{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.05011286681715576, "eval_steps": 500, "global_step": 222, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.1428571428571427e-07, "loss": 0.0007, "step": 1 }, { "epoch": 0.0, "learning_rate": 2.2857142857142855e-07, "loss": -0.0016, "step": 2 }, { "epoch": 0.0, "learning_rate": 3.428571428571428e-07, "loss": -0.0009, "step": 3 }, { "epoch": 0.0, "learning_rate": 4.571428571428571e-07, "loss": -0.0002, "step": 4 }, { "epoch": 0.0, "learning_rate": 5.714285714285714e-07, "loss": 0.0005, "step": 5 }, { "epoch": 0.0, "learning_rate": 6.857142857142856e-07, "loss": -0.0004, "step": 6 }, { "epoch": 0.0, "learning_rate": 8e-07, "loss": -0.0013, "step": 7 }, { "epoch": 0.0, "learning_rate": 7.999572983069959e-07, "loss": -0.0003, "step": 8 }, { "epoch": 0.0, "learning_rate": 7.99829202345157e-07, "loss": -0.0009, "step": 9 }, { "epoch": 0.0, "learning_rate": 7.996157394640553e-07, "loss": -0.0003, "step": 10 }, { "epoch": 0.0, "learning_rate": 7.993169552398228e-07, "loss": -0.0001, "step": 11 }, { "epoch": 0.0, "learning_rate": 7.989329134654206e-07, "loss": -0.0005, "step": 12 }, { "epoch": 0.0, "learning_rate": 7.984636961370187e-07, "loss": -0.0007, "step": 13 }, { "epoch": 0.0, "learning_rate": 7.979094034364882e-07, "loss": 0.0007, "step": 14 }, { "epoch": 0.0, "learning_rate": 7.972701537100132e-07, "loss": -0.0001, "step": 15 }, { "epoch": 0.0, "learning_rate": 7.965460834428213e-07, "loss": -0.0011, "step": 16 }, { "epoch": 0.0, "learning_rate": 7.957373472300441e-07, "loss": -0.0019, "step": 17 }, { "epoch": 0.0, "learning_rate": 7.948441177437086e-07, "loss": -0.0009, "step": 18 }, { "epoch": 0.0, "learning_rate": 7.938665856958715e-07, "loss": 0.0006, "step": 19 }, { "epoch": 0.0, "learning_rate": 7.928049597979001e-07, "loss": -0.0015, "step": 20 }, { "epoch": 0.0, "learning_rate": 7.916594667159099e-07, "loss": -0.0008, "step": 21 }, { "epoch": 0.0, "learning_rate": 7.904303510223708e-07, "loss": -0.0003, "step": 22 }, { "epoch": 0.01, "learning_rate": 7.891178751438877e-07, "loss": 0.0001, "step": 23 }, { "epoch": 0.01, "learning_rate": 7.87722319305171e-07, "loss": -0.0004, "step": 24 }, { "epoch": 0.01, "learning_rate": 7.862439814692054e-07, "loss": -0.0007, "step": 25 }, { "epoch": 0.01, "learning_rate": 7.846831772736331e-07, "loss": -0.0009, "step": 26 }, { "epoch": 0.01, "learning_rate": 7.830402399633623e-07, "loss": -0.0017, "step": 27 }, { "epoch": 0.01, "learning_rate": 7.813155203194161e-07, "loss": -0.0016, "step": 28 }, { "epoch": 0.01, "learning_rate": 7.795093865840382e-07, "loss": -0.0009, "step": 29 }, { "epoch": 0.01, "learning_rate": 7.776222243820702e-07, "loss": -0.001, "step": 30 }, { "epoch": 0.01, "learning_rate": 7.75654436638617e-07, "loss": -0.0018, "step": 31 }, { "epoch": 0.01, "learning_rate": 7.736064434930192e-07, "loss": -0.0006, "step": 32 }, { "epoch": 0.01, "learning_rate": 7.714786822091497e-07, "loss": -0.0024, "step": 33 }, { "epoch": 0.01, "learning_rate": 7.692716070820541e-07, "loss": -0.0011, "step": 34 }, { "epoch": 0.01, "learning_rate": 7.66985689340955e-07, "loss": -0.0024, "step": 35 }, { "epoch": 0.01, "learning_rate": 7.646214170486406e-07, "loss": -0.0011, "step": 36 }, { "epoch": 0.01, "learning_rate": 7.621792949972586e-07, "loss": -0.0007, "step": 37 }, { "epoch": 0.01, "learning_rate": 7.596598446005399e-07, "loss": -0.0017, "step": 38 }, { "epoch": 0.01, "learning_rate": 7.570636037824713e-07, "loss": 0.001, "step": 39 }, { "epoch": 0.01, "learning_rate": 7.543911268624445e-07, "loss": -0.001, "step": 40 }, { "epoch": 0.01, "learning_rate": 7.516429844369049e-07, "loss": -0.0003, "step": 41 }, { "epoch": 0.01, "learning_rate": 7.48819763257523e-07, "loss": -0.0002, "step": 42 }, { "epoch": 0.01, "learning_rate": 7.459220661059195e-07, "loss": -0.0007, "step": 43 }, { "epoch": 0.01, "learning_rate": 7.429505116649652e-07, "loss": 0.0002, "step": 44 }, { "epoch": 0.01, "learning_rate": 7.399057343866876e-07, "loss": -0.0021, "step": 45 }, { "epoch": 0.01, "learning_rate": 7.367883843568095e-07, "loss": -0.0019, "step": 46 }, { "epoch": 0.01, "learning_rate": 7.335991271559511e-07, "loss": 0.0003, "step": 47 }, { "epoch": 0.01, "learning_rate": 7.303386437175216e-07, "loss": -0.0016, "step": 48 }, { "epoch": 0.01, "learning_rate": 7.270076301823353e-07, "loss": -0.0018, "step": 49 }, { "epoch": 0.01, "learning_rate": 7.23606797749979e-07, "loss": -0.0007, "step": 50 }, { "epoch": 0.01, "learning_rate": 7.201368725269651e-07, "loss": -0.0004, "step": 51 }, { "epoch": 0.01, "learning_rate": 7.165985953717016e-07, "loss": -0.0012, "step": 52 }, { "epoch": 0.01, "learning_rate": 7.12992721736313e-07, "loss": -0.0005, "step": 53 }, { "epoch": 0.01, "learning_rate": 7.093200215053441e-07, "loss": -0.0008, "step": 54 }, { "epoch": 0.01, "learning_rate": 7.055812788313836e-07, "loss": -0.0008, "step": 55 }, { "epoch": 0.01, "learning_rate": 7.01777291967641e-07, "loss": 0.0004, "step": 56 }, { "epoch": 0.01, "learning_rate": 6.979088730975127e-07, "loss": -0.0011, "step": 57 }, { "epoch": 0.01, "learning_rate": 6.939768481611736e-07, "loss": -0.0003, "step": 58 }, { "epoch": 0.01, "learning_rate": 6.899820566792321e-07, "loss": -0.0006, "step": 59 }, { "epoch": 0.01, "learning_rate": 6.859253515734858e-07, "loss": -0.001, "step": 60 }, { "epoch": 0.01, "learning_rate": 6.818075989848146e-07, "loss": -0.0008, "step": 61 }, { "epoch": 0.01, "learning_rate": 6.776296780882536e-07, "loss": -0.0009, "step": 62 }, { "epoch": 0.01, "learning_rate": 6.733924809052798e-07, "loss": -0.0007, "step": 63 }, { "epoch": 0.01, "learning_rate": 6.6909691211336e-07, "loss": -0.0007, "step": 64 }, { "epoch": 0.01, "learning_rate": 6.647438888527933e-07, "loss": -0.0011, "step": 65 }, { "epoch": 0.01, "learning_rate": 6.603343405308942e-07, "loss": -0.0007, "step": 66 }, { "epoch": 0.02, "learning_rate": 6.558692086235565e-07, "loss": -0.0011, "step": 67 }, { "epoch": 0.02, "learning_rate": 6.513494464742397e-07, "loss": -0.0001, "step": 68 }, { "epoch": 0.02, "learning_rate": 6.467760190904226e-07, "loss": -0.0017, "step": 69 }, { "epoch": 0.02, "learning_rate": 6.421499029375658e-07, "loss": -0.0009, "step": 70 }, { "epoch": 0.02, "learning_rate": 6.374720857306283e-07, "loss": 0.0003, "step": 71 }, { "epoch": 0.02, "learning_rate": 6.327435662231812e-07, "loss": -0.0018, "step": 72 }, { "epoch": 0.02, "learning_rate": 6.279653539941665e-07, "loss": -0.0006, "step": 73 }, { "epoch": 0.02, "learning_rate": 6.231384692323428e-07, "loss": -0.0002, "step": 74 }, { "epoch": 0.02, "learning_rate": 6.182639425184666e-07, "loss": -0.0009, "step": 75 }, { "epoch": 0.02, "learning_rate": 6.133428146052541e-07, "loss": -0.0007, "step": 76 }, { "epoch": 0.02, "learning_rate": 6.083761361951721e-07, "loss": 0.0001, "step": 77 }, { "epoch": 0.02, "learning_rate": 6.033649677161044e-07, "loss": -0.0009, "step": 78 }, { "epoch": 0.02, "learning_rate": 5.983103790949407e-07, "loss": -0.0, "step": 79 }, { "epoch": 0.02, "learning_rate": 5.932134495291391e-07, "loss": 0.0005, "step": 80 }, { "epoch": 0.02, "learning_rate": 5.880752672563073e-07, "loss": -0.0008, "step": 81 }, { "epoch": 0.02, "learning_rate": 5.828969293218554e-07, "loss": -0.0006, "step": 82 }, { "epoch": 0.02, "learning_rate": 5.776795413447672e-07, "loss": -0.0008, "step": 83 }, { "epoch": 0.02, "learning_rate": 5.72424217281541e-07, "loss": -0.0005, "step": 84 }, { "epoch": 0.02, "learning_rate": 5.671320791883507e-07, "loss": -0.0013, "step": 85 }, { "epoch": 0.02, "learning_rate": 5.618042569814775e-07, "loss": -0.0006, "step": 86 }, { "epoch": 0.02, "learning_rate": 5.564418881960623e-07, "loss": -0.0007, "step": 87 }, { "epoch": 0.02, "learning_rate": 5.510461177432338e-07, "loss": -0.0006, "step": 88 }, { "epoch": 0.02, "learning_rate": 5.456180976656585e-07, "loss": -0.0007, "step": 89 }, { "epoch": 0.02, "learning_rate": 5.401589868915718e-07, "loss": 0.0001, "step": 90 }, { "epoch": 0.02, "learning_rate": 5.346699509873349e-07, "loss": -0.0007, "step": 91 }, { "epoch": 0.02, "learning_rate": 5.291521619085784e-07, "loss": -0.0007, "step": 92 }, { "epoch": 0.02, "learning_rate": 5.23606797749979e-07, "loss": -0.0002, "step": 93 }, { "epoch": 0.02, "learning_rate": 5.180350424937258e-07, "loss": 0.0002, "step": 94 }, { "epoch": 0.02, "learning_rate": 5.124380857567313e-07, "loss": -0.0006, "step": 95 }, { "epoch": 0.02, "learning_rate": 5.068171225366372e-07, "loss": -0.0015, "step": 96 }, { "epoch": 0.02, "learning_rate": 5.011733529566722e-07, "loss": -0.0004, "step": 97 }, { "epoch": 0.02, "learning_rate": 4.955079820094167e-07, "loss": -0.001, "step": 98 }, { "epoch": 0.02, "learning_rate": 4.898222192995253e-07, "loss": 0.0002, "step": 99 }, { "epoch": 0.02, "learning_rate": 4.841172787854665e-07, "loss": -0.0012, "step": 100 }, { "epoch": 0.02, "learning_rate": 4.783943785203326e-07, "loss": -0.0026, "step": 101 }, { "epoch": 0.02, "learning_rate": 4.726547403917745e-07, "loss": -0.0004, "step": 102 }, { "epoch": 0.02, "learning_rate": 4.6689958986111913e-07, "loss": -0.0005, "step": 103 }, { "epoch": 0.02, "learning_rate": 4.6113015570172204e-07, "loss": 0.0, "step": 104 }, { "epoch": 0.02, "learning_rate": 4.553476697366145e-07, "loss": -0.0001, "step": 105 }, { "epoch": 0.02, "learning_rate": 4.4955336657549915e-07, "loss": -0.0026, "step": 106 }, { "epoch": 0.02, "learning_rate": 4.4374848335114983e-07, "loss": 0.0018, "step": 107 }, { "epoch": 0.02, "learning_rate": 4.3793425945527306e-07, "loss": 0.0016, "step": 108 }, { "epoch": 0.02, "learning_rate": 4.321119362738884e-07, "loss": -0.0006, "step": 109 }, { "epoch": 0.02, "learning_rate": 4.2628275692228115e-07, "loss": -0.0004, "step": 110 }, { "epoch": 0.03, "learning_rate": 4.2044796597958684e-07, "loss": -0.0009, "step": 111 }, { "epoch": 0.03, "learning_rate": 4.146088092230636e-07, "loss": -0.0007, "step": 112 }, { "epoch": 0.03, "learning_rate": 4.087665333621072e-07, "loss": -0.0003, "step": 113 }, { "epoch": 0.03, "learning_rate": 4.029223857720692e-07, "loss": -0.0003, "step": 114 }, { "epoch": 0.03, "learning_rate": 3.970776142279308e-07, "loss": 0.0, "step": 115 }, { "epoch": 0.03, "learning_rate": 3.9123346663789275e-07, "loss": -0.0006, "step": 116 }, { "epoch": 0.03, "learning_rate": 3.853911907769364e-07, "loss": -0.0015, "step": 117 }, { "epoch": 0.03, "learning_rate": 3.795520340204132e-07, "loss": -0.0006, "step": 118 }, { "epoch": 0.03, "learning_rate": 3.7371724307771886e-07, "loss": -0.0001, "step": 119 }, { "epoch": 0.03, "learning_rate": 3.6788806372611154e-07, "loss": 0.0003, "step": 120 }, { "epoch": 0.03, "learning_rate": 3.6206574054472696e-07, "loss": -0.0003, "step": 121 }, { "epoch": 0.03, "learning_rate": 3.562515166488503e-07, "loss": -0.0002, "step": 122 }, { "epoch": 0.03, "learning_rate": 3.5044663342450076e-07, "loss": -0.0007, "step": 123 }, { "epoch": 0.03, "learning_rate": 3.446523302633854e-07, "loss": -0.0015, "step": 124 }, { "epoch": 0.03, "learning_rate": 3.3886984429827797e-07, "loss": -0.0004, "step": 125 }, { "epoch": 0.03, "learning_rate": 3.331004101388809e-07, "loss": -0.0008, "step": 126 }, { "epoch": 0.03, "learning_rate": 3.2734525960822544e-07, "loss": -0.0003, "step": 127 }, { "epoch": 0.03, "learning_rate": 3.2160562147966743e-07, "loss": -0.001, "step": 128 }, { "epoch": 0.03, "learning_rate": 3.158827212145336e-07, "loss": -0.0001, "step": 129 }, { "epoch": 0.03, "learning_rate": 3.1017778070047474e-07, "loss": -0.0016, "step": 130 }, { "epoch": 0.03, "learning_rate": 3.044920179905832e-07, "loss": -0.0015, "step": 131 }, { "epoch": 0.03, "learning_rate": 2.9882664704332764e-07, "loss": -0.0005, "step": 132 }, { "epoch": 0.03, "learning_rate": 2.9318287746336285e-07, "loss": -0.0003, "step": 133 }, { "epoch": 0.03, "learning_rate": 2.8756191424326873e-07, "loss": -0.0009, "step": 134 }, { "epoch": 0.03, "learning_rate": 2.819649575062742e-07, "loss": -0.0007, "step": 135 }, { "epoch": 0.03, "learning_rate": 2.7639320225002107e-07, "loss": -0.001, "step": 136 }, { "epoch": 0.03, "learning_rate": 2.708478380914216e-07, "loss": 0.0002, "step": 137 }, { "epoch": 0.03, "learning_rate": 2.653300490126652e-07, "loss": -0.0004, "step": 138 }, { "epoch": 0.03, "learning_rate": 2.598410131084282e-07, "loss": 0.0005, "step": 139 }, { "epoch": 0.03, "learning_rate": 2.543819023343413e-07, "loss": -0.0003, "step": 140 }, { "epoch": 0.03, "learning_rate": 2.4895388225676614e-07, "loss": -0.0001, "step": 141 }, { "epoch": 0.03, "learning_rate": 2.4355811180393763e-07, "loss": -0.0005, "step": 142 }, { "epoch": 0.03, "learning_rate": 2.3819574301852254e-07, "loss": -0.0001, "step": 143 }, { "epoch": 0.03, "learning_rate": 2.3286792081164936e-07, "loss": -0.0015, "step": 144 }, { "epoch": 0.03, "learning_rate": 2.2757578271845907e-07, "loss": -0.0007, "step": 145 }, { "epoch": 0.03, "learning_rate": 2.223204586552329e-07, "loss": -0.0004, "step": 146 }, { "epoch": 0.03, "learning_rate": 2.1710307067814458e-07, "loss": -0.0013, "step": 147 }, { "epoch": 0.03, "learning_rate": 2.119247327436926e-07, "loss": 0.0006, "step": 148 }, { "epoch": 0.03, "learning_rate": 2.0678655047086084e-07, "loss": -0.0008, "step": 149 }, { "epoch": 0.03, "learning_rate": 2.0168962090505928e-07, "loss": -0.0007, "step": 150 }, { "epoch": 0.03, "learning_rate": 1.9663503228389558e-07, "loss": -0.0016, "step": 151 }, { "epoch": 0.03, "learning_rate": 1.9162386380482794e-07, "loss": 0.0003, "step": 152 }, { "epoch": 0.03, "learning_rate": 1.8665718539474603e-07, "loss": -0.0011, "step": 153 }, { "epoch": 0.03, "learning_rate": 1.817360574815334e-07, "loss": -0.0008, "step": 154 }, { "epoch": 0.03, "learning_rate": 1.7686153076765704e-07, "loss": -0.0022, "step": 155 }, { "epoch": 0.04, "learning_rate": 1.7203464600583347e-07, "loss": -0.0006, "step": 156 }, { "epoch": 0.04, "learning_rate": 1.6725643377681893e-07, "loss": -0.0, "step": 157 }, { "epoch": 0.04, "learning_rate": 1.6252791426937177e-07, "loss": 0.0004, "step": 158 }, { "epoch": 0.04, "learning_rate": 1.5785009706243417e-07, "loss": -0.001, "step": 159 }, { "epoch": 0.04, "learning_rate": 1.5322398090957744e-07, "loss": -0.0002, "step": 160 }, { "epoch": 0.04, "learning_rate": 1.4865055352576028e-07, "loss": -0.0008, "step": 161 }, { "epoch": 0.04, "learning_rate": 1.4413079137644356e-07, "loss": -0.0012, "step": 162 }, { "epoch": 0.04, "learning_rate": 1.3966565946910578e-07, "loss": -0.0006, "step": 163 }, { "epoch": 0.04, "learning_rate": 1.3525611114720669e-07, "loss": 0.0003, "step": 164 }, { "epoch": 0.04, "learning_rate": 1.3090308788663999e-07, "loss": -0.0004, "step": 165 }, { "epoch": 0.04, "learning_rate": 1.2660751909472027e-07, "loss": -0.0009, "step": 166 }, { "epoch": 0.04, "learning_rate": 1.223703219117464e-07, "loss": -0.0004, "step": 167 }, { "epoch": 0.04, "learning_rate": 1.181924010151853e-07, "loss": -0.0004, "step": 168 }, { "epoch": 0.04, "learning_rate": 1.1407464842651431e-07, "loss": -0.0009, "step": 169 }, { "epoch": 0.04, "learning_rate": 1.1001794332076793e-07, "loss": -0.0012, "step": 170 }, { "epoch": 0.04, "learning_rate": 1.0602315183882642e-07, "loss": -0.0002, "step": 171 }, { "epoch": 0.04, "learning_rate": 1.0209112690248725e-07, "loss": -0.0001, "step": 172 }, { "epoch": 0.04, "learning_rate": 9.822270803235895e-08, "loss": -0.0013, "step": 173 }, { "epoch": 0.04, "learning_rate": 9.441872116861636e-08, "loss": 0.0004, "step": 174 }, { "epoch": 0.04, "learning_rate": 9.0679978494656e-08, "loss": -0.0024, "step": 175 }, { "epoch": 0.04, "learning_rate": 8.700727826368708e-08, "loss": -0.0006, "step": 176 }, { "epoch": 0.04, "learning_rate": 8.340140462829839e-08, "loss": -0.0001, "step": 177 }, { "epoch": 0.04, "learning_rate": 7.986312747303503e-08, "loss": -0.002, "step": 178 }, { "epoch": 0.04, "learning_rate": 7.639320225002106e-08, "loss": -0.0017, "step": 179 }, { "epoch": 0.04, "learning_rate": 7.299236981766466e-08, "loss": 0.0003, "step": 180 }, { "epoch": 0.04, "learning_rate": 6.966135628247834e-08, "loss": -0.0009, "step": 181 }, { "epoch": 0.04, "learning_rate": 6.640087284404888e-08, "loss": -0.0006, "step": 182 }, { "epoch": 0.04, "learning_rate": 6.321161564319033e-08, "loss": -0.0012, "step": 183 }, { "epoch": 0.04, "learning_rate": 6.009426561331241e-08, "loss": -0.0002, "step": 184 }, { "epoch": 0.04, "learning_rate": 5.704948833503484e-08, "loss": -0.0014, "step": 185 }, { "epoch": 0.04, "learning_rate": 5.4077933894080484e-08, "loss": -0.0002, "step": 186 }, { "epoch": 0.04, "learning_rate": 5.118023674247691e-08, "loss": -0.0007, "step": 187 }, { "epoch": 0.04, "learning_rate": 4.835701556309515e-08, "loss": -0.0012, "step": 188 }, { "epoch": 0.04, "learning_rate": 4.560887313755537e-08, "loss": -0.0002, "step": 189 }, { "epoch": 0.04, "learning_rate": 4.293639621752874e-08, "loss": -0.0, "step": 190 }, { "epoch": 0.04, "learning_rate": 4.03401553994601e-08, "loss": -0.0009, "step": 191 }, { "epoch": 0.04, "learning_rate": 3.782070500274135e-08, "loss": -0.0004, "step": 192 }, { "epoch": 0.04, "learning_rate": 3.5378582951359495e-08, "loss": -0.0005, "step": 193 }, { "epoch": 0.04, "learning_rate": 3.301431065904503e-08, "loss": -0.0004, "step": 194 }, { "epoch": 0.04, "learning_rate": 3.072839291794591e-08, "loss": -0.0009, "step": 195 }, { "epoch": 0.04, "learning_rate": 2.852131779085032e-08, "loss": 0.0007, "step": 196 }, { "epoch": 0.04, "learning_rate": 2.639355650698083e-08, "loss": -0.0003, "step": 197 }, { "epoch": 0.04, "learning_rate": 2.4345563361383025e-08, "loss": -0.0011, "step": 198 }, { "epoch": 0.04, "learning_rate": 2.237777561792975e-08, "loss": -0.0012, "step": 199 }, { "epoch": 0.05, "learning_rate": 2.0490613415961744e-08, "loss": 0.0001, "step": 200 }, { "epoch": 0.05, "learning_rate": 1.868447968058393e-08, "loss": -0.001, "step": 201 }, { "epoch": 0.05, "learning_rate": 1.695976003663766e-08, "loss": -0.0016, "step": 202 }, { "epoch": 0.05, "learning_rate": 1.5316822726366785e-08, "loss": -0.0015, "step": 203 }, { "epoch": 0.05, "learning_rate": 1.37560185307946e-08, "loss": -0.0008, "step": 204 }, { "epoch": 0.05, "learning_rate": 1.2277680694828996e-08, "loss": -0.0014, "step": 205 }, { "epoch": 0.05, "learning_rate": 1.0882124856112218e-08, "loss": -0.0007, "step": 206 }, { "epoch": 0.05, "learning_rate": 9.569648977629175e-09, "loss": 0.0004, "step": 207 }, { "epoch": 0.05, "learning_rate": 8.340533284090012e-09, "loss": -0.0015, "step": 208 }, { "epoch": 0.05, "learning_rate": 7.1950402020998895e-09, "loss": -0.002, "step": 209 }, { "epoch": 0.05, "learning_rate": 6.133414304128371e-09, "loss": 0.0001, "step": 210 }, { "epoch": 0.05, "learning_rate": 5.1558822562913774e-09, "loss": -0.0, "step": 211 }, { "epoch": 0.05, "learning_rate": 4.2626527699558985e-09, "loss": -0.0007, "step": 212 }, { "epoch": 0.05, "learning_rate": 3.4539165571785713e-09, "loss": -0.0002, "step": 213 }, { "epoch": 0.05, "learning_rate": 2.729846289986737e-09, "loss": 0.0, "step": 214 }, { "epoch": 0.05, "learning_rate": 2.0905965635117152e-09, "loss": -0.0003, "step": 215 }, { "epoch": 0.05, "learning_rate": 1.5363038629813808e-09, "loss": -0.0012, "step": 216 }, { "epoch": 0.05, "learning_rate": 1.0670865345793423e-09, "loss": -0.0019, "step": 217 }, { "epoch": 0.05, "learning_rate": 6.830447601771983e-10, "loss": -0.0018, "step": 218 }, { "epoch": 0.05, "learning_rate": 3.8426053594471375e-10, "loss": -0.0012, "step": 219 }, { "epoch": 0.05, "learning_rate": 1.7079765484293574e-10, "loss": -0.0003, "step": 220 }, { "epoch": 0.05, "learning_rate": 4.270169300397697e-11, "loss": 0.0001, "step": 221 }, { "epoch": 0.05, "learning_rate": 0.0, "loss": -0.0002, "step": 222 }, { "epoch": 0.05, "step": 222, "total_flos": 6.51669001961472e+16, "train_loss": -0.0006611563352556266, "train_runtime": 59896.0215, "train_samples_per_second": 0.177, "train_steps_per_second": 0.004 } ], "logging_steps": 1.0, "max_steps": 222, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 200, "total_flos": 6.51669001961472e+16, "train_batch_size": 48, "trial_name": null, "trial_params": null }