| { | |
| "best_metric": 0.7202807068824768, | |
| "best_model_checkpoint": "/media/xzuyn/c41d93ed-83a7-4577-9d38-5062b7e14438/xzuyn/Documents/AI_ROCM_5-6/LLaMa-LoRA-Tuner/data/lora_models/LLaMa-2-WizardLM-Uncensored-Tulu-Format-50K-7B-LoRA/checkpoint-30000", | |
| "epoch": 0.6, | |
| "global_step": 30000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1136, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.997999599919985e-05, | |
| "loss": 0.9595, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.995999199839968e-05, | |
| "loss": 0.8936, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.993998799759952e-05, | |
| "loss": 0.9237, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.992038407681536e-05, | |
| "loss": 0.9005, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.99003800760152e-05, | |
| "loss": 0.8433, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.988037607521505e-05, | |
| "loss": 0.863, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.98603720744149e-05, | |
| "loss": 0.8634, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.984036807361474e-05, | |
| "loss": 0.8463, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.982036407281457e-05, | |
| "loss": 0.8527, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.980036007201441e-05, | |
| "loss": 0.8259, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.978035607121424e-05, | |
| "loss": 0.8473, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.976035207041409e-05, | |
| "loss": 0.8462, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.974034806961393e-05, | |
| "loss": 0.8609, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.972034406881376e-05, | |
| "loss": 0.8238, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.970034006801361e-05, | |
| "loss": 0.8212, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.968033606721344e-05, | |
| "loss": 0.8391, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.966033206641328e-05, | |
| "loss": 0.774, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.964032806561313e-05, | |
| "loss": 0.8149, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.962032406481296e-05, | |
| "loss": 0.8794, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.96003200640128e-05, | |
| "loss": 0.7747, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.958031606321264e-05, | |
| "loss": 0.8191, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.956031206241248e-05, | |
| "loss": 0.8219, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.954030806161232e-05, | |
| "loss": 0.8191, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.952030406081217e-05, | |
| "loss": 0.8662, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.950030006001201e-05, | |
| "loss": 0.787, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.948029605921184e-05, | |
| "loss": 0.8668, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.946029205841169e-05, | |
| "loss": 0.7817, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.944028805761153e-05, | |
| "loss": 0.8021, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.942028405681136e-05, | |
| "loss": 0.8324, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.940028005601121e-05, | |
| "loss": 0.8631, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.938027605521105e-05, | |
| "loss": 0.797, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.936027205441088e-05, | |
| "loss": 0.8224, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.934026805361073e-05, | |
| "loss": 0.8164, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.932026405281056e-05, | |
| "loss": 0.7975, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.930046009201841e-05, | |
| "loss": 0.8468, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.928045609121824e-05, | |
| "loss": 0.8169, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.926045209041809e-05, | |
| "loss": 0.7761, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.924044808961793e-05, | |
| "loss": 0.8042, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.922044408881776e-05, | |
| "loss": 0.8146, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.920044008801761e-05, | |
| "loss": 0.8243, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.918043608721744e-05, | |
| "loss": 0.8315, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.916043208641728e-05, | |
| "loss": 0.8014, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.914042808561713e-05, | |
| "loss": 0.8241, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.912042408481696e-05, | |
| "loss": 0.8233, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.91004200840168e-05, | |
| "loss": 0.7748, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.908041608321665e-05, | |
| "loss": 0.8197, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.906041208241649e-05, | |
| "loss": 0.8195, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.904040808161634e-05, | |
| "loss": 0.7864, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.902040408081617e-05, | |
| "loss": 0.7812, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.900040008001601e-05, | |
| "loss": 0.7966, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.898039607921584e-05, | |
| "loss": 0.793, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.896039207841569e-05, | |
| "loss": 0.8568, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.894038807761553e-05, | |
| "loss": 0.7614, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.892038407681536e-05, | |
| "loss": 0.8171, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.890038007601521e-05, | |
| "loss": 0.7649, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.888037607521505e-05, | |
| "loss": 0.8074, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.886037207441488e-05, | |
| "loss": 0.7591, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.884036807361473e-05, | |
| "loss": 0.8102, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.882036407281456e-05, | |
| "loss": 0.784, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.88003600720144e-05, | |
| "loss": 0.8367, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.878035607121425e-05, | |
| "loss": 0.8418, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.876035207041409e-05, | |
| "loss": 0.798, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.874034806961394e-05, | |
| "loss": 0.7928, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.872034406881377e-05, | |
| "loss": 0.8262, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.870034006801361e-05, | |
| "loss": 0.7965, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.868033606721346e-05, | |
| "loss": 0.7952, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.866033206641329e-05, | |
| "loss": 0.7987, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.864032806561313e-05, | |
| "loss": 0.7939, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.862032406481296e-05, | |
| "loss": 0.7484, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.860032006401281e-05, | |
| "loss": 0.819, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.858031606321265e-05, | |
| "loss": 0.747, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.856031206241248e-05, | |
| "loss": 0.8029, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.854030806161233e-05, | |
| "loss": 0.9013, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.852030406081216e-05, | |
| "loss": 0.8337, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.8500300060012e-05, | |
| "loss": 0.7927, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.848029605921185e-05, | |
| "loss": 0.8725, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.846029205841169e-05, | |
| "loss": 0.7865, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.844028805761154e-05, | |
| "loss": 0.7726, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.842048409681936e-05, | |
| "loss": 0.7868, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.840048009601921e-05, | |
| "loss": 0.7918, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.838047609521905e-05, | |
| "loss": 0.7924, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.836047209441888e-05, | |
| "loss": 0.8036, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.834046809361873e-05, | |
| "loss": 0.7844, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.832046409281857e-05, | |
| "loss": 0.8124, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.830046009201842e-05, | |
| "loss": 0.837, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.828045609121826e-05, | |
| "loss": 0.8315, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.826045209041809e-05, | |
| "loss": 0.7738, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.824044808961794e-05, | |
| "loss": 0.8164, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.822044408881777e-05, | |
| "loss": 0.7529, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.820044008801761e-05, | |
| "loss": 0.8037, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.818043608721746e-05, | |
| "loss": 0.8537, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.816043208641729e-05, | |
| "loss": 0.8241, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.814042808561713e-05, | |
| "loss": 0.7729, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.812042408481696e-05, | |
| "loss": 0.8116, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.81004200840168e-05, | |
| "loss": 0.8254, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.808041608321665e-05, | |
| "loss": 0.8285, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.806041208241648e-05, | |
| "loss": 0.7577, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.804040808161633e-05, | |
| "loss": 0.8024, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.802040408081616e-05, | |
| "loss": 0.8078, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 0.740916907787323, | |
| "eval_runtime": 689.7578, | |
| "eval_samples_per_second": 7.211, | |
| "eval_steps_per_second": 0.902, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.8000400080016e-05, | |
| "loss": 0.7849, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.798039607921585e-05, | |
| "loss": 0.7931, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.796039207841569e-05, | |
| "loss": 0.7884, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.794038807761554e-05, | |
| "loss": 0.7949, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.792038407681537e-05, | |
| "loss": 0.7916, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.790038007601521e-05, | |
| "loss": 0.8072, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.788037607521506e-05, | |
| "loss": 0.7965, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.786037207441489e-05, | |
| "loss": 0.8039, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.784036807361473e-05, | |
| "loss": 0.8011, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.782036407281456e-05, | |
| "loss": 0.8014, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.78003600720144e-05, | |
| "loss": 0.7835, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.778035607121425e-05, | |
| "loss": 0.8469, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.776035207041408e-05, | |
| "loss": 0.8018, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.774034806961393e-05, | |
| "loss": 0.7839, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.772034406881376e-05, | |
| "loss": 0.8312, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.770054010802161e-05, | |
| "loss": 0.7886, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.768053610722146e-05, | |
| "loss": 0.8096, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.766053210642129e-05, | |
| "loss": 0.7806, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.764052810562113e-05, | |
| "loss": 0.8512, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.762052410482096e-05, | |
| "loss": 0.8475, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.76005201040208e-05, | |
| "loss": 0.749, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.758051610322065e-05, | |
| "loss": 0.8245, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.756051210242048e-05, | |
| "loss": 0.7966, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.754050810162033e-05, | |
| "loss": 0.773, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.752050410082017e-05, | |
| "loss": 0.7876, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.750050010002001e-05, | |
| "loss": 0.7982, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.748049609921986e-05, | |
| "loss": 0.7509, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.746049209841969e-05, | |
| "loss": 0.7815, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.744048809761953e-05, | |
| "loss": 0.8516, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.742048409681937e-05, | |
| "loss": 0.7684, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.740048009601921e-05, | |
| "loss": 0.7889, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.738047609521905e-05, | |
| "loss": 0.7882, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.736047209441889e-05, | |
| "loss": 0.8196, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.734046809361873e-05, | |
| "loss": 0.7889, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.732046409281856e-05, | |
| "loss": 0.7931, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.73004600920184e-05, | |
| "loss": 0.8172, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.728045609121825e-05, | |
| "loss": 0.8364, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.726045209041808e-05, | |
| "loss": 0.822, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.724044808961793e-05, | |
| "loss": 0.8546, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.722044408881777e-05, | |
| "loss": 0.7977, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.720044008801761e-05, | |
| "loss": 0.7591, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.718063612722545e-05, | |
| "loss": 0.7787, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.716063212642529e-05, | |
| "loss": 0.8361, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.714062812562513e-05, | |
| "loss": 0.8153, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.712062412482496e-05, | |
| "loss": 0.776, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.71006201240248e-05, | |
| "loss": 0.7491, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.708061612322465e-05, | |
| "loss": 0.7975, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.70606121224245e-05, | |
| "loss": 0.7637, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.704060812162434e-05, | |
| "loss": 0.7977, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.702060412082417e-05, | |
| "loss": 0.8041, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.700060012002401e-05, | |
| "loss": 0.7997, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.698059611922386e-05, | |
| "loss": 0.8221, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.696059211842369e-05, | |
| "loss": 0.7657, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.694058811762353e-05, | |
| "loss": 0.8091, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.692058411682336e-05, | |
| "loss": 0.7692, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.690058011602321e-05, | |
| "loss": 0.7988, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.688057611522305e-05, | |
| "loss": 0.7457, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.686057211442288e-05, | |
| "loss": 0.8223, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.684056811362273e-05, | |
| "loss": 0.8366, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.682056411282256e-05, | |
| "loss": 0.7959, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.680076015203041e-05, | |
| "loss": 0.7944, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.678075615123024e-05, | |
| "loss": 0.7645, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.676075215043009e-05, | |
| "loss": 0.8111, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.674074814962993e-05, | |
| "loss": 0.7907, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.672074414882976e-05, | |
| "loss": 0.8108, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.670074014802961e-05, | |
| "loss": 0.8119, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.668073614722945e-05, | |
| "loss": 0.8109, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.666073214642928e-05, | |
| "loss": 0.7862, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.664072814562913e-05, | |
| "loss": 0.7886, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.662072414482897e-05, | |
| "loss": 0.786, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.660072014402882e-05, | |
| "loss": 0.8014, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.658071614322865e-05, | |
| "loss": 0.7486, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.65607121424285e-05, | |
| "loss": 0.7688, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.654070814162834e-05, | |
| "loss": 0.7815, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.652070414082817e-05, | |
| "loss": 0.7857, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.650070014002801e-05, | |
| "loss": 0.777, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.648069613922786e-05, | |
| "loss": 0.8186, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.646069213842769e-05, | |
| "loss": 0.7765, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.644068813762753e-05, | |
| "loss": 0.8268, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.642068413682736e-05, | |
| "loss": 0.7864, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.640068013602721e-05, | |
| "loss": 0.7741, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.638067613522705e-05, | |
| "loss": 0.831, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.636067213442688e-05, | |
| "loss": 0.7912, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.634066813362673e-05, | |
| "loss": 0.7864, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.632066413282656e-05, | |
| "loss": 0.7639, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.63006601320264e-05, | |
| "loss": 0.7874, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.628065613122625e-05, | |
| "loss": 0.7642, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.626065213042609e-05, | |
| "loss": 0.756, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.624064812962594e-05, | |
| "loss": 0.8036, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.622064412882577e-05, | |
| "loss": 0.7537, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.620064012802561e-05, | |
| "loss": 0.8193, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.618063612722546e-05, | |
| "loss": 0.8036, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.616063212642529e-05, | |
| "loss": 0.784, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.614062812562513e-05, | |
| "loss": 0.7861, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.612062412482496e-05, | |
| "loss": 0.7882, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.610062012402481e-05, | |
| "loss": 0.8048, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.608061612322465e-05, | |
| "loss": 0.8544, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.606061212242448e-05, | |
| "loss": 0.7827, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.604060812162433e-05, | |
| "loss": 0.7641, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.602060412082416e-05, | |
| "loss": 0.7146, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 0.7286937832832336, | |
| "eval_runtime": 689.1627, | |
| "eval_samples_per_second": 7.217, | |
| "eval_steps_per_second": 0.903, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.6000600120024e-05, | |
| "loss": 0.7801, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.598059611922385e-05, | |
| "loss": 0.7803, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.596059211842369e-05, | |
| "loss": 0.7981, | |
| "step": 20300 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.594058811762354e-05, | |
| "loss": 0.755, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.592058411682337e-05, | |
| "loss": 0.7743, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.590098019603922e-05, | |
| "loss": 0.7358, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.588097619523905e-05, | |
| "loss": 0.7851, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.586097219443889e-05, | |
| "loss": 0.7746, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.584096819363874e-05, | |
| "loss": 0.8246, | |
| "step": 20900 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.582096419283857e-05, | |
| "loss": 0.7833, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.580096019203841e-05, | |
| "loss": 0.7666, | |
| "step": 21100 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.578095619123824e-05, | |
| "loss": 0.8575, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.576095219043809e-05, | |
| "loss": 0.7934, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.574094818963793e-05, | |
| "loss": 0.7473, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.572094418883776e-05, | |
| "loss": 0.7455, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.570094018803761e-05, | |
| "loss": 0.7527, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.568093618723745e-05, | |
| "loss": 0.7906, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.56609321864373e-05, | |
| "loss": 0.8384, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.564092818563714e-05, | |
| "loss": 0.7897, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.562092418483697e-05, | |
| "loss": 0.7459, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.560092018403682e-05, | |
| "loss": 0.7838, | |
| "step": 22100 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.558091618323666e-05, | |
| "loss": 0.7745, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.556091218243649e-05, | |
| "loss": 0.7985, | |
| "step": 22300 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.554090818163634e-05, | |
| "loss": 0.7945, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.552090418083617e-05, | |
| "loss": 0.7985, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.550090018003601e-05, | |
| "loss": 0.7785, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.548089617923586e-05, | |
| "loss": 0.7642, | |
| "step": 22700 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.546089217843569e-05, | |
| "loss": 0.7953, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.544088817763553e-05, | |
| "loss": 0.7848, | |
| "step": 22900 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.542088417683536e-05, | |
| "loss": 0.7388, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.540088017603521e-05, | |
| "loss": 0.7947, | |
| "step": 23100 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.538087617523505e-05, | |
| "loss": 0.7464, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.53608721744349e-05, | |
| "loss": 0.7696, | |
| "step": 23300 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.534086817363474e-05, | |
| "loss": 0.7563, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.532086417283457e-05, | |
| "loss": 0.7522, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.530086017203442e-05, | |
| "loss": 0.772, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.528085617123426e-05, | |
| "loss": 0.7654, | |
| "step": 23700 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.526085217043409e-05, | |
| "loss": 0.759, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.524084816963394e-05, | |
| "loss": 0.8032, | |
| "step": 23900 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.522084416883377e-05, | |
| "loss": 0.7638, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.520084016803361e-05, | |
| "loss": 0.7974, | |
| "step": 24100 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.518083616723346e-05, | |
| "loss": 0.7647, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.516083216643329e-05, | |
| "loss": 0.8016, | |
| "step": 24300 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.514082816563313e-05, | |
| "loss": 0.7749, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.512082416483296e-05, | |
| "loss": 0.7573, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.51008201640328e-05, | |
| "loss": 0.7476, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.508081616323265e-05, | |
| "loss": 0.7848, | |
| "step": 24700 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.50608121624325e-05, | |
| "loss": 0.7994, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.504100820164034e-05, | |
| "loss": 0.8056, | |
| "step": 24900 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.502100420084017e-05, | |
| "loss": 0.8066, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.500100020004001e-05, | |
| "loss": 0.7506, | |
| "step": 25100 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.498099619923986e-05, | |
| "loss": 0.8018, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.496099219843969e-05, | |
| "loss": 0.7451, | |
| "step": 25300 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.494098819763953e-05, | |
| "loss": 0.764, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.492098419683938e-05, | |
| "loss": 0.8212, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.490098019603922e-05, | |
| "loss": 0.7954, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.488097619523905e-05, | |
| "loss": 0.7965, | |
| "step": 25700 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.48609721944389e-05, | |
| "loss": 0.7795, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.484096819363874e-05, | |
| "loss": 0.7993, | |
| "step": 25900 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.482096419283857e-05, | |
| "loss": 0.7951, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.480096019203842e-05, | |
| "loss": 0.7432, | |
| "step": 26100 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.478095619123826e-05, | |
| "loss": 0.7844, | |
| "step": 26200 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.476095219043809e-05, | |
| "loss": 0.7029, | |
| "step": 26300 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.474094818963793e-05, | |
| "loss": 0.7613, | |
| "step": 26400 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.472094418883777e-05, | |
| "loss": 0.7896, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.470094018803761e-05, | |
| "loss": 0.7674, | |
| "step": 26600 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.468093618723745e-05, | |
| "loss": 0.8013, | |
| "step": 26700 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.466093218643729e-05, | |
| "loss": 0.811, | |
| "step": 26800 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.464092818563713e-05, | |
| "loss": 0.8235, | |
| "step": 26900 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.462092418483696e-05, | |
| "loss": 0.7717, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.46009201840368e-05, | |
| "loss": 0.7364, | |
| "step": 27100 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.458111622324465e-05, | |
| "loss": 0.7663, | |
| "step": 27200 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.456111222244449e-05, | |
| "loss": 0.771, | |
| "step": 27300 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.454110822164433e-05, | |
| "loss": 0.7583, | |
| "step": 27400 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.452110422084417e-05, | |
| "loss": 0.7672, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.450110022004401e-05, | |
| "loss": 0.7706, | |
| "step": 27600 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.448109621924385e-05, | |
| "loss": 0.7676, | |
| "step": 27700 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.446109221844369e-05, | |
| "loss": 0.7859, | |
| "step": 27800 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.444108821764353e-05, | |
| "loss": 0.7573, | |
| "step": 27900 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.442108421684337e-05, | |
| "loss": 0.8289, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.440108021604322e-05, | |
| "loss": 0.7757, | |
| "step": 28100 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.438107621524306e-05, | |
| "loss": 0.7688, | |
| "step": 28200 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.43610722144429e-05, | |
| "loss": 0.7941, | |
| "step": 28300 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.434106821364274e-05, | |
| "loss": 0.7829, | |
| "step": 28400 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.432106421284257e-05, | |
| "loss": 0.7744, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.430106021204241e-05, | |
| "loss": 0.805, | |
| "step": 28600 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.428105621124226e-05, | |
| "loss": 0.776, | |
| "step": 28700 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.426105221044209e-05, | |
| "loss": 0.7839, | |
| "step": 28800 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.424104820964193e-05, | |
| "loss": 0.7894, | |
| "step": 28900 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.422104420884177e-05, | |
| "loss": 0.8138, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.420104020804161e-05, | |
| "loss": 0.7834, | |
| "step": 29100 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.418103620724145e-05, | |
| "loss": 0.7323, | |
| "step": 29200 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.416103220644129e-05, | |
| "loss": 0.79, | |
| "step": 29300 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.414102820564113e-05, | |
| "loss": 0.773, | |
| "step": 29400 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.412102420484097e-05, | |
| "loss": 0.7607, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.410102020404082e-05, | |
| "loss": 0.7581, | |
| "step": 29600 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.408101620324066e-05, | |
| "loss": 0.7528, | |
| "step": 29700 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.40610122024405e-05, | |
| "loss": 0.7755, | |
| "step": 29800 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.404100820164034e-05, | |
| "loss": 0.7633, | |
| "step": 29900 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.402100420084017e-05, | |
| "loss": 0.7352, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_loss": 0.7202807068824768, | |
| "eval_runtime": 707.6198, | |
| "eval_samples_per_second": 7.029, | |
| "eval_steps_per_second": 0.879, | |
| "step": 30000 | |
| } | |
| ], | |
| "max_steps": 500000, | |
| "num_train_epochs": 10, | |
| "total_flos": 4.209735834528645e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |