| {"loss": 1.1136, "learning_rate": 0.0001, "epoch": 0.0, "step": 100} | |
| {"loss": 0.9595, "learning_rate": 9.997999599919985e-05, "epoch": 0.0, "step": 200} | |
| {"loss": 0.8936, "learning_rate": 9.995999199839968e-05, "epoch": 0.01, "step": 300} | |
| {"loss": 0.9237, "learning_rate": 9.993998799759952e-05, "epoch": 0.01, "step": 400} | |
| {"loss": 0.9005, "learning_rate": 9.992038407681536e-05, "epoch": 0.01, "step": 500} | |
| {"loss": 0.8433, "learning_rate": 9.99003800760152e-05, "epoch": 0.01, "step": 600} | |
| {"loss": 0.863, "learning_rate": 9.988037607521505e-05, "epoch": 0.01, "step": 700} | |
| {"loss": 0.8634, "learning_rate": 9.98603720744149e-05, "epoch": 0.02, "step": 800} | |
| {"loss": 0.8463, "learning_rate": 9.984036807361474e-05, "epoch": 0.02, "step": 900} | |
| {"loss": 0.8527, "learning_rate": 9.982036407281457e-05, "epoch": 0.02, "step": 1000} | |
| {"loss": 0.8259, "learning_rate": 9.980036007201441e-05, "epoch": 0.02, "step": 1100} | |
| {"loss": 0.8473, "learning_rate": 9.978035607121424e-05, "epoch": 0.02, "step": 1200} | |
| {"loss": 0.8462, "learning_rate": 9.976035207041409e-05, "epoch": 0.03, "step": 1300} | |
| {"loss": 0.8609, "learning_rate": 9.974034806961393e-05, "epoch": 0.03, "step": 1400} | |
| {"loss": 0.8238, "learning_rate": 9.972034406881376e-05, "epoch": 0.03, "step": 1500} | |
| {"loss": 0.8212, "learning_rate": 9.970034006801361e-05, "epoch": 0.03, "step": 1600} | |
| {"loss": 0.8391, "learning_rate": 9.968033606721344e-05, "epoch": 0.03, "step": 1700} | |
| {"loss": 0.774, "learning_rate": 9.966033206641328e-05, "epoch": 0.04, "step": 1800} | |
| {"loss": 0.8149, "learning_rate": 9.964032806561313e-05, "epoch": 0.04, "step": 1900} | |
| {"loss": 0.8794, "learning_rate": 9.962032406481296e-05, "epoch": 0.04, "step": 2000} | |
| {"loss": 0.7747, "learning_rate": 9.96003200640128e-05, "epoch": 0.04, "step": 2100} | |
| {"loss": 0.8191, "learning_rate": 9.958031606321264e-05, "epoch": 0.04, "step": 2200} | |
| {"loss": 0.8219, "learning_rate": 9.956031206241248e-05, "epoch": 0.05, "step": 2300} | |
| {"loss": 0.8191, "learning_rate": 9.954030806161232e-05, "epoch": 0.05, "step": 2400} | |
| {"loss": 0.8662, "learning_rate": 9.952030406081217e-05, "epoch": 0.05, "step": 2500} | |
| {"loss": 0.787, "learning_rate": 9.950030006001201e-05, "epoch": 0.05, "step": 2600} | |
| {"loss": 0.8668, "learning_rate": 9.948029605921184e-05, "epoch": 0.05, "step": 2700} | |
| {"loss": 0.7817, "learning_rate": 9.946029205841169e-05, "epoch": 0.06, "step": 2800} | |
| {"loss": 0.8021, "learning_rate": 9.944028805761153e-05, "epoch": 0.06, "step": 2900} | |
| {"loss": 0.8324, "learning_rate": 9.942028405681136e-05, "epoch": 0.06, "step": 3000} | |
| {"loss": 0.8631, "learning_rate": 9.940028005601121e-05, "epoch": 0.06, "step": 3100} | |
| {"loss": 0.797, "learning_rate": 9.938027605521105e-05, "epoch": 0.06, "step": 3200} | |
| {"loss": 0.8224, "learning_rate": 9.936027205441088e-05, "epoch": 0.07, "step": 3300} | |
| {"loss": 0.8164, "learning_rate": 9.934026805361073e-05, "epoch": 0.07, "step": 3400} | |
| {"loss": 0.7975, "learning_rate": 9.932026405281056e-05, "epoch": 0.07, "step": 3500} | |
| {"loss": 0.8468, "learning_rate": 9.930046009201841e-05, "epoch": 0.07, "step": 3600} | |
| {"loss": 0.8169, "learning_rate": 9.928045609121824e-05, "epoch": 0.07, "step": 3700} | |
| {"loss": 0.7761, "learning_rate": 9.926045209041809e-05, "epoch": 0.08, "step": 3800} | |
| {"loss": 0.8042, "learning_rate": 9.924044808961793e-05, "epoch": 0.08, "step": 3900} | |
| {"loss": 0.8146, "learning_rate": 9.922044408881776e-05, "epoch": 0.08, "step": 4000} | |
| {"loss": 0.8243, "learning_rate": 9.920044008801761e-05, "epoch": 0.08, "step": 4100} | |
| {"loss": 0.8315, "learning_rate": 9.918043608721744e-05, "epoch": 0.08, "step": 4200} | |
| {"loss": 0.8014, "learning_rate": 9.916043208641728e-05, "epoch": 0.09, "step": 4300} | |
| {"loss": 0.8241, "learning_rate": 9.914042808561713e-05, "epoch": 0.09, "step": 4400} | |
| {"loss": 0.8233, "learning_rate": 9.912042408481696e-05, "epoch": 0.09, "step": 4500} | |
| {"loss": 0.7748, "learning_rate": 9.91004200840168e-05, "epoch": 0.09, "step": 4600} | |
| {"loss": 0.8197, "learning_rate": 9.908041608321665e-05, "epoch": 0.09, "step": 4700} | |
| {"loss": 0.8195, "learning_rate": 9.906041208241649e-05, "epoch": 0.1, "step": 4800} | |
| {"loss": 0.7864, "learning_rate": 9.904040808161634e-05, "epoch": 0.1, "step": 4900} | |
| {"loss": 0.7812, "learning_rate": 9.902040408081617e-05, "epoch": 0.1, "step": 5000} | |
| {"loss": 0.7966, "learning_rate": 9.900040008001601e-05, "epoch": 0.1, "step": 5100} | |
| {"loss": 0.793, "learning_rate": 9.898039607921584e-05, "epoch": 0.1, "step": 5200} | |
| {"loss": 0.8568, "learning_rate": 9.896039207841569e-05, "epoch": 0.11, "step": 5300} | |
| {"loss": 0.7614, "learning_rate": 9.894038807761553e-05, "epoch": 0.11, "step": 5400} | |
| {"loss": 0.8171, "learning_rate": 9.892038407681536e-05, "epoch": 0.11, "step": 5500} | |
| {"loss": 0.7649, "learning_rate": 9.890038007601521e-05, "epoch": 0.11, "step": 5600} | |
| {"loss": 0.8074, "learning_rate": 9.888037607521505e-05, "epoch": 0.11, "step": 5700} | |
| {"loss": 0.7591, "learning_rate": 9.886037207441488e-05, "epoch": 0.12, "step": 5800} | |
| {"loss": 0.8102, "learning_rate": 9.884036807361473e-05, "epoch": 0.12, "step": 5900} | |
| {"loss": 0.784, "learning_rate": 9.882036407281456e-05, "epoch": 0.12, "step": 6000} | |
| {"loss": 0.8367, "learning_rate": 9.88003600720144e-05, "epoch": 0.12, "step": 6100} | |
| {"loss": 0.8418, "learning_rate": 9.878035607121425e-05, "epoch": 0.12, "step": 6200} | |
| {"loss": 0.798, "learning_rate": 9.876035207041409e-05, "epoch": 0.13, "step": 6300} | |
| {"loss": 0.7928, "learning_rate": 9.874034806961394e-05, "epoch": 0.13, "step": 6400} | |
| {"loss": 0.8262, "learning_rate": 9.872034406881377e-05, "epoch": 0.13, "step": 6500} | |
| {"loss": 0.7965, "learning_rate": 9.870034006801361e-05, "epoch": 0.13, "step": 6600} | |
| {"loss": 0.7952, "learning_rate": 9.868033606721346e-05, "epoch": 0.13, "step": 6700} | |
| {"loss": 0.7987, "learning_rate": 9.866033206641329e-05, "epoch": 0.14, "step": 6800} | |
| {"loss": 0.7939, "learning_rate": 9.864032806561313e-05, "epoch": 0.14, "step": 6900} | |
| {"loss": 0.7484, "learning_rate": 9.862032406481296e-05, "epoch": 0.14, "step": 7000} | |
| {"loss": 0.819, "learning_rate": 9.860032006401281e-05, "epoch": 0.14, "step": 7100} | |
| {"loss": 0.747, "learning_rate": 9.858031606321265e-05, "epoch": 0.14, "step": 7200} | |
| {"loss": 0.8029, "learning_rate": 9.856031206241248e-05, "epoch": 0.15, "step": 7300} | |
| {"loss": 0.9013, "learning_rate": 9.854030806161233e-05, "epoch": 0.15, "step": 7400} | |
| {"loss": 0.8337, "learning_rate": 9.852030406081216e-05, "epoch": 0.15, "step": 7500} | |
| {"loss": 0.7927, "learning_rate": 9.8500300060012e-05, "epoch": 0.15, "step": 7600} | |
| {"loss": 0.8725, "learning_rate": 9.848029605921185e-05, "epoch": 0.15, "step": 7700} | |
| {"loss": 0.7865, "learning_rate": 9.846029205841169e-05, "epoch": 0.16, "step": 7800} | |
| {"loss": 0.7726, "learning_rate": 9.844028805761154e-05, "epoch": 0.16, "step": 7900} | |
| {"loss": 0.7868, "learning_rate": 9.842048409681936e-05, "epoch": 0.16, "step": 8000} | |
| {"loss": 0.7918, "learning_rate": 9.840048009601921e-05, "epoch": 0.16, "step": 8100} | |
| {"loss": 0.7924, "learning_rate": 9.838047609521905e-05, "epoch": 0.16, "step": 8200} | |
| {"loss": 0.8036, "learning_rate": 9.836047209441888e-05, "epoch": 0.17, "step": 8300} | |
| {"loss": 0.7844, "learning_rate": 9.834046809361873e-05, "epoch": 0.17, "step": 8400} | |
| {"loss": 0.8124, "learning_rate": 9.832046409281857e-05, "epoch": 0.17, "step": 8500} | |
| {"loss": 0.837, "learning_rate": 9.830046009201842e-05, "epoch": 0.17, "step": 8600} | |
| {"loss": 0.8315, "learning_rate": 9.828045609121826e-05, "epoch": 0.17, "step": 8700} | |
| {"loss": 0.7738, "learning_rate": 9.826045209041809e-05, "epoch": 0.18, "step": 8800} | |
| {"loss": 0.8164, "learning_rate": 9.824044808961794e-05, "epoch": 0.18, "step": 8900} | |
| {"loss": 0.7529, "learning_rate": 9.822044408881777e-05, "epoch": 0.18, "step": 9000} | |
| {"loss": 0.8037, "learning_rate": 9.820044008801761e-05, "epoch": 0.18, "step": 9100} | |
| {"loss": 0.8537, "learning_rate": 9.818043608721746e-05, "epoch": 0.18, "step": 9200} | |
| {"loss": 0.8241, "learning_rate": 9.816043208641729e-05, "epoch": 0.19, "step": 9300} | |
| {"loss": 0.7729, "learning_rate": 9.814042808561713e-05, "epoch": 0.19, "step": 9400} | |
| {"loss": 0.8116, "learning_rate": 9.812042408481696e-05, "epoch": 0.19, "step": 9500} | |
| {"loss": 0.8254, "learning_rate": 9.81004200840168e-05, "epoch": 0.19, "step": 9600} | |
| {"loss": 0.8285, "learning_rate": 9.808041608321665e-05, "epoch": 0.19, "step": 9700} | |
| {"loss": 0.7577, "learning_rate": 9.806041208241648e-05, "epoch": 0.2, "step": 9800} | |
| {"loss": 0.8024, "learning_rate": 9.804040808161633e-05, "epoch": 0.2, "step": 9900} | |
| {"loss": 0.8078, "learning_rate": 9.802040408081616e-05, "epoch": 0.2, "step": 10000} | |
| {"eval_loss": 0.740916907787323, "eval_runtime": 689.7578, "eval_samples_per_second": 7.211, "eval_steps_per_second": 0.902, "epoch": 0.2, "step": 10000} | |
| {"loss": 0.7849, "learning_rate": 9.8000400080016e-05, "epoch": 0.2, "step": 10100} | |
| {"loss": 0.7931, "learning_rate": 9.798039607921585e-05, "epoch": 0.2, "step": 10200} | |
| {"loss": 0.7884, "learning_rate": 9.796039207841569e-05, "epoch": 0.21, "step": 10300} | |
| {"loss": 0.7949, "learning_rate": 9.794038807761554e-05, "epoch": 0.21, "step": 10400} | |
| {"loss": 0.7916, "learning_rate": 9.792038407681537e-05, "epoch": 0.21, "step": 10500} | |
| {"loss": 0.8072, "learning_rate": 9.790038007601521e-05, "epoch": 0.21, "step": 10600} | |
| {"loss": 0.7965, "learning_rate": 9.788037607521506e-05, "epoch": 0.21, "step": 10700} | |
| {"loss": 0.8039, "learning_rate": 9.786037207441489e-05, "epoch": 0.22, "step": 10800} | |
| {"loss": 0.8011, "learning_rate": 9.784036807361473e-05, "epoch": 0.22, "step": 10900} | |
| {"loss": 0.8014, "learning_rate": 9.782036407281456e-05, "epoch": 0.22, "step": 11000} | |
| {"loss": 0.7835, "learning_rate": 9.78003600720144e-05, "epoch": 0.22, "step": 11100} | |
| {"loss": 0.8469, "learning_rate": 9.778035607121425e-05, "epoch": 0.22, "step": 11200} | |
| {"loss": 0.8018, "learning_rate": 9.776035207041408e-05, "epoch": 0.23, "step": 11300} | |
| {"loss": 0.7839, "learning_rate": 9.774034806961393e-05, "epoch": 0.23, "step": 11400} | |
| {"loss": 0.8312, "learning_rate": 9.772034406881376e-05, "epoch": 0.23, "step": 11500} | |
| {"loss": 0.7886, "learning_rate": 9.770054010802161e-05, "epoch": 0.23, "step": 11600} | |
| {"loss": 0.8096, "learning_rate": 9.768053610722146e-05, "epoch": 0.23, "step": 11700} | |
| {"loss": 0.7806, "learning_rate": 9.766053210642129e-05, "epoch": 0.24, "step": 11800} | |
| {"loss": 0.8512, "learning_rate": 9.764052810562113e-05, "epoch": 0.24, "step": 11900} | |
| {"loss": 0.8475, "learning_rate": 9.762052410482096e-05, "epoch": 0.24, "step": 12000} | |
| {"loss": 0.749, "learning_rate": 9.76005201040208e-05, "epoch": 0.24, "step": 12100} | |
| {"loss": 0.8245, "learning_rate": 9.758051610322065e-05, "epoch": 0.24, "step": 12200} | |
| {"loss": 0.7966, "learning_rate": 9.756051210242048e-05, "epoch": 0.25, "step": 12300} | |
| {"loss": 0.773, "learning_rate": 9.754050810162033e-05, "epoch": 0.25, "step": 12400} | |
| {"loss": 0.7876, "learning_rate": 9.752050410082017e-05, "epoch": 0.25, "step": 12500} | |
| {"loss": 0.7982, "learning_rate": 9.750050010002001e-05, "epoch": 0.25, "step": 12600} | |
| {"loss": 0.7509, "learning_rate": 9.748049609921986e-05, "epoch": 0.25, "step": 12700} | |
| {"loss": 0.7815, "learning_rate": 9.746049209841969e-05, "epoch": 0.26, "step": 12800} | |
| {"loss": 0.8516, "learning_rate": 9.744048809761953e-05, "epoch": 0.26, "step": 12900} | |
| {"loss": 0.7684, "learning_rate": 9.742048409681937e-05, "epoch": 0.26, "step": 13000} | |
| {"loss": 0.7889, "learning_rate": 9.740048009601921e-05, "epoch": 0.26, "step": 13100} | |
| {"loss": 0.7882, "learning_rate": 9.738047609521905e-05, "epoch": 0.26, "step": 13200} | |
| {"loss": 0.8196, "learning_rate": 9.736047209441889e-05, "epoch": 0.27, "step": 13300} | |
| {"loss": 0.7889, "learning_rate": 9.734046809361873e-05, "epoch": 0.27, "step": 13400} | |
| {"loss": 0.7931, "learning_rate": 9.732046409281856e-05, "epoch": 0.27, "step": 13500} | |
| {"loss": 0.8172, "learning_rate": 9.73004600920184e-05, "epoch": 0.27, "step": 13600} | |
| {"loss": 0.8364, "learning_rate": 9.728045609121825e-05, "epoch": 0.27, "step": 13700} | |
| {"loss": 0.822, "learning_rate": 9.726045209041808e-05, "epoch": 0.28, "step": 13800} | |
| {"loss": 0.8546, "learning_rate": 9.724044808961793e-05, "epoch": 0.28, "step": 13900} | |
| {"loss": 0.7977, "learning_rate": 9.722044408881777e-05, "epoch": 0.28, "step": 14000} | |
| {"loss": 0.7591, "learning_rate": 9.720044008801761e-05, "epoch": 0.28, "step": 14100} | |
| {"loss": 0.7787, "learning_rate": 9.718063612722545e-05, "epoch": 0.28, "step": 14200} | |
| {"loss": 0.8361, "learning_rate": 9.716063212642529e-05, "epoch": 0.29, "step": 14300} | |
| {"loss": 0.8153, "learning_rate": 9.714062812562513e-05, "epoch": 0.29, "step": 14400} | |
| {"loss": 0.776, "learning_rate": 9.712062412482496e-05, "epoch": 0.29, "step": 14500} | |
| {"loss": 0.7491, "learning_rate": 9.71006201240248e-05, "epoch": 0.29, "step": 14600} | |
| {"loss": 0.7975, "learning_rate": 9.708061612322465e-05, "epoch": 0.29, "step": 14700} | |
| {"loss": 0.7637, "learning_rate": 9.70606121224245e-05, "epoch": 0.3, "step": 14800} | |
| {"loss": 0.7977, "learning_rate": 9.704060812162434e-05, "epoch": 0.3, "step": 14900} | |
| {"loss": 0.8041, "learning_rate": 9.702060412082417e-05, "epoch": 0.3, "step": 15000} | |
| {"loss": 0.7997, "learning_rate": 9.700060012002401e-05, "epoch": 0.3, "step": 15100} | |
| {"loss": 0.8221, "learning_rate": 9.698059611922386e-05, "epoch": 0.3, "step": 15200} | |
| {"loss": 0.7657, "learning_rate": 9.696059211842369e-05, "epoch": 0.31, "step": 15300} | |
| {"loss": 0.8091, "learning_rate": 9.694058811762353e-05, "epoch": 0.31, "step": 15400} | |
| {"loss": 0.7692, "learning_rate": 9.692058411682336e-05, "epoch": 0.31, "step": 15500} | |
| {"loss": 0.7988, "learning_rate": 9.690058011602321e-05, "epoch": 0.31, "step": 15600} | |
| {"loss": 0.7457, "learning_rate": 9.688057611522305e-05, "epoch": 0.31, "step": 15700} | |
| {"loss": 0.8223, "learning_rate": 9.686057211442288e-05, "epoch": 0.32, "step": 15800} | |
| {"loss": 0.8366, "learning_rate": 9.684056811362273e-05, "epoch": 0.32, "step": 15900} | |
| {"loss": 0.7959, "learning_rate": 9.682056411282256e-05, "epoch": 0.32, "step": 16000} | |
| {"loss": 0.7944, "learning_rate": 9.680076015203041e-05, "epoch": 0.32, "step": 16100} | |
| {"loss": 0.7645, "learning_rate": 9.678075615123024e-05, "epoch": 0.32, "step": 16200} | |
| {"loss": 0.8111, "learning_rate": 9.676075215043009e-05, "epoch": 0.33, "step": 16300} | |
| {"loss": 0.7907, "learning_rate": 9.674074814962993e-05, "epoch": 0.33, "step": 16400} | |
| {"loss": 0.8108, "learning_rate": 9.672074414882976e-05, "epoch": 0.33, "step": 16500} | |
| {"loss": 0.8119, "learning_rate": 9.670074014802961e-05, "epoch": 0.33, "step": 16600} | |
| {"loss": 0.8109, "learning_rate": 9.668073614722945e-05, "epoch": 0.33, "step": 16700} | |
| {"loss": 0.7862, "learning_rate": 9.666073214642928e-05, "epoch": 0.34, "step": 16800} | |
| {"loss": 0.7886, "learning_rate": 9.664072814562913e-05, "epoch": 0.34, "step": 16900} | |
| {"loss": 0.786, "learning_rate": 9.662072414482897e-05, "epoch": 0.34, "step": 17000} | |
| {"loss": 0.8014, "learning_rate": 9.660072014402882e-05, "epoch": 0.34, "step": 17100} | |
| {"loss": 0.7486, "learning_rate": 9.658071614322865e-05, "epoch": 0.34, "step": 17200} | |
| {"loss": 0.7688, "learning_rate": 9.65607121424285e-05, "epoch": 0.35, "step": 17300} | |
| {"loss": 0.7815, "learning_rate": 9.654070814162834e-05, "epoch": 0.35, "step": 17400} | |
| {"loss": 0.7857, "learning_rate": 9.652070414082817e-05, "epoch": 0.35, "step": 17500} | |
| {"loss": 0.777, "learning_rate": 9.650070014002801e-05, "epoch": 0.35, "step": 17600} | |
| {"loss": 0.8186, "learning_rate": 9.648069613922786e-05, "epoch": 0.35, "step": 17700} | |
| {"loss": 0.7765, "learning_rate": 9.646069213842769e-05, "epoch": 0.36, "step": 17800} | |
| {"loss": 0.8268, "learning_rate": 9.644068813762753e-05, "epoch": 0.36, "step": 17900} | |
| {"loss": 0.7864, "learning_rate": 9.642068413682736e-05, "epoch": 0.36, "step": 18000} | |
| {"loss": 0.7741, "learning_rate": 9.640068013602721e-05, "epoch": 0.36, "step": 18100} | |
| {"loss": 0.831, "learning_rate": 9.638067613522705e-05, "epoch": 0.36, "step": 18200} | |
| {"loss": 0.7912, "learning_rate": 9.636067213442688e-05, "epoch": 0.37, "step": 18300} | |
| {"loss": 0.7864, "learning_rate": 9.634066813362673e-05, "epoch": 0.37, "step": 18400} | |
| {"loss": 0.7639, "learning_rate": 9.632066413282656e-05, "epoch": 0.37, "step": 18500} | |
| {"loss": 0.7874, "learning_rate": 9.63006601320264e-05, "epoch": 0.37, "step": 18600} | |
| {"loss": 0.7642, "learning_rate": 9.628065613122625e-05, "epoch": 0.37, "step": 18700} | |
| {"loss": 0.756, "learning_rate": 9.626065213042609e-05, "epoch": 0.38, "step": 18800} | |
| {"loss": 0.8036, "learning_rate": 9.624064812962594e-05, "epoch": 0.38, "step": 18900} | |
| {"loss": 0.7537, "learning_rate": 9.622064412882577e-05, "epoch": 0.38, "step": 19000} | |
| {"loss": 0.8193, "learning_rate": 9.620064012802561e-05, "epoch": 0.38, "step": 19100} | |
| {"loss": 0.8036, "learning_rate": 9.618063612722546e-05, "epoch": 0.38, "step": 19200} | |
| {"loss": 0.784, "learning_rate": 9.616063212642529e-05, "epoch": 0.39, "step": 19300} | |
| {"loss": 0.7861, "learning_rate": 9.614062812562513e-05, "epoch": 0.39, "step": 19400} | |
| {"loss": 0.7882, "learning_rate": 9.612062412482496e-05, "epoch": 0.39, "step": 19500} | |
| {"loss": 0.8048, "learning_rate": 9.610062012402481e-05, "epoch": 0.39, "step": 19600} | |
| {"loss": 0.8544, "learning_rate": 9.608061612322465e-05, "epoch": 0.39, "step": 19700} | |
| {"loss": 0.7827, "learning_rate": 9.606061212242448e-05, "epoch": 0.4, "step": 19800} | |
| {"loss": 0.7641, "learning_rate": 9.604060812162433e-05, "epoch": 0.4, "step": 19900} | |
| {"loss": 0.7146, "learning_rate": 9.602060412082416e-05, "epoch": 0.4, "step": 20000} | |
| {"eval_loss": 0.7286937832832336, "eval_runtime": 689.1627, "eval_samples_per_second": 7.217, "eval_steps_per_second": 0.903, "epoch": 0.4, "step": 20000} | |
| {"loss": 0.7801, "learning_rate": 9.6000600120024e-05, "epoch": 0.4, "step": 20100} | |
| {"loss": 0.7803, "learning_rate": 9.598059611922385e-05, "epoch": 0.4, "step": 20200} | |
| {"loss": 0.7981, "learning_rate": 9.596059211842369e-05, "epoch": 0.41, "step": 20300} | |
| {"loss": 0.755, "learning_rate": 9.594058811762354e-05, "epoch": 0.41, "step": 20400} | |
| {"loss": 0.7743, "learning_rate": 9.592058411682337e-05, "epoch": 0.41, "step": 20500} | |
| {"loss": 0.7358, "learning_rate": 9.590098019603922e-05, "epoch": 0.41, "step": 20600} | |
| {"loss": 0.7851, "learning_rate": 9.588097619523905e-05, "epoch": 0.41, "step": 20700} | |
| {"loss": 0.7746, "learning_rate": 9.586097219443889e-05, "epoch": 0.42, "step": 20800} | |
| {"loss": 0.8246, "learning_rate": 9.584096819363874e-05, "epoch": 0.42, "step": 20900} | |
| {"loss": 0.7833, "learning_rate": 9.582096419283857e-05, "epoch": 0.42, "step": 21000} | |
| {"loss": 0.7666, "learning_rate": 9.580096019203841e-05, "epoch": 0.42, "step": 21100} | |
| {"loss": 0.8575, "learning_rate": 9.578095619123824e-05, "epoch": 0.42, "step": 21200} | |
| {"loss": 0.7934, "learning_rate": 9.576095219043809e-05, "epoch": 0.43, "step": 21300} | |
| {"loss": 0.7473, "learning_rate": 9.574094818963793e-05, "epoch": 0.43, "step": 21400} | |
| {"loss": 0.7455, "learning_rate": 9.572094418883776e-05, "epoch": 0.43, "step": 21500} | |
| {"loss": 0.7527, "learning_rate": 9.570094018803761e-05, "epoch": 0.43, "step": 21600} | |
| {"loss": 0.7906, "learning_rate": 9.568093618723745e-05, "epoch": 0.43, "step": 21700} | |
| {"loss": 0.8384, "learning_rate": 9.56609321864373e-05, "epoch": 0.44, "step": 21800} | |
| {"loss": 0.7897, "learning_rate": 9.564092818563714e-05, "epoch": 0.44, "step": 21900} | |
| {"loss": 0.7459, "learning_rate": 9.562092418483697e-05, "epoch": 0.44, "step": 22000} | |
| {"loss": 0.7838, "learning_rate": 9.560092018403682e-05, "epoch": 0.44, "step": 22100} | |
| {"loss": 0.7745, "learning_rate": 9.558091618323666e-05, "epoch": 0.44, "step": 22200} | |
| {"loss": 0.7985, "learning_rate": 9.556091218243649e-05, "epoch": 0.45, "step": 22300} | |
| {"loss": 0.7945, "learning_rate": 9.554090818163634e-05, "epoch": 0.45, "step": 22400} | |
| {"loss": 0.7985, "learning_rate": 9.552090418083617e-05, "epoch": 0.45, "step": 22500} | |
| {"loss": 0.7785, "learning_rate": 9.550090018003601e-05, "epoch": 0.45, "step": 22600} | |
| {"loss": 0.7642, "learning_rate": 9.548089617923586e-05, "epoch": 0.45, "step": 22700} | |
| {"loss": 0.7953, "learning_rate": 9.546089217843569e-05, "epoch": 0.46, "step": 22800} | |
| {"loss": 0.7848, "learning_rate": 9.544088817763553e-05, "epoch": 0.46, "step": 22900} | |
| {"loss": 0.7388, "learning_rate": 9.542088417683536e-05, "epoch": 0.46, "step": 23000} | |
| {"loss": 0.7947, "learning_rate": 9.540088017603521e-05, "epoch": 0.46, "step": 23100} | |
| {"loss": 0.7464, "learning_rate": 9.538087617523505e-05, "epoch": 0.46, "step": 23200} | |
| {"loss": 0.7696, "learning_rate": 9.53608721744349e-05, "epoch": 0.47, "step": 23300} | |
| {"loss": 0.7563, "learning_rate": 9.534086817363474e-05, "epoch": 0.47, "step": 23400} | |
| {"loss": 0.7522, "learning_rate": 9.532086417283457e-05, "epoch": 0.47, "step": 23500} | |
| {"loss": 0.772, "learning_rate": 9.530086017203442e-05, "epoch": 0.47, "step": 23600} | |
| {"loss": 0.7654, "learning_rate": 9.528085617123426e-05, "epoch": 0.47, "step": 23700} | |
| {"loss": 0.759, "learning_rate": 9.526085217043409e-05, "epoch": 0.48, "step": 23800} | |
| {"loss": 0.8032, "learning_rate": 9.524084816963394e-05, "epoch": 0.48, "step": 23900} | |
| {"loss": 0.7638, "learning_rate": 9.522084416883377e-05, "epoch": 0.48, "step": 24000} | |
| {"loss": 0.7974, "learning_rate": 9.520084016803361e-05, "epoch": 0.48, "step": 24100} | |
| {"loss": 0.7647, "learning_rate": 9.518083616723346e-05, "epoch": 0.48, "step": 24200} | |
| {"loss": 0.8016, "learning_rate": 9.516083216643329e-05, "epoch": 0.49, "step": 24300} | |
| {"loss": 0.7749, "learning_rate": 9.514082816563313e-05, "epoch": 0.49, "step": 24400} | |
| {"loss": 0.7573, "learning_rate": 9.512082416483296e-05, "epoch": 0.49, "step": 24500} | |
| {"loss": 0.7476, "learning_rate": 9.51008201640328e-05, "epoch": 0.49, "step": 24600} | |
| {"loss": 0.7848, "learning_rate": 9.508081616323265e-05, "epoch": 0.49, "step": 24700} | |
| {"loss": 0.7994, "learning_rate": 9.50608121624325e-05, "epoch": 0.5, "step": 24800} | |
| {"loss": 0.8056, "learning_rate": 9.504100820164034e-05, "epoch": 0.5, "step": 24900} | |
| {"loss": 0.8066, "learning_rate": 9.502100420084017e-05, "epoch": 0.5, "step": 25000} | |
| {"loss": 0.7506, "learning_rate": 9.500100020004001e-05, "epoch": 0.5, "step": 25100} | |
| {"loss": 0.8018, "learning_rate": 9.498099619923986e-05, "epoch": 0.5, "step": 25200} | |
| {"loss": 0.7451, "learning_rate": 9.496099219843969e-05, "epoch": 0.51, "step": 25300} | |
| {"loss": 0.764, "learning_rate": 9.494098819763953e-05, "epoch": 0.51, "step": 25400} | |
| {"loss": 0.8212, "learning_rate": 9.492098419683938e-05, "epoch": 0.51, "step": 25500} | |
| {"loss": 0.7954, "learning_rate": 9.490098019603922e-05, "epoch": 0.51, "step": 25600} | |
| {"loss": 0.7965, "learning_rate": 9.488097619523905e-05, "epoch": 0.51, "step": 25700} | |
| {"loss": 0.7795, "learning_rate": 9.48609721944389e-05, "epoch": 0.52, "step": 25800} | |
| {"loss": 0.7993, "learning_rate": 9.484096819363874e-05, "epoch": 0.52, "step": 25900} | |
| {"loss": 0.7951, "learning_rate": 9.482096419283857e-05, "epoch": 0.52, "step": 26000} | |
| {"loss": 0.7432, "learning_rate": 9.480096019203842e-05, "epoch": 0.52, "step": 26100} | |
| {"loss": 0.7844, "learning_rate": 9.478095619123826e-05, "epoch": 0.52, "step": 26200} | |
| {"loss": 0.7029, "learning_rate": 9.476095219043809e-05, "epoch": 0.53, "step": 26300} | |
| {"loss": 0.7613, "learning_rate": 9.474094818963793e-05, "epoch": 0.53, "step": 26400} | |
| {"loss": 0.7896, "learning_rate": 9.472094418883777e-05, "epoch": 0.53, "step": 26500} | |
| {"loss": 0.7674, "learning_rate": 9.470094018803761e-05, "epoch": 0.53, "step": 26600} | |
| {"loss": 0.8013, "learning_rate": 9.468093618723745e-05, "epoch": 0.53, "step": 26700} | |
| {"loss": 0.811, "learning_rate": 9.466093218643729e-05, "epoch": 0.54, "step": 26800} | |
| {"loss": 0.8235, "learning_rate": 9.464092818563713e-05, "epoch": 0.54, "step": 26900} | |
| {"loss": 0.7717, "learning_rate": 9.462092418483696e-05, "epoch": 0.54, "step": 27000} | |
| {"loss": 0.7364, "learning_rate": 9.46009201840368e-05, "epoch": 0.54, "step": 27100} | |
| {"loss": 0.7663, "learning_rate": 9.458111622324465e-05, "epoch": 0.54, "step": 27200} | |
| {"loss": 0.771, "learning_rate": 9.456111222244449e-05, "epoch": 0.55, "step": 27300} | |
| {"loss": 0.7583, "learning_rate": 9.454110822164433e-05, "epoch": 0.55, "step": 27400} | |
| {"loss": 0.7672, "learning_rate": 9.452110422084417e-05, "epoch": 0.55, "step": 27500} | |
| {"loss": 0.7706, "learning_rate": 9.450110022004401e-05, "epoch": 0.55, "step": 27600} | |
| {"loss": 0.7676, "learning_rate": 9.448109621924385e-05, "epoch": 0.55, "step": 27700} | |
| {"loss": 0.7859, "learning_rate": 9.446109221844369e-05, "epoch": 0.56, "step": 27800} | |
| {"loss": 0.7573, "learning_rate": 9.444108821764353e-05, "epoch": 0.56, "step": 27900} | |
| {"loss": 0.8289, "learning_rate": 9.442108421684337e-05, "epoch": 0.56, "step": 28000} | |
| {"loss": 0.7757, "learning_rate": 9.440108021604322e-05, "epoch": 0.56, "step": 28100} | |
| {"loss": 0.7688, "learning_rate": 9.438107621524306e-05, "epoch": 0.56, "step": 28200} | |
| {"loss": 0.7941, "learning_rate": 9.43610722144429e-05, "epoch": 0.57, "step": 28300} | |
| {"loss": 0.7829, "learning_rate": 9.434106821364274e-05, "epoch": 0.57, "step": 28400} | |
| {"loss": 0.7744, "learning_rate": 9.432106421284257e-05, "epoch": 0.57, "step": 28500} | |
| {"loss": 0.805, "learning_rate": 9.430106021204241e-05, "epoch": 0.57, "step": 28600} | |
| {"loss": 0.776, "learning_rate": 9.428105621124226e-05, "epoch": 0.57, "step": 28700} | |
| {"loss": 0.7839, "learning_rate": 9.426105221044209e-05, "epoch": 0.58, "step": 28800} | |
| {"loss": 0.7894, "learning_rate": 9.424104820964193e-05, "epoch": 0.58, "step": 28900} | |
| {"loss": 0.8138, "learning_rate": 9.422104420884177e-05, "epoch": 0.58, "step": 29000} | |
| {"loss": 0.7834, "learning_rate": 9.420104020804161e-05, "epoch": 0.58, "step": 29100} | |
| {"loss": 0.7323, "learning_rate": 9.418103620724145e-05, "epoch": 0.58, "step": 29200} | |
| {"loss": 0.79, "learning_rate": 9.416103220644129e-05, "epoch": 0.59, "step": 29300} | |
| {"loss": 0.773, "learning_rate": 9.414102820564113e-05, "epoch": 0.59, "step": 29400} | |
| {"loss": 0.7607, "learning_rate": 9.412102420484097e-05, "epoch": 0.59, "step": 29500} | |
| {"loss": 0.7581, "learning_rate": 9.410102020404082e-05, "epoch": 0.59, "step": 29600} | |
| {"loss": 0.7528, "learning_rate": 9.408101620324066e-05, "epoch": 0.59, "step": 29700} | |
| {"loss": 0.7755, "learning_rate": 9.40610122024405e-05, "epoch": 0.6, "step": 29800} | |
| {"loss": 0.7633, "learning_rate": 9.404100820164034e-05, "epoch": 0.6, "step": 29900} | |
| {"loss": 0.7352, "learning_rate": 9.402100420084017e-05, "epoch": 0.6, "step": 30000} | |
| {"eval_loss": 0.7202807068824768, "eval_runtime": 707.6198, "eval_samples_per_second": 7.029, "eval_steps_per_second": 0.879, "epoch": 0.6, "step": 30000} | |
| {"loss": 0.7437, "learning_rate": 9.400100020004001e-05, "epoch": 0.6, "step": 30100} | |
| {"loss": 0.7887, "learning_rate": 9.398099619923986e-05, "epoch": 0.6, "step": 30200} | |
| {"loss": 0.7586, "learning_rate": 9.396099219843969e-05, "epoch": 0.61, "step": 30300} | |
| {"loss": 0.7925, "learning_rate": 9.394098819763953e-05, "epoch": 0.61, "step": 30400} | |
| {"loss": 0.7947, "learning_rate": 9.392098419683936e-05, "epoch": 0.61, "step": 30500} | |
| {"loss": 0.7701, "learning_rate": 9.390098019603921e-05, "epoch": 0.61, "step": 30600} | |
| {"loss": 0.7631, "learning_rate": 9.388117623524706e-05, "epoch": 0.61, "step": 30700} | |
| {"loss": 0.7977, "learning_rate": 9.38611722344469e-05, "epoch": 0.62, "step": 30800} | |
| {"loss": 0.7941, "learning_rate": 9.384116823364674e-05, "epoch": 0.62, "step": 30900} | |
| {"loss": 0.7929, "learning_rate": 9.382116423284657e-05, "epoch": 0.62, "step": 31000} | |
| {"loss": 0.7871, "learning_rate": 9.380116023204641e-05, "epoch": 0.62, "step": 31100} | |
| {"loss": 0.7449, "learning_rate": 9.378115623124626e-05, "epoch": 0.62, "step": 31200} | |
| {"loss": 0.7994, "learning_rate": 9.376115223044609e-05, "epoch": 0.63, "step": 31300} | |
| {"loss": 0.8307, "learning_rate": 9.374114822964593e-05, "epoch": 0.63, "step": 31400} | |
| {"loss": 0.7521, "learning_rate": 9.372114422884576e-05, "epoch": 0.63, "step": 31500} | |
| {"loss": 0.7767, "learning_rate": 9.370114022804561e-05, "epoch": 0.63, "step": 31600} | |
| {"loss": 0.7686, "learning_rate": 9.368113622724545e-05, "epoch": 0.63, "step": 31700} | |
| {"loss": 0.7903, "learning_rate": 9.36611322264453e-05, "epoch": 0.64, "step": 31800} | |
| {"loss": 0.7247, "learning_rate": 9.364112822564514e-05, "epoch": 0.64, "step": 31900} | |
| {"loss": 0.7425, "learning_rate": 9.362112422484497e-05, "epoch": 0.64, "step": 32000} | |
| {"loss": 0.7607, "learning_rate": 9.360112022404482e-05, "epoch": 0.64, "step": 32100} | |
| {"loss": 0.7745, "learning_rate": 9.358111622324466e-05, "epoch": 0.64, "step": 32200} | |
| {"loss": 0.7873, "learning_rate": 9.356111222244449e-05, "epoch": 0.65, "step": 32300} | |
| {"loss": 0.7496, "learning_rate": 9.354110822164434e-05, "epoch": 0.65, "step": 32400} | |
| {"loss": 0.7789, "learning_rate": 9.352110422084417e-05, "epoch": 0.65, "step": 32500} | |
| {"loss": 0.7862, "learning_rate": 9.350110022004401e-05, "epoch": 0.65, "step": 32600} | |
| {"loss": 0.7496, "learning_rate": 9.348109621924386e-05, "epoch": 0.65, "step": 32700} | |
| {"loss": 0.7601, "learning_rate": 9.346109221844369e-05, "epoch": 0.66, "step": 32800} | |
| {"loss": 0.822, "learning_rate": 9.344108821764353e-05, "epoch": 0.66, "step": 32900} | |
| {"loss": 0.7935, "learning_rate": 9.342108421684336e-05, "epoch": 0.66, "step": 33000} | |
| {"loss": 0.7496, "learning_rate": 9.340108021604321e-05, "epoch": 0.66, "step": 33100} | |
| {"loss": 0.7517, "learning_rate": 9.338107621524305e-05, "epoch": 0.66, "step": 33200} | |
| {"loss": 0.7648, "learning_rate": 9.33610722144429e-05, "epoch": 0.67, "step": 33300} | |
| {"loss": 0.7855, "learning_rate": 9.334106821364274e-05, "epoch": 0.67, "step": 33400} | |
| {"loss": 0.8075, "learning_rate": 9.332106421284257e-05, "epoch": 0.67, "step": 33500} | |
| {"loss": 0.7918, "learning_rate": 9.330106021204242e-05, "epoch": 0.67, "step": 33600} | |
| {"loss": 0.7654, "learning_rate": 9.328105621124226e-05, "epoch": 0.67, "step": 33700} | |
| {"loss": 0.7865, "learning_rate": 9.326105221044209e-05, "epoch": 0.68, "step": 33800} | |
| {"loss": 0.7665, "learning_rate": 9.324104820964194e-05, "epoch": 0.68, "step": 33900} | |
| {"loss": 0.7662, "learning_rate": 9.322104420884177e-05, "epoch": 0.68, "step": 34000} | |
| {"loss": 0.7713, "learning_rate": 9.320104020804161e-05, "epoch": 0.68, "step": 34100} | |
| {"loss": 0.7528, "learning_rate": 9.318103620724146e-05, "epoch": 0.68, "step": 34200} | |
| {"loss": 0.7992, "learning_rate": 9.316103220644129e-05, "epoch": 0.69, "step": 34300} | |
| {"loss": 0.762, "learning_rate": 9.314102820564113e-05, "epoch": 0.69, "step": 34400} | |
| {"loss": 0.7782, "learning_rate": 9.312122424484897e-05, "epoch": 0.69, "step": 34500} | |
| {"loss": 0.764, "learning_rate": 9.310122024404882e-05, "epoch": 0.69, "step": 34600} | |
| {"loss": 0.7462, "learning_rate": 9.308121624324866e-05, "epoch": 0.69, "step": 34700} | |
| {"loss": 0.7737, "learning_rate": 9.306121224244849e-05, "epoch": 0.7, "step": 34800} | |
| {"loss": 0.7289, "learning_rate": 9.304120824164834e-05, "epoch": 0.7, "step": 34900} | |
| {"loss": 0.7715, "learning_rate": 9.302120424084817e-05, "epoch": 0.7, "step": 35000} | |
| {"loss": 0.7211, "learning_rate": 9.300120024004801e-05, "epoch": 0.7, "step": 35100} | |
| {"loss": 0.7473, "learning_rate": 9.298119623924786e-05, "epoch": 0.7, "step": 35200} | |
| {"loss": 0.7805, "learning_rate": 9.296119223844769e-05, "epoch": 0.71, "step": 35300} | |
| {"loss": 0.7617, "learning_rate": 9.294118823764753e-05, "epoch": 0.71, "step": 35400} | |
| {"loss": 0.7964, "learning_rate": 9.292118423684736e-05, "epoch": 0.71, "step": 35500} | |
| {"loss": 0.7889, "learning_rate": 9.290118023604721e-05, "epoch": 0.71, "step": 35600} | |
| {"loss": 0.7584, "learning_rate": 9.288117623524705e-05, "epoch": 0.71, "step": 35700} | |
| {"loss": 0.8073, "learning_rate": 9.28611722344469e-05, "epoch": 0.72, "step": 35800} | |
| {"loss": 0.7515, "learning_rate": 9.284116823364674e-05, "epoch": 0.72, "step": 35900} | |
| {"loss": 0.7605, "learning_rate": 9.282116423284657e-05, "epoch": 0.72, "step": 36000} | |
| {"loss": 0.7869, "learning_rate": 9.280116023204642e-05, "epoch": 0.72, "step": 36100} | |
| {"loss": 0.7255, "learning_rate": 9.278115623124626e-05, "epoch": 0.72, "step": 36200} | |
| {"loss": 0.7429, "learning_rate": 9.276115223044609e-05, "epoch": 0.73, "step": 36300} | |
| {"loss": 0.7826, "learning_rate": 9.274114822964594e-05, "epoch": 0.73, "step": 36400} | |
| {"loss": 0.7448, "learning_rate": 9.272114422884577e-05, "epoch": 0.73, "step": 36500} | |
| {"loss": 0.7862, "learning_rate": 9.270114022804561e-05, "epoch": 0.73, "step": 36600} | |
| {"loss": 0.7566, "learning_rate": 9.268113622724546e-05, "epoch": 0.73, "step": 36700} | |
| {"loss": 0.7131, "learning_rate": 9.266113222644529e-05, "epoch": 0.74, "step": 36800} | |
| {"loss": 0.77, "learning_rate": 9.264112822564513e-05, "epoch": 0.74, "step": 36900} | |
| {"loss": 0.7562, "learning_rate": 9.262112422484496e-05, "epoch": 0.74, "step": 37000} | |
| {"loss": 0.7475, "learning_rate": 9.260112022404481e-05, "epoch": 0.74, "step": 37100} | |
| {"loss": 0.7773, "learning_rate": 9.258111622324465e-05, "epoch": 0.74, "step": 37200} | |
| {"loss": 0.7671, "learning_rate": 9.25611122224445e-05, "epoch": 0.75, "step": 37300} | |
| {"loss": 0.7903, "learning_rate": 9.254130826165234e-05, "epoch": 0.75, "step": 37400} | |
| {"loss": 0.8081, "learning_rate": 9.252130426085217e-05, "epoch": 0.75, "step": 37500} | |
| {"loss": 0.7324, "learning_rate": 9.250130026005201e-05, "epoch": 0.75, "step": 37600} | |
| {"loss": 0.7595, "learning_rate": 9.248129625925186e-05, "epoch": 0.75, "step": 37700} | |
| {"loss": 0.774, "learning_rate": 9.246129225845169e-05, "epoch": 0.76, "step": 37800} | |
| {"loss": 0.8066, "learning_rate": 9.244128825765153e-05, "epoch": 0.76, "step": 37900} | |
| {"loss": 0.7878, "learning_rate": 9.242128425685138e-05, "epoch": 0.76, "step": 38000} | |
| {"loss": 0.781, "learning_rate": 9.240128025605122e-05, "epoch": 0.76, "step": 38100} | |
| {"loss": 0.733, "learning_rate": 9.238127625525106e-05, "epoch": 0.76, "step": 38200} | |
| {"loss": 0.7704, "learning_rate": 9.23612722544509e-05, "epoch": 0.77, "step": 38300} | |
| {"loss": 0.7342, "learning_rate": 9.234126825365074e-05, "epoch": 0.77, "step": 38400} | |
| {"loss": 0.7681, "learning_rate": 9.232126425285057e-05, "epoch": 0.77, "step": 38500} | |
| {"loss": 0.7074, "learning_rate": 9.230126025205042e-05, "epoch": 0.77, "step": 38600} | |
| {"loss": 0.7909, "learning_rate": 9.228125625125026e-05, "epoch": 0.77, "step": 38700} | |
| {"loss": 0.7581, "learning_rate": 9.226125225045009e-05, "epoch": 0.78, "step": 38800} | |
| {"loss": 0.7346, "learning_rate": 9.224124824964994e-05, "epoch": 0.78, "step": 38900} | |
| {"loss": 0.7564, "learning_rate": 9.222124424884977e-05, "epoch": 0.78, "step": 39000} | |
| {"loss": 0.8259, "learning_rate": 9.220124024804961e-05, "epoch": 0.78, "step": 39100} | |
| {"loss": 0.7828, "learning_rate": 9.218123624724946e-05, "epoch": 0.78, "step": 39200} | |
| {"loss": 0.7757, "learning_rate": 9.216123224644929e-05, "epoch": 0.79, "step": 39300} | |
| {"loss": 0.7402, "learning_rate": 9.214122824564913e-05, "epoch": 0.79, "step": 39400} | |
| {"loss": 0.807, "learning_rate": 9.212122424484898e-05, "epoch": 0.79, "step": 39500} | |
| {"loss": 0.7574, "learning_rate": 9.210122024404882e-05, "epoch": 0.79, "step": 39600} | |
| {"loss": 0.7471, "learning_rate": 9.208121624324866e-05, "epoch": 0.79, "step": 39700} | |
| {"loss": 0.7779, "learning_rate": 9.20612122424485e-05, "epoch": 0.8, "step": 39800} | |
| {"loss": 0.7553, "learning_rate": 9.204140828165634e-05, "epoch": 0.8, "step": 39900} | |
| {"loss": 0.796, "learning_rate": 9.202140428085617e-05, "epoch": 0.8, "step": 40000} | |
| {"eval_loss": 0.7169719338417053, "eval_runtime": 690.7702, "eval_samples_per_second": 7.201, "eval_steps_per_second": 0.9, "epoch": 0.8, "step": 40000} | |
| {"loss": 0.7667, "learning_rate": 9.200140028005601e-05, "epoch": 0.8, "step": 40100} | |
| {"loss": 0.7845, "learning_rate": 9.198139627925586e-05, "epoch": 0.8, "step": 40200} | |
| {"loss": 0.7691, "learning_rate": 9.19613922784557e-05, "epoch": 0.81, "step": 40300} | |
| {"loss": 0.7522, "learning_rate": 9.194138827765554e-05, "epoch": 0.81, "step": 40400} | |
| {"loss": 0.7589, "learning_rate": 9.192138427685538e-05, "epoch": 0.81, "step": 40500} | |
| {"loss": 0.7768, "learning_rate": 9.190138027605522e-05, "epoch": 0.81, "step": 40600} | |
| {"loss": 0.7543, "learning_rate": 9.188137627525506e-05, "epoch": 0.81, "step": 40700} | |
| {"loss": 0.8211, "learning_rate": 9.18613722744549e-05, "epoch": 0.82, "step": 40800} | |
| {"loss": 0.7923, "learning_rate": 9.184136827365474e-05, "epoch": 0.82, "step": 40900} | |
| {"loss": 0.8029, "learning_rate": 9.182136427285457e-05, "epoch": 0.82, "step": 41000} | |
| {"loss": 0.7672, "learning_rate": 9.180136027205441e-05, "epoch": 0.82, "step": 41100} | |
| {"loss": 0.7621, "learning_rate": 9.178135627125426e-05, "epoch": 0.82, "step": 41200} | |
| {"loss": 0.799, "learning_rate": 9.176135227045409e-05, "epoch": 0.83, "step": 41300} | |
| {"loss": 0.7748, "learning_rate": 9.174134826965393e-05, "epoch": 0.83, "step": 41400} | |
| {"loss": 0.7332, "learning_rate": 9.172134426885377e-05, "epoch": 0.83, "step": 41500} | |
| {"loss": 0.7714, "learning_rate": 9.170134026805361e-05, "epoch": 0.83, "step": 41600} | |
| {"train_runtime": 13773.4091, "train_samples_per_second": 36.302, "train_steps_per_second": 36.302, "total_flos": 5.849606367276564e+17, "train_loss": 0.7897592906171692, "epoch": 0.83, "step": 41608} |