| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 9.27246790299572, |
| "eval_steps": 500, |
| "global_step": 6500, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.07132667617689016, |
| "grad_norm": 8.374612808227539, |
| "learning_rate": 2.9400000000000002e-06, |
| "loss": 1.4876, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.14265335235378032, |
| "grad_norm": 1.39237642288208, |
| "learning_rate": 2.9772093023255814e-06, |
| "loss": 1.2975, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.21398002853067047, |
| "grad_norm": 1.192126989364624, |
| "learning_rate": 2.953953488372093e-06, |
| "loss": 1.2197, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.28530670470756064, |
| "grad_norm": 1.450819730758667, |
| "learning_rate": 2.9306976744186045e-06, |
| "loss": 1.1853, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.3566333808844508, |
| "grad_norm": 1.5788187980651855, |
| "learning_rate": 2.9074418604651165e-06, |
| "loss": 1.1815, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.42796005706134094, |
| "grad_norm": 1.7196673154830933, |
| "learning_rate": 2.884186046511628e-06, |
| "loss": 1.1645, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.4992867332382311, |
| "grad_norm": 1.5300871133804321, |
| "learning_rate": 2.8609302325581396e-06, |
| "loss": 1.1449, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.5706134094151213, |
| "grad_norm": 1.5986438989639282, |
| "learning_rate": 2.837674418604651e-06, |
| "loss": 1.142, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.6419400855920114, |
| "grad_norm": 1.7197072505950928, |
| "learning_rate": 2.8144186046511628e-06, |
| "loss": 1.135, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.7132667617689016, |
| "grad_norm": 1.5155837535858154, |
| "learning_rate": 2.7911627906976743e-06, |
| "loss": 1.118, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.7845934379457917, |
| "grad_norm": 1.4595504999160767, |
| "learning_rate": 2.767906976744186e-06, |
| "loss": 1.1237, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.8559201141226819, |
| "grad_norm": 1.3575159311294556, |
| "learning_rate": 2.7446511627906975e-06, |
| "loss": 1.1367, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.927246790299572, |
| "grad_norm": 1.3612666130065918, |
| "learning_rate": 2.7213953488372094e-06, |
| "loss": 1.1206, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.9985734664764622, |
| "grad_norm": 1.3593813180923462, |
| "learning_rate": 2.698139534883721e-06, |
| "loss": 1.1385, |
| "step": 700 |
| }, |
| { |
| "epoch": 1.0699001426533523, |
| "grad_norm": 0.8913443088531494, |
| "learning_rate": 2.6748837209302326e-06, |
| "loss": 1.108, |
| "step": 750 |
| }, |
| { |
| "epoch": 1.1412268188302426, |
| "grad_norm": 0.9751181602478027, |
| "learning_rate": 2.651627906976744e-06, |
| "loss": 1.0941, |
| "step": 800 |
| }, |
| { |
| "epoch": 1.2125534950071326, |
| "grad_norm": 0.9726682901382446, |
| "learning_rate": 2.6283720930232557e-06, |
| "loss": 1.1093, |
| "step": 850 |
| }, |
| { |
| "epoch": 1.2838801711840229, |
| "grad_norm": 0.87285315990448, |
| "learning_rate": 2.6051162790697673e-06, |
| "loss": 1.135, |
| "step": 900 |
| }, |
| { |
| "epoch": 1.355206847360913, |
| "grad_norm": 0.8699735403060913, |
| "learning_rate": 2.581860465116279e-06, |
| "loss": 1.104, |
| "step": 950 |
| }, |
| { |
| "epoch": 1.4265335235378032, |
| "grad_norm": 1.0313692092895508, |
| "learning_rate": 2.558604651162791e-06, |
| "loss": 1.1197, |
| "step": 1000 |
| }, |
| { |
| "epoch": 1.4978601997146934, |
| "grad_norm": 0.9118053317070007, |
| "learning_rate": 2.5353488372093024e-06, |
| "loss": 1.096, |
| "step": 1050 |
| }, |
| { |
| "epoch": 1.5691868758915835, |
| "grad_norm": 0.9036098122596741, |
| "learning_rate": 2.512093023255814e-06, |
| "loss": 1.0938, |
| "step": 1100 |
| }, |
| { |
| "epoch": 1.6405135520684735, |
| "grad_norm": 1.0469361543655396, |
| "learning_rate": 2.4888372093023255e-06, |
| "loss": 1.093, |
| "step": 1150 |
| }, |
| { |
| "epoch": 1.7118402282453637, |
| "grad_norm": 0.9310563206672668, |
| "learning_rate": 2.4655813953488375e-06, |
| "loss": 1.1114, |
| "step": 1200 |
| }, |
| { |
| "epoch": 1.783166904422254, |
| "grad_norm": 0.8712385296821594, |
| "learning_rate": 2.442325581395349e-06, |
| "loss": 1.0992, |
| "step": 1250 |
| }, |
| { |
| "epoch": 1.854493580599144, |
| "grad_norm": 0.9832571148872375, |
| "learning_rate": 2.4190697674418606e-06, |
| "loss": 1.0896, |
| "step": 1300 |
| }, |
| { |
| "epoch": 1.925820256776034, |
| "grad_norm": 0.9631858468055725, |
| "learning_rate": 2.395813953488372e-06, |
| "loss": 1.0729, |
| "step": 1350 |
| }, |
| { |
| "epoch": 1.9971469329529246, |
| "grad_norm": 0.9085518717765808, |
| "learning_rate": 2.3725581395348838e-06, |
| "loss": 1.1032, |
| "step": 1400 |
| }, |
| { |
| "epoch": 2.0684736091298146, |
| "grad_norm": 0.7245323061943054, |
| "learning_rate": 2.3493023255813958e-06, |
| "loss": 1.1042, |
| "step": 1450 |
| }, |
| { |
| "epoch": 2.1398002853067046, |
| "grad_norm": 0.8104898929595947, |
| "learning_rate": 2.3260465116279073e-06, |
| "loss": 1.0829, |
| "step": 1500 |
| }, |
| { |
| "epoch": 2.2111269614835947, |
| "grad_norm": 0.7655696272850037, |
| "learning_rate": 2.302790697674419e-06, |
| "loss": 1.0701, |
| "step": 1550 |
| }, |
| { |
| "epoch": 2.282453637660485, |
| "grad_norm": 0.6674259305000305, |
| "learning_rate": 2.2795348837209304e-06, |
| "loss": 1.0831, |
| "step": 1600 |
| }, |
| { |
| "epoch": 2.353780313837375, |
| "grad_norm": 0.7704442739486694, |
| "learning_rate": 2.256279069767442e-06, |
| "loss": 1.0916, |
| "step": 1650 |
| }, |
| { |
| "epoch": 2.425106990014265, |
| "grad_norm": 0.9494684934616089, |
| "learning_rate": 2.2330232558139536e-06, |
| "loss": 1.0756, |
| "step": 1700 |
| }, |
| { |
| "epoch": 2.4964336661911553, |
| "grad_norm": 0.8056609630584717, |
| "learning_rate": 2.209767441860465e-06, |
| "loss": 1.0921, |
| "step": 1750 |
| }, |
| { |
| "epoch": 2.5677603423680457, |
| "grad_norm": 0.736752986907959, |
| "learning_rate": 2.1865116279069767e-06, |
| "loss": 1.1026, |
| "step": 1800 |
| }, |
| { |
| "epoch": 2.6390870185449358, |
| "grad_norm": 0.676659882068634, |
| "learning_rate": 2.1632558139534887e-06, |
| "loss": 1.127, |
| "step": 1850 |
| }, |
| { |
| "epoch": 2.710413694721826, |
| "grad_norm": 0.7117391228675842, |
| "learning_rate": 2.1400000000000003e-06, |
| "loss": 1.064, |
| "step": 1900 |
| }, |
| { |
| "epoch": 2.7817403708987163, |
| "grad_norm": 0.649662971496582, |
| "learning_rate": 2.116744186046512e-06, |
| "loss": 1.0887, |
| "step": 1950 |
| }, |
| { |
| "epoch": 2.8530670470756063, |
| "grad_norm": 0.7031757831573486, |
| "learning_rate": 2.0934883720930234e-06, |
| "loss": 1.084, |
| "step": 2000 |
| }, |
| { |
| "epoch": 2.9243937232524964, |
| "grad_norm": 0.8348050117492676, |
| "learning_rate": 2.070232558139535e-06, |
| "loss": 1.0606, |
| "step": 2050 |
| }, |
| { |
| "epoch": 2.995720399429387, |
| "grad_norm": 0.7338863015174866, |
| "learning_rate": 2.0469767441860465e-06, |
| "loss": 1.0748, |
| "step": 2100 |
| }, |
| { |
| "epoch": 3.067047075606277, |
| "grad_norm": 0.8033592700958252, |
| "learning_rate": 2.023720930232558e-06, |
| "loss": 1.0861, |
| "step": 2150 |
| }, |
| { |
| "epoch": 3.138373751783167, |
| "grad_norm": 0.7755851745605469, |
| "learning_rate": 2.0004651162790697e-06, |
| "loss": 1.0839, |
| "step": 2200 |
| }, |
| { |
| "epoch": 3.209700427960057, |
| "grad_norm": 0.7161579728126526, |
| "learning_rate": 1.9772093023255812e-06, |
| "loss": 1.0688, |
| "step": 2250 |
| }, |
| { |
| "epoch": 3.281027104136947, |
| "grad_norm": 0.854751706123352, |
| "learning_rate": 1.953953488372093e-06, |
| "loss": 1.0622, |
| "step": 2300 |
| }, |
| { |
| "epoch": 3.3523537803138375, |
| "grad_norm": 0.7753428220748901, |
| "learning_rate": 1.9306976744186048e-06, |
| "loss": 1.0677, |
| "step": 2350 |
| }, |
| { |
| "epoch": 3.4236804564907275, |
| "grad_norm": 0.7227015495300293, |
| "learning_rate": 1.9074418604651163e-06, |
| "loss": 1.0706, |
| "step": 2400 |
| }, |
| { |
| "epoch": 3.4950071326676175, |
| "grad_norm": 0.7708733677864075, |
| "learning_rate": 1.884186046511628e-06, |
| "loss": 1.0783, |
| "step": 2450 |
| }, |
| { |
| "epoch": 3.566333808844508, |
| "grad_norm": 0.7089836001396179, |
| "learning_rate": 1.8609302325581395e-06, |
| "loss": 1.0783, |
| "step": 2500 |
| }, |
| { |
| "epoch": 3.637660485021398, |
| "grad_norm": 0.7261233329772949, |
| "learning_rate": 1.837674418604651e-06, |
| "loss": 1.0997, |
| "step": 2550 |
| }, |
| { |
| "epoch": 3.708987161198288, |
| "grad_norm": 0.792870819568634, |
| "learning_rate": 1.8144186046511628e-06, |
| "loss": 1.0904, |
| "step": 2600 |
| }, |
| { |
| "epoch": 3.7803138373751786, |
| "grad_norm": 0.7912273406982422, |
| "learning_rate": 1.7911627906976744e-06, |
| "loss": 1.0609, |
| "step": 2650 |
| }, |
| { |
| "epoch": 3.8516405135520686, |
| "grad_norm": 0.8693161010742188, |
| "learning_rate": 1.7679069767441861e-06, |
| "loss": 1.0507, |
| "step": 2700 |
| }, |
| { |
| "epoch": 3.9229671897289586, |
| "grad_norm": 0.8193188905715942, |
| "learning_rate": 1.7446511627906977e-06, |
| "loss": 1.0802, |
| "step": 2750 |
| }, |
| { |
| "epoch": 3.9942938659058487, |
| "grad_norm": 0.8383765816688538, |
| "learning_rate": 1.7213953488372095e-06, |
| "loss": 1.0747, |
| "step": 2800 |
| }, |
| { |
| "epoch": 4.065620542082739, |
| "grad_norm": 0.6324388980865479, |
| "learning_rate": 1.698139534883721e-06, |
| "loss": 1.0643, |
| "step": 2850 |
| }, |
| { |
| "epoch": 4.136947218259629, |
| "grad_norm": 0.6263940930366516, |
| "learning_rate": 1.6748837209302326e-06, |
| "loss": 1.0707, |
| "step": 2900 |
| }, |
| { |
| "epoch": 4.20827389443652, |
| "grad_norm": 0.7516465187072754, |
| "learning_rate": 1.6516279069767442e-06, |
| "loss": 1.0723, |
| "step": 2950 |
| }, |
| { |
| "epoch": 4.279600570613409, |
| "grad_norm": 0.6298231482505798, |
| "learning_rate": 1.6283720930232558e-06, |
| "loss": 1.0902, |
| "step": 3000 |
| }, |
| { |
| "epoch": 4.3509272467903, |
| "grad_norm": 0.7128252387046814, |
| "learning_rate": 1.6051162790697673e-06, |
| "loss": 1.08, |
| "step": 3050 |
| }, |
| { |
| "epoch": 4.422253922967189, |
| "grad_norm": 0.8887324333190918, |
| "learning_rate": 1.5818604651162793e-06, |
| "loss": 1.064, |
| "step": 3100 |
| }, |
| { |
| "epoch": 4.49358059914408, |
| "grad_norm": 0.5942298769950867, |
| "learning_rate": 1.5586046511627909e-06, |
| "loss": 1.0748, |
| "step": 3150 |
| }, |
| { |
| "epoch": 4.56490727532097, |
| "grad_norm": 0.6519397497177124, |
| "learning_rate": 1.5353488372093024e-06, |
| "loss": 1.049, |
| "step": 3200 |
| }, |
| { |
| "epoch": 4.63623395149786, |
| "grad_norm": 0.8501943945884705, |
| "learning_rate": 1.512093023255814e-06, |
| "loss": 1.0689, |
| "step": 3250 |
| }, |
| { |
| "epoch": 4.70756062767475, |
| "grad_norm": 0.8405245542526245, |
| "learning_rate": 1.4888372093023256e-06, |
| "loss": 1.089, |
| "step": 3300 |
| }, |
| { |
| "epoch": 4.778887303851641, |
| "grad_norm": 0.8166652321815491, |
| "learning_rate": 1.4655813953488371e-06, |
| "loss": 1.0583, |
| "step": 3350 |
| }, |
| { |
| "epoch": 4.85021398002853, |
| "grad_norm": 0.6082524061203003, |
| "learning_rate": 1.442325581395349e-06, |
| "loss": 1.0721, |
| "step": 3400 |
| }, |
| { |
| "epoch": 4.921540656205421, |
| "grad_norm": 0.7854369282722473, |
| "learning_rate": 1.4190697674418605e-06, |
| "loss": 1.0536, |
| "step": 3450 |
| }, |
| { |
| "epoch": 4.9928673323823105, |
| "grad_norm": 0.662228524684906, |
| "learning_rate": 1.395813953488372e-06, |
| "loss": 1.0562, |
| "step": 3500 |
| }, |
| { |
| "epoch": 5.064194008559201, |
| "grad_norm": 0.5415034294128418, |
| "learning_rate": 1.3725581395348836e-06, |
| "loss": 1.0698, |
| "step": 3550 |
| }, |
| { |
| "epoch": 5.1355206847360915, |
| "grad_norm": 0.5144311785697937, |
| "learning_rate": 1.3493023255813954e-06, |
| "loss": 1.0739, |
| "step": 3600 |
| }, |
| { |
| "epoch": 5.206847360912981, |
| "grad_norm": 0.6178659796714783, |
| "learning_rate": 1.326046511627907e-06, |
| "loss": 1.0592, |
| "step": 3650 |
| }, |
| { |
| "epoch": 5.2781740370898715, |
| "grad_norm": 0.625819981098175, |
| "learning_rate": 1.3027906976744185e-06, |
| "loss": 1.0627, |
| "step": 3700 |
| }, |
| { |
| "epoch": 5.349500713266762, |
| "grad_norm": 0.605794370174408, |
| "learning_rate": 1.2795348837209303e-06, |
| "loss": 1.0903, |
| "step": 3750 |
| }, |
| { |
| "epoch": 5.420827389443652, |
| "grad_norm": 0.5412514209747314, |
| "learning_rate": 1.2562790697674418e-06, |
| "loss": 1.038, |
| "step": 3800 |
| }, |
| { |
| "epoch": 5.492154065620542, |
| "grad_norm": 0.5697770118713379, |
| "learning_rate": 1.2330232558139536e-06, |
| "loss": 1.0812, |
| "step": 3850 |
| }, |
| { |
| "epoch": 5.563480741797433, |
| "grad_norm": 0.6216000318527222, |
| "learning_rate": 1.2097674418604652e-06, |
| "loss": 1.0546, |
| "step": 3900 |
| }, |
| { |
| "epoch": 5.634807417974322, |
| "grad_norm": 0.6266453266143799, |
| "learning_rate": 1.1865116279069768e-06, |
| "loss": 1.0684, |
| "step": 3950 |
| }, |
| { |
| "epoch": 5.706134094151213, |
| "grad_norm": 0.5270595550537109, |
| "learning_rate": 1.1632558139534885e-06, |
| "loss": 1.0808, |
| "step": 4000 |
| }, |
| { |
| "epoch": 5.777460770328103, |
| "grad_norm": 0.5590298771858215, |
| "learning_rate": 1.14e-06, |
| "loss": 1.0685, |
| "step": 4050 |
| }, |
| { |
| "epoch": 5.848787446504993, |
| "grad_norm": 0.5806601643562317, |
| "learning_rate": 1.1167441860465117e-06, |
| "loss": 1.0524, |
| "step": 4100 |
| }, |
| { |
| "epoch": 5.920114122681883, |
| "grad_norm": 0.595583975315094, |
| "learning_rate": 1.0934883720930232e-06, |
| "loss": 1.0502, |
| "step": 4150 |
| }, |
| { |
| "epoch": 5.991440798858774, |
| "grad_norm": 0.5921752452850342, |
| "learning_rate": 1.070232558139535e-06, |
| "loss": 1.0579, |
| "step": 4200 |
| }, |
| { |
| "epoch": 6.062767475035663, |
| "grad_norm": 0.49118900299072266, |
| "learning_rate": 1.0469767441860466e-06, |
| "loss": 1.0414, |
| "step": 4250 |
| }, |
| { |
| "epoch": 6.134094151212554, |
| "grad_norm": 0.5080968141555786, |
| "learning_rate": 1.0237209302325581e-06, |
| "loss": 1.0669, |
| "step": 4300 |
| }, |
| { |
| "epoch": 6.205420827389443, |
| "grad_norm": 0.5190466642379761, |
| "learning_rate": 1.0004651162790697e-06, |
| "loss": 1.0815, |
| "step": 4350 |
| }, |
| { |
| "epoch": 6.276747503566334, |
| "grad_norm": 0.571312427520752, |
| "learning_rate": 9.772093023255815e-07, |
| "loss": 1.0586, |
| "step": 4400 |
| }, |
| { |
| "epoch": 6.348074179743224, |
| "grad_norm": 0.5268988013267517, |
| "learning_rate": 9.53953488372093e-07, |
| "loss": 1.0493, |
| "step": 4450 |
| }, |
| { |
| "epoch": 6.419400855920114, |
| "grad_norm": 0.5871033072471619, |
| "learning_rate": 9.306976744186046e-07, |
| "loss": 1.0639, |
| "step": 4500 |
| }, |
| { |
| "epoch": 6.490727532097004, |
| "grad_norm": 0.49360474944114685, |
| "learning_rate": 9.074418604651163e-07, |
| "loss": 1.0544, |
| "step": 4550 |
| }, |
| { |
| "epoch": 6.562054208273894, |
| "grad_norm": 0.5026892423629761, |
| "learning_rate": 8.841860465116279e-07, |
| "loss": 1.0788, |
| "step": 4600 |
| }, |
| { |
| "epoch": 6.633380884450784, |
| "grad_norm": 0.6054779291152954, |
| "learning_rate": 8.609302325581396e-07, |
| "loss": 1.0689, |
| "step": 4650 |
| }, |
| { |
| "epoch": 6.704707560627675, |
| "grad_norm": 0.5217434167861938, |
| "learning_rate": 8.376744186046512e-07, |
| "loss": 1.051, |
| "step": 4700 |
| }, |
| { |
| "epoch": 6.7760342368045645, |
| "grad_norm": 0.47993195056915283, |
| "learning_rate": 8.144186046511627e-07, |
| "loss": 1.0578, |
| "step": 4750 |
| }, |
| { |
| "epoch": 6.847360912981455, |
| "grad_norm": 0.5778261423110962, |
| "learning_rate": 7.911627906976745e-07, |
| "loss": 1.0706, |
| "step": 4800 |
| }, |
| { |
| "epoch": 6.9186875891583455, |
| "grad_norm": 0.5452645421028137, |
| "learning_rate": 7.679069767441861e-07, |
| "loss": 1.0663, |
| "step": 4850 |
| }, |
| { |
| "epoch": 6.990014265335235, |
| "grad_norm": 0.5666407346725464, |
| "learning_rate": 7.446511627906977e-07, |
| "loss": 1.0599, |
| "step": 4900 |
| }, |
| { |
| "epoch": 7.0613409415121255, |
| "grad_norm": 0.5317495465278625, |
| "learning_rate": 7.213953488372093e-07, |
| "loss": 1.0537, |
| "step": 4950 |
| }, |
| { |
| "epoch": 7.132667617689016, |
| "grad_norm": 0.5068672299385071, |
| "learning_rate": 6.981395348837209e-07, |
| "loss": 1.0779, |
| "step": 5000 |
| }, |
| { |
| "epoch": 7.203994293865906, |
| "grad_norm": 0.5887683629989624, |
| "learning_rate": 6.748837209302326e-07, |
| "loss": 1.057, |
| "step": 5050 |
| }, |
| { |
| "epoch": 7.275320970042796, |
| "grad_norm": 0.4751645028591156, |
| "learning_rate": 6.516279069767442e-07, |
| "loss": 1.0531, |
| "step": 5100 |
| }, |
| { |
| "epoch": 7.346647646219687, |
| "grad_norm": 0.4856860935688019, |
| "learning_rate": 6.283720930232559e-07, |
| "loss": 1.0575, |
| "step": 5150 |
| }, |
| { |
| "epoch": 7.417974322396576, |
| "grad_norm": 0.5266442894935608, |
| "learning_rate": 6.051162790697675e-07, |
| "loss": 1.0607, |
| "step": 5200 |
| }, |
| { |
| "epoch": 7.489300998573467, |
| "grad_norm": 0.5558798909187317, |
| "learning_rate": 5.818604651162791e-07, |
| "loss": 1.0653, |
| "step": 5250 |
| }, |
| { |
| "epoch": 7.560627674750357, |
| "grad_norm": 0.5203211307525635, |
| "learning_rate": 5.586046511627907e-07, |
| "loss": 1.0583, |
| "step": 5300 |
| }, |
| { |
| "epoch": 7.631954350927247, |
| "grad_norm": 0.4732428789138794, |
| "learning_rate": 5.353488372093024e-07, |
| "loss": 1.0534, |
| "step": 5350 |
| }, |
| { |
| "epoch": 7.703281027104137, |
| "grad_norm": 0.6876837015151978, |
| "learning_rate": 5.120930232558139e-07, |
| "loss": 1.0562, |
| "step": 5400 |
| }, |
| { |
| "epoch": 7.774607703281027, |
| "grad_norm": 0.5170673727989197, |
| "learning_rate": 4.888372093023256e-07, |
| "loss": 1.066, |
| "step": 5450 |
| }, |
| { |
| "epoch": 7.845934379457917, |
| "grad_norm": 0.5236818790435791, |
| "learning_rate": 4.6558139534883723e-07, |
| "loss": 1.0644, |
| "step": 5500 |
| }, |
| { |
| "epoch": 7.917261055634808, |
| "grad_norm": 0.47422999143600464, |
| "learning_rate": 4.423255813953489e-07, |
| "loss": 1.0594, |
| "step": 5550 |
| }, |
| { |
| "epoch": 7.988587731811697, |
| "grad_norm": 0.5537230968475342, |
| "learning_rate": 4.1906976744186046e-07, |
| "loss": 1.0578, |
| "step": 5600 |
| }, |
| { |
| "epoch": 8.059914407988588, |
| "grad_norm": 0.5376424789428711, |
| "learning_rate": 3.9581395348837213e-07, |
| "loss": 1.0523, |
| "step": 5650 |
| }, |
| { |
| "epoch": 8.131241084165477, |
| "grad_norm": 0.4635964334011078, |
| "learning_rate": 3.725581395348837e-07, |
| "loss": 1.0559, |
| "step": 5700 |
| }, |
| { |
| "epoch": 8.202567760342369, |
| "grad_norm": 0.46095508337020874, |
| "learning_rate": 3.493023255813953e-07, |
| "loss": 1.0825, |
| "step": 5750 |
| }, |
| { |
| "epoch": 8.273894436519258, |
| "grad_norm": 0.48418450355529785, |
| "learning_rate": 3.26046511627907e-07, |
| "loss": 1.0541, |
| "step": 5800 |
| }, |
| { |
| "epoch": 8.345221112696148, |
| "grad_norm": 0.45845749974250793, |
| "learning_rate": 3.027906976744186e-07, |
| "loss": 1.0731, |
| "step": 5850 |
| }, |
| { |
| "epoch": 8.41654778887304, |
| "grad_norm": 0.5616029500961304, |
| "learning_rate": 2.795348837209302e-07, |
| "loss": 1.055, |
| "step": 5900 |
| }, |
| { |
| "epoch": 8.487874465049929, |
| "grad_norm": 0.5239433646202087, |
| "learning_rate": 2.5627906976744184e-07, |
| "loss": 1.0522, |
| "step": 5950 |
| }, |
| { |
| "epoch": 8.559201141226819, |
| "grad_norm": 0.46216270327568054, |
| "learning_rate": 2.3302325581395349e-07, |
| "loss": 1.0674, |
| "step": 6000 |
| }, |
| { |
| "epoch": 8.63052781740371, |
| "grad_norm": 0.479060560464859, |
| "learning_rate": 2.097674418604651e-07, |
| "loss": 1.0417, |
| "step": 6050 |
| }, |
| { |
| "epoch": 8.7018544935806, |
| "grad_norm": 0.5477082133293152, |
| "learning_rate": 1.8651162790697675e-07, |
| "loss": 1.0656, |
| "step": 6100 |
| }, |
| { |
| "epoch": 8.773181169757489, |
| "grad_norm": 0.5336242914199829, |
| "learning_rate": 1.632558139534884e-07, |
| "loss": 1.0703, |
| "step": 6150 |
| }, |
| { |
| "epoch": 8.844507845934379, |
| "grad_norm": 0.5367247462272644, |
| "learning_rate": 1.4e-07, |
| "loss": 1.0451, |
| "step": 6200 |
| }, |
| { |
| "epoch": 8.91583452211127, |
| "grad_norm": 0.4902244806289673, |
| "learning_rate": 1.1674418604651164e-07, |
| "loss": 1.0561, |
| "step": 6250 |
| }, |
| { |
| "epoch": 8.98716119828816, |
| "grad_norm": 0.5425043106079102, |
| "learning_rate": 9.348837209302326e-08, |
| "loss": 1.0481, |
| "step": 6300 |
| }, |
| { |
| "epoch": 9.05848787446505, |
| "grad_norm": 0.5019351840019226, |
| "learning_rate": 7.023255813953488e-08, |
| "loss": 1.0592, |
| "step": 6350 |
| }, |
| { |
| "epoch": 9.12981455064194, |
| "grad_norm": 0.45389461517333984, |
| "learning_rate": 4.697674418604651e-08, |
| "loss": 1.0553, |
| "step": 6400 |
| }, |
| { |
| "epoch": 9.20114122681883, |
| "grad_norm": 0.4804913103580475, |
| "learning_rate": 2.372093023255814e-08, |
| "loss": 1.0812, |
| "step": 6450 |
| }, |
| { |
| "epoch": 9.27246790299572, |
| "grad_norm": 0.469777375459671, |
| "learning_rate": 4.651162790697675e-10, |
| "loss": 1.0647, |
| "step": 6500 |
| } |
| ], |
| "logging_steps": 50, |
| "max_steps": 6500, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 10, |
| "save_steps": 1000, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1.572171625708462e+18, |
| "train_batch_size": 48, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
