| { | |
| "best_metric": 0.9624110460281372, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-100", | |
| "epoch": 3.0049751243781095, | |
| "eval_steps": 50, | |
| "global_step": 151, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01990049751243781, | |
| "grad_norm": 5.543930530548096, | |
| "learning_rate": 1e-05, | |
| "loss": 5.5751, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01990049751243781, | |
| "eval_loss": 9.599178314208984, | |
| "eval_runtime": 6.162, | |
| "eval_samples_per_second": 13.794, | |
| "eval_steps_per_second": 3.57, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.03980099502487562, | |
| "grad_norm": 5.9391279220581055, | |
| "learning_rate": 2e-05, | |
| "loss": 6.2605, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.05970149253731343, | |
| "grad_norm": 10.055806159973145, | |
| "learning_rate": 3e-05, | |
| "loss": 6.6465, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.07960199004975124, | |
| "grad_norm": 24.726375579833984, | |
| "learning_rate": 4e-05, | |
| "loss": 9.6417, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.09950248756218906, | |
| "grad_norm": 21.94843292236328, | |
| "learning_rate": 5e-05, | |
| "loss": 9.1886, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.11940298507462686, | |
| "grad_norm": 17.2749080657959, | |
| "learning_rate": 6e-05, | |
| "loss": 7.9442, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.13930348258706468, | |
| "grad_norm": 20.374170303344727, | |
| "learning_rate": 7e-05, | |
| "loss": 5.9546, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.15920398009950248, | |
| "grad_norm": 14.176255226135254, | |
| "learning_rate": 8e-05, | |
| "loss": 3.7086, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.1791044776119403, | |
| "grad_norm": 8.863327026367188, | |
| "learning_rate": 9e-05, | |
| "loss": 1.8624, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.19900497512437812, | |
| "grad_norm": 7.542079925537109, | |
| "learning_rate": 0.0001, | |
| "loss": 5.9968, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.21890547263681592, | |
| "grad_norm": 15.19139289855957, | |
| "learning_rate": 9.998758966336295e-05, | |
| "loss": 8.1119, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.23880597014925373, | |
| "grad_norm": 18.45025634765625, | |
| "learning_rate": 9.995036481411004e-05, | |
| "loss": 8.9081, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.25870646766169153, | |
| "grad_norm": 12.949933052062988, | |
| "learning_rate": 9.988834393115767e-05, | |
| "loss": 4.0368, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.27860696517412936, | |
| "grad_norm": 12.047260284423828, | |
| "learning_rate": 9.980155780250727e-05, | |
| "loss": 4.0461, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.29850746268656714, | |
| "grad_norm": 10.137738227844238, | |
| "learning_rate": 9.969004950996175e-05, | |
| "loss": 3.5263, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.31840796019900497, | |
| "grad_norm": 7.845066547393799, | |
| "learning_rate": 9.9553874407739e-05, | |
| "loss": 1.8586, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.3383084577114428, | |
| "grad_norm": 1.7427806854248047, | |
| "learning_rate": 9.939310009499348e-05, | |
| "loss": 0.4681, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.3582089552238806, | |
| "grad_norm": 1.3990861177444458, | |
| "learning_rate": 9.92078063822589e-05, | |
| "loss": 0.3307, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.3781094527363184, | |
| "grad_norm": 0.9510717988014221, | |
| "learning_rate": 9.899808525182935e-05, | |
| "loss": 0.148, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.39800995024875624, | |
| "grad_norm": 0.7023572325706482, | |
| "learning_rate": 9.876404081209796e-05, | |
| "loss": 0.0863, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.417910447761194, | |
| "grad_norm": 2.599980354309082, | |
| "learning_rate": 9.850578924587614e-05, | |
| "loss": 0.8357, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.43781094527363185, | |
| "grad_norm": 8.273324966430664, | |
| "learning_rate": 9.822345875271883e-05, | |
| "loss": 3.8429, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.4577114427860697, | |
| "grad_norm": 10.221076965332031, | |
| "learning_rate": 9.791718948528458e-05, | |
| "loss": 4.1774, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.47761194029850745, | |
| "grad_norm": 10.803452491760254, | |
| "learning_rate": 9.758713347976179e-05, | |
| "loss": 4.457, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.4975124378109453, | |
| "grad_norm": 9.160074234008789, | |
| "learning_rate": 9.723345458039594e-05, | |
| "loss": 2.9867, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.5174129353233831, | |
| "grad_norm": 4.31099796295166, | |
| "learning_rate": 9.685632835815518e-05, | |
| "loss": 2.5178, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.5373134328358209, | |
| "grad_norm": 3.5618479251861572, | |
| "learning_rate": 9.645594202357439e-05, | |
| "loss": 2.5185, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.5572139303482587, | |
| "grad_norm": 2.6641221046447754, | |
| "learning_rate": 9.603249433382144e-05, | |
| "loss": 1.2579, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.5771144278606966, | |
| "grad_norm": 1.0730020999908447, | |
| "learning_rate": 9.558619549403147e-05, | |
| "loss": 0.3654, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.5970149253731343, | |
| "grad_norm": 0.2987671196460724, | |
| "learning_rate": 9.511726705295817e-05, | |
| "loss": 0.0464, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.6169154228855721, | |
| "grad_norm": 0.8347283005714417, | |
| "learning_rate": 9.462594179299406e-05, | |
| "loss": 0.2107, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.6368159203980099, | |
| "grad_norm": 0.8029500246047974, | |
| "learning_rate": 9.41124636146141e-05, | |
| "loss": 0.179, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.6567164179104478, | |
| "grad_norm": 1.6079643964767456, | |
| "learning_rate": 9.357708741530025e-05, | |
| "loss": 0.5489, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.6766169154228856, | |
| "grad_norm": 7.646543502807617, | |
| "learning_rate": 9.302007896300698e-05, | |
| "loss": 3.2044, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.6965174129353234, | |
| "grad_norm": 8.641079902648926, | |
| "learning_rate": 9.244171476423037e-05, | |
| "loss": 3.0777, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.7164179104477612, | |
| "grad_norm": 11.796111106872559, | |
| "learning_rate": 9.184228192674667e-05, | |
| "loss": 4.0429, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.736318407960199, | |
| "grad_norm": 5.918362617492676, | |
| "learning_rate": 9.122207801708802e-05, | |
| "loss": 2.5737, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.7562189054726368, | |
| "grad_norm": 6.796441078186035, | |
| "learning_rate": 9.058141091282656e-05, | |
| "loss": 2.2759, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.7761194029850746, | |
| "grad_norm": 5.166509628295898, | |
| "learning_rate": 8.992059864973972e-05, | |
| "loss": 2.3741, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.7960199004975125, | |
| "grad_norm": 9.372264862060547, | |
| "learning_rate": 8.923996926393305e-05, | |
| "loss": 0.6126, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.8159203980099502, | |
| "grad_norm": 0.7963209748268127, | |
| "learning_rate": 8.853986062899868e-05, | |
| "loss": 0.2505, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.835820895522388, | |
| "grad_norm": 0.8974098563194275, | |
| "learning_rate": 8.782062028829028e-05, | |
| "loss": 0.3499, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.8557213930348259, | |
| "grad_norm": 0.8836018443107605, | |
| "learning_rate": 8.708260528239788e-05, | |
| "loss": 0.336, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.8756218905472637, | |
| "grad_norm": 0.41947388648986816, | |
| "learning_rate": 8.632618197190816e-05, | |
| "loss": 0.0833, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.8955223880597015, | |
| "grad_norm": 0.6316033601760864, | |
| "learning_rate": 8.555172585553805e-05, | |
| "loss": 0.1746, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.9154228855721394, | |
| "grad_norm": 3.0872931480407715, | |
| "learning_rate": 8.475962138373213e-05, | |
| "loss": 1.8755, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.9353233830845771, | |
| "grad_norm": 5.99851655960083, | |
| "learning_rate": 8.395026176781627e-05, | |
| "loss": 3.2634, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.9552238805970149, | |
| "grad_norm": 8.195245742797852, | |
| "learning_rate": 8.312404878480222e-05, | |
| "loss": 3.2091, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.9751243781094527, | |
| "grad_norm": 2.1140687465667725, | |
| "learning_rate": 8.228139257794012e-05, | |
| "loss": 1.2906, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.9950248756218906, | |
| "grad_norm": 2.3306195735931396, | |
| "learning_rate": 8.142271145311783e-05, | |
| "loss": 1.0865, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.9950248756218906, | |
| "eval_loss": 1.0756102800369263, | |
| "eval_runtime": 6.3397, | |
| "eval_samples_per_second": 13.408, | |
| "eval_steps_per_second": 3.47, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.0149253731343284, | |
| "grad_norm": 9.008943557739258, | |
| "learning_rate": 8.054843167120827e-05, | |
| "loss": 3.9621, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 1.0348258706467661, | |
| "grad_norm": 2.001354455947876, | |
| "learning_rate": 7.965898723646776e-05, | |
| "loss": 1.5088, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 1.054726368159204, | |
| "grad_norm": 2.112483024597168, | |
| "learning_rate": 7.875481968109052e-05, | |
| "loss": 1.5239, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 1.0746268656716418, | |
| "grad_norm": 1.9544909000396729, | |
| "learning_rate": 7.783637784602609e-05, | |
| "loss": 0.975, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 1.0945273631840795, | |
| "grad_norm": 0.5338630676269531, | |
| "learning_rate": 7.690411765816864e-05, | |
| "loss": 0.1268, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 1.1144278606965174, | |
| "grad_norm": 0.5618958473205566, | |
| "learning_rate": 7.595850190402876e-05, | |
| "loss": 0.1182, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 1.1343283582089552, | |
| "grad_norm": 0.6160061955451965, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 0.144, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 1.154228855721393, | |
| "grad_norm": 0.5106895565986633, | |
| "learning_rate": 7.402908775933419e-05, | |
| "loss": 0.0888, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 1.1741293532338308, | |
| "grad_norm": 0.39127403497695923, | |
| "learning_rate": 7.304624715594139e-05, | |
| "loss": 0.0848, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 1.1940298507462686, | |
| "grad_norm": 3.96272611618042, | |
| "learning_rate": 7.205196608513159e-05, | |
| "loss": 1.3122, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.2139303482587065, | |
| "grad_norm": 4.137331008911133, | |
| "learning_rate": 7.104673812141675e-05, | |
| "loss": 2.0647, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 1.2338308457711442, | |
| "grad_norm": 9.125043869018555, | |
| "learning_rate": 7.003106227349399e-05, | |
| "loss": 2.4258, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 1.2537313432835822, | |
| "grad_norm": 4.514773845672607, | |
| "learning_rate": 6.900544273653074e-05, | |
| "loss": 2.0476, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 1.2736318407960199, | |
| "grad_norm": 2.9408376216888428, | |
| "learning_rate": 6.797038864187564e-05, | |
| "loss": 1.5427, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 1.2935323383084576, | |
| "grad_norm": 2.8879308700561523, | |
| "learning_rate": 6.692641380431879e-05, | |
| "loss": 1.7285, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 1.3134328358208955, | |
| "grad_norm": 1.3168548345565796, | |
| "learning_rate": 6.587403646702714e-05, | |
| "loss": 0.5157, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 0.7307369112968445, | |
| "learning_rate": 6.481377904428171e-05, | |
| "loss": 0.138, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 1.3532338308457712, | |
| "grad_norm": 0.3926345705986023, | |
| "learning_rate": 6.374616786214402e-05, | |
| "loss": 0.0664, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.373134328358209, | |
| "grad_norm": 0.3398008346557617, | |
| "learning_rate": 6.26717328971808e-05, | |
| "loss": 0.0546, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 1.3930348258706466, | |
| "grad_norm": 0.7794240713119507, | |
| "learning_rate": 6.159100751337642e-05, | |
| "loss": 0.1423, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.4129353233830846, | |
| "grad_norm": 0.28816598653793335, | |
| "learning_rate": 6.0504528197363894e-05, | |
| "loss": 0.0374, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 1.4328358208955223, | |
| "grad_norm": 2.6052889823913574, | |
| "learning_rate": 5.9412834292105676e-05, | |
| "loss": 1.3742, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.4527363184079602, | |
| "grad_norm": 5.297976970672607, | |
| "learning_rate": 5.831646772915651e-05, | |
| "loss": 2.3129, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 1.472636815920398, | |
| "grad_norm": 6.222244739532471, | |
| "learning_rate": 5.721597275964133e-05, | |
| "loss": 2.4451, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.4925373134328357, | |
| "grad_norm": 3.1612331867218018, | |
| "learning_rate": 5.6111895684081725e-05, | |
| "loss": 1.5285, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.5124378109452736, | |
| "grad_norm": 2.373920440673828, | |
| "learning_rate": 5.5004784581204927e-05, | |
| "loss": 1.4736, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.5323383084577116, | |
| "grad_norm": 2.493252992630005, | |
| "learning_rate": 5.389518903587017e-05, | |
| "loss": 1.5259, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.5522388059701493, | |
| "grad_norm": 2.429081678390503, | |
| "learning_rate": 5.2783659866247424e-05, | |
| "loss": 1.6323, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.572139303482587, | |
| "grad_norm": 0.9949527382850647, | |
| "learning_rate": 5.167074885038373e-05, | |
| "loss": 0.3188, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.5920398009950247, | |
| "grad_norm": 0.5965785980224609, | |
| "learning_rate": 5.055700845229327e-05, | |
| "loss": 0.1503, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.6119402985074627, | |
| "grad_norm": 0.5434532165527344, | |
| "learning_rate": 4.944299154770673e-05, | |
| "loss": 0.094, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.6318407960199006, | |
| "grad_norm": 0.7457742691040039, | |
| "learning_rate": 4.832925114961629e-05, | |
| "loss": 0.164, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.6517412935323383, | |
| "grad_norm": 0.3958088457584381, | |
| "learning_rate": 4.72163401337526e-05, | |
| "loss": 0.0606, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.671641791044776, | |
| "grad_norm": 1.8904391527175903, | |
| "learning_rate": 4.610481096412984e-05, | |
| "loss": 0.9604, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.6915422885572138, | |
| "grad_norm": 4.544404029846191, | |
| "learning_rate": 4.4995215418795085e-05, | |
| "loss": 2.215, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.7114427860696517, | |
| "grad_norm": 5.509747505187988, | |
| "learning_rate": 4.388810431591829e-05, | |
| "loss": 2.0638, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.7313432835820897, | |
| "grad_norm": 2.859410047531128, | |
| "learning_rate": 4.278402724035867e-05, | |
| "loss": 1.693, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.7512437810945274, | |
| "grad_norm": 2.292931079864502, | |
| "learning_rate": 4.1683532270843504e-05, | |
| "loss": 1.3341, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.771144278606965, | |
| "grad_norm": 2.5066726207733154, | |
| "learning_rate": 4.058716570789433e-05, | |
| "loss": 1.4565, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.7910447761194028, | |
| "grad_norm": 1.7927272319793701, | |
| "learning_rate": 3.94954718026361e-05, | |
| "loss": 0.892, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.8109452736318408, | |
| "grad_norm": 0.5884804725646973, | |
| "learning_rate": 3.840899248662358e-05, | |
| "loss": 0.1595, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.8308457711442787, | |
| "grad_norm": 1.1309645175933838, | |
| "learning_rate": 3.7328267102819225e-05, | |
| "loss": 0.2703, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.8507462686567164, | |
| "grad_norm": 0.31311914324760437, | |
| "learning_rate": 3.6253832137856e-05, | |
| "loss": 0.0479, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.8706467661691542, | |
| "grad_norm": 0.5945075154304504, | |
| "learning_rate": 3.5186220955718306e-05, | |
| "loss": 0.1501, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.890547263681592, | |
| "grad_norm": 1.0878932476043701, | |
| "learning_rate": 3.4125963532972873e-05, | |
| "loss": 0.4793, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.9104477611940298, | |
| "grad_norm": 3.067518472671509, | |
| "learning_rate": 3.307358619568123e-05, | |
| "loss": 1.7691, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.9303482587064678, | |
| "grad_norm": 3.476691722869873, | |
| "learning_rate": 3.202961135812437e-05, | |
| "loss": 1.928, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.9502487562189055, | |
| "grad_norm": 4.594018459320068, | |
| "learning_rate": 3.0994557263469265e-05, | |
| "loss": 2.3617, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.9701492537313432, | |
| "grad_norm": 2.5807111263275146, | |
| "learning_rate": 2.996893772650602e-05, | |
| "loss": 1.4338, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.9900497512437811, | |
| "grad_norm": 0.2568559944629669, | |
| "learning_rate": 2.895326187858326e-05, | |
| "loss": 0.0345, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.9900497512437811, | |
| "eval_loss": 0.9624110460281372, | |
| "eval_runtime": 6.3475, | |
| "eval_samples_per_second": 13.391, | |
| "eval_steps_per_second": 3.466, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 2.009950248756219, | |
| "grad_norm": 15.313194274902344, | |
| "learning_rate": 2.7948033914868415e-05, | |
| "loss": 3.6961, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 2.029850746268657, | |
| "grad_norm": 1.6831941604614258, | |
| "learning_rate": 2.69537528440586e-05, | |
| "loss": 0.9623, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 2.0497512437810945, | |
| "grad_norm": 1.7447772026062012, | |
| "learning_rate": 2.5970912240665813e-05, | |
| "loss": 1.0607, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 2.0696517412935322, | |
| "grad_norm": 1.8062071800231934, | |
| "learning_rate": 2.500000000000001e-05, | |
| "loss": 1.0382, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 2.08955223880597, | |
| "grad_norm": 0.533903181552887, | |
| "learning_rate": 2.4041498095971253e-05, | |
| "loss": 0.0953, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 2.109452736318408, | |
| "grad_norm": 0.41501080989837646, | |
| "learning_rate": 2.3095882341831372e-05, | |
| "loss": 0.0889, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 2.129353233830846, | |
| "grad_norm": 0.605328381061554, | |
| "learning_rate": 2.216362215397393e-05, | |
| "loss": 0.1302, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 2.1492537313432836, | |
| "grad_norm": 0.3093937039375305, | |
| "learning_rate": 2.124518031890948e-05, | |
| "loss": 0.0564, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 2.1691542288557213, | |
| "grad_norm": 0.321236252784729, | |
| "learning_rate": 2.0341012763532243e-05, | |
| "loss": 0.0479, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 2.189054726368159, | |
| "grad_norm": 1.6769386529922485, | |
| "learning_rate": 1.945156832879174e-05, | |
| "loss": 0.6905, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 2.208955223880597, | |
| "grad_norm": 3.2930898666381836, | |
| "learning_rate": 1.8577288546882167e-05, | |
| "loss": 1.2683, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 2.228855721393035, | |
| "grad_norm": 3.6497483253479004, | |
| "learning_rate": 1.771860742205988e-05, | |
| "loss": 1.4517, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 2.2487562189054726, | |
| "grad_norm": 2.7759523391723633, | |
| "learning_rate": 1.687595121519778e-05, | |
| "loss": 1.2682, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 2.2686567164179103, | |
| "grad_norm": 1.7020666599273682, | |
| "learning_rate": 1.604973823218376e-05, | |
| "loss": 0.8624, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 2.288557213930348, | |
| "grad_norm": 2.058931589126587, | |
| "learning_rate": 1.5240378616267886e-05, | |
| "loss": 0.9802, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 2.308457711442786, | |
| "grad_norm": 1.6504006385803223, | |
| "learning_rate": 1.4448274144461965e-05, | |
| "loss": 0.626, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 2.328358208955224, | |
| "grad_norm": 0.36916500329971313, | |
| "learning_rate": 1.367381802809185e-05, | |
| "loss": 0.0499, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 2.3482587064676617, | |
| "grad_norm": 0.4427722692489624, | |
| "learning_rate": 1.2917394717602121e-05, | |
| "loss": 0.0567, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 2.3681592039800994, | |
| "grad_norm": 0.24329277873039246, | |
| "learning_rate": 1.2179379711709737e-05, | |
| "loss": 0.0243, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 2.388059701492537, | |
| "grad_norm": 0.7087410688400269, | |
| "learning_rate": 1.1460139371001338e-05, | |
| "loss": 0.155, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.4079601990049753, | |
| "grad_norm": 0.24490374326705933, | |
| "learning_rate": 1.0760030736066951e-05, | |
| "loss": 0.011, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 2.427860696517413, | |
| "grad_norm": 2.3325791358947754, | |
| "learning_rate": 1.0079401350260287e-05, | |
| "loss": 0.6917, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 2.4477611940298507, | |
| "grad_norm": 3.353835344314575, | |
| "learning_rate": 9.41858908717344e-06, | |
| "loss": 1.2289, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 2.4676616915422884, | |
| "grad_norm": 4.807076930999756, | |
| "learning_rate": 8.777921982911996e-06, | |
| "loss": 1.6531, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 2.487562189054726, | |
| "grad_norm": 3.2822203636169434, | |
| "learning_rate": 8.157718073253351e-06, | |
| "loss": 1.3298, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 2.5074626865671643, | |
| "grad_norm": 2.1589744091033936, | |
| "learning_rate": 7.558285235769646e-06, | |
| "loss": 0.8649, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 2.527363184079602, | |
| "grad_norm": 2.269338607788086, | |
| "learning_rate": 6.979921036993042e-06, | |
| "loss": 1.0175, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 2.5472636815920398, | |
| "grad_norm": 1.7215580940246582, | |
| "learning_rate": 6.422912584699753e-06, | |
| "loss": 0.4796, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 2.5671641791044775, | |
| "grad_norm": 0.5773996710777283, | |
| "learning_rate": 5.887536385385917e-06, | |
| "loss": 0.0872, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 2.587064676616915, | |
| "grad_norm": 0.3628270924091339, | |
| "learning_rate": 5.374058207005944e-06, | |
| "loss": 0.0363, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 2.6069651741293534, | |
| "grad_norm": 0.5935407876968384, | |
| "learning_rate": 4.882732947041818e-06, | |
| "loss": 0.1109, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 2.626865671641791, | |
| "grad_norm": 0.3369639217853546, | |
| "learning_rate": 4.413804505968533e-06, | |
| "loss": 0.0402, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 2.646766169154229, | |
| "grad_norm": 0.4762958288192749, | |
| "learning_rate": 3.967505666178556e-06, | |
| "loss": 0.0557, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 2.6666666666666665, | |
| "grad_norm": 2.259021759033203, | |
| "learning_rate": 3.544057976425619e-06, | |
| "loss": 0.8966, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 2.6865671641791042, | |
| "grad_norm": 3.387164831161499, | |
| "learning_rate": 3.1436716418448307e-06, | |
| "loss": 1.3748, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 2.7064676616915424, | |
| "grad_norm": 4.192432880401611, | |
| "learning_rate": 2.7665454196040664e-06, | |
| "loss": 1.4131, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 2.72636815920398, | |
| "grad_norm": 2.7902915477752686, | |
| "learning_rate": 2.4128665202382326e-06, | |
| "loss": 1.1285, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 2.746268656716418, | |
| "grad_norm": 2.0142769813537598, | |
| "learning_rate": 2.0828105147154273e-06, | |
| "loss": 0.7524, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 2.7661691542288556, | |
| "grad_norm": 2.4600789546966553, | |
| "learning_rate": 1.7765412472811771e-06, | |
| "loss": 1.0153, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 2.7860696517412933, | |
| "grad_norm": 2.070169448852539, | |
| "learning_rate": 1.4942107541238704e-06, | |
| "loss": 0.8548, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.8059701492537314, | |
| "grad_norm": 0.9298514127731323, | |
| "learning_rate": 1.2359591879020526e-06, | |
| "loss": 0.2304, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 2.825870646766169, | |
| "grad_norm": 0.287707656621933, | |
| "learning_rate": 1.0019147481706625e-06, | |
| "loss": 0.0342, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 2.845771144278607, | |
| "grad_norm": 0.24002987146377563, | |
| "learning_rate": 7.921936177411049e-07, | |
| "loss": 0.0208, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 2.8656716417910446, | |
| "grad_norm": 0.5002749562263489, | |
| "learning_rate": 6.06899905006525e-07, | |
| "loss": 0.0661, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 2.8855721393034823, | |
| "grad_norm": 0.4782179594039917, | |
| "learning_rate": 4.461255922609986e-07, | |
| "loss": 0.088, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 2.9054726368159205, | |
| "grad_norm": 1.0259336233139038, | |
| "learning_rate": 3.0995049003826325e-07, | |
| "loss": 0.2354, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 2.925373134328358, | |
| "grad_norm": 4.005832195281982, | |
| "learning_rate": 1.984421974927375e-07, | |
| "loss": 1.3838, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 2.945273631840796, | |
| "grad_norm": 3.5791056156158447, | |
| "learning_rate": 1.1165606884234181e-07, | |
| "loss": 1.5911, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 2.965174129353234, | |
| "grad_norm": 3.579153299331665, | |
| "learning_rate": 4.963518588996796e-08, | |
| "loss": 1.3113, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 2.9850746268656714, | |
| "grad_norm": 0.6860787868499756, | |
| "learning_rate": 1.2410336637047605e-08, | |
| "loss": 0.1386, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.9850746268656714, | |
| "eval_loss": 1.0078964233398438, | |
| "eval_runtime": 6.3591, | |
| "eval_samples_per_second": 13.367, | |
| "eval_steps_per_second": 3.46, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 3.0049751243781095, | |
| "grad_norm": 7.453361511230469, | |
| "learning_rate": 0.0, | |
| "loss": 2.4948, | |
| "step": 151 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 151, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 1 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.2778555039062426e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |