| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.35522388059701493, | |
| "eval_steps": 500, | |
| "global_step": 119, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0029850746268656717, | |
| "grad_norm": 8.253287036233855, | |
| "learning_rate": 9.090909090909091e-07, | |
| "loss": 2.2247, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.005970149253731343, | |
| "grad_norm": 7.395570676912434, | |
| "learning_rate": 1.8181818181818183e-06, | |
| "loss": 2.1255, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.008955223880597015, | |
| "grad_norm": 8.228093047465732, | |
| "learning_rate": 2.7272727272727272e-06, | |
| "loss": 2.1028, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.011940298507462687, | |
| "grad_norm": 7.134217440402169, | |
| "learning_rate": 3.6363636363636366e-06, | |
| "loss": 2.0029, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.014925373134328358, | |
| "grad_norm": 5.937798020763942, | |
| "learning_rate": 4.5454545454545455e-06, | |
| "loss": 1.907, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01791044776119403, | |
| "grad_norm": 5.559897081671283, | |
| "learning_rate": 5.4545454545454545e-06, | |
| "loss": 1.859, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.020895522388059702, | |
| "grad_norm": 5.0306441137723485, | |
| "learning_rate": 6.363636363636364e-06, | |
| "loss": 1.8198, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.023880597014925373, | |
| "grad_norm": 3.514372505025641, | |
| "learning_rate": 7.272727272727273e-06, | |
| "loss": 1.5112, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.026865671641791045, | |
| "grad_norm": 3.1293116769998894, | |
| "learning_rate": 8.181818181818183e-06, | |
| "loss": 1.4412, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.029850746268656716, | |
| "grad_norm": 2.7656396768471456, | |
| "learning_rate": 9.090909090909091e-06, | |
| "loss": 1.4058, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03283582089552239, | |
| "grad_norm": 5.352646410247892, | |
| "learning_rate": 1e-05, | |
| "loss": 1.4371, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.03582089552238806, | |
| "grad_norm": 4.381930202952016, | |
| "learning_rate": 9.99976495753613e-06, | |
| "loss": 1.4176, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.03880597014925373, | |
| "grad_norm": 3.1514253661202765, | |
| "learning_rate": 9.999059852242508e-06, | |
| "loss": 1.2973, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.041791044776119404, | |
| "grad_norm": 2.6485142204402696, | |
| "learning_rate": 9.997884750411004e-06, | |
| "loss": 1.1784, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.04477611940298507, | |
| "grad_norm": 2.7946518315041007, | |
| "learning_rate": 9.996239762521152e-06, | |
| "loss": 1.3108, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.04776119402985075, | |
| "grad_norm": 2.6114018043960003, | |
| "learning_rate": 9.994125043229753e-06, | |
| "loss": 1.102, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.050746268656716415, | |
| "grad_norm": 2.3027638181918495, | |
| "learning_rate": 9.991540791356342e-06, | |
| "loss": 1.0699, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.05373134328358209, | |
| "grad_norm": 2.204185879725156, | |
| "learning_rate": 9.98848724986449e-06, | |
| "loss": 1.1473, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.056716417910447764, | |
| "grad_norm": 2.2090761147091817, | |
| "learning_rate": 9.98496470583896e-06, | |
| "loss": 1.1816, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.05970149253731343, | |
| "grad_norm": 2.0223479706782737, | |
| "learning_rate": 9.980973490458728e-06, | |
| "loss": 1.1217, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0626865671641791, | |
| "grad_norm": 2.080473927553376, | |
| "learning_rate": 9.976513978965829e-06, | |
| "loss": 1.0251, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.06567164179104477, | |
| "grad_norm": 2.3488846800450744, | |
| "learning_rate": 9.971586590630094e-06, | |
| "loss": 1.0279, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.06865671641791045, | |
| "grad_norm": 2.1532960090217212, | |
| "learning_rate": 9.966191788709716e-06, | |
| "loss": 1.0126, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.07164179104477612, | |
| "grad_norm": 2.055129377186085, | |
| "learning_rate": 9.960330080407712e-06, | |
| "loss": 1.0072, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.07462686567164178, | |
| "grad_norm": 2.156877426415811, | |
| "learning_rate": 9.954002016824226e-06, | |
| "loss": 1.0349, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07761194029850746, | |
| "grad_norm": 1.9914221696004062, | |
| "learning_rate": 9.947208192904722e-06, | |
| "loss": 1.0244, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.08059701492537313, | |
| "grad_norm": 1.9887204088063186, | |
| "learning_rate": 9.939949247384046e-06, | |
| "loss": 0.9912, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.08358208955223881, | |
| "grad_norm": 2.110799590941847, | |
| "learning_rate": 9.93222586272637e-06, | |
| "loss": 1.0377, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.08656716417910448, | |
| "grad_norm": 2.0583073599588473, | |
| "learning_rate": 9.924038765061042e-06, | |
| "loss": 1.0139, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.08955223880597014, | |
| "grad_norm": 2.1723065119138196, | |
| "learning_rate": 9.915388724114301e-06, | |
| "loss": 0.9376, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.09253731343283582, | |
| "grad_norm": 2.064005981064233, | |
| "learning_rate": 9.906276553136924e-06, | |
| "loss": 1.0226, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0955223880597015, | |
| "grad_norm": 2.122144912833922, | |
| "learning_rate": 9.896703108827758e-06, | |
| "loss": 0.9483, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.09850746268656717, | |
| "grad_norm": 2.1876975651489246, | |
| "learning_rate": 9.886669291253178e-06, | |
| "loss": 0.8942, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.10149253731343283, | |
| "grad_norm": 1.9606368757720636, | |
| "learning_rate": 9.876176043762467e-06, | |
| "loss": 0.885, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.1044776119402985, | |
| "grad_norm": 2.27926831852693, | |
| "learning_rate": 9.86522435289912e-06, | |
| "loss": 0.9491, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.10746268656716418, | |
| "grad_norm": 2.030550748336965, | |
| "learning_rate": 9.853815248308101e-06, | |
| "loss": 0.982, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.11044776119402985, | |
| "grad_norm": 2.0075365017841995, | |
| "learning_rate": 9.841949802639031e-06, | |
| "loss": 0.9843, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.11343283582089553, | |
| "grad_norm": 2.459727689278767, | |
| "learning_rate": 9.829629131445342e-06, | |
| "loss": 0.8796, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.11641791044776119, | |
| "grad_norm": 1.980360220017672, | |
| "learning_rate": 9.816854393079402e-06, | |
| "loss": 0.8793, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.11940298507462686, | |
| "grad_norm": 2.0803429090292433, | |
| "learning_rate": 9.803626788583603e-06, | |
| "loss": 0.8749, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.12238805970149254, | |
| "grad_norm": 2.0518013006773446, | |
| "learning_rate": 9.789947561577445e-06, | |
| "loss": 0.9123, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.1253731343283582, | |
| "grad_norm": 2.0868313477992695, | |
| "learning_rate": 9.775817998140615e-06, | |
| "loss": 0.884, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.12835820895522387, | |
| "grad_norm": 2.133757511141339, | |
| "learning_rate": 9.761239426692077e-06, | |
| "loss": 0.8846, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.13134328358208955, | |
| "grad_norm": 2.5042273900185332, | |
| "learning_rate": 9.74621321786517e-06, | |
| "loss": 0.9755, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.13432835820895522, | |
| "grad_norm": 1.9478417312433638, | |
| "learning_rate": 9.730740784378755e-06, | |
| "loss": 0.8857, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.1373134328358209, | |
| "grad_norm": 1.9676701142325428, | |
| "learning_rate": 9.71482358090438e-06, | |
| "loss": 0.8642, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.14029850746268657, | |
| "grad_norm": 1.920433325570768, | |
| "learning_rate": 9.698463103929542e-06, | |
| "loss": 0.8977, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.14328358208955225, | |
| "grad_norm": 1.9460499143315086, | |
| "learning_rate": 9.681660891616967e-06, | |
| "loss": 0.902, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.14626865671641792, | |
| "grad_norm": 1.968155074678708, | |
| "learning_rate": 9.664418523660004e-06, | |
| "loss": 0.9046, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.14925373134328357, | |
| "grad_norm": 1.9490820429685933, | |
| "learning_rate": 9.646737621134112e-06, | |
| "loss": 0.943, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.15223880597014924, | |
| "grad_norm": 2.08429080101051, | |
| "learning_rate": 9.628619846344453e-06, | |
| "loss": 0.9337, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.15522388059701492, | |
| "grad_norm": 1.930901797857114, | |
| "learning_rate": 9.610066902669593e-06, | |
| "loss": 0.8928, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.1582089552238806, | |
| "grad_norm": 1.9013217162350948, | |
| "learning_rate": 9.591080534401371e-06, | |
| "loss": 0.8905, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.16119402985074627, | |
| "grad_norm": 1.9982220077949842, | |
| "learning_rate": 9.571662526580898e-06, | |
| "loss": 0.8843, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.16417910447761194, | |
| "grad_norm": 1.9851369521318882, | |
| "learning_rate": 9.551814704830734e-06, | |
| "loss": 0.935, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.16716417910447762, | |
| "grad_norm": 2.1245353835937437, | |
| "learning_rate": 9.531538935183252e-06, | |
| "loss": 0.8666, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.1701492537313433, | |
| "grad_norm": 2.0723319083497245, | |
| "learning_rate": 9.51083712390519e-06, | |
| "loss": 0.9155, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.17313432835820897, | |
| "grad_norm": 1.7064471958774803, | |
| "learning_rate": 9.48971121731844e-06, | |
| "loss": 0.7976, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.1761194029850746, | |
| "grad_norm": 2.205553839755198, | |
| "learning_rate": 9.468163201617063e-06, | |
| "loss": 0.8575, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.1791044776119403, | |
| "grad_norm": 2.123341193411675, | |
| "learning_rate": 9.446195102680531e-06, | |
| "loss": 0.8779, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.18208955223880596, | |
| "grad_norm": 2.019140032242156, | |
| "learning_rate": 9.423808985883289e-06, | |
| "loss": 0.8087, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.18507462686567164, | |
| "grad_norm": 1.9693330173987202, | |
| "learning_rate": 9.401006955900555e-06, | |
| "loss": 0.8838, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.1880597014925373, | |
| "grad_norm": 1.7933631080982435, | |
| "learning_rate": 9.377791156510456e-06, | |
| "loss": 0.8881, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.191044776119403, | |
| "grad_norm": 2.064611639976522, | |
| "learning_rate": 9.35416377039246e-06, | |
| "loss": 0.9213, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.19402985074626866, | |
| "grad_norm": 2.108631144160596, | |
| "learning_rate": 9.330127018922195e-06, | |
| "loss": 0.8718, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.19701492537313434, | |
| "grad_norm": 2.0324645592765624, | |
| "learning_rate": 9.305683161962569e-06, | |
| "loss": 0.9008, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 2.433917861345567, | |
| "learning_rate": 9.280834497651334e-06, | |
| "loss": 0.8843, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.20298507462686566, | |
| "grad_norm": 2.1425405132153377, | |
| "learning_rate": 9.255583362184998e-06, | |
| "loss": 0.91, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.20597014925373133, | |
| "grad_norm": 2.1530253028764794, | |
| "learning_rate": 9.229932129599206e-06, | |
| "loss": 0.9011, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.208955223880597, | |
| "grad_norm": 1.883401707305491, | |
| "learning_rate": 9.203883211545517e-06, | |
| "loss": 0.8988, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.21194029850746268, | |
| "grad_norm": 1.9557201234597423, | |
| "learning_rate": 9.177439057064684e-06, | |
| "loss": 0.8674, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.21492537313432836, | |
| "grad_norm": 1.9754846823177397, | |
| "learning_rate": 9.150602152356394e-06, | |
| "loss": 0.8917, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.21791044776119403, | |
| "grad_norm": 2.1092131375976333, | |
| "learning_rate": 9.123375020545534e-06, | |
| "loss": 0.9798, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.2208955223880597, | |
| "grad_norm": 1.9481297510729838, | |
| "learning_rate": 9.09576022144496e-06, | |
| "loss": 0.8525, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.22388059701492538, | |
| "grad_norm": 2.1761431025018845, | |
| "learning_rate": 9.067760351314838e-06, | |
| "loss": 0.8726, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.22686567164179106, | |
| "grad_norm": 2.2204565282035276, | |
| "learning_rate": 9.039378042618556e-06, | |
| "loss": 1.0044, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.2298507462686567, | |
| "grad_norm": 1.9277579355394385, | |
| "learning_rate": 9.01061596377522e-06, | |
| "loss": 0.8209, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.23283582089552238, | |
| "grad_norm": 2.0131002774833076, | |
| "learning_rate": 8.981476818908778e-06, | |
| "loss": 0.9414, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.23582089552238805, | |
| "grad_norm": 1.8291261816939248, | |
| "learning_rate": 8.951963347593797e-06, | |
| "loss": 0.832, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.23880597014925373, | |
| "grad_norm": 2.174063739192292, | |
| "learning_rate": 8.92207832459788e-06, | |
| "loss": 0.9025, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.2417910447761194, | |
| "grad_norm": 2.1708714478771687, | |
| "learning_rate": 8.891824559620801e-06, | |
| "loss": 0.8268, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.24477611940298508, | |
| "grad_norm": 2.027241742869845, | |
| "learning_rate": 8.861204897030346e-06, | |
| "loss": 0.7233, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.24776119402985075, | |
| "grad_norm": 1.9258937377265744, | |
| "learning_rate": 8.83022221559489e-06, | |
| "loss": 0.7375, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.2507462686567164, | |
| "grad_norm": 1.9152686230971898, | |
| "learning_rate": 8.798879428212748e-06, | |
| "loss": 0.8811, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.2537313432835821, | |
| "grad_norm": 1.7807746386758827, | |
| "learning_rate": 8.767179481638303e-06, | |
| "loss": 0.8278, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.25671641791044775, | |
| "grad_norm": 1.8289188389465045, | |
| "learning_rate": 8.735125356204982e-06, | |
| "loss": 0.8347, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.25970149253731345, | |
| "grad_norm": 2.245431341266184, | |
| "learning_rate": 8.702720065545024e-06, | |
| "loss": 0.7732, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.2626865671641791, | |
| "grad_norm": 1.958611723882418, | |
| "learning_rate": 8.669966656306176e-06, | |
| "loss": 0.7941, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.2656716417910448, | |
| "grad_norm": 1.9764966388329304, | |
| "learning_rate": 8.636868207865244e-06, | |
| "loss": 0.8525, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.26865671641791045, | |
| "grad_norm": 2.1735209259103483, | |
| "learning_rate": 8.603427832038574e-06, | |
| "loss": 0.8746, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.2716417910447761, | |
| "grad_norm": 2.0038856668883813, | |
| "learning_rate": 8.569648672789496e-06, | |
| "loss": 0.8601, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.2746268656716418, | |
| "grad_norm": 2.055939688066048, | |
| "learning_rate": 8.535533905932739e-06, | |
| "loss": 0.85, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.27761194029850744, | |
| "grad_norm": 1.9848488069313877, | |
| "learning_rate": 8.501086738835843e-06, | |
| "loss": 0.8632, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.28059701492537314, | |
| "grad_norm": 2.113603493312776, | |
| "learning_rate": 8.466310410117622e-06, | |
| "loss": 0.8752, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.2835820895522388, | |
| "grad_norm": 2.1479673113866333, | |
| "learning_rate": 8.43120818934367e-06, | |
| "loss": 0.8588, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.2865671641791045, | |
| "grad_norm": 2.053881585055401, | |
| "learning_rate": 8.395783376718967e-06, | |
| "loss": 0.8122, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.28955223880597014, | |
| "grad_norm": 1.9411012651755073, | |
| "learning_rate": 8.360039302777614e-06, | |
| "loss": 0.8059, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.29253731343283584, | |
| "grad_norm": 1.8508350536791303, | |
| "learning_rate": 8.323979328069689e-06, | |
| "loss": 0.833, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.2955223880597015, | |
| "grad_norm": 1.8935253815844801, | |
| "learning_rate": 8.28760684284532e-06, | |
| "loss": 0.8401, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.29850746268656714, | |
| "grad_norm": 2.2781390311354577, | |
| "learning_rate": 8.25092526673592e-06, | |
| "loss": 0.909, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.30149253731343284, | |
| "grad_norm": 2.0985691350237876, | |
| "learning_rate": 8.213938048432697e-06, | |
| "loss": 0.8366, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.3044776119402985, | |
| "grad_norm": 2.0596354416831835, | |
| "learning_rate": 8.176648665362426e-06, | |
| "loss": 0.7292, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.3074626865671642, | |
| "grad_norm": 1.9898953284718084, | |
| "learning_rate": 8.139060623360494e-06, | |
| "loss": 0.8331, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.31044776119402984, | |
| "grad_norm": 2.1541238474940476, | |
| "learning_rate": 8.101177456341301e-06, | |
| "loss": 0.8119, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.31343283582089554, | |
| "grad_norm": 1.945555185400502, | |
| "learning_rate": 8.063002725966014e-06, | |
| "loss": 0.817, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.3164179104477612, | |
| "grad_norm": 1.7990902555417547, | |
| "learning_rate": 8.024540021307709e-06, | |
| "loss": 0.7763, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.3194029850746269, | |
| "grad_norm": 1.9977298309774547, | |
| "learning_rate": 7.985792958513932e-06, | |
| "loss": 0.857, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.32238805970149254, | |
| "grad_norm": 1.8852500490644626, | |
| "learning_rate": 7.946765180466725e-06, | |
| "loss": 0.7729, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.3253731343283582, | |
| "grad_norm": 2.076072697679506, | |
| "learning_rate": 7.907460356440133e-06, | |
| "loss": 0.8789, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.3283582089552239, | |
| "grad_norm": 1.9583549363047221, | |
| "learning_rate": 7.86788218175523e-06, | |
| "loss": 0.824, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.33134328358208953, | |
| "grad_norm": 1.8457221620577013, | |
| "learning_rate": 7.828034377432694e-06, | |
| "loss": 0.8298, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.33432835820895523, | |
| "grad_norm": 1.9836364032355813, | |
| "learning_rate": 7.787920689842965e-06, | |
| "loss": 0.8754, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.3373134328358209, | |
| "grad_norm": 1.9566209100212486, | |
| "learning_rate": 7.747544890354031e-06, | |
| "loss": 0.8364, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.3402985074626866, | |
| "grad_norm": 1.784255320047874, | |
| "learning_rate": 7.706910774976849e-06, | |
| "loss": 0.7379, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.34328358208955223, | |
| "grad_norm": 2.028920304841314, | |
| "learning_rate": 7.666022164008458e-06, | |
| "loss": 0.8175, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.34626865671641793, | |
| "grad_norm": 1.8633625964435614, | |
| "learning_rate": 7.624882901672801e-06, | |
| "loss": 0.8038, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.3492537313432836, | |
| "grad_norm": 2.0316374775972723, | |
| "learning_rate": 7.5834968557593155e-06, | |
| "loss": 0.8876, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.3522388059701492, | |
| "grad_norm": 1.9425263272714322, | |
| "learning_rate": 7.541867917259278e-06, | |
| "loss": 0.9399, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.35522388059701493, | |
| "grad_norm": 1.9570130825151233, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.7776, | |
| "step": 119 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 335, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 17, | |
| "total_flos": 9966949359616.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |