| { |
| "best_metric": 1.3468055725097656, |
| "best_model_checkpoint": "miner_id_24/checkpoint-350", |
| "epoch": 0.0827178729689808, |
| "eval_steps": 50, |
| "global_step": 350, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0002363367799113737, |
| "grad_norm": 0.962568998336792, |
| "learning_rate": 1e-05, |
| "loss": 1.5766, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.0002363367799113737, |
| "eval_loss": 2.307421922683716, |
| "eval_runtime": 544.1619, |
| "eval_samples_per_second": 13.097, |
| "eval_steps_per_second": 3.275, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.0004726735598227474, |
| "grad_norm": 0.836689829826355, |
| "learning_rate": 2e-05, |
| "loss": 1.3413, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.0007090103397341211, |
| "grad_norm": 0.9246699213981628, |
| "learning_rate": 3e-05, |
| "loss": 1.4062, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.0009453471196454948, |
| "grad_norm": 1.0338189601898193, |
| "learning_rate": 4e-05, |
| "loss": 1.5972, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.0011816838995568684, |
| "grad_norm": 0.7805269956588745, |
| "learning_rate": 5e-05, |
| "loss": 1.8295, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.0014180206794682422, |
| "grad_norm": 0.7270001173019409, |
| "learning_rate": 6e-05, |
| "loss": 1.5079, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.001654357459379616, |
| "grad_norm": 1.0106533765792847, |
| "learning_rate": 7e-05, |
| "loss": 1.452, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.0018906942392909896, |
| "grad_norm": 0.8681199550628662, |
| "learning_rate": 8e-05, |
| "loss": 1.7515, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.0021270310192023634, |
| "grad_norm": 0.9652785658836365, |
| "learning_rate": 9e-05, |
| "loss": 1.6405, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.002363367799113737, |
| "grad_norm": 0.8977751135826111, |
| "learning_rate": 0.0001, |
| "loss": 1.7423, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.002599704579025111, |
| "grad_norm": 0.8928416967391968, |
| "learning_rate": 9.99983777858264e-05, |
| "loss": 1.5599, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.0028360413589364844, |
| "grad_norm": 1.0562834739685059, |
| "learning_rate": 9.999351124856874e-05, |
| "loss": 1.8838, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.0030723781388478583, |
| "grad_norm": 1.0885255336761475, |
| "learning_rate": 9.998540070400966e-05, |
| "loss": 1.6491, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.003308714918759232, |
| "grad_norm": 1.4191340208053589, |
| "learning_rate": 9.997404667843075e-05, |
| "loss": 1.3198, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.0035450516986706058, |
| "grad_norm": 1.09172523021698, |
| "learning_rate": 9.995944990857849e-05, |
| "loss": 1.553, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.0037813884785819793, |
| "grad_norm": 1.0828980207443237, |
| "learning_rate": 9.994161134161634e-05, |
| "loss": 1.5503, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.004017725258493353, |
| "grad_norm": 1.0516393184661865, |
| "learning_rate": 9.992053213506334e-05, |
| "loss": 1.6532, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.004254062038404727, |
| "grad_norm": 1.0532169342041016, |
| "learning_rate": 9.989621365671902e-05, |
| "loss": 1.363, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.0044903988183161, |
| "grad_norm": 0.8288836479187012, |
| "learning_rate": 9.986865748457457e-05, |
| "loss": 1.3737, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.004726735598227474, |
| "grad_norm": 1.2724401950836182, |
| "learning_rate": 9.983786540671051e-05, |
| "loss": 1.4115, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.004963072378138848, |
| "grad_norm": 1.1288785934448242, |
| "learning_rate": 9.980383942118066e-05, |
| "loss": 1.5479, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.005199409158050222, |
| "grad_norm": 0.9786710739135742, |
| "learning_rate": 9.976658173588244e-05, |
| "loss": 1.634, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.005435745937961595, |
| "grad_norm": 0.9771286845207214, |
| "learning_rate": 9.972609476841367e-05, |
| "loss": 1.4783, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.005672082717872969, |
| "grad_norm": 1.1386462450027466, |
| "learning_rate": 9.968238114591566e-05, |
| "loss": 1.3894, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.005908419497784343, |
| "grad_norm": 0.932101845741272, |
| "learning_rate": 9.96354437049027e-05, |
| "loss": 1.7236, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.006144756277695717, |
| "grad_norm": 0.9714637994766235, |
| "learning_rate": 9.95852854910781e-05, |
| "loss": 1.4703, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.00638109305760709, |
| "grad_norm": 1.0149480104446411, |
| "learning_rate": 9.953190975913647e-05, |
| "loss": 1.5565, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.006617429837518464, |
| "grad_norm": 1.1948692798614502, |
| "learning_rate": 9.947531997255256e-05, |
| "loss": 1.3331, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.006853766617429837, |
| "grad_norm": 0.9405235052108765, |
| "learning_rate": 9.941551980335652e-05, |
| "loss": 1.5302, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.0070901033973412115, |
| "grad_norm": 0.8745080828666687, |
| "learning_rate": 9.935251313189564e-05, |
| "loss": 1.5957, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.007326440177252585, |
| "grad_norm": 0.8737711310386658, |
| "learning_rate": 9.928630404658255e-05, |
| "loss": 1.3771, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.0075627769571639586, |
| "grad_norm": 0.940108597278595, |
| "learning_rate": 9.921689684362989e-05, |
| "loss": 1.6437, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.007799113737075332, |
| "grad_norm": 0.8977481722831726, |
| "learning_rate": 9.914429602677162e-05, |
| "loss": 1.4883, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.008035450516986706, |
| "grad_norm": 1.0117844343185425, |
| "learning_rate": 9.906850630697068e-05, |
| "loss": 1.3731, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.00827178729689808, |
| "grad_norm": 1.182795524597168, |
| "learning_rate": 9.898953260211338e-05, |
| "loss": 1.5906, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.008508124076809453, |
| "grad_norm": 0.8277016282081604, |
| "learning_rate": 9.890738003669029e-05, |
| "loss": 1.3558, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.008744460856720827, |
| "grad_norm": 0.8979318737983704, |
| "learning_rate": 9.882205394146361e-05, |
| "loss": 1.6458, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.0089807976366322, |
| "grad_norm": 0.9339172840118408, |
| "learning_rate": 9.87335598531214e-05, |
| "loss": 1.4363, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.009217134416543574, |
| "grad_norm": 0.9349825382232666, |
| "learning_rate": 9.864190351391822e-05, |
| "loss": 1.5653, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.009453471196454948, |
| "grad_norm": 0.9626030325889587, |
| "learning_rate": 9.85470908713026e-05, |
| "loss": 1.5098, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.009689807976366323, |
| "grad_norm": 0.9268250465393066, |
| "learning_rate": 9.844912807753104e-05, |
| "loss": 1.5328, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.009926144756277696, |
| "grad_norm": 0.9651059508323669, |
| "learning_rate": 9.834802148926882e-05, |
| "loss": 1.7049, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.01016248153618907, |
| "grad_norm": 0.9303879737854004, |
| "learning_rate": 9.824377766717759e-05, |
| "loss": 1.6356, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.010398818316100443, |
| "grad_norm": 0.8670615553855896, |
| "learning_rate": 9.813640337548954e-05, |
| "loss": 1.7882, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.010635155096011817, |
| "grad_norm": 0.8428188562393188, |
| "learning_rate": 9.802590558156862e-05, |
| "loss": 1.6789, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.01087149187592319, |
| "grad_norm": 0.8441392779350281, |
| "learning_rate": 9.791229145545831e-05, |
| "loss": 1.621, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.011107828655834564, |
| "grad_norm": 1.016707181930542, |
| "learning_rate": 9.779556836941645e-05, |
| "loss": 1.6904, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.011344165435745937, |
| "grad_norm": 1.208844780921936, |
| "learning_rate": 9.767574389743682e-05, |
| "loss": 1.7754, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.011580502215657311, |
| "grad_norm": 1.6422576904296875, |
| "learning_rate": 9.755282581475769e-05, |
| "loss": 1.756, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.011816838995568686, |
| "grad_norm": 1.860146403312683, |
| "learning_rate": 9.742682209735727e-05, |
| "loss": 1.7089, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.011816838995568686, |
| "eval_loss": 1.5719951391220093, |
| "eval_runtime": 546.3214, |
| "eval_samples_per_second": 13.045, |
| "eval_steps_per_second": 3.262, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.01205317577548006, |
| "grad_norm": 1.3921040296554565, |
| "learning_rate": 9.729774092143627e-05, |
| "loss": 1.2475, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.012289512555391433, |
| "grad_norm": 1.468597650527954, |
| "learning_rate": 9.716559066288715e-05, |
| "loss": 1.5212, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.012525849335302807, |
| "grad_norm": 0.9106165170669556, |
| "learning_rate": 9.703037989675087e-05, |
| "loss": 1.3007, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.01276218611521418, |
| "grad_norm": 0.7389653325080872, |
| "learning_rate": 9.689211739666023e-05, |
| "loss": 1.7474, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.012998522895125554, |
| "grad_norm": 0.5950703620910645, |
| "learning_rate": 9.675081213427076e-05, |
| "loss": 1.447, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.013234859675036927, |
| "grad_norm": 0.7542724609375, |
| "learning_rate": 9.66064732786784e-05, |
| "loss": 1.2154, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.0134711964549483, |
| "grad_norm": 0.615863025188446, |
| "learning_rate": 9.645911019582467e-05, |
| "loss": 1.4581, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.013707533234859674, |
| "grad_norm": 0.6719561815261841, |
| "learning_rate": 9.630873244788883e-05, |
| "loss": 1.7195, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.01394387001477105, |
| "grad_norm": 0.8025861978530884, |
| "learning_rate": 9.615534979266745e-05, |
| "loss": 1.4424, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.014180206794682423, |
| "grad_norm": 0.8593571186065674, |
| "learning_rate": 9.599897218294122e-05, |
| "loss": 1.2772, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.014416543574593797, |
| "grad_norm": 0.8140683174133301, |
| "learning_rate": 9.583960976582913e-05, |
| "loss": 1.2782, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.01465288035450517, |
| "grad_norm": 0.8652300834655762, |
| "learning_rate": 9.567727288213005e-05, |
| "loss": 1.2384, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.014889217134416544, |
| "grad_norm": 0.9477348327636719, |
| "learning_rate": 9.551197206565173e-05, |
| "loss": 1.1725, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.015125553914327917, |
| "grad_norm": 0.7401777505874634, |
| "learning_rate": 9.534371804252728e-05, |
| "loss": 1.339, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.01536189069423929, |
| "grad_norm": 0.8628498315811157, |
| "learning_rate": 9.517252173051911e-05, |
| "loss": 1.354, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.015598227474150664, |
| "grad_norm": 0.858137309551239, |
| "learning_rate": 9.49983942383106e-05, |
| "loss": 1.4769, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.015834564254062038, |
| "grad_norm": 0.7888755798339844, |
| "learning_rate": 9.482134686478519e-05, |
| "loss": 1.457, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.016070901033973413, |
| "grad_norm": 0.796262264251709, |
| "learning_rate": 9.464139109829321e-05, |
| "loss": 1.1534, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.016307237813884785, |
| "grad_norm": 0.9140974879264832, |
| "learning_rate": 9.445853861590647e-05, |
| "loss": 1.3714, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.01654357459379616, |
| "grad_norm": 0.7814958095550537, |
| "learning_rate": 9.42728012826605e-05, |
| "loss": 1.2061, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.01677991137370753, |
| "grad_norm": 0.7970321178436279, |
| "learning_rate": 9.408419115078471e-05, |
| "loss": 1.3835, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.017016248153618907, |
| "grad_norm": 0.7458755970001221, |
| "learning_rate": 9.389272045892024e-05, |
| "loss": 1.3189, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.017252584933530282, |
| "grad_norm": 0.8076595664024353, |
| "learning_rate": 9.36984016313259e-05, |
| "loss": 1.3228, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.017488921713441654, |
| "grad_norm": 0.8528698086738586, |
| "learning_rate": 9.350124727707197e-05, |
| "loss": 1.2153, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.01772525849335303, |
| "grad_norm": 0.8382695317268372, |
| "learning_rate": 9.330127018922194e-05, |
| "loss": 1.3955, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.0179615952732644, |
| "grad_norm": 0.8453637957572937, |
| "learning_rate": 9.309848334400246e-05, |
| "loss": 1.2161, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.018197932053175776, |
| "grad_norm": 0.8528453707695007, |
| "learning_rate": 9.289289989996133e-05, |
| "loss": 1.3031, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.018434268833087148, |
| "grad_norm": 0.8365333676338196, |
| "learning_rate": 9.268453319711363e-05, |
| "loss": 1.5613, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.018670605612998523, |
| "grad_norm": 0.7652196884155273, |
| "learning_rate": 9.247339675607605e-05, |
| "loss": 1.2893, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.018906942392909895, |
| "grad_norm": 0.8892850875854492, |
| "learning_rate": 9.225950427718975e-05, |
| "loss": 1.291, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.01914327917282127, |
| "grad_norm": 0.7519558668136597, |
| "learning_rate": 9.204286963963111e-05, |
| "loss": 1.306, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.019379615952732646, |
| "grad_norm": 0.7413076758384705, |
| "learning_rate": 9.182350690051133e-05, |
| "loss": 1.3465, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.019615952732644017, |
| "grad_norm": 0.8368196487426758, |
| "learning_rate": 9.160143029396422e-05, |
| "loss": 1.377, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.019852289512555393, |
| "grad_norm": 0.7968633770942688, |
| "learning_rate": 9.13766542302225e-05, |
| "loss": 1.428, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.020088626292466764, |
| "grad_norm": 1.0490226745605469, |
| "learning_rate": 9.114919329468282e-05, |
| "loss": 1.6381, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.02032496307237814, |
| "grad_norm": 0.8110454678535461, |
| "learning_rate": 9.091906224695935e-05, |
| "loss": 1.6322, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.02056129985228951, |
| "grad_norm": 0.9556715488433838, |
| "learning_rate": 9.068627601992598e-05, |
| "loss": 1.6105, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.020797636632200887, |
| "grad_norm": 0.7827744483947754, |
| "learning_rate": 9.045084971874738e-05, |
| "loss": 1.6826, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.02103397341211226, |
| "grad_norm": 0.9120392799377441, |
| "learning_rate": 9.021279861989885e-05, |
| "loss": 1.6796, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.021270310192023634, |
| "grad_norm": 0.8531889915466309, |
| "learning_rate": 8.997213817017507e-05, |
| "loss": 1.5029, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.02150664697193501, |
| "grad_norm": 0.8959422707557678, |
| "learning_rate": 8.972888398568772e-05, |
| "loss": 1.6082, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.02174298375184638, |
| "grad_norm": 0.8365451693534851, |
| "learning_rate": 8.948305185085225e-05, |
| "loss": 1.637, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.021979320531757756, |
| "grad_norm": 0.8199298977851868, |
| "learning_rate": 8.92346577173636e-05, |
| "loss": 1.7106, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.022215657311669128, |
| "grad_norm": 0.8373245596885681, |
| "learning_rate": 8.898371770316111e-05, |
| "loss": 1.7262, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.022451994091580503, |
| "grad_norm": 0.9236546158790588, |
| "learning_rate": 8.873024809138272e-05, |
| "loss": 1.6798, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.022688330871491875, |
| "grad_norm": 0.8642488718032837, |
| "learning_rate": 8.847426532930831e-05, |
| "loss": 1.6958, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.02292466765140325, |
| "grad_norm": 1.0294744968414307, |
| "learning_rate": 8.821578602729242e-05, |
| "loss": 1.6575, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.023161004431314622, |
| "grad_norm": 1.3789857625961304, |
| "learning_rate": 8.795482695768658e-05, |
| "loss": 1.7238, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.023397341211225997, |
| "grad_norm": 1.3623933792114258, |
| "learning_rate": 8.769140505375085e-05, |
| "loss": 1.4895, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.023633677991137372, |
| "grad_norm": 2.5840892791748047, |
| "learning_rate": 8.742553740855506e-05, |
| "loss": 1.5329, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.023633677991137372, |
| "eval_loss": 1.501173734664917, |
| "eval_runtime": 546.1114, |
| "eval_samples_per_second": 13.05, |
| "eval_steps_per_second": 3.263, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.023870014771048744, |
| "grad_norm": 0.8037025332450867, |
| "learning_rate": 8.715724127386972e-05, |
| "loss": 1.6607, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.02410635155096012, |
| "grad_norm": 0.872471034526825, |
| "learning_rate": 8.688653405904652e-05, |
| "loss": 1.4072, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.02434268833087149, |
| "grad_norm": 0.8332801461219788, |
| "learning_rate": 8.661343332988869e-05, |
| "loss": 1.7243, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.024579025110782866, |
| "grad_norm": 0.7240092754364014, |
| "learning_rate": 8.633795680751116e-05, |
| "loss": 1.5283, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.024815361890694238, |
| "grad_norm": 0.7378276586532593, |
| "learning_rate": 8.606012236719073e-05, |
| "loss": 1.9285, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.025051698670605613, |
| "grad_norm": 0.6166480183601379, |
| "learning_rate": 8.577994803720606e-05, |
| "loss": 1.5375, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.025288035450516985, |
| "grad_norm": 0.5955683588981628, |
| "learning_rate": 8.549745199766792e-05, |
| "loss": 1.4995, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.02552437223042836, |
| "grad_norm": 0.6268562078475952, |
| "learning_rate": 8.521265257933948e-05, |
| "loss": 1.4955, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.025760709010339736, |
| "grad_norm": 0.6342216730117798, |
| "learning_rate": 8.492556826244687e-05, |
| "loss": 1.6321, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.025997045790251107, |
| "grad_norm": 0.6816225051879883, |
| "learning_rate": 8.463621767547998e-05, |
| "loss": 1.4456, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.026233382570162483, |
| "grad_norm": 0.6341848373413086, |
| "learning_rate": 8.434461959398376e-05, |
| "loss": 1.5821, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.026469719350073855, |
| "grad_norm": 0.7276953458786011, |
| "learning_rate": 8.405079293933986e-05, |
| "loss": 1.6067, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.02670605612998523, |
| "grad_norm": 0.770680844783783, |
| "learning_rate": 8.375475677753881e-05, |
| "loss": 1.3039, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.0269423929098966, |
| "grad_norm": 0.7676824927330017, |
| "learning_rate": 8.345653031794292e-05, |
| "loss": 1.6077, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.027178729689807977, |
| "grad_norm": 0.795630931854248, |
| "learning_rate": 8.315613291203976e-05, |
| "loss": 1.287, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.02741506646971935, |
| "grad_norm": 0.8530545830726624, |
| "learning_rate": 8.285358405218655e-05, |
| "loss": 1.1443, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.027651403249630724, |
| "grad_norm": 0.8104541301727295, |
| "learning_rate": 8.25489033703452e-05, |
| "loss": 1.2917, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.0278877400295421, |
| "grad_norm": 0.8394821286201477, |
| "learning_rate": 8.224211063680853e-05, |
| "loss": 1.499, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.02812407680945347, |
| "grad_norm": 0.8282003402709961, |
| "learning_rate": 8.19332257589174e-05, |
| "loss": 1.4619, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.028360413589364846, |
| "grad_norm": 0.8042457103729248, |
| "learning_rate": 8.162226877976887e-05, |
| "loss": 1.2362, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.028596750369276218, |
| "grad_norm": 0.7527115345001221, |
| "learning_rate": 8.130925987691569e-05, |
| "loss": 1.3841, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.028833087149187593, |
| "grad_norm": 0.7210351228713989, |
| "learning_rate": 8.099421936105702e-05, |
| "loss": 1.6026, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.029069423929098965, |
| "grad_norm": 0.7347239255905151, |
| "learning_rate": 8.067716767472045e-05, |
| "loss": 1.4895, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.02930576070901034, |
| "grad_norm": 0.7784380316734314, |
| "learning_rate": 8.035812539093557e-05, |
| "loss": 1.3074, |
| "step": 124 |
| }, |
| { |
| "epoch": 0.029542097488921712, |
| "grad_norm": 0.7604086995124817, |
| "learning_rate": 8.003711321189895e-05, |
| "loss": 1.2598, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.029778434268833087, |
| "grad_norm": 0.7531232237815857, |
| "learning_rate": 7.971415196763088e-05, |
| "loss": 1.4353, |
| "step": 126 |
| }, |
| { |
| "epoch": 0.030014771048744462, |
| "grad_norm": 0.8189026117324829, |
| "learning_rate": 7.938926261462366e-05, |
| "loss": 1.1932, |
| "step": 127 |
| }, |
| { |
| "epoch": 0.030251107828655834, |
| "grad_norm": 0.7173908352851868, |
| "learning_rate": 7.906246623448183e-05, |
| "loss": 1.2883, |
| "step": 128 |
| }, |
| { |
| "epoch": 0.03048744460856721, |
| "grad_norm": 0.7799598574638367, |
| "learning_rate": 7.873378403255419e-05, |
| "loss": 1.4449, |
| "step": 129 |
| }, |
| { |
| "epoch": 0.03072378138847858, |
| "grad_norm": 0.7680227756500244, |
| "learning_rate": 7.840323733655778e-05, |
| "loss": 1.4208, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.030960118168389957, |
| "grad_norm": 1.027715802192688, |
| "learning_rate": 7.807084759519405e-05, |
| "loss": 1.4461, |
| "step": 131 |
| }, |
| { |
| "epoch": 0.03119645494830133, |
| "grad_norm": 0.7231760025024414, |
| "learning_rate": 7.773663637675694e-05, |
| "loss": 1.2763, |
| "step": 132 |
| }, |
| { |
| "epoch": 0.0314327917282127, |
| "grad_norm": 0.8921284079551697, |
| "learning_rate": 7.740062536773352e-05, |
| "loss": 1.3582, |
| "step": 133 |
| }, |
| { |
| "epoch": 0.031669128508124075, |
| "grad_norm": 0.7310764789581299, |
| "learning_rate": 7.706283637139658e-05, |
| "loss": 1.3699, |
| "step": 134 |
| }, |
| { |
| "epoch": 0.03190546528803545, |
| "grad_norm": 0.8492346405982971, |
| "learning_rate": 7.672329130639005e-05, |
| "loss": 1.4663, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.032141802067946826, |
| "grad_norm": 0.9474092721939087, |
| "learning_rate": 7.638201220530665e-05, |
| "loss": 1.5147, |
| "step": 136 |
| }, |
| { |
| "epoch": 0.0323781388478582, |
| "grad_norm": 0.7737847566604614, |
| "learning_rate": 7.603902121325813e-05, |
| "loss": 1.4654, |
| "step": 137 |
| }, |
| { |
| "epoch": 0.03261447562776957, |
| "grad_norm": 0.7279232740402222, |
| "learning_rate": 7.569434058643844e-05, |
| "loss": 1.5139, |
| "step": 138 |
| }, |
| { |
| "epoch": 0.032850812407680945, |
| "grad_norm": 0.842126727104187, |
| "learning_rate": 7.534799269067953e-05, |
| "loss": 1.4975, |
| "step": 139 |
| }, |
| { |
| "epoch": 0.03308714918759232, |
| "grad_norm": 0.8013493418693542, |
| "learning_rate": 7.500000000000001e-05, |
| "loss": 1.3718, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.033323485967503695, |
| "grad_norm": 0.9007391929626465, |
| "learning_rate": 7.465038509514688e-05, |
| "loss": 1.5933, |
| "step": 141 |
| }, |
| { |
| "epoch": 0.03355982274741506, |
| "grad_norm": 0.7821772694587708, |
| "learning_rate": 7.42991706621303e-05, |
| "loss": 1.6399, |
| "step": 142 |
| }, |
| { |
| "epoch": 0.03379615952732644, |
| "grad_norm": 0.8712095618247986, |
| "learning_rate": 7.394637949075154e-05, |
| "loss": 1.6031, |
| "step": 143 |
| }, |
| { |
| "epoch": 0.034032496307237814, |
| "grad_norm": 0.8085826635360718, |
| "learning_rate": 7.35920344731241e-05, |
| "loss": 1.6938, |
| "step": 144 |
| }, |
| { |
| "epoch": 0.03426883308714919, |
| "grad_norm": 0.8882783651351929, |
| "learning_rate": 7.323615860218843e-05, |
| "loss": 1.7896, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.034505169867060564, |
| "grad_norm": 0.9685528874397278, |
| "learning_rate": 7.287877497021978e-05, |
| "loss": 1.5817, |
| "step": 146 |
| }, |
| { |
| "epoch": 0.03474150664697193, |
| "grad_norm": 0.9416519999504089, |
| "learning_rate": 7.251990676732984e-05, |
| "loss": 1.6405, |
| "step": 147 |
| }, |
| { |
| "epoch": 0.03497784342688331, |
| "grad_norm": 1.1194108724594116, |
| "learning_rate": 7.215957727996207e-05, |
| "loss": 1.5226, |
| "step": 148 |
| }, |
| { |
| "epoch": 0.03521418020679468, |
| "grad_norm": 1.6036629676818848, |
| "learning_rate": 7.179780988938051e-05, |
| "loss": 1.7582, |
| "step": 149 |
| }, |
| { |
| "epoch": 0.03545051698670606, |
| "grad_norm": 2.10256028175354, |
| "learning_rate": 7.143462807015271e-05, |
| "loss": 1.3764, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.03545051698670606, |
| "eval_loss": 1.4293397665023804, |
| "eval_runtime": 545.786, |
| "eval_samples_per_second": 13.058, |
| "eval_steps_per_second": 3.265, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.03568685376661743, |
| "grad_norm": 0.5102832317352295, |
| "learning_rate": 7.107005538862646e-05, |
| "loss": 1.0989, |
| "step": 151 |
| }, |
| { |
| "epoch": 0.0359231905465288, |
| "grad_norm": 0.5750204920768738, |
| "learning_rate": 7.07041155014006e-05, |
| "loss": 1.2694, |
| "step": 152 |
| }, |
| { |
| "epoch": 0.03615952732644018, |
| "grad_norm": 0.656173586845398, |
| "learning_rate": 7.033683215379002e-05, |
| "loss": 1.3851, |
| "step": 153 |
| }, |
| { |
| "epoch": 0.03639586410635155, |
| "grad_norm": 0.6856642365455627, |
| "learning_rate": 6.996822917828477e-05, |
| "loss": 1.5173, |
| "step": 154 |
| }, |
| { |
| "epoch": 0.03663220088626293, |
| "grad_norm": 0.7456484436988831, |
| "learning_rate": 6.959833049300377e-05, |
| "loss": 1.5841, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.036868537666174296, |
| "grad_norm": 0.7046188116073608, |
| "learning_rate": 6.922716010014255e-05, |
| "loss": 1.5951, |
| "step": 156 |
| }, |
| { |
| "epoch": 0.03710487444608567, |
| "grad_norm": 0.5908671617507935, |
| "learning_rate": 6.885474208441603e-05, |
| "loss": 1.3802, |
| "step": 157 |
| }, |
| { |
| "epoch": 0.03734121122599705, |
| "grad_norm": 0.820754885673523, |
| "learning_rate": 6.848110061149556e-05, |
| "loss": 1.6758, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.03757754800590842, |
| "grad_norm": 0.684035062789917, |
| "learning_rate": 6.810625992644085e-05, |
| "loss": 1.3755, |
| "step": 159 |
| }, |
| { |
| "epoch": 0.03781388478581979, |
| "grad_norm": 0.6925457119941711, |
| "learning_rate": 6.773024435212678e-05, |
| "loss": 1.2663, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.038050221565731165, |
| "grad_norm": 0.6184958815574646, |
| "learning_rate": 6.735307828766515e-05, |
| "loss": 1.2938, |
| "step": 161 |
| }, |
| { |
| "epoch": 0.03828655834564254, |
| "grad_norm": 0.6940076351165771, |
| "learning_rate": 6.697478620682137e-05, |
| "loss": 1.4158, |
| "step": 162 |
| }, |
| { |
| "epoch": 0.038522895125553916, |
| "grad_norm": 0.8101726174354553, |
| "learning_rate": 6.659539265642643e-05, |
| "loss": 1.4058, |
| "step": 163 |
| }, |
| { |
| "epoch": 0.03875923190546529, |
| "grad_norm": 0.7321621179580688, |
| "learning_rate": 6.621492225478414e-05, |
| "loss": 1.3675, |
| "step": 164 |
| }, |
| { |
| "epoch": 0.03899556868537666, |
| "grad_norm": 0.661636233329773, |
| "learning_rate": 6.583339969007363e-05, |
| "loss": 1.4018, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.039231905465288035, |
| "grad_norm": 0.8307713866233826, |
| "learning_rate": 6.545084971874738e-05, |
| "loss": 1.2124, |
| "step": 166 |
| }, |
| { |
| "epoch": 0.03946824224519941, |
| "grad_norm": 0.6810014843940735, |
| "learning_rate": 6.506729716392481e-05, |
| "loss": 1.396, |
| "step": 167 |
| }, |
| { |
| "epoch": 0.039704579025110785, |
| "grad_norm": 1.1103947162628174, |
| "learning_rate": 6.468276691378155e-05, |
| "loss": 1.1694, |
| "step": 168 |
| }, |
| { |
| "epoch": 0.039940915805022154, |
| "grad_norm": 1.0087090730667114, |
| "learning_rate": 6.429728391993446e-05, |
| "loss": 1.1331, |
| "step": 169 |
| }, |
| { |
| "epoch": 0.04017725258493353, |
| "grad_norm": 0.7625798583030701, |
| "learning_rate": 6.391087319582264e-05, |
| "loss": 1.0443, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.040413589364844904, |
| "grad_norm": 0.9178735613822937, |
| "learning_rate": 6.35235598150842e-05, |
| "loss": 0.9701, |
| "step": 171 |
| }, |
| { |
| "epoch": 0.04064992614475628, |
| "grad_norm": 0.7063561081886292, |
| "learning_rate": 6.313536890992935e-05, |
| "loss": 1.2148, |
| "step": 172 |
| }, |
| { |
| "epoch": 0.040886262924667655, |
| "grad_norm": 0.865037202835083, |
| "learning_rate": 6.274632566950967e-05, |
| "loss": 1.1229, |
| "step": 173 |
| }, |
| { |
| "epoch": 0.04112259970457902, |
| "grad_norm": 0.8387466073036194, |
| "learning_rate": 6.235645533828349e-05, |
| "loss": 1.1997, |
| "step": 174 |
| }, |
| { |
| "epoch": 0.0413589364844904, |
| "grad_norm": 0.8075777888298035, |
| "learning_rate": 6.19657832143779e-05, |
| "loss": 1.4105, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.04159527326440177, |
| "grad_norm": 0.7226337194442749, |
| "learning_rate": 6.157433464794716e-05, |
| "loss": 1.2907, |
| "step": 176 |
| }, |
| { |
| "epoch": 0.04183161004431315, |
| "grad_norm": 0.8077397346496582, |
| "learning_rate": 6.118213503952779e-05, |
| "loss": 1.2005, |
| "step": 177 |
| }, |
| { |
| "epoch": 0.04206794682422452, |
| "grad_norm": 0.6361984014511108, |
| "learning_rate": 6.078920983839031e-05, |
| "loss": 1.4436, |
| "step": 178 |
| }, |
| { |
| "epoch": 0.04230428360413589, |
| "grad_norm": 0.7160913944244385, |
| "learning_rate": 6.0395584540887963e-05, |
| "loss": 1.3491, |
| "step": 179 |
| }, |
| { |
| "epoch": 0.04254062038404727, |
| "grad_norm": 0.7092442512512207, |
| "learning_rate": 6.0001284688802226e-05, |
| "loss": 1.4114, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.04277695716395864, |
| "grad_norm": 0.6893870830535889, |
| "learning_rate": 5.960633586768543e-05, |
| "loss": 1.2432, |
| "step": 181 |
| }, |
| { |
| "epoch": 0.04301329394387002, |
| "grad_norm": 0.701185405254364, |
| "learning_rate": 5.921076370520058e-05, |
| "loss": 1.555, |
| "step": 182 |
| }, |
| { |
| "epoch": 0.043249630723781386, |
| "grad_norm": 0.7154824733734131, |
| "learning_rate": 5.8814593869458455e-05, |
| "loss": 1.4316, |
| "step": 183 |
| }, |
| { |
| "epoch": 0.04348596750369276, |
| "grad_norm": 0.7668738961219788, |
| "learning_rate": 5.841785206735192e-05, |
| "loss": 1.3753, |
| "step": 184 |
| }, |
| { |
| "epoch": 0.04372230428360414, |
| "grad_norm": 0.6547796726226807, |
| "learning_rate": 5.8020564042888015e-05, |
| "loss": 1.309, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.04395864106351551, |
| "grad_norm": 0.7471556663513184, |
| "learning_rate": 5.762275557551727e-05, |
| "loss": 1.4151, |
| "step": 186 |
| }, |
| { |
| "epoch": 0.04419497784342688, |
| "grad_norm": 0.8444517254829407, |
| "learning_rate": 5.7224452478461064e-05, |
| "loss": 1.5483, |
| "step": 187 |
| }, |
| { |
| "epoch": 0.044431314623338256, |
| "grad_norm": 0.7207075357437134, |
| "learning_rate": 5.682568059703659e-05, |
| "loss": 1.3984, |
| "step": 188 |
| }, |
| { |
| "epoch": 0.04466765140324963, |
| "grad_norm": 0.7106743454933167, |
| "learning_rate": 5.642646580697973e-05, |
| "loss": 1.6566, |
| "step": 189 |
| }, |
| { |
| "epoch": 0.044903988183161006, |
| "grad_norm": 0.7666814923286438, |
| "learning_rate": 5.602683401276615e-05, |
| "loss": 1.5597, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.04514032496307238, |
| "grad_norm": 0.8037781715393066, |
| "learning_rate": 5.562681114593028e-05, |
| "loss": 1.2949, |
| "step": 191 |
| }, |
| { |
| "epoch": 0.04537666174298375, |
| "grad_norm": 0.8395581245422363, |
| "learning_rate": 5.522642316338268e-05, |
| "loss": 1.6436, |
| "step": 192 |
| }, |
| { |
| "epoch": 0.045612998522895125, |
| "grad_norm": 0.7195446491241455, |
| "learning_rate": 5.482569604572576e-05, |
| "loss": 1.5842, |
| "step": 193 |
| }, |
| { |
| "epoch": 0.0458493353028065, |
| "grad_norm": 0.7659729719161987, |
| "learning_rate": 5.442465579556793e-05, |
| "loss": 1.6308, |
| "step": 194 |
| }, |
| { |
| "epoch": 0.046085672082717875, |
| "grad_norm": 0.8340838551521301, |
| "learning_rate": 5.402332843583631e-05, |
| "loss": 1.6474, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.046322008862629244, |
| "grad_norm": 1.0423740148544312, |
| "learning_rate": 5.3621740008088126e-05, |
| "loss": 1.8423, |
| "step": 196 |
| }, |
| { |
| "epoch": 0.04655834564254062, |
| "grad_norm": 0.9392973780632019, |
| "learning_rate": 5.321991657082097e-05, |
| "loss": 1.716, |
| "step": 197 |
| }, |
| { |
| "epoch": 0.046794682422451994, |
| "grad_norm": 1.356037974357605, |
| "learning_rate": 5.281788419778187e-05, |
| "loss": 1.6955, |
| "step": 198 |
| }, |
| { |
| "epoch": 0.04703101920236337, |
| "grad_norm": 1.3172742128372192, |
| "learning_rate": 5.2415668976275355e-05, |
| "loss": 1.5057, |
| "step": 199 |
| }, |
| { |
| "epoch": 0.047267355982274745, |
| "grad_norm": 2.301046133041382, |
| "learning_rate": 5.201329700547076e-05, |
| "loss": 1.4726, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.047267355982274745, |
| "eval_loss": 1.3950382471084595, |
| "eval_runtime": 546.6233, |
| "eval_samples_per_second": 13.038, |
| "eval_steps_per_second": 3.26, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.04750369276218611, |
| "grad_norm": 0.5470572710037231, |
| "learning_rate": 5.161079439470866e-05, |
| "loss": 0.833, |
| "step": 201 |
| }, |
| { |
| "epoch": 0.04774002954209749, |
| "grad_norm": 0.5119298696517944, |
| "learning_rate": 5.1208187261806615e-05, |
| "loss": 1.2667, |
| "step": 202 |
| }, |
| { |
| "epoch": 0.047976366322008863, |
| "grad_norm": 0.5891231894493103, |
| "learning_rate": 5.080550173136457e-05, |
| "loss": 1.6066, |
| "step": 203 |
| }, |
| { |
| "epoch": 0.04821270310192024, |
| "grad_norm": 0.5852164626121521, |
| "learning_rate": 5.0402763933069496e-05, |
| "loss": 1.4913, |
| "step": 204 |
| }, |
| { |
| "epoch": 0.04844903988183161, |
| "grad_norm": 0.5890781283378601, |
| "learning_rate": 5e-05, |
| "loss": 1.5169, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.04868537666174298, |
| "grad_norm": 0.6997533440589905, |
| "learning_rate": 4.9597236066930516e-05, |
| "loss": 1.3982, |
| "step": 206 |
| }, |
| { |
| "epoch": 0.04892171344165436, |
| "grad_norm": 0.6835668087005615, |
| "learning_rate": 4.919449826863544e-05, |
| "loss": 1.4295, |
| "step": 207 |
| }, |
| { |
| "epoch": 0.04915805022156573, |
| "grad_norm": 0.6673969626426697, |
| "learning_rate": 4.87918127381934e-05, |
| "loss": 1.5904, |
| "step": 208 |
| }, |
| { |
| "epoch": 0.04939438700147711, |
| "grad_norm": 0.8849208950996399, |
| "learning_rate": 4.8389205605291365e-05, |
| "loss": 1.2838, |
| "step": 209 |
| }, |
| { |
| "epoch": 0.049630723781388476, |
| "grad_norm": 0.7788513898849487, |
| "learning_rate": 4.798670299452926e-05, |
| "loss": 1.0962, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.04986706056129985, |
| "grad_norm": 0.8474761247634888, |
| "learning_rate": 4.758433102372466e-05, |
| "loss": 1.3638, |
| "step": 211 |
| }, |
| { |
| "epoch": 0.05010339734121123, |
| "grad_norm": 0.6527062058448792, |
| "learning_rate": 4.7182115802218126e-05, |
| "loss": 1.3278, |
| "step": 212 |
| }, |
| { |
| "epoch": 0.0503397341211226, |
| "grad_norm": 0.9855383038520813, |
| "learning_rate": 4.678008342917903e-05, |
| "loss": 0.9405, |
| "step": 213 |
| }, |
| { |
| "epoch": 0.05057607090103397, |
| "grad_norm": 0.6667340993881226, |
| "learning_rate": 4.6378259991911886e-05, |
| "loss": 1.3972, |
| "step": 214 |
| }, |
| { |
| "epoch": 0.050812407680945346, |
| "grad_norm": 0.9174965620040894, |
| "learning_rate": 4.597667156416371e-05, |
| "loss": 1.3544, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.05104874446085672, |
| "grad_norm": 0.8398842215538025, |
| "learning_rate": 4.5575344204432084e-05, |
| "loss": 1.3297, |
| "step": 216 |
| }, |
| { |
| "epoch": 0.051285081240768096, |
| "grad_norm": 0.7070477604866028, |
| "learning_rate": 4.5174303954274244e-05, |
| "loss": 1.1856, |
| "step": 217 |
| }, |
| { |
| "epoch": 0.05152141802067947, |
| "grad_norm": 0.7112593650817871, |
| "learning_rate": 4.477357683661734e-05, |
| "loss": 1.1494, |
| "step": 218 |
| }, |
| { |
| "epoch": 0.05175775480059084, |
| "grad_norm": 0.6220516562461853, |
| "learning_rate": 4.437318885406973e-05, |
| "loss": 1.1729, |
| "step": 219 |
| }, |
| { |
| "epoch": 0.051994091580502215, |
| "grad_norm": 0.7269288301467896, |
| "learning_rate": 4.397316598723385e-05, |
| "loss": 1.2121, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.05223042836041359, |
| "grad_norm": 0.6351245641708374, |
| "learning_rate": 4.3573534193020274e-05, |
| "loss": 1.0508, |
| "step": 221 |
| }, |
| { |
| "epoch": 0.052466765140324965, |
| "grad_norm": 0.6852203011512756, |
| "learning_rate": 4.317431940296343e-05, |
| "loss": 1.6863, |
| "step": 222 |
| }, |
| { |
| "epoch": 0.052703101920236334, |
| "grad_norm": 0.8352505564689636, |
| "learning_rate": 4.277554752153895e-05, |
| "loss": 1.3719, |
| "step": 223 |
| }, |
| { |
| "epoch": 0.05293943870014771, |
| "grad_norm": 0.8485846519470215, |
| "learning_rate": 4.237724442448273e-05, |
| "loss": 1.0385, |
| "step": 224 |
| }, |
| { |
| "epoch": 0.053175775480059084, |
| "grad_norm": 0.771316647529602, |
| "learning_rate": 4.197943595711198e-05, |
| "loss": 1.1393, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.05341211225997046, |
| "grad_norm": 0.6911978721618652, |
| "learning_rate": 4.1582147932648074e-05, |
| "loss": 1.2856, |
| "step": 226 |
| }, |
| { |
| "epoch": 0.053648449039881835, |
| "grad_norm": 0.7352108955383301, |
| "learning_rate": 4.118540613054156e-05, |
| "loss": 0.9995, |
| "step": 227 |
| }, |
| { |
| "epoch": 0.0538847858197932, |
| "grad_norm": 0.6404459476470947, |
| "learning_rate": 4.078923629479943e-05, |
| "loss": 1.3882, |
| "step": 228 |
| }, |
| { |
| "epoch": 0.05412112259970458, |
| "grad_norm": 0.6022409200668335, |
| "learning_rate": 4.039366413231458e-05, |
| "loss": 1.3075, |
| "step": 229 |
| }, |
| { |
| "epoch": 0.054357459379615954, |
| "grad_norm": 0.713438868522644, |
| "learning_rate": 3.9998715311197785e-05, |
| "loss": 1.2688, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.05459379615952733, |
| "grad_norm": 0.699632465839386, |
| "learning_rate": 3.960441545911204e-05, |
| "loss": 1.418, |
| "step": 231 |
| }, |
| { |
| "epoch": 0.0548301329394387, |
| "grad_norm": 0.6836651563644409, |
| "learning_rate": 3.92107901616097e-05, |
| "loss": 1.3571, |
| "step": 232 |
| }, |
| { |
| "epoch": 0.05506646971935007, |
| "grad_norm": 0.672203779220581, |
| "learning_rate": 3.8817864960472236e-05, |
| "loss": 1.4165, |
| "step": 233 |
| }, |
| { |
| "epoch": 0.05530280649926145, |
| "grad_norm": 0.6799800992012024, |
| "learning_rate": 3.842566535205286e-05, |
| "loss": 1.2292, |
| "step": 234 |
| }, |
| { |
| "epoch": 0.05553914327917282, |
| "grad_norm": 0.677742600440979, |
| "learning_rate": 3.803421678562213e-05, |
| "loss": 1.3764, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.0557754800590842, |
| "grad_norm": 0.6180285215377808, |
| "learning_rate": 3.764354466171652e-05, |
| "loss": 1.3424, |
| "step": 236 |
| }, |
| { |
| "epoch": 0.056011816838995566, |
| "grad_norm": 0.6665123105049133, |
| "learning_rate": 3.725367433049033e-05, |
| "loss": 1.1586, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.05624815361890694, |
| "grad_norm": 0.7062758207321167, |
| "learning_rate": 3.6864631090070655e-05, |
| "loss": 1.2893, |
| "step": 238 |
| }, |
| { |
| "epoch": 0.05648449039881832, |
| "grad_norm": 0.707490861415863, |
| "learning_rate": 3.6476440184915815e-05, |
| "loss": 1.4726, |
| "step": 239 |
| }, |
| { |
| "epoch": 0.05672082717872969, |
| "grad_norm": 0.7802096605300903, |
| "learning_rate": 3.608912680417737e-05, |
| "loss": 1.6563, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.05695716395864106, |
| "grad_norm": 0.7431864738464355, |
| "learning_rate": 3.570271608006555e-05, |
| "loss": 1.6515, |
| "step": 241 |
| }, |
| { |
| "epoch": 0.057193500738552436, |
| "grad_norm": 0.8141834139823914, |
| "learning_rate": 3.531723308621847e-05, |
| "loss": 1.5237, |
| "step": 242 |
| }, |
| { |
| "epoch": 0.05742983751846381, |
| "grad_norm": 0.7511299848556519, |
| "learning_rate": 3.493270283607522e-05, |
| "loss": 1.5226, |
| "step": 243 |
| }, |
| { |
| "epoch": 0.057666174298375186, |
| "grad_norm": 0.7921032905578613, |
| "learning_rate": 3.4549150281252636e-05, |
| "loss": 1.6995, |
| "step": 244 |
| }, |
| { |
| "epoch": 0.05790251107828656, |
| "grad_norm": 0.7483624815940857, |
| "learning_rate": 3.4166600309926387e-05, |
| "loss": 1.5628, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.05813884785819793, |
| "grad_norm": 0.8765662312507629, |
| "learning_rate": 3.3785077745215873e-05, |
| "loss": 1.7283, |
| "step": 246 |
| }, |
| { |
| "epoch": 0.058375184638109305, |
| "grad_norm": 0.9644213318824768, |
| "learning_rate": 3.340460734357359e-05, |
| "loss": 1.7183, |
| "step": 247 |
| }, |
| { |
| "epoch": 0.05861152141802068, |
| "grad_norm": 1.148491621017456, |
| "learning_rate": 3.3025213793178646e-05, |
| "loss": 1.7021, |
| "step": 248 |
| }, |
| { |
| "epoch": 0.058847858197932056, |
| "grad_norm": 1.2919323444366455, |
| "learning_rate": 3.264692171233485e-05, |
| "loss": 1.5816, |
| "step": 249 |
| }, |
| { |
| "epoch": 0.059084194977843424, |
| "grad_norm": 1.6092400550842285, |
| "learning_rate": 3.226975564787322e-05, |
| "loss": 1.498, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.059084194977843424, |
| "eval_loss": 1.3724737167358398, |
| "eval_runtime": 546.408, |
| "eval_samples_per_second": 13.043, |
| "eval_steps_per_second": 3.261, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.0593205317577548, |
| "grad_norm": 0.5109966993331909, |
| "learning_rate": 3.189374007355917e-05, |
| "loss": 1.4638, |
| "step": 251 |
| }, |
| { |
| "epoch": 0.059556868537666174, |
| "grad_norm": 0.6506799459457397, |
| "learning_rate": 3.151889938850445e-05, |
| "loss": 1.0217, |
| "step": 252 |
| }, |
| { |
| "epoch": 0.05979320531757755, |
| "grad_norm": 0.515404999256134, |
| "learning_rate": 3.114525791558398e-05, |
| "loss": 1.2415, |
| "step": 253 |
| }, |
| { |
| "epoch": 0.060029542097488925, |
| "grad_norm": 0.4935321807861328, |
| "learning_rate": 3.0772839899857464e-05, |
| "loss": 1.6383, |
| "step": 254 |
| }, |
| { |
| "epoch": 0.06026587887740029, |
| "grad_norm": 0.4957164227962494, |
| "learning_rate": 3.0401669506996256e-05, |
| "loss": 1.6438, |
| "step": 255 |
| }, |
| { |
| "epoch": 0.06050221565731167, |
| "grad_norm": 0.5981006026268005, |
| "learning_rate": 3.003177082171523e-05, |
| "loss": 1.3765, |
| "step": 256 |
| }, |
| { |
| "epoch": 0.060738552437223044, |
| "grad_norm": 0.5776561498641968, |
| "learning_rate": 2.9663167846209998e-05, |
| "loss": 1.6361, |
| "step": 257 |
| }, |
| { |
| "epoch": 0.06097488921713442, |
| "grad_norm": 0.5799050331115723, |
| "learning_rate": 2.9295884498599414e-05, |
| "loss": 1.284, |
| "step": 258 |
| }, |
| { |
| "epoch": 0.06121122599704579, |
| "grad_norm": 0.5518852472305298, |
| "learning_rate": 2.8929944611373554e-05, |
| "loss": 1.5062, |
| "step": 259 |
| }, |
| { |
| "epoch": 0.06144756277695716, |
| "grad_norm": 0.6372984647750854, |
| "learning_rate": 2.8565371929847284e-05, |
| "loss": 1.4555, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.06168389955686854, |
| "grad_norm": 0.6874483227729797, |
| "learning_rate": 2.8202190110619493e-05, |
| "loss": 1.416, |
| "step": 261 |
| }, |
| { |
| "epoch": 0.06192023633677991, |
| "grad_norm": 0.7429825663566589, |
| "learning_rate": 2.784042272003794e-05, |
| "loss": 1.3298, |
| "step": 262 |
| }, |
| { |
| "epoch": 0.06215657311669129, |
| "grad_norm": 0.7798938751220703, |
| "learning_rate": 2.7480093232670158e-05, |
| "loss": 1.2861, |
| "step": 263 |
| }, |
| { |
| "epoch": 0.06239290989660266, |
| "grad_norm": 0.6920987367630005, |
| "learning_rate": 2.712122502978024e-05, |
| "loss": 1.3256, |
| "step": 264 |
| }, |
| { |
| "epoch": 0.06262924667651404, |
| "grad_norm": 0.8484948873519897, |
| "learning_rate": 2.6763841397811573e-05, |
| "loss": 1.1311, |
| "step": 265 |
| }, |
| { |
| "epoch": 0.0628655834564254, |
| "grad_norm": 0.7573161721229553, |
| "learning_rate": 2.64079655268759e-05, |
| "loss": 1.1854, |
| "step": 266 |
| }, |
| { |
| "epoch": 0.06310192023633678, |
| "grad_norm": 0.6577363610267639, |
| "learning_rate": 2.605362050924848e-05, |
| "loss": 1.1889, |
| "step": 267 |
| }, |
| { |
| "epoch": 0.06333825701624815, |
| "grad_norm": 0.7484895586967468, |
| "learning_rate": 2.57008293378697e-05, |
| "loss": 1.1099, |
| "step": 268 |
| }, |
| { |
| "epoch": 0.06357459379615953, |
| "grad_norm": 0.7348273992538452, |
| "learning_rate": 2.534961490485313e-05, |
| "loss": 1.3144, |
| "step": 269 |
| }, |
| { |
| "epoch": 0.0638109305760709, |
| "grad_norm": 0.6651597619056702, |
| "learning_rate": 2.500000000000001e-05, |
| "loss": 1.3081, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.06404726735598228, |
| "grad_norm": 0.9223187565803528, |
| "learning_rate": 2.4652007309320498e-05, |
| "loss": 1.0502, |
| "step": 271 |
| }, |
| { |
| "epoch": 0.06428360413589365, |
| "grad_norm": 0.701023280620575, |
| "learning_rate": 2.430565941356157e-05, |
| "loss": 1.0347, |
| "step": 272 |
| }, |
| { |
| "epoch": 0.06451994091580503, |
| "grad_norm": 0.754749596118927, |
| "learning_rate": 2.3960978786741877e-05, |
| "loss": 0.8906, |
| "step": 273 |
| }, |
| { |
| "epoch": 0.0647562776957164, |
| "grad_norm": 0.6555619835853577, |
| "learning_rate": 2.361798779469336e-05, |
| "loss": 1.2904, |
| "step": 274 |
| }, |
| { |
| "epoch": 0.06499261447562776, |
| "grad_norm": 0.679719090461731, |
| "learning_rate": 2.3276708693609943e-05, |
| "loss": 1.3425, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.06522895125553914, |
| "grad_norm": 0.7636399269104004, |
| "learning_rate": 2.2937163628603435e-05, |
| "loss": 1.5006, |
| "step": 276 |
| }, |
| { |
| "epoch": 0.06546528803545051, |
| "grad_norm": 0.8542468547821045, |
| "learning_rate": 2.259937463226651e-05, |
| "loss": 1.3895, |
| "step": 277 |
| }, |
| { |
| "epoch": 0.06570162481536189, |
| "grad_norm": 0.712738037109375, |
| "learning_rate": 2.2263363623243054e-05, |
| "loss": 1.3828, |
| "step": 278 |
| }, |
| { |
| "epoch": 0.06593796159527326, |
| "grad_norm": 0.7575670480728149, |
| "learning_rate": 2.192915240480596e-05, |
| "loss": 1.1332, |
| "step": 279 |
| }, |
| { |
| "epoch": 0.06617429837518464, |
| "grad_norm": 0.8040823936462402, |
| "learning_rate": 2.1596762663442218e-05, |
| "loss": 1.3518, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.06641063515509602, |
| "grad_norm": 0.7066543698310852, |
| "learning_rate": 2.1266215967445824e-05, |
| "loss": 1.4215, |
| "step": 281 |
| }, |
| { |
| "epoch": 0.06664697193500739, |
| "grad_norm": 0.6851208209991455, |
| "learning_rate": 2.0937533765518187e-05, |
| "loss": 1.3533, |
| "step": 282 |
| }, |
| { |
| "epoch": 0.06688330871491877, |
| "grad_norm": 0.7184422016143799, |
| "learning_rate": 2.061073738537635e-05, |
| "loss": 1.366, |
| "step": 283 |
| }, |
| { |
| "epoch": 0.06711964549483013, |
| "grad_norm": 0.7543905973434448, |
| "learning_rate": 2.0285848032369137e-05, |
| "loss": 1.2897, |
| "step": 284 |
| }, |
| { |
| "epoch": 0.0673559822747415, |
| "grad_norm": 0.7322332262992859, |
| "learning_rate": 1.996288678810105e-05, |
| "loss": 1.3138, |
| "step": 285 |
| }, |
| { |
| "epoch": 0.06759231905465288, |
| "grad_norm": 0.7079530954360962, |
| "learning_rate": 1.9641874609064443e-05, |
| "loss": 1.2877, |
| "step": 286 |
| }, |
| { |
| "epoch": 0.06782865583456425, |
| "grad_norm": 0.6339015960693359, |
| "learning_rate": 1.932283232527956e-05, |
| "loss": 1.4494, |
| "step": 287 |
| }, |
| { |
| "epoch": 0.06806499261447563, |
| "grad_norm": 0.6862327456474304, |
| "learning_rate": 1.9005780638942982e-05, |
| "loss": 1.3859, |
| "step": 288 |
| }, |
| { |
| "epoch": 0.068301329394387, |
| "grad_norm": 0.6487401723861694, |
| "learning_rate": 1.8690740123084316e-05, |
| "loss": 1.4005, |
| "step": 289 |
| }, |
| { |
| "epoch": 0.06853766617429838, |
| "grad_norm": 0.69569331407547, |
| "learning_rate": 1.837773122023114e-05, |
| "loss": 1.4787, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.06877400295420975, |
| "grad_norm": 0.7482574582099915, |
| "learning_rate": 1.8066774241082612e-05, |
| "loss": 1.6109, |
| "step": 291 |
| }, |
| { |
| "epoch": 0.06901033973412113, |
| "grad_norm": 0.7999292016029358, |
| "learning_rate": 1.7757889363191483e-05, |
| "loss": 1.3041, |
| "step": 292 |
| }, |
| { |
| "epoch": 0.06924667651403249, |
| "grad_norm": 0.8274068236351013, |
| "learning_rate": 1.745109662965481e-05, |
| "loss": 1.5867, |
| "step": 293 |
| }, |
| { |
| "epoch": 0.06948301329394387, |
| "grad_norm": 0.6892622709274292, |
| "learning_rate": 1.714641594781347e-05, |
| "loss": 1.662, |
| "step": 294 |
| }, |
| { |
| "epoch": 0.06971935007385524, |
| "grad_norm": 0.7591934204101562, |
| "learning_rate": 1.684386708796025e-05, |
| "loss": 1.6833, |
| "step": 295 |
| }, |
| { |
| "epoch": 0.06995568685376662, |
| "grad_norm": 0.8044524192810059, |
| "learning_rate": 1.6543469682057106e-05, |
| "loss": 1.707, |
| "step": 296 |
| }, |
| { |
| "epoch": 0.07019202363367799, |
| "grad_norm": 0.8446269035339355, |
| "learning_rate": 1.62452432224612e-05, |
| "loss": 1.7677, |
| "step": 297 |
| }, |
| { |
| "epoch": 0.07042836041358937, |
| "grad_norm": 0.9677338004112244, |
| "learning_rate": 1.5949207060660138e-05, |
| "loss": 1.6272, |
| "step": 298 |
| }, |
| { |
| "epoch": 0.07066469719350074, |
| "grad_norm": 1.271749496459961, |
| "learning_rate": 1.5655380406016235e-05, |
| "loss": 1.5857, |
| "step": 299 |
| }, |
| { |
| "epoch": 0.07090103397341212, |
| "grad_norm": 1.752280831336975, |
| "learning_rate": 1.536378232452003e-05, |
| "loss": 1.4372, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.07090103397341212, |
| "eval_loss": 1.3569902181625366, |
| "eval_runtime": 547.1469, |
| "eval_samples_per_second": 13.026, |
| "eval_steps_per_second": 3.257, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.07113737075332349, |
| "grad_norm": 0.5056349039077759, |
| "learning_rate": 1.5074431737553157e-05, |
| "loss": 0.909, |
| "step": 301 |
| }, |
| { |
| "epoch": 0.07137370753323485, |
| "grad_norm": 0.6225544214248657, |
| "learning_rate": 1.4787347420660541e-05, |
| "loss": 1.0927, |
| "step": 302 |
| }, |
| { |
| "epoch": 0.07161004431314623, |
| "grad_norm": 0.451829195022583, |
| "learning_rate": 1.4502548002332088e-05, |
| "loss": 1.2154, |
| "step": 303 |
| }, |
| { |
| "epoch": 0.0718463810930576, |
| "grad_norm": 0.46706512570381165, |
| "learning_rate": 1.422005196279395e-05, |
| "loss": 1.5182, |
| "step": 304 |
| }, |
| { |
| "epoch": 0.07208271787296898, |
| "grad_norm": 0.45082706212997437, |
| "learning_rate": 1.3939877632809278e-05, |
| "loss": 1.5319, |
| "step": 305 |
| }, |
| { |
| "epoch": 0.07231905465288035, |
| "grad_norm": 0.45447584986686707, |
| "learning_rate": 1.3662043192488849e-05, |
| "loss": 1.4201, |
| "step": 306 |
| }, |
| { |
| "epoch": 0.07255539143279173, |
| "grad_norm": 0.4832976758480072, |
| "learning_rate": 1.338656667011134e-05, |
| "loss": 1.1821, |
| "step": 307 |
| }, |
| { |
| "epoch": 0.0727917282127031, |
| "grad_norm": 0.5570776462554932, |
| "learning_rate": 1.3113465940953495e-05, |
| "loss": 1.4962, |
| "step": 308 |
| }, |
| { |
| "epoch": 0.07302806499261448, |
| "grad_norm": 0.6216040849685669, |
| "learning_rate": 1.2842758726130283e-05, |
| "loss": 1.4007, |
| "step": 309 |
| }, |
| { |
| "epoch": 0.07326440177252586, |
| "grad_norm": 0.6087969541549683, |
| "learning_rate": 1.257446259144494e-05, |
| "loss": 1.468, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.07350073855243722, |
| "grad_norm": 0.6260071396827698, |
| "learning_rate": 1.2308594946249163e-05, |
| "loss": 1.4255, |
| "step": 311 |
| }, |
| { |
| "epoch": 0.07373707533234859, |
| "grad_norm": 0.6277833580970764, |
| "learning_rate": 1.204517304231343e-05, |
| "loss": 1.3669, |
| "step": 312 |
| }, |
| { |
| "epoch": 0.07397341211225997, |
| "grad_norm": 0.6244453191757202, |
| "learning_rate": 1.178421397270758e-05, |
| "loss": 1.1709, |
| "step": 313 |
| }, |
| { |
| "epoch": 0.07420974889217134, |
| "grad_norm": 0.5879858136177063, |
| "learning_rate": 1.1525734670691701e-05, |
| "loss": 1.3037, |
| "step": 314 |
| }, |
| { |
| "epoch": 0.07444608567208272, |
| "grad_norm": 0.8364067077636719, |
| "learning_rate": 1.1269751908617277e-05, |
| "loss": 1.1557, |
| "step": 315 |
| }, |
| { |
| "epoch": 0.0746824224519941, |
| "grad_norm": 0.7973691821098328, |
| "learning_rate": 1.1016282296838887e-05, |
| "loss": 1.195, |
| "step": 316 |
| }, |
| { |
| "epoch": 0.07491875923190547, |
| "grad_norm": 0.6769034266471863, |
| "learning_rate": 1.0765342282636416e-05, |
| "loss": 1.4524, |
| "step": 317 |
| }, |
| { |
| "epoch": 0.07515509601181684, |
| "grad_norm": 0.6845118403434753, |
| "learning_rate": 1.0516948149147754e-05, |
| "loss": 1.1724, |
| "step": 318 |
| }, |
| { |
| "epoch": 0.07539143279172822, |
| "grad_norm": 0.6323385238647461, |
| "learning_rate": 1.0271116014312293e-05, |
| "loss": 1.3095, |
| "step": 319 |
| }, |
| { |
| "epoch": 0.07562776957163958, |
| "grad_norm": 0.686935544013977, |
| "learning_rate": 1.0027861829824952e-05, |
| "loss": 1.2477, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.07586410635155096, |
| "grad_norm": 0.7167879939079285, |
| "learning_rate": 9.787201380101157e-06, |
| "loss": 1.4439, |
| "step": 321 |
| }, |
| { |
| "epoch": 0.07610044313146233, |
| "grad_norm": 0.7184543609619141, |
| "learning_rate": 9.549150281252633e-06, |
| "loss": 1.4287, |
| "step": 322 |
| }, |
| { |
| "epoch": 0.0763367799113737, |
| "grad_norm": 0.7863267064094543, |
| "learning_rate": 9.313723980074018e-06, |
| "loss": 1.0322, |
| "step": 323 |
| }, |
| { |
| "epoch": 0.07657311669128508, |
| "grad_norm": 0.6955685615539551, |
| "learning_rate": 9.080937753040646e-06, |
| "loss": 0.968, |
| "step": 324 |
| }, |
| { |
| "epoch": 0.07680945347119646, |
| "grad_norm": 0.6205272674560547, |
| "learning_rate": 8.850806705317183e-06, |
| "loss": 1.346, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.07704579025110783, |
| "grad_norm": 0.6382667422294617, |
| "learning_rate": 8.623345769777514e-06, |
| "loss": 1.5221, |
| "step": 326 |
| }, |
| { |
| "epoch": 0.07728212703101921, |
| "grad_norm": 0.6864624619483948, |
| "learning_rate": 8.398569706035792e-06, |
| "loss": 1.4055, |
| "step": 327 |
| }, |
| { |
| "epoch": 0.07751846381093058, |
| "grad_norm": 0.9323960542678833, |
| "learning_rate": 8.176493099488663e-06, |
| "loss": 1.2108, |
| "step": 328 |
| }, |
| { |
| "epoch": 0.07775480059084194, |
| "grad_norm": 0.6769596934318542, |
| "learning_rate": 7.957130360368898e-06, |
| "loss": 1.3255, |
| "step": 329 |
| }, |
| { |
| "epoch": 0.07799113737075332, |
| "grad_norm": 0.7400510907173157, |
| "learning_rate": 7.740495722810271e-06, |
| "loss": 1.1148, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.0782274741506647, |
| "grad_norm": 0.650111734867096, |
| "learning_rate": 7.526603243923957e-06, |
| "loss": 1.2401, |
| "step": 331 |
| }, |
| { |
| "epoch": 0.07846381093057607, |
| "grad_norm": 0.6910921335220337, |
| "learning_rate": 7.315466802886401e-06, |
| "loss": 1.4438, |
| "step": 332 |
| }, |
| { |
| "epoch": 0.07870014771048744, |
| "grad_norm": 0.6273283958435059, |
| "learning_rate": 7.107100100038671e-06, |
| "loss": 1.3027, |
| "step": 333 |
| }, |
| { |
| "epoch": 0.07893648449039882, |
| "grad_norm": 0.7488106489181519, |
| "learning_rate": 6.901516655997536e-06, |
| "loss": 1.5159, |
| "step": 334 |
| }, |
| { |
| "epoch": 0.0791728212703102, |
| "grad_norm": 0.6988032460212708, |
| "learning_rate": 6.698729810778065e-06, |
| "loss": 1.3121, |
| "step": 335 |
| }, |
| { |
| "epoch": 0.07940915805022157, |
| "grad_norm": 0.8176699876785278, |
| "learning_rate": 6.498752722928042e-06, |
| "loss": 1.3366, |
| "step": 336 |
| }, |
| { |
| "epoch": 0.07964549483013295, |
| "grad_norm": 0.7137824892997742, |
| "learning_rate": 6.301598368674105e-06, |
| "loss": 1.4861, |
| "step": 337 |
| }, |
| { |
| "epoch": 0.07988183161004431, |
| "grad_norm": 0.8342055678367615, |
| "learning_rate": 6.107279541079769e-06, |
| "loss": 1.4292, |
| "step": 338 |
| }, |
| { |
| "epoch": 0.08011816838995568, |
| "grad_norm": 0.7024268507957458, |
| "learning_rate": 5.915808849215304e-06, |
| "loss": 1.4164, |
| "step": 339 |
| }, |
| { |
| "epoch": 0.08035450516986706, |
| "grad_norm": 0.7405518293380737, |
| "learning_rate": 5.727198717339511e-06, |
| "loss": 1.4451, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.08059084194977843, |
| "grad_norm": 0.8962257504463196, |
| "learning_rate": 5.54146138409355e-06, |
| "loss": 1.285, |
| "step": 341 |
| }, |
| { |
| "epoch": 0.08082717872968981, |
| "grad_norm": 0.7109900712966919, |
| "learning_rate": 5.358608901706802e-06, |
| "loss": 1.4931, |
| "step": 342 |
| }, |
| { |
| "epoch": 0.08106351550960118, |
| "grad_norm": 0.8523017168045044, |
| "learning_rate": 5.178653135214812e-06, |
| "loss": 1.4101, |
| "step": 343 |
| }, |
| { |
| "epoch": 0.08129985228951256, |
| "grad_norm": 0.6992009878158569, |
| "learning_rate": 5.001605761689398e-06, |
| "loss": 1.7812, |
| "step": 344 |
| }, |
| { |
| "epoch": 0.08153618906942393, |
| "grad_norm": 0.7525827288627625, |
| "learning_rate": 4.827478269480895e-06, |
| "loss": 1.6047, |
| "step": 345 |
| }, |
| { |
| "epoch": 0.08177252584933531, |
| "grad_norm": 0.7274166345596313, |
| "learning_rate": 4.65628195747273e-06, |
| "loss": 1.655, |
| "step": 346 |
| }, |
| { |
| "epoch": 0.08200886262924667, |
| "grad_norm": 0.8694523572921753, |
| "learning_rate": 4.488027934348271e-06, |
| "loss": 1.6647, |
| "step": 347 |
| }, |
| { |
| "epoch": 0.08224519940915805, |
| "grad_norm": 0.977834165096283, |
| "learning_rate": 4.322727117869951e-06, |
| "loss": 1.6009, |
| "step": 348 |
| }, |
| { |
| "epoch": 0.08248153618906942, |
| "grad_norm": 1.1253348588943481, |
| "learning_rate": 4.16039023417088e-06, |
| "loss": 1.6383, |
| "step": 349 |
| }, |
| { |
| "epoch": 0.0827178729689808, |
| "grad_norm": 1.2942062616348267, |
| "learning_rate": 4.001027817058789e-06, |
| "loss": 1.4725, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.0827178729689808, |
| "eval_loss": 1.3468055725097656, |
| "eval_runtime": 546.0894, |
| "eval_samples_per_second": 13.051, |
| "eval_steps_per_second": 3.263, |
| "step": 350 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 400, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 1, |
| "save_steps": 50, |
| "stateful_callbacks": { |
| "EarlyStoppingCallback": { |
| "args": { |
| "early_stopping_patience": 5, |
| "early_stopping_threshold": 0.0 |
| }, |
| "attributes": { |
| "early_stopping_patience_counter": 0 |
| } |
| }, |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 5.321281649026007e+17, |
| "train_batch_size": 8, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
