| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 1.0, |
| "eval_steps": 500, |
| "global_step": 1465, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0006825938566552901, |
| "grad_norm": 0.5078125, |
| "learning_rate": 0.0, |
| "loss": 0.8745, |
| "num_tokens": 407394.0, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.0013651877133105802, |
| "grad_norm": 0.5234375, |
| "learning_rate": 6.756756756756758e-07, |
| "loss": 0.8532, |
| "num_tokens": 782681.0, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.0020477815699658703, |
| "grad_norm": 0.51953125, |
| "learning_rate": 1.3513513513513515e-06, |
| "loss": 0.9001, |
| "num_tokens": 1168734.0, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.0027303754266211604, |
| "grad_norm": 0.50390625, |
| "learning_rate": 2.0270270270270273e-06, |
| "loss": 0.8964, |
| "num_tokens": 1580346.0, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.0034129692832764505, |
| "grad_norm": 0.5390625, |
| "learning_rate": 2.702702702702703e-06, |
| "loss": 0.8467, |
| "num_tokens": 1934239.0, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.004095563139931741, |
| "grad_norm": 0.50390625, |
| "learning_rate": 3.3783783783783788e-06, |
| "loss": 0.8491, |
| "num_tokens": 2316765.0, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.00477815699658703, |
| "grad_norm": 0.55078125, |
| "learning_rate": 4.0540540540540545e-06, |
| "loss": 0.8391, |
| "num_tokens": 2636133.0, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.005460750853242321, |
| "grad_norm": 0.5078125, |
| "learning_rate": 4.72972972972973e-06, |
| "loss": 0.8519, |
| "num_tokens": 2964392.0, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.0061433447098976105, |
| "grad_norm": 0.4375, |
| "learning_rate": 5.405405405405406e-06, |
| "loss": 0.8597, |
| "num_tokens": 3349382.0, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.006825938566552901, |
| "grad_norm": 0.419921875, |
| "learning_rate": 6.081081081081082e-06, |
| "loss": 0.826, |
| "num_tokens": 3759778.0, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.007508532423208191, |
| "grad_norm": 0.43359375, |
| "learning_rate": 6.7567567567567575e-06, |
| "loss": 0.8169, |
| "num_tokens": 4119269.0, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.008191126279863481, |
| "grad_norm": 0.404296875, |
| "learning_rate": 7.432432432432433e-06, |
| "loss": 0.9005, |
| "num_tokens": 4534024.0, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.008873720136518772, |
| "grad_norm": 0.4140625, |
| "learning_rate": 8.108108108108109e-06, |
| "loss": 0.8424, |
| "num_tokens": 4896346.0, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.00955631399317406, |
| "grad_norm": 0.40625, |
| "learning_rate": 8.783783783783785e-06, |
| "loss": 0.8994, |
| "num_tokens": 5260048.0, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.010238907849829351, |
| "grad_norm": 0.37890625, |
| "learning_rate": 9.45945945945946e-06, |
| "loss": 0.8602, |
| "num_tokens": 5645472.0, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.010921501706484642, |
| "grad_norm": 0.33984375, |
| "learning_rate": 1.0135135135135136e-05, |
| "loss": 0.8641, |
| "num_tokens": 6059998.0, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.011604095563139932, |
| "grad_norm": 0.3046875, |
| "learning_rate": 1.0810810810810812e-05, |
| "loss": 0.8391, |
| "num_tokens": 6481157.0, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.012286689419795221, |
| "grad_norm": 0.298828125, |
| "learning_rate": 1.1486486486486488e-05, |
| "loss": 0.8549, |
| "num_tokens": 6894460.0, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.012969283276450512, |
| "grad_norm": 0.310546875, |
| "learning_rate": 1.2162162162162164e-05, |
| "loss": 0.838, |
| "num_tokens": 7260818.0, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.013651877133105802, |
| "grad_norm": 0.283203125, |
| "learning_rate": 1.2837837837837838e-05, |
| "loss": 0.8383, |
| "num_tokens": 7670069.0, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.014334470989761093, |
| "grad_norm": 0.287109375, |
| "learning_rate": 1.3513513513513515e-05, |
| "loss": 0.8603, |
| "num_tokens": 8029113.0, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.015017064846416382, |
| "grad_norm": 0.28125, |
| "learning_rate": 1.4189189189189189e-05, |
| "loss": 0.8239, |
| "num_tokens": 8384714.0, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.015699658703071672, |
| "grad_norm": 0.2578125, |
| "learning_rate": 1.4864864864864867e-05, |
| "loss": 0.8397, |
| "num_tokens": 8760277.0, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.016382252559726963, |
| "grad_norm": 0.25390625, |
| "learning_rate": 1.554054054054054e-05, |
| "loss": 0.8326, |
| "num_tokens": 9132115.0, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.017064846416382253, |
| "grad_norm": 0.2421875, |
| "learning_rate": 1.6216216216216218e-05, |
| "loss": 0.8228, |
| "num_tokens": 9514590.0, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.017747440273037544, |
| "grad_norm": 0.228515625, |
| "learning_rate": 1.6891891891891892e-05, |
| "loss": 0.8118, |
| "num_tokens": 9880881.0, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.018430034129692834, |
| "grad_norm": 0.2255859375, |
| "learning_rate": 1.756756756756757e-05, |
| "loss": 0.8132, |
| "num_tokens": 10245704.0, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.01911262798634812, |
| "grad_norm": 0.2216796875, |
| "learning_rate": 1.8243243243243244e-05, |
| "loss": 0.8125, |
| "num_tokens": 10574251.0, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.019795221843003412, |
| "grad_norm": 0.1943359375, |
| "learning_rate": 1.891891891891892e-05, |
| "loss": 0.7904, |
| "num_tokens": 10975160.0, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.020477815699658702, |
| "grad_norm": 0.1865234375, |
| "learning_rate": 1.9594594594594595e-05, |
| "loss": 0.7817, |
| "num_tokens": 11370767.0, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.021160409556313993, |
| "grad_norm": 0.1953125, |
| "learning_rate": 2.0270270270270273e-05, |
| "loss": 0.8256, |
| "num_tokens": 11701894.0, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.021843003412969283, |
| "grad_norm": 0.19140625, |
| "learning_rate": 2.0945945945945947e-05, |
| "loss": 0.789, |
| "num_tokens": 12017501.0, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.022525597269624574, |
| "grad_norm": 0.1689453125, |
| "learning_rate": 2.1621621621621624e-05, |
| "loss": 0.8203, |
| "num_tokens": 12402292.0, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.023208191126279865, |
| "grad_norm": 0.16015625, |
| "learning_rate": 2.2297297297297298e-05, |
| "loss": 0.8185, |
| "num_tokens": 12855236.0, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.023890784982935155, |
| "grad_norm": 0.16015625, |
| "learning_rate": 2.2972972972972976e-05, |
| "loss": 0.7784, |
| "num_tokens": 13221634.0, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.024573378839590442, |
| "grad_norm": 0.1513671875, |
| "learning_rate": 2.364864864864865e-05, |
| "loss": 0.7785, |
| "num_tokens": 13589043.0, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.025255972696245733, |
| "grad_norm": 0.146484375, |
| "learning_rate": 2.4324324324324327e-05, |
| "loss": 0.7852, |
| "num_tokens": 13946703.0, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.025938566552901023, |
| "grad_norm": 0.1435546875, |
| "learning_rate": 2.5e-05, |
| "loss": 0.765, |
| "num_tokens": 14328885.0, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.026621160409556314, |
| "grad_norm": 0.138671875, |
| "learning_rate": 2.5675675675675675e-05, |
| "loss": 0.7731, |
| "num_tokens": 14730912.0, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.027303754266211604, |
| "grad_norm": 0.142578125, |
| "learning_rate": 2.635135135135135e-05, |
| "loss": 0.7835, |
| "num_tokens": 15120586.0, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.027986348122866895, |
| "grad_norm": 0.1435546875, |
| "learning_rate": 2.702702702702703e-05, |
| "loss": 0.772, |
| "num_tokens": 15472277.0, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.028668941979522185, |
| "grad_norm": 0.13671875, |
| "learning_rate": 2.7702702702702704e-05, |
| "loss": 0.8304, |
| "num_tokens": 15908239.0, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.029351535836177476, |
| "grad_norm": 0.130859375, |
| "learning_rate": 2.8378378378378378e-05, |
| "loss": 0.7694, |
| "num_tokens": 16312333.0, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.030034129692832763, |
| "grad_norm": 0.12451171875, |
| "learning_rate": 2.9054054054054052e-05, |
| "loss": 0.7828, |
| "num_tokens": 16712844.0, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.030716723549488054, |
| "grad_norm": 0.12890625, |
| "learning_rate": 2.9729729729729733e-05, |
| "loss": 0.7395, |
| "num_tokens": 17045024.0, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.031399317406143344, |
| "grad_norm": 0.1201171875, |
| "learning_rate": 3.0405405405405407e-05, |
| "loss": 0.7812, |
| "num_tokens": 17405361.0, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.032081911262798635, |
| "grad_norm": 0.11865234375, |
| "learning_rate": 3.108108108108108e-05, |
| "loss": 0.7626, |
| "num_tokens": 17779922.0, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.032764505119453925, |
| "grad_norm": 0.1162109375, |
| "learning_rate": 3.175675675675676e-05, |
| "loss": 0.7832, |
| "num_tokens": 18176343.0, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.033447098976109216, |
| "grad_norm": 0.111328125, |
| "learning_rate": 3.2432432432432436e-05, |
| "loss": 0.7527, |
| "num_tokens": 18572324.0, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.034129692832764506, |
| "grad_norm": 0.1162109375, |
| "learning_rate": 3.310810810810811e-05, |
| "loss": 0.7253, |
| "num_tokens": 18909989.0, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.0348122866894198, |
| "grad_norm": 0.10986328125, |
| "learning_rate": 3.3783783783783784e-05, |
| "loss": 0.7465, |
| "num_tokens": 19317339.0, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.03549488054607509, |
| "grad_norm": 0.1142578125, |
| "learning_rate": 3.445945945945946e-05, |
| "loss": 0.7881, |
| "num_tokens": 19661637.0, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.03617747440273038, |
| "grad_norm": 0.1064453125, |
| "learning_rate": 3.513513513513514e-05, |
| "loss": 0.7507, |
| "num_tokens": 20081469.0, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.03686006825938567, |
| "grad_norm": 0.1083984375, |
| "learning_rate": 3.581081081081081e-05, |
| "loss": 0.7288, |
| "num_tokens": 20450059.0, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.03754266211604096, |
| "grad_norm": 0.1025390625, |
| "learning_rate": 3.648648648648649e-05, |
| "loss": 0.7096, |
| "num_tokens": 20794568.0, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.03822525597269624, |
| "grad_norm": 0.1025390625, |
| "learning_rate": 3.7162162162162165e-05, |
| "loss": 0.7353, |
| "num_tokens": 21171278.0, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.03890784982935153, |
| "grad_norm": 0.11083984375, |
| "learning_rate": 3.783783783783784e-05, |
| "loss": 0.7584, |
| "num_tokens": 21502503.0, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.039590443686006824, |
| "grad_norm": 0.10205078125, |
| "learning_rate": 3.851351351351351e-05, |
| "loss": 0.731, |
| "num_tokens": 21905681.0, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.040273037542662114, |
| "grad_norm": 0.10595703125, |
| "learning_rate": 3.918918918918919e-05, |
| "loss": 0.8025, |
| "num_tokens": 22279466.0, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.040955631399317405, |
| "grad_norm": 0.1015625, |
| "learning_rate": 3.986486486486487e-05, |
| "loss": 0.7227, |
| "num_tokens": 22645536.0, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.041638225255972695, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 4.0540540540540545e-05, |
| "loss": 0.724, |
| "num_tokens": 23069580.0, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.042320819112627986, |
| "grad_norm": 0.10302734375, |
| "learning_rate": 4.1216216216216216e-05, |
| "loss": 0.7279, |
| "num_tokens": 23393055.0, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.043003412969283276, |
| "grad_norm": 0.095703125, |
| "learning_rate": 4.189189189189189e-05, |
| "loss": 0.7787, |
| "num_tokens": 23853519.0, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.04368600682593857, |
| "grad_norm": 0.095703125, |
| "learning_rate": 4.256756756756757e-05, |
| "loss": 0.7723, |
| "num_tokens": 24235234.0, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.04436860068259386, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 4.324324324324325e-05, |
| "loss": 0.7345, |
| "num_tokens": 24609009.0, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.04505119453924915, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.391891891891892e-05, |
| "loss": 0.7388, |
| "num_tokens": 25009744.0, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.04573378839590444, |
| "grad_norm": 0.095703125, |
| "learning_rate": 4.4594594594594596e-05, |
| "loss": 0.7306, |
| "num_tokens": 25400295.0, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.04641638225255973, |
| "grad_norm": 0.09375, |
| "learning_rate": 4.5270270270270274e-05, |
| "loss": 0.7625, |
| "num_tokens": 25827145.0, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.04709897610921502, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.594594594594595e-05, |
| "loss": 0.7277, |
| "num_tokens": 26256417.0, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.04778156996587031, |
| "grad_norm": 0.095703125, |
| "learning_rate": 4.662162162162162e-05, |
| "loss": 0.7285, |
| "num_tokens": 26641035.0, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.048464163822525594, |
| "grad_norm": 0.091796875, |
| "learning_rate": 4.72972972972973e-05, |
| "loss": 0.6822, |
| "num_tokens": 27002707.0, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.049146757679180884, |
| "grad_norm": 0.09814453125, |
| "learning_rate": 4.797297297297298e-05, |
| "loss": 0.7129, |
| "num_tokens": 27348038.0, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.049829351535836175, |
| "grad_norm": 0.10205078125, |
| "learning_rate": 4.8648648648648654e-05, |
| "loss": 0.7233, |
| "num_tokens": 27704845.0, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.050511945392491465, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.9324324324324325e-05, |
| "loss": 0.6843, |
| "num_tokens": 28067434.0, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.051194539249146756, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 5e-05, |
| "loss": 0.7024, |
| "num_tokens": 28445983.0, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.05187713310580205, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.999994261507436e-05, |
| "loss": 0.7298, |
| "num_tokens": 28855065.0, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.05255972696245734, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 4.999977046059014e-05, |
| "loss": 0.6869, |
| "num_tokens": 29260490.0, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.05324232081911263, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 4.9999483537425486e-05, |
| "loss": 0.7325, |
| "num_tokens": 29688420.0, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.05392491467576792, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 4.999908184704396e-05, |
| "loss": 0.7348, |
| "num_tokens": 30043725.0, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.05460750853242321, |
| "grad_norm": 0.09375, |
| "learning_rate": 4.999856539149453e-05, |
| "loss": 0.7227, |
| "num_tokens": 30423007.0, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.0552901023890785, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.9997934173411596e-05, |
| "loss": 0.725, |
| "num_tokens": 30814889.0, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.05597269624573379, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.9997188196014907e-05, |
| "loss": 0.7081, |
| "num_tokens": 31180315.0, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.05665529010238908, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.999632746310961e-05, |
| "loss": 0.6569, |
| "num_tokens": 31550238.0, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.05733788395904437, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.999535197908619e-05, |
| "loss": 0.7457, |
| "num_tokens": 31947817.0, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.05802047781569966, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.99942617489205e-05, |
| "loss": 0.7383, |
| "num_tokens": 32356676.0, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.05870307167235495, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 4.999305677817367e-05, |
| "loss": 0.7201, |
| "num_tokens": 32751234.0, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.059385665529010236, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.999173707299211e-05, |
| "loss": 0.7204, |
| "num_tokens": 33100578.0, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.060068259385665526, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 4.999030264010747e-05, |
| "loss": 0.729, |
| "num_tokens": 33495983.0, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.06075085324232082, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.9988753486836635e-05, |
| "loss": 0.7198, |
| "num_tokens": 33872217.0, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.06143344709897611, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.998708962108165e-05, |
| "loss": 0.6787, |
| "num_tokens": 34237001.0, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.0621160409556314, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 4.998531105132971e-05, |
| "loss": 0.7097, |
| "num_tokens": 34608500.0, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.06279863481228669, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.998341778665308e-05, |
| "loss": 0.7188, |
| "num_tokens": 34966319.0, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.06348122866894199, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.998140983670907e-05, |
| "loss": 0.7124, |
| "num_tokens": 35353782.0, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.06416382252559727, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.9979287211740004e-05, |
| "loss": 0.7217, |
| "num_tokens": 35699847.0, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.06484641638225255, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 4.9977049922573155e-05, |
| "loss": 0.6902, |
| "num_tokens": 36074352.0, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.06552901023890785, |
| "grad_norm": 0.1005859375, |
| "learning_rate": 4.9974697980620654e-05, |
| "loss": 0.7434, |
| "num_tokens": 36434152.0, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.06621160409556313, |
| "grad_norm": 0.0859375, |
| "learning_rate": 4.997223139787949e-05, |
| "loss": 0.7608, |
| "num_tokens": 36841329.0, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.06689419795221843, |
| "grad_norm": 0.087890625, |
| "learning_rate": 4.996965018693142e-05, |
| "loss": 0.7327, |
| "num_tokens": 37239299.0, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.06757679180887372, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.996695436094287e-05, |
| "loss": 0.6706, |
| "num_tokens": 37591054.0, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.06825938566552901, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.9964143933664945e-05, |
| "loss": 0.7028, |
| "num_tokens": 37967720.0, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.0689419795221843, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 4.996121891943332e-05, |
| "loss": 0.7295, |
| "num_tokens": 38363954.0, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.0696245733788396, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.995817933316812e-05, |
| "loss": 0.753, |
| "num_tokens": 38743158.0, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.07030716723549488, |
| "grad_norm": 0.091796875, |
| "learning_rate": 4.9955025190373935e-05, |
| "loss": 0.6928, |
| "num_tokens": 39085101.0, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.07098976109215017, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 4.995175650713968e-05, |
| "loss": 0.694, |
| "num_tokens": 39435486.0, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.07167235494880546, |
| "grad_norm": 0.087890625, |
| "learning_rate": 4.99483733001385e-05, |
| "loss": 0.7347, |
| "num_tokens": 39853351.0, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.07235494880546076, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 4.994487558662776e-05, |
| "loss": 0.7109, |
| "num_tokens": 40232996.0, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.07303754266211604, |
| "grad_norm": 0.091796875, |
| "learning_rate": 4.994126338444887e-05, |
| "loss": 0.6806, |
| "num_tokens": 40612858.0, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.07372013651877134, |
| "grad_norm": 0.087890625, |
| "learning_rate": 4.993753671202726e-05, |
| "loss": 0.7138, |
| "num_tokens": 40977915.0, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.07440273037542662, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.993369558837224e-05, |
| "loss": 0.7438, |
| "num_tokens": 41358972.0, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.07508532423208192, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 4.9929740033076915e-05, |
| "loss": 0.7171, |
| "num_tokens": 41833783.0, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.0757679180887372, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 4.992567006631814e-05, |
| "loss": 0.6927, |
| "num_tokens": 42209618.0, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.07645051194539249, |
| "grad_norm": 0.091796875, |
| "learning_rate": 4.9921485708856305e-05, |
| "loss": 0.697, |
| "num_tokens": 42552299.0, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.07713310580204778, |
| "grad_norm": 0.091796875, |
| "learning_rate": 4.9917186982035354e-05, |
| "loss": 0.7069, |
| "num_tokens": 42906241.0, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.07781569965870307, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.991277390778256e-05, |
| "loss": 0.6857, |
| "num_tokens": 43291373.0, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.07849829351535836, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.9908246508608506e-05, |
| "loss": 0.7539, |
| "num_tokens": 43682290.0, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.07918088737201365, |
| "grad_norm": 0.091796875, |
| "learning_rate": 4.990360480760693e-05, |
| "loss": 0.7055, |
| "num_tokens": 44019445.0, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.07986348122866894, |
| "grad_norm": 0.0859375, |
| "learning_rate": 4.989884882845459e-05, |
| "loss": 0.7112, |
| "num_tokens": 44440480.0, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.08054607508532423, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 4.989397859541118e-05, |
| "loss": 0.6937, |
| "num_tokens": 44772682.0, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.08122866894197953, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 4.9888994133319184e-05, |
| "loss": 0.6495, |
| "num_tokens": 45158854.0, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.08191126279863481, |
| "grad_norm": 0.087890625, |
| "learning_rate": 4.9883895467603764e-05, |
| "loss": 0.7414, |
| "num_tokens": 45550959.0, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.08259385665529011, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 4.98786826242726e-05, |
| "loss": 0.7404, |
| "num_tokens": 45886449.0, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.08327645051194539, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.987335562991581e-05, |
| "loss": 0.6948, |
| "num_tokens": 46250221.0, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.08395904436860069, |
| "grad_norm": 0.08349609375, |
| "learning_rate": 4.9867914511705745e-05, |
| "loss": 0.704, |
| "num_tokens": 46633079.0, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.08464163822525597, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.986235929739692e-05, |
| "loss": 0.6996, |
| "num_tokens": 47060774.0, |
| "step": 124 |
| }, |
| { |
| "epoch": 0.08532423208191127, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 4.985669001532583e-05, |
| "loss": 0.7328, |
| "num_tokens": 47474526.0, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.08600682593856655, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 4.985090669441081e-05, |
| "loss": 0.7302, |
| "num_tokens": 47866053.0, |
| "step": 126 |
| }, |
| { |
| "epoch": 0.08668941979522184, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 4.984500936415191e-05, |
| "loss": 0.6713, |
| "num_tokens": 48229683.0, |
| "step": 127 |
| }, |
| { |
| "epoch": 0.08737201365187713, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.98389980546307e-05, |
| "loss": 0.7107, |
| "num_tokens": 48610658.0, |
| "step": 128 |
| }, |
| { |
| "epoch": 0.08805460750853242, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.983287279651019e-05, |
| "loss": 0.6758, |
| "num_tokens": 48973153.0, |
| "step": 129 |
| }, |
| { |
| "epoch": 0.08873720136518772, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 4.982663362103456e-05, |
| "loss": 0.729, |
| "num_tokens": 49377672.0, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.089419795221843, |
| "grad_norm": 0.09814453125, |
| "learning_rate": 4.9820280560029155e-05, |
| "loss": 0.7365, |
| "num_tokens": 49764225.0, |
| "step": 131 |
| }, |
| { |
| "epoch": 0.0901023890784983, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.981381364590015e-05, |
| "loss": 0.6672, |
| "num_tokens": 50106626.0, |
| "step": 132 |
| }, |
| { |
| "epoch": 0.09078498293515358, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 4.980723291163454e-05, |
| "loss": 0.679, |
| "num_tokens": 50475699.0, |
| "step": 133 |
| }, |
| { |
| "epoch": 0.09146757679180888, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 4.980053839079986e-05, |
| "loss": 0.6938, |
| "num_tokens": 50819408.0, |
| "step": 134 |
| }, |
| { |
| "epoch": 0.09215017064846416, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 4.979373011754409e-05, |
| "loss": 0.7192, |
| "num_tokens": 51201531.0, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.09283276450511946, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 4.9786808126595414e-05, |
| "loss": 0.7014, |
| "num_tokens": 51601184.0, |
| "step": 136 |
| }, |
| { |
| "epoch": 0.09351535836177474, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.97797724532621e-05, |
| "loss": 0.6848, |
| "num_tokens": 51974261.0, |
| "step": 137 |
| }, |
| { |
| "epoch": 0.09419795221843004, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 4.977262313343228e-05, |
| "loss": 0.6897, |
| "num_tokens": 52287696.0, |
| "step": 138 |
| }, |
| { |
| "epoch": 0.09488054607508532, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.976536020357381e-05, |
| "loss": 0.7072, |
| "num_tokens": 52652681.0, |
| "step": 139 |
| }, |
| { |
| "epoch": 0.09556313993174062, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.975798370073404e-05, |
| "loss": 0.7115, |
| "num_tokens": 53019691.0, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.0962457337883959, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.975049366253962e-05, |
| "loss": 0.6631, |
| "num_tokens": 53400025.0, |
| "step": 141 |
| }, |
| { |
| "epoch": 0.09692832764505119, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 4.974289012719637e-05, |
| "loss": 0.7049, |
| "num_tokens": 53784787.0, |
| "step": 142 |
| }, |
| { |
| "epoch": 0.09761092150170649, |
| "grad_norm": 0.083984375, |
| "learning_rate": 4.9735173133489035e-05, |
| "loss": 0.697, |
| "num_tokens": 54208023.0, |
| "step": 143 |
| }, |
| { |
| "epoch": 0.09829351535836177, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.9727342720781076e-05, |
| "loss": 0.6875, |
| "num_tokens": 54553341.0, |
| "step": 144 |
| }, |
| { |
| "epoch": 0.09897610921501707, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.971939892901451e-05, |
| "loss": 0.7061, |
| "num_tokens": 54928756.0, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.09965870307167235, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 4.9711341798709696e-05, |
| "loss": 0.6605, |
| "num_tokens": 55316921.0, |
| "step": 146 |
| }, |
| { |
| "epoch": 0.10034129692832765, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.970317137096509e-05, |
| "loss": 0.7317, |
| "num_tokens": 55683380.0, |
| "step": 147 |
| }, |
| { |
| "epoch": 0.10102389078498293, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.969488768745709e-05, |
| "loss": 0.7348, |
| "num_tokens": 56088690.0, |
| "step": 148 |
| }, |
| { |
| "epoch": 0.10170648464163823, |
| "grad_norm": 0.083984375, |
| "learning_rate": 4.9686490790439795e-05, |
| "loss": 0.7091, |
| "num_tokens": 56519495.0, |
| "step": 149 |
| }, |
| { |
| "epoch": 0.10238907849829351, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.9677980722744774e-05, |
| "loss": 0.7269, |
| "num_tokens": 56919336.0, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.10307167235494881, |
| "grad_norm": 0.0859375, |
| "learning_rate": 4.96693575277809e-05, |
| "loss": 0.666, |
| "num_tokens": 57293646.0, |
| "step": 151 |
| }, |
| { |
| "epoch": 0.1037542662116041, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.9660621249534064e-05, |
| "loss": 0.6722, |
| "num_tokens": 57666806.0, |
| "step": 152 |
| }, |
| { |
| "epoch": 0.10443686006825939, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.965177193256699e-05, |
| "loss": 0.6865, |
| "num_tokens": 58065270.0, |
| "step": 153 |
| }, |
| { |
| "epoch": 0.10511945392491467, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.964280962201901e-05, |
| "loss": 0.6837, |
| "num_tokens": 58443664.0, |
| "step": 154 |
| }, |
| { |
| "epoch": 0.10580204778156997, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.9633734363605805e-05, |
| "loss": 0.6955, |
| "num_tokens": 58769487.0, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.10648464163822526, |
| "grad_norm": 0.083984375, |
| "learning_rate": 4.962454620361921e-05, |
| "loss": 0.6963, |
| "num_tokens": 59157201.0, |
| "step": 156 |
| }, |
| { |
| "epoch": 0.10716723549488055, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 4.9615245188926954e-05, |
| "loss": 0.6577, |
| "num_tokens": 59509606.0, |
| "step": 157 |
| }, |
| { |
| "epoch": 0.10784982935153584, |
| "grad_norm": 0.08349609375, |
| "learning_rate": 4.9605831366972386e-05, |
| "loss": 0.6797, |
| "num_tokens": 59898621.0, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.10853242320819112, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 4.959630478577435e-05, |
| "loss": 0.7205, |
| "num_tokens": 60297079.0, |
| "step": 159 |
| }, |
| { |
| "epoch": 0.10921501706484642, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 4.9586665493926773e-05, |
| "loss": 0.6875, |
| "num_tokens": 60660649.0, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.1098976109215017, |
| "grad_norm": 0.09375, |
| "learning_rate": 4.9576913540598563e-05, |
| "loss": 0.67, |
| "num_tokens": 61008988.0, |
| "step": 161 |
| }, |
| { |
| "epoch": 0.110580204778157, |
| "grad_norm": 0.08251953125, |
| "learning_rate": 4.9567048975533307e-05, |
| "loss": 0.6721, |
| "num_tokens": 61428893.0, |
| "step": 162 |
| }, |
| { |
| "epoch": 0.11126279863481228, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 4.9557071849048954e-05, |
| "loss": 0.6662, |
| "num_tokens": 61839146.0, |
| "step": 163 |
| }, |
| { |
| "epoch": 0.11194539249146758, |
| "grad_norm": 0.08349609375, |
| "learning_rate": 4.954698221203767e-05, |
| "loss": 0.6697, |
| "num_tokens": 62289847.0, |
| "step": 164 |
| }, |
| { |
| "epoch": 0.11262798634812286, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.9536780115965514e-05, |
| "loss": 0.6725, |
| "num_tokens": 62673288.0, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.11331058020477816, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 4.9526465612872165e-05, |
| "loss": 0.6942, |
| "num_tokens": 63099694.0, |
| "step": 166 |
| }, |
| { |
| "epoch": 0.11399317406143344, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 4.951603875537069e-05, |
| "loss": 0.6841, |
| "num_tokens": 63476469.0, |
| "step": 167 |
| }, |
| { |
| "epoch": 0.11467576791808874, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.950549959664727e-05, |
| "loss": 0.6537, |
| "num_tokens": 63828724.0, |
| "step": 168 |
| }, |
| { |
| "epoch": 0.11535836177474403, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 4.94948481904609e-05, |
| "loss": 0.6884, |
| "num_tokens": 64211499.0, |
| "step": 169 |
| }, |
| { |
| "epoch": 0.11604095563139932, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.9484084591143154e-05, |
| "loss": 0.698, |
| "num_tokens": 64575366.0, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.1167235494880546, |
| "grad_norm": 0.083984375, |
| "learning_rate": 4.9473208853597894e-05, |
| "loss": 0.6742, |
| "num_tokens": 64966868.0, |
| "step": 171 |
| }, |
| { |
| "epoch": 0.1174061433447099, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 4.946222103330096e-05, |
| "loss": 0.6875, |
| "num_tokens": 65367333.0, |
| "step": 172 |
| }, |
| { |
| "epoch": 0.11808873720136519, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.9451121186299946e-05, |
| "loss": 0.7029, |
| "num_tokens": 65736026.0, |
| "step": 173 |
| }, |
| { |
| "epoch": 0.11877133105802047, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.9439909369213854e-05, |
| "loss": 0.7062, |
| "num_tokens": 66102656.0, |
| "step": 174 |
| }, |
| { |
| "epoch": 0.11945392491467577, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 4.942858563923285e-05, |
| "loss": 0.6795, |
| "num_tokens": 66496209.0, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.12013651877133105, |
| "grad_norm": 0.095703125, |
| "learning_rate": 4.941715005411795e-05, |
| "loss": 0.7091, |
| "num_tokens": 66851018.0, |
| "step": 176 |
| }, |
| { |
| "epoch": 0.12081911262798635, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.940560267220072e-05, |
| "loss": 0.7153, |
| "num_tokens": 67280438.0, |
| "step": 177 |
| }, |
| { |
| "epoch": 0.12150170648464163, |
| "grad_norm": 0.087890625, |
| "learning_rate": 4.9393943552382995e-05, |
| "loss": 0.719, |
| "num_tokens": 67668386.0, |
| "step": 178 |
| }, |
| { |
| "epoch": 0.12218430034129693, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.9382172754136574e-05, |
| "loss": 0.7129, |
| "num_tokens": 68062992.0, |
| "step": 179 |
| }, |
| { |
| "epoch": 0.12286689419795221, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 4.9370290337502924e-05, |
| "loss": 0.672, |
| "num_tokens": 68438217.0, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.12354948805460751, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 4.935829636309283e-05, |
| "loss": 0.7262, |
| "num_tokens": 68847142.0, |
| "step": 181 |
| }, |
| { |
| "epoch": 0.1242320819112628, |
| "grad_norm": 0.08349609375, |
| "learning_rate": 4.9346190892086174e-05, |
| "loss": 0.6441, |
| "num_tokens": 69235431.0, |
| "step": 182 |
| }, |
| { |
| "epoch": 0.12491467576791809, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 4.933397398623152e-05, |
| "loss": 0.6972, |
| "num_tokens": 69633179.0, |
| "step": 183 |
| }, |
| { |
| "epoch": 0.12559726962457338, |
| "grad_norm": 0.095703125, |
| "learning_rate": 4.9321645707845866e-05, |
| "loss": 0.6941, |
| "num_tokens": 70003209.0, |
| "step": 184 |
| }, |
| { |
| "epoch": 0.12627986348122866, |
| "grad_norm": 0.091796875, |
| "learning_rate": 4.930920611981431e-05, |
| "loss": 0.6989, |
| "num_tokens": 70344364.0, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.12696245733788397, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 4.9296655285589723e-05, |
| "loss": 0.6941, |
| "num_tokens": 70746075.0, |
| "step": 186 |
| }, |
| { |
| "epoch": 0.12764505119453926, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.928399326919245e-05, |
| "loss": 0.715, |
| "num_tokens": 71119582.0, |
| "step": 187 |
| }, |
| { |
| "epoch": 0.12832764505119454, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.927122013520992e-05, |
| "loss": 0.7195, |
| "num_tokens": 71485426.0, |
| "step": 188 |
| }, |
| { |
| "epoch": 0.12901023890784982, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.925833594879641e-05, |
| "loss": 0.6734, |
| "num_tokens": 71881005.0, |
| "step": 189 |
| }, |
| { |
| "epoch": 0.1296928327645051, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 4.9245340775672634e-05, |
| "loss": 0.6575, |
| "num_tokens": 72264786.0, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.13037542662116042, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 4.923223468212542e-05, |
| "loss": 0.6558, |
| "num_tokens": 72673955.0, |
| "step": 191 |
| }, |
| { |
| "epoch": 0.1310580204778157, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.921901773500744e-05, |
| "loss": 0.6518, |
| "num_tokens": 73042938.0, |
| "step": 192 |
| }, |
| { |
| "epoch": 0.13174061433447098, |
| "grad_norm": 0.0830078125, |
| "learning_rate": 4.920569000173675e-05, |
| "loss": 0.6865, |
| "num_tokens": 73457127.0, |
| "step": 193 |
| }, |
| { |
| "epoch": 0.13242320819112627, |
| "grad_norm": 0.0859375, |
| "learning_rate": 4.919225155029657e-05, |
| "loss": 0.6765, |
| "num_tokens": 73892838.0, |
| "step": 194 |
| }, |
| { |
| "epoch": 0.13310580204778158, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.917870244923486e-05, |
| "loss": 0.6466, |
| "num_tokens": 74233642.0, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.13378839590443686, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.916504276766399e-05, |
| "loss": 0.6856, |
| "num_tokens": 74640449.0, |
| "step": 196 |
| }, |
| { |
| "epoch": 0.13447098976109215, |
| "grad_norm": 0.08349609375, |
| "learning_rate": 4.915127257526037e-05, |
| "loss": 0.69, |
| "num_tokens": 75037788.0, |
| "step": 197 |
| }, |
| { |
| "epoch": 0.13515358361774743, |
| "grad_norm": 0.095703125, |
| "learning_rate": 4.9137391942264154e-05, |
| "loss": 0.7131, |
| "num_tokens": 75406305.0, |
| "step": 198 |
| }, |
| { |
| "epoch": 0.13583617747440274, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 4.91234009394788e-05, |
| "loss": 0.7173, |
| "num_tokens": 75791398.0, |
| "step": 199 |
| }, |
| { |
| "epoch": 0.13651877133105803, |
| "grad_norm": 0.0859375, |
| "learning_rate": 4.910929963827078e-05, |
| "loss": 0.6772, |
| "num_tokens": 76173511.0, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.1372013651877133, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.909508811056915e-05, |
| "loss": 0.655, |
| "num_tokens": 76579919.0, |
| "step": 201 |
| }, |
| { |
| "epoch": 0.1378839590443686, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.908076642886526e-05, |
| "loss": 0.6901, |
| "num_tokens": 76957202.0, |
| "step": 202 |
| }, |
| { |
| "epoch": 0.1385665529010239, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.906633466621232e-05, |
| "loss": 0.6855, |
| "num_tokens": 77327975.0, |
| "step": 203 |
| }, |
| { |
| "epoch": 0.1392491467576792, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 4.905179289622504e-05, |
| "loss": 0.7126, |
| "num_tokens": 77729665.0, |
| "step": 204 |
| }, |
| { |
| "epoch": 0.13993174061433447, |
| "grad_norm": 0.08203125, |
| "learning_rate": 4.903714119307929e-05, |
| "loss": 0.6657, |
| "num_tokens": 78183541.0, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.14061433447098975, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 4.902237963151168e-05, |
| "loss": 0.6999, |
| "num_tokens": 78580930.0, |
| "step": 206 |
| }, |
| { |
| "epoch": 0.14129692832764504, |
| "grad_norm": 0.087890625, |
| "learning_rate": 4.900750828681919e-05, |
| "loss": 0.6876, |
| "num_tokens": 78967091.0, |
| "step": 207 |
| }, |
| { |
| "epoch": 0.14197952218430035, |
| "grad_norm": 0.0859375, |
| "learning_rate": 4.899252723485881e-05, |
| "loss": 0.6708, |
| "num_tokens": 79372643.0, |
| "step": 208 |
| }, |
| { |
| "epoch": 0.14266211604095563, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 4.8977436552047104e-05, |
| "loss": 0.6766, |
| "num_tokens": 79768505.0, |
| "step": 209 |
| }, |
| { |
| "epoch": 0.14334470989761092, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 4.8962236315359896e-05, |
| "loss": 0.6739, |
| "num_tokens": 80166162.0, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.1440273037542662, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 4.894692660233178e-05, |
| "loss": 0.6561, |
| "num_tokens": 80521430.0, |
| "step": 211 |
| }, |
| { |
| "epoch": 0.1447098976109215, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.8931507491055814e-05, |
| "loss": 0.6784, |
| "num_tokens": 80897702.0, |
| "step": 212 |
| }, |
| { |
| "epoch": 0.1453924914675768, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 4.891597906018306e-05, |
| "loss": 0.6904, |
| "num_tokens": 81295809.0, |
| "step": 213 |
| }, |
| { |
| "epoch": 0.14607508532423208, |
| "grad_norm": 0.09375, |
| "learning_rate": 4.890034138892222e-05, |
| "loss": 0.6867, |
| "num_tokens": 81629870.0, |
| "step": 214 |
| }, |
| { |
| "epoch": 0.14675767918088736, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.8884594557039224e-05, |
| "loss": 0.7011, |
| "num_tokens": 81987283.0, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.14744027303754267, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.886873864485679e-05, |
| "loss": 0.7124, |
| "num_tokens": 82372528.0, |
| "step": 216 |
| }, |
| { |
| "epoch": 0.14812286689419796, |
| "grad_norm": 0.0859375, |
| "learning_rate": 4.885277373325407e-05, |
| "loss": 0.7124, |
| "num_tokens": 82790757.0, |
| "step": 217 |
| }, |
| { |
| "epoch": 0.14880546075085324, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.88366999036662e-05, |
| "loss": 0.6752, |
| "num_tokens": 83163839.0, |
| "step": 218 |
| }, |
| { |
| "epoch": 0.14948805460750852, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 4.8820517238083876e-05, |
| "loss": 0.7047, |
| "num_tokens": 83552435.0, |
| "step": 219 |
| }, |
| { |
| "epoch": 0.15017064846416384, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 4.8804225819052994e-05, |
| "loss": 0.6534, |
| "num_tokens": 83915537.0, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.15085324232081912, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 4.8787825729674156e-05, |
| "loss": 0.6351, |
| "num_tokens": 84253929.0, |
| "step": 221 |
| }, |
| { |
| "epoch": 0.1515358361774744, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 4.877131705360229e-05, |
| "loss": 0.6855, |
| "num_tokens": 84573619.0, |
| "step": 222 |
| }, |
| { |
| "epoch": 0.1522184300341297, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 4.875469987504619e-05, |
| "loss": 0.6447, |
| "num_tokens": 84949885.0, |
| "step": 223 |
| }, |
| { |
| "epoch": 0.15290102389078497, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.873797427876815e-05, |
| "loss": 0.6741, |
| "num_tokens": 85344649.0, |
| "step": 224 |
| }, |
| { |
| "epoch": 0.15358361774744028, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 4.872114035008345e-05, |
| "loss": 0.6674, |
| "num_tokens": 85690047.0, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.15426621160409557, |
| "grad_norm": 0.087890625, |
| "learning_rate": 4.870419817486e-05, |
| "loss": 0.7036, |
| "num_tokens": 86103365.0, |
| "step": 226 |
| }, |
| { |
| "epoch": 0.15494880546075085, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 4.868714783951783e-05, |
| "loss": 0.7408, |
| "num_tokens": 86504191.0, |
| "step": 227 |
| }, |
| { |
| "epoch": 0.15563139931740613, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.866998943102868e-05, |
| "loss": 0.712, |
| "num_tokens": 86911965.0, |
| "step": 228 |
| }, |
| { |
| "epoch": 0.15631399317406144, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.86527230369156e-05, |
| "loss": 0.6902, |
| "num_tokens": 87261787.0, |
| "step": 229 |
| }, |
| { |
| "epoch": 0.15699658703071673, |
| "grad_norm": 0.087890625, |
| "learning_rate": 4.863534874525241e-05, |
| "loss": 0.6923, |
| "num_tokens": 87705484.0, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.157679180887372, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.861786664466334e-05, |
| "loss": 0.6695, |
| "num_tokens": 88059252.0, |
| "step": 231 |
| }, |
| { |
| "epoch": 0.1583617747440273, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 4.8600276824322514e-05, |
| "loss": 0.6767, |
| "num_tokens": 88439748.0, |
| "step": 232 |
| }, |
| { |
| "epoch": 0.1590443686006826, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 4.858257937395355e-05, |
| "loss": 0.6794, |
| "num_tokens": 88866120.0, |
| "step": 233 |
| }, |
| { |
| "epoch": 0.1597269624573379, |
| "grad_norm": 0.087890625, |
| "learning_rate": 4.8564774383829045e-05, |
| "loss": 0.6443, |
| "num_tokens": 89240523.0, |
| "step": 234 |
| }, |
| { |
| "epoch": 0.16040955631399317, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 4.854686194477017e-05, |
| "loss": 0.7049, |
| "num_tokens": 89594618.0, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.16109215017064846, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 4.852884214814616e-05, |
| "loss": 0.6856, |
| "num_tokens": 89988899.0, |
| "step": 236 |
| }, |
| { |
| "epoch": 0.16177474402730374, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 4.8510715085873885e-05, |
| "loss": 0.6673, |
| "num_tokens": 90396165.0, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.16245733788395905, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 4.849248085041734e-05, |
| "loss": 0.633, |
| "num_tokens": 90754788.0, |
| "step": 238 |
| }, |
| { |
| "epoch": 0.16313993174061434, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 4.847413953478723e-05, |
| "loss": 0.7026, |
| "num_tokens": 91167914.0, |
| "step": 239 |
| }, |
| { |
| "epoch": 0.16382252559726962, |
| "grad_norm": 0.08349609375, |
| "learning_rate": 4.845569123254044e-05, |
| "loss": 0.6669, |
| "num_tokens": 91557529.0, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.1645051194539249, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 4.843713603777958e-05, |
| "loss": 0.6695, |
| "num_tokens": 91951842.0, |
| "step": 241 |
| }, |
| { |
| "epoch": 0.16518771331058021, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 4.8418474045152534e-05, |
| "loss": 0.6772, |
| "num_tokens": 92296868.0, |
| "step": 242 |
| }, |
| { |
| "epoch": 0.1658703071672355, |
| "grad_norm": 0.0830078125, |
| "learning_rate": 4.839970534985191e-05, |
| "loss": 0.6992, |
| "num_tokens": 92755075.0, |
| "step": 243 |
| }, |
| { |
| "epoch": 0.16655290102389078, |
| "grad_norm": 0.10986328125, |
| "learning_rate": 4.838083004761462e-05, |
| "loss": 0.669, |
| "num_tokens": 93143891.0, |
| "step": 244 |
| }, |
| { |
| "epoch": 0.16723549488054607, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.836184823472136e-05, |
| "loss": 0.6458, |
| "num_tokens": 93508244.0, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.16791808873720138, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 4.834276000799613e-05, |
| "loss": 0.6597, |
| "num_tokens": 93900770.0, |
| "step": 246 |
| }, |
| { |
| "epoch": 0.16860068259385666, |
| "grad_norm": 0.09375, |
| "learning_rate": 4.832356546480571e-05, |
| "loss": 0.7026, |
| "num_tokens": 94289608.0, |
| "step": 247 |
| }, |
| { |
| "epoch": 0.16928327645051194, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.8304264703059226e-05, |
| "loss": 0.7146, |
| "num_tokens": 94658991.0, |
| "step": 248 |
| }, |
| { |
| "epoch": 0.16996587030716723, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 4.828485782120757e-05, |
| "loss": 0.6746, |
| "num_tokens": 95068405.0, |
| "step": 249 |
| }, |
| { |
| "epoch": 0.17064846416382254, |
| "grad_norm": 0.09716796875, |
| "learning_rate": 4.8265344918242974e-05, |
| "loss": 0.699, |
| "num_tokens": 95453893.0, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.17133105802047782, |
| "grad_norm": 0.09375, |
| "learning_rate": 4.824572609369846e-05, |
| "loss": 0.655, |
| "num_tokens": 95795599.0, |
| "step": 251 |
| }, |
| { |
| "epoch": 0.1720136518771331, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.822600144764734e-05, |
| "loss": 0.6675, |
| "num_tokens": 96137688.0, |
| "step": 252 |
| }, |
| { |
| "epoch": 0.1726962457337884, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.820617108070272e-05, |
| "loss": 0.7117, |
| "num_tokens": 96522727.0, |
| "step": 253 |
| }, |
| { |
| "epoch": 0.17337883959044367, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 4.818623509401694e-05, |
| "loss": 0.7088, |
| "num_tokens": 96903419.0, |
| "step": 254 |
| }, |
| { |
| "epoch": 0.17406143344709898, |
| "grad_norm": 0.0859375, |
| "learning_rate": 4.8166193589281154e-05, |
| "loss": 0.6602, |
| "num_tokens": 97319183.0, |
| "step": 255 |
| }, |
| { |
| "epoch": 0.17474402730375427, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.81460466687247e-05, |
| "loss": 0.6776, |
| "num_tokens": 97771307.0, |
| "step": 256 |
| }, |
| { |
| "epoch": 0.17542662116040955, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 4.8125794435114655e-05, |
| "loss": 0.6988, |
| "num_tokens": 98183260.0, |
| "step": 257 |
| }, |
| { |
| "epoch": 0.17610921501706484, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.810543699175527e-05, |
| "loss": 0.673, |
| "num_tokens": 98548095.0, |
| "step": 258 |
| }, |
| { |
| "epoch": 0.17679180887372015, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 4.808497444248748e-05, |
| "loss": 0.6234, |
| "num_tokens": 98962864.0, |
| "step": 259 |
| }, |
| { |
| "epoch": 0.17747440273037543, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 4.806440689168833e-05, |
| "loss": 0.6914, |
| "num_tokens": 99400041.0, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.1781569965870307, |
| "grad_norm": 0.09814453125, |
| "learning_rate": 4.804373444427048e-05, |
| "loss": 0.7153, |
| "num_tokens": 99790696.0, |
| "step": 261 |
| }, |
| { |
| "epoch": 0.178839590443686, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 4.802295720568164e-05, |
| "loss": 0.6694, |
| "num_tokens": 100188329.0, |
| "step": 262 |
| }, |
| { |
| "epoch": 0.1795221843003413, |
| "grad_norm": 0.08203125, |
| "learning_rate": 4.800207528190409e-05, |
| "loss": 0.6907, |
| "num_tokens": 100632528.0, |
| "step": 263 |
| }, |
| { |
| "epoch": 0.1802047781569966, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.798108877945402e-05, |
| "loss": 0.6463, |
| "num_tokens": 100987045.0, |
| "step": 264 |
| }, |
| { |
| "epoch": 0.18088737201365188, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 4.795999780538113e-05, |
| "loss": 0.6768, |
| "num_tokens": 101373875.0, |
| "step": 265 |
| }, |
| { |
| "epoch": 0.18156996587030716, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.793880246726801e-05, |
| "loss": 0.6743, |
| "num_tokens": 101767775.0, |
| "step": 266 |
| }, |
| { |
| "epoch": 0.18225255972696247, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 4.7917502873229564e-05, |
| "loss": 0.672, |
| "num_tokens": 102145317.0, |
| "step": 267 |
| }, |
| { |
| "epoch": 0.18293515358361775, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 4.789609913191253e-05, |
| "loss": 0.6828, |
| "num_tokens": 102519503.0, |
| "step": 268 |
| }, |
| { |
| "epoch": 0.18361774744027304, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.787459135249485e-05, |
| "loss": 0.6498, |
| "num_tokens": 102902403.0, |
| "step": 269 |
| }, |
| { |
| "epoch": 0.18430034129692832, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 4.785297964468519e-05, |
| "loss": 0.6923, |
| "num_tokens": 103240516.0, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.1849829351535836, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.783126411872233e-05, |
| "loss": 0.7118, |
| "num_tokens": 103602366.0, |
| "step": 271 |
| }, |
| { |
| "epoch": 0.18566552901023892, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.780944488537461e-05, |
| "loss": 0.677, |
| "num_tokens": 103993075.0, |
| "step": 272 |
| }, |
| { |
| "epoch": 0.1863481228668942, |
| "grad_norm": 0.10498046875, |
| "learning_rate": 4.7787522055939373e-05, |
| "loss": 0.6898, |
| "num_tokens": 104334773.0, |
| "step": 273 |
| }, |
| { |
| "epoch": 0.18703071672354948, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 4.7765495742242376e-05, |
| "loss": 0.6895, |
| "num_tokens": 104695258.0, |
| "step": 274 |
| }, |
| { |
| "epoch": 0.18771331058020477, |
| "grad_norm": 0.091796875, |
| "learning_rate": 4.7743366056637266e-05, |
| "loss": 0.6566, |
| "num_tokens": 105061908.0, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.18839590443686008, |
| "grad_norm": 0.0986328125, |
| "learning_rate": 4.772113311200496e-05, |
| "loss": 0.7222, |
| "num_tokens": 105416210.0, |
| "step": 276 |
| }, |
| { |
| "epoch": 0.18907849829351536, |
| "grad_norm": 0.10205078125, |
| "learning_rate": 4.76987970217531e-05, |
| "loss": 0.6653, |
| "num_tokens": 105754103.0, |
| "step": 277 |
| }, |
| { |
| "epoch": 0.18976109215017065, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 4.767635789981544e-05, |
| "loss": 0.68, |
| "num_tokens": 106110302.0, |
| "step": 278 |
| }, |
| { |
| "epoch": 0.19044368600682593, |
| "grad_norm": 0.08349609375, |
| "learning_rate": 4.76538158606513e-05, |
| "loss": 0.6748, |
| "num_tokens": 106514304.0, |
| "step": 279 |
| }, |
| { |
| "epoch": 0.19112627986348124, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.7631171019244974e-05, |
| "loss": 0.7368, |
| "num_tokens": 106919294.0, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.19180887372013652, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.760842349110513e-05, |
| "loss": 0.6186, |
| "num_tokens": 107296551.0, |
| "step": 281 |
| }, |
| { |
| "epoch": 0.1924914675767918, |
| "grad_norm": 0.09716796875, |
| "learning_rate": 4.7585573392264244e-05, |
| "loss": 0.6481, |
| "num_tokens": 107637243.0, |
| "step": 282 |
| }, |
| { |
| "epoch": 0.1931740614334471, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 4.756262083927795e-05, |
| "loss": 0.6437, |
| "num_tokens": 108033548.0, |
| "step": 283 |
| }, |
| { |
| "epoch": 0.19385665529010238, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.753956594922456e-05, |
| "loss": 0.6961, |
| "num_tokens": 108417815.0, |
| "step": 284 |
| }, |
| { |
| "epoch": 0.1945392491467577, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 4.7516408839704316e-05, |
| "loss": 0.644, |
| "num_tokens": 108819219.0, |
| "step": 285 |
| }, |
| { |
| "epoch": 0.19522184300341297, |
| "grad_norm": 0.0859375, |
| "learning_rate": 4.749314962883892e-05, |
| "loss": 0.6934, |
| "num_tokens": 109255162.0, |
| "step": 286 |
| }, |
| { |
| "epoch": 0.19590443686006825, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.7469788435270876e-05, |
| "loss": 0.6654, |
| "num_tokens": 109658924.0, |
| "step": 287 |
| }, |
| { |
| "epoch": 0.19658703071672354, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 4.744632537816289e-05, |
| "loss": 0.6538, |
| "num_tokens": 110042198.0, |
| "step": 288 |
| }, |
| { |
| "epoch": 0.19726962457337885, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.7422760577197226e-05, |
| "loss": 0.6951, |
| "num_tokens": 110424707.0, |
| "step": 289 |
| }, |
| { |
| "epoch": 0.19795221843003413, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 4.739909415257518e-05, |
| "loss": 0.6667, |
| "num_tokens": 110858993.0, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.19863481228668942, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.7375326225016394e-05, |
| "loss": 0.6414, |
| "num_tokens": 111191365.0, |
| "step": 291 |
| }, |
| { |
| "epoch": 0.1993174061433447, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.735145691575827e-05, |
| "loss": 0.7033, |
| "num_tokens": 111584364.0, |
| "step": 292 |
| }, |
| { |
| "epoch": 0.2, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.732748634655534e-05, |
| "loss": 0.6741, |
| "num_tokens": 111990417.0, |
| "step": 293 |
| }, |
| { |
| "epoch": 0.2006825938566553, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.730341463967866e-05, |
| "loss": 0.6786, |
| "num_tokens": 112327973.0, |
| "step": 294 |
| }, |
| { |
| "epoch": 0.20136518771331058, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.727924191791518e-05, |
| "loss": 0.6936, |
| "num_tokens": 112732095.0, |
| "step": 295 |
| }, |
| { |
| "epoch": 0.20204778156996586, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 4.725496830456708e-05, |
| "loss": 0.7114, |
| "num_tokens": 113104930.0, |
| "step": 296 |
| }, |
| { |
| "epoch": 0.20273037542662117, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.723059392345123e-05, |
| "loss": 0.658, |
| "num_tokens": 113467137.0, |
| "step": 297 |
| }, |
| { |
| "epoch": 0.20341296928327646, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 4.720611889889848e-05, |
| "loss": 0.6705, |
| "num_tokens": 113853495.0, |
| "step": 298 |
| }, |
| { |
| "epoch": 0.20409556313993174, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.718154335575304e-05, |
| "loss": 0.6619, |
| "num_tokens": 114222263.0, |
| "step": 299 |
| }, |
| { |
| "epoch": 0.20477815699658702, |
| "grad_norm": 0.09814453125, |
| "learning_rate": 4.7156867419371866e-05, |
| "loss": 0.6966, |
| "num_tokens": 114540376.0, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.2054607508532423, |
| "grad_norm": 0.087890625, |
| "learning_rate": 4.7132091215624e-05, |
| "loss": 0.6882, |
| "num_tokens": 114954829.0, |
| "step": 301 |
| }, |
| { |
| "epoch": 0.20614334470989762, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.7107214870889935e-05, |
| "loss": 0.6719, |
| "num_tokens": 115302505.0, |
| "step": 302 |
| }, |
| { |
| "epoch": 0.2068259385665529, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.7082238512060994e-05, |
| "loss": 0.6484, |
| "num_tokens": 115656774.0, |
| "step": 303 |
| }, |
| { |
| "epoch": 0.2075085324232082, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.705716226653863e-05, |
| "loss": 0.6559, |
| "num_tokens": 116064777.0, |
| "step": 304 |
| }, |
| { |
| "epoch": 0.20819112627986347, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 4.703198626223383e-05, |
| "loss": 0.6494, |
| "num_tokens": 116464215.0, |
| "step": 305 |
| }, |
| { |
| "epoch": 0.20887372013651878, |
| "grad_norm": 0.091796875, |
| "learning_rate": 4.7006710627566425e-05, |
| "loss": 0.6747, |
| "num_tokens": 116842042.0, |
| "step": 306 |
| }, |
| { |
| "epoch": 0.20955631399317406, |
| "grad_norm": 0.0986328125, |
| "learning_rate": 4.698133549146444e-05, |
| "loss": 0.6703, |
| "num_tokens": 117194351.0, |
| "step": 307 |
| }, |
| { |
| "epoch": 0.21023890784982935, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.695586098336347e-05, |
| "loss": 0.666, |
| "num_tokens": 117614028.0, |
| "step": 308 |
| }, |
| { |
| "epoch": 0.21092150170648463, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.6930287233205976e-05, |
| "loss": 0.647, |
| "num_tokens": 118090761.0, |
| "step": 309 |
| }, |
| { |
| "epoch": 0.21160409556313994, |
| "grad_norm": 0.09814453125, |
| "learning_rate": 4.6904614371440654e-05, |
| "loss": 0.6498, |
| "num_tokens": 118465258.0, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.21228668941979523, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.687884252902174e-05, |
| "loss": 0.6763, |
| "num_tokens": 118873320.0, |
| "step": 311 |
| }, |
| { |
| "epoch": 0.2129692832764505, |
| "grad_norm": 0.09375, |
| "learning_rate": 4.6852971837408384e-05, |
| "loss": 0.6872, |
| "num_tokens": 119211632.0, |
| "step": 312 |
| }, |
| { |
| "epoch": 0.2136518771331058, |
| "grad_norm": 0.09912109375, |
| "learning_rate": 4.682700242856391e-05, |
| "loss": 0.6435, |
| "num_tokens": 119530047.0, |
| "step": 313 |
| }, |
| { |
| "epoch": 0.2143344709897611, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 4.680093443495525e-05, |
| "loss": 0.6431, |
| "num_tokens": 119915702.0, |
| "step": 314 |
| }, |
| { |
| "epoch": 0.2150170648464164, |
| "grad_norm": 0.087890625, |
| "learning_rate": 4.677476798955213e-05, |
| "loss": 0.7039, |
| "num_tokens": 120323497.0, |
| "step": 315 |
| }, |
| { |
| "epoch": 0.21569965870307167, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.674850322582655e-05, |
| "loss": 0.6902, |
| "num_tokens": 120743724.0, |
| "step": 316 |
| }, |
| { |
| "epoch": 0.21638225255972696, |
| "grad_norm": 0.1005859375, |
| "learning_rate": 4.672214027775196e-05, |
| "loss": 0.6958, |
| "num_tokens": 121112731.0, |
| "step": 317 |
| }, |
| { |
| "epoch": 0.21706484641638224, |
| "grad_norm": 0.08251953125, |
| "learning_rate": 4.6695679279802653e-05, |
| "loss": 0.6731, |
| "num_tokens": 121575838.0, |
| "step": 318 |
| }, |
| { |
| "epoch": 0.21774744027303755, |
| "grad_norm": 0.091796875, |
| "learning_rate": 4.6669120366953055e-05, |
| "loss": 0.6435, |
| "num_tokens": 121942930.0, |
| "step": 319 |
| }, |
| { |
| "epoch": 0.21843003412969283, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.664246367467707e-05, |
| "loss": 0.669, |
| "num_tokens": 122331818.0, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.21911262798634812, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.661570933894734e-05, |
| "loss": 0.6981, |
| "num_tokens": 122724192.0, |
| "step": 321 |
| }, |
| { |
| "epoch": 0.2197952218430034, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.658885749623459e-05, |
| "loss": 0.6518, |
| "num_tokens": 123078889.0, |
| "step": 322 |
| }, |
| { |
| "epoch": 0.2204778156996587, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.6561908283506895e-05, |
| "loss": 0.6681, |
| "num_tokens": 123474922.0, |
| "step": 323 |
| }, |
| { |
| "epoch": 0.221160409556314, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.653486183822903e-05, |
| "loss": 0.6851, |
| "num_tokens": 123833149.0, |
| "step": 324 |
| }, |
| { |
| "epoch": 0.22184300341296928, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.6507718298361716e-05, |
| "loss": 0.6855, |
| "num_tokens": 124260967.0, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.22252559726962456, |
| "grad_norm": 0.099609375, |
| "learning_rate": 4.648047780236096e-05, |
| "loss": 0.6968, |
| "num_tokens": 124592227.0, |
| "step": 326 |
| }, |
| { |
| "epoch": 0.22320819112627988, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.645314048917733e-05, |
| "loss": 0.6754, |
| "num_tokens": 124995781.0, |
| "step": 327 |
| }, |
| { |
| "epoch": 0.22389078498293516, |
| "grad_norm": 0.09912109375, |
| "learning_rate": 4.642570649825524e-05, |
| "loss": 0.6799, |
| "num_tokens": 125361254.0, |
| "step": 328 |
| }, |
| { |
| "epoch": 0.22457337883959044, |
| "grad_norm": 0.10205078125, |
| "learning_rate": 4.6398175969532245e-05, |
| "loss": 0.6685, |
| "num_tokens": 125723116.0, |
| "step": 329 |
| }, |
| { |
| "epoch": 0.22525597269624573, |
| "grad_norm": 0.0859375, |
| "learning_rate": 4.637054904343833e-05, |
| "loss": 0.6476, |
| "num_tokens": 126115279.0, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.225938566552901, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.63428258608952e-05, |
| "loss": 0.6755, |
| "num_tokens": 126487824.0, |
| "step": 331 |
| }, |
| { |
| "epoch": 0.22662116040955632, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.631500656331554e-05, |
| "loss": 0.6247, |
| "num_tokens": 126907825.0, |
| "step": 332 |
| }, |
| { |
| "epoch": 0.2273037542662116, |
| "grad_norm": 0.0859375, |
| "learning_rate": 4.628709129260231e-05, |
| "loss": 0.672, |
| "num_tokens": 127319584.0, |
| "step": 333 |
| }, |
| { |
| "epoch": 0.2279863481228669, |
| "grad_norm": 0.083984375, |
| "learning_rate": 4.6259080191148025e-05, |
| "loss": 0.6432, |
| "num_tokens": 127725420.0, |
| "step": 334 |
| }, |
| { |
| "epoch": 0.22866894197952217, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.623097340183401e-05, |
| "loss": 0.6567, |
| "num_tokens": 128096244.0, |
| "step": 335 |
| }, |
| { |
| "epoch": 0.22935153583617748, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 4.62027710680297e-05, |
| "loss": 0.6898, |
| "num_tokens": 128456105.0, |
| "step": 336 |
| }, |
| { |
| "epoch": 0.23003412969283277, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.617447333359187e-05, |
| "loss": 0.657, |
| "num_tokens": 128814273.0, |
| "step": 337 |
| }, |
| { |
| "epoch": 0.23071672354948805, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 4.614608034286392e-05, |
| "loss": 0.6803, |
| "num_tokens": 129297757.0, |
| "step": 338 |
| }, |
| { |
| "epoch": 0.23139931740614333, |
| "grad_norm": 0.1005859375, |
| "learning_rate": 4.611759224067518e-05, |
| "loss": 0.7249, |
| "num_tokens": 129646174.0, |
| "step": 339 |
| }, |
| { |
| "epoch": 0.23208191126279865, |
| "grad_norm": 0.09765625, |
| "learning_rate": 4.60890091723401e-05, |
| "loss": 0.6674, |
| "num_tokens": 129992439.0, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.23276450511945393, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.606033128365754e-05, |
| "loss": 0.6747, |
| "num_tokens": 130381982.0, |
| "step": 341 |
| }, |
| { |
| "epoch": 0.2334470989761092, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.6031558720910026e-05, |
| "loss": 0.6504, |
| "num_tokens": 130725575.0, |
| "step": 342 |
| }, |
| { |
| "epoch": 0.2341296928327645, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.600269163086302e-05, |
| "loss": 0.6954, |
| "num_tokens": 131097305.0, |
| "step": 343 |
| }, |
| { |
| "epoch": 0.2348122866894198, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 4.597373016076415e-05, |
| "loss": 0.6611, |
| "num_tokens": 131458096.0, |
| "step": 344 |
| }, |
| { |
| "epoch": 0.2354948805460751, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.5944674458342473e-05, |
| "loss": 0.7214, |
| "num_tokens": 131869001.0, |
| "step": 345 |
| }, |
| { |
| "epoch": 0.23617747440273038, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.59155246718077e-05, |
| "loss": 0.7091, |
| "num_tokens": 132250335.0, |
| "step": 346 |
| }, |
| { |
| "epoch": 0.23686006825938566, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.5886280949849466e-05, |
| "loss": 0.6459, |
| "num_tokens": 132696648.0, |
| "step": 347 |
| }, |
| { |
| "epoch": 0.23754266211604094, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 4.585694344163654e-05, |
| "loss": 0.6993, |
| "num_tokens": 133066855.0, |
| "step": 348 |
| }, |
| { |
| "epoch": 0.23822525597269625, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.5827512296816114e-05, |
| "loss": 0.6749, |
| "num_tokens": 133444440.0, |
| "step": 349 |
| }, |
| { |
| "epoch": 0.23890784982935154, |
| "grad_norm": 0.0986328125, |
| "learning_rate": 4.579798766551298e-05, |
| "loss": 0.7206, |
| "num_tokens": 133787994.0, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.23959044368600682, |
| "grad_norm": 0.1005859375, |
| "learning_rate": 4.5768369698328814e-05, |
| "loss": 0.6417, |
| "num_tokens": 134111179.0, |
| "step": 351 |
| }, |
| { |
| "epoch": 0.2402730375426621, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 4.573865854634139e-05, |
| "loss": 0.6732, |
| "num_tokens": 134487914.0, |
| "step": 352 |
| }, |
| { |
| "epoch": 0.24095563139931742, |
| "grad_norm": 0.09716796875, |
| "learning_rate": 4.570885436110378e-05, |
| "loss": 0.66, |
| "num_tokens": 134832765.0, |
| "step": 353 |
| }, |
| { |
| "epoch": 0.2416382252559727, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.567895729464362e-05, |
| "loss": 0.6599, |
| "num_tokens": 135209662.0, |
| "step": 354 |
| }, |
| { |
| "epoch": 0.24232081911262798, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.564896749946234e-05, |
| "loss": 0.6604, |
| "num_tokens": 135563718.0, |
| "step": 355 |
| }, |
| { |
| "epoch": 0.24300341296928327, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.561888512853436e-05, |
| "loss": 0.6545, |
| "num_tokens": 135919722.0, |
| "step": 356 |
| }, |
| { |
| "epoch": 0.24368600682593858, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 4.55887103353063e-05, |
| "loss": 0.68, |
| "num_tokens": 136321694.0, |
| "step": 357 |
| }, |
| { |
| "epoch": 0.24436860068259386, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.555844327369624e-05, |
| "loss": 0.6302, |
| "num_tokens": 136752419.0, |
| "step": 358 |
| }, |
| { |
| "epoch": 0.24505119453924915, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.5528084098092885e-05, |
| "loss": 0.6993, |
| "num_tokens": 137135458.0, |
| "step": 359 |
| }, |
| { |
| "epoch": 0.24573378839590443, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.549763296335481e-05, |
| "loss": 0.6859, |
| "num_tokens": 137544925.0, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.24641638225255974, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.5467090024809686e-05, |
| "loss": 0.6611, |
| "num_tokens": 137916918.0, |
| "step": 361 |
| }, |
| { |
| "epoch": 0.24709897610921502, |
| "grad_norm": 0.095703125, |
| "learning_rate": 4.543645543825345e-05, |
| "loss": 0.6525, |
| "num_tokens": 138290744.0, |
| "step": 362 |
| }, |
| { |
| "epoch": 0.2477815699658703, |
| "grad_norm": 0.09814453125, |
| "learning_rate": 4.54057293599495e-05, |
| "loss": 0.6779, |
| "num_tokens": 138645821.0, |
| "step": 363 |
| }, |
| { |
| "epoch": 0.2484641638225256, |
| "grad_norm": 0.091796875, |
| "learning_rate": 4.5374911946627964e-05, |
| "loss": 0.639, |
| "num_tokens": 139025382.0, |
| "step": 364 |
| }, |
| { |
| "epoch": 0.24914675767918087, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.534400335548484e-05, |
| "loss": 0.6562, |
| "num_tokens": 139435802.0, |
| "step": 365 |
| }, |
| { |
| "epoch": 0.24982935153583619, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.5313003744181185e-05, |
| "loss": 0.6633, |
| "num_tokens": 139826460.0, |
| "step": 366 |
| }, |
| { |
| "epoch": 0.25051194539249144, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 4.52819132708424e-05, |
| "loss": 0.6496, |
| "num_tokens": 140163935.0, |
| "step": 367 |
| }, |
| { |
| "epoch": 0.25119453924914675, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 4.525073209405732e-05, |
| "loss": 0.6489, |
| "num_tokens": 140480505.0, |
| "step": 368 |
| }, |
| { |
| "epoch": 0.25187713310580206, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.521946037287745e-05, |
| "loss": 0.6629, |
| "num_tokens": 140894545.0, |
| "step": 369 |
| }, |
| { |
| "epoch": 0.2525597269624573, |
| "grad_norm": 0.10107421875, |
| "learning_rate": 4.518809826681616e-05, |
| "loss": 0.6819, |
| "num_tokens": 141219364.0, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.25324232081911263, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 4.5156645935847866e-05, |
| "loss": 0.6502, |
| "num_tokens": 141685436.0, |
| "step": 371 |
| }, |
| { |
| "epoch": 0.25392491467576794, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 4.51251035404072e-05, |
| "loss": 0.6386, |
| "num_tokens": 142090074.0, |
| "step": 372 |
| }, |
| { |
| "epoch": 0.2546075085324232, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 4.5093471241388213e-05, |
| "loss": 0.7487, |
| "num_tokens": 142514467.0, |
| "step": 373 |
| }, |
| { |
| "epoch": 0.2552901023890785, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.5061749200143535e-05, |
| "loss": 0.7012, |
| "num_tokens": 142900927.0, |
| "step": 374 |
| }, |
| { |
| "epoch": 0.25597269624573377, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.5029937578483566e-05, |
| "loss": 0.6751, |
| "num_tokens": 143301347.0, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.2566552901023891, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 4.499803653867564e-05, |
| "loss": 0.6869, |
| "num_tokens": 143683127.0, |
| "step": 376 |
| }, |
| { |
| "epoch": 0.2573378839590444, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.4966046243443214e-05, |
| "loss": 0.6172, |
| "num_tokens": 144058279.0, |
| "step": 377 |
| }, |
| { |
| "epoch": 0.25802047781569964, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.493396685596501e-05, |
| "loss": 0.6877, |
| "num_tokens": 144449641.0, |
| "step": 378 |
| }, |
| { |
| "epoch": 0.25870307167235496, |
| "grad_norm": 0.1015625, |
| "learning_rate": 4.4901798539874215e-05, |
| "loss": 0.6113, |
| "num_tokens": 144742915.0, |
| "step": 379 |
| }, |
| { |
| "epoch": 0.2593856655290102, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.486954145925761e-05, |
| "loss": 0.6868, |
| "num_tokens": 145160611.0, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.2600682593856655, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 4.4837195778654774e-05, |
| "loss": 0.6491, |
| "num_tokens": 145534695.0, |
| "step": 381 |
| }, |
| { |
| "epoch": 0.26075085324232083, |
| "grad_norm": 0.095703125, |
| "learning_rate": 4.480476166305722e-05, |
| "loss": 0.6716, |
| "num_tokens": 145925715.0, |
| "step": 382 |
| }, |
| { |
| "epoch": 0.2614334470989761, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 4.4772239277907544e-05, |
| "loss": 0.6605, |
| "num_tokens": 146352774.0, |
| "step": 383 |
| }, |
| { |
| "epoch": 0.2621160409556314, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 4.473962878909861e-05, |
| "loss": 0.6306, |
| "num_tokens": 146708294.0, |
| "step": 384 |
| }, |
| { |
| "epoch": 0.2627986348122867, |
| "grad_norm": 0.095703125, |
| "learning_rate": 4.470693036297268e-05, |
| "loss": 0.6385, |
| "num_tokens": 147031619.0, |
| "step": 385 |
| }, |
| { |
| "epoch": 0.26348122866894197, |
| "grad_norm": 0.0986328125, |
| "learning_rate": 4.467414416632058e-05, |
| "loss": 0.675, |
| "num_tokens": 147391776.0, |
| "step": 386 |
| }, |
| { |
| "epoch": 0.2641638225255973, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 4.464127036638083e-05, |
| "loss": 0.6502, |
| "num_tokens": 147779041.0, |
| "step": 387 |
| }, |
| { |
| "epoch": 0.26484641638225254, |
| "grad_norm": 0.09375, |
| "learning_rate": 4.4608309130838824e-05, |
| "loss": 0.6584, |
| "num_tokens": 148158504.0, |
| "step": 388 |
| }, |
| { |
| "epoch": 0.26552901023890785, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.457526062782594e-05, |
| "loss": 0.6596, |
| "num_tokens": 148608505.0, |
| "step": 389 |
| }, |
| { |
| "epoch": 0.26621160409556316, |
| "grad_norm": 0.09814453125, |
| "learning_rate": 4.454212502591871e-05, |
| "loss": 0.6875, |
| "num_tokens": 148964661.0, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.2668941979522184, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 4.4508902494137916e-05, |
| "loss": 0.7036, |
| "num_tokens": 149373333.0, |
| "step": 391 |
| }, |
| { |
| "epoch": 0.2675767918088737, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.447559320194781e-05, |
| "loss": 0.6616, |
| "num_tokens": 149775756.0, |
| "step": 392 |
| }, |
| { |
| "epoch": 0.26825938566552904, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.4442197319255145e-05, |
| "loss": 0.6581, |
| "num_tokens": 150213143.0, |
| "step": 393 |
| }, |
| { |
| "epoch": 0.2689419795221843, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.4408715016408405e-05, |
| "loss": 0.6815, |
| "num_tokens": 150572364.0, |
| "step": 394 |
| }, |
| { |
| "epoch": 0.2696245733788396, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 4.437514646419687e-05, |
| "loss": 0.6777, |
| "num_tokens": 150937307.0, |
| "step": 395 |
| }, |
| { |
| "epoch": 0.27030716723549486, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 4.434149183384977e-05, |
| "loss": 0.6331, |
| "num_tokens": 151283062.0, |
| "step": 396 |
| }, |
| { |
| "epoch": 0.27098976109215017, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.430775129703543e-05, |
| "loss": 0.6995, |
| "num_tokens": 151691785.0, |
| "step": 397 |
| }, |
| { |
| "epoch": 0.2716723549488055, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.427392502586033e-05, |
| "loss": 0.661, |
| "num_tokens": 152071105.0, |
| "step": 398 |
| }, |
| { |
| "epoch": 0.27235494880546074, |
| "grad_norm": 0.0859375, |
| "learning_rate": 4.42400131928683e-05, |
| "loss": 0.6621, |
| "num_tokens": 152484454.0, |
| "step": 399 |
| }, |
| { |
| "epoch": 0.27303754266211605, |
| "grad_norm": 0.09375, |
| "learning_rate": 4.4206015971039625e-05, |
| "loss": 0.6593, |
| "num_tokens": 152834849.0, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.2737201365187713, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 4.417193353379012e-05, |
| "loss": 0.6165, |
| "num_tokens": 153226832.0, |
| "step": 401 |
| }, |
| { |
| "epoch": 0.2744027303754266, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 4.4137766054970293e-05, |
| "loss": 0.6474, |
| "num_tokens": 153588418.0, |
| "step": 402 |
| }, |
| { |
| "epoch": 0.27508532423208193, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.4103513708864416e-05, |
| "loss": 0.6652, |
| "num_tokens": 153970683.0, |
| "step": 403 |
| }, |
| { |
| "epoch": 0.2757679180887372, |
| "grad_norm": 0.095703125, |
| "learning_rate": 4.4069176670189696e-05, |
| "loss": 0.648, |
| "num_tokens": 154329185.0, |
| "step": 404 |
| }, |
| { |
| "epoch": 0.2764505119453925, |
| "grad_norm": 0.083984375, |
| "learning_rate": 4.403475511409531e-05, |
| "loss": 0.6531, |
| "num_tokens": 154739335.0, |
| "step": 405 |
| }, |
| { |
| "epoch": 0.2771331058020478, |
| "grad_norm": 0.091796875, |
| "learning_rate": 4.400024921616156e-05, |
| "loss": 0.7196, |
| "num_tokens": 155179775.0, |
| "step": 406 |
| }, |
| { |
| "epoch": 0.27781569965870306, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.396565915239899e-05, |
| "loss": 0.6576, |
| "num_tokens": 155526504.0, |
| "step": 407 |
| }, |
| { |
| "epoch": 0.2784982935153584, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.3930985099247415e-05, |
| "loss": 0.6733, |
| "num_tokens": 155897832.0, |
| "step": 408 |
| }, |
| { |
| "epoch": 0.27918088737201363, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.389622723357512e-05, |
| "loss": 0.7002, |
| "num_tokens": 156311761.0, |
| "step": 409 |
| }, |
| { |
| "epoch": 0.27986348122866894, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 4.386138573267788e-05, |
| "loss": 0.6482, |
| "num_tokens": 156709373.0, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.28054607508532425, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.3826460774278084e-05, |
| "loss": 0.6538, |
| "num_tokens": 157058533.0, |
| "step": 411 |
| }, |
| { |
| "epoch": 0.2812286689419795, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 4.3791452536523846e-05, |
| "loss": 0.7021, |
| "num_tokens": 157424054.0, |
| "step": 412 |
| }, |
| { |
| "epoch": 0.2819112627986348, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.3756361197988056e-05, |
| "loss": 0.6269, |
| "num_tokens": 157793850.0, |
| "step": 413 |
| }, |
| { |
| "epoch": 0.2825938566552901, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.3721186937667505e-05, |
| "loss": 0.6686, |
| "num_tokens": 158148458.0, |
| "step": 414 |
| }, |
| { |
| "epoch": 0.2832764505119454, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 4.368592993498194e-05, |
| "loss": 0.6137, |
| "num_tokens": 158541316.0, |
| "step": 415 |
| }, |
| { |
| "epoch": 0.2839590443686007, |
| "grad_norm": 0.091796875, |
| "learning_rate": 4.36505903697732e-05, |
| "loss": 0.6805, |
| "num_tokens": 158926018.0, |
| "step": 416 |
| }, |
| { |
| "epoch": 0.28464163822525596, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.3615168422304235e-05, |
| "loss": 0.7111, |
| "num_tokens": 159330507.0, |
| "step": 417 |
| }, |
| { |
| "epoch": 0.28532423208191127, |
| "grad_norm": 0.09375, |
| "learning_rate": 4.3579664273258243e-05, |
| "loss": 0.6387, |
| "num_tokens": 159680025.0, |
| "step": 418 |
| }, |
| { |
| "epoch": 0.2860068259385666, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 4.354407810373769e-05, |
| "loss": 0.6883, |
| "num_tokens": 160076536.0, |
| "step": 419 |
| }, |
| { |
| "epoch": 0.28668941979522183, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 4.350841009526346e-05, |
| "loss": 0.667, |
| "num_tokens": 160459756.0, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.28737201365187715, |
| "grad_norm": 0.09716796875, |
| "learning_rate": 4.3472660429773843e-05, |
| "loss": 0.6526, |
| "num_tokens": 160783158.0, |
| "step": 421 |
| }, |
| { |
| "epoch": 0.2880546075085324, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.343682928962369e-05, |
| "loss": 0.6545, |
| "num_tokens": 161198309.0, |
| "step": 422 |
| }, |
| { |
| "epoch": 0.2887372013651877, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.340091685758341e-05, |
| "loss": 0.6605, |
| "num_tokens": 161596202.0, |
| "step": 423 |
| }, |
| { |
| "epoch": 0.289419795221843, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 4.3364923316838114e-05, |
| "loss": 0.663, |
| "num_tokens": 161999350.0, |
| "step": 424 |
| }, |
| { |
| "epoch": 0.2901023890784983, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.33288488509866e-05, |
| "loss": 0.6725, |
| "num_tokens": 162358277.0, |
| "step": 425 |
| }, |
| { |
| "epoch": 0.2907849829351536, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.3292693644040486e-05, |
| "loss": 0.697, |
| "num_tokens": 162753562.0, |
| "step": 426 |
| }, |
| { |
| "epoch": 0.29146757679180885, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.3256457880423214e-05, |
| "loss": 0.6697, |
| "num_tokens": 163116958.0, |
| "step": 427 |
| }, |
| { |
| "epoch": 0.29215017064846416, |
| "grad_norm": 0.09375, |
| "learning_rate": 4.322014174496915e-05, |
| "loss": 0.7068, |
| "num_tokens": 163544269.0, |
| "step": 428 |
| }, |
| { |
| "epoch": 0.29283276450511947, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 4.318374542292264e-05, |
| "loss": 0.6447, |
| "num_tokens": 163995392.0, |
| "step": 429 |
| }, |
| { |
| "epoch": 0.2935153583617747, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.3147269099937014e-05, |
| "loss": 0.6786, |
| "num_tokens": 164401378.0, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.29419795221843004, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.3110712962073716e-05, |
| "loss": 0.6405, |
| "num_tokens": 164789059.0, |
| "step": 431 |
| }, |
| { |
| "epoch": 0.29488054607508535, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 4.30740771958013e-05, |
| "loss": 0.6329, |
| "num_tokens": 165121965.0, |
| "step": 432 |
| }, |
| { |
| "epoch": 0.2955631399317406, |
| "grad_norm": 0.087890625, |
| "learning_rate": 4.30373619879945e-05, |
| "loss": 0.6224, |
| "num_tokens": 165516420.0, |
| "step": 433 |
| }, |
| { |
| "epoch": 0.2962457337883959, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 4.3000567525933266e-05, |
| "loss": 0.67, |
| "num_tokens": 165906662.0, |
| "step": 434 |
| }, |
| { |
| "epoch": 0.29692832764505117, |
| "grad_norm": 0.09375, |
| "learning_rate": 4.2963693997301814e-05, |
| "loss": 0.646, |
| "num_tokens": 166276916.0, |
| "step": 435 |
| }, |
| { |
| "epoch": 0.2976109215017065, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 4.292674159018768e-05, |
| "loss": 0.6514, |
| "num_tokens": 166716902.0, |
| "step": 436 |
| }, |
| { |
| "epoch": 0.2982935153583618, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.2889710493080726e-05, |
| "loss": 0.6384, |
| "num_tokens": 167063298.0, |
| "step": 437 |
| }, |
| { |
| "epoch": 0.29897610921501705, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 4.285260089487225e-05, |
| "loss": 0.6215, |
| "num_tokens": 167486308.0, |
| "step": 438 |
| }, |
| { |
| "epoch": 0.29965870307167236, |
| "grad_norm": 0.09375, |
| "learning_rate": 4.281541298485392e-05, |
| "loss": 0.6669, |
| "num_tokens": 167838078.0, |
| "step": 439 |
| }, |
| { |
| "epoch": 0.3003412969283277, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 4.2778146952716884e-05, |
| "loss": 0.6822, |
| "num_tokens": 168250987.0, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.30102389078498293, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 4.27408029885508e-05, |
| "loss": 0.6922, |
| "num_tokens": 168667286.0, |
| "step": 441 |
| }, |
| { |
| "epoch": 0.30170648464163824, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.270338128284283e-05, |
| "loss": 0.6353, |
| "num_tokens": 169054451.0, |
| "step": 442 |
| }, |
| { |
| "epoch": 0.3023890784982935, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 4.266588202647668e-05, |
| "loss": 0.6858, |
| "num_tokens": 169469643.0, |
| "step": 443 |
| }, |
| { |
| "epoch": 0.3030716723549488, |
| "grad_norm": 0.087890625, |
| "learning_rate": 4.262830541073165e-05, |
| "loss": 0.6911, |
| "num_tokens": 169868165.0, |
| "step": 444 |
| }, |
| { |
| "epoch": 0.3037542662116041, |
| "grad_norm": 0.09814453125, |
| "learning_rate": 4.259065162728163e-05, |
| "loss": 0.6666, |
| "num_tokens": 170217424.0, |
| "step": 445 |
| }, |
| { |
| "epoch": 0.3044368600682594, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 4.2552920868194146e-05, |
| "loss": 0.6701, |
| "num_tokens": 170612853.0, |
| "step": 446 |
| }, |
| { |
| "epoch": 0.3051194539249147, |
| "grad_norm": 0.099609375, |
| "learning_rate": 4.2515113325929345e-05, |
| "loss": 0.6474, |
| "num_tokens": 170968208.0, |
| "step": 447 |
| }, |
| { |
| "epoch": 0.30580204778156994, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.247722919333906e-05, |
| "loss": 0.6475, |
| "num_tokens": 171346652.0, |
| "step": 448 |
| }, |
| { |
| "epoch": 0.30648464163822525, |
| "grad_norm": 0.087890625, |
| "learning_rate": 4.243926866366579e-05, |
| "loss": 0.5972, |
| "num_tokens": 171754367.0, |
| "step": 449 |
| }, |
| { |
| "epoch": 0.30716723549488056, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.240123193054172e-05, |
| "loss": 0.682, |
| "num_tokens": 172160313.0, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.3078498293515358, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 4.236311918798775e-05, |
| "loss": 0.6485, |
| "num_tokens": 172567536.0, |
| "step": 451 |
| }, |
| { |
| "epoch": 0.30853242320819113, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 4.2324930630412505e-05, |
| "loss": 0.6538, |
| "num_tokens": 172985074.0, |
| "step": 452 |
| }, |
| { |
| "epoch": 0.30921501706484644, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.22866664526113e-05, |
| "loss": 0.6664, |
| "num_tokens": 173358620.0, |
| "step": 453 |
| }, |
| { |
| "epoch": 0.3098976109215017, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.224832684976523e-05, |
| "loss": 0.6733, |
| "num_tokens": 173741900.0, |
| "step": 454 |
| }, |
| { |
| "epoch": 0.310580204778157, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.2209912017440066e-05, |
| "loss": 0.6707, |
| "num_tokens": 174112017.0, |
| "step": 455 |
| }, |
| { |
| "epoch": 0.31126279863481227, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.217142215158537e-05, |
| "loss": 0.6611, |
| "num_tokens": 174524101.0, |
| "step": 456 |
| }, |
| { |
| "epoch": 0.3119453924914676, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.21328574485334e-05, |
| "loss": 0.6726, |
| "num_tokens": 174941045.0, |
| "step": 457 |
| }, |
| { |
| "epoch": 0.3126279863481229, |
| "grad_norm": 0.0830078125, |
| "learning_rate": 4.209421810499818e-05, |
| "loss": 0.6579, |
| "num_tokens": 175386228.0, |
| "step": 458 |
| }, |
| { |
| "epoch": 0.31331058020477814, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.205550431807444e-05, |
| "loss": 0.6242, |
| "num_tokens": 175729092.0, |
| "step": 459 |
| }, |
| { |
| "epoch": 0.31399317406143346, |
| "grad_norm": 0.10107421875, |
| "learning_rate": 4.2016716285236655e-05, |
| "loss": 0.6487, |
| "num_tokens": 176115268.0, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.3146757679180887, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.1977854204338016e-05, |
| "loss": 0.6524, |
| "num_tokens": 176510000.0, |
| "step": 461 |
| }, |
| { |
| "epoch": 0.315358361774744, |
| "grad_norm": 0.083984375, |
| "learning_rate": 4.1938918273609425e-05, |
| "loss": 0.6445, |
| "num_tokens": 176926040.0, |
| "step": 462 |
| }, |
| { |
| "epoch": 0.31604095563139933, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.189990869165847e-05, |
| "loss": 0.6283, |
| "num_tokens": 177291446.0, |
| "step": 463 |
| }, |
| { |
| "epoch": 0.3167235494880546, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.186082565746845e-05, |
| "loss": 0.6244, |
| "num_tokens": 177680634.0, |
| "step": 464 |
| }, |
| { |
| "epoch": 0.3174061433447099, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.182166937039731e-05, |
| "loss": 0.6284, |
| "num_tokens": 178049137.0, |
| "step": 465 |
| }, |
| { |
| "epoch": 0.3180887372013652, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.178244003017667e-05, |
| "loss": 0.6729, |
| "num_tokens": 178419127.0, |
| "step": 466 |
| }, |
| { |
| "epoch": 0.31877133105802047, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.174313783691077e-05, |
| "loss": 0.6476, |
| "num_tokens": 178799262.0, |
| "step": 467 |
| }, |
| { |
| "epoch": 0.3194539249146758, |
| "grad_norm": 0.1025390625, |
| "learning_rate": 4.1703762991075475e-05, |
| "loss": 0.6523, |
| "num_tokens": 179142011.0, |
| "step": 468 |
| }, |
| { |
| "epoch": 0.32013651877133104, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 4.166431569351724e-05, |
| "loss": 0.6705, |
| "num_tokens": 179540343.0, |
| "step": 469 |
| }, |
| { |
| "epoch": 0.32081911262798635, |
| "grad_norm": 0.087890625, |
| "learning_rate": 4.1624796145452075e-05, |
| "loss": 0.668, |
| "num_tokens": 179947208.0, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.32150170648464166, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 4.1585204548464554e-05, |
| "loss": 0.6884, |
| "num_tokens": 180363552.0, |
| "step": 471 |
| }, |
| { |
| "epoch": 0.3221843003412969, |
| "grad_norm": 0.10107421875, |
| "learning_rate": 4.154554110450676e-05, |
| "loss": 0.6668, |
| "num_tokens": 180719896.0, |
| "step": 472 |
| }, |
| { |
| "epoch": 0.3228668941979522, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 4.150580601589723e-05, |
| "loss": 0.6784, |
| "num_tokens": 181070784.0, |
| "step": 473 |
| }, |
| { |
| "epoch": 0.3235494880546075, |
| "grad_norm": 0.0830078125, |
| "learning_rate": 4.146599948531998e-05, |
| "loss": 0.6996, |
| "num_tokens": 181530619.0, |
| "step": 474 |
| }, |
| { |
| "epoch": 0.3242320819112628, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 4.1426121715823455e-05, |
| "loss": 0.613, |
| "num_tokens": 181891310.0, |
| "step": 475 |
| }, |
| { |
| "epoch": 0.3249146757679181, |
| "grad_norm": 0.0986328125, |
| "learning_rate": 4.138617291081943e-05, |
| "loss": 0.6487, |
| "num_tokens": 182272816.0, |
| "step": 476 |
| }, |
| { |
| "epoch": 0.32559726962457336, |
| "grad_norm": 0.09375, |
| "learning_rate": 4.1346153274082086e-05, |
| "loss": 0.6519, |
| "num_tokens": 182640008.0, |
| "step": 477 |
| }, |
| { |
| "epoch": 0.32627986348122867, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 4.130606300974686e-05, |
| "loss": 0.6855, |
| "num_tokens": 183000223.0, |
| "step": 478 |
| }, |
| { |
| "epoch": 0.326962457337884, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.126590232230949e-05, |
| "loss": 0.6525, |
| "num_tokens": 183375015.0, |
| "step": 479 |
| }, |
| { |
| "epoch": 0.32764505119453924, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 4.1225671416624886e-05, |
| "loss": 0.6555, |
| "num_tokens": 183776326.0, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.32832764505119455, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 4.118537049790619e-05, |
| "loss": 0.6559, |
| "num_tokens": 184119649.0, |
| "step": 481 |
| }, |
| { |
| "epoch": 0.3290102389078498, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 4.114499977172364e-05, |
| "loss": 0.662, |
| "num_tokens": 184511279.0, |
| "step": 482 |
| }, |
| { |
| "epoch": 0.3296928327645051, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 4.1104559444003545e-05, |
| "loss": 0.6753, |
| "num_tokens": 184943461.0, |
| "step": 483 |
| }, |
| { |
| "epoch": 0.33037542662116043, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 4.106404972102726e-05, |
| "loss": 0.6844, |
| "num_tokens": 185316534.0, |
| "step": 484 |
| }, |
| { |
| "epoch": 0.3310580204778157, |
| "grad_norm": 0.09375, |
| "learning_rate": 4.1023470809430115e-05, |
| "loss": 0.7202, |
| "num_tokens": 185721070.0, |
| "step": 485 |
| }, |
| { |
| "epoch": 0.331740614334471, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 4.098282291620036e-05, |
| "loss": 0.6629, |
| "num_tokens": 186156118.0, |
| "step": 486 |
| }, |
| { |
| "epoch": 0.3324232081911263, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 4.094210624867812e-05, |
| "loss": 0.6702, |
| "num_tokens": 186575131.0, |
| "step": 487 |
| }, |
| { |
| "epoch": 0.33310580204778156, |
| "grad_norm": 0.0986328125, |
| "learning_rate": 4.090132101455431e-05, |
| "loss": 0.6544, |
| "num_tokens": 186953339.0, |
| "step": 488 |
| }, |
| { |
| "epoch": 0.3337883959044369, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.086046742186962e-05, |
| "loss": 0.6803, |
| "num_tokens": 187351835.0, |
| "step": 489 |
| }, |
| { |
| "epoch": 0.33447098976109213, |
| "grad_norm": 0.099609375, |
| "learning_rate": 4.081954567901341e-05, |
| "loss": 0.6776, |
| "num_tokens": 187730193.0, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.33515358361774744, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 4.0778555994722676e-05, |
| "loss": 0.6785, |
| "num_tokens": 188114011.0, |
| "step": 491 |
| }, |
| { |
| "epoch": 0.33583617747440275, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 4.0737498578080974e-05, |
| "loss": 0.6326, |
| "num_tokens": 188462453.0, |
| "step": 492 |
| }, |
| { |
| "epoch": 0.336518771331058, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 4.069637363851735e-05, |
| "loss": 0.6498, |
| "num_tokens": 188890662.0, |
| "step": 493 |
| }, |
| { |
| "epoch": 0.3372013651877133, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 4.065518138580528e-05, |
| "loss": 0.6627, |
| "num_tokens": 189248524.0, |
| "step": 494 |
| }, |
| { |
| "epoch": 0.3378839590443686, |
| "grad_norm": 0.08984375, |
| "learning_rate": 4.0613922030061593e-05, |
| "loss": 0.6389, |
| "num_tokens": 189611145.0, |
| "step": 495 |
| }, |
| { |
| "epoch": 0.3385665529010239, |
| "grad_norm": 0.095703125, |
| "learning_rate": 4.0572595781745414e-05, |
| "loss": 0.6334, |
| "num_tokens": 189960238.0, |
| "step": 496 |
| }, |
| { |
| "epoch": 0.3392491467576792, |
| "grad_norm": 0.09765625, |
| "learning_rate": 4.0531202851657064e-05, |
| "loss": 0.6809, |
| "num_tokens": 190336045.0, |
| "step": 497 |
| }, |
| { |
| "epoch": 0.33993174061433445, |
| "grad_norm": 0.0830078125, |
| "learning_rate": 4.048974345093701e-05, |
| "loss": 0.6608, |
| "num_tokens": 190805418.0, |
| "step": 498 |
| }, |
| { |
| "epoch": 0.34061433447098977, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.044821779106476e-05, |
| "loss": 0.6522, |
| "num_tokens": 191175891.0, |
| "step": 499 |
| }, |
| { |
| "epoch": 0.3412969283276451, |
| "grad_norm": 0.09814453125, |
| "learning_rate": 4.040662608385783e-05, |
| "loss": 0.6549, |
| "num_tokens": 191543591.0, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.34197952218430033, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 4.036496854147064e-05, |
| "loss": 0.651, |
| "num_tokens": 191930617.0, |
| "step": 501 |
| }, |
| { |
| "epoch": 0.34266211604095564, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 4.032324537639338e-05, |
| "loss": 0.6423, |
| "num_tokens": 192340206.0, |
| "step": 502 |
| }, |
| { |
| "epoch": 0.3433447098976109, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 4.0281456801451004e-05, |
| "loss": 0.6392, |
| "num_tokens": 192728819.0, |
| "step": 503 |
| }, |
| { |
| "epoch": 0.3440273037542662, |
| "grad_norm": 0.09814453125, |
| "learning_rate": 4.023960302980214e-05, |
| "loss": 0.6641, |
| "num_tokens": 193077410.0, |
| "step": 504 |
| }, |
| { |
| "epoch": 0.3447098976109215, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 4.019768427493792e-05, |
| "loss": 0.6568, |
| "num_tokens": 193451364.0, |
| "step": 505 |
| }, |
| { |
| "epoch": 0.3453924914675768, |
| "grad_norm": 0.087890625, |
| "learning_rate": 4.0155700750681004e-05, |
| "loss": 0.6602, |
| "num_tokens": 193846836.0, |
| "step": 506 |
| }, |
| { |
| "epoch": 0.3460750853242321, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 4.0113652671184385e-05, |
| "loss": 0.677, |
| "num_tokens": 194226506.0, |
| "step": 507 |
| }, |
| { |
| "epoch": 0.34675767918088735, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 4.007154025093039e-05, |
| "loss": 0.6609, |
| "num_tokens": 194601094.0, |
| "step": 508 |
| }, |
| { |
| "epoch": 0.34744027303754266, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 4.0029363704729486e-05, |
| "loss": 0.6865, |
| "num_tokens": 195004588.0, |
| "step": 509 |
| }, |
| { |
| "epoch": 0.34812286689419797, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 3.99871232477193e-05, |
| "loss": 0.6292, |
| "num_tokens": 195399830.0, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.3488054607508532, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 3.9944819095363404e-05, |
| "loss": 0.6449, |
| "num_tokens": 195819744.0, |
| "step": 511 |
| }, |
| { |
| "epoch": 0.34948805460750854, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 3.990245146345033e-05, |
| "loss": 0.6393, |
| "num_tokens": 196166544.0, |
| "step": 512 |
| }, |
| { |
| "epoch": 0.35017064846416385, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 3.986002056809235e-05, |
| "loss": 0.6309, |
| "num_tokens": 196547362.0, |
| "step": 513 |
| }, |
| { |
| "epoch": 0.3508532423208191, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 3.981752662572449e-05, |
| "loss": 0.654, |
| "num_tokens": 196928205.0, |
| "step": 514 |
| }, |
| { |
| "epoch": 0.3515358361774744, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 3.9774969853103334e-05, |
| "loss": 0.6283, |
| "num_tokens": 197289380.0, |
| "step": 515 |
| }, |
| { |
| "epoch": 0.35221843003412967, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 3.973235046730598e-05, |
| "loss": 0.6967, |
| "num_tokens": 197694444.0, |
| "step": 516 |
| }, |
| { |
| "epoch": 0.352901023890785, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 3.968966868572888e-05, |
| "loss": 0.6672, |
| "num_tokens": 198084017.0, |
| "step": 517 |
| }, |
| { |
| "epoch": 0.3535836177474403, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 3.9646924726086794e-05, |
| "loss": 0.6365, |
| "num_tokens": 198457611.0, |
| "step": 518 |
| }, |
| { |
| "epoch": 0.35426621160409555, |
| "grad_norm": 0.08984375, |
| "learning_rate": 3.9604118806411624e-05, |
| "loss": 0.596, |
| "num_tokens": 198824465.0, |
| "step": 519 |
| }, |
| { |
| "epoch": 0.35494880546075086, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 3.9561251145051325e-05, |
| "loss": 0.6599, |
| "num_tokens": 199248080.0, |
| "step": 520 |
| }, |
| { |
| "epoch": 0.3556313993174061, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 3.951832196066879e-05, |
| "loss": 0.6823, |
| "num_tokens": 199619540.0, |
| "step": 521 |
| }, |
| { |
| "epoch": 0.3563139931740614, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 3.947533147224073e-05, |
| "loss": 0.6247, |
| "num_tokens": 200026026.0, |
| "step": 522 |
| }, |
| { |
| "epoch": 0.35699658703071674, |
| "grad_norm": 0.10009765625, |
| "learning_rate": 3.943227989905659e-05, |
| "loss": 0.6756, |
| "num_tokens": 200374588.0, |
| "step": 523 |
| }, |
| { |
| "epoch": 0.357679180887372, |
| "grad_norm": 0.0986328125, |
| "learning_rate": 3.938916746071735e-05, |
| "loss": 0.7299, |
| "num_tokens": 200769347.0, |
| "step": 524 |
| }, |
| { |
| "epoch": 0.3583617747440273, |
| "grad_norm": 0.1005859375, |
| "learning_rate": 3.934599437713449e-05, |
| "loss": 0.6787, |
| "num_tokens": 201135153.0, |
| "step": 525 |
| }, |
| { |
| "epoch": 0.3590443686006826, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 3.9302760868528824e-05, |
| "loss": 0.6332, |
| "num_tokens": 201473004.0, |
| "step": 526 |
| }, |
| { |
| "epoch": 0.3597269624573379, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 3.925946715542941e-05, |
| "loss": 0.6684, |
| "num_tokens": 201855512.0, |
| "step": 527 |
| }, |
| { |
| "epoch": 0.3604095563139932, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 3.9216113458672354e-05, |
| "loss": 0.6776, |
| "num_tokens": 202224422.0, |
| "step": 528 |
| }, |
| { |
| "epoch": 0.36109215017064844, |
| "grad_norm": 0.09716796875, |
| "learning_rate": 3.917269999939978e-05, |
| "loss": 0.6689, |
| "num_tokens": 202608738.0, |
| "step": 529 |
| }, |
| { |
| "epoch": 0.36177474402730375, |
| "grad_norm": 0.09912109375, |
| "learning_rate": 3.912922699905861e-05, |
| "loss": 0.612, |
| "num_tokens": 202936389.0, |
| "step": 530 |
| }, |
| { |
| "epoch": 0.36245733788395906, |
| "grad_norm": 0.09912109375, |
| "learning_rate": 3.9085694679399544e-05, |
| "loss": 0.6489, |
| "num_tokens": 203262618.0, |
| "step": 531 |
| }, |
| { |
| "epoch": 0.3631399317406143, |
| "grad_norm": 0.09814453125, |
| "learning_rate": 3.904210326247577e-05, |
| "loss": 0.6524, |
| "num_tokens": 203606843.0, |
| "step": 532 |
| }, |
| { |
| "epoch": 0.36382252559726963, |
| "grad_norm": 0.091796875, |
| "learning_rate": 3.899845297064201e-05, |
| "loss": 0.6398, |
| "num_tokens": 203987258.0, |
| "step": 533 |
| }, |
| { |
| "epoch": 0.36450511945392494, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 3.8954744026553245e-05, |
| "loss": 0.6805, |
| "num_tokens": 204386876.0, |
| "step": 534 |
| }, |
| { |
| "epoch": 0.3651877133105802, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 3.8910976653163655e-05, |
| "loss": 0.6615, |
| "num_tokens": 204803047.0, |
| "step": 535 |
| }, |
| { |
| "epoch": 0.3658703071672355, |
| "grad_norm": 0.08984375, |
| "learning_rate": 3.886715107372547e-05, |
| "loss": 0.6617, |
| "num_tokens": 205206396.0, |
| "step": 536 |
| }, |
| { |
| "epoch": 0.36655290102389076, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 3.8823267511787804e-05, |
| "loss": 0.6401, |
| "num_tokens": 205572867.0, |
| "step": 537 |
| }, |
| { |
| "epoch": 0.3672354948805461, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 3.877932619119554e-05, |
| "loss": 0.6636, |
| "num_tokens": 205989659.0, |
| "step": 538 |
| }, |
| { |
| "epoch": 0.3679180887372014, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 3.873532733608818e-05, |
| "loss": 0.6788, |
| "num_tokens": 206428066.0, |
| "step": 539 |
| }, |
| { |
| "epoch": 0.36860068259385664, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 3.869127117089871e-05, |
| "loss": 0.6258, |
| "num_tokens": 206795183.0, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.36928327645051195, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 3.864715792035244e-05, |
| "loss": 0.6733, |
| "num_tokens": 207189392.0, |
| "step": 541 |
| }, |
| { |
| "epoch": 0.3699658703071672, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 3.860298780946587e-05, |
| "loss": 0.6193, |
| "num_tokens": 207574477.0, |
| "step": 542 |
| }, |
| { |
| "epoch": 0.3706484641638225, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 3.855876106354553e-05, |
| "loss": 0.6823, |
| "num_tokens": 207977577.0, |
| "step": 543 |
| }, |
| { |
| "epoch": 0.37133105802047783, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 3.8514477908186845e-05, |
| "loss": 0.6316, |
| "num_tokens": 208389718.0, |
| "step": 544 |
| }, |
| { |
| "epoch": 0.3720136518771331, |
| "grad_norm": 0.087890625, |
| "learning_rate": 3.847013856927298e-05, |
| "loss": 0.6435, |
| "num_tokens": 208806680.0, |
| "step": 545 |
| }, |
| { |
| "epoch": 0.3726962457337884, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 3.84257432729737e-05, |
| "loss": 0.6474, |
| "num_tokens": 209176140.0, |
| "step": 546 |
| }, |
| { |
| "epoch": 0.3733788395904437, |
| "grad_norm": 0.09912109375, |
| "learning_rate": 3.838129224574415e-05, |
| "loss": 0.6536, |
| "num_tokens": 209517993.0, |
| "step": 547 |
| }, |
| { |
| "epoch": 0.37406143344709897, |
| "grad_norm": 0.091796875, |
| "learning_rate": 3.833678571432382e-05, |
| "loss": 0.6657, |
| "num_tokens": 209890732.0, |
| "step": 548 |
| }, |
| { |
| "epoch": 0.3747440273037543, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 3.829222390573527e-05, |
| "loss": 0.6238, |
| "num_tokens": 210246927.0, |
| "step": 549 |
| }, |
| { |
| "epoch": 0.37542662116040953, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 3.824760704728303e-05, |
| "loss": 0.6624, |
| "num_tokens": 210620229.0, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.37610921501706485, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 3.8202935366552467e-05, |
| "loss": 0.6506, |
| "num_tokens": 210981205.0, |
| "step": 551 |
| }, |
| { |
| "epoch": 0.37679180887372016, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 3.8158209091408544e-05, |
| "loss": 0.6704, |
| "num_tokens": 211320650.0, |
| "step": 552 |
| }, |
| { |
| "epoch": 0.3774744027303754, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 3.8113428449994754e-05, |
| "loss": 0.6679, |
| "num_tokens": 211706784.0, |
| "step": 553 |
| }, |
| { |
| "epoch": 0.3781569965870307, |
| "grad_norm": 0.08984375, |
| "learning_rate": 3.806859367073185e-05, |
| "loss": 0.6521, |
| "num_tokens": 212108214.0, |
| "step": 554 |
| }, |
| { |
| "epoch": 0.378839590443686, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 3.802370498231677e-05, |
| "loss": 0.664, |
| "num_tokens": 212481056.0, |
| "step": 555 |
| }, |
| { |
| "epoch": 0.3795221843003413, |
| "grad_norm": 0.08984375, |
| "learning_rate": 3.797876261372143e-05, |
| "loss": 0.64, |
| "num_tokens": 212867487.0, |
| "step": 556 |
| }, |
| { |
| "epoch": 0.3802047781569966, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 3.793376679419157e-05, |
| "loss": 0.6352, |
| "num_tokens": 213208531.0, |
| "step": 557 |
| }, |
| { |
| "epoch": 0.38088737201365186, |
| "grad_norm": 0.09375, |
| "learning_rate": 3.788871775324556e-05, |
| "loss": 0.654, |
| "num_tokens": 213553690.0, |
| "step": 558 |
| }, |
| { |
| "epoch": 0.38156996587030717, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 3.7843615720673256e-05, |
| "loss": 0.6678, |
| "num_tokens": 213940682.0, |
| "step": 559 |
| }, |
| { |
| "epoch": 0.3822525597269625, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 3.7798460926534814e-05, |
| "loss": 0.6683, |
| "num_tokens": 214296967.0, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.38293515358361774, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 3.775325360115952e-05, |
| "loss": 0.6293, |
| "num_tokens": 214705968.0, |
| "step": 561 |
| }, |
| { |
| "epoch": 0.38361774744027305, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 3.7707993975144634e-05, |
| "loss": 0.6756, |
| "num_tokens": 215111668.0, |
| "step": 562 |
| }, |
| { |
| "epoch": 0.3843003412969283, |
| "grad_norm": 0.0986328125, |
| "learning_rate": 3.766268227935416e-05, |
| "loss": 0.7153, |
| "num_tokens": 215488136.0, |
| "step": 563 |
| }, |
| { |
| "epoch": 0.3849829351535836, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 3.761731874491775e-05, |
| "loss": 0.6707, |
| "num_tokens": 215856298.0, |
| "step": 564 |
| }, |
| { |
| "epoch": 0.3856655290102389, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 3.757190360322943e-05, |
| "loss": 0.6469, |
| "num_tokens": 216266686.0, |
| "step": 565 |
| }, |
| { |
| "epoch": 0.3863481228668942, |
| "grad_norm": 0.09716796875, |
| "learning_rate": 3.7526437085946504e-05, |
| "loss": 0.6636, |
| "num_tokens": 216653248.0, |
| "step": 566 |
| }, |
| { |
| "epoch": 0.3870307167235495, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 3.748091942498832e-05, |
| "loss": 0.6647, |
| "num_tokens": 217049083.0, |
| "step": 567 |
| }, |
| { |
| "epoch": 0.38771331058020475, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 3.7435350852535105e-05, |
| "loss": 0.6452, |
| "num_tokens": 217413017.0, |
| "step": 568 |
| }, |
| { |
| "epoch": 0.38839590443686006, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 3.738973160102679e-05, |
| "loss": 0.6235, |
| "num_tokens": 217757086.0, |
| "step": 569 |
| }, |
| { |
| "epoch": 0.3890784982935154, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 3.7344061903161796e-05, |
| "loss": 0.6716, |
| "num_tokens": 218179883.0, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.38976109215017063, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 3.7298341991895883e-05, |
| "loss": 0.6684, |
| "num_tokens": 218560618.0, |
| "step": 571 |
| }, |
| { |
| "epoch": 0.39044368600682594, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 3.725257210044094e-05, |
| "loss": 0.6662, |
| "num_tokens": 218981219.0, |
| "step": 572 |
| }, |
| { |
| "epoch": 0.39112627986348125, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 3.720675246226379e-05, |
| "loss": 0.6267, |
| "num_tokens": 219372993.0, |
| "step": 573 |
| }, |
| { |
| "epoch": 0.3918088737201365, |
| "grad_norm": 0.087890625, |
| "learning_rate": 3.716088331108501e-05, |
| "loss": 0.6547, |
| "num_tokens": 219794699.0, |
| "step": 574 |
| }, |
| { |
| "epoch": 0.3924914675767918, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 3.711496488087775e-05, |
| "loss": 0.6674, |
| "num_tokens": 220144855.0, |
| "step": 575 |
| }, |
| { |
| "epoch": 0.3931740614334471, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 3.7068997405866515e-05, |
| "loss": 0.6363, |
| "num_tokens": 220538949.0, |
| "step": 576 |
| }, |
| { |
| "epoch": 0.3938566552901024, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 3.702298112052597e-05, |
| "loss": 0.6486, |
| "num_tokens": 220910557.0, |
| "step": 577 |
| }, |
| { |
| "epoch": 0.3945392491467577, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 3.697691625957978e-05, |
| "loss": 0.6712, |
| "num_tokens": 221354810.0, |
| "step": 578 |
| }, |
| { |
| "epoch": 0.39522184300341295, |
| "grad_norm": 0.095703125, |
| "learning_rate": 3.693080305799938e-05, |
| "loss": 0.715, |
| "num_tokens": 221745503.0, |
| "step": 579 |
| }, |
| { |
| "epoch": 0.39590443686006827, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 3.6884641751002784e-05, |
| "loss": 0.629, |
| "num_tokens": 222093783.0, |
| "step": 580 |
| }, |
| { |
| "epoch": 0.3965870307167236, |
| "grad_norm": 0.09912109375, |
| "learning_rate": 3.6838432574053375e-05, |
| "loss": 0.6446, |
| "num_tokens": 222464319.0, |
| "step": 581 |
| }, |
| { |
| "epoch": 0.39726962457337883, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 3.6792175762858745e-05, |
| "loss": 0.6848, |
| "num_tokens": 222876213.0, |
| "step": 582 |
| }, |
| { |
| "epoch": 0.39795221843003414, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 3.6745871553369414e-05, |
| "loss": 0.648, |
| "num_tokens": 223204000.0, |
| "step": 583 |
| }, |
| { |
| "epoch": 0.3986348122866894, |
| "grad_norm": 0.09375, |
| "learning_rate": 3.669952018177772e-05, |
| "loss": 0.6412, |
| "num_tokens": 223566716.0, |
| "step": 584 |
| }, |
| { |
| "epoch": 0.3993174061433447, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 3.6653121884516555e-05, |
| "loss": 0.6033, |
| "num_tokens": 223929151.0, |
| "step": 585 |
| }, |
| { |
| "epoch": 0.4, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 3.660667689825817e-05, |
| "loss": 0.6908, |
| "num_tokens": 224318688.0, |
| "step": 586 |
| }, |
| { |
| "epoch": 0.4006825938566553, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 3.656018545991296e-05, |
| "loss": 0.6566, |
| "num_tokens": 224697865.0, |
| "step": 587 |
| }, |
| { |
| "epoch": 0.4013651877133106, |
| "grad_norm": 0.08984375, |
| "learning_rate": 3.65136478066283e-05, |
| "loss": 0.6298, |
| "num_tokens": 225061467.0, |
| "step": 588 |
| }, |
| { |
| "epoch": 0.40204778156996585, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 3.646706417578727e-05, |
| "loss": 0.6601, |
| "num_tokens": 225470520.0, |
| "step": 589 |
| }, |
| { |
| "epoch": 0.40273037542662116, |
| "grad_norm": 0.09375, |
| "learning_rate": 3.6420434805007476e-05, |
| "loss": 0.6329, |
| "num_tokens": 225865685.0, |
| "step": 590 |
| }, |
| { |
| "epoch": 0.40341296928327647, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 3.637375993213986e-05, |
| "loss": 0.6188, |
| "num_tokens": 226186168.0, |
| "step": 591 |
| }, |
| { |
| "epoch": 0.4040955631399317, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 3.632703979526746e-05, |
| "loss": 0.6217, |
| "num_tokens": 226555606.0, |
| "step": 592 |
| }, |
| { |
| "epoch": 0.40477815699658704, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 3.628027463270419e-05, |
| "loss": 0.7018, |
| "num_tokens": 226969310.0, |
| "step": 593 |
| }, |
| { |
| "epoch": 0.40546075085324235, |
| "grad_norm": 0.0859375, |
| "learning_rate": 3.623346468299363e-05, |
| "loss": 0.6502, |
| "num_tokens": 227350204.0, |
| "step": 594 |
| }, |
| { |
| "epoch": 0.4061433447098976, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 3.618661018490784e-05, |
| "loss": 0.6463, |
| "num_tokens": 227786006.0, |
| "step": 595 |
| }, |
| { |
| "epoch": 0.4068259385665529, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 3.6139711377446086e-05, |
| "loss": 0.6335, |
| "num_tokens": 228128719.0, |
| "step": 596 |
| }, |
| { |
| "epoch": 0.40750853242320817, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 3.6092768499833655e-05, |
| "loss": 0.6503, |
| "num_tokens": 228531765.0, |
| "step": 597 |
| }, |
| { |
| "epoch": 0.4081911262798635, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 3.6045781791520665e-05, |
| "loss": 0.6735, |
| "num_tokens": 228974046.0, |
| "step": 598 |
| }, |
| { |
| "epoch": 0.4088737201365188, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 3.599875149218077e-05, |
| "loss": 0.6356, |
| "num_tokens": 229365037.0, |
| "step": 599 |
| }, |
| { |
| "epoch": 0.40955631399317405, |
| "grad_norm": 0.091796875, |
| "learning_rate": 3.595167784170998e-05, |
| "loss": 0.6544, |
| "num_tokens": 229748231.0, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.41023890784982936, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 3.5904561080225474e-05, |
| "loss": 0.628, |
| "num_tokens": 230131351.0, |
| "step": 601 |
| }, |
| { |
| "epoch": 0.4109215017064846, |
| "grad_norm": 0.083984375, |
| "learning_rate": 3.585740144806428e-05, |
| "loss": 0.6462, |
| "num_tokens": 230550361.0, |
| "step": 602 |
| }, |
| { |
| "epoch": 0.4116040955631399, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 3.5810199185782124e-05, |
| "loss": 0.6627, |
| "num_tokens": 230909521.0, |
| "step": 603 |
| }, |
| { |
| "epoch": 0.41228668941979524, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 3.5762954534152214e-05, |
| "loss": 0.6722, |
| "num_tokens": 231280257.0, |
| "step": 604 |
| }, |
| { |
| "epoch": 0.4129692832764505, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 3.571566773416394e-05, |
| "loss": 0.6614, |
| "num_tokens": 231716185.0, |
| "step": 605 |
| }, |
| { |
| "epoch": 0.4136518771331058, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 3.566833902702171e-05, |
| "loss": 0.6655, |
| "num_tokens": 232066273.0, |
| "step": 606 |
| }, |
| { |
| "epoch": 0.4143344709897611, |
| "grad_norm": 0.0859375, |
| "learning_rate": 3.5620968654143686e-05, |
| "loss": 0.6778, |
| "num_tokens": 232506102.0, |
| "step": 607 |
| }, |
| { |
| "epoch": 0.4150170648464164, |
| "grad_norm": 0.09716796875, |
| "learning_rate": 3.557355685716056e-05, |
| "loss": 0.6706, |
| "num_tokens": 232863356.0, |
| "step": 608 |
| }, |
| { |
| "epoch": 0.4156996587030717, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 3.552610387791433e-05, |
| "loss": 0.703, |
| "num_tokens": 233268645.0, |
| "step": 609 |
| }, |
| { |
| "epoch": 0.41638225255972694, |
| "grad_norm": 0.08984375, |
| "learning_rate": 3.5478609958457056e-05, |
| "loss": 0.6621, |
| "num_tokens": 233655900.0, |
| "step": 610 |
| }, |
| { |
| "epoch": 0.41706484641638225, |
| "grad_norm": 0.08984375, |
| "learning_rate": 3.543107534104963e-05, |
| "loss": 0.6456, |
| "num_tokens": 234037627.0, |
| "step": 611 |
| }, |
| { |
| "epoch": 0.41774744027303756, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 3.538350026816053e-05, |
| "loss": 0.6515, |
| "num_tokens": 234414778.0, |
| "step": 612 |
| }, |
| { |
| "epoch": 0.4184300341296928, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 3.533588498246463e-05, |
| "loss": 0.6787, |
| "num_tokens": 234788834.0, |
| "step": 613 |
| }, |
| { |
| "epoch": 0.41911262798634813, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 3.5288229726841864e-05, |
| "loss": 0.6847, |
| "num_tokens": 235155148.0, |
| "step": 614 |
| }, |
| { |
| "epoch": 0.4197952218430034, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 3.524053474437611e-05, |
| "loss": 0.6396, |
| "num_tokens": 235502719.0, |
| "step": 615 |
| }, |
| { |
| "epoch": 0.4204778156996587, |
| "grad_norm": 0.091796875, |
| "learning_rate": 3.519280027835383e-05, |
| "loss": 0.6511, |
| "num_tokens": 235871545.0, |
| "step": 616 |
| }, |
| { |
| "epoch": 0.421160409556314, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 3.5145026572262934e-05, |
| "loss": 0.684, |
| "num_tokens": 236288957.0, |
| "step": 617 |
| }, |
| { |
| "epoch": 0.42184300341296926, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 3.509721386979147e-05, |
| "loss": 0.653, |
| "num_tokens": 236698850.0, |
| "step": 618 |
| }, |
| { |
| "epoch": 0.4225255972696246, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 3.504936241482641e-05, |
| "loss": 0.6675, |
| "num_tokens": 237076298.0, |
| "step": 619 |
| }, |
| { |
| "epoch": 0.4232081911262799, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 3.500147245145238e-05, |
| "loss": 0.6816, |
| "num_tokens": 237433481.0, |
| "step": 620 |
| }, |
| { |
| "epoch": 0.42389078498293514, |
| "grad_norm": 0.095703125, |
| "learning_rate": 3.495354422395046e-05, |
| "loss": 0.6268, |
| "num_tokens": 237793537.0, |
| "step": 621 |
| }, |
| { |
| "epoch": 0.42457337883959045, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 3.490557797679688e-05, |
| "loss": 0.6532, |
| "num_tokens": 238210414.0, |
| "step": 622 |
| }, |
| { |
| "epoch": 0.4252559726962457, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 3.4857573954661835e-05, |
| "loss": 0.67, |
| "num_tokens": 238584169.0, |
| "step": 623 |
| }, |
| { |
| "epoch": 0.425938566552901, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 3.480953240240818e-05, |
| "loss": 0.6746, |
| "num_tokens": 238959325.0, |
| "step": 624 |
| }, |
| { |
| "epoch": 0.42662116040955633, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 3.476145356509023e-05, |
| "loss": 0.6498, |
| "num_tokens": 239310469.0, |
| "step": 625 |
| }, |
| { |
| "epoch": 0.4273037542662116, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 3.471333768795246e-05, |
| "loss": 0.6478, |
| "num_tokens": 239702762.0, |
| "step": 626 |
| }, |
| { |
| "epoch": 0.4279863481228669, |
| "grad_norm": 0.08984375, |
| "learning_rate": 3.4665185016428306e-05, |
| "loss": 0.6808, |
| "num_tokens": 240087384.0, |
| "step": 627 |
| }, |
| { |
| "epoch": 0.4286689419795222, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 3.461699579613888e-05, |
| "loss": 0.6413, |
| "num_tokens": 240452678.0, |
| "step": 628 |
| }, |
| { |
| "epoch": 0.42935153583617747, |
| "grad_norm": 0.095703125, |
| "learning_rate": 3.4568770272891706e-05, |
| "loss": 0.6734, |
| "num_tokens": 240802942.0, |
| "step": 629 |
| }, |
| { |
| "epoch": 0.4300341296928328, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 3.452050869267951e-05, |
| "loss": 0.6647, |
| "num_tokens": 241153898.0, |
| "step": 630 |
| }, |
| { |
| "epoch": 0.43071672354948803, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 3.447221130167892e-05, |
| "loss": 0.6656, |
| "num_tokens": 241514517.0, |
| "step": 631 |
| }, |
| { |
| "epoch": 0.43139931740614335, |
| "grad_norm": 0.091796875, |
| "learning_rate": 3.442387834624927e-05, |
| "loss": 0.7002, |
| "num_tokens": 241904037.0, |
| "step": 632 |
| }, |
| { |
| "epoch": 0.43208191126279866, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 3.437551007293124e-05, |
| "loss": 0.6605, |
| "num_tokens": 242280144.0, |
| "step": 633 |
| }, |
| { |
| "epoch": 0.4327645051194539, |
| "grad_norm": 0.09375, |
| "learning_rate": 3.4327106728445734e-05, |
| "loss": 0.7022, |
| "num_tokens": 242657445.0, |
| "step": 634 |
| }, |
| { |
| "epoch": 0.4334470989761092, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 3.427866855969249e-05, |
| "loss": 0.6249, |
| "num_tokens": 243032696.0, |
| "step": 635 |
| }, |
| { |
| "epoch": 0.4341296928327645, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 3.4230195813748905e-05, |
| "loss": 0.6806, |
| "num_tokens": 243458287.0, |
| "step": 636 |
| }, |
| { |
| "epoch": 0.4348122866894198, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 3.418168873786877e-05, |
| "loss": 0.6613, |
| "num_tokens": 243830297.0, |
| "step": 637 |
| }, |
| { |
| "epoch": 0.4354948805460751, |
| "grad_norm": 0.0859375, |
| "learning_rate": 3.4133147579480944e-05, |
| "loss": 0.6391, |
| "num_tokens": 244232660.0, |
| "step": 638 |
| }, |
| { |
| "epoch": 0.43617747440273036, |
| "grad_norm": 0.091796875, |
| "learning_rate": 3.408457258618818e-05, |
| "loss": 0.6181, |
| "num_tokens": 244607365.0, |
| "step": 639 |
| }, |
| { |
| "epoch": 0.43686006825938567, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 3.4035964005765786e-05, |
| "loss": 0.6188, |
| "num_tokens": 245008254.0, |
| "step": 640 |
| }, |
| { |
| "epoch": 0.437542662116041, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 3.398732208616043e-05, |
| "loss": 0.6865, |
| "num_tokens": 245437181.0, |
| "step": 641 |
| }, |
| { |
| "epoch": 0.43822525597269624, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 3.3938647075488785e-05, |
| "loss": 0.6701, |
| "num_tokens": 245862358.0, |
| "step": 642 |
| }, |
| { |
| "epoch": 0.43890784982935155, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 3.3889939222036374e-05, |
| "loss": 0.6521, |
| "num_tokens": 246239469.0, |
| "step": 643 |
| }, |
| { |
| "epoch": 0.4395904436860068, |
| "grad_norm": 0.1015625, |
| "learning_rate": 3.384119877425622e-05, |
| "loss": 0.7379, |
| "num_tokens": 246656797.0, |
| "step": 644 |
| }, |
| { |
| "epoch": 0.4402730375426621, |
| "grad_norm": 0.09375, |
| "learning_rate": 3.379242598076761e-05, |
| "loss": 0.6732, |
| "num_tokens": 247016964.0, |
| "step": 645 |
| }, |
| { |
| "epoch": 0.4409556313993174, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 3.374362109035482e-05, |
| "loss": 0.6574, |
| "num_tokens": 247404358.0, |
| "step": 646 |
| }, |
| { |
| "epoch": 0.4416382252559727, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 3.369478435196585e-05, |
| "loss": 0.641, |
| "num_tokens": 247774775.0, |
| "step": 647 |
| }, |
| { |
| "epoch": 0.442320819112628, |
| "grad_norm": 0.09912109375, |
| "learning_rate": 3.364591601471115e-05, |
| "loss": 0.6236, |
| "num_tokens": 248075907.0, |
| "step": 648 |
| }, |
| { |
| "epoch": 0.44300341296928325, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 3.3597016327862365e-05, |
| "loss": 0.643, |
| "num_tokens": 248447489.0, |
| "step": 649 |
| }, |
| { |
| "epoch": 0.44368600682593856, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 3.354808554085103e-05, |
| "loss": 0.679, |
| "num_tokens": 248791992.0, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.4443686006825939, |
| "grad_norm": 0.087890625, |
| "learning_rate": 3.3499123903267345e-05, |
| "loss": 0.6368, |
| "num_tokens": 249191737.0, |
| "step": 651 |
| }, |
| { |
| "epoch": 0.44505119453924913, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 3.345013166485884e-05, |
| "loss": 0.676, |
| "num_tokens": 249580003.0, |
| "step": 652 |
| }, |
| { |
| "epoch": 0.44573378839590444, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 3.340110907552917e-05, |
| "loss": 0.6785, |
| "num_tokens": 249921522.0, |
| "step": 653 |
| }, |
| { |
| "epoch": 0.44641638225255975, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 3.3352056385336784e-05, |
| "loss": 0.6562, |
| "num_tokens": 250331702.0, |
| "step": 654 |
| }, |
| { |
| "epoch": 0.447098976109215, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 3.330297384449369e-05, |
| "loss": 0.6585, |
| "num_tokens": 250713070.0, |
| "step": 655 |
| }, |
| { |
| "epoch": 0.4477815699658703, |
| "grad_norm": 0.095703125, |
| "learning_rate": 3.3253861703364135e-05, |
| "loss": 0.6533, |
| "num_tokens": 251067840.0, |
| "step": 656 |
| }, |
| { |
| "epoch": 0.4484641638225256, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 3.3204720212463374e-05, |
| "loss": 0.6926, |
| "num_tokens": 251458713.0, |
| "step": 657 |
| }, |
| { |
| "epoch": 0.4491467576791809, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 3.315554962245638e-05, |
| "loss": 0.6447, |
| "num_tokens": 251854949.0, |
| "step": 658 |
| }, |
| { |
| "epoch": 0.4498293515358362, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 3.3106350184156545e-05, |
| "loss": 0.6183, |
| "num_tokens": 252231527.0, |
| "step": 659 |
| }, |
| { |
| "epoch": 0.45051194539249145, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 3.3057122148524394e-05, |
| "loss": 0.6417, |
| "num_tokens": 252648795.0, |
| "step": 660 |
| }, |
| { |
| "epoch": 0.45119453924914676, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 3.300786576666636e-05, |
| "loss": 0.6659, |
| "num_tokens": 253057627.0, |
| "step": 661 |
| }, |
| { |
| "epoch": 0.451877133105802, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 3.2958581289833425e-05, |
| "loss": 0.6901, |
| "num_tokens": 253453331.0, |
| "step": 662 |
| }, |
| { |
| "epoch": 0.45255972696245733, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 3.2909268969419924e-05, |
| "loss": 0.6699, |
| "num_tokens": 253785018.0, |
| "step": 663 |
| }, |
| { |
| "epoch": 0.45324232081911264, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 3.28599290569622e-05, |
| "loss": 0.6202, |
| "num_tokens": 254132239.0, |
| "step": 664 |
| }, |
| { |
| "epoch": 0.4539249146757679, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 3.2810561804137316e-05, |
| "loss": 0.66, |
| "num_tokens": 254539106.0, |
| "step": 665 |
| }, |
| { |
| "epoch": 0.4546075085324232, |
| "grad_norm": 0.10205078125, |
| "learning_rate": 3.276116746276183e-05, |
| "loss": 0.6572, |
| "num_tokens": 254858100.0, |
| "step": 666 |
| }, |
| { |
| "epoch": 0.4552901023890785, |
| "grad_norm": 0.08251953125, |
| "learning_rate": 3.271174628479048e-05, |
| "loss": 0.6645, |
| "num_tokens": 255329584.0, |
| "step": 667 |
| }, |
| { |
| "epoch": 0.4559726962457338, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 3.266229852231485e-05, |
| "loss": 0.6206, |
| "num_tokens": 255669662.0, |
| "step": 668 |
| }, |
| { |
| "epoch": 0.4566552901023891, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 3.261282442756217e-05, |
| "loss": 0.6852, |
| "num_tokens": 256092984.0, |
| "step": 669 |
| }, |
| { |
| "epoch": 0.45733788395904434, |
| "grad_norm": 0.091796875, |
| "learning_rate": 3.2563324252893964e-05, |
| "loss": 0.6583, |
| "num_tokens": 256460069.0, |
| "step": 670 |
| }, |
| { |
| "epoch": 0.45802047781569966, |
| "grad_norm": 0.08984375, |
| "learning_rate": 3.2513798250804804e-05, |
| "loss": 0.6388, |
| "num_tokens": 256852272.0, |
| "step": 671 |
| }, |
| { |
| "epoch": 0.45870307167235497, |
| "grad_norm": 0.08984375, |
| "learning_rate": 3.2464246673920994e-05, |
| "loss": 0.6296, |
| "num_tokens": 257216416.0, |
| "step": 672 |
| }, |
| { |
| "epoch": 0.4593856655290102, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 3.241466977499929e-05, |
| "loss": 0.6277, |
| "num_tokens": 257607252.0, |
| "step": 673 |
| }, |
| { |
| "epoch": 0.46006825938566553, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 3.236506780692563e-05, |
| "loss": 0.6896, |
| "num_tokens": 258025518.0, |
| "step": 674 |
| }, |
| { |
| "epoch": 0.46075085324232085, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 3.2315441022713796e-05, |
| "loss": 0.6125, |
| "num_tokens": 258439462.0, |
| "step": 675 |
| }, |
| { |
| "epoch": 0.4614334470989761, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 3.226578967550419e-05, |
| "loss": 0.6536, |
| "num_tokens": 258782974.0, |
| "step": 676 |
| }, |
| { |
| "epoch": 0.4621160409556314, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 3.221611401856248e-05, |
| "loss": 0.6501, |
| "num_tokens": 259233406.0, |
| "step": 677 |
| }, |
| { |
| "epoch": 0.46279863481228667, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 3.2166414305278334e-05, |
| "loss": 0.6614, |
| "num_tokens": 259646125.0, |
| "step": 678 |
| }, |
| { |
| "epoch": 0.463481228668942, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 3.211669078916417e-05, |
| "loss": 0.6632, |
| "num_tokens": 260050686.0, |
| "step": 679 |
| }, |
| { |
| "epoch": 0.4641638225255973, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 3.206694372385375e-05, |
| "loss": 0.6842, |
| "num_tokens": 260499223.0, |
| "step": 680 |
| }, |
| { |
| "epoch": 0.46484641638225255, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 3.201717336310103e-05, |
| "loss": 0.6283, |
| "num_tokens": 260830815.0, |
| "step": 681 |
| }, |
| { |
| "epoch": 0.46552901023890786, |
| "grad_norm": 0.087890625, |
| "learning_rate": 3.1967379960778736e-05, |
| "loss": 0.6532, |
| "num_tokens": 261226426.0, |
| "step": 682 |
| }, |
| { |
| "epoch": 0.4662116040955631, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 3.191756377087718e-05, |
| "loss": 0.6486, |
| "num_tokens": 261618465.0, |
| "step": 683 |
| }, |
| { |
| "epoch": 0.4668941979522184, |
| "grad_norm": 0.08984375, |
| "learning_rate": 3.1867725047502856e-05, |
| "loss": 0.6387, |
| "num_tokens": 261963028.0, |
| "step": 684 |
| }, |
| { |
| "epoch": 0.46757679180887374, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 3.1817864044877236e-05, |
| "loss": 0.655, |
| "num_tokens": 262428104.0, |
| "step": 685 |
| }, |
| { |
| "epoch": 0.468259385665529, |
| "grad_norm": 0.10107421875, |
| "learning_rate": 3.176798101733542e-05, |
| "loss": 0.6079, |
| "num_tokens": 262715994.0, |
| "step": 686 |
| }, |
| { |
| "epoch": 0.4689419795221843, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 3.1718076219324875e-05, |
| "loss": 0.6603, |
| "num_tokens": 263118521.0, |
| "step": 687 |
| }, |
| { |
| "epoch": 0.4696245733788396, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 3.1668149905404085e-05, |
| "loss": 0.6093, |
| "num_tokens": 263506418.0, |
| "step": 688 |
| }, |
| { |
| "epoch": 0.47030716723549487, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 3.161820233024131e-05, |
| "loss": 0.6716, |
| "num_tokens": 263896216.0, |
| "step": 689 |
| }, |
| { |
| "epoch": 0.4709897610921502, |
| "grad_norm": 0.08984375, |
| "learning_rate": 3.156823374861324e-05, |
| "loss": 0.6822, |
| "num_tokens": 264331327.0, |
| "step": 690 |
| }, |
| { |
| "epoch": 0.47167235494880544, |
| "grad_norm": 0.09912109375, |
| "learning_rate": 3.151824441540373e-05, |
| "loss": 0.6472, |
| "num_tokens": 264681982.0, |
| "step": 691 |
| }, |
| { |
| "epoch": 0.47235494880546075, |
| "grad_norm": 0.091796875, |
| "learning_rate": 3.146823458560249e-05, |
| "loss": 0.6387, |
| "num_tokens": 265054203.0, |
| "step": 692 |
| }, |
| { |
| "epoch": 0.47303754266211606, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 3.141820451430377e-05, |
| "loss": 0.6531, |
| "num_tokens": 265441871.0, |
| "step": 693 |
| }, |
| { |
| "epoch": 0.4737201365187713, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 3.136815445670507e-05, |
| "loss": 0.6733, |
| "num_tokens": 265813543.0, |
| "step": 694 |
| }, |
| { |
| "epoch": 0.47440273037542663, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 3.131808466810586e-05, |
| "loss": 0.6648, |
| "num_tokens": 266242185.0, |
| "step": 695 |
| }, |
| { |
| "epoch": 0.4750853242320819, |
| "grad_norm": 0.09375, |
| "learning_rate": 3.126799540390622e-05, |
| "loss": 0.6932, |
| "num_tokens": 266600989.0, |
| "step": 696 |
| }, |
| { |
| "epoch": 0.4757679180887372, |
| "grad_norm": 0.08984375, |
| "learning_rate": 3.1217886919605585e-05, |
| "loss": 0.6252, |
| "num_tokens": 266976693.0, |
| "step": 697 |
| }, |
| { |
| "epoch": 0.4764505119453925, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 3.116775947080146e-05, |
| "loss": 0.6467, |
| "num_tokens": 267376602.0, |
| "step": 698 |
| }, |
| { |
| "epoch": 0.47713310580204776, |
| "grad_norm": 0.09375, |
| "learning_rate": 3.111761331318804e-05, |
| "loss": 0.6547, |
| "num_tokens": 267752773.0, |
| "step": 699 |
| }, |
| { |
| "epoch": 0.4778156996587031, |
| "grad_norm": 0.091796875, |
| "learning_rate": 3.1067448702554976e-05, |
| "loss": 0.6575, |
| "num_tokens": 268142845.0, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.4784982935153584, |
| "grad_norm": 0.0986328125, |
| "learning_rate": 3.101726589478604e-05, |
| "loss": 0.688, |
| "num_tokens": 268524027.0, |
| "step": 701 |
| }, |
| { |
| "epoch": 0.47918088737201364, |
| "grad_norm": 0.087890625, |
| "learning_rate": 3.0967065145857835e-05, |
| "loss": 0.6672, |
| "num_tokens": 268905971.0, |
| "step": 702 |
| }, |
| { |
| "epoch": 0.47986348122866895, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 3.091684671183845e-05, |
| "loss": 0.681, |
| "num_tokens": 269318397.0, |
| "step": 703 |
| }, |
| { |
| "epoch": 0.4805460750853242, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 3.086661084888624e-05, |
| "loss": 0.6421, |
| "num_tokens": 269703965.0, |
| "step": 704 |
| }, |
| { |
| "epoch": 0.4812286689419795, |
| "grad_norm": 0.09375, |
| "learning_rate": 3.08163578132484e-05, |
| "loss": 0.6716, |
| "num_tokens": 270078357.0, |
| "step": 705 |
| }, |
| { |
| "epoch": 0.48191126279863483, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 3.076608786125976e-05, |
| "loss": 0.6529, |
| "num_tokens": 270480666.0, |
| "step": 706 |
| }, |
| { |
| "epoch": 0.4825938566552901, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 3.0715801249341425e-05, |
| "loss": 0.6396, |
| "num_tokens": 270867450.0, |
| "step": 707 |
| }, |
| { |
| "epoch": 0.4832764505119454, |
| "grad_norm": 0.10009765625, |
| "learning_rate": 3.0665498233999487e-05, |
| "loss": 0.6981, |
| "num_tokens": 271245798.0, |
| "step": 708 |
| }, |
| { |
| "epoch": 0.48395904436860065, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 3.06151790718237e-05, |
| "loss": 0.6785, |
| "num_tokens": 271639559.0, |
| "step": 709 |
| }, |
| { |
| "epoch": 0.48464163822525597, |
| "grad_norm": 0.091796875, |
| "learning_rate": 3.056484401948618e-05, |
| "loss": 0.6435, |
| "num_tokens": 272018748.0, |
| "step": 710 |
| }, |
| { |
| "epoch": 0.4853242320819113, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 3.0514493333740117e-05, |
| "loss": 0.693, |
| "num_tokens": 272436671.0, |
| "step": 711 |
| }, |
| { |
| "epoch": 0.48600682593856653, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 3.0464127271418426e-05, |
| "loss": 0.6441, |
| "num_tokens": 272840494.0, |
| "step": 712 |
| }, |
| { |
| "epoch": 0.48668941979522184, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 3.041374608943246e-05, |
| "loss": 0.6693, |
| "num_tokens": 273202534.0, |
| "step": 713 |
| }, |
| { |
| "epoch": 0.48737201365187716, |
| "grad_norm": 0.0859375, |
| "learning_rate": 3.0363350044770707e-05, |
| "loss": 0.6883, |
| "num_tokens": 273636834.0, |
| "step": 714 |
| }, |
| { |
| "epoch": 0.4880546075085324, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 3.0312939394497442e-05, |
| "loss": 0.6778, |
| "num_tokens": 274027158.0, |
| "step": 715 |
| }, |
| { |
| "epoch": 0.4887372013651877, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 3.0262514395751474e-05, |
| "loss": 0.6646, |
| "num_tokens": 274447377.0, |
| "step": 716 |
| }, |
| { |
| "epoch": 0.489419795221843, |
| "grad_norm": 0.09375, |
| "learning_rate": 3.021207530574477e-05, |
| "loss": 0.6272, |
| "num_tokens": 274790151.0, |
| "step": 717 |
| }, |
| { |
| "epoch": 0.4901023890784983, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 3.01616223817612e-05, |
| "loss": 0.6882, |
| "num_tokens": 275165972.0, |
| "step": 718 |
| }, |
| { |
| "epoch": 0.4907849829351536, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 3.0111155881155196e-05, |
| "loss": 0.6475, |
| "num_tokens": 275546853.0, |
| "step": 719 |
| }, |
| { |
| "epoch": 0.49146757679180886, |
| "grad_norm": 0.09375, |
| "learning_rate": 3.0060676061350423e-05, |
| "loss": 0.7032, |
| "num_tokens": 275897357.0, |
| "step": 720 |
| }, |
| { |
| "epoch": 0.49215017064846417, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 3.0010183179838508e-05, |
| "loss": 0.6526, |
| "num_tokens": 276256992.0, |
| "step": 721 |
| }, |
| { |
| "epoch": 0.4928327645051195, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 2.9959677494177694e-05, |
| "loss": 0.606, |
| "num_tokens": 276604249.0, |
| "step": 722 |
| }, |
| { |
| "epoch": 0.49351535836177474, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 2.9909159261991525e-05, |
| "loss": 0.6393, |
| "num_tokens": 276930026.0, |
| "step": 723 |
| }, |
| { |
| "epoch": 0.49419795221843005, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 2.9858628740967582e-05, |
| "loss": 0.6217, |
| "num_tokens": 277317649.0, |
| "step": 724 |
| }, |
| { |
| "epoch": 0.4948805460750853, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 2.9808086188856084e-05, |
| "loss": 0.6497, |
| "num_tokens": 277733458.0, |
| "step": 725 |
| }, |
| { |
| "epoch": 0.4955631399317406, |
| "grad_norm": 0.091796875, |
| "learning_rate": 2.9757531863468645e-05, |
| "loss": 0.6723, |
| "num_tokens": 278094118.0, |
| "step": 726 |
| }, |
| { |
| "epoch": 0.4962457337883959, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 2.970696602267693e-05, |
| "loss": 0.6722, |
| "num_tokens": 278473395.0, |
| "step": 727 |
| }, |
| { |
| "epoch": 0.4969283276450512, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 2.965638892441133e-05, |
| "loss": 0.6326, |
| "num_tokens": 278820111.0, |
| "step": 728 |
| }, |
| { |
| "epoch": 0.4976109215017065, |
| "grad_norm": 0.087890625, |
| "learning_rate": 2.9605800826659698e-05, |
| "loss": 0.6694, |
| "num_tokens": 279209545.0, |
| "step": 729 |
| }, |
| { |
| "epoch": 0.49829351535836175, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 2.9555201987465932e-05, |
| "loss": 0.6479, |
| "num_tokens": 279550428.0, |
| "step": 730 |
| }, |
| { |
| "epoch": 0.49897610921501706, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 2.9504592664928765e-05, |
| "loss": 0.6447, |
| "num_tokens": 279948393.0, |
| "step": 731 |
| }, |
| { |
| "epoch": 0.49965870307167237, |
| "grad_norm": 0.0830078125, |
| "learning_rate": 2.94539731172004e-05, |
| "loss": 0.647, |
| "num_tokens": 280381456.0, |
| "step": 732 |
| }, |
| { |
| "epoch": 0.5003412969283276, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 2.9403343602485185e-05, |
| "loss": 0.6138, |
| "num_tokens": 280766082.0, |
| "step": 733 |
| }, |
| { |
| "epoch": 0.5010238907849829, |
| "grad_norm": 0.09716796875, |
| "learning_rate": 2.935270437903832e-05, |
| "loss": 0.6664, |
| "num_tokens": 281109538.0, |
| "step": 734 |
| }, |
| { |
| "epoch": 0.5017064846416383, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 2.9302055705164516e-05, |
| "loss": 0.6366, |
| "num_tokens": 281502549.0, |
| "step": 735 |
| }, |
| { |
| "epoch": 0.5023890784982935, |
| "grad_norm": 0.091796875, |
| "learning_rate": 2.9251397839216704e-05, |
| "loss": 0.6624, |
| "num_tokens": 281904780.0, |
| "step": 736 |
| }, |
| { |
| "epoch": 0.5030716723549488, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 2.9200731039594682e-05, |
| "loss": 0.6703, |
| "num_tokens": 282329702.0, |
| "step": 737 |
| }, |
| { |
| "epoch": 0.5037542662116041, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 2.915005556474384e-05, |
| "loss": 0.6719, |
| "num_tokens": 282759680.0, |
| "step": 738 |
| }, |
| { |
| "epoch": 0.5044368600682594, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 2.909937167315381e-05, |
| "loss": 0.654, |
| "num_tokens": 283098514.0, |
| "step": 739 |
| }, |
| { |
| "epoch": 0.5051194539249146, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 2.9048679623357155e-05, |
| "loss": 0.628, |
| "num_tokens": 283508120.0, |
| "step": 740 |
| }, |
| { |
| "epoch": 0.50580204778157, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 2.899797967392805e-05, |
| "loss": 0.6217, |
| "num_tokens": 283877954.0, |
| "step": 741 |
| }, |
| { |
| "epoch": 0.5064846416382253, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 2.8947272083480975e-05, |
| "loss": 0.6151, |
| "num_tokens": 284260339.0, |
| "step": 742 |
| }, |
| { |
| "epoch": 0.5071672354948805, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 2.8896557110669364e-05, |
| "loss": 0.6636, |
| "num_tokens": 284645235.0, |
| "step": 743 |
| }, |
| { |
| "epoch": 0.5078498293515359, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 2.8845835014184346e-05, |
| "loss": 0.6533, |
| "num_tokens": 285059652.0, |
| "step": 744 |
| }, |
| { |
| "epoch": 0.5085324232081911, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 2.8795106052753356e-05, |
| "loss": 0.694, |
| "num_tokens": 285480194.0, |
| "step": 745 |
| }, |
| { |
| "epoch": 0.5092150170648464, |
| "grad_norm": 0.09912109375, |
| "learning_rate": 2.8744370485138846e-05, |
| "loss": 0.66, |
| "num_tokens": 285866364.0, |
| "step": 746 |
| }, |
| { |
| "epoch": 0.5098976109215017, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 2.8693628570136978e-05, |
| "loss": 0.6513, |
| "num_tokens": 286255756.0, |
| "step": 747 |
| }, |
| { |
| "epoch": 0.510580204778157, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 2.8642880566576298e-05, |
| "loss": 0.6365, |
| "num_tokens": 286620725.0, |
| "step": 748 |
| }, |
| { |
| "epoch": 0.5112627986348123, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 2.859212673331637e-05, |
| "loss": 0.6934, |
| "num_tokens": 286970996.0, |
| "step": 749 |
| }, |
| { |
| "epoch": 0.5119453924914675, |
| "grad_norm": 0.08349609375, |
| "learning_rate": 2.854136732924656e-05, |
| "loss": 0.6921, |
| "num_tokens": 287419124.0, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.5126279863481229, |
| "grad_norm": 0.091796875, |
| "learning_rate": 2.849060261328458e-05, |
| "loss": 0.6315, |
| "num_tokens": 287777603.0, |
| "step": 751 |
| }, |
| { |
| "epoch": 0.5133105802047782, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 2.8439832844375286e-05, |
| "loss": 0.6103, |
| "num_tokens": 288153082.0, |
| "step": 752 |
| }, |
| { |
| "epoch": 0.5139931740614334, |
| "grad_norm": 0.09814453125, |
| "learning_rate": 2.83890582814893e-05, |
| "loss": 0.632, |
| "num_tokens": 288504661.0, |
| "step": 753 |
| }, |
| { |
| "epoch": 0.5146757679180888, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 2.8338279183621668e-05, |
| "loss": 0.6239, |
| "num_tokens": 288900643.0, |
| "step": 754 |
| }, |
| { |
| "epoch": 0.515358361774744, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 2.828749580979061e-05, |
| "loss": 0.6659, |
| "num_tokens": 289306762.0, |
| "step": 755 |
| }, |
| { |
| "epoch": 0.5160409556313993, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 2.8236708419036136e-05, |
| "loss": 0.6515, |
| "num_tokens": 289684165.0, |
| "step": 756 |
| }, |
| { |
| "epoch": 0.5167235494880547, |
| "grad_norm": 0.08984375, |
| "learning_rate": 2.8185917270418728e-05, |
| "loss": 0.6557, |
| "num_tokens": 290051136.0, |
| "step": 757 |
| }, |
| { |
| "epoch": 0.5174061433447099, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 2.8135122623018084e-05, |
| "loss": 0.6383, |
| "num_tokens": 290409489.0, |
| "step": 758 |
| }, |
| { |
| "epoch": 0.5180887372013652, |
| "grad_norm": 0.099609375, |
| "learning_rate": 2.8084324735931695e-05, |
| "loss": 0.6591, |
| "num_tokens": 290746721.0, |
| "step": 759 |
| }, |
| { |
| "epoch": 0.5187713310580204, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 2.8033523868273625e-05, |
| "loss": 0.6567, |
| "num_tokens": 291105578.0, |
| "step": 760 |
| }, |
| { |
| "epoch": 0.5194539249146758, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 2.7982720279173108e-05, |
| "loss": 0.7089, |
| "num_tokens": 291473492.0, |
| "step": 761 |
| }, |
| { |
| "epoch": 0.520136518771331, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 2.793191422777327e-05, |
| "loss": 0.6712, |
| "num_tokens": 291881581.0, |
| "step": 762 |
| }, |
| { |
| "epoch": 0.5208191126279863, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 2.7881105973229804e-05, |
| "loss": 0.6241, |
| "num_tokens": 292279908.0, |
| "step": 763 |
| }, |
| { |
| "epoch": 0.5215017064846417, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 2.7830295774709625e-05, |
| "loss": 0.6499, |
| "num_tokens": 292678201.0, |
| "step": 764 |
| }, |
| { |
| "epoch": 0.5221843003412969, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 2.777948389138959e-05, |
| "loss": 0.6493, |
| "num_tokens": 293082816.0, |
| "step": 765 |
| }, |
| { |
| "epoch": 0.5228668941979522, |
| "grad_norm": 0.08984375, |
| "learning_rate": 2.7728670582455124e-05, |
| "loss": 0.6733, |
| "num_tokens": 293461025.0, |
| "step": 766 |
| }, |
| { |
| "epoch": 0.5235494880546075, |
| "grad_norm": 0.09716796875, |
| "learning_rate": 2.767785610709894e-05, |
| "loss": 0.6301, |
| "num_tokens": 293786213.0, |
| "step": 767 |
| }, |
| { |
| "epoch": 0.5242320819112628, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 2.762704072451968e-05, |
| "loss": 0.6543, |
| "num_tokens": 294170957.0, |
| "step": 768 |
| }, |
| { |
| "epoch": 0.5249146757679181, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 2.757622469392065e-05, |
| "loss": 0.6506, |
| "num_tokens": 294579183.0, |
| "step": 769 |
| }, |
| { |
| "epoch": 0.5255972696245734, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 2.7525408274508424e-05, |
| "loss": 0.6409, |
| "num_tokens": 294930434.0, |
| "step": 770 |
| }, |
| { |
| "epoch": 0.5262798634812287, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 2.747459172549158e-05, |
| "loss": 0.6766, |
| "num_tokens": 295342976.0, |
| "step": 771 |
| }, |
| { |
| "epoch": 0.5269624573378839, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 2.7423775306079362e-05, |
| "loss": 0.6648, |
| "num_tokens": 295707641.0, |
| "step": 772 |
| }, |
| { |
| "epoch": 0.5276450511945392, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 2.7372959275480327e-05, |
| "loss": 0.6194, |
| "num_tokens": 296052149.0, |
| "step": 773 |
| }, |
| { |
| "epoch": 0.5283276450511946, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 2.732214389290107e-05, |
| "loss": 0.6576, |
| "num_tokens": 296448682.0, |
| "step": 774 |
| }, |
| { |
| "epoch": 0.5290102389078498, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 2.727132941754488e-05, |
| "loss": 0.671, |
| "num_tokens": 296842807.0, |
| "step": 775 |
| }, |
| { |
| "epoch": 0.5296928327645051, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 2.7220516108610412e-05, |
| "loss": 0.6631, |
| "num_tokens": 297240581.0, |
| "step": 776 |
| }, |
| { |
| "epoch": 0.5303754266211604, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 2.7169704225290384e-05, |
| "loss": 0.6653, |
| "num_tokens": 297674896.0, |
| "step": 777 |
| }, |
| { |
| "epoch": 0.5310580204778157, |
| "grad_norm": 0.0986328125, |
| "learning_rate": 2.7118894026770208e-05, |
| "loss": 0.6314, |
| "num_tokens": 298026899.0, |
| "step": 778 |
| }, |
| { |
| "epoch": 0.531740614334471, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 2.7068085772226736e-05, |
| "loss": 0.6692, |
| "num_tokens": 298392971.0, |
| "step": 779 |
| }, |
| { |
| "epoch": 0.5324232081911263, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 2.70172797208269e-05, |
| "loss": 0.6553, |
| "num_tokens": 298726998.0, |
| "step": 780 |
| }, |
| { |
| "epoch": 0.5331058020477816, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 2.696647613172638e-05, |
| "loss": 0.6585, |
| "num_tokens": 299129920.0, |
| "step": 781 |
| }, |
| { |
| "epoch": 0.5337883959044368, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 2.6915675264068307e-05, |
| "loss": 0.6686, |
| "num_tokens": 299512924.0, |
| "step": 782 |
| }, |
| { |
| "epoch": 0.5344709897610922, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 2.686487737698192e-05, |
| "loss": 0.6779, |
| "num_tokens": 299952331.0, |
| "step": 783 |
| }, |
| { |
| "epoch": 0.5351535836177475, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 2.6814082729581274e-05, |
| "loss": 0.6139, |
| "num_tokens": 300291986.0, |
| "step": 784 |
| }, |
| { |
| "epoch": 0.5358361774744027, |
| "grad_norm": 0.087890625, |
| "learning_rate": 2.676329158096388e-05, |
| "loss": 0.616, |
| "num_tokens": 300660125.0, |
| "step": 785 |
| }, |
| { |
| "epoch": 0.5365187713310581, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 2.6712504190209393e-05, |
| "loss": 0.674, |
| "num_tokens": 300999528.0, |
| "step": 786 |
| }, |
| { |
| "epoch": 0.5372013651877133, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 2.6661720816378338e-05, |
| "loss": 0.6306, |
| "num_tokens": 301331664.0, |
| "step": 787 |
| }, |
| { |
| "epoch": 0.5378839590443686, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 2.6610941718510705e-05, |
| "loss": 0.6245, |
| "num_tokens": 301709973.0, |
| "step": 788 |
| }, |
| { |
| "epoch": 0.5385665529010238, |
| "grad_norm": 0.091796875, |
| "learning_rate": 2.6560167155624716e-05, |
| "loss": 0.6026, |
| "num_tokens": 302072602.0, |
| "step": 789 |
| }, |
| { |
| "epoch": 0.5392491467576792, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 2.650939738671543e-05, |
| "loss": 0.6382, |
| "num_tokens": 302437719.0, |
| "step": 790 |
| }, |
| { |
| "epoch": 0.5399317406143345, |
| "grad_norm": 0.15625, |
| "learning_rate": 2.6458632670753457e-05, |
| "loss": 0.6726, |
| "num_tokens": 302835101.0, |
| "step": 791 |
| }, |
| { |
| "epoch": 0.5406143344709897, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 2.6407873266683635e-05, |
| "loss": 0.6644, |
| "num_tokens": 303226079.0, |
| "step": 792 |
| }, |
| { |
| "epoch": 0.5412969283276451, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 2.6357119433423717e-05, |
| "loss": 0.6988, |
| "num_tokens": 303656303.0, |
| "step": 793 |
| }, |
| { |
| "epoch": 0.5419795221843003, |
| "grad_norm": 0.103515625, |
| "learning_rate": 2.6306371429863024e-05, |
| "loss": 0.6787, |
| "num_tokens": 303974277.0, |
| "step": 794 |
| }, |
| { |
| "epoch": 0.5426621160409556, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 2.6255629514861156e-05, |
| "loss": 0.6207, |
| "num_tokens": 304343518.0, |
| "step": 795 |
| }, |
| { |
| "epoch": 0.543344709897611, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 2.620489394724665e-05, |
| "loss": 0.6825, |
| "num_tokens": 304708601.0, |
| "step": 796 |
| }, |
| { |
| "epoch": 0.5440273037542662, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 2.6154164985815656e-05, |
| "loss": 0.6768, |
| "num_tokens": 305105643.0, |
| "step": 797 |
| }, |
| { |
| "epoch": 0.5447098976109215, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 2.610344288933064e-05, |
| "loss": 0.6513, |
| "num_tokens": 305446546.0, |
| "step": 798 |
| }, |
| { |
| "epoch": 0.5453924914675768, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 2.6052727916519033e-05, |
| "loss": 0.6825, |
| "num_tokens": 305866540.0, |
| "step": 799 |
| }, |
| { |
| "epoch": 0.5460750853242321, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 2.6002020326071958e-05, |
| "loss": 0.6192, |
| "num_tokens": 306180961.0, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.5467576791808874, |
| "grad_norm": 0.08984375, |
| "learning_rate": 2.595132037664285e-05, |
| "loss": 0.6664, |
| "num_tokens": 306579502.0, |
| "step": 801 |
| }, |
| { |
| "epoch": 0.5474402730375426, |
| "grad_norm": 0.08984375, |
| "learning_rate": 2.5900628326846193e-05, |
| "loss": 0.6773, |
| "num_tokens": 306975404.0, |
| "step": 802 |
| }, |
| { |
| "epoch": 0.548122866894198, |
| "grad_norm": 0.09375, |
| "learning_rate": 2.584994443525617e-05, |
| "loss": 0.7199, |
| "num_tokens": 307342635.0, |
| "step": 803 |
| }, |
| { |
| "epoch": 0.5488054607508532, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 2.5799268960405327e-05, |
| "loss": 0.6419, |
| "num_tokens": 307690583.0, |
| "step": 804 |
| }, |
| { |
| "epoch": 0.5494880546075085, |
| "grad_norm": 0.091796875, |
| "learning_rate": 2.5748602160783308e-05, |
| "loss": 0.6385, |
| "num_tokens": 308076053.0, |
| "step": 805 |
| }, |
| { |
| "epoch": 0.5501706484641639, |
| "grad_norm": 0.1005859375, |
| "learning_rate": 2.5697944294835486e-05, |
| "loss": 0.6979, |
| "num_tokens": 308452506.0, |
| "step": 806 |
| }, |
| { |
| "epoch": 0.5508532423208191, |
| "grad_norm": 0.091796875, |
| "learning_rate": 2.5647295620961687e-05, |
| "loss": 0.6404, |
| "num_tokens": 308834833.0, |
| "step": 807 |
| }, |
| { |
| "epoch": 0.5515358361774744, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 2.5596656397514817e-05, |
| "loss": 0.6608, |
| "num_tokens": 309233628.0, |
| "step": 808 |
| }, |
| { |
| "epoch": 0.5522184300341297, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 2.5546026882799605e-05, |
| "loss": 0.6558, |
| "num_tokens": 309667811.0, |
| "step": 809 |
| }, |
| { |
| "epoch": 0.552901023890785, |
| "grad_norm": 0.095703125, |
| "learning_rate": 2.5495407335071247e-05, |
| "loss": 0.6286, |
| "num_tokens": 310003006.0, |
| "step": 810 |
| }, |
| { |
| "epoch": 0.5535836177474402, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 2.5444798012534084e-05, |
| "loss": 0.6118, |
| "num_tokens": 310395308.0, |
| "step": 811 |
| }, |
| { |
| "epoch": 0.5542662116040956, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 2.539419917334031e-05, |
| "loss": 0.6562, |
| "num_tokens": 310778707.0, |
| "step": 812 |
| }, |
| { |
| "epoch": 0.5549488054607509, |
| "grad_norm": 0.09375, |
| "learning_rate": 2.534361107558867e-05, |
| "loss": 0.6592, |
| "num_tokens": 311133666.0, |
| "step": 813 |
| }, |
| { |
| "epoch": 0.5556313993174061, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 2.529303397732307e-05, |
| "loss": 0.6821, |
| "num_tokens": 311570174.0, |
| "step": 814 |
| }, |
| { |
| "epoch": 0.5563139931740614, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 2.5242468136531354e-05, |
| "loss": 0.616, |
| "num_tokens": 311908102.0, |
| "step": 815 |
| }, |
| { |
| "epoch": 0.5569965870307167, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 2.519191381114393e-05, |
| "loss": 0.6539, |
| "num_tokens": 312313029.0, |
| "step": 816 |
| }, |
| { |
| "epoch": 0.557679180887372, |
| "grad_norm": 0.099609375, |
| "learning_rate": 2.514137125903243e-05, |
| "loss": 0.6719, |
| "num_tokens": 312653211.0, |
| "step": 817 |
| }, |
| { |
| "epoch": 0.5583617747440273, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 2.509084073800848e-05, |
| "loss": 0.6368, |
| "num_tokens": 312986060.0, |
| "step": 818 |
| }, |
| { |
| "epoch": 0.5590443686006826, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 2.5040322505822318e-05, |
| "loss": 0.6357, |
| "num_tokens": 313375277.0, |
| "step": 819 |
| }, |
| { |
| "epoch": 0.5597269624573379, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 2.4989816820161497e-05, |
| "loss": 0.6583, |
| "num_tokens": 313777123.0, |
| "step": 820 |
| }, |
| { |
| "epoch": 0.5604095563139931, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 2.493932393864958e-05, |
| "loss": 0.6398, |
| "num_tokens": 314205571.0, |
| "step": 821 |
| }, |
| { |
| "epoch": 0.5610921501706485, |
| "grad_norm": 0.091796875, |
| "learning_rate": 2.4888844118844813e-05, |
| "loss": 0.6733, |
| "num_tokens": 314582075.0, |
| "step": 822 |
| }, |
| { |
| "epoch": 0.5617747440273038, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 2.4838377618238807e-05, |
| "loss": 0.6564, |
| "num_tokens": 314953513.0, |
| "step": 823 |
| }, |
| { |
| "epoch": 0.562457337883959, |
| "grad_norm": 0.091796875, |
| "learning_rate": 2.4787924694255237e-05, |
| "loss": 0.6129, |
| "num_tokens": 315341686.0, |
| "step": 824 |
| }, |
| { |
| "epoch": 0.5631399317406144, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 2.4737485604248534e-05, |
| "loss": 0.6487, |
| "num_tokens": 315742426.0, |
| "step": 825 |
| }, |
| { |
| "epoch": 0.5638225255972696, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 2.4687060605502566e-05, |
| "loss": 0.6838, |
| "num_tokens": 316110672.0, |
| "step": 826 |
| }, |
| { |
| "epoch": 0.5645051194539249, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 2.4636649955229302e-05, |
| "loss": 0.6732, |
| "num_tokens": 316526615.0, |
| "step": 827 |
| }, |
| { |
| "epoch": 0.5651877133105802, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 2.4586253910567537e-05, |
| "loss": 0.6495, |
| "num_tokens": 316915178.0, |
| "step": 828 |
| }, |
| { |
| "epoch": 0.5658703071672355, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 2.4535872728581583e-05, |
| "loss": 0.6378, |
| "num_tokens": 317244368.0, |
| "step": 829 |
| }, |
| { |
| "epoch": 0.5665529010238908, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 2.4485506666259895e-05, |
| "loss": 0.6423, |
| "num_tokens": 317577413.0, |
| "step": 830 |
| }, |
| { |
| "epoch": 0.567235494880546, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 2.4435155980513823e-05, |
| "loss": 0.7158, |
| "num_tokens": 317974088.0, |
| "step": 831 |
| }, |
| { |
| "epoch": 0.5679180887372014, |
| "grad_norm": 0.0859375, |
| "learning_rate": 2.438482092817631e-05, |
| "loss": 0.5908, |
| "num_tokens": 318358238.0, |
| "step": 832 |
| }, |
| { |
| "epoch": 0.5686006825938567, |
| "grad_norm": 0.0810546875, |
| "learning_rate": 2.4334501766000522e-05, |
| "loss": 0.6456, |
| "num_tokens": 318805556.0, |
| "step": 833 |
| }, |
| { |
| "epoch": 0.5692832764505119, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 2.4284198750658577e-05, |
| "loss": 0.6892, |
| "num_tokens": 319201775.0, |
| "step": 834 |
| }, |
| { |
| "epoch": 0.5699658703071673, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 2.4233912138740244e-05, |
| "loss": 0.6102, |
| "num_tokens": 319557900.0, |
| "step": 835 |
| }, |
| { |
| "epoch": 0.5706484641638225, |
| "grad_norm": 0.1005859375, |
| "learning_rate": 2.418364218675161e-05, |
| "loss": 0.6293, |
| "num_tokens": 319865284.0, |
| "step": 836 |
| }, |
| { |
| "epoch": 0.5713310580204778, |
| "grad_norm": 0.08349609375, |
| "learning_rate": 2.4133389151113773e-05, |
| "loss": 0.668, |
| "num_tokens": 320314209.0, |
| "step": 837 |
| }, |
| { |
| "epoch": 0.5720136518771332, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 2.408315328816155e-05, |
| "loss": 0.6982, |
| "num_tokens": 320704525.0, |
| "step": 838 |
| }, |
| { |
| "epoch": 0.5726962457337884, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 2.4032934854142174e-05, |
| "loss": 0.6382, |
| "num_tokens": 321096447.0, |
| "step": 839 |
| }, |
| { |
| "epoch": 0.5733788395904437, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 2.3982734105213967e-05, |
| "loss": 0.6374, |
| "num_tokens": 321462994.0, |
| "step": 840 |
| }, |
| { |
| "epoch": 0.5740614334470989, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 2.3932551297445026e-05, |
| "loss": 0.6401, |
| "num_tokens": 321832669.0, |
| "step": 841 |
| }, |
| { |
| "epoch": 0.5747440273037543, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 2.3882386686811968e-05, |
| "loss": 0.623, |
| "num_tokens": 322192193.0, |
| "step": 842 |
| }, |
| { |
| "epoch": 0.5754266211604095, |
| "grad_norm": 0.095703125, |
| "learning_rate": 2.3832240529198553e-05, |
| "loss": 0.6597, |
| "num_tokens": 322550077.0, |
| "step": 843 |
| }, |
| { |
| "epoch": 0.5761092150170648, |
| "grad_norm": 0.08984375, |
| "learning_rate": 2.3782113080394414e-05, |
| "loss": 0.6576, |
| "num_tokens": 322963579.0, |
| "step": 844 |
| }, |
| { |
| "epoch": 0.5767918088737202, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 2.373200459609379e-05, |
| "loss": 0.6644, |
| "num_tokens": 323355363.0, |
| "step": 845 |
| }, |
| { |
| "epoch": 0.5774744027303754, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 2.368191533189415e-05, |
| "loss": 0.6355, |
| "num_tokens": 323705445.0, |
| "step": 846 |
| }, |
| { |
| "epoch": 0.5781569965870307, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 2.3631845543294928e-05, |
| "loss": 0.6644, |
| "num_tokens": 324075379.0, |
| "step": 847 |
| }, |
| { |
| "epoch": 0.578839590443686, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 2.3581795485696235e-05, |
| "loss": 0.6919, |
| "num_tokens": 324473225.0, |
| "step": 848 |
| }, |
| { |
| "epoch": 0.5795221843003413, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 2.353176541439753e-05, |
| "loss": 0.6378, |
| "num_tokens": 324824406.0, |
| "step": 849 |
| }, |
| { |
| "epoch": 0.5802047781569966, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 2.348175558459628e-05, |
| "loss": 0.6162, |
| "num_tokens": 325155802.0, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.5808873720136519, |
| "grad_norm": 0.08984375, |
| "learning_rate": 2.3431766251386773e-05, |
| "loss": 0.6071, |
| "num_tokens": 325513601.0, |
| "step": 851 |
| }, |
| { |
| "epoch": 0.5815699658703072, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 2.3381797669758705e-05, |
| "loss": 0.6391, |
| "num_tokens": 325839096.0, |
| "step": 852 |
| }, |
| { |
| "epoch": 0.5822525597269624, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 2.333185009459592e-05, |
| "loss": 0.6671, |
| "num_tokens": 326213481.0, |
| "step": 853 |
| }, |
| { |
| "epoch": 0.5829351535836177, |
| "grad_norm": 0.09375, |
| "learning_rate": 2.3281923780675123e-05, |
| "loss": 0.6833, |
| "num_tokens": 326574240.0, |
| "step": 854 |
| }, |
| { |
| "epoch": 0.5836177474402731, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 2.323201898266458e-05, |
| "loss": 0.6844, |
| "num_tokens": 326956942.0, |
| "step": 855 |
| }, |
| { |
| "epoch": 0.5843003412969283, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 2.3182135955122776e-05, |
| "loss": 0.6453, |
| "num_tokens": 327385241.0, |
| "step": 856 |
| }, |
| { |
| "epoch": 0.5849829351535836, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 2.3132274952497153e-05, |
| "loss": 0.6914, |
| "num_tokens": 327805772.0, |
| "step": 857 |
| }, |
| { |
| "epoch": 0.5856655290102389, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 2.308243622912283e-05, |
| "loss": 0.6415, |
| "num_tokens": 328190019.0, |
| "step": 858 |
| }, |
| { |
| "epoch": 0.5863481228668942, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 2.3032620039221263e-05, |
| "loss": 0.6272, |
| "num_tokens": 328544118.0, |
| "step": 859 |
| }, |
| { |
| "epoch": 0.5870307167235495, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 2.2982826636898975e-05, |
| "loss": 0.6555, |
| "num_tokens": 328885222.0, |
| "step": 860 |
| }, |
| { |
| "epoch": 0.5877133105802048, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 2.2933056276146255e-05, |
| "loss": 0.6543, |
| "num_tokens": 329282375.0, |
| "step": 861 |
| }, |
| { |
| "epoch": 0.5883959044368601, |
| "grad_norm": 0.083984375, |
| "learning_rate": 2.2883309210835845e-05, |
| "loss": 0.6752, |
| "num_tokens": 329725499.0, |
| "step": 862 |
| }, |
| { |
| "epoch": 0.5890784982935153, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 2.2833585694721665e-05, |
| "loss": 0.6315, |
| "num_tokens": 330121174.0, |
| "step": 863 |
| }, |
| { |
| "epoch": 0.5897610921501707, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 2.2783885981437526e-05, |
| "loss": 0.6403, |
| "num_tokens": 330470508.0, |
| "step": 864 |
| }, |
| { |
| "epoch": 0.590443686006826, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 2.2734210324495812e-05, |
| "loss": 0.6529, |
| "num_tokens": 330878184.0, |
| "step": 865 |
| }, |
| { |
| "epoch": 0.5911262798634812, |
| "grad_norm": 0.08251953125, |
| "learning_rate": 2.2684558977286202e-05, |
| "loss": 0.624, |
| "num_tokens": 331302118.0, |
| "step": 866 |
| }, |
| { |
| "epoch": 0.5918088737201365, |
| "grad_norm": 0.09765625, |
| "learning_rate": 2.263493219307437e-05, |
| "loss": 0.5808, |
| "num_tokens": 331624127.0, |
| "step": 867 |
| }, |
| { |
| "epoch": 0.5924914675767918, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 2.258533022500071e-05, |
| "loss": 0.6358, |
| "num_tokens": 332053832.0, |
| "step": 868 |
| }, |
| { |
| "epoch": 0.5931740614334471, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 2.253575332607901e-05, |
| "loss": 0.634, |
| "num_tokens": 332380991.0, |
| "step": 869 |
| }, |
| { |
| "epoch": 0.5938566552901023, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 2.24862017491952e-05, |
| "loss": 0.6937, |
| "num_tokens": 332821824.0, |
| "step": 870 |
| }, |
| { |
| "epoch": 0.5945392491467577, |
| "grad_norm": 0.091796875, |
| "learning_rate": 2.2436675747106044e-05, |
| "loss": 0.656, |
| "num_tokens": 333207990.0, |
| "step": 871 |
| }, |
| { |
| "epoch": 0.595221843003413, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 2.2387175572437842e-05, |
| "loss": 0.6607, |
| "num_tokens": 333620629.0, |
| "step": 872 |
| }, |
| { |
| "epoch": 0.5959044368600682, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 2.2337701477685153e-05, |
| "loss": 0.6498, |
| "num_tokens": 333995540.0, |
| "step": 873 |
| }, |
| { |
| "epoch": 0.5965870307167236, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 2.2288253715209528e-05, |
| "loss": 0.6498, |
| "num_tokens": 334388364.0, |
| "step": 874 |
| }, |
| { |
| "epoch": 0.5972696245733788, |
| "grad_norm": 0.087890625, |
| "learning_rate": 2.2238832537238174e-05, |
| "loss": 0.6726, |
| "num_tokens": 334798404.0, |
| "step": 875 |
| }, |
| { |
| "epoch": 0.5979522184300341, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 2.2189438195862693e-05, |
| "loss": 0.6281, |
| "num_tokens": 335133840.0, |
| "step": 876 |
| }, |
| { |
| "epoch": 0.5986348122866895, |
| "grad_norm": 0.09375, |
| "learning_rate": 2.2140070943037816e-05, |
| "loss": 0.6564, |
| "num_tokens": 335503527.0, |
| "step": 877 |
| }, |
| { |
| "epoch": 0.5993174061433447, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 2.209073103058008e-05, |
| "loss": 0.6197, |
| "num_tokens": 335917253.0, |
| "step": 878 |
| }, |
| { |
| "epoch": 0.6, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 2.2041418710166577e-05, |
| "loss": 0.609, |
| "num_tokens": 336302485.0, |
| "step": 879 |
| }, |
| { |
| "epoch": 0.6006825938566553, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 2.199213423333365e-05, |
| "loss": 0.6209, |
| "num_tokens": 336651049.0, |
| "step": 880 |
| }, |
| { |
| "epoch": 0.6013651877133106, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 2.194287785147561e-05, |
| "loss": 0.6289, |
| "num_tokens": 336991723.0, |
| "step": 881 |
| }, |
| { |
| "epoch": 0.6020477815699659, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 2.1893649815843463e-05, |
| "loss": 0.6293, |
| "num_tokens": 337338753.0, |
| "step": 882 |
| }, |
| { |
| "epoch": 0.6027303754266211, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 2.184445037754362e-05, |
| "loss": 0.6513, |
| "num_tokens": 337723146.0, |
| "step": 883 |
| }, |
| { |
| "epoch": 0.6034129692832765, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 2.1795279787536625e-05, |
| "loss": 0.6923, |
| "num_tokens": 338103395.0, |
| "step": 884 |
| }, |
| { |
| "epoch": 0.6040955631399317, |
| "grad_norm": 0.0986328125, |
| "learning_rate": 2.174613829663587e-05, |
| "loss": 0.6568, |
| "num_tokens": 338438000.0, |
| "step": 885 |
| }, |
| { |
| "epoch": 0.604778156996587, |
| "grad_norm": 0.087890625, |
| "learning_rate": 2.1697026155506318e-05, |
| "loss": 0.7018, |
| "num_tokens": 338875474.0, |
| "step": 886 |
| }, |
| { |
| "epoch": 0.6054607508532424, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 2.1647943614663225e-05, |
| "loss": 0.6373, |
| "num_tokens": 339275347.0, |
| "step": 887 |
| }, |
| { |
| "epoch": 0.6061433447098976, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 2.1598890924470844e-05, |
| "loss": 0.6617, |
| "num_tokens": 339662984.0, |
| "step": 888 |
| }, |
| { |
| "epoch": 0.6068259385665529, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 2.1549868335141167e-05, |
| "loss": 0.6489, |
| "num_tokens": 339993893.0, |
| "step": 889 |
| }, |
| { |
| "epoch": 0.6075085324232082, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 2.1500876096732664e-05, |
| "loss": 0.6489, |
| "num_tokens": 340414967.0, |
| "step": 890 |
| }, |
| { |
| "epoch": 0.6081911262798635, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 2.1451914459148977e-05, |
| "loss": 0.6689, |
| "num_tokens": 340803552.0, |
| "step": 891 |
| }, |
| { |
| "epoch": 0.6088737201365187, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 2.140298367213764e-05, |
| "loss": 0.6753, |
| "num_tokens": 341167229.0, |
| "step": 892 |
| }, |
| { |
| "epoch": 0.6095563139931741, |
| "grad_norm": 0.0986328125, |
| "learning_rate": 2.1354083985288846e-05, |
| "loss": 0.688, |
| "num_tokens": 341512610.0, |
| "step": 893 |
| }, |
| { |
| "epoch": 0.6102389078498294, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 2.130521564803416e-05, |
| "loss": 0.6472, |
| "num_tokens": 341904114.0, |
| "step": 894 |
| }, |
| { |
| "epoch": 0.6109215017064846, |
| "grad_norm": 0.091796875, |
| "learning_rate": 2.1256378909645187e-05, |
| "loss": 0.6209, |
| "num_tokens": 342244656.0, |
| "step": 895 |
| }, |
| { |
| "epoch": 0.6116040955631399, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 2.1207574019232397e-05, |
| "loss": 0.6398, |
| "num_tokens": 342631176.0, |
| "step": 896 |
| }, |
| { |
| "epoch": 0.6122866894197952, |
| "grad_norm": 0.10009765625, |
| "learning_rate": 2.1158801225743785e-05, |
| "loss": 0.6242, |
| "num_tokens": 343018919.0, |
| "step": 897 |
| }, |
| { |
| "epoch": 0.6129692832764505, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 2.111006077796363e-05, |
| "loss": 0.6382, |
| "num_tokens": 343447603.0, |
| "step": 898 |
| }, |
| { |
| "epoch": 0.6136518771331058, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 2.1061352924511217e-05, |
| "loss": 0.652, |
| "num_tokens": 343854813.0, |
| "step": 899 |
| }, |
| { |
| "epoch": 0.6143344709897611, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 2.101267791383959e-05, |
| "loss": 0.6862, |
| "num_tokens": 344295726.0, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.6150170648464164, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 2.0964035994234223e-05, |
| "loss": 0.6396, |
| "num_tokens": 344637344.0, |
| "step": 901 |
| }, |
| { |
| "epoch": 0.6156996587030716, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 2.0915427413811834e-05, |
| "loss": 0.6536, |
| "num_tokens": 345056526.0, |
| "step": 902 |
| }, |
| { |
| "epoch": 0.616382252559727, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 2.086685242051906e-05, |
| "loss": 0.697, |
| "num_tokens": 345488874.0, |
| "step": 903 |
| }, |
| { |
| "epoch": 0.6170648464163823, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 2.081831126213124e-05, |
| "loss": 0.6849, |
| "num_tokens": 345892417.0, |
| "step": 904 |
| }, |
| { |
| "epoch": 0.6177474402730375, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 2.0769804186251096e-05, |
| "loss": 0.6307, |
| "num_tokens": 346263433.0, |
| "step": 905 |
| }, |
| { |
| "epoch": 0.6184300341296929, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 2.0721331440307512e-05, |
| "loss": 0.6905, |
| "num_tokens": 346640671.0, |
| "step": 906 |
| }, |
| { |
| "epoch": 0.6191126279863481, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 2.067289327155427e-05, |
| "loss": 0.652, |
| "num_tokens": 347026155.0, |
| "step": 907 |
| }, |
| { |
| "epoch": 0.6197952218430034, |
| "grad_norm": 0.087890625, |
| "learning_rate": 2.0624489927068764e-05, |
| "loss": 0.6245, |
| "num_tokens": 347400789.0, |
| "step": 908 |
| }, |
| { |
| "epoch": 0.6204778156996587, |
| "grad_norm": 0.091796875, |
| "learning_rate": 2.057612165375074e-05, |
| "loss": 0.6787, |
| "num_tokens": 347768749.0, |
| "step": 909 |
| }, |
| { |
| "epoch": 0.621160409556314, |
| "grad_norm": 0.08984375, |
| "learning_rate": 2.052778869832108e-05, |
| "loss": 0.6355, |
| "num_tokens": 348137882.0, |
| "step": 910 |
| }, |
| { |
| "epoch": 0.6218430034129693, |
| "grad_norm": 0.08984375, |
| "learning_rate": 2.0479491307320502e-05, |
| "loss": 0.6904, |
| "num_tokens": 348521681.0, |
| "step": 911 |
| }, |
| { |
| "epoch": 0.6225255972696245, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 2.0431229727108303e-05, |
| "loss": 0.663, |
| "num_tokens": 348922473.0, |
| "step": 912 |
| }, |
| { |
| "epoch": 0.6232081911262799, |
| "grad_norm": 0.095703125, |
| "learning_rate": 2.0383004203861134e-05, |
| "loss": 0.6542, |
| "num_tokens": 349251762.0, |
| "step": 913 |
| }, |
| { |
| "epoch": 0.6238907849829352, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 2.03348149835717e-05, |
| "loss": 0.6403, |
| "num_tokens": 349602080.0, |
| "step": 914 |
| }, |
| { |
| "epoch": 0.6245733788395904, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 2.0286662312047537e-05, |
| "loss": 0.6718, |
| "num_tokens": 349987787.0, |
| "step": 915 |
| }, |
| { |
| "epoch": 0.6252559726962458, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 2.0238546434909773e-05, |
| "loss": 0.6195, |
| "num_tokens": 350337330.0, |
| "step": 916 |
| }, |
| { |
| "epoch": 0.625938566552901, |
| "grad_norm": 0.0859375, |
| "learning_rate": 2.0190467597591816e-05, |
| "loss": 0.6676, |
| "num_tokens": 350777005.0, |
| "step": 917 |
| }, |
| { |
| "epoch": 0.6266211604095563, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 2.0142426045338163e-05, |
| "loss": 0.678, |
| "num_tokens": 351147173.0, |
| "step": 918 |
| }, |
| { |
| "epoch": 0.6273037542662117, |
| "grad_norm": 0.087890625, |
| "learning_rate": 2.0094422023203114e-05, |
| "loss": 0.6152, |
| "num_tokens": 351513334.0, |
| "step": 919 |
| }, |
| { |
| "epoch": 0.6279863481228669, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 2.0046455776049545e-05, |
| "loss": 0.6113, |
| "num_tokens": 351875623.0, |
| "step": 920 |
| }, |
| { |
| "epoch": 0.6286689419795222, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 1.9998527548547626e-05, |
| "loss": 0.6319, |
| "num_tokens": 352254072.0, |
| "step": 921 |
| }, |
| { |
| "epoch": 0.6293515358361774, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 1.9950637585173592e-05, |
| "loss": 0.6526, |
| "num_tokens": 352670878.0, |
| "step": 922 |
| }, |
| { |
| "epoch": 0.6300341296928328, |
| "grad_norm": 0.09716796875, |
| "learning_rate": 1.9902786130208527e-05, |
| "loss": 0.645, |
| "num_tokens": 353018417.0, |
| "step": 923 |
| }, |
| { |
| "epoch": 0.630716723549488, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 1.9854973427737068e-05, |
| "loss": 0.6673, |
| "num_tokens": 353420688.0, |
| "step": 924 |
| }, |
| { |
| "epoch": 0.6313993174061433, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 1.980719972164617e-05, |
| "loss": 0.6414, |
| "num_tokens": 353796065.0, |
| "step": 925 |
| }, |
| { |
| "epoch": 0.6320819112627987, |
| "grad_norm": 0.08984375, |
| "learning_rate": 1.9759465255623905e-05, |
| "loss": 0.6629, |
| "num_tokens": 354165105.0, |
| "step": 926 |
| }, |
| { |
| "epoch": 0.6327645051194539, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.971177027315815e-05, |
| "loss": 0.6337, |
| "num_tokens": 354564534.0, |
| "step": 927 |
| }, |
| { |
| "epoch": 0.6334470989761092, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 1.9664115017535382e-05, |
| "loss": 0.6357, |
| "num_tokens": 354937113.0, |
| "step": 928 |
| }, |
| { |
| "epoch": 0.6341296928327645, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 1.9616499731839473e-05, |
| "loss": 0.6455, |
| "num_tokens": 355287616.0, |
| "step": 929 |
| }, |
| { |
| "epoch": 0.6348122866894198, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 1.9568924658950376e-05, |
| "loss": 0.6484, |
| "num_tokens": 355697743.0, |
| "step": 930 |
| }, |
| { |
| "epoch": 0.6354948805460751, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 1.9521390041542946e-05, |
| "loss": 0.6284, |
| "num_tokens": 356120934.0, |
| "step": 931 |
| }, |
| { |
| "epoch": 0.6361774744027304, |
| "grad_norm": 0.08984375, |
| "learning_rate": 1.9473896122085673e-05, |
| "loss": 0.6499, |
| "num_tokens": 356499885.0, |
| "step": 932 |
| }, |
| { |
| "epoch": 0.6368600682593857, |
| "grad_norm": 0.08984375, |
| "learning_rate": 1.9426443142839447e-05, |
| "loss": 0.6802, |
| "num_tokens": 356873062.0, |
| "step": 933 |
| }, |
| { |
| "epoch": 0.6375426621160409, |
| "grad_norm": 0.09814453125, |
| "learning_rate": 1.9379031345856326e-05, |
| "loss": 0.6543, |
| "num_tokens": 357179004.0, |
| "step": 934 |
| }, |
| { |
| "epoch": 0.6382252559726962, |
| "grad_norm": 0.08984375, |
| "learning_rate": 1.9331660972978294e-05, |
| "loss": 0.7311, |
| "num_tokens": 357607500.0, |
| "step": 935 |
| }, |
| { |
| "epoch": 0.6389078498293516, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 1.9284332265836065e-05, |
| "loss": 0.653, |
| "num_tokens": 357978919.0, |
| "step": 936 |
| }, |
| { |
| "epoch": 0.6395904436860068, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.923704546584779e-05, |
| "loss": 0.6495, |
| "num_tokens": 358329202.0, |
| "step": 937 |
| }, |
| { |
| "epoch": 0.6402730375426621, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 1.918980081421787e-05, |
| "loss": 0.6999, |
| "num_tokens": 358730949.0, |
| "step": 938 |
| }, |
| { |
| "epoch": 0.6409556313993174, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 1.914259855193573e-05, |
| "loss": 0.6496, |
| "num_tokens": 359085717.0, |
| "step": 939 |
| }, |
| { |
| "epoch": 0.6416382252559727, |
| "grad_norm": 0.0859375, |
| "learning_rate": 1.909543891977454e-05, |
| "loss": 0.6705, |
| "num_tokens": 359496768.0, |
| "step": 940 |
| }, |
| { |
| "epoch": 0.642320819112628, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 1.9048322158290016e-05, |
| "loss": 0.6131, |
| "num_tokens": 359905689.0, |
| "step": 941 |
| }, |
| { |
| "epoch": 0.6430034129692833, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.9001248507819235e-05, |
| "loss": 0.6204, |
| "num_tokens": 360262389.0, |
| "step": 942 |
| }, |
| { |
| "epoch": 0.6436860068259386, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.895421820847934e-05, |
| "loss": 0.6475, |
| "num_tokens": 360661995.0, |
| "step": 943 |
| }, |
| { |
| "epoch": 0.6443686006825938, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 1.890723150016634e-05, |
| "loss": 0.6527, |
| "num_tokens": 361035596.0, |
| "step": 944 |
| }, |
| { |
| "epoch": 0.6450511945392492, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 1.8860288622553922e-05, |
| "loss": 0.6626, |
| "num_tokens": 361389017.0, |
| "step": 945 |
| }, |
| { |
| "epoch": 0.6457337883959045, |
| "grad_norm": 0.091796875, |
| "learning_rate": 1.8813389815092163e-05, |
| "loss": 0.6313, |
| "num_tokens": 361726207.0, |
| "step": 946 |
| }, |
| { |
| "epoch": 0.6464163822525597, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 1.8766535317006373e-05, |
| "loss": 0.6546, |
| "num_tokens": 362114480.0, |
| "step": 947 |
| }, |
| { |
| "epoch": 0.647098976109215, |
| "grad_norm": 0.0859375, |
| "learning_rate": 1.8719725367295815e-05, |
| "loss": 0.647, |
| "num_tokens": 362499080.0, |
| "step": 948 |
| }, |
| { |
| "epoch": 0.6477815699658703, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 1.867296020473255e-05, |
| "loss": 0.637, |
| "num_tokens": 362841135.0, |
| "step": 949 |
| }, |
| { |
| "epoch": 0.6484641638225256, |
| "grad_norm": 0.0859375, |
| "learning_rate": 1.862624006786014e-05, |
| "loss": 0.6487, |
| "num_tokens": 363249804.0, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.6491467576791808, |
| "grad_norm": 0.09765625, |
| "learning_rate": 1.8579565194992533e-05, |
| "loss": 0.6748, |
| "num_tokens": 363577492.0, |
| "step": 951 |
| }, |
| { |
| "epoch": 0.6498293515358362, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.8532935824212744e-05, |
| "loss": 0.6584, |
| "num_tokens": 363983408.0, |
| "step": 952 |
| }, |
| { |
| "epoch": 0.6505119453924915, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 1.8486352193371715e-05, |
| "loss": 0.6194, |
| "num_tokens": 364358124.0, |
| "step": 953 |
| }, |
| { |
| "epoch": 0.6511945392491467, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 1.8439814540087043e-05, |
| "loss": 0.6821, |
| "num_tokens": 364764247.0, |
| "step": 954 |
| }, |
| { |
| "epoch": 0.6518771331058021, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 1.839332310174184e-05, |
| "loss": 0.6045, |
| "num_tokens": 365084839.0, |
| "step": 955 |
| }, |
| { |
| "epoch": 0.6525597269624573, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 1.834687811548345e-05, |
| "loss": 0.6581, |
| "num_tokens": 365470138.0, |
| "step": 956 |
| }, |
| { |
| "epoch": 0.6532423208191126, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 1.830047981822228e-05, |
| "loss": 0.6388, |
| "num_tokens": 365786681.0, |
| "step": 957 |
| }, |
| { |
| "epoch": 0.653924914675768, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 1.8254128446630592e-05, |
| "loss": 0.6506, |
| "num_tokens": 366167846.0, |
| "step": 958 |
| }, |
| { |
| "epoch": 0.6546075085324232, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 1.8207824237141267e-05, |
| "loss": 0.6175, |
| "num_tokens": 366534988.0, |
| "step": 959 |
| }, |
| { |
| "epoch": 0.6552901023890785, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 1.8161567425946623e-05, |
| "loss": 0.6305, |
| "num_tokens": 366923656.0, |
| "step": 960 |
| }, |
| { |
| "epoch": 0.6559726962457337, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 1.811535824899722e-05, |
| "loss": 0.6573, |
| "num_tokens": 367341454.0, |
| "step": 961 |
| }, |
| { |
| "epoch": 0.6566552901023891, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 1.8069196942000616e-05, |
| "loss": 0.6402, |
| "num_tokens": 367739953.0, |
| "step": 962 |
| }, |
| { |
| "epoch": 0.6573378839590444, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 1.802308374042022e-05, |
| "loss": 0.6784, |
| "num_tokens": 368149784.0, |
| "step": 963 |
| }, |
| { |
| "epoch": 0.6580204778156996, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 1.7977018879474035e-05, |
| "loss": 0.6592, |
| "num_tokens": 368511121.0, |
| "step": 964 |
| }, |
| { |
| "epoch": 0.658703071672355, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 1.7931002594133494e-05, |
| "loss": 0.6992, |
| "num_tokens": 368897813.0, |
| "step": 965 |
| }, |
| { |
| "epoch": 0.6593856655290102, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 1.7885035119122256e-05, |
| "loss": 0.6264, |
| "num_tokens": 369246657.0, |
| "step": 966 |
| }, |
| { |
| "epoch": 0.6600682593856655, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 1.7839116688914997e-05, |
| "loss": 0.6799, |
| "num_tokens": 369614246.0, |
| "step": 967 |
| }, |
| { |
| "epoch": 0.6607508532423209, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 1.7793247537736213e-05, |
| "loss": 0.65, |
| "num_tokens": 370006857.0, |
| "step": 968 |
| }, |
| { |
| "epoch": 0.6614334470989761, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 1.774742789955906e-05, |
| "loss": 0.6463, |
| "num_tokens": 370391477.0, |
| "step": 969 |
| }, |
| { |
| "epoch": 0.6621160409556314, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 1.7701658008104122e-05, |
| "loss": 0.6631, |
| "num_tokens": 370786942.0, |
| "step": 970 |
| }, |
| { |
| "epoch": 0.6627986348122867, |
| "grad_norm": 0.09375, |
| "learning_rate": 1.7655938096838203e-05, |
| "loss": 0.6284, |
| "num_tokens": 371137111.0, |
| "step": 971 |
| }, |
| { |
| "epoch": 0.663481228668942, |
| "grad_norm": 0.09716796875, |
| "learning_rate": 1.7610268398973222e-05, |
| "loss": 0.6118, |
| "num_tokens": 371467486.0, |
| "step": 972 |
| }, |
| { |
| "epoch": 0.6641638225255972, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 1.75646491474649e-05, |
| "loss": 0.6164, |
| "num_tokens": 371828200.0, |
| "step": 973 |
| }, |
| { |
| "epoch": 0.6648464163822526, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 1.7519080575011686e-05, |
| "loss": 0.646, |
| "num_tokens": 372219313.0, |
| "step": 974 |
| }, |
| { |
| "epoch": 0.6655290102389079, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 1.7473562914053505e-05, |
| "loss": 0.6374, |
| "num_tokens": 372630696.0, |
| "step": 975 |
| }, |
| { |
| "epoch": 0.6662116040955631, |
| "grad_norm": 0.091796875, |
| "learning_rate": 1.7428096396770567e-05, |
| "loss": 0.6407, |
| "num_tokens": 372999701.0, |
| "step": 976 |
| }, |
| { |
| "epoch": 0.6668941979522184, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 1.738268125508225e-05, |
| "loss": 0.6909, |
| "num_tokens": 373432059.0, |
| "step": 977 |
| }, |
| { |
| "epoch": 0.6675767918088737, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 1.7337317720645847e-05, |
| "loss": 0.6381, |
| "num_tokens": 373837387.0, |
| "step": 978 |
| }, |
| { |
| "epoch": 0.668259385665529, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.729200602485537e-05, |
| "loss": 0.6519, |
| "num_tokens": 374231253.0, |
| "step": 979 |
| }, |
| { |
| "epoch": 0.6689419795221843, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 1.7246746398840484e-05, |
| "loss": 0.6557, |
| "num_tokens": 374628223.0, |
| "step": 980 |
| }, |
| { |
| "epoch": 0.6696245733788396, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 1.7201539073465198e-05, |
| "loss": 0.6366, |
| "num_tokens": 374995482.0, |
| "step": 981 |
| }, |
| { |
| "epoch": 0.6703071672354949, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.715638427932675e-05, |
| "loss": 0.6565, |
| "num_tokens": 375413718.0, |
| "step": 982 |
| }, |
| { |
| "epoch": 0.6709897610921501, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 1.7111282246754444e-05, |
| "loss": 0.6243, |
| "num_tokens": 375745662.0, |
| "step": 983 |
| }, |
| { |
| "epoch": 0.6716723549488055, |
| "grad_norm": 0.091796875, |
| "learning_rate": 1.7066233205808436e-05, |
| "loss": 0.6393, |
| "num_tokens": 376113589.0, |
| "step": 984 |
| }, |
| { |
| "epoch": 0.6723549488054608, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.702123738627857e-05, |
| "loss": 0.6237, |
| "num_tokens": 376493989.0, |
| "step": 985 |
| }, |
| { |
| "epoch": 0.673037542662116, |
| "grad_norm": 0.087890625, |
| "learning_rate": 1.6976295017683238e-05, |
| "loss": 0.6239, |
| "num_tokens": 376901460.0, |
| "step": 986 |
| }, |
| { |
| "epoch": 0.6737201365187714, |
| "grad_norm": 0.0859375, |
| "learning_rate": 1.6931406329268163e-05, |
| "loss": 0.6318, |
| "num_tokens": 377284481.0, |
| "step": 987 |
| }, |
| { |
| "epoch": 0.6744027303754266, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 1.688657155000525e-05, |
| "loss": 0.6213, |
| "num_tokens": 377657268.0, |
| "step": 988 |
| }, |
| { |
| "epoch": 0.6750853242320819, |
| "grad_norm": 0.091796875, |
| "learning_rate": 1.684179090859145e-05, |
| "loss": 0.6779, |
| "num_tokens": 378061402.0, |
| "step": 989 |
| }, |
| { |
| "epoch": 0.6757679180887372, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 1.679706463344754e-05, |
| "loss": 0.6491, |
| "num_tokens": 378392822.0, |
| "step": 990 |
| }, |
| { |
| "epoch": 0.6764505119453925, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 1.6752392952716978e-05, |
| "loss": 0.6228, |
| "num_tokens": 378794121.0, |
| "step": 991 |
| }, |
| { |
| "epoch": 0.6771331058020478, |
| "grad_norm": 0.091796875, |
| "learning_rate": 1.6707776094264745e-05, |
| "loss": 0.6572, |
| "num_tokens": 379167932.0, |
| "step": 992 |
| }, |
| { |
| "epoch": 0.677815699658703, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 1.6663214285676198e-05, |
| "loss": 0.6102, |
| "num_tokens": 379512553.0, |
| "step": 993 |
| }, |
| { |
| "epoch": 0.6784982935153584, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 1.6618707754255853e-05, |
| "loss": 0.5991, |
| "num_tokens": 379906899.0, |
| "step": 994 |
| }, |
| { |
| "epoch": 0.6791808873720137, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 1.657425672702631e-05, |
| "loss": 0.656, |
| "num_tokens": 380308538.0, |
| "step": 995 |
| }, |
| { |
| "epoch": 0.6798634812286689, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 1.6529861430727024e-05, |
| "loss": 0.6508, |
| "num_tokens": 380660755.0, |
| "step": 996 |
| }, |
| { |
| "epoch": 0.6805460750853243, |
| "grad_norm": 0.09765625, |
| "learning_rate": 1.648552209181316e-05, |
| "loss": 0.6956, |
| "num_tokens": 381022131.0, |
| "step": 997 |
| }, |
| { |
| "epoch": 0.6812286689419795, |
| "grad_norm": 0.087890625, |
| "learning_rate": 1.644123893645448e-05, |
| "loss": 0.6622, |
| "num_tokens": 381426426.0, |
| "step": 998 |
| }, |
| { |
| "epoch": 0.6819112627986348, |
| "grad_norm": 0.091796875, |
| "learning_rate": 1.6397012190534138e-05, |
| "loss": 0.6258, |
| "num_tokens": 381759607.0, |
| "step": 999 |
| }, |
| { |
| "epoch": 0.6825938566552902, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 1.6352842079647567e-05, |
| "loss": 0.638, |
| "num_tokens": 382166430.0, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.6832764505119454, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.6308728829101295e-05, |
| "loss": 0.6837, |
| "num_tokens": 382606196.0, |
| "step": 1001 |
| }, |
| { |
| "epoch": 0.6839590443686007, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 1.6264672663911828e-05, |
| "loss": 0.6461, |
| "num_tokens": 382974639.0, |
| "step": 1002 |
| }, |
| { |
| "epoch": 0.6846416382252559, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 1.6220673808804465e-05, |
| "loss": 0.6744, |
| "num_tokens": 383361189.0, |
| "step": 1003 |
| }, |
| { |
| "epoch": 0.6853242320819113, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 1.617673248821221e-05, |
| "loss": 0.6602, |
| "num_tokens": 383718573.0, |
| "step": 1004 |
| }, |
| { |
| "epoch": 0.6860068259385665, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 1.6132848926274537e-05, |
| "loss": 0.6518, |
| "num_tokens": 384075218.0, |
| "step": 1005 |
| }, |
| { |
| "epoch": 0.6866894197952218, |
| "grad_norm": 0.091796875, |
| "learning_rate": 1.608902334683635e-05, |
| "loss": 0.6177, |
| "num_tokens": 384430379.0, |
| "step": 1006 |
| }, |
| { |
| "epoch": 0.6873720136518772, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 1.6045255973446764e-05, |
| "loss": 0.6617, |
| "num_tokens": 384812573.0, |
| "step": 1007 |
| }, |
| { |
| "epoch": 0.6880546075085324, |
| "grad_norm": 0.08349609375, |
| "learning_rate": 1.6001547029357994e-05, |
| "loss": 0.6211, |
| "num_tokens": 385231970.0, |
| "step": 1008 |
| }, |
| { |
| "epoch": 0.6887372013651877, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 1.595789673752423e-05, |
| "loss": 0.6428, |
| "num_tokens": 385547548.0, |
| "step": 1009 |
| }, |
| { |
| "epoch": 0.689419795221843, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 1.5914305320600474e-05, |
| "loss": 0.657, |
| "num_tokens": 385900887.0, |
| "step": 1010 |
| }, |
| { |
| "epoch": 0.6901023890784983, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 1.5870773000941386e-05, |
| "loss": 0.634, |
| "num_tokens": 386274798.0, |
| "step": 1011 |
| }, |
| { |
| "epoch": 0.6907849829351536, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.582730000060023e-05, |
| "loss": 0.6777, |
| "num_tokens": 386681135.0, |
| "step": 1012 |
| }, |
| { |
| "epoch": 0.6914675767918089, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 1.5783886541327652e-05, |
| "loss": 0.6615, |
| "num_tokens": 387057029.0, |
| "step": 1013 |
| }, |
| { |
| "epoch": 0.6921501706484642, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 1.5740532844570592e-05, |
| "loss": 0.6697, |
| "num_tokens": 387470680.0, |
| "step": 1014 |
| }, |
| { |
| "epoch": 0.6928327645051194, |
| "grad_norm": 0.10009765625, |
| "learning_rate": 1.5697239131471175e-05, |
| "loss": 0.6394, |
| "num_tokens": 387782669.0, |
| "step": 1015 |
| }, |
| { |
| "epoch": 0.6935153583617747, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 1.565400562286552e-05, |
| "loss": 0.6333, |
| "num_tokens": 388124060.0, |
| "step": 1016 |
| }, |
| { |
| "epoch": 0.6941979522184301, |
| "grad_norm": 0.083984375, |
| "learning_rate": 1.561083253928266e-05, |
| "loss": 0.6474, |
| "num_tokens": 388531886.0, |
| "step": 1017 |
| }, |
| { |
| "epoch": 0.6948805460750853, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 1.5567720100943418e-05, |
| "loss": 0.6558, |
| "num_tokens": 388916855.0, |
| "step": 1018 |
| }, |
| { |
| "epoch": 0.6955631399317406, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 1.5524668527759273e-05, |
| "loss": 0.6253, |
| "num_tokens": 389262728.0, |
| "step": 1019 |
| }, |
| { |
| "epoch": 0.6962457337883959, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 1.5481678039331217e-05, |
| "loss": 0.6442, |
| "num_tokens": 389650899.0, |
| "step": 1020 |
| }, |
| { |
| "epoch": 0.6969283276450512, |
| "grad_norm": 0.08984375, |
| "learning_rate": 1.543874885494868e-05, |
| "loss": 0.6306, |
| "num_tokens": 390026303.0, |
| "step": 1021 |
| }, |
| { |
| "epoch": 0.6976109215017064, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 1.5395881193588385e-05, |
| "loss": 0.6342, |
| "num_tokens": 390401858.0, |
| "step": 1022 |
| }, |
| { |
| "epoch": 0.6982935153583618, |
| "grad_norm": 0.08984375, |
| "learning_rate": 1.5353075273913215e-05, |
| "loss": 0.6951, |
| "num_tokens": 390811201.0, |
| "step": 1023 |
| }, |
| { |
| "epoch": 0.6989761092150171, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 1.5310331314271127e-05, |
| "loss": 0.6885, |
| "num_tokens": 391235955.0, |
| "step": 1024 |
| }, |
| { |
| "epoch": 0.6996587030716723, |
| "grad_norm": 0.09375, |
| "learning_rate": 1.526764953269403e-05, |
| "loss": 0.6776, |
| "num_tokens": 391632321.0, |
| "step": 1025 |
| }, |
| { |
| "epoch": 0.7003412969283277, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 1.5225030146896668e-05, |
| "loss": 0.6635, |
| "num_tokens": 392039121.0, |
| "step": 1026 |
| }, |
| { |
| "epoch": 0.701023890784983, |
| "grad_norm": 0.08984375, |
| "learning_rate": 1.5182473374275513e-05, |
| "loss": 0.6112, |
| "num_tokens": 392442423.0, |
| "step": 1027 |
| }, |
| { |
| "epoch": 0.7017064846416382, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 1.5139979431907653e-05, |
| "loss": 0.6689, |
| "num_tokens": 392784272.0, |
| "step": 1028 |
| }, |
| { |
| "epoch": 0.7023890784982935, |
| "grad_norm": 0.091796875, |
| "learning_rate": 1.5097548536549675e-05, |
| "loss": 0.6575, |
| "num_tokens": 393134534.0, |
| "step": 1029 |
| }, |
| { |
| "epoch": 0.7030716723549488, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 1.5055180904636602e-05, |
| "loss": 0.6281, |
| "num_tokens": 393493604.0, |
| "step": 1030 |
| }, |
| { |
| "epoch": 0.7037542662116041, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 1.5012876752280709e-05, |
| "loss": 0.6565, |
| "num_tokens": 393843182.0, |
| "step": 1031 |
| }, |
| { |
| "epoch": 0.7044368600682593, |
| "grad_norm": 0.09375, |
| "learning_rate": 1.4970636295270519e-05, |
| "loss": 0.6988, |
| "num_tokens": 394205653.0, |
| "step": 1032 |
| }, |
| { |
| "epoch": 0.7051194539249147, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 1.4928459749069625e-05, |
| "loss": 0.6654, |
| "num_tokens": 394575170.0, |
| "step": 1033 |
| }, |
| { |
| "epoch": 0.70580204778157, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 1.4886347328815614e-05, |
| "loss": 0.6319, |
| "num_tokens": 394952885.0, |
| "step": 1034 |
| }, |
| { |
| "epoch": 0.7064846416382252, |
| "grad_norm": 0.091796875, |
| "learning_rate": 1.4844299249319e-05, |
| "loss": 0.6609, |
| "num_tokens": 395329185.0, |
| "step": 1035 |
| }, |
| { |
| "epoch": 0.7071672354948806, |
| "grad_norm": 0.0859375, |
| "learning_rate": 1.4802315725062088e-05, |
| "loss": 0.6729, |
| "num_tokens": 395748348.0, |
| "step": 1036 |
| }, |
| { |
| "epoch": 0.7078498293515358, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 1.4760396970197868e-05, |
| "loss": 0.6192, |
| "num_tokens": 396102552.0, |
| "step": 1037 |
| }, |
| { |
| "epoch": 0.7085324232081911, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 1.4718543198548998e-05, |
| "loss": 0.6706, |
| "num_tokens": 396473191.0, |
| "step": 1038 |
| }, |
| { |
| "epoch": 0.7092150170648465, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 1.4676754623606634e-05, |
| "loss": 0.7055, |
| "num_tokens": 396854982.0, |
| "step": 1039 |
| }, |
| { |
| "epoch": 0.7098976109215017, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 1.4635031458529366e-05, |
| "loss": 0.6872, |
| "num_tokens": 397224172.0, |
| "step": 1040 |
| }, |
| { |
| "epoch": 0.710580204778157, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 1.4593373916142167e-05, |
| "loss": 0.6352, |
| "num_tokens": 397638473.0, |
| "step": 1041 |
| }, |
| { |
| "epoch": 0.7112627986348122, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.4551782208935245e-05, |
| "loss": 0.653, |
| "num_tokens": 398020413.0, |
| "step": 1042 |
| }, |
| { |
| "epoch": 0.7119453924914676, |
| "grad_norm": 0.0859375, |
| "learning_rate": 1.4510256549063006e-05, |
| "loss": 0.6446, |
| "num_tokens": 398417000.0, |
| "step": 1043 |
| }, |
| { |
| "epoch": 0.7126279863481229, |
| "grad_norm": 0.0859375, |
| "learning_rate": 1.4468797148342946e-05, |
| "loss": 0.6053, |
| "num_tokens": 398778614.0, |
| "step": 1044 |
| }, |
| { |
| "epoch": 0.7133105802047781, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 1.4427404218254598e-05, |
| "loss": 0.6679, |
| "num_tokens": 399158135.0, |
| "step": 1045 |
| }, |
| { |
| "epoch": 0.7139931740614335, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 1.4386077969938409e-05, |
| "loss": 0.6481, |
| "num_tokens": 399501986.0, |
| "step": 1046 |
| }, |
| { |
| "epoch": 0.7146757679180887, |
| "grad_norm": 0.09375, |
| "learning_rate": 1.4344818614194721e-05, |
| "loss": 0.6481, |
| "num_tokens": 399874589.0, |
| "step": 1047 |
| }, |
| { |
| "epoch": 0.715358361774744, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 1.4303626361482653e-05, |
| "loss": 0.6782, |
| "num_tokens": 400263698.0, |
| "step": 1048 |
| }, |
| { |
| "epoch": 0.7160409556313994, |
| "grad_norm": 0.09375, |
| "learning_rate": 1.4262501421919031e-05, |
| "loss": 0.7101, |
| "num_tokens": 400640595.0, |
| "step": 1049 |
| }, |
| { |
| "epoch": 0.7167235494880546, |
| "grad_norm": 0.08984375, |
| "learning_rate": 1.422144400527733e-05, |
| "loss": 0.6242, |
| "num_tokens": 400990040.0, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.7174061433447099, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 1.4180454320986591e-05, |
| "loss": 0.6624, |
| "num_tokens": 401337431.0, |
| "step": 1051 |
| }, |
| { |
| "epoch": 0.7180887372013652, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 1.4139532578130383e-05, |
| "loss": 0.6398, |
| "num_tokens": 401728632.0, |
| "step": 1052 |
| }, |
| { |
| "epoch": 0.7187713310580205, |
| "grad_norm": 0.087890625, |
| "learning_rate": 1.4098678985445696e-05, |
| "loss": 0.6331, |
| "num_tokens": 402106021.0, |
| "step": 1053 |
| }, |
| { |
| "epoch": 0.7194539249146757, |
| "grad_norm": 0.08984375, |
| "learning_rate": 1.4057893751321887e-05, |
| "loss": 0.6231, |
| "num_tokens": 402473563.0, |
| "step": 1054 |
| }, |
| { |
| "epoch": 0.7201365187713311, |
| "grad_norm": 0.0859375, |
| "learning_rate": 1.401717708379964e-05, |
| "loss": 0.6309, |
| "num_tokens": 402861165.0, |
| "step": 1055 |
| }, |
| { |
| "epoch": 0.7208191126279864, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 1.3976529190569897e-05, |
| "loss": 0.6907, |
| "num_tokens": 403229159.0, |
| "step": 1056 |
| }, |
| { |
| "epoch": 0.7215017064846416, |
| "grad_norm": 0.0859375, |
| "learning_rate": 1.3935950278972747e-05, |
| "loss": 0.6218, |
| "num_tokens": 403592546.0, |
| "step": 1057 |
| }, |
| { |
| "epoch": 0.7221843003412969, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.3895440555996464e-05, |
| "loss": 0.6286, |
| "num_tokens": 403992643.0, |
| "step": 1058 |
| }, |
| { |
| "epoch": 0.7228668941979522, |
| "grad_norm": 0.091796875, |
| "learning_rate": 1.3855000228276369e-05, |
| "loss": 0.6517, |
| "num_tokens": 404354006.0, |
| "step": 1059 |
| }, |
| { |
| "epoch": 0.7235494880546075, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.3814629502093803e-05, |
| "loss": 0.6646, |
| "num_tokens": 404752387.0, |
| "step": 1060 |
| }, |
| { |
| "epoch": 0.7242320819112628, |
| "grad_norm": 0.091796875, |
| "learning_rate": 1.3774328583375107e-05, |
| "loss": 0.65, |
| "num_tokens": 405147619.0, |
| "step": 1061 |
| }, |
| { |
| "epoch": 0.7249146757679181, |
| "grad_norm": 0.08984375, |
| "learning_rate": 1.3734097677690522e-05, |
| "loss": 0.6196, |
| "num_tokens": 405526287.0, |
| "step": 1062 |
| }, |
| { |
| "epoch": 0.7255972696245734, |
| "grad_norm": 0.08984375, |
| "learning_rate": 1.3693936990253142e-05, |
| "loss": 0.6763, |
| "num_tokens": 405900294.0, |
| "step": 1063 |
| }, |
| { |
| "epoch": 0.7262798634812286, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 1.3653846725917917e-05, |
| "loss": 0.6204, |
| "num_tokens": 406230534.0, |
| "step": 1064 |
| }, |
| { |
| "epoch": 0.726962457337884, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 1.3613827089180576e-05, |
| "loss": 0.6922, |
| "num_tokens": 406680793.0, |
| "step": 1065 |
| }, |
| { |
| "epoch": 0.7276450511945393, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 1.357387828417655e-05, |
| "loss": 0.6589, |
| "num_tokens": 407046774.0, |
| "step": 1066 |
| }, |
| { |
| "epoch": 0.7283276450511945, |
| "grad_norm": 0.08251953125, |
| "learning_rate": 1.353400051468002e-05, |
| "loss": 0.6636, |
| "num_tokens": 407507174.0, |
| "step": 1067 |
| }, |
| { |
| "epoch": 0.7290102389078499, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 1.3494193984102776e-05, |
| "loss": 0.6336, |
| "num_tokens": 407859909.0, |
| "step": 1068 |
| }, |
| { |
| "epoch": 0.7296928327645051, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.3454458895493249e-05, |
| "loss": 0.6326, |
| "num_tokens": 408267778.0, |
| "step": 1069 |
| }, |
| { |
| "epoch": 0.7303754266211604, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 1.3414795451535448e-05, |
| "loss": 0.6622, |
| "num_tokens": 408650161.0, |
| "step": 1070 |
| }, |
| { |
| "epoch": 0.7310580204778157, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 1.337520385454793e-05, |
| "loss": 0.6653, |
| "num_tokens": 409036294.0, |
| "step": 1071 |
| }, |
| { |
| "epoch": 0.731740614334471, |
| "grad_norm": 0.095703125, |
| "learning_rate": 1.3335684306482763e-05, |
| "loss": 0.6689, |
| "num_tokens": 409376992.0, |
| "step": 1072 |
| }, |
| { |
| "epoch": 0.7324232081911263, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 1.3296237008924528e-05, |
| "loss": 0.6494, |
| "num_tokens": 409752221.0, |
| "step": 1073 |
| }, |
| { |
| "epoch": 0.7331058020477815, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 1.3256862163089234e-05, |
| "loss": 0.6362, |
| "num_tokens": 410156018.0, |
| "step": 1074 |
| }, |
| { |
| "epoch": 0.7337883959044369, |
| "grad_norm": 0.09375, |
| "learning_rate": 1.3217559969823334e-05, |
| "loss": 0.6701, |
| "num_tokens": 410510602.0, |
| "step": 1075 |
| }, |
| { |
| "epoch": 0.7344709897610922, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.3178330629602698e-05, |
| "loss": 0.6586, |
| "num_tokens": 410934140.0, |
| "step": 1076 |
| }, |
| { |
| "epoch": 0.7351535836177474, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 1.3139174342531552e-05, |
| "loss": 0.6268, |
| "num_tokens": 411298191.0, |
| "step": 1077 |
| }, |
| { |
| "epoch": 0.7358361774744028, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 1.310009130834153e-05, |
| "loss": 0.6751, |
| "num_tokens": 411654765.0, |
| "step": 1078 |
| }, |
| { |
| "epoch": 0.736518771331058, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.3061081726390584e-05, |
| "loss": 0.6596, |
| "num_tokens": 412026613.0, |
| "step": 1079 |
| }, |
| { |
| "epoch": 0.7372013651877133, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 1.3022145795661988e-05, |
| "loss": 0.6484, |
| "num_tokens": 412454003.0, |
| "step": 1080 |
| }, |
| { |
| "epoch": 0.7378839590443687, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 1.2983283714763345e-05, |
| "loss": 0.6876, |
| "num_tokens": 412875960.0, |
| "step": 1081 |
| }, |
| { |
| "epoch": 0.7385665529010239, |
| "grad_norm": 0.08984375, |
| "learning_rate": 1.2944495681925568e-05, |
| "loss": 0.671, |
| "num_tokens": 413265671.0, |
| "step": 1082 |
| }, |
| { |
| "epoch": 0.7392491467576792, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 1.2905781895001823e-05, |
| "loss": 0.6457, |
| "num_tokens": 413635871.0, |
| "step": 1083 |
| }, |
| { |
| "epoch": 0.7399317406143344, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 1.2867142551466602e-05, |
| "loss": 0.689, |
| "num_tokens": 414058458.0, |
| "step": 1084 |
| }, |
| { |
| "epoch": 0.7406143344709898, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 1.2828577848414636e-05, |
| "loss": 0.6417, |
| "num_tokens": 414425737.0, |
| "step": 1085 |
| }, |
| { |
| "epoch": 0.741296928327645, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 1.279008798255993e-05, |
| "loss": 0.6709, |
| "num_tokens": 414789117.0, |
| "step": 1086 |
| }, |
| { |
| "epoch": 0.7419795221843003, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 1.2751673150234773e-05, |
| "loss": 0.6346, |
| "num_tokens": 415145213.0, |
| "step": 1087 |
| }, |
| { |
| "epoch": 0.7426621160409557, |
| "grad_norm": 0.08984375, |
| "learning_rate": 1.2713333547388706e-05, |
| "loss": 0.646, |
| "num_tokens": 415533226.0, |
| "step": 1088 |
| }, |
| { |
| "epoch": 0.7433447098976109, |
| "grad_norm": 0.08154296875, |
| "learning_rate": 1.2675069369587506e-05, |
| "loss": 0.6909, |
| "num_tokens": 416017674.0, |
| "step": 1089 |
| }, |
| { |
| "epoch": 0.7440273037542662, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 1.2636880812012253e-05, |
| "loss": 0.634, |
| "num_tokens": 416440659.0, |
| "step": 1090 |
| }, |
| { |
| "epoch": 0.7447098976109215, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 1.2598768069458289e-05, |
| "loss": 0.6853, |
| "num_tokens": 416841489.0, |
| "step": 1091 |
| }, |
| { |
| "epoch": 0.7453924914675768, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 1.2560731336334215e-05, |
| "loss": 0.6445, |
| "num_tokens": 417177327.0, |
| "step": 1092 |
| }, |
| { |
| "epoch": 0.7460750853242321, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 1.2522770806660939e-05, |
| "loss": 0.6562, |
| "num_tokens": 417577274.0, |
| "step": 1093 |
| }, |
| { |
| "epoch": 0.7467576791808874, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 1.2484886674070662e-05, |
| "loss": 0.6204, |
| "num_tokens": 417938669.0, |
| "step": 1094 |
| }, |
| { |
| "epoch": 0.7474402730375427, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 1.2447079131805857e-05, |
| "loss": 0.6532, |
| "num_tokens": 418334906.0, |
| "step": 1095 |
| }, |
| { |
| "epoch": 0.7481228668941979, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 1.240934837271837e-05, |
| "loss": 0.605, |
| "num_tokens": 418683963.0, |
| "step": 1096 |
| }, |
| { |
| "epoch": 0.7488054607508532, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 1.2371694589268357e-05, |
| "loss": 0.6393, |
| "num_tokens": 419026720.0, |
| "step": 1097 |
| }, |
| { |
| "epoch": 0.7494880546075086, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 1.233411797352332e-05, |
| "loss": 0.628, |
| "num_tokens": 419362555.0, |
| "step": 1098 |
| }, |
| { |
| "epoch": 0.7501706484641638, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 1.2296618717157178e-05, |
| "loss": 0.6649, |
| "num_tokens": 419748312.0, |
| "step": 1099 |
| }, |
| { |
| "epoch": 0.7508532423208191, |
| "grad_norm": 0.091796875, |
| "learning_rate": 1.2259197011449203e-05, |
| "loss": 0.6547, |
| "num_tokens": 420121223.0, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.7515358361774744, |
| "grad_norm": 0.09814453125, |
| "learning_rate": 1.2221853047283121e-05, |
| "loss": 0.633, |
| "num_tokens": 420455135.0, |
| "step": 1101 |
| }, |
| { |
| "epoch": 0.7522184300341297, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 1.2184587015146094e-05, |
| "loss": 0.6469, |
| "num_tokens": 420851181.0, |
| "step": 1102 |
| }, |
| { |
| "epoch": 0.752901023890785, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 1.2147399105127762e-05, |
| "loss": 0.6593, |
| "num_tokens": 421241253.0, |
| "step": 1103 |
| }, |
| { |
| "epoch": 0.7535836177474403, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 1.2110289506919273e-05, |
| "loss": 0.6103, |
| "num_tokens": 421581023.0, |
| "step": 1104 |
| }, |
| { |
| "epoch": 0.7542662116040956, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 1.2073258409812328e-05, |
| "loss": 0.6422, |
| "num_tokens": 421975487.0, |
| "step": 1105 |
| }, |
| { |
| "epoch": 0.7549488054607508, |
| "grad_norm": 0.1005859375, |
| "learning_rate": 1.2036306002698193e-05, |
| "loss": 0.6784, |
| "num_tokens": 422279490.0, |
| "step": 1106 |
| }, |
| { |
| "epoch": 0.7556313993174062, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 1.1999432474066744e-05, |
| "loss": 0.6231, |
| "num_tokens": 422646056.0, |
| "step": 1107 |
| }, |
| { |
| "epoch": 0.7563139931740614, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 1.1962638012005508e-05, |
| "loss": 0.6501, |
| "num_tokens": 423004511.0, |
| "step": 1108 |
| }, |
| { |
| "epoch": 0.7569965870307167, |
| "grad_norm": 0.095703125, |
| "learning_rate": 1.1925922804198702e-05, |
| "loss": 0.6453, |
| "num_tokens": 423364356.0, |
| "step": 1109 |
| }, |
| { |
| "epoch": 0.757679180887372, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 1.1889287037926289e-05, |
| "loss": 0.6599, |
| "num_tokens": 423755589.0, |
| "step": 1110 |
| }, |
| { |
| "epoch": 0.7583617747440273, |
| "grad_norm": 0.0859375, |
| "learning_rate": 1.1852730900062995e-05, |
| "loss": 0.6623, |
| "num_tokens": 424162917.0, |
| "step": 1111 |
| }, |
| { |
| "epoch": 0.7590443686006826, |
| "grad_norm": 0.08984375, |
| "learning_rate": 1.1816254577077367e-05, |
| "loss": 0.6371, |
| "num_tokens": 424526345.0, |
| "step": 1112 |
| }, |
| { |
| "epoch": 0.7597269624573378, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 1.1779858255030847e-05, |
| "loss": 0.6529, |
| "num_tokens": 424877727.0, |
| "step": 1113 |
| }, |
| { |
| "epoch": 0.7604095563139932, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 1.1743542119576797e-05, |
| "loss": 0.6032, |
| "num_tokens": 425246012.0, |
| "step": 1114 |
| }, |
| { |
| "epoch": 0.7610921501706485, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 1.1707306355959516e-05, |
| "loss": 0.6482, |
| "num_tokens": 425585640.0, |
| "step": 1115 |
| }, |
| { |
| "epoch": 0.7617747440273037, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 1.1671151149013399e-05, |
| "loss": 0.6472, |
| "num_tokens": 425949087.0, |
| "step": 1116 |
| }, |
| { |
| "epoch": 0.7624573378839591, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 1.1635076683161892e-05, |
| "loss": 0.6377, |
| "num_tokens": 426349742.0, |
| "step": 1117 |
| }, |
| { |
| "epoch": 0.7631399317406143, |
| "grad_norm": 0.08984375, |
| "learning_rate": 1.159908314241659e-05, |
| "loss": 0.6364, |
| "num_tokens": 426755093.0, |
| "step": 1118 |
| }, |
| { |
| "epoch": 0.7638225255972696, |
| "grad_norm": 0.087890625, |
| "learning_rate": 1.1563170710376318e-05, |
| "loss": 0.6822, |
| "num_tokens": 427135471.0, |
| "step": 1119 |
| }, |
| { |
| "epoch": 0.764505119453925, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 1.152733957022617e-05, |
| "loss": 0.5975, |
| "num_tokens": 427476418.0, |
| "step": 1120 |
| }, |
| { |
| "epoch": 0.7651877133105802, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 1.149158990473655e-05, |
| "loss": 0.6551, |
| "num_tokens": 427826974.0, |
| "step": 1121 |
| }, |
| { |
| "epoch": 0.7658703071672355, |
| "grad_norm": 0.087890625, |
| "learning_rate": 1.1455921896262311e-05, |
| "loss": 0.6267, |
| "num_tokens": 428203527.0, |
| "step": 1122 |
| }, |
| { |
| "epoch": 0.7665529010238907, |
| "grad_norm": 0.087890625, |
| "learning_rate": 1.1420335726741765e-05, |
| "loss": 0.6344, |
| "num_tokens": 428574499.0, |
| "step": 1123 |
| }, |
| { |
| "epoch": 0.7672354948805461, |
| "grad_norm": 0.087890625, |
| "learning_rate": 1.1384831577695763e-05, |
| "loss": 0.648, |
| "num_tokens": 428955771.0, |
| "step": 1124 |
| }, |
| { |
| "epoch": 0.7679180887372014, |
| "grad_norm": 0.08203125, |
| "learning_rate": 1.1349409630226804e-05, |
| "loss": 0.6499, |
| "num_tokens": 429400796.0, |
| "step": 1125 |
| }, |
| { |
| "epoch": 0.7686006825938566, |
| "grad_norm": 0.083984375, |
| "learning_rate": 1.1314070065018066e-05, |
| "loss": 0.6448, |
| "num_tokens": 429815558.0, |
| "step": 1126 |
| }, |
| { |
| "epoch": 0.769283276450512, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 1.1278813062332508e-05, |
| "loss": 0.6498, |
| "num_tokens": 430186939.0, |
| "step": 1127 |
| }, |
| { |
| "epoch": 0.7699658703071672, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 1.1243638802011954e-05, |
| "loss": 0.6251, |
| "num_tokens": 430579114.0, |
| "step": 1128 |
| }, |
| { |
| "epoch": 0.7706484641638225, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 1.1208547463476162e-05, |
| "loss": 0.6631, |
| "num_tokens": 431000699.0, |
| "step": 1129 |
| }, |
| { |
| "epoch": 0.7713310580204779, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.1173539225721916e-05, |
| "loss": 0.6262, |
| "num_tokens": 431359520.0, |
| "step": 1130 |
| }, |
| { |
| "epoch": 0.7720136518771331, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 1.1138614267322126e-05, |
| "loss": 0.6078, |
| "num_tokens": 431703704.0, |
| "step": 1131 |
| }, |
| { |
| "epoch": 0.7726962457337884, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 1.1103772766424886e-05, |
| "loss": 0.6778, |
| "num_tokens": 432102812.0, |
| "step": 1132 |
| }, |
| { |
| "epoch": 0.7733788395904437, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 1.1069014900752592e-05, |
| "loss": 0.6828, |
| "num_tokens": 432489869.0, |
| "step": 1133 |
| }, |
| { |
| "epoch": 0.774061433447099, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 1.1034340847601023e-05, |
| "loss": 0.6241, |
| "num_tokens": 432902685.0, |
| "step": 1134 |
| }, |
| { |
| "epoch": 0.7747440273037542, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.0999750783838442e-05, |
| "loss": 0.6549, |
| "num_tokens": 433284774.0, |
| "step": 1135 |
| }, |
| { |
| "epoch": 0.7754266211604095, |
| "grad_norm": 0.087890625, |
| "learning_rate": 1.0965244885904699e-05, |
| "loss": 0.661, |
| "num_tokens": 433668868.0, |
| "step": 1136 |
| }, |
| { |
| "epoch": 0.7761092150170649, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 1.0930823329810311e-05, |
| "loss": 0.6733, |
| "num_tokens": 434081034.0, |
| "step": 1137 |
| }, |
| { |
| "epoch": 0.7767918088737201, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.0896486291135583e-05, |
| "loss": 0.6325, |
| "num_tokens": 434454435.0, |
| "step": 1138 |
| }, |
| { |
| "epoch": 0.7774744027303754, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 1.0862233945029712e-05, |
| "loss": 0.5977, |
| "num_tokens": 434783120.0, |
| "step": 1139 |
| }, |
| { |
| "epoch": 0.7781569965870307, |
| "grad_norm": 0.087890625, |
| "learning_rate": 1.0828066466209891e-05, |
| "loss": 0.642, |
| "num_tokens": 435149765.0, |
| "step": 1140 |
| }, |
| { |
| "epoch": 0.778839590443686, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 1.079398402896038e-05, |
| "loss": 0.6302, |
| "num_tokens": 435559701.0, |
| "step": 1141 |
| }, |
| { |
| "epoch": 0.7795221843003413, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 1.0759986807131706e-05, |
| "loss": 0.6595, |
| "num_tokens": 435958115.0, |
| "step": 1142 |
| }, |
| { |
| "epoch": 0.7802047781569966, |
| "grad_norm": 0.087890625, |
| "learning_rate": 1.0726074974139683e-05, |
| "loss": 0.6339, |
| "num_tokens": 436364850.0, |
| "step": 1143 |
| }, |
| { |
| "epoch": 0.7808873720136519, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 1.0692248702964575e-05, |
| "loss": 0.6689, |
| "num_tokens": 436764088.0, |
| "step": 1144 |
| }, |
| { |
| "epoch": 0.7815699658703071, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 1.0658508166150224e-05, |
| "loss": 0.6364, |
| "num_tokens": 437135437.0, |
| "step": 1145 |
| }, |
| { |
| "epoch": 0.7822525597269625, |
| "grad_norm": 0.09375, |
| "learning_rate": 1.0624853535803136e-05, |
| "loss": 0.6343, |
| "num_tokens": 437487699.0, |
| "step": 1146 |
| }, |
| { |
| "epoch": 0.7829351535836178, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 1.0591284983591595e-05, |
| "loss": 0.6321, |
| "num_tokens": 437861464.0, |
| "step": 1147 |
| }, |
| { |
| "epoch": 0.783617747440273, |
| "grad_norm": 0.08349609375, |
| "learning_rate": 1.0557802680744857e-05, |
| "loss": 0.6645, |
| "num_tokens": 438289022.0, |
| "step": 1148 |
| }, |
| { |
| "epoch": 0.7843003412969284, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 1.05244067980522e-05, |
| "loss": 0.6422, |
| "num_tokens": 438635088.0, |
| "step": 1149 |
| }, |
| { |
| "epoch": 0.7849829351535836, |
| "grad_norm": 0.08251953125, |
| "learning_rate": 1.0491097505862085e-05, |
| "loss": 0.6588, |
| "num_tokens": 439079263.0, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.7856655290102389, |
| "grad_norm": 0.08203125, |
| "learning_rate": 1.04578749740813e-05, |
| "loss": 0.6688, |
| "num_tokens": 439530553.0, |
| "step": 1151 |
| }, |
| { |
| "epoch": 0.7863481228668942, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 1.0424739372174067e-05, |
| "loss": 0.6734, |
| "num_tokens": 439960260.0, |
| "step": 1152 |
| }, |
| { |
| "epoch": 0.7870307167235495, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.0391690869161183e-05, |
| "loss": 0.673, |
| "num_tokens": 440357634.0, |
| "step": 1153 |
| }, |
| { |
| "epoch": 0.7877133105802048, |
| "grad_norm": 0.1162109375, |
| "learning_rate": 1.0358729633619182e-05, |
| "loss": 0.6519, |
| "num_tokens": 440781977.0, |
| "step": 1154 |
| }, |
| { |
| "epoch": 0.78839590443686, |
| "grad_norm": 0.0859375, |
| "learning_rate": 1.0325855833679438e-05, |
| "loss": 0.6495, |
| "num_tokens": 441195758.0, |
| "step": 1155 |
| }, |
| { |
| "epoch": 0.7890784982935154, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 1.0293069637027328e-05, |
| "loss": 0.6627, |
| "num_tokens": 441594791.0, |
| "step": 1156 |
| }, |
| { |
| "epoch": 0.7897610921501707, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 1.0260371210901399e-05, |
| "loss": 0.6618, |
| "num_tokens": 441995349.0, |
| "step": 1157 |
| }, |
| { |
| "epoch": 0.7904436860068259, |
| "grad_norm": 0.087890625, |
| "learning_rate": 1.0227760722092465e-05, |
| "loss": 0.6753, |
| "num_tokens": 442406971.0, |
| "step": 1158 |
| }, |
| { |
| "epoch": 0.7911262798634813, |
| "grad_norm": 0.09375, |
| "learning_rate": 1.0195238336942792e-05, |
| "loss": 0.6384, |
| "num_tokens": 442717453.0, |
| "step": 1159 |
| }, |
| { |
| "epoch": 0.7918088737201365, |
| "grad_norm": 0.08984375, |
| "learning_rate": 1.0162804221345235e-05, |
| "loss": 0.6757, |
| "num_tokens": 443118717.0, |
| "step": 1160 |
| }, |
| { |
| "epoch": 0.7924914675767918, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 1.0130458540742394e-05, |
| "loss": 0.6576, |
| "num_tokens": 443520329.0, |
| "step": 1161 |
| }, |
| { |
| "epoch": 0.7931740614334472, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 1.0098201460125792e-05, |
| "loss": 0.6324, |
| "num_tokens": 443865827.0, |
| "step": 1162 |
| }, |
| { |
| "epoch": 0.7938566552901024, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 1.0066033144034992e-05, |
| "loss": 0.6497, |
| "num_tokens": 444279796.0, |
| "step": 1163 |
| }, |
| { |
| "epoch": 0.7945392491467577, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 1.0033953756556785e-05, |
| "loss": 0.6425, |
| "num_tokens": 444622517.0, |
| "step": 1164 |
| }, |
| { |
| "epoch": 0.7952218430034129, |
| "grad_norm": 0.09765625, |
| "learning_rate": 1.0001963461324357e-05, |
| "loss": 0.6649, |
| "num_tokens": 444948197.0, |
| "step": 1165 |
| }, |
| { |
| "epoch": 0.7959044368600683, |
| "grad_norm": 0.087890625, |
| "learning_rate": 9.970062421516441e-06, |
| "loss": 0.6555, |
| "num_tokens": 445353092.0, |
| "step": 1166 |
| }, |
| { |
| "epoch": 0.7965870307167235, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 9.938250799856466e-06, |
| "loss": 0.6847, |
| "num_tokens": 445778668.0, |
| "step": 1167 |
| }, |
| { |
| "epoch": 0.7972696245733788, |
| "grad_norm": 0.087890625, |
| "learning_rate": 9.906528758611792e-06, |
| "loss": 0.6295, |
| "num_tokens": 446185709.0, |
| "step": 1168 |
| }, |
| { |
| "epoch": 0.7979522184300342, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 9.874896459592806e-06, |
| "loss": 0.6236, |
| "num_tokens": 446578569.0, |
| "step": 1169 |
| }, |
| { |
| "epoch": 0.7986348122866894, |
| "grad_norm": 0.09765625, |
| "learning_rate": 9.843354064152135e-06, |
| "loss": 0.6437, |
| "num_tokens": 446925677.0, |
| "step": 1170 |
| }, |
| { |
| "epoch": 0.7993174061433447, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 9.811901733183841e-06, |
| "loss": 0.6686, |
| "num_tokens": 447278961.0, |
| "step": 1171 |
| }, |
| { |
| "epoch": 0.8, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 9.780539627122561e-06, |
| "loss": 0.6248, |
| "num_tokens": 447628657.0, |
| "step": 1172 |
| }, |
| { |
| "epoch": 0.8006825938566553, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 9.749267905942686e-06, |
| "loss": 0.6478, |
| "num_tokens": 447994583.0, |
| "step": 1173 |
| }, |
| { |
| "epoch": 0.8013651877133106, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 9.718086729157602e-06, |
| "loss": 0.6697, |
| "num_tokens": 448367981.0, |
| "step": 1174 |
| }, |
| { |
| "epoch": 0.8020477815699659, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 9.686996255818815e-06, |
| "loss": 0.6594, |
| "num_tokens": 448739177.0, |
| "step": 1175 |
| }, |
| { |
| "epoch": 0.8027303754266212, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 9.655996644515167e-06, |
| "loss": 0.6512, |
| "num_tokens": 449148652.0, |
| "step": 1176 |
| }, |
| { |
| "epoch": 0.8034129692832764, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 9.625088053372036e-06, |
| "loss": 0.6832, |
| "num_tokens": 449557489.0, |
| "step": 1177 |
| }, |
| { |
| "epoch": 0.8040955631399317, |
| "grad_norm": 0.0986328125, |
| "learning_rate": 9.594270640050501e-06, |
| "loss": 0.6877, |
| "num_tokens": 449897223.0, |
| "step": 1178 |
| }, |
| { |
| "epoch": 0.8047781569965871, |
| "grad_norm": 0.0859375, |
| "learning_rate": 9.56354456174656e-06, |
| "loss": 0.6907, |
| "num_tokens": 450306825.0, |
| "step": 1179 |
| }, |
| { |
| "epoch": 0.8054607508532423, |
| "grad_norm": 0.080078125, |
| "learning_rate": 9.532909975190317e-06, |
| "loss": 0.657, |
| "num_tokens": 450761568.0, |
| "step": 1180 |
| }, |
| { |
| "epoch": 0.8061433447098976, |
| "grad_norm": 0.08984375, |
| "learning_rate": 9.502367036645197e-06, |
| "loss": 0.6345, |
| "num_tokens": 451135481.0, |
| "step": 1181 |
| }, |
| { |
| "epoch": 0.8068259385665529, |
| "grad_norm": 0.087890625, |
| "learning_rate": 9.471915901907124e-06, |
| "loss": 0.6666, |
| "num_tokens": 451541008.0, |
| "step": 1182 |
| }, |
| { |
| "epoch": 0.8075085324232082, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 9.441556726303768e-06, |
| "loss": 0.6612, |
| "num_tokens": 451946776.0, |
| "step": 1183 |
| }, |
| { |
| "epoch": 0.8081911262798634, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 9.411289664693702e-06, |
| "loss": 0.6467, |
| "num_tokens": 452288819.0, |
| "step": 1184 |
| }, |
| { |
| "epoch": 0.8088737201365188, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 9.381114871465644e-06, |
| "loss": 0.6133, |
| "num_tokens": 452625293.0, |
| "step": 1185 |
| }, |
| { |
| "epoch": 0.8095563139931741, |
| "grad_norm": 0.0859375, |
| "learning_rate": 9.35103250053766e-06, |
| "loss": 0.6512, |
| "num_tokens": 453021462.0, |
| "step": 1186 |
| }, |
| { |
| "epoch": 0.8102389078498293, |
| "grad_norm": 0.08251953125, |
| "learning_rate": 9.321042705356381e-06, |
| "loss": 0.6455, |
| "num_tokens": 453466997.0, |
| "step": 1187 |
| }, |
| { |
| "epoch": 0.8109215017064847, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 9.29114563889623e-06, |
| "loss": 0.6233, |
| "num_tokens": 453827598.0, |
| "step": 1188 |
| }, |
| { |
| "epoch": 0.81160409556314, |
| "grad_norm": 0.087890625, |
| "learning_rate": 9.261341453658618e-06, |
| "loss": 0.6599, |
| "num_tokens": 454208685.0, |
| "step": 1189 |
| }, |
| { |
| "epoch": 0.8122866894197952, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 9.231630301671188e-06, |
| "loss": 0.6595, |
| "num_tokens": 454558803.0, |
| "step": 1190 |
| }, |
| { |
| "epoch": 0.8129692832764505, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 9.20201233448702e-06, |
| "loss": 0.6273, |
| "num_tokens": 454926906.0, |
| "step": 1191 |
| }, |
| { |
| "epoch": 0.8136518771331058, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 9.172487703183897e-06, |
| "loss": 0.6479, |
| "num_tokens": 455311583.0, |
| "step": 1192 |
| }, |
| { |
| "epoch": 0.8143344709897611, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 9.143056558363463e-06, |
| "loss": 0.6308, |
| "num_tokens": 455715873.0, |
| "step": 1193 |
| }, |
| { |
| "epoch": 0.8150170648464163, |
| "grad_norm": 0.087890625, |
| "learning_rate": 9.113719050150543e-06, |
| "loss": 0.6105, |
| "num_tokens": 456064370.0, |
| "step": 1194 |
| }, |
| { |
| "epoch": 0.8156996587030717, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 9.084475328192308e-06, |
| "loss": 0.6466, |
| "num_tokens": 456494011.0, |
| "step": 1195 |
| }, |
| { |
| "epoch": 0.816382252559727, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 9.055325541657529e-06, |
| "loss": 0.6406, |
| "num_tokens": 456828576.0, |
| "step": 1196 |
| }, |
| { |
| "epoch": 0.8170648464163822, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 9.026269839235852e-06, |
| "loss": 0.6442, |
| "num_tokens": 457219184.0, |
| "step": 1197 |
| }, |
| { |
| "epoch": 0.8177474402730376, |
| "grad_norm": 0.08984375, |
| "learning_rate": 8.997308369136989e-06, |
| "loss": 0.6904, |
| "num_tokens": 457593165.0, |
| "step": 1198 |
| }, |
| { |
| "epoch": 0.8184300341296928, |
| "grad_norm": 0.08203125, |
| "learning_rate": 8.968441279089984e-06, |
| "loss": 0.6374, |
| "num_tokens": 458028159.0, |
| "step": 1199 |
| }, |
| { |
| "epoch": 0.8191126279863481, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 8.939668716342473e-06, |
| "loss": 0.6961, |
| "num_tokens": 458401399.0, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.8197952218430035, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 8.910990827659905e-06, |
| "loss": 0.6759, |
| "num_tokens": 458808862.0, |
| "step": 1201 |
| }, |
| { |
| "epoch": 0.8204778156996587, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 8.882407759324812e-06, |
| "loss": 0.6674, |
| "num_tokens": 459146583.0, |
| "step": 1202 |
| }, |
| { |
| "epoch": 0.821160409556314, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 8.853919657136076e-06, |
| "loss": 0.6733, |
| "num_tokens": 459578580.0, |
| "step": 1203 |
| }, |
| { |
| "epoch": 0.8218430034129692, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 8.825526666408138e-06, |
| "loss": 0.6548, |
| "num_tokens": 459918930.0, |
| "step": 1204 |
| }, |
| { |
| "epoch": 0.8225255972696246, |
| "grad_norm": 0.091796875, |
| "learning_rate": 8.797228931970305e-06, |
| "loss": 0.6742, |
| "num_tokens": 460303546.0, |
| "step": 1205 |
| }, |
| { |
| "epoch": 0.8232081911262799, |
| "grad_norm": 0.095703125, |
| "learning_rate": 8.769026598165993e-06, |
| "loss": 0.6435, |
| "num_tokens": 460623328.0, |
| "step": 1206 |
| }, |
| { |
| "epoch": 0.8238907849829351, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 8.740919808851985e-06, |
| "loss": 0.6441, |
| "num_tokens": 460992905.0, |
| "step": 1207 |
| }, |
| { |
| "epoch": 0.8245733788395905, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 8.712908707397692e-06, |
| "loss": 0.6453, |
| "num_tokens": 461346019.0, |
| "step": 1208 |
| }, |
| { |
| "epoch": 0.8252559726962457, |
| "grad_norm": 0.0830078125, |
| "learning_rate": 8.684993436684466e-06, |
| "loss": 0.6435, |
| "num_tokens": 461786553.0, |
| "step": 1209 |
| }, |
| { |
| "epoch": 0.825938566552901, |
| "grad_norm": 0.095703125, |
| "learning_rate": 8.657174139104807e-06, |
| "loss": 0.6172, |
| "num_tokens": 462132037.0, |
| "step": 1210 |
| }, |
| { |
| "epoch": 0.8266211604095564, |
| "grad_norm": 0.08056640625, |
| "learning_rate": 8.629450956561674e-06, |
| "loss": 0.6436, |
| "num_tokens": 462574179.0, |
| "step": 1211 |
| }, |
| { |
| "epoch": 0.8273037542662116, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 8.601824030467762e-06, |
| "loss": 0.6749, |
| "num_tokens": 462979260.0, |
| "step": 1212 |
| }, |
| { |
| "epoch": 0.8279863481228669, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 8.574293501744762e-06, |
| "loss": 0.6509, |
| "num_tokens": 463366323.0, |
| "step": 1213 |
| }, |
| { |
| "epoch": 0.8286689419795222, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 8.546859510822672e-06, |
| "loss": 0.603, |
| "num_tokens": 463718124.0, |
| "step": 1214 |
| }, |
| { |
| "epoch": 0.8293515358361775, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 8.519522197639043e-06, |
| "loss": 0.6162, |
| "num_tokens": 464065062.0, |
| "step": 1215 |
| }, |
| { |
| "epoch": 0.8300341296928327, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 8.49228170163829e-06, |
| "loss": 0.6711, |
| "num_tokens": 464460679.0, |
| "step": 1216 |
| }, |
| { |
| "epoch": 0.830716723549488, |
| "grad_norm": 0.08984375, |
| "learning_rate": 8.46513816177098e-06, |
| "loss": 0.6421, |
| "num_tokens": 464848866.0, |
| "step": 1217 |
| }, |
| { |
| "epoch": 0.8313993174061434, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 8.438091716493114e-06, |
| "loss": 0.6511, |
| "num_tokens": 465216487.0, |
| "step": 1218 |
| }, |
| { |
| "epoch": 0.8320819112627986, |
| "grad_norm": 0.08349609375, |
| "learning_rate": 8.411142503765415e-06, |
| "loss": 0.693, |
| "num_tokens": 465666655.0, |
| "step": 1219 |
| }, |
| { |
| "epoch": 0.8327645051194539, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 8.384290661052662e-06, |
| "loss": 0.6205, |
| "num_tokens": 466050672.0, |
| "step": 1220 |
| }, |
| { |
| "epoch": 0.8334470989761092, |
| "grad_norm": 0.091796875, |
| "learning_rate": 8.357536325322935e-06, |
| "loss": 0.6497, |
| "num_tokens": 466404932.0, |
| "step": 1221 |
| }, |
| { |
| "epoch": 0.8341296928327645, |
| "grad_norm": 0.08251953125, |
| "learning_rate": 8.330879633046945e-06, |
| "loss": 0.663, |
| "num_tokens": 466842284.0, |
| "step": 1222 |
| }, |
| { |
| "epoch": 0.8348122866894198, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 8.304320720197355e-06, |
| "loss": 0.6785, |
| "num_tokens": 467262888.0, |
| "step": 1223 |
| }, |
| { |
| "epoch": 0.8354948805460751, |
| "grad_norm": 0.087890625, |
| "learning_rate": 8.277859722248052e-06, |
| "loss": 0.6373, |
| "num_tokens": 467663696.0, |
| "step": 1224 |
| }, |
| { |
| "epoch": 0.8361774744027304, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 8.251496774173451e-06, |
| "loss": 0.6052, |
| "num_tokens": 468039326.0, |
| "step": 1225 |
| }, |
| { |
| "epoch": 0.8368600682593856, |
| "grad_norm": 0.0986328125, |
| "learning_rate": 8.22523201044787e-06, |
| "loss": 0.6079, |
| "num_tokens": 468332879.0, |
| "step": 1226 |
| }, |
| { |
| "epoch": 0.837542662116041, |
| "grad_norm": 0.08984375, |
| "learning_rate": 8.199065565044764e-06, |
| "loss": 0.6594, |
| "num_tokens": 468750924.0, |
| "step": 1227 |
| }, |
| { |
| "epoch": 0.8382252559726963, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 8.172997571436092e-06, |
| "loss": 0.6281, |
| "num_tokens": 469093445.0, |
| "step": 1228 |
| }, |
| { |
| "epoch": 0.8389078498293515, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 8.147028162591625e-06, |
| "loss": 0.6644, |
| "num_tokens": 469508535.0, |
| "step": 1229 |
| }, |
| { |
| "epoch": 0.8395904436860068, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 8.121157470978268e-06, |
| "loss": 0.6468, |
| "num_tokens": 469868923.0, |
| "step": 1230 |
| }, |
| { |
| "epoch": 0.8402730375426621, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 8.09538562855935e-06, |
| "loss": 0.6264, |
| "num_tokens": 470202533.0, |
| "step": 1231 |
| }, |
| { |
| "epoch": 0.8409556313993174, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 8.069712766794027e-06, |
| "loss": 0.644, |
| "num_tokens": 470565413.0, |
| "step": 1232 |
| }, |
| { |
| "epoch": 0.8416382252559726, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 8.044139016636535e-06, |
| "loss": 0.6511, |
| "num_tokens": 470959530.0, |
| "step": 1233 |
| }, |
| { |
| "epoch": 0.842320819112628, |
| "grad_norm": 0.087890625, |
| "learning_rate": 8.01866450853556e-06, |
| "loss": 0.6402, |
| "num_tokens": 471348091.0, |
| "step": 1234 |
| }, |
| { |
| "epoch": 0.8430034129692833, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 7.993289372433582e-06, |
| "loss": 0.6488, |
| "num_tokens": 471694644.0, |
| "step": 1235 |
| }, |
| { |
| "epoch": 0.8436860068259385, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 7.96801373776617e-06, |
| "loss": 0.6578, |
| "num_tokens": 472064862.0, |
| "step": 1236 |
| }, |
| { |
| "epoch": 0.8443686006825939, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 7.942837733461368e-06, |
| "loss": 0.6135, |
| "num_tokens": 472412332.0, |
| "step": 1237 |
| }, |
| { |
| "epoch": 0.8450511945392492, |
| "grad_norm": 0.08154296875, |
| "learning_rate": 7.917761487939008e-06, |
| "loss": 0.6792, |
| "num_tokens": 472893712.0, |
| "step": 1238 |
| }, |
| { |
| "epoch": 0.8457337883959044, |
| "grad_norm": 0.091796875, |
| "learning_rate": 7.892785129110063e-06, |
| "loss": 0.6332, |
| "num_tokens": 473286759.0, |
| "step": 1239 |
| }, |
| { |
| "epoch": 0.8464163822525598, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 7.867908784376006e-06, |
| "loss": 0.6238, |
| "num_tokens": 473648633.0, |
| "step": 1240 |
| }, |
| { |
| "epoch": 0.847098976109215, |
| "grad_norm": 0.083984375, |
| "learning_rate": 7.843132580628141e-06, |
| "loss": 0.6668, |
| "num_tokens": 474064383.0, |
| "step": 1241 |
| }, |
| { |
| "epoch": 0.8477815699658703, |
| "grad_norm": 0.08349609375, |
| "learning_rate": 7.818456644246964e-06, |
| "loss": 0.6571, |
| "num_tokens": 474478109.0, |
| "step": 1242 |
| }, |
| { |
| "epoch": 0.8484641638225257, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 7.793881101101524e-06, |
| "loss": 0.6639, |
| "num_tokens": 474860191.0, |
| "step": 1243 |
| }, |
| { |
| "epoch": 0.8491467576791809, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 7.769406076548773e-06, |
| "loss": 0.6371, |
| "num_tokens": 475216150.0, |
| "step": 1244 |
| }, |
| { |
| "epoch": 0.8498293515358362, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 7.745031695432923e-06, |
| "loss": 0.6648, |
| "num_tokens": 475597050.0, |
| "step": 1245 |
| }, |
| { |
| "epoch": 0.8505119453924914, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 7.720758082084834e-06, |
| "loss": 0.6797, |
| "num_tokens": 475981591.0, |
| "step": 1246 |
| }, |
| { |
| "epoch": 0.8511945392491468, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 7.696585360321347e-06, |
| "loss": 0.6787, |
| "num_tokens": 476323962.0, |
| "step": 1247 |
| }, |
| { |
| "epoch": 0.851877133105802, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 7.67251365344466e-06, |
| "loss": 0.6071, |
| "num_tokens": 476674609.0, |
| "step": 1248 |
| }, |
| { |
| "epoch": 0.8525597269624573, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 7.648543084241735e-06, |
| "loss": 0.6885, |
| "num_tokens": 477111104.0, |
| "step": 1249 |
| }, |
| { |
| "epoch": 0.8532423208191127, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 7.624673774983614e-06, |
| "loss": 0.6535, |
| "num_tokens": 477480036.0, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.8539249146757679, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 7.600905847424824e-06, |
| "loss": 0.6546, |
| "num_tokens": 477853375.0, |
| "step": 1251 |
| }, |
| { |
| "epoch": 0.8546075085324232, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 7.57723942280278e-06, |
| "loss": 0.6649, |
| "num_tokens": 478232122.0, |
| "step": 1252 |
| }, |
| { |
| "epoch": 0.8552901023890785, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 7.5536746218371205e-06, |
| "loss": 0.6518, |
| "num_tokens": 478648408.0, |
| "step": 1253 |
| }, |
| { |
| "epoch": 0.8559726962457338, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 7.530211564729121e-06, |
| "loss": 0.6492, |
| "num_tokens": 479034020.0, |
| "step": 1254 |
| }, |
| { |
| "epoch": 0.856655290102389, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 7.506850371161078e-06, |
| "loss": 0.6721, |
| "num_tokens": 479450987.0, |
| "step": 1255 |
| }, |
| { |
| "epoch": 0.8573378839590444, |
| "grad_norm": 0.08349609375, |
| "learning_rate": 7.483591160295693e-06, |
| "loss": 0.6726, |
| "num_tokens": 479905336.0, |
| "step": 1256 |
| }, |
| { |
| "epoch": 0.8580204778156997, |
| "grad_norm": 0.0859375, |
| "learning_rate": 7.460434050775449e-06, |
| "loss": 0.6538, |
| "num_tokens": 480304561.0, |
| "step": 1257 |
| }, |
| { |
| "epoch": 0.8587030716723549, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 7.4373791607220455e-06, |
| "loss": 0.6738, |
| "num_tokens": 480688339.0, |
| "step": 1258 |
| }, |
| { |
| "epoch": 0.8593856655290102, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 7.414426607735763e-06, |
| "loss": 0.6585, |
| "num_tokens": 481053539.0, |
| "step": 1259 |
| }, |
| { |
| "epoch": 0.8600682593856656, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 7.391576508894865e-06, |
| "loss": 0.6639, |
| "num_tokens": 481428209.0, |
| "step": 1260 |
| }, |
| { |
| "epoch": 0.8607508532423208, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 7.368828980755027e-06, |
| "loss": 0.6377, |
| "num_tokens": 481836377.0, |
| "step": 1261 |
| }, |
| { |
| "epoch": 0.8614334470989761, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 7.346184139348704e-06, |
| "loss": 0.6416, |
| "num_tokens": 482163428.0, |
| "step": 1262 |
| }, |
| { |
| "epoch": 0.8621160409556314, |
| "grad_norm": 0.091796875, |
| "learning_rate": 7.323642100184569e-06, |
| "loss": 0.634, |
| "num_tokens": 482501545.0, |
| "step": 1263 |
| }, |
| { |
| "epoch": 0.8627986348122867, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 7.301202978246907e-06, |
| "loss": 0.6614, |
| "num_tokens": 482932881.0, |
| "step": 1264 |
| }, |
| { |
| "epoch": 0.863481228668942, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 7.27886688799504e-06, |
| "loss": 0.6577, |
| "num_tokens": 483299671.0, |
| "step": 1265 |
| }, |
| { |
| "epoch": 0.8641638225255973, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 7.256633943362737e-06, |
| "loss": 0.6528, |
| "num_tokens": 483666833.0, |
| "step": 1266 |
| }, |
| { |
| "epoch": 0.8648464163822526, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 7.234504257757628e-06, |
| "loss": 0.677, |
| "num_tokens": 484087297.0, |
| "step": 1267 |
| }, |
| { |
| "epoch": 0.8655290102389078, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 7.2124779440606344e-06, |
| "loss": 0.676, |
| "num_tokens": 484503092.0, |
| "step": 1268 |
| }, |
| { |
| "epoch": 0.8662116040955632, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 7.190555114625394e-06, |
| "loss": 0.6844, |
| "num_tokens": 484886083.0, |
| "step": 1269 |
| }, |
| { |
| "epoch": 0.8668941979522184, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 7.1687358812776715e-06, |
| "loss": 0.6611, |
| "num_tokens": 485251856.0, |
| "step": 1270 |
| }, |
| { |
| "epoch": 0.8675767918088737, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 7.147020355314808e-06, |
| "loss": 0.6525, |
| "num_tokens": 485643845.0, |
| "step": 1271 |
| }, |
| { |
| "epoch": 0.868259385665529, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 7.125408647505155e-06, |
| "loss": 0.6429, |
| "num_tokens": 486021829.0, |
| "step": 1272 |
| }, |
| { |
| "epoch": 0.8689419795221843, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 7.103900868087482e-06, |
| "loss": 0.617, |
| "num_tokens": 486421963.0, |
| "step": 1273 |
| }, |
| { |
| "epoch": 0.8696245733788396, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 7.082497126770438e-06, |
| "loss": 0.6591, |
| "num_tokens": 486834100.0, |
| "step": 1274 |
| }, |
| { |
| "epoch": 0.8703071672354948, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 7.061197532731992e-06, |
| "loss": 0.6345, |
| "num_tokens": 487198446.0, |
| "step": 1275 |
| }, |
| { |
| "epoch": 0.8709897610921502, |
| "grad_norm": 0.09619140625, |
| "learning_rate": 7.040002194618872e-06, |
| "loss": 0.6657, |
| "num_tokens": 487536405.0, |
| "step": 1276 |
| }, |
| { |
| "epoch": 0.8716723549488055, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 7.0189112205459876e-06, |
| "loss": 0.6457, |
| "num_tokens": 487938500.0, |
| "step": 1277 |
| }, |
| { |
| "epoch": 0.8723549488054607, |
| "grad_norm": 0.0830078125, |
| "learning_rate": 6.997924718095924e-06, |
| "loss": 0.6421, |
| "num_tokens": 488359267.0, |
| "step": 1278 |
| }, |
| { |
| "epoch": 0.8730375426621161, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 6.9770427943183585e-06, |
| "loss": 0.6094, |
| "num_tokens": 488734453.0, |
| "step": 1279 |
| }, |
| { |
| "epoch": 0.8737201365187713, |
| "grad_norm": 0.08984375, |
| "learning_rate": 6.956265555729524e-06, |
| "loss": 0.6505, |
| "num_tokens": 489140124.0, |
| "step": 1280 |
| }, |
| { |
| "epoch": 0.8744027303754266, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 6.9355931083116735e-06, |
| "loss": 0.6277, |
| "num_tokens": 489479112.0, |
| "step": 1281 |
| }, |
| { |
| "epoch": 0.875085324232082, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 6.915025557512528e-06, |
| "loss": 0.6537, |
| "num_tokens": 489904444.0, |
| "step": 1282 |
| }, |
| { |
| "epoch": 0.8757679180887372, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 6.894563008244731e-06, |
| "loss": 0.6576, |
| "num_tokens": 490259833.0, |
| "step": 1283 |
| }, |
| { |
| "epoch": 0.8764505119453925, |
| "grad_norm": 0.087890625, |
| "learning_rate": 6.8742055648853515e-06, |
| "loss": 0.6217, |
| "num_tokens": 490647543.0, |
| "step": 1284 |
| }, |
| { |
| "epoch": 0.8771331058020477, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 6.853953331275304e-06, |
| "loss": 0.6508, |
| "num_tokens": 491054868.0, |
| "step": 1285 |
| }, |
| { |
| "epoch": 0.8778156996587031, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 6.8338064107188505e-06, |
| "loss": 0.6332, |
| "num_tokens": 491477035.0, |
| "step": 1286 |
| }, |
| { |
| "epoch": 0.8784982935153584, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 6.813764905983061e-06, |
| "loss": 0.6412, |
| "num_tokens": 491861895.0, |
| "step": 1287 |
| }, |
| { |
| "epoch": 0.8791808873720136, |
| "grad_norm": 0.0830078125, |
| "learning_rate": 6.79382891929729e-06, |
| "loss": 0.6513, |
| "num_tokens": 492277323.0, |
| "step": 1288 |
| }, |
| { |
| "epoch": 0.879863481228669, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 6.773998552352662e-06, |
| "loss": 0.6429, |
| "num_tokens": 492612171.0, |
| "step": 1289 |
| }, |
| { |
| "epoch": 0.8805460750853242, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 6.754273906301544e-06, |
| "loss": 0.6566, |
| "num_tokens": 492974236.0, |
| "step": 1290 |
| }, |
| { |
| "epoch": 0.8812286689419795, |
| "grad_norm": 0.0859375, |
| "learning_rate": 6.73465508175703e-06, |
| "loss": 0.6808, |
| "num_tokens": 493405894.0, |
| "step": 1291 |
| }, |
| { |
| "epoch": 0.8819112627986349, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 6.715142178792435e-06, |
| "loss": 0.6205, |
| "num_tokens": 493802599.0, |
| "step": 1292 |
| }, |
| { |
| "epoch": 0.8825938566552901, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 6.695735296940784e-06, |
| "loss": 0.6627, |
| "num_tokens": 494201024.0, |
| "step": 1293 |
| }, |
| { |
| "epoch": 0.8832764505119454, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 6.676434535194294e-06, |
| "loss": 0.6949, |
| "num_tokens": 494599724.0, |
| "step": 1294 |
| }, |
| { |
| "epoch": 0.8839590443686007, |
| "grad_norm": 0.08056640625, |
| "learning_rate": 6.6572399920038775e-06, |
| "loss": 0.6856, |
| "num_tokens": 495062198.0, |
| "step": 1295 |
| }, |
| { |
| "epoch": 0.884641638225256, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 6.638151765278641e-06, |
| "loss": 0.6525, |
| "num_tokens": 495459710.0, |
| "step": 1296 |
| }, |
| { |
| "epoch": 0.8853242320819112, |
| "grad_norm": 0.091796875, |
| "learning_rate": 6.61916995238538e-06, |
| "loss": 0.619, |
| "num_tokens": 495803027.0, |
| "step": 1297 |
| }, |
| { |
| "epoch": 0.8860068259385665, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 6.6002946501480936e-06, |
| "loss": 0.6217, |
| "num_tokens": 496232066.0, |
| "step": 1298 |
| }, |
| { |
| "epoch": 0.8866894197952219, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 6.581525954847471e-06, |
| "loss": 0.6206, |
| "num_tokens": 496623086.0, |
| "step": 1299 |
| }, |
| { |
| "epoch": 0.8873720136518771, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 6.562863962220414e-06, |
| "loss": 0.6818, |
| "num_tokens": 496987511.0, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.8880546075085324, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 6.5443087674595614e-06, |
| "loss": 0.6346, |
| "num_tokens": 497347539.0, |
| "step": 1301 |
| }, |
| { |
| "epoch": 0.8887372013651877, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 6.525860465212774e-06, |
| "loss": 0.6829, |
| "num_tokens": 497766535.0, |
| "step": 1302 |
| }, |
| { |
| "epoch": 0.889419795221843, |
| "grad_norm": 0.0859375, |
| "learning_rate": 6.507519149582659e-06, |
| "loss": 0.6585, |
| "num_tokens": 498155097.0, |
| "step": 1303 |
| }, |
| { |
| "epoch": 0.8901023890784983, |
| "grad_norm": 0.087890625, |
| "learning_rate": 6.489284914126121e-06, |
| "loss": 0.6673, |
| "num_tokens": 498572700.0, |
| "step": 1304 |
| }, |
| { |
| "epoch": 0.8907849829351536, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 6.47115785185384e-06, |
| "loss": 0.6292, |
| "num_tokens": 498971917.0, |
| "step": 1305 |
| }, |
| { |
| "epoch": 0.8914675767918089, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 6.453138055229828e-06, |
| "loss": 0.6496, |
| "num_tokens": 499395757.0, |
| "step": 1306 |
| }, |
| { |
| "epoch": 0.8921501706484641, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 6.435225616170952e-06, |
| "loss": 0.6412, |
| "num_tokens": 499768342.0, |
| "step": 1307 |
| }, |
| { |
| "epoch": 0.8928327645051195, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 6.417420626046457e-06, |
| "loss": 0.6584, |
| "num_tokens": 500168759.0, |
| "step": 1308 |
| }, |
| { |
| "epoch": 0.8935153583617748, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 6.399723175677487e-06, |
| "loss": 0.6482, |
| "num_tokens": 500587148.0, |
| "step": 1309 |
| }, |
| { |
| "epoch": 0.89419795221843, |
| "grad_norm": 0.087890625, |
| "learning_rate": 6.382133355336667e-06, |
| "loss": 0.6232, |
| "num_tokens": 500947010.0, |
| "step": 1310 |
| }, |
| { |
| "epoch": 0.8948805460750853, |
| "grad_norm": 0.08251953125, |
| "learning_rate": 6.364651254747593e-06, |
| "loss": 0.6766, |
| "num_tokens": 501405813.0, |
| "step": 1311 |
| }, |
| { |
| "epoch": 0.8955631399317406, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 6.347276963084405e-06, |
| "loss": 0.6628, |
| "num_tokens": 501807587.0, |
| "step": 1312 |
| }, |
| { |
| "epoch": 0.8962457337883959, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 6.330010568971316e-06, |
| "loss": 0.6391, |
| "num_tokens": 502170916.0, |
| "step": 1313 |
| }, |
| { |
| "epoch": 0.8969283276450511, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 6.3128521604821755e-06, |
| "loss": 0.6681, |
| "num_tokens": 502552329.0, |
| "step": 1314 |
| }, |
| { |
| "epoch": 0.8976109215017065, |
| "grad_norm": 0.087890625, |
| "learning_rate": 6.29580182514e-06, |
| "loss": 0.6693, |
| "num_tokens": 502958987.0, |
| "step": 1315 |
| }, |
| { |
| "epoch": 0.8982935153583618, |
| "grad_norm": 0.08056640625, |
| "learning_rate": 6.2788596499165496e-06, |
| "loss": 0.6402, |
| "num_tokens": 503407695.0, |
| "step": 1316 |
| }, |
| { |
| "epoch": 0.898976109215017, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 6.262025721231857e-06, |
| "loss": 0.6635, |
| "num_tokens": 503767045.0, |
| "step": 1317 |
| }, |
| { |
| "epoch": 0.8996587030716724, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 6.245300124953813e-06, |
| "loss": 0.6423, |
| "num_tokens": 504144988.0, |
| "step": 1318 |
| }, |
| { |
| "epoch": 0.9003412969283277, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 6.228682946397719e-06, |
| "loss": 0.6489, |
| "num_tokens": 504511747.0, |
| "step": 1319 |
| }, |
| { |
| "epoch": 0.9010238907849829, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 6.212174270325845e-06, |
| "loss": 0.6269, |
| "num_tokens": 504893463.0, |
| "step": 1320 |
| }, |
| { |
| "epoch": 0.9017064846416383, |
| "grad_norm": 0.087890625, |
| "learning_rate": 6.195774180947006e-06, |
| "loss": 0.6321, |
| "num_tokens": 505274632.0, |
| "step": 1321 |
| }, |
| { |
| "epoch": 0.9023890784982935, |
| "grad_norm": 0.08203125, |
| "learning_rate": 6.179482761916123e-06, |
| "loss": 0.6502, |
| "num_tokens": 505718073.0, |
| "step": 1322 |
| }, |
| { |
| "epoch": 0.9030716723549488, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 6.163300096333806e-06, |
| "loss": 0.6879, |
| "num_tokens": 506090578.0, |
| "step": 1323 |
| }, |
| { |
| "epoch": 0.903754266211604, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 6.147226266745932e-06, |
| "loss": 0.6395, |
| "num_tokens": 506453806.0, |
| "step": 1324 |
| }, |
| { |
| "epoch": 0.9044368600682594, |
| "grad_norm": 0.08251953125, |
| "learning_rate": 6.1312613551432135e-06, |
| "loss": 0.6257, |
| "num_tokens": 506861395.0, |
| "step": 1325 |
| }, |
| { |
| "epoch": 0.9051194539249147, |
| "grad_norm": 0.091796875, |
| "learning_rate": 6.115405442960781e-06, |
| "loss": 0.6645, |
| "num_tokens": 507221671.0, |
| "step": 1326 |
| }, |
| { |
| "epoch": 0.9058020477815699, |
| "grad_norm": 0.0859375, |
| "learning_rate": 6.0996586110777815e-06, |
| "loss": 0.6598, |
| "num_tokens": 507622936.0, |
| "step": 1327 |
| }, |
| { |
| "epoch": 0.9064846416382253, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 6.084020939816946e-06, |
| "loss": 0.638, |
| "num_tokens": 507979226.0, |
| "step": 1328 |
| }, |
| { |
| "epoch": 0.9071672354948805, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 6.068492508944193e-06, |
| "loss": 0.6426, |
| "num_tokens": 508329436.0, |
| "step": 1329 |
| }, |
| { |
| "epoch": 0.9078498293515358, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 6.053073397668226e-06, |
| "loss": 0.6421, |
| "num_tokens": 508690960.0, |
| "step": 1330 |
| }, |
| { |
| "epoch": 0.9085324232081912, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 6.037763684640113e-06, |
| "loss": 0.6186, |
| "num_tokens": 509044002.0, |
| "step": 1331 |
| }, |
| { |
| "epoch": 0.9092150170648464, |
| "grad_norm": 0.09814453125, |
| "learning_rate": 6.022563447952896e-06, |
| "loss": 0.6725, |
| "num_tokens": 509412864.0, |
| "step": 1332 |
| }, |
| { |
| "epoch": 0.9098976109215017, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 6.0074727651412e-06, |
| "loss": 0.6647, |
| "num_tokens": 509848958.0, |
| "step": 1333 |
| }, |
| { |
| "epoch": 0.910580204778157, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 5.992491713180819e-06, |
| "loss": 0.6378, |
| "num_tokens": 510213306.0, |
| "step": 1334 |
| }, |
| { |
| "epoch": 0.9112627986348123, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 5.977620368488328e-06, |
| "loss": 0.6455, |
| "num_tokens": 510627944.0, |
| "step": 1335 |
| }, |
| { |
| "epoch": 0.9119453924914676, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 5.962858806920716e-06, |
| "loss": 0.6538, |
| "num_tokens": 510994784.0, |
| "step": 1336 |
| }, |
| { |
| "epoch": 0.9126279863481229, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 5.948207103774966e-06, |
| "loss": 0.6139, |
| "num_tokens": 511354220.0, |
| "step": 1337 |
| }, |
| { |
| "epoch": 0.9133105802047782, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 5.933665333787686e-06, |
| "loss": 0.6809, |
| "num_tokens": 511781960.0, |
| "step": 1338 |
| }, |
| { |
| "epoch": 0.9139931740614334, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 5.919233571134742e-06, |
| "loss": 0.6854, |
| "num_tokens": 512183340.0, |
| "step": 1339 |
| }, |
| { |
| "epoch": 0.9146757679180887, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 5.904911889430853e-06, |
| "loss": 0.6821, |
| "num_tokens": 512558436.0, |
| "step": 1340 |
| }, |
| { |
| "epoch": 0.9153583617747441, |
| "grad_norm": 0.091796875, |
| "learning_rate": 5.8907003617292294e-06, |
| "loss": 0.631, |
| "num_tokens": 512899634.0, |
| "step": 1341 |
| }, |
| { |
| "epoch": 0.9160409556313993, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 5.876599060521206e-06, |
| "loss": 0.6691, |
| "num_tokens": 513284117.0, |
| "step": 1342 |
| }, |
| { |
| "epoch": 0.9167235494880546, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 5.8626080577358526e-06, |
| "loss": 0.6231, |
| "num_tokens": 513639477.0, |
| "step": 1343 |
| }, |
| { |
| "epoch": 0.9174061433447099, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 5.848727424739629e-06, |
| "loss": 0.6029, |
| "num_tokens": 514038300.0, |
| "step": 1344 |
| }, |
| { |
| "epoch": 0.9180887372013652, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 5.834957232336018e-06, |
| "loss": 0.6575, |
| "num_tokens": 514407245.0, |
| "step": 1345 |
| }, |
| { |
| "epoch": 0.9187713310580204, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 5.82129755076514e-06, |
| "loss": 0.6137, |
| "num_tokens": 514797002.0, |
| "step": 1346 |
| }, |
| { |
| "epoch": 0.9194539249146758, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 5.8077484497034306e-06, |
| "loss": 0.6529, |
| "num_tokens": 515153030.0, |
| "step": 1347 |
| }, |
| { |
| "epoch": 0.9201365187713311, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 5.794309998263256e-06, |
| "loss": 0.6166, |
| "num_tokens": 515517449.0, |
| "step": 1348 |
| }, |
| { |
| "epoch": 0.9208191126279863, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 5.780982264992571e-06, |
| "loss": 0.6898, |
| "num_tokens": 515939848.0, |
| "step": 1349 |
| }, |
| { |
| "epoch": 0.9215017064846417, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 5.7677653178745805e-06, |
| "loss": 0.6963, |
| "num_tokens": 516301089.0, |
| "step": 1350 |
| }, |
| { |
| "epoch": 0.922184300341297, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 5.754659224327375e-06, |
| "loss": 0.6301, |
| "num_tokens": 516650879.0, |
| "step": 1351 |
| }, |
| { |
| "epoch": 0.9228668941979522, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 5.741664051203592e-06, |
| "loss": 0.657, |
| "num_tokens": 517060505.0, |
| "step": 1352 |
| }, |
| { |
| "epoch": 0.9235494880546075, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 5.728779864790079e-06, |
| "loss": 0.6758, |
| "num_tokens": 517461503.0, |
| "step": 1353 |
| }, |
| { |
| "epoch": 0.9242320819112628, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 5.716006730807559e-06, |
| "loss": 0.6976, |
| "num_tokens": 517812271.0, |
| "step": 1354 |
| }, |
| { |
| "epoch": 0.9249146757679181, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 5.7033447144102785e-06, |
| "loss": 0.6545, |
| "num_tokens": 518223298.0, |
| "step": 1355 |
| }, |
| { |
| "epoch": 0.9255972696245733, |
| "grad_norm": 0.09375, |
| "learning_rate": 5.690793880185699e-06, |
| "loss": 0.6566, |
| "num_tokens": 518582017.0, |
| "step": 1356 |
| }, |
| { |
| "epoch": 0.9262798634812287, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 5.6783542921541426e-06, |
| "loss": 0.6495, |
| "num_tokens": 518955685.0, |
| "step": 1357 |
| }, |
| { |
| "epoch": 0.926962457337884, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 5.666026013768487e-06, |
| "loss": 0.6262, |
| "num_tokens": 519300948.0, |
| "step": 1358 |
| }, |
| { |
| "epoch": 0.9276450511945392, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 5.65380910791383e-06, |
| "loss": 0.6588, |
| "num_tokens": 519669495.0, |
| "step": 1359 |
| }, |
| { |
| "epoch": 0.9283276450511946, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 5.641703636907171e-06, |
| "loss": 0.678, |
| "num_tokens": 520051962.0, |
| "step": 1360 |
| }, |
| { |
| "epoch": 0.9290102389078498, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 5.6297096624970815e-06, |
| "loss": 0.6249, |
| "num_tokens": 520405287.0, |
| "step": 1361 |
| }, |
| { |
| "epoch": 0.9296928327645051, |
| "grad_norm": 0.0947265625, |
| "learning_rate": 5.617827245863427e-06, |
| "loss": 0.6416, |
| "num_tokens": 520768023.0, |
| "step": 1362 |
| }, |
| { |
| "epoch": 0.9303754266211605, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 5.606056447617009e-06, |
| "loss": 0.6282, |
| "num_tokens": 521126358.0, |
| "step": 1363 |
| }, |
| { |
| "epoch": 0.9310580204778157, |
| "grad_norm": 0.0859375, |
| "learning_rate": 5.594397327799285e-06, |
| "loss": 0.6578, |
| "num_tokens": 521526236.0, |
| "step": 1364 |
| }, |
| { |
| "epoch": 0.931740614334471, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 5.582849945882055e-06, |
| "loss": 0.627, |
| "num_tokens": 521910187.0, |
| "step": 1365 |
| }, |
| { |
| "epoch": 0.9324232081911262, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 5.57141436076715e-06, |
| "loss": 0.6502, |
| "num_tokens": 522278881.0, |
| "step": 1366 |
| }, |
| { |
| "epoch": 0.9331058020477816, |
| "grad_norm": 0.0859375, |
| "learning_rate": 5.560090630786145e-06, |
| "loss": 0.6382, |
| "num_tokens": 522675014.0, |
| "step": 1367 |
| }, |
| { |
| "epoch": 0.9337883959044369, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 5.548878813700058e-06, |
| "loss": 0.5973, |
| "num_tokens": 523056725.0, |
| "step": 1368 |
| }, |
| { |
| "epoch": 0.9344709897610921, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 5.5377789666990435e-06, |
| "loss": 0.6418, |
| "num_tokens": 523472668.0, |
| "step": 1369 |
| }, |
| { |
| "epoch": 0.9351535836177475, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 5.526791146402112e-06, |
| "loss": 0.7105, |
| "num_tokens": 523880084.0, |
| "step": 1370 |
| }, |
| { |
| "epoch": 0.9358361774744027, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 5.515915408856848e-06, |
| "loss": 0.6482, |
| "num_tokens": 524273413.0, |
| "step": 1371 |
| }, |
| { |
| "epoch": 0.936518771331058, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 5.505151809539106e-06, |
| "loss": 0.6564, |
| "num_tokens": 524621998.0, |
| "step": 1372 |
| }, |
| { |
| "epoch": 0.9372013651877134, |
| "grad_norm": 0.0966796875, |
| "learning_rate": 5.494500403352739e-06, |
| "loss": 0.6738, |
| "num_tokens": 524982312.0, |
| "step": 1373 |
| }, |
| { |
| "epoch": 0.9378839590443686, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 5.483961244629314e-06, |
| "loss": 0.6895, |
| "num_tokens": 525445228.0, |
| "step": 1374 |
| }, |
| { |
| "epoch": 0.9385665529010239, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 5.473534387127838e-06, |
| "loss": 0.6087, |
| "num_tokens": 525832920.0, |
| "step": 1375 |
| }, |
| { |
| "epoch": 0.9392491467576792, |
| "grad_norm": 0.08984375, |
| "learning_rate": 5.463219884034489e-06, |
| "loss": 0.6777, |
| "num_tokens": 526255048.0, |
| "step": 1376 |
| }, |
| { |
| "epoch": 0.9399317406143345, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 5.453017787962329e-06, |
| "loss": 0.6382, |
| "num_tokens": 526635531.0, |
| "step": 1377 |
| }, |
| { |
| "epoch": 0.9406143344709897, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 5.442928150951053e-06, |
| "loss": 0.6567, |
| "num_tokens": 527026047.0, |
| "step": 1378 |
| }, |
| { |
| "epoch": 0.941296928327645, |
| "grad_norm": 0.0859375, |
| "learning_rate": 5.4329510244667055e-06, |
| "loss": 0.6572, |
| "num_tokens": 527399527.0, |
| "step": 1379 |
| }, |
| { |
| "epoch": 0.9419795221843004, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 5.423086459401437e-06, |
| "loss": 0.6629, |
| "num_tokens": 527753736.0, |
| "step": 1380 |
| }, |
| { |
| "epoch": 0.9426621160409556, |
| "grad_norm": 0.083984375, |
| "learning_rate": 5.4133345060732336e-06, |
| "loss": 0.6535, |
| "num_tokens": 528179295.0, |
| "step": 1381 |
| }, |
| { |
| "epoch": 0.9433447098976109, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 5.40369521422566e-06, |
| "loss": 0.6302, |
| "num_tokens": 528582126.0, |
| "step": 1382 |
| }, |
| { |
| "epoch": 0.9440273037542662, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 5.394168633027613e-06, |
| "loss": 0.7068, |
| "num_tokens": 528988386.0, |
| "step": 1383 |
| }, |
| { |
| "epoch": 0.9447098976109215, |
| "grad_norm": 0.091796875, |
| "learning_rate": 5.384754811073055e-06, |
| "loss": 0.6499, |
| "num_tokens": 529361801.0, |
| "step": 1384 |
| }, |
| { |
| "epoch": 0.9453924914675768, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 5.37545379638079e-06, |
| "loss": 0.6535, |
| "num_tokens": 529711197.0, |
| "step": 1385 |
| }, |
| { |
| "epoch": 0.9460750853242321, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 5.366265636394199e-06, |
| "loss": 0.6419, |
| "num_tokens": 530106526.0, |
| "step": 1386 |
| }, |
| { |
| "epoch": 0.9467576791808874, |
| "grad_norm": 0.091796875, |
| "learning_rate": 5.357190377980996e-06, |
| "loss": 0.6289, |
| "num_tokens": 530439137.0, |
| "step": 1387 |
| }, |
| { |
| "epoch": 0.9474402730375426, |
| "grad_norm": 0.08984375, |
| "learning_rate": 5.3482280674330136e-06, |
| "loss": 0.6975, |
| "num_tokens": 530843865.0, |
| "step": 1388 |
| }, |
| { |
| "epoch": 0.948122866894198, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 5.339378750465943e-06, |
| "loss": 0.6284, |
| "num_tokens": 531183148.0, |
| "step": 1389 |
| }, |
| { |
| "epoch": 0.9488054607508533, |
| "grad_norm": 0.091796875, |
| "learning_rate": 5.330642472219102e-06, |
| "loss": 0.6641, |
| "num_tokens": 531553862.0, |
| "step": 1390 |
| }, |
| { |
| "epoch": 0.9494880546075085, |
| "grad_norm": 0.08984375, |
| "learning_rate": 5.322019277255226e-06, |
| "loss": 0.652, |
| "num_tokens": 531927487.0, |
| "step": 1391 |
| }, |
| { |
| "epoch": 0.9501706484641638, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 5.3135092095602105e-06, |
| "loss": 0.6664, |
| "num_tokens": 532305154.0, |
| "step": 1392 |
| }, |
| { |
| "epoch": 0.9508532423208191, |
| "grad_norm": 0.09375, |
| "learning_rate": 5.305112312542913e-06, |
| "loss": 0.6881, |
| "num_tokens": 532688223.0, |
| "step": 1393 |
| }, |
| { |
| "epoch": 0.9515358361774744, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 5.296828629034914e-06, |
| "loss": 0.6605, |
| "num_tokens": 533064460.0, |
| "step": 1394 |
| }, |
| { |
| "epoch": 0.9522184300341296, |
| "grad_norm": 0.08984375, |
| "learning_rate": 5.28865820129031e-06, |
| "loss": 0.6554, |
| "num_tokens": 533448215.0, |
| "step": 1395 |
| }, |
| { |
| "epoch": 0.952901023890785, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 5.280601070985488e-06, |
| "loss": 0.6604, |
| "num_tokens": 533840482.0, |
| "step": 1396 |
| }, |
| { |
| "epoch": 0.9535836177474403, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 5.272657279218926e-06, |
| "loss": 0.6434, |
| "num_tokens": 534201616.0, |
| "step": 1397 |
| }, |
| { |
| "epoch": 0.9542662116040955, |
| "grad_norm": 0.087890625, |
| "learning_rate": 5.264826866510967e-06, |
| "loss": 0.6639, |
| "num_tokens": 534587138.0, |
| "step": 1398 |
| }, |
| { |
| "epoch": 0.9549488054607509, |
| "grad_norm": 0.087890625, |
| "learning_rate": 5.257109872803629e-06, |
| "loss": 0.6243, |
| "num_tokens": 534962041.0, |
| "step": 1399 |
| }, |
| { |
| "epoch": 0.9556313993174061, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 5.24950633746038e-06, |
| "loss": 0.6479, |
| "num_tokens": 535341346.0, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.9563139931740614, |
| "grad_norm": 0.083984375, |
| "learning_rate": 5.2420162992659665e-06, |
| "loss": 0.6107, |
| "num_tokens": 535755631.0, |
| "step": 1401 |
| }, |
| { |
| "epoch": 0.9569965870307168, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 5.234639796426189e-06, |
| "loss": 0.6292, |
| "num_tokens": 536098184.0, |
| "step": 1402 |
| }, |
| { |
| "epoch": 0.957679180887372, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 5.227376866567717e-06, |
| "loss": 0.6496, |
| "num_tokens": 536504433.0, |
| "step": 1403 |
| }, |
| { |
| "epoch": 0.9583617747440273, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 5.220227546737905e-06, |
| "loss": 0.6564, |
| "num_tokens": 536922129.0, |
| "step": 1404 |
| }, |
| { |
| "epoch": 0.9590443686006825, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 5.213191873404592e-06, |
| "loss": 0.6291, |
| "num_tokens": 537336227.0, |
| "step": 1405 |
| }, |
| { |
| "epoch": 0.9597269624573379, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 5.2062698824559174e-06, |
| "loss": 0.6276, |
| "num_tokens": 537717714.0, |
| "step": 1406 |
| }, |
| { |
| "epoch": 0.9604095563139932, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 5.199461609200142e-06, |
| "loss": 0.6655, |
| "num_tokens": 538121187.0, |
| "step": 1407 |
| }, |
| { |
| "epoch": 0.9610921501706484, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 5.1927670883654665e-06, |
| "loss": 0.6679, |
| "num_tokens": 538510097.0, |
| "step": 1408 |
| }, |
| { |
| "epoch": 0.9617747440273038, |
| "grad_norm": 0.08984375, |
| "learning_rate": 5.186186354099854e-06, |
| "loss": 0.6392, |
| "num_tokens": 538869301.0, |
| "step": 1409 |
| }, |
| { |
| "epoch": 0.962457337883959, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 5.179719439970854e-06, |
| "loss": 0.6066, |
| "num_tokens": 539226192.0, |
| "step": 1410 |
| }, |
| { |
| "epoch": 0.9631399317406143, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 5.173366378965439e-06, |
| "loss": 0.664, |
| "num_tokens": 539615473.0, |
| "step": 1411 |
| }, |
| { |
| "epoch": 0.9638225255972697, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 5.167127203489819e-06, |
| "loss": 0.6491, |
| "num_tokens": 540037038.0, |
| "step": 1412 |
| }, |
| { |
| "epoch": 0.9645051194539249, |
| "grad_norm": 0.09765625, |
| "learning_rate": 5.161001945369297e-06, |
| "loss": 0.6324, |
| "num_tokens": 540364652.0, |
| "step": 1413 |
| }, |
| { |
| "epoch": 0.9651877133105802, |
| "grad_norm": 0.09912109375, |
| "learning_rate": 5.154990635848092e-06, |
| "loss": 0.6369, |
| "num_tokens": 540682753.0, |
| "step": 1414 |
| }, |
| { |
| "epoch": 0.9658703071672355, |
| "grad_norm": 0.091796875, |
| "learning_rate": 5.14909330558919e-06, |
| "loss": 0.6627, |
| "num_tokens": 541043880.0, |
| "step": 1415 |
| }, |
| { |
| "epoch": 0.9665529010238908, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 5.143309984674172e-06, |
| "loss": 0.6881, |
| "num_tokens": 541421088.0, |
| "step": 1416 |
| }, |
| { |
| "epoch": 0.967235494880546, |
| "grad_norm": 0.08984375, |
| "learning_rate": 5.1376407026030834e-06, |
| "loss": 0.6735, |
| "num_tokens": 541784415.0, |
| "step": 1417 |
| }, |
| { |
| "epoch": 0.9679180887372013, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 5.132085488294262e-06, |
| "loss": 0.6053, |
| "num_tokens": 542148721.0, |
| "step": 1418 |
| }, |
| { |
| "epoch": 0.9686006825938567, |
| "grad_norm": 0.0908203125, |
| "learning_rate": 5.1266443700842e-06, |
| "loss": 0.6295, |
| "num_tokens": 542500193.0, |
| "step": 1419 |
| }, |
| { |
| "epoch": 0.9692832764505119, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 5.121317375727405e-06, |
| "loss": 0.6624, |
| "num_tokens": 542913487.0, |
| "step": 1420 |
| }, |
| { |
| "epoch": 0.9699658703071672, |
| "grad_norm": 0.083984375, |
| "learning_rate": 5.116104532396245e-06, |
| "loss": 0.6195, |
| "num_tokens": 543310490.0, |
| "step": 1421 |
| }, |
| { |
| "epoch": 0.9706484641638226, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 5.111005866680819e-06, |
| "loss": 0.6805, |
| "num_tokens": 543709489.0, |
| "step": 1422 |
| }, |
| { |
| "epoch": 0.9713310580204778, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 5.106021404588827e-06, |
| "loss": 0.7197, |
| "num_tokens": 544131831.0, |
| "step": 1423 |
| }, |
| { |
| "epoch": 0.9720136518771331, |
| "grad_norm": 0.10107421875, |
| "learning_rate": 5.101151171545417e-06, |
| "loss": 0.6878, |
| "num_tokens": 544482942.0, |
| "step": 1424 |
| }, |
| { |
| "epoch": 0.9726962457337884, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 5.0963951923930745e-06, |
| "loss": 0.6347, |
| "num_tokens": 544851416.0, |
| "step": 1425 |
| }, |
| { |
| "epoch": 0.9733788395904437, |
| "grad_norm": 0.08984375, |
| "learning_rate": 5.091753491391495e-06, |
| "loss": 0.6317, |
| "num_tokens": 545221293.0, |
| "step": 1426 |
| }, |
| { |
| "epoch": 0.974061433447099, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 5.087226092217444e-06, |
| "loss": 0.7013, |
| "num_tokens": 545616055.0, |
| "step": 1427 |
| }, |
| { |
| "epoch": 0.9747440273037543, |
| "grad_norm": 0.09912109375, |
| "learning_rate": 5.082813017964652e-06, |
| "loss": 0.6009, |
| "num_tokens": 545899652.0, |
| "step": 1428 |
| }, |
| { |
| "epoch": 0.9754266211604096, |
| "grad_norm": 0.09326171875, |
| "learning_rate": 5.078514291143693e-06, |
| "loss": 0.6658, |
| "num_tokens": 546257896.0, |
| "step": 1429 |
| }, |
| { |
| "epoch": 0.9761092150170648, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 5.074329933681866e-06, |
| "loss": 0.684, |
| "num_tokens": 546668069.0, |
| "step": 1430 |
| }, |
| { |
| "epoch": 0.9767918088737202, |
| "grad_norm": 0.087890625, |
| "learning_rate": 5.070259966923086e-06, |
| "loss": 0.6504, |
| "num_tokens": 547025827.0, |
| "step": 1431 |
| }, |
| { |
| "epoch": 0.9774744027303754, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 5.066304411627771e-06, |
| "loss": 0.6275, |
| "num_tokens": 547384681.0, |
| "step": 1432 |
| }, |
| { |
| "epoch": 0.9781569965870307, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 5.062463287972747e-06, |
| "loss": 0.672, |
| "num_tokens": 547814225.0, |
| "step": 1433 |
| }, |
| { |
| "epoch": 0.978839590443686, |
| "grad_norm": 0.0830078125, |
| "learning_rate": 5.058736615551134e-06, |
| "loss": 0.6357, |
| "num_tokens": 548234763.0, |
| "step": 1434 |
| }, |
| { |
| "epoch": 0.9795221843003413, |
| "grad_norm": 0.091796875, |
| "learning_rate": 5.055124413372245e-06, |
| "loss": 0.6572, |
| "num_tokens": 548580998.0, |
| "step": 1435 |
| }, |
| { |
| "epoch": 0.9802047781569966, |
| "grad_norm": 0.08984375, |
| "learning_rate": 5.051626699861503e-06, |
| "loss": 0.6495, |
| "num_tokens": 548956744.0, |
| "step": 1436 |
| }, |
| { |
| "epoch": 0.9808873720136518, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 5.04824349286033e-06, |
| "loss": 0.6528, |
| "num_tokens": 549304722.0, |
| "step": 1437 |
| }, |
| { |
| "epoch": 0.9815699658703072, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 5.044974809626069e-06, |
| "loss": 0.6503, |
| "num_tokens": 549678469.0, |
| "step": 1438 |
| }, |
| { |
| "epoch": 0.9822525597269625, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 5.041820666831883e-06, |
| "loss": 0.6359, |
| "num_tokens": 550089480.0, |
| "step": 1439 |
| }, |
| { |
| "epoch": 0.9829351535836177, |
| "grad_norm": 0.0927734375, |
| "learning_rate": 5.038781080566688e-06, |
| "loss": 0.6511, |
| "num_tokens": 550468465.0, |
| "step": 1440 |
| }, |
| { |
| "epoch": 0.9836177474402731, |
| "grad_norm": 0.09130859375, |
| "learning_rate": 5.035856066335055e-06, |
| "loss": 0.6427, |
| "num_tokens": 550819756.0, |
| "step": 1441 |
| }, |
| { |
| "epoch": 0.9843003412969283, |
| "grad_norm": 0.09423828125, |
| "learning_rate": 5.033045639057135e-06, |
| "loss": 0.6171, |
| "num_tokens": 551168125.0, |
| "step": 1442 |
| }, |
| { |
| "epoch": 0.9849829351535836, |
| "grad_norm": 0.08544921875, |
| "learning_rate": 5.030349813068592e-06, |
| "loss": 0.6588, |
| "num_tokens": 551575993.0, |
| "step": 1443 |
| }, |
| { |
| "epoch": 0.985665529010239, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 5.027768602120511e-06, |
| "loss": 0.679, |
| "num_tokens": 552037158.0, |
| "step": 1444 |
| }, |
| { |
| "epoch": 0.9863481228668942, |
| "grad_norm": 0.091796875, |
| "learning_rate": 5.025302019379348e-06, |
| "loss": 0.6588, |
| "num_tokens": 552397091.0, |
| "step": 1445 |
| }, |
| { |
| "epoch": 0.9870307167235495, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 5.022950077426851e-06, |
| "loss": 0.6316, |
| "num_tokens": 552764464.0, |
| "step": 1446 |
| }, |
| { |
| "epoch": 0.9877133105802047, |
| "grad_norm": 0.0888671875, |
| "learning_rate": 5.020712788259997e-06, |
| "loss": 0.5986, |
| "num_tokens": 553136719.0, |
| "step": 1447 |
| }, |
| { |
| "epoch": 0.9883959044368601, |
| "grad_norm": 0.091796875, |
| "learning_rate": 5.018590163290938e-06, |
| "loss": 0.6561, |
| "num_tokens": 553531687.0, |
| "step": 1448 |
| }, |
| { |
| "epoch": 0.9890784982935154, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 5.016582213346931e-06, |
| "loss": 0.6692, |
| "num_tokens": 553945058.0, |
| "step": 1449 |
| }, |
| { |
| "epoch": 0.9897610921501706, |
| "grad_norm": 0.08642578125, |
| "learning_rate": 5.014688948670296e-06, |
| "loss": 0.6163, |
| "num_tokens": 554315309.0, |
| "step": 1450 |
| }, |
| { |
| "epoch": 0.990443686006826, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 5.01291037891835e-06, |
| "loss": 0.6488, |
| "num_tokens": 554683452.0, |
| "step": 1451 |
| }, |
| { |
| "epoch": 0.9911262798634812, |
| "grad_norm": 0.09375, |
| "learning_rate": 5.01124651316337e-06, |
| "loss": 0.6262, |
| "num_tokens": 555035565.0, |
| "step": 1452 |
| }, |
| { |
| "epoch": 0.9918088737201365, |
| "grad_norm": 0.0849609375, |
| "learning_rate": 5.009697359892536e-06, |
| "loss": 0.664, |
| "num_tokens": 555434452.0, |
| "step": 1453 |
| }, |
| { |
| "epoch": 0.9924914675767919, |
| "grad_norm": 0.09716796875, |
| "learning_rate": 5.008262927007901e-06, |
| "loss": 0.6745, |
| "num_tokens": 555777623.0, |
| "step": 1454 |
| }, |
| { |
| "epoch": 0.9931740614334471, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 5.006943221826337e-06, |
| "loss": 0.6941, |
| "num_tokens": 556147373.0, |
| "step": 1455 |
| }, |
| { |
| "epoch": 0.9938566552901024, |
| "grad_norm": 0.08740234375, |
| "learning_rate": 5.005738251079502e-06, |
| "loss": 0.6767, |
| "num_tokens": 556544624.0, |
| "step": 1456 |
| }, |
| { |
| "epoch": 0.9945392491467577, |
| "grad_norm": 0.09228515625, |
| "learning_rate": 5.004648020913811e-06, |
| "loss": 0.6647, |
| "num_tokens": 556901995.0, |
| "step": 1457 |
| }, |
| { |
| "epoch": 0.995221843003413, |
| "grad_norm": 0.08837890625, |
| "learning_rate": 5.003672536890401e-06, |
| "loss": 0.6182, |
| "num_tokens": 557284239.0, |
| "step": 1458 |
| }, |
| { |
| "epoch": 0.9959044368600682, |
| "grad_norm": 0.08447265625, |
| "learning_rate": 5.002811803985101e-06, |
| "loss": 0.655, |
| "num_tokens": 557725176.0, |
| "step": 1459 |
| }, |
| { |
| "epoch": 0.9965870307167235, |
| "grad_norm": 0.0859375, |
| "learning_rate": 5.002065826588409e-06, |
| "loss": 0.6155, |
| "num_tokens": 558093284.0, |
| "step": 1460 |
| }, |
| { |
| "epoch": 0.9972696245733789, |
| "grad_norm": 0.09521484375, |
| "learning_rate": 5.001434608505466e-06, |
| "loss": 0.6573, |
| "num_tokens": 558448894.0, |
| "step": 1461 |
| }, |
| { |
| "epoch": 0.9979522184300341, |
| "grad_norm": 0.09033203125, |
| "learning_rate": 5.000918152956042e-06, |
| "loss": 0.6626, |
| "num_tokens": 558834290.0, |
| "step": 1462 |
| }, |
| { |
| "epoch": 0.9986348122866894, |
| "grad_norm": 0.091796875, |
| "learning_rate": 5.000516462574515e-06, |
| "loss": 0.6463, |
| "num_tokens": 559204196.0, |
| "step": 1463 |
| }, |
| { |
| "epoch": 0.9993174061433447, |
| "grad_norm": 0.08935546875, |
| "learning_rate": 5.000229539409863e-06, |
| "loss": 0.6569, |
| "num_tokens": 559597755.0, |
| "step": 1464 |
| }, |
| { |
| "epoch": 1.0, |
| "grad_norm": 0.0869140625, |
| "learning_rate": 5.000057384925645e-06, |
| "loss": 0.6576, |
| "num_tokens": 560009809.0, |
| "step": 1465 |
| }, |
| { |
| "epoch": 1.0, |
| "step": 1465, |
| "total_flos": 1.0370150057822388e+19, |
| "train_loss": 0.0, |
| "train_runtime": 2.5915, |
| "train_samples_per_second": 36169.037, |
| "train_steps_per_second": 565.304 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 1465, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 1, |
| "save_steps": 100, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1.0370150057822388e+19, |
| "train_batch_size": 16, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
