| { | |
| "best_metric": 10.30843448638916, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.09062075215224287, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00045310376076121433, | |
| "grad_norm": 0.08431881666183472, | |
| "learning_rate": 1e-05, | |
| "loss": 10.3752, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00045310376076121433, | |
| "eval_loss": 10.37207317352295, | |
| "eval_runtime": 8.5826, | |
| "eval_samples_per_second": 433.086, | |
| "eval_steps_per_second": 108.359, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0009062075215224287, | |
| "grad_norm": 0.07532705366611481, | |
| "learning_rate": 2e-05, | |
| "loss": 10.374, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.001359311282283643, | |
| "grad_norm": 0.08897168189287186, | |
| "learning_rate": 3e-05, | |
| "loss": 10.3721, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0018124150430448573, | |
| "grad_norm": 0.08756265044212341, | |
| "learning_rate": 4e-05, | |
| "loss": 10.3729, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0022655188038060714, | |
| "grad_norm": 0.08504622429609299, | |
| "learning_rate": 5e-05, | |
| "loss": 10.3698, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.002718622564567286, | |
| "grad_norm": 0.08428767323493958, | |
| "learning_rate": 6e-05, | |
| "loss": 10.3702, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0031717263253285004, | |
| "grad_norm": 0.09279818087816238, | |
| "learning_rate": 7e-05, | |
| "loss": 10.3744, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0036248300860897147, | |
| "grad_norm": 0.08182211965322495, | |
| "learning_rate": 8e-05, | |
| "loss": 10.3754, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0040779338468509285, | |
| "grad_norm": 0.08619166910648346, | |
| "learning_rate": 9e-05, | |
| "loss": 10.3703, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.004531037607612143, | |
| "grad_norm": 0.0850154384970665, | |
| "learning_rate": 0.0001, | |
| "loss": 10.3702, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.004984141368373357, | |
| "grad_norm": 0.09280817955732346, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 10.3708, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.005437245129134572, | |
| "grad_norm": 0.09724012762308121, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 10.367, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0058903488898957865, | |
| "grad_norm": 0.09114246070384979, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 10.3663, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.006343452650657001, | |
| "grad_norm": 0.10436821728944778, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 10.3722, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.006796556411418215, | |
| "grad_norm": 0.11564914882183075, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 10.3689, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.007249660172179429, | |
| "grad_norm": 0.11810274422168732, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 10.3647, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.007702763932940644, | |
| "grad_norm": 0.1179056242108345, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 10.366, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.008155867693701857, | |
| "grad_norm": 0.12273578345775604, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 10.3671, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.008608971454463071, | |
| "grad_norm": 0.12481386959552765, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 10.3636, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.009062075215224286, | |
| "grad_norm": 0.1384822130203247, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 10.3666, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0095151789759855, | |
| "grad_norm": 0.14004549384117126, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 10.3644, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.009968282736746714, | |
| "grad_norm": 0.15692494809627533, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 10.3632, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.01042138649750793, | |
| "grad_norm": 0.1507769227027893, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 10.3606, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.010874490258269144, | |
| "grad_norm": 0.15059030055999756, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 10.3591, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.011327594019030359, | |
| "grad_norm": 0.16990287601947784, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 10.3594, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.011780697779791573, | |
| "grad_norm": 0.2083251029253006, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 10.3592, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.012233801540552787, | |
| "grad_norm": 0.17906531691551208, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 10.3543, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.012686905301314002, | |
| "grad_norm": 0.19057442247867584, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 10.3545, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.013140009062075216, | |
| "grad_norm": 0.2047286331653595, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 10.3528, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.01359311282283643, | |
| "grad_norm": 0.17271780967712402, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 10.3537, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.014046216583597644, | |
| "grad_norm": 0.14873968064785004, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 10.3531, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.014499320344358859, | |
| "grad_norm": 0.16192154586315155, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 10.348, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.014952424105120073, | |
| "grad_norm": 0.16878123581409454, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 10.3497, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.015405527865881287, | |
| "grad_norm": 0.14054249227046967, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 10.3442, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0158586316266425, | |
| "grad_norm": 0.15284615755081177, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 10.3439, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.016311735387403714, | |
| "grad_norm": 0.15226201713085175, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 10.3469, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.01676483914816493, | |
| "grad_norm": 0.12173519283533096, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 10.3453, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.017217942908926143, | |
| "grad_norm": 0.10521087050437927, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 10.3455, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.017671046669687357, | |
| "grad_norm": 0.13628210127353668, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 10.3405, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.01812415043044857, | |
| "grad_norm": 0.11509867757558823, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 10.3447, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.018577254191209785, | |
| "grad_norm": 0.09881282597780228, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 10.3418, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.019030357951971, | |
| "grad_norm": 0.09307833015918732, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 10.343, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.019483461712732214, | |
| "grad_norm": 0.09937003999948502, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 10.3447, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.01993656547349343, | |
| "grad_norm": 0.07688029110431671, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 10.3377, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.020389669234254643, | |
| "grad_norm": 0.09292032569646835, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 10.3418, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.02084277299501586, | |
| "grad_norm": 0.0978633388876915, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 10.3383, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.021295876755777075, | |
| "grad_norm": 0.1081092581152916, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 10.3408, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.02174898051653829, | |
| "grad_norm": 0.07476416975259781, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 10.3369, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.022202084277299503, | |
| "grad_norm": 0.09883634746074677, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 10.3381, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.022655188038060717, | |
| "grad_norm": 0.09351437538862228, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 10.3379, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.022655188038060717, | |
| "eval_loss": 10.337263107299805, | |
| "eval_runtime": 8.6146, | |
| "eval_samples_per_second": 431.476, | |
| "eval_steps_per_second": 107.956, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.023108291798821932, | |
| "grad_norm": 0.07879135757684708, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 10.3383, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.023561395559583146, | |
| "grad_norm": 0.09073659032583237, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 10.3387, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.02401449932034436, | |
| "grad_norm": 0.0914374589920044, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 10.339, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.024467603081105575, | |
| "grad_norm": 0.0849669799208641, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 10.3369, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.02492070684186679, | |
| "grad_norm": 0.10950816422700882, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 10.3342, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.025373810602628003, | |
| "grad_norm": 0.08910530060529709, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 10.3356, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.025826914363389217, | |
| "grad_norm": 0.0773700401186943, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 10.3339, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.02628001812415043, | |
| "grad_norm": 0.08261220902204514, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 10.3366, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.026733121884911646, | |
| "grad_norm": 0.09103437513113022, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 10.3357, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.02718622564567286, | |
| "grad_norm": 0.08540646731853485, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 10.3342, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.027639329406434075, | |
| "grad_norm": 0.09951465576887131, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 10.335, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.02809243316719529, | |
| "grad_norm": 0.09674787521362305, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 10.3338, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.028545536927956503, | |
| "grad_norm": 0.0789952427148819, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 10.3327, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.028998640688717717, | |
| "grad_norm": 0.10217826813459396, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 10.3291, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.02945174444947893, | |
| "grad_norm": 0.09129117429256439, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 10.3307, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.029904848210240146, | |
| "grad_norm": 0.10249340534210205, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 10.3316, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.03035795197100136, | |
| "grad_norm": 0.08941692113876343, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 10.3352, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.030811055731762575, | |
| "grad_norm": 0.09472766518592834, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 10.3325, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.031264159492523785, | |
| "grad_norm": 0.11177093535661697, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 10.3258, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.031717263253285, | |
| "grad_norm": 0.09918520599603653, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 10.3262, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.032170367014046214, | |
| "grad_norm": 0.09618629515171051, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 10.3268, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.03262347077480743, | |
| "grad_norm": 0.09960244596004486, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 10.3307, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.03307657453556864, | |
| "grad_norm": 0.10685873031616211, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 10.3286, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.03352967829632986, | |
| "grad_norm": 0.09603196382522583, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 10.3275, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.03398278205709107, | |
| "grad_norm": 0.09453380107879639, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 10.3289, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.034435885817852285, | |
| "grad_norm": 0.11593970656394958, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 10.3266, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0348889895786135, | |
| "grad_norm": 0.09383190423250198, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 10.3261, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.035342093339374714, | |
| "grad_norm": 0.09254466742277145, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 10.3345, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.03579519710013593, | |
| "grad_norm": 0.11111216992139816, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 10.3257, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.03624830086089714, | |
| "grad_norm": 0.08759641647338867, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 10.3283, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.03670140462165836, | |
| "grad_norm": 0.09093410521745682, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 10.3242, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.03715450838241957, | |
| "grad_norm": 0.08180715143680573, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 10.3308, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.037607612143180785, | |
| "grad_norm": 0.07752928137779236, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 10.3243, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.038060715903942, | |
| "grad_norm": 0.0943979024887085, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 10.3275, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.038513819664703214, | |
| "grad_norm": 0.08871487528085709, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 10.3202, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.03896692342546443, | |
| "grad_norm": 0.08199674636125565, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 10.32, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.03942002718622564, | |
| "grad_norm": 0.07866522669792175, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 10.3252, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.03987313094698686, | |
| "grad_norm": 0.08581995218992233, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 10.3221, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.04032623470774807, | |
| "grad_norm": 0.09769885241985321, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 10.3204, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.040779338468509285, | |
| "grad_norm": 0.08246136456727982, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 10.3208, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0412324422292705, | |
| "grad_norm": 0.0928322970867157, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 10.3226, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.04168554599003172, | |
| "grad_norm": 0.09226018935441971, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 10.3217, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.042138649750792935, | |
| "grad_norm": 0.09736492484807968, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 10.322, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.04259175351155415, | |
| "grad_norm": 0.07728290557861328, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 10.3216, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.043044857272315364, | |
| "grad_norm": 0.09013911336660385, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 10.3246, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.04349796103307658, | |
| "grad_norm": 0.10085926204919815, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 10.3163, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.04395106479383779, | |
| "grad_norm": 0.09105587005615234, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 10.3195, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.044404168554599006, | |
| "grad_norm": 0.09174185991287231, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 10.3247, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.04485727231536022, | |
| "grad_norm": 0.09219998121261597, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 10.3223, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.045310376076121435, | |
| "grad_norm": 0.07694579660892487, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 10.3189, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.045310376076121435, | |
| "eval_loss": 10.3190279006958, | |
| "eval_runtime": 8.6177, | |
| "eval_samples_per_second": 431.32, | |
| "eval_steps_per_second": 107.917, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04576347983688265, | |
| "grad_norm": 0.1055775135755539, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 10.3193, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.046216583597643864, | |
| "grad_norm": 0.08020050078630447, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 10.3183, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.04666968735840508, | |
| "grad_norm": 0.08927895128726959, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 10.3206, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.04712279111916629, | |
| "grad_norm": 0.09862085431814194, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 10.318, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.047575894879927506, | |
| "grad_norm": 0.07221733033657074, | |
| "learning_rate": 5e-05, | |
| "loss": 10.3162, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.04802899864068872, | |
| "grad_norm": 0.08651195466518402, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 10.3202, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.048482102401449935, | |
| "grad_norm": 0.0794462263584137, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 10.3171, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.04893520616221115, | |
| "grad_norm": 0.07811540365219116, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 10.3215, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.049388309922972363, | |
| "grad_norm": 0.07207513600587845, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 10.314, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.04984141368373358, | |
| "grad_norm": 0.08736873418092728, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 10.319, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.05029451744449479, | |
| "grad_norm": 0.07394842058420181, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 10.321, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.050747621205256006, | |
| "grad_norm": 0.07760969549417496, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 10.3187, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.05120072496601722, | |
| "grad_norm": 0.07013378292322159, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 10.3126, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.051653828726778435, | |
| "grad_norm": 0.0694650188088417, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 10.318, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.05210693248753965, | |
| "grad_norm": 0.09646458923816681, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 10.3161, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.05256003624830086, | |
| "grad_norm": 0.0897815078496933, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 10.3154, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.05301314000906208, | |
| "grad_norm": 0.07170182466506958, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 10.3183, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.05346624376982329, | |
| "grad_norm": 0.0803888589143753, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 10.315, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.053919347530584506, | |
| "grad_norm": 0.09444740414619446, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 10.3113, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.05437245129134572, | |
| "grad_norm": 0.08745145052671432, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 10.3124, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.054825555052106935, | |
| "grad_norm": 0.07487913221120834, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 10.3141, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.05527865881286815, | |
| "grad_norm": 0.08770157396793365, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 10.3139, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.05573176257362936, | |
| "grad_norm": 0.08678803592920303, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 10.3179, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.05618486633439058, | |
| "grad_norm": 0.08807584643363953, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 10.3116, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.05663797009515179, | |
| "grad_norm": 0.09391863644123077, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 10.3145, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.057091073855913006, | |
| "grad_norm": 0.0772278681397438, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 10.3106, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.05754417761667422, | |
| "grad_norm": 0.08135335892438889, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 10.3175, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.057997281377435435, | |
| "grad_norm": 0.08163835853338242, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 10.3158, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.05845038513819665, | |
| "grad_norm": 0.0771145299077034, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 10.3126, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.05890348889895786, | |
| "grad_norm": 0.08987168222665787, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 10.3138, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.05935659265971908, | |
| "grad_norm": 0.06634359061717987, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 10.3159, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.05980969642048029, | |
| "grad_norm": 0.09013017266988754, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 10.3127, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.060262800181241506, | |
| "grad_norm": 0.07322061061859131, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 10.3118, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.06071590394200272, | |
| "grad_norm": 0.08726721256971359, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 10.3101, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.061169007702763935, | |
| "grad_norm": 0.0875997245311737, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 10.3114, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.06162211146352515, | |
| "grad_norm": 0.08566056191921234, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 10.3147, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.06207521522428636, | |
| "grad_norm": 0.09365395456552505, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 10.3165, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.06252831898504757, | |
| "grad_norm": 0.07719623297452927, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 10.3117, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.06298142274580878, | |
| "grad_norm": 0.09921330958604813, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 10.3118, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.06343452650657, | |
| "grad_norm": 0.091233991086483, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 10.3106, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.06388763026733121, | |
| "grad_norm": 0.08959236741065979, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 10.3147, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.06434073402809243, | |
| "grad_norm": 0.08307244628667831, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 10.3116, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.06479383778885364, | |
| "grad_norm": 0.10779538005590439, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 10.3043, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.06524694154961486, | |
| "grad_norm": 0.09790661931037903, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 10.3038, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.06570004531037607, | |
| "grad_norm": 0.09899360686540604, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 10.3114, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.06615314907113728, | |
| "grad_norm": 0.07933146506547928, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 10.3129, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.0666062528318985, | |
| "grad_norm": 0.08410552144050598, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 10.3105, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.06705935659265971, | |
| "grad_norm": 0.08882565051317215, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 10.3107, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.06751246035342093, | |
| "grad_norm": 0.10251981765031815, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 10.3087, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.06796556411418214, | |
| "grad_norm": 0.08644380420446396, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 10.3145, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.06796556411418214, | |
| "eval_loss": 10.309906959533691, | |
| "eval_runtime": 8.5961, | |
| "eval_samples_per_second": 432.403, | |
| "eval_steps_per_second": 108.188, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.06841866787494336, | |
| "grad_norm": 0.08348909765481949, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 10.3128, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.06887177163570457, | |
| "grad_norm": 0.09379616379737854, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 10.3159, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.06932487539646578, | |
| "grad_norm": 0.0836254358291626, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 10.3124, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.069777979157227, | |
| "grad_norm": 0.09745253622531891, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 10.3097, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.07023108291798821, | |
| "grad_norm": 0.08137994259595871, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 10.3102, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.07068418667874943, | |
| "grad_norm": 0.12498476356267929, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 10.3071, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.07113729043951064, | |
| "grad_norm": 0.1092490404844284, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 10.3114, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.07159039420027186, | |
| "grad_norm": 0.09169293940067291, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 10.309, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.07204349796103307, | |
| "grad_norm": 0.11189093440771103, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 10.3033, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.07249660172179428, | |
| "grad_norm": 0.08791095018386841, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 10.3112, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.0729497054825555, | |
| "grad_norm": 0.08543276786804199, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 10.31, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.07340280924331671, | |
| "grad_norm": 0.08431566506624222, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 10.3044, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.07385591300407793, | |
| "grad_norm": 0.07814370840787888, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 10.3143, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.07430901676483914, | |
| "grad_norm": 0.09191855788230896, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 10.3017, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.07476212052560036, | |
| "grad_norm": 0.09807488322257996, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 10.3141, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.07521522428636157, | |
| "grad_norm": 0.07995956391096115, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 10.314, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.07566832804712278, | |
| "grad_norm": 0.07811577618122101, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 10.3099, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.076121431807884, | |
| "grad_norm": 0.08100374788045883, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 10.3058, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.07657453556864521, | |
| "grad_norm": 0.07173501700162888, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 10.3133, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.07702763932940643, | |
| "grad_norm": 0.09132947772741318, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 10.3105, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.07748074309016764, | |
| "grad_norm": 0.07854383438825607, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 10.3081, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.07793384685092886, | |
| "grad_norm": 0.0775444358587265, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 10.3118, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.07838695061169007, | |
| "grad_norm": 0.08046716451644897, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 10.3063, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.07884005437245128, | |
| "grad_norm": 0.0875779464840889, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 10.316, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.0792931581332125, | |
| "grad_norm": 0.07684172689914703, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 10.3075, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.07974626189397371, | |
| "grad_norm": 0.08233340829610825, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 10.3063, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.08019936565473493, | |
| "grad_norm": 0.08705272525548935, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 10.3084, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.08065246941549614, | |
| "grad_norm": 0.08085699379444122, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 10.3118, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.08110557317625736, | |
| "grad_norm": 0.08175158500671387, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 10.3074, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.08155867693701857, | |
| "grad_norm": 0.0819338709115982, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 10.3111, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.08201178069777978, | |
| "grad_norm": 0.08131764829158783, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 10.3073, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.082464884458541, | |
| "grad_norm": 0.09145107865333557, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 10.3095, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.08291798821930221, | |
| "grad_norm": 0.10062360763549805, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 10.3068, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.08337109198006344, | |
| "grad_norm": 0.0973200649023056, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 10.3095, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.08382419574082466, | |
| "grad_norm": 0.08843878656625748, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 10.3099, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.08427729950158587, | |
| "grad_norm": 0.08757731318473816, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 10.3083, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.08473040326234708, | |
| "grad_norm": 0.07891438901424408, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 10.3117, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.0851835070231083, | |
| "grad_norm": 0.08320674300193787, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 10.3063, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.08563661078386951, | |
| "grad_norm": 0.09015924483537674, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 10.3128, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.08608971454463073, | |
| "grad_norm": 0.09279254823923111, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 10.3092, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.08654281830539194, | |
| "grad_norm": 0.09104158729314804, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 10.3095, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.08699592206615316, | |
| "grad_norm": 0.09526855498552322, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 10.3143, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.08744902582691437, | |
| "grad_norm": 0.08919867873191833, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 10.312, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.08790212958767558, | |
| "grad_norm": 0.08875775337219238, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 10.3029, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.0883552333484368, | |
| "grad_norm": 0.07775101065635681, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 10.3073, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.08880833710919801, | |
| "grad_norm": 0.07984912395477295, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 10.3121, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.08926144086995923, | |
| "grad_norm": 0.08644933253526688, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 10.3098, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.08971454463072044, | |
| "grad_norm": 0.07364819943904877, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 10.3072, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.09016764839148166, | |
| "grad_norm": 0.08544598519802094, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 10.3105, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.09062075215224287, | |
| "grad_norm": 0.10996384173631668, | |
| "learning_rate": 0.0, | |
| "loss": 10.3074, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.09062075215224287, | |
| "eval_loss": 10.30843448638916, | |
| "eval_runtime": 8.601, | |
| "eval_samples_per_second": 432.157, | |
| "eval_steps_per_second": 108.127, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 23035211415552.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |