }, |
| { |
| "epoch": 0.21787999308317482, |
| "grad_norm": 0.7008628845214844, |
| "learning_rate": 8.244719592134014e-05, |
| "loss": 2.1542, |
| "step": 315 |
| }, |
| { |
| "epoch": 0.21857167560089918, |
| "grad_norm": 0.8683955669403076, |
| "learning_rate": 8.237436270939549e-05, |
| "loss": 1.6697, |
| "step": 316 |
| }, |
| { |
| "epoch": 0.21926335811862355, |
| "grad_norm": 0.7530885338783264, |
| "learning_rate": 8.230152949745085e-05, |
| "loss": 1.2485, |
| "step": 317 |
| }, |
| { |
| "epoch": 0.2199550406363479, |
| "grad_norm": 0.787668764591217, |
| "learning_rate": 8.22286962855062e-05, |
| "loss": 1.1665, |
| "step": 318 |
| }, |
| { |
| "epoch": 0.22064672315407227, |
| "grad_norm": 0.7304548025131226, |
| "learning_rate": 8.215586307356154e-05, |
| "loss": 2.0012, |
| "step": 319 |
| }, |
| { |
| "epoch": 0.22133840567179663, |
| "grad_norm": 0.6933902502059937, |
| "learning_rate": 8.20830298616169e-05, |
| "loss": 1.8554, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.222030088189521, |
| "grad_norm": 1.5794916152954102, |
| "learning_rate": 8.201019664967225e-05, |
| "loss": 1.6497, |
| "step": 321 |
| }, |
| { |
| "epoch": 0.2227217707072454, |
| "grad_norm": 0.788654625415802, |
| "learning_rate": 8.193736343772761e-05, |
| "loss": 1.9427, |
| "step": 322 |
| }, |
| { |
| "epoch": 0.22341345322496975, |
| "grad_norm": 0.6310513019561768, |
| "learning_rate": 8.186453022578296e-05, |
| "loss": 1.1892, |
| "step": 323 |
| }, |
| { |
| "epoch": 0.2241051357426941, |
| "grad_norm": 0.7976911664009094, |
| "learning_rate": 8.17916970138383e-05, |
| "loss": 1.8037, |
| "step": 324 |
| }, |
| { |
| "epoch": 0.22479681826041847, |
| "grad_norm": 0.8486120104789734, |
| "learning_rate": 8.171886380189367e-05, |
| "loss": 0.8769, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.22548850077814284, |
| "grad_norm": 0.6910994648933411, |
| "learning_rate": 8.164603058994901e-05, |
| "loss": 1.2035, |
| "step": 326 |
| }, |
| { |
| "epoch": 0.2261801832958672, |
| "grad_norm": 0.7402215003967285, |
| "learning_rate": 8.157319737800438e-05, |
| "loss": 1.8359, |
| "step": 327 |
| }, |
| { |
| "epoch": 0.22687186581359156, |
| "grad_norm": 0.8741387724876404, |
| "learning_rate": 8.150036416605972e-05, |
| "loss": 2.1804, |
| "step": 328 |
| }, |
| { |
| "epoch": 0.22756354833131592, |
| "grad_norm": 0.7987048625946045, |
| "learning_rate": 8.142753095411508e-05, |
| "loss": 1.8705, |
| "step": 329 |
| }, |
| { |
| "epoch": 0.22825523084904029, |
| "grad_norm": 0.9527040123939514, |
| "learning_rate": 8.135469774217043e-05, |
| "loss": 1.9079, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.22894691336676465, |
| "grad_norm": 0.7261909246444702, |
| "learning_rate": 8.128186453022578e-05, |
| "loss": 1.9903, |
| "step": 331 |
| }, |
| { |
| "epoch": 0.229638595884489, |
| "grad_norm": 0.7130205035209656, |
| "learning_rate": 8.120903131828114e-05, |
| "loss": 1.9952, |
| "step": 332 |
| }, |
| { |
| "epoch": 0.23033027840221337, |
| "grad_norm": 0.6477673649787903, |
| "learning_rate": 8.113619810633649e-05, |
| "loss": 1.4532, |
| "step": 333 |
| }, |
| { |
| "epoch": 0.23102196091993774, |
| "grad_norm": 0.7342897653579712, |
| "learning_rate": 8.106336489439185e-05, |
| "loss": 1.9345, |
| "step": 334 |
| }, |
| { |
| "epoch": 0.23171364343766213, |
| "grad_norm": 0.9584718346595764, |
| "learning_rate": 8.09905316824472e-05, |
| "loss": 2.1087, |
| "step": 335 |
| }, |
| { |
| "epoch": 0.2324053259553865, |
| "grad_norm": 0.7027872204780579, |
| "learning_rate": 8.091769847050256e-05, |
| "loss": 1.9372, |
| "step": 336 |
| }, |
| { |
| "epoch": 0.23309700847311085, |
| "grad_norm": 0.5855789184570312, |
| "learning_rate": 8.08448652585579e-05, |
| "loss": 1.7773, |
| "step": 337 |
| }, |
| { |
| "epoch": 0.2337886909908352, |
| "grad_norm": 0.6886003017425537, |
| "learning_rate": 8.077203204661325e-05, |
| "loss": 1.7705, |
| "step": 338 |
| }, |
| { |
| "epoch": 0.23448037350855958, |
| "grad_norm": 0.6997604966163635, |
| "learning_rate": 8.069919883466861e-05, |
| "loss": 1.3358, |
| "step": 339 |
| }, |
| { |
| "epoch": 0.23517205602628394, |
| "grad_norm": 0.6345038414001465, |
| "learning_rate": 8.062636562272396e-05, |
| "loss": 1.2833, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.2358637385440083, |
| "grad_norm": 0.6772000789642334, |
| "learning_rate": 8.055353241077932e-05, |
| "loss": 1.5009, |
| "step": 341 |
| }, |
| { |
| "epoch": 0.23655542106173266, |
| "grad_norm": 0.7856854200363159, |
| "learning_rate": 8.048069919883467e-05, |
| "loss": 1.8073, |
| "step": 342 |
| }, |
| { |
| "epoch": 0.23724710357945702, |
| "grad_norm": 0.729583740234375, |
| "learning_rate": 8.040786598689003e-05, |
| "loss": 2.3742, |
| "step": 343 |
| }, |
| { |
| "epoch": 0.2379387860971814, |
| "grad_norm": 0.6015090346336365, |
| "learning_rate": 8.033503277494538e-05, |
| "loss": 1.0806, |
| "step": 344 |
| }, |
| { |
| "epoch": 0.23863046861490575, |
| "grad_norm": 0.6671364307403564, |
| "learning_rate": 8.026219956300073e-05, |
| "loss": 1.3656, |
| "step": 345 |
| }, |
| { |
| "epoch": 0.2393221511326301, |
| "grad_norm": 0.7358818054199219, |
| "learning_rate": 8.018936635105609e-05, |
| "loss": 1.769, |
| "step": 346 |
| }, |
| { |
| "epoch": 0.24001383365035447, |
| "grad_norm": 0.697853147983551, |
| "learning_rate": 8.011653313911143e-05, |
| "loss": 1.596, |
| "step": 347 |
| }, |
| { |
| "epoch": 0.24070551616807886, |
| "grad_norm": 0.6559710502624512, |
| "learning_rate": 8.00436999271668e-05, |
| "loss": 1.4413, |
| "step": 348 |
| }, |
| { |
| "epoch": 0.24139719868580323, |
| "grad_norm": 0.770010232925415, |
| "learning_rate": 7.997086671522214e-05, |
| "loss": 2.2503, |
| "step": 349 |
| }, |
| { |
| "epoch": 0.2420888812035276, |
| "grad_norm": 0.6468133926391602, |
| "learning_rate": 7.98980335032775e-05, |
| "loss": 2.0731, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.24278056372125195, |
| "grad_norm": 0.8330799341201782, |
| "learning_rate": 7.982520029133285e-05, |
| "loss": 1.5362, |
| "step": 351 |
| }, |
| { |
| "epoch": 0.24347224623897631, |
| "grad_norm": 1.0678739547729492, |
| "learning_rate": 7.97523670793882e-05, |
| "loss": 1.814, |
| "step": 352 |
| }, |
| { |
| "epoch": 0.24416392875670068, |
| "grad_norm": 0.8693723678588867, |
| "learning_rate": 7.967953386744356e-05, |
| "loss": 2.3669, |
| "step": 353 |
| }, |
| { |
| "epoch": 0.24485561127442504, |
| "grad_norm": 0.7095674872398376, |
| "learning_rate": 7.960670065549891e-05, |
| "loss": 1.2215, |
| "step": 354 |
| }, |
| { |
| "epoch": 0.2455472937921494, |
| "grad_norm": 0.7264676094055176, |
| "learning_rate": 7.953386744355427e-05, |
| "loss": 1.7455, |
| "step": 355 |
| }, |
| { |
| "epoch": 0.24623897630987376, |
| "grad_norm": 0.7226927280426025, |
| "learning_rate": 7.946103423160962e-05, |
| "loss": 1.722, |
| "step": 356 |
| }, |
| { |
| "epoch": 0.24693065882759813, |
| "grad_norm": 0.8582272529602051, |
| "learning_rate": 7.938820101966498e-05, |
| "loss": 2.2391, |
| "step": 357 |
| }, |
| { |
| "epoch": 0.2476223413453225, |
| "grad_norm": 0.633518636226654, |
| "learning_rate": 7.931536780772032e-05, |
| "loss": 1.8529, |
| "step": 358 |
| }, |
| { |
| "epoch": 0.24831402386304685, |
| "grad_norm": 0.9155897498130798, |
| "learning_rate": 7.924253459577567e-05, |
| "loss": 1.5049, |
| "step": 359 |
| }, |
| { |
| "epoch": 0.2490057063807712, |
| "grad_norm": 0.8335781097412109, |
| "learning_rate": 7.916970138383103e-05, |
| "loss": 2.0309, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.2496973888984956, |
| "grad_norm": 0.7219622731208801, |
| "learning_rate": 7.909686817188638e-05, |
| "loss": 1.1584, |
| "step": 361 |
| }, |
| { |
| "epoch": 0.25038907141621997, |
| "grad_norm": 0.8952975869178772, |
| "learning_rate": 7.902403495994174e-05, |
| "loss": 2.1489, |
| "step": 362 |
| }, |
| { |
| "epoch": 0.2510807539339443, |
| "grad_norm": 0.7409587502479553, |
| "learning_rate": 7.895120174799709e-05, |
| "loss": 2.0858, |
| "step": 363 |
| }, |
| { |
| "epoch": 0.2517724364516687, |
| "grad_norm": 0.5998899936676025, |
| "learning_rate": 7.887836853605244e-05, |
| "loss": 1.0058, |
| "step": 364 |
| }, |
| { |
| "epoch": 0.252464118969393, |
| "grad_norm": 0.7379968762397766, |
| "learning_rate": 7.88055353241078e-05, |
| "loss": 1.2178, |
| "step": 365 |
| }, |
| { |
| "epoch": 0.2531558014871174, |
| "grad_norm": 1.2425731420516968, |
| "learning_rate": 7.873270211216315e-05, |
| "loss": 1.5815, |
| "step": 366 |
| }, |
| { |
| "epoch": 0.25384748400484175, |
| "grad_norm": 0.7375080585479736, |
| "learning_rate": 7.86598689002185e-05, |
| "loss": 2.2113, |
| "step": 367 |
| }, |
| { |
| "epoch": 0.25453916652256614, |
| "grad_norm": 1.1793482303619385, |
| "learning_rate": 7.858703568827385e-05, |
| "loss": 1.7387, |
| "step": 368 |
| }, |
| { |
| "epoch": 0.25523084904029053, |
| "grad_norm": 1.397199273109436, |
| "learning_rate": 7.851420247632922e-05, |
| "loss": 1.288, |
| "step": 369 |
| }, |
| { |
| "epoch": 0.25592253155801487, |
| "grad_norm": 0.7143409252166748, |
| "learning_rate": 7.844136926438456e-05, |
| "loss": 1.6329, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.25661421407573926, |
| "grad_norm": 0.6984686255455017, |
| "learning_rate": 7.836853605243991e-05, |
| "loss": 1.7279, |
| "step": 371 |
| }, |
| { |
| "epoch": 0.2573058965934636, |
| "grad_norm": 0.7117382287979126, |
| "learning_rate": 7.829570284049527e-05, |
| "loss": 1.8195, |
| "step": 372 |
| }, |
| { |
| "epoch": 0.257997579111188, |
| "grad_norm": 0.7039769887924194, |
| "learning_rate": 7.822286962855062e-05, |
| "loss": 1.8619, |
| "step": 373 |
| }, |
| { |
| "epoch": 0.2586892616289123, |
| "grad_norm": 1.0917625427246094, |
| "learning_rate": 7.815003641660598e-05, |
| "loss": 1.8889, |
| "step": 374 |
| }, |
| { |
| "epoch": 0.2593809441466367, |
| "grad_norm": 0.6653533577919006, |
| "learning_rate": 7.807720320466133e-05, |
| "loss": 1.8691, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.26007262666436104, |
| "grad_norm": 0.7204722166061401, |
| "learning_rate": 7.800436999271669e-05, |
| "loss": 1.4544, |
| "step": 376 |
| }, |
| { |
| "epoch": 0.26076430918208543, |
| "grad_norm": 0.523810863494873, |
| "learning_rate": 7.793153678077204e-05, |
| "loss": 0.9303, |
| "step": 377 |
| }, |
| { |
| "epoch": 0.26145599169980976, |
| "grad_norm": 0.6838625073432922, |
| "learning_rate": 7.785870356882738e-05, |
| "loss": 1.6911, |
| "step": 378 |
| }, |
| { |
| "epoch": 0.26214767421753415, |
| "grad_norm": 0.7531947493553162, |
| "learning_rate": 7.778587035688274e-05, |
| "loss": 1.8134, |
| "step": 379 |
| }, |
| { |
| "epoch": 0.2628393567352585, |
| "grad_norm": 0.7090226411819458, |
| "learning_rate": 7.771303714493809e-05, |
| "loss": 2.1142, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.2635310392529829, |
| "grad_norm": 0.7452750205993652, |
| "learning_rate": 7.764020393299345e-05, |
| "loss": 1.3778, |
| "step": 381 |
| }, |
| { |
| "epoch": 0.26422272177070727, |
| "grad_norm": 0.709818422794342, |
| "learning_rate": 7.75673707210488e-05, |
| "loss": 1.0387, |
| "step": 382 |
| }, |
| { |
| "epoch": 0.2649144042884316, |
| "grad_norm": 0.7260245680809021, |
| "learning_rate": 7.749453750910416e-05, |
| "loss": 1.5823, |
| "step": 383 |
| }, |
| { |
| "epoch": 0.265606086806156, |
| "grad_norm": 1.0123919248580933, |
| "learning_rate": 7.742170429715951e-05, |
| "loss": 1.7737, |
| "step": 384 |
| }, |
| { |
| "epoch": 0.26629776932388033, |
| "grad_norm": 0.92912757396698, |
| "learning_rate": 7.734887108521486e-05, |
| "loss": 1.0985, |
| "step": 385 |
| }, |
| { |
| "epoch": 0.2669894518416047, |
| "grad_norm": 0.7285180687904358, |
| "learning_rate": 7.727603787327022e-05, |
| "loss": 1.6727, |
| "step": 386 |
| }, |
| { |
| "epoch": 0.26768113435932905, |
| "grad_norm": 0.9794962406158447, |
| "learning_rate": 7.720320466132557e-05, |
| "loss": 2.2499, |
| "step": 387 |
| }, |
| { |
| "epoch": 0.26837281687705344, |
| "grad_norm": 0.8266851902008057, |
| "learning_rate": 7.713037144938093e-05, |
| "loss": 1.7979, |
| "step": 388 |
| }, |
| { |
| "epoch": 0.2690644993947778, |
| "grad_norm": 0.8111055493354797, |
| "learning_rate": 7.705753823743627e-05, |
| "loss": 1.6586, |
| "step": 389 |
| }, |
| { |
| "epoch": 0.26975618191250217, |
| "grad_norm": 0.76368248462677, |
| "learning_rate": 7.698470502549163e-05, |
| "loss": 1.6407, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.2704478644302265, |
| "grad_norm": 0.7116588950157166, |
| "learning_rate": 7.691187181354698e-05, |
| "loss": 2.0222, |
| "step": 391 |
| }, |
| { |
| "epoch": 0.2711395469479509, |
| "grad_norm": 0.7411038875579834, |
| "learning_rate": 7.683903860160233e-05, |
| "loss": 1.9575, |
| "step": 392 |
| }, |
| { |
| "epoch": 0.27183122946567523, |
| "grad_norm": 0.6649416089057922, |
| "learning_rate": 7.676620538965769e-05, |
| "loss": 1.0947, |
| "step": 393 |
| }, |
| { |
| "epoch": 0.2725229119833996, |
| "grad_norm": 0.8183586597442627, |
| "learning_rate": 7.669337217771304e-05, |
| "loss": 1.5777, |
| "step": 394 |
| }, |
| { |
| "epoch": 0.273214594501124, |
| "grad_norm": 0.7400631904602051, |
| "learning_rate": 7.66205389657684e-05, |
| "loss": 1.9312, |
| "step": 395 |
| }, |
| { |
| "epoch": 0.27390627701884834, |
| "grad_norm": 0.6919555068016052, |
| "learning_rate": 7.654770575382375e-05, |
| "loss": 1.8406, |
| "step": 396 |
| }, |
| { |
| "epoch": 0.27459795953657273, |
| "grad_norm": 1.2769502401351929, |
| "learning_rate": 7.647487254187911e-05, |
| "loss": 2.2262, |
| "step": 397 |
| }, |
| { |
| "epoch": 0.27528964205429707, |
| "grad_norm": 0.9513121247291565, |
| "learning_rate": 7.640203932993446e-05, |
| "loss": 2.0865, |
| "step": 398 |
| }, |
| { |
| "epoch": 0.27598132457202146, |
| "grad_norm": 0.7086082100868225, |
| "learning_rate": 7.63292061179898e-05, |
| "loss": 1.8785, |
| "step": 399 |
| }, |
| { |
| "epoch": 0.2766730070897458, |
| "grad_norm": 0.7725284695625305, |
| "learning_rate": 7.625637290604516e-05, |
| "loss": 1.6341, |
| "step": 400 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 1446, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 1, |
| "save_steps": 50, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1.2834936847761408e+16, |
| "train_batch_size": 2, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
