| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 102, | |
| "global_step": 102, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00980392156862745, | |
| "grad_norm": 2.500279188156128, | |
| "learning_rate": 1e-05, | |
| "loss": 2.3876, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0196078431372549, | |
| "grad_norm": 2.360731840133667, | |
| "learning_rate": 9.901960784313727e-06, | |
| "loss": 2.3224, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.029411764705882353, | |
| "grad_norm": 2.111973285675049, | |
| "learning_rate": 9.803921568627451e-06, | |
| "loss": 2.3377, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0392156862745098, | |
| "grad_norm": 1.7793817520141602, | |
| "learning_rate": 9.705882352941177e-06, | |
| "loss": 2.2592, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.049019607843137254, | |
| "grad_norm": 1.3986390829086304, | |
| "learning_rate": 9.607843137254903e-06, | |
| "loss": 2.2263, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.058823529411764705, | |
| "grad_norm": 1.1015570163726807, | |
| "learning_rate": 9.509803921568628e-06, | |
| "loss": 2.2227, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.06862745098039216, | |
| "grad_norm": 0.9321516752243042, | |
| "learning_rate": 9.411764705882354e-06, | |
| "loss": 2.1582, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0784313725490196, | |
| "grad_norm": 0.8922587633132935, | |
| "learning_rate": 9.31372549019608e-06, | |
| "loss": 2.1078, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.08823529411764706, | |
| "grad_norm": 0.9146968126296997, | |
| "learning_rate": 9.215686274509804e-06, | |
| "loss": 2.0554, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.09803921568627451, | |
| "grad_norm": 0.9473930597305298, | |
| "learning_rate": 9.11764705882353e-06, | |
| "loss": 2.0022, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.10784313725490197, | |
| "grad_norm": 0.9462062120437622, | |
| "learning_rate": 9.019607843137256e-06, | |
| "loss": 2.0462, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.11764705882352941, | |
| "grad_norm": 0.959138810634613, | |
| "learning_rate": 8.921568627450982e-06, | |
| "loss": 2.0612, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.12745098039215685, | |
| "grad_norm": 0.9512083530426025, | |
| "learning_rate": 8.823529411764707e-06, | |
| "loss": 1.9063, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.13725490196078433, | |
| "grad_norm": 0.9136151671409607, | |
| "learning_rate": 8.725490196078433e-06, | |
| "loss": 2.0061, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.14705882352941177, | |
| "grad_norm": 0.8621863722801208, | |
| "learning_rate": 8.627450980392157e-06, | |
| "loss": 1.9088, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.1568627450980392, | |
| "grad_norm": 0.8003532290458679, | |
| "learning_rate": 8.529411764705883e-06, | |
| "loss": 1.8919, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.16666666666666666, | |
| "grad_norm": 0.7299250364303589, | |
| "learning_rate": 8.43137254901961e-06, | |
| "loss": 1.8571, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.17647058823529413, | |
| "grad_norm": 0.6769949197769165, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 1.8633, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.18627450980392157, | |
| "grad_norm": 0.6527869701385498, | |
| "learning_rate": 8.23529411764706e-06, | |
| "loss": 1.7669, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.19607843137254902, | |
| "grad_norm": 0.6449242234230042, | |
| "learning_rate": 8.137254901960784e-06, | |
| "loss": 1.7996, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.20588235294117646, | |
| "grad_norm": 0.6480177044868469, | |
| "learning_rate": 8.03921568627451e-06, | |
| "loss": 1.7738, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.21568627450980393, | |
| "grad_norm": 0.6246963739395142, | |
| "learning_rate": 7.941176470588236e-06, | |
| "loss": 1.7386, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.22549019607843138, | |
| "grad_norm": 0.6275233626365662, | |
| "learning_rate": 7.84313725490196e-06, | |
| "loss": 1.8029, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.23529411764705882, | |
| "grad_norm": 0.5825409889221191, | |
| "learning_rate": 7.745098039215687e-06, | |
| "loss": 1.6673, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.24509803921568626, | |
| "grad_norm": 0.5436587929725647, | |
| "learning_rate": 7.647058823529411e-06, | |
| "loss": 1.6992, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.2549019607843137, | |
| "grad_norm": 0.5287730097770691, | |
| "learning_rate": 7.549019607843138e-06, | |
| "loss": 1.6691, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.2647058823529412, | |
| "grad_norm": 0.5117806792259216, | |
| "learning_rate": 7.450980392156863e-06, | |
| "loss": 1.6736, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.27450980392156865, | |
| "grad_norm": 0.5041622519493103, | |
| "learning_rate": 7.352941176470589e-06, | |
| "loss": 1.6874, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.28431372549019607, | |
| "grad_norm": 0.47983038425445557, | |
| "learning_rate": 7.2549019607843145e-06, | |
| "loss": 1.6406, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.29411764705882354, | |
| "grad_norm": 0.4921911358833313, | |
| "learning_rate": 7.15686274509804e-06, | |
| "loss": 1.6127, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.30392156862745096, | |
| "grad_norm": 0.48269519209861755, | |
| "learning_rate": 7.058823529411766e-06, | |
| "loss": 1.6611, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.3137254901960784, | |
| "grad_norm": 0.5123285055160522, | |
| "learning_rate": 6.96078431372549e-06, | |
| "loss": 1.6019, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.3235294117647059, | |
| "grad_norm": 0.4910086691379547, | |
| "learning_rate": 6.862745098039216e-06, | |
| "loss": 1.5817, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 0.49109798669815063, | |
| "learning_rate": 6.764705882352942e-06, | |
| "loss": 1.5944, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.3431372549019608, | |
| "grad_norm": 0.4968704581260681, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 1.5942, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.35294117647058826, | |
| "grad_norm": 0.47293922305107117, | |
| "learning_rate": 6.568627450980393e-06, | |
| "loss": 1.564, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.3627450980392157, | |
| "grad_norm": 0.44330501556396484, | |
| "learning_rate": 6.470588235294119e-06, | |
| "loss": 1.5762, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.37254901960784315, | |
| "grad_norm": 0.4455674886703491, | |
| "learning_rate": 6.372549019607843e-06, | |
| "loss": 1.5843, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.38235294117647056, | |
| "grad_norm": 0.41862353682518005, | |
| "learning_rate": 6.274509803921569e-06, | |
| "loss": 1.5127, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.39215686274509803, | |
| "grad_norm": 0.4261449873447418, | |
| "learning_rate": 6.176470588235295e-06, | |
| "loss": 1.5167, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.4019607843137255, | |
| "grad_norm": 0.4266234040260315, | |
| "learning_rate": 6.07843137254902e-06, | |
| "loss": 1.5048, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.4117647058823529, | |
| "grad_norm": 0.432207852602005, | |
| "learning_rate": 5.980392156862746e-06, | |
| "loss": 1.4852, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.4215686274509804, | |
| "grad_norm": 0.40910154581069946, | |
| "learning_rate": 5.882352941176471e-06, | |
| "loss": 1.4988, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.43137254901960786, | |
| "grad_norm": 0.4706984758377075, | |
| "learning_rate": 5.784313725490197e-06, | |
| "loss": 1.5645, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.4411764705882353, | |
| "grad_norm": 0.40101850032806396, | |
| "learning_rate": 5.686274509803922e-06, | |
| "loss": 1.4994, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.45098039215686275, | |
| "grad_norm": 0.40438491106033325, | |
| "learning_rate": 5.588235294117647e-06, | |
| "loss": 1.5272, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.46078431372549017, | |
| "grad_norm": 0.39020276069641113, | |
| "learning_rate": 5.4901960784313735e-06, | |
| "loss": 1.4562, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.47058823529411764, | |
| "grad_norm": 0.37909573316574097, | |
| "learning_rate": 5.392156862745098e-06, | |
| "loss": 1.4873, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.4803921568627451, | |
| "grad_norm": 0.3828938603401184, | |
| "learning_rate": 5.294117647058824e-06, | |
| "loss": 1.4798, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.49019607843137253, | |
| "grad_norm": 0.367044597864151, | |
| "learning_rate": 5.19607843137255e-06, | |
| "loss": 1.4328, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.36673134565353394, | |
| "learning_rate": 5.098039215686274e-06, | |
| "loss": 1.4307, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.5098039215686274, | |
| "grad_norm": 0.3599296808242798, | |
| "learning_rate": 5e-06, | |
| "loss": 1.4069, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.5196078431372549, | |
| "grad_norm": 0.3603789806365967, | |
| "learning_rate": 4.901960784313726e-06, | |
| "loss": 1.4296, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.5294117647058824, | |
| "grad_norm": 0.3620091676712036, | |
| "learning_rate": 4.803921568627452e-06, | |
| "loss": 1.4063, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.5392156862745098, | |
| "grad_norm": 0.33629655838012695, | |
| "learning_rate": 4.705882352941177e-06, | |
| "loss": 1.421, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.5490196078431373, | |
| "grad_norm": 0.3413692116737366, | |
| "learning_rate": 4.607843137254902e-06, | |
| "loss": 1.4518, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.5588235294117647, | |
| "grad_norm": 0.3161819279193878, | |
| "learning_rate": 4.509803921568628e-06, | |
| "loss": 1.3914, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.5686274509803921, | |
| "grad_norm": 0.3277413547039032, | |
| "learning_rate": 4.411764705882353e-06, | |
| "loss": 1.389, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.5784313725490197, | |
| "grad_norm": 0.3128764033317566, | |
| "learning_rate": 4.313725490196079e-06, | |
| "loss": 1.395, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.5882352941176471, | |
| "grad_norm": 0.31412273645401, | |
| "learning_rate": 4.215686274509805e-06, | |
| "loss": 1.3735, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.5980392156862745, | |
| "grad_norm": 0.3194039762020111, | |
| "learning_rate": 4.11764705882353e-06, | |
| "loss": 1.4189, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.6078431372549019, | |
| "grad_norm": 0.3120681941509247, | |
| "learning_rate": 4.019607843137255e-06, | |
| "loss": 1.3911, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.6176470588235294, | |
| "grad_norm": 0.3160897493362427, | |
| "learning_rate": 3.92156862745098e-06, | |
| "loss": 1.3543, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.6274509803921569, | |
| "grad_norm": 0.31680962443351746, | |
| "learning_rate": 3.8235294117647055e-06, | |
| "loss": 1.3603, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.6372549019607843, | |
| "grad_norm": 0.31749823689460754, | |
| "learning_rate": 3.7254901960784316e-06, | |
| "loss": 1.4025, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.6470588235294118, | |
| "grad_norm": 0.3060447573661804, | |
| "learning_rate": 3.6274509803921573e-06, | |
| "loss": 1.3412, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.6568627450980392, | |
| "grad_norm": 0.3055213391780853, | |
| "learning_rate": 3.529411764705883e-06, | |
| "loss": 1.3913, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 0.3184523582458496, | |
| "learning_rate": 3.431372549019608e-06, | |
| "loss": 1.3994, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.6764705882352942, | |
| "grad_norm": 0.30899760127067566, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 1.3694, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.6862745098039216, | |
| "grad_norm": 0.29854196310043335, | |
| "learning_rate": 3.2352941176470594e-06, | |
| "loss": 1.3715, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.696078431372549, | |
| "grad_norm": 0.298645555973053, | |
| "learning_rate": 3.1372549019607846e-06, | |
| "loss": 1.3476, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.7058823529411765, | |
| "grad_norm": 0.2904506027698517, | |
| "learning_rate": 3.03921568627451e-06, | |
| "loss": 1.3476, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.7156862745098039, | |
| "grad_norm": 0.2939620912075043, | |
| "learning_rate": 2.9411764705882355e-06, | |
| "loss": 1.3666, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.7254901960784313, | |
| "grad_norm": 0.29882726073265076, | |
| "learning_rate": 2.843137254901961e-06, | |
| "loss": 1.347, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.7352941176470589, | |
| "grad_norm": 0.31818199157714844, | |
| "learning_rate": 2.7450980392156867e-06, | |
| "loss": 1.3821, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.7450980392156863, | |
| "grad_norm": 0.29951557517051697, | |
| "learning_rate": 2.647058823529412e-06, | |
| "loss": 1.3443, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.7549019607843137, | |
| "grad_norm": 0.2927483320236206, | |
| "learning_rate": 2.549019607843137e-06, | |
| "loss": 1.3258, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.7647058823529411, | |
| "grad_norm": 0.2869877219200134, | |
| "learning_rate": 2.450980392156863e-06, | |
| "loss": 1.3304, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.7745098039215687, | |
| "grad_norm": 0.29732242226600647, | |
| "learning_rate": 2.3529411764705885e-06, | |
| "loss": 1.3465, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.7843137254901961, | |
| "grad_norm": 0.30281081795692444, | |
| "learning_rate": 2.254901960784314e-06, | |
| "loss": 1.319, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.7941176470588235, | |
| "grad_norm": 0.2886880040168762, | |
| "learning_rate": 2.1568627450980393e-06, | |
| "loss": 1.3312, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.803921568627451, | |
| "grad_norm": 0.29117289185523987, | |
| "learning_rate": 2.058823529411765e-06, | |
| "loss": 1.3463, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.8137254901960784, | |
| "grad_norm": 0.28929680585861206, | |
| "learning_rate": 1.96078431372549e-06, | |
| "loss": 1.3469, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.8235294117647058, | |
| "grad_norm": 0.28210750222206116, | |
| "learning_rate": 1.8627450980392158e-06, | |
| "loss": 1.3329, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "grad_norm": 0.282003790140152, | |
| "learning_rate": 1.7647058823529414e-06, | |
| "loss": 1.2958, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.8431372549019608, | |
| "grad_norm": 0.3041072487831116, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 1.3422, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.8529411764705882, | |
| "grad_norm": 0.28907510638237, | |
| "learning_rate": 1.5686274509803923e-06, | |
| "loss": 1.355, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.8627450980392157, | |
| "grad_norm": 0.28159913420677185, | |
| "learning_rate": 1.4705882352941177e-06, | |
| "loss": 1.3366, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.8725490196078431, | |
| "grad_norm": 0.28043174743652344, | |
| "learning_rate": 1.3725490196078434e-06, | |
| "loss": 1.3384, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.8823529411764706, | |
| "grad_norm": 0.29932349920272827, | |
| "learning_rate": 1.2745098039215686e-06, | |
| "loss": 1.2866, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.8921568627450981, | |
| "grad_norm": 0.2869769036769867, | |
| "learning_rate": 1.1764705882352942e-06, | |
| "loss": 1.3329, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.9019607843137255, | |
| "grad_norm": 0.27412521839141846, | |
| "learning_rate": 1.0784313725490197e-06, | |
| "loss": 1.3273, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.9117647058823529, | |
| "grad_norm": 0.291384220123291, | |
| "learning_rate": 9.80392156862745e-07, | |
| "loss": 1.3566, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.9215686274509803, | |
| "grad_norm": 0.28981104493141174, | |
| "learning_rate": 8.823529411764707e-07, | |
| "loss": 1.3279, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.9313725490196079, | |
| "grad_norm": 0.28091126680374146, | |
| "learning_rate": 7.843137254901962e-07, | |
| "loss": 1.2947, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.9411764705882353, | |
| "grad_norm": 0.28839319944381714, | |
| "learning_rate": 6.862745098039217e-07, | |
| "loss": 1.3098, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.9509803921568627, | |
| "grad_norm": 0.27996671199798584, | |
| "learning_rate": 5.882352941176471e-07, | |
| "loss": 1.3001, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.9607843137254902, | |
| "grad_norm": 0.27735868096351624, | |
| "learning_rate": 4.901960784313725e-07, | |
| "loss": 1.3123, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.9705882352941176, | |
| "grad_norm": 0.2749045193195343, | |
| "learning_rate": 3.921568627450981e-07, | |
| "loss": 1.3244, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.9803921568627451, | |
| "grad_norm": 0.28808385133743286, | |
| "learning_rate": 2.9411764705882356e-07, | |
| "loss": 1.2847, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.9901960784313726, | |
| "grad_norm": 0.28120607137680054, | |
| "learning_rate": 1.9607843137254904e-07, | |
| "loss": 1.2952, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.2849506437778473, | |
| "learning_rate": 9.803921568627452e-08, | |
| "loss": 1.3353, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 1.329064965248108, | |
| "eval_runtime": 2.3109, | |
| "eval_samples_per_second": 6.491, | |
| "eval_steps_per_second": 0.865, | |
| "step": 102 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 102, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 0, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.1010882465169408e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |