| { | |
| "best_metric": 1.3433974981307983, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-100", | |
| "epoch": 2.606060606060606, | |
| "eval_steps": 50, | |
| "global_step": 150, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.017316017316017316, | |
| "grad_norm": 0.3772251009941101, | |
| "learning_rate": 1e-05, | |
| "loss": 1.0925, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.017316017316017316, | |
| "eval_loss": 1.5536484718322754, | |
| "eval_runtime": 4.5189, | |
| "eval_samples_per_second": 21.465, | |
| "eval_steps_per_second": 5.532, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.03463203463203463, | |
| "grad_norm": 0.5300730466842651, | |
| "learning_rate": 2e-05, | |
| "loss": 1.3261, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.05194805194805195, | |
| "grad_norm": 0.552025556564331, | |
| "learning_rate": 3e-05, | |
| "loss": 1.3591, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.06926406926406926, | |
| "grad_norm": 0.6423289775848389, | |
| "learning_rate": 4e-05, | |
| "loss": 1.3797, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.08658008658008658, | |
| "grad_norm": 0.5506105422973633, | |
| "learning_rate": 5e-05, | |
| "loss": 1.4092, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.1038961038961039, | |
| "grad_norm": 0.5402395129203796, | |
| "learning_rate": 6e-05, | |
| "loss": 1.3346, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.12121212121212122, | |
| "grad_norm": 0.5220448970794678, | |
| "learning_rate": 7e-05, | |
| "loss": 1.3439, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.13852813852813853, | |
| "grad_norm": 0.5838501453399658, | |
| "learning_rate": 8e-05, | |
| "loss": 1.5076, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.15584415584415584, | |
| "grad_norm": 0.5467644333839417, | |
| "learning_rate": 9e-05, | |
| "loss": 1.3178, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.17316017316017315, | |
| "grad_norm": 0.6578097939491272, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4318, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.19047619047619047, | |
| "grad_norm": 0.7482398152351379, | |
| "learning_rate": 9.999071352056675e-05, | |
| "loss": 1.4785, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.2077922077922078, | |
| "grad_norm": 0.9860633015632629, | |
| "learning_rate": 9.9962857531815e-05, | |
| "loss": 1.5532, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.22510822510822512, | |
| "grad_norm": 1.0673537254333496, | |
| "learning_rate": 9.99164423811074e-05, | |
| "loss": 1.5532, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.24242424242424243, | |
| "grad_norm": 2.2215752601623535, | |
| "learning_rate": 9.985148530977767e-05, | |
| "loss": 1.7363, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.2597402597402597, | |
| "grad_norm": 0.5617103576660156, | |
| "learning_rate": 9.976801044672608e-05, | |
| "loss": 1.3188, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.27705627705627706, | |
| "grad_norm": 0.4882018566131592, | |
| "learning_rate": 9.966604879945659e-05, | |
| "loss": 1.2916, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.2943722943722944, | |
| "grad_norm": 0.46946704387664795, | |
| "learning_rate": 9.954563824255878e-05, | |
| "loss": 1.2448, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.3116883116883117, | |
| "grad_norm": 0.46930521726608276, | |
| "learning_rate": 9.940682350363912e-05, | |
| "loss": 1.293, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.329004329004329, | |
| "grad_norm": 0.45845523476600647, | |
| "learning_rate": 9.924965614670629e-05, | |
| "loss": 1.2965, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.3463203463203463, | |
| "grad_norm": 0.4072847366333008, | |
| "learning_rate": 9.907419455301741e-05, | |
| "loss": 1.2336, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.36363636363636365, | |
| "grad_norm": 0.4443903863430023, | |
| "learning_rate": 9.888050389939172e-05, | |
| "loss": 1.3432, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.38095238095238093, | |
| "grad_norm": 0.4643585681915283, | |
| "learning_rate": 9.866865613400008e-05, | |
| "loss": 1.2724, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.39826839826839827, | |
| "grad_norm": 0.4974319636821747, | |
| "learning_rate": 9.843872994963911e-05, | |
| "loss": 1.3752, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.4155844155844156, | |
| "grad_norm": 0.6184501051902771, | |
| "learning_rate": 9.819081075450014e-05, | |
| "loss": 1.6313, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.4329004329004329, | |
| "grad_norm": 0.72726970911026, | |
| "learning_rate": 9.792499064044342e-05, | |
| "loss": 1.4269, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.45021645021645024, | |
| "grad_norm": 0.7648056149482727, | |
| "learning_rate": 9.764136834878986e-05, | |
| "loss": 1.3242, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.4675324675324675, | |
| "grad_norm": 1.0451995134353638, | |
| "learning_rate": 9.734004923364257e-05, | |
| "loss": 1.6421, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.48484848484848486, | |
| "grad_norm": 1.6422125101089478, | |
| "learning_rate": 9.702114522275216e-05, | |
| "loss": 1.5802, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.5021645021645021, | |
| "grad_norm": 0.30660855770111084, | |
| "learning_rate": 9.66847747759402e-05, | |
| "loss": 1.2923, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.5194805194805194, | |
| "grad_norm": 0.33507832884788513, | |
| "learning_rate": 9.63310628410961e-05, | |
| "loss": 1.3251, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.5367965367965368, | |
| "grad_norm": 0.3787473142147064, | |
| "learning_rate": 9.596014080776423e-05, | |
| "loss": 1.1753, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.5541125541125541, | |
| "grad_norm": 0.3825725018978119, | |
| "learning_rate": 9.557214645833792e-05, | |
| "loss": 1.2146, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.5714285714285714, | |
| "grad_norm": 0.3974904417991638, | |
| "learning_rate": 9.516722391687902e-05, | |
| "loss": 1.2126, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.5887445887445888, | |
| "grad_norm": 0.4273703396320343, | |
| "learning_rate": 9.474552359558166e-05, | |
| "loss": 1.2017, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.6060606060606061, | |
| "grad_norm": 0.44417670369148254, | |
| "learning_rate": 9.43072021389003e-05, | |
| "loss": 1.3947, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.6233766233766234, | |
| "grad_norm": 0.4899435341358185, | |
| "learning_rate": 9.38524223653626e-05, | |
| "loss": 1.3964, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.6406926406926406, | |
| "grad_norm": 0.49582576751708984, | |
| "learning_rate": 9.338135320708911e-05, | |
| "loss": 1.201, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.658008658008658, | |
| "grad_norm": 0.618787944316864, | |
| "learning_rate": 9.289416964704185e-05, | |
| "loss": 1.3723, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.6753246753246753, | |
| "grad_norm": 0.6022177934646606, | |
| "learning_rate": 9.239105265402525e-05, | |
| "loss": 1.2341, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.6926406926406926, | |
| "grad_norm": 0.7559783458709717, | |
| "learning_rate": 9.187218911546362e-05, | |
| "loss": 1.5585, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.70995670995671, | |
| "grad_norm": 0.9588459730148315, | |
| "learning_rate": 9.133777176798013e-05, | |
| "loss": 1.535, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.7272727272727273, | |
| "grad_norm": 1.7360554933547974, | |
| "learning_rate": 9.078799912580304e-05, | |
| "loss": 1.668, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.7445887445887446, | |
| "grad_norm": 0.2811432182788849, | |
| "learning_rate": 9.022307540702576e-05, | |
| "loss": 1.1934, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.7619047619047619, | |
| "grad_norm": 0.31521645188331604, | |
| "learning_rate": 8.964321045774807e-05, | |
| "loss": 1.1976, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.7792207792207793, | |
| "grad_norm": 0.3455949127674103, | |
| "learning_rate": 8.904861967412703e-05, | |
| "loss": 1.316, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.7965367965367965, | |
| "grad_norm": 0.3525890111923218, | |
| "learning_rate": 8.843952392236594e-05, | |
| "loss": 1.2997, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.8138528138528138, | |
| "grad_norm": 0.37616753578186035, | |
| "learning_rate": 8.781614945667169e-05, | |
| "loss": 1.254, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.8311688311688312, | |
| "grad_norm": 0.4022103250026703, | |
| "learning_rate": 8.717872783521047e-05, | |
| "loss": 1.3993, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.8484848484848485, | |
| "grad_norm": 0.42644044756889343, | |
| "learning_rate": 8.65274958340934e-05, | |
| "loss": 1.245, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.8658008658008658, | |
| "grad_norm": 0.5010473132133484, | |
| "learning_rate": 8.586269535942385e-05, | |
| "loss": 1.2409, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.8658008658008658, | |
| "eval_loss": 1.3599679470062256, | |
| "eval_runtime": 4.6022, | |
| "eval_samples_per_second": 21.077, | |
| "eval_steps_per_second": 5.432, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.8831168831168831, | |
| "grad_norm": 0.5181626081466675, | |
| "learning_rate": 8.518457335743926e-05, | |
| "loss": 1.4489, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.9004329004329005, | |
| "grad_norm": 0.5752385854721069, | |
| "learning_rate": 8.449338172278059e-05, | |
| "loss": 1.5145, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.9177489177489178, | |
| "grad_norm": 0.65139240026474, | |
| "learning_rate": 8.378937720492384e-05, | |
| "loss": 1.4028, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.935064935064935, | |
| "grad_norm": 0.8399358987808228, | |
| "learning_rate": 8.307282131280804e-05, | |
| "loss": 1.5549, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.9523809523809523, | |
| "grad_norm": 0.9761013388633728, | |
| "learning_rate": 8.23439802176954e-05, | |
| "loss": 1.5483, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.9696969696969697, | |
| "grad_norm": 1.5134563446044922, | |
| "learning_rate": 8.160312465429952e-05, | |
| "loss": 1.4882, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.987012987012987, | |
| "grad_norm": 0.37874630093574524, | |
| "learning_rate": 8.085052982021847e-05, | |
| "loss": 1.2874, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 1.0086580086580086, | |
| "grad_norm": 0.8670079708099365, | |
| "learning_rate": 8.008647527371023e-05, | |
| "loss": 1.9227, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 1.025974025974026, | |
| "grad_norm": 0.28728023171424866, | |
| "learning_rate": 7.931124482984802e-05, | |
| "loss": 1.2465, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 1.0432900432900434, | |
| "grad_norm": 0.2963547110557556, | |
| "learning_rate": 7.85251264550948e-05, | |
| "loss": 1.1765, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.0606060606060606, | |
| "grad_norm": 0.3141789138317108, | |
| "learning_rate": 7.772841216033533e-05, | |
| "loss": 1.1228, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 1.077922077922078, | |
| "grad_norm": 0.32667580246925354, | |
| "learning_rate": 7.692139789240611e-05, | |
| "loss": 1.1514, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 1.0952380952380953, | |
| "grad_norm": 0.3543757200241089, | |
| "learning_rate": 7.610438342416319e-05, | |
| "loss": 1.0842, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 1.1125541125541125, | |
| "grad_norm": 0.367701917886734, | |
| "learning_rate": 7.527767224312883e-05, | |
| "loss": 1.1141, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 1.12987012987013, | |
| "grad_norm": 0.4170190989971161, | |
| "learning_rate": 7.44415714387582e-05, | |
| "loss": 1.3028, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 1.1471861471861473, | |
| "grad_norm": 0.44387149810791016, | |
| "learning_rate": 7.359639158836828e-05, | |
| "loss": 1.1789, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 1.1645021645021645, | |
| "grad_norm": 0.46756917238235474, | |
| "learning_rate": 7.274244664177097e-05, | |
| "loss": 1.2253, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 1.1818181818181819, | |
| "grad_norm": 0.5271068215370178, | |
| "learning_rate": 7.188005380465364e-05, | |
| "loss": 1.0931, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.199134199134199, | |
| "grad_norm": 0.5228838920593262, | |
| "learning_rate": 7.10095334207501e-05, | |
| "loss": 0.9346, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 1.2164502164502164, | |
| "grad_norm": 0.7063124775886536, | |
| "learning_rate": 7.013120885284598e-05, | |
| "loss": 1.1747, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.2337662337662338, | |
| "grad_norm": 1.0697877407073975, | |
| "learning_rate": 6.924540636266272e-05, | |
| "loss": 1.2499, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 1.251082251082251, | |
| "grad_norm": 0.7471345067024231, | |
| "learning_rate": 6.835245498966461e-05, | |
| "loss": 0.8715, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.2683982683982684, | |
| "grad_norm": 0.3865669071674347, | |
| "learning_rate": 6.745268642883404e-05, | |
| "loss": 1.2091, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 1.2857142857142856, | |
| "grad_norm": 0.3613302409648895, | |
| "learning_rate": 6.654643490746042e-05, | |
| "loss": 1.1159, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.303030303030303, | |
| "grad_norm": 0.38617178797721863, | |
| "learning_rate": 6.563403706098833e-05, | |
| "loss": 1.1253, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.3203463203463204, | |
| "grad_norm": 0.4121956527233124, | |
| "learning_rate": 6.471583180797121e-05, | |
| "loss": 1.1525, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.3376623376623376, | |
| "grad_norm": 0.42778268456459045, | |
| "learning_rate": 6.379216022417696e-05, | |
| "loss": 1.1418, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.354978354978355, | |
| "grad_norm": 0.47594478726387024, | |
| "learning_rate": 6.286336541589224e-05, | |
| "loss": 1.2353, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.3722943722943723, | |
| "grad_norm": 0.46286624670028687, | |
| "learning_rate": 6.192979239247243e-05, | |
| "loss": 1.1018, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.3896103896103895, | |
| "grad_norm": 0.4925539195537567, | |
| "learning_rate": 6.0991787938184784e-05, | |
| "loss": 1.0574, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.406926406926407, | |
| "grad_norm": 0.5687847137451172, | |
| "learning_rate": 6.004970048339226e-05, | |
| "loss": 1.2203, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.4242424242424243, | |
| "grad_norm": 0.6070426106452942, | |
| "learning_rate": 5.910387997512573e-05, | |
| "loss": 1.2845, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.4415584415584415, | |
| "grad_norm": 0.6566668152809143, | |
| "learning_rate": 5.8154677747093134e-05, | |
| "loss": 1.1189, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.4588744588744589, | |
| "grad_norm": 0.8260652422904968, | |
| "learning_rate": 5.7202446389173223e-05, | |
| "loss": 1.1718, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.4761904761904763, | |
| "grad_norm": 1.1566524505615234, | |
| "learning_rate": 5.624753961644281e-05, | |
| "loss": 0.8529, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.4935064935064934, | |
| "grad_norm": 1.3062056303024292, | |
| "learning_rate": 5.5290312137786146e-05, | |
| "loss": 0.9415, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.5108225108225108, | |
| "grad_norm": 0.34991952776908875, | |
| "learning_rate": 5.433111952413495e-05, | |
| "loss": 1.213, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.5281385281385282, | |
| "grad_norm": 0.3678584098815918, | |
| "learning_rate": 5.33703180763884e-05, | |
| "loss": 1.1127, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.5454545454545454, | |
| "grad_norm": 0.3730531334877014, | |
| "learning_rate": 5.240826469306187e-05, | |
| "loss": 1.1213, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.5627705627705628, | |
| "grad_norm": 0.385711669921875, | |
| "learning_rate": 5.144531673771363e-05, | |
| "loss": 1.0606, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.5800865800865802, | |
| "grad_norm": 0.4201517701148987, | |
| "learning_rate": 5.048183190619904e-05, | |
| "loss": 1.1155, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.5974025974025974, | |
| "grad_norm": 0.43522825837135315, | |
| "learning_rate": 4.951816809380097e-05, | |
| "loss": 1.1036, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.6147186147186146, | |
| "grad_norm": 0.46107161045074463, | |
| "learning_rate": 4.855468326228638e-05, | |
| "loss": 1.0929, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.6320346320346322, | |
| "grad_norm": 0.5166659355163574, | |
| "learning_rate": 4.759173530693814e-05, | |
| "loss": 1.1499, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.6493506493506493, | |
| "grad_norm": 0.5847046375274658, | |
| "learning_rate": 4.6629681923611603e-05, | |
| "loss": 1.2884, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.6666666666666665, | |
| "grad_norm": 0.6265720725059509, | |
| "learning_rate": 4.566888047586507e-05, | |
| "loss": 1.2245, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.6839826839826841, | |
| "grad_norm": 0.7204932570457458, | |
| "learning_rate": 4.4709687862213866e-05, | |
| "loss": 1.0221, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.7012987012987013, | |
| "grad_norm": 0.8253524899482727, | |
| "learning_rate": 4.3752460383557195e-05, | |
| "loss": 0.915, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.7186147186147185, | |
| "grad_norm": 1.2134804725646973, | |
| "learning_rate": 4.27975536108268e-05, | |
| "loss": 1.0184, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.7359307359307359, | |
| "grad_norm": 0.7511556148529053, | |
| "learning_rate": 4.1845322252906864e-05, | |
| "loss": 0.8647, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.7359307359307359, | |
| "eval_loss": 1.3433974981307983, | |
| "eval_runtime": 4.5988, | |
| "eval_samples_per_second": 21.092, | |
| "eval_steps_per_second": 5.436, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.7532467532467533, | |
| "grad_norm": 0.34484952688217163, | |
| "learning_rate": 4.0896120024874286e-05, | |
| "loss": 1.1188, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.7705627705627704, | |
| "grad_norm": 0.3626050055027008, | |
| "learning_rate": 3.9950299516607766e-05, | |
| "loss": 1.1234, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.7878787878787878, | |
| "grad_norm": 0.4060533940792084, | |
| "learning_rate": 3.900821206181521e-05, | |
| "loss": 1.1722, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.8051948051948052, | |
| "grad_norm": 0.4080195724964142, | |
| "learning_rate": 3.8070207607527584e-05, | |
| "loss": 0.9997, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.8225108225108224, | |
| "grad_norm": 0.4614916741847992, | |
| "learning_rate": 3.713663458410779e-05, | |
| "loss": 1.2573, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.8398268398268398, | |
| "grad_norm": 0.45603105425834656, | |
| "learning_rate": 3.620783977582305e-05, | |
| "loss": 1.108, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.8571428571428572, | |
| "grad_norm": 0.4706602990627289, | |
| "learning_rate": 3.528416819202881e-05, | |
| "loss": 1.0272, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.8744588744588744, | |
| "grad_norm": 0.5074113011360168, | |
| "learning_rate": 3.43659629390117e-05, | |
| "loss": 1.0969, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.8917748917748918, | |
| "grad_norm": 0.571370005607605, | |
| "learning_rate": 3.345356509253959e-05, | |
| "loss": 1.1148, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.9090909090909092, | |
| "grad_norm": 0.6326013803482056, | |
| "learning_rate": 3.254731357116597e-05, | |
| "loss": 1.0579, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.9264069264069263, | |
| "grad_norm": 0.7040982842445374, | |
| "learning_rate": 3.16475450103354e-05, | |
| "loss": 1.0999, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.9437229437229437, | |
| "grad_norm": 0.7916378378868103, | |
| "learning_rate": 3.0754593637337276e-05, | |
| "loss": 1.0317, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.9610389610389611, | |
| "grad_norm": 1.2524980306625366, | |
| "learning_rate": 2.986879114715403e-05, | |
| "loss": 1.208, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.9783549783549783, | |
| "grad_norm": 0.991152822971344, | |
| "learning_rate": 2.899046657924992e-05, | |
| "loss": 1.0673, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.9956709956709957, | |
| "grad_norm": 1.3673672676086426, | |
| "learning_rate": 2.8119946195346375e-05, | |
| "loss": 1.5977, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 2.017316017316017, | |
| "grad_norm": 0.32585984468460083, | |
| "learning_rate": 2.7257553358229034e-05, | |
| "loss": 1.1775, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 2.034632034632035, | |
| "grad_norm": 0.3496079444885254, | |
| "learning_rate": 2.6403608411631742e-05, | |
| "loss": 1.2228, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 2.051948051948052, | |
| "grad_norm": 0.3493185341358185, | |
| "learning_rate": 2.555842856124182e-05, | |
| "loss": 0.9992, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 2.069264069264069, | |
| "grad_norm": 0.38740429282188416, | |
| "learning_rate": 2.472232775687119e-05, | |
| "loss": 1.0707, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 2.0865800865800868, | |
| "grad_norm": 0.42126643657684326, | |
| "learning_rate": 2.389561657583681e-05, | |
| "loss": 1.0792, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.103896103896104, | |
| "grad_norm": 0.4259890615940094, | |
| "learning_rate": 2.30786021075939e-05, | |
| "loss": 1.0755, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 2.121212121212121, | |
| "grad_norm": 0.4546581506729126, | |
| "learning_rate": 2.2271587839664672e-05, | |
| "loss": 1.0404, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 2.1385281385281387, | |
| "grad_norm": 0.4937589466571808, | |
| "learning_rate": 2.1474873544905205e-05, | |
| "loss": 1.0073, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 2.155844155844156, | |
| "grad_norm": 0.5410916805267334, | |
| "learning_rate": 2.0688755170151996e-05, | |
| "loss": 1.1053, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 2.173160173160173, | |
| "grad_norm": 0.5611327290534973, | |
| "learning_rate": 1.9913524726289784e-05, | |
| "loss": 0.9988, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 2.1904761904761907, | |
| "grad_norm": 0.6241786479949951, | |
| "learning_rate": 1.914947017978153e-05, | |
| "loss": 0.9437, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 2.207792207792208, | |
| "grad_norm": 0.6713927984237671, | |
| "learning_rate": 1.8396875345700497e-05, | |
| "loss": 0.8584, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 2.225108225108225, | |
| "grad_norm": 0.9032663702964783, | |
| "learning_rate": 1.76560197823046e-05, | |
| "loss": 0.7647, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 2.242424242424242, | |
| "grad_norm": 1.1723631620407104, | |
| "learning_rate": 1.692717868719195e-05, | |
| "loss": 0.6741, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 2.25974025974026, | |
| "grad_norm": 0.3166380226612091, | |
| "learning_rate": 1.621062279507617e-05, | |
| "loss": 1.0231, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 2.277056277056277, | |
| "grad_norm": 0.36407145857810974, | |
| "learning_rate": 1.550661827721941e-05, | |
| "loss": 1.0707, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 2.2943722943722946, | |
| "grad_norm": 0.39266106486320496, | |
| "learning_rate": 1.4815426642560754e-05, | |
| "loss": 1.0085, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 2.311688311688312, | |
| "grad_norm": 0.4319557845592499, | |
| "learning_rate": 1.413730464057616e-05, | |
| "loss": 1.0623, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 2.329004329004329, | |
| "grad_norm": 0.46388888359069824, | |
| "learning_rate": 1.3472504165906613e-05, | |
| "loss": 1.098, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 2.346320346320346, | |
| "grad_norm": 0.46383485198020935, | |
| "learning_rate": 1.2821272164789544e-05, | |
| "loss": 1.0293, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 2.3636363636363638, | |
| "grad_norm": 0.49216514825820923, | |
| "learning_rate": 1.2183850543328312e-05, | |
| "loss": 0.9614, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 2.380952380952381, | |
| "grad_norm": 0.5347785949707031, | |
| "learning_rate": 1.156047607763407e-05, | |
| "loss": 1.0165, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 2.398268398268398, | |
| "grad_norm": 0.576884388923645, | |
| "learning_rate": 1.0951380325872979e-05, | |
| "loss": 0.9611, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 2.4155844155844157, | |
| "grad_norm": 0.6156734824180603, | |
| "learning_rate": 1.0356789542251938e-05, | |
| "loss": 0.9602, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 2.432900432900433, | |
| "grad_norm": 0.7232640981674194, | |
| "learning_rate": 9.776924592974256e-06, | |
| "loss": 0.9805, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.45021645021645, | |
| "grad_norm": 0.7894703149795532, | |
| "learning_rate": 9.212000874196953e-06, | |
| "loss": 0.8696, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 2.4675324675324677, | |
| "grad_norm": 0.9049820899963379, | |
| "learning_rate": 8.662228232019876e-06, | |
| "loss": 0.6752, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 2.484848484848485, | |
| "grad_norm": 1.1644245386123657, | |
| "learning_rate": 8.127810884536403e-06, | |
| "loss": 0.5053, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 2.502164502164502, | |
| "grad_norm": 0.36669421195983887, | |
| "learning_rate": 7.60894734597476e-06, | |
| "loss": 1.1323, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 2.5194805194805197, | |
| "grad_norm": 0.4047780930995941, | |
| "learning_rate": 7.105830352958142e-06, | |
| "loss": 1.1258, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 2.536796536796537, | |
| "grad_norm": 0.42274975776672363, | |
| "learning_rate": 6.618646792910893e-06, | |
| "loss": 1.0484, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 2.554112554112554, | |
| "grad_norm": 0.42739108204841614, | |
| "learning_rate": 6.147577634637414e-06, | |
| "loss": 0.9409, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 2.571428571428571, | |
| "grad_norm": 0.4765530228614807, | |
| "learning_rate": 5.692797861099719e-06, | |
| "loss": 1.0724, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 2.588744588744589, | |
| "grad_norm": 0.49571213126182556, | |
| "learning_rate": 5.25447640441834e-06, | |
| "loss": 0.9421, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 2.606060606060606, | |
| "grad_norm": 0.5144370794296265, | |
| "learning_rate": 4.832776083120982e-06, | |
| "loss": 1.0029, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.606060606060606, | |
| "eval_loss": 1.354403018951416, | |
| "eval_runtime": 4.6289, | |
| "eval_samples_per_second": 20.955, | |
| "eval_steps_per_second": 5.401, | |
| "step": 150 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 173, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 1 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 8.886296965545984e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |