| { | |
| "best_metric": 1.154872179031372, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.27481964960494676, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0013740982480247338, | |
| "grad_norm": 0.10447467863559723, | |
| "learning_rate": 1e-05, | |
| "loss": 0.685, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0013740982480247338, | |
| "eval_loss": 3.1940252780914307, | |
| "eval_runtime": 34.4278, | |
| "eval_samples_per_second": 35.611, | |
| "eval_steps_per_second": 8.917, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0027481964960494676, | |
| "grad_norm": 0.12629619240760803, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8576, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.004122294744074201, | |
| "grad_norm": 0.12000161409378052, | |
| "learning_rate": 3e-05, | |
| "loss": 0.8004, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.005496392992098935, | |
| "grad_norm": 0.15668775141239166, | |
| "learning_rate": 4e-05, | |
| "loss": 0.9421, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.006870491240123669, | |
| "grad_norm": 0.1253223568201065, | |
| "learning_rate": 5e-05, | |
| "loss": 0.91, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.008244589488148402, | |
| "grad_norm": 0.1712837964296341, | |
| "learning_rate": 6e-05, | |
| "loss": 0.853, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.009618687736173137, | |
| "grad_norm": 0.14983898401260376, | |
| "learning_rate": 7e-05, | |
| "loss": 0.9085, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.01099278598419787, | |
| "grad_norm": 0.1604943573474884, | |
| "learning_rate": 8e-05, | |
| "loss": 0.9668, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.012366884232222604, | |
| "grad_norm": 0.1646648794412613, | |
| "learning_rate": 9e-05, | |
| "loss": 0.9474, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.013740982480247337, | |
| "grad_norm": 0.14233022928237915, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9307, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01511508072827207, | |
| "grad_norm": 0.14303773641586304, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.0272, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.016489178976296804, | |
| "grad_norm": 0.13132160902023315, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.9655, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.017863277224321537, | |
| "grad_norm": 0.1405678391456604, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.9454, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.019237375472346274, | |
| "grad_norm": 0.1712307631969452, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.901, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.020611473720371008, | |
| "grad_norm": 0.17713665962219238, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.1299, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02198557196839574, | |
| "grad_norm": 0.1875886619091034, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.03, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.023359670216420474, | |
| "grad_norm": 0.16109849512577057, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.9476, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.024733768464445208, | |
| "grad_norm": 0.1598743498325348, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.9316, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.02610786671246994, | |
| "grad_norm": 0.15691572427749634, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.9573, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.027481964960494674, | |
| "grad_norm": 0.1694604903459549, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.9552, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.028856063208519408, | |
| "grad_norm": 0.16706468164920807, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.0703, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.03023016145654414, | |
| "grad_norm": 0.16619545221328735, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.1266, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.03160425970456888, | |
| "grad_norm": 0.17776361107826233, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.9576, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.03297835795259361, | |
| "grad_norm": 0.1760794222354889, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.0205, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.034352456200618345, | |
| "grad_norm": 0.1899178922176361, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.0423, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.035726554448643075, | |
| "grad_norm": 0.20989012718200684, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.0949, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.03710065269666781, | |
| "grad_norm": 0.22795264422893524, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 1.0732, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.03847475094469255, | |
| "grad_norm": 0.24893148243427277, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.1178, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.03984884919271728, | |
| "grad_norm": 0.3269492983818054, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.2364, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.041222947440742015, | |
| "grad_norm": 0.2281530201435089, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.1341, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.042597045688766745, | |
| "grad_norm": 0.2597139775753021, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 1.1221, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.04397114393679148, | |
| "grad_norm": 0.9067723751068115, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 1.9614, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.04534524218481621, | |
| "grad_norm": 4.838332176208496, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 6.601, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.04671934043284095, | |
| "grad_norm": 9.844199180603027, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 10.6881, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.04809343868086568, | |
| "grad_norm": 10.059837341308594, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 9.3037, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.049467536928890415, | |
| "grad_norm": 8.158489227294922, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 7.7695, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.05084163517691515, | |
| "grad_norm": 8.049012184143066, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 7.3775, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.05221573342493988, | |
| "grad_norm": 8.838970184326172, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 7.0543, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.05358983167296462, | |
| "grad_norm": 11.280040740966797, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 6.6935, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.05496392992098935, | |
| "grad_norm": 10.341362953186035, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 5.954, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.056338028169014086, | |
| "grad_norm": 2.2097039222717285, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 2.2857, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.057712126417038816, | |
| "grad_norm": 2.333188533782959, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 1.4563, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.05908622466506355, | |
| "grad_norm": 2.797895669937134, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 1.5016, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.06046032291308828, | |
| "grad_norm": 2.8829591274261475, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 1.3985, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.06183442116111302, | |
| "grad_norm": 2.3330905437469482, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 1.4629, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.06320851940913776, | |
| "grad_norm": 1.7073644399642944, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 1.4424, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.06458261765716249, | |
| "grad_norm": 1.1441912651062012, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 1.305, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.06595671590518722, | |
| "grad_norm": 1.0069657564163208, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 1.547, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.06733081415321196, | |
| "grad_norm": 0.795106828212738, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 1.3543, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.06870491240123669, | |
| "grad_norm": 0.7250723242759705, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 1.5456, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06870491240123669, | |
| "eval_loss": 1.583877444267273, | |
| "eval_runtime": 34.2929, | |
| "eval_samples_per_second": 35.751, | |
| "eval_steps_per_second": 8.952, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.07007901064926142, | |
| "grad_norm": 0.3846101462841034, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.831, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.07145310889728615, | |
| "grad_norm": 0.4264107644557953, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 1.0423, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0728272071453109, | |
| "grad_norm": 0.3170631229877472, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.874, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.07420130539333562, | |
| "grad_norm": 0.26041364669799805, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.9679, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.07557540364136035, | |
| "grad_norm": 0.23548555374145508, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.825, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0769495018893851, | |
| "grad_norm": 0.20205585658550262, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.8373, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.07832360013740983, | |
| "grad_norm": 0.19775812327861786, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.9506, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.07969769838543456, | |
| "grad_norm": 0.15645460784435272, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.9405, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.08107179663345929, | |
| "grad_norm": 0.14678984880447388, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.8962, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.08244589488148403, | |
| "grad_norm": 0.1445266157388687, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.851, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.08381999312950876, | |
| "grad_norm": 0.21242700517177582, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.036, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.08519409137753349, | |
| "grad_norm": 0.17483900487422943, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.9628, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.08656818962555823, | |
| "grad_norm": 0.14225730299949646, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.9495, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.08794228787358296, | |
| "grad_norm": 0.1447581648826599, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.8811, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.0893163861216077, | |
| "grad_norm": 0.14672036468982697, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.9486, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.09069048436963242, | |
| "grad_norm": 0.1578073501586914, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.9383, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.09206458261765717, | |
| "grad_norm": 0.1623167097568512, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.0042, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.0934386808656819, | |
| "grad_norm": 0.16508011519908905, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.9384, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.09481277911370663, | |
| "grad_norm": 0.15675126016139984, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.9524, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.09618687736173136, | |
| "grad_norm": 0.1692209541797638, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.0562, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0975609756097561, | |
| "grad_norm": 0.17102833092212677, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.8768, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.09893507385778083, | |
| "grad_norm": 0.16048406064510345, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.8628, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.10030917210580556, | |
| "grad_norm": 0.18186289072036743, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.98, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.1016832703538303, | |
| "grad_norm": 0.17549650371074677, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.8624, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.10305736860185503, | |
| "grad_norm": 0.17140643298625946, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 1.0334, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.10443146684987976, | |
| "grad_norm": 0.2252202332019806, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.0619, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.1058055650979045, | |
| "grad_norm": 0.22643177211284637, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 1.0897, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.10717966334592924, | |
| "grad_norm": 0.2238086313009262, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.1413, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.10855376159395397, | |
| "grad_norm": 0.2520258128643036, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.1535, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.1099278598419787, | |
| "grad_norm": 0.2497335523366928, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 1.0523, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.11130195809000344, | |
| "grad_norm": 0.4694143533706665, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.3923, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.11267605633802817, | |
| "grad_norm": 2.4829208850860596, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 4.44, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.1140501545860529, | |
| "grad_norm": 1.9080414772033691, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 2.6943, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.11542425283407763, | |
| "grad_norm": 8.531490325927734, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 5.4883, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.11679835108210238, | |
| "grad_norm": 8.17274284362793, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 5.6548, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.1181724493301271, | |
| "grad_norm": 7.879384994506836, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 5.3349, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.11954654757815183, | |
| "grad_norm": 6.465800762176514, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 4.1707, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.12092064582617656, | |
| "grad_norm": 6.287056922912598, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 4.6775, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.12229474407420131, | |
| "grad_norm": 5.547956943511963, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 3.763, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.12366884232222604, | |
| "grad_norm": 4.8003129959106445, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 3.0735, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.12504294057025078, | |
| "grad_norm": 0.3764083683490753, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 1.1216, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.1264170388182755, | |
| "grad_norm": 0.365705668926239, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.8125, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.12779113706630024, | |
| "grad_norm": 0.3882658779621124, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.9153, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.12916523531432497, | |
| "grad_norm": 0.4514417350292206, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.8798, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.1305393335623497, | |
| "grad_norm": 0.5178213715553284, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 1.3223, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.13191343181037443, | |
| "grad_norm": 0.5428131818771362, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 1.3461, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.13328753005839916, | |
| "grad_norm": 0.49981769919395447, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 1.085, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.13466162830642392, | |
| "grad_norm": 0.5136014223098755, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 1.1512, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.13603572655444865, | |
| "grad_norm": 0.512438952922821, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 1.2455, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.13740982480247338, | |
| "grad_norm": 0.5576941967010498, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 1.5722, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.13740982480247338, | |
| "eval_loss": 1.2312716245651245, | |
| "eval_runtime": 34.26, | |
| "eval_samples_per_second": 35.785, | |
| "eval_steps_per_second": 8.961, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1387839230504981, | |
| "grad_norm": 0.2255360633134842, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.7779, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.14015802129852284, | |
| "grad_norm": 0.23751790821552277, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.8338, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.14153211954654757, | |
| "grad_norm": 0.2663083076477051, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.9188, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.1429062177945723, | |
| "grad_norm": 0.22981339693069458, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.8188, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.14428031604259706, | |
| "grad_norm": 0.236619770526886, | |
| "learning_rate": 5e-05, | |
| "loss": 0.8803, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.1456544142906218, | |
| "grad_norm": 0.23290015757083893, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.9059, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.14702851253864652, | |
| "grad_norm": 0.19473782181739807, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.8494, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.14840261078667125, | |
| "grad_norm": 0.26345762610435486, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.8652, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.14977670903469598, | |
| "grad_norm": 0.17493213713169098, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.8151, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.1511508072827207, | |
| "grad_norm": 0.18951790034770966, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.8168, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.15252490553074544, | |
| "grad_norm": 0.18956202268600464, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.8934, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.1538990037787702, | |
| "grad_norm": 0.212601900100708, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.9969, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.15527310202679492, | |
| "grad_norm": 0.15459412336349487, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.8822, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.15664720027481965, | |
| "grad_norm": 0.18324562907218933, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.9027, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.15802129852284438, | |
| "grad_norm": 0.18034571409225464, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.9191, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.1593953967708691, | |
| "grad_norm": 0.16426022350788116, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.9042, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.16076949501889384, | |
| "grad_norm": 0.15972800552845, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.8766, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.16214359326691857, | |
| "grad_norm": 0.15054263174533844, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.9411, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.16351769151494333, | |
| "grad_norm": 0.15989775955677032, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.9101, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.16489178976296806, | |
| "grad_norm": 0.169904425740242, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.8387, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.1662658880109928, | |
| "grad_norm": 0.16550658643245697, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.9908, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.16763998625901752, | |
| "grad_norm": 0.1593722254037857, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.8916, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.16901408450704225, | |
| "grad_norm": 0.1593838334083557, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.9206, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.17038818275506698, | |
| "grad_norm": 0.152430459856987, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.881, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.1717622810030917, | |
| "grad_norm": 0.18957607448101044, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 1.0454, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.17313637925111647, | |
| "grad_norm": 0.18225038051605225, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 1.0129, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.1745104774991412, | |
| "grad_norm": 0.1942896544933319, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 1.0398, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.17588457574716593, | |
| "grad_norm": 0.20318052172660828, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 1.0716, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.17725867399519066, | |
| "grad_norm": 0.23516058921813965, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 1.0764, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.1786327722432154, | |
| "grad_norm": 0.22030310332775116, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 1.0809, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.18000687049124012, | |
| "grad_norm": 0.2949303090572357, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 1.311, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.18138096873926485, | |
| "grad_norm": 0.2638658583164215, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.1203, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.1827550669872896, | |
| "grad_norm": 0.21188068389892578, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.8101, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.18412916523531433, | |
| "grad_norm": 1.5809319019317627, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 3.1598, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.18550326348333906, | |
| "grad_norm": 1.4560643434524536, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 2.7776, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.1868773617313638, | |
| "grad_norm": 4.069778919219971, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 2.8172, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.18825145997938852, | |
| "grad_norm": 3.78922700881958, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 2.4094, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.18962555822741325, | |
| "grad_norm": 0.9479630589485168, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.7767, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.19099965647543798, | |
| "grad_norm": 0.9855109453201294, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 1.6067, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.19237375472346271, | |
| "grad_norm": 3.8103370666503906, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 2.5768, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.19374785297148747, | |
| "grad_norm": 3.6094367504119873, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 2.1034, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.1951219512195122, | |
| "grad_norm": 0.40120404958724976, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 1.0799, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.19649604946753693, | |
| "grad_norm": 0.2957104444503784, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.8669, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.19787014771556166, | |
| "grad_norm": 0.3517855107784271, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 1.1585, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.1992442459635864, | |
| "grad_norm": 0.2185845673084259, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.8003, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.20061834421161112, | |
| "grad_norm": 0.2852717638015747, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.8969, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.20199244245963585, | |
| "grad_norm": 0.2719075679779053, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.8089, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.2033665407076606, | |
| "grad_norm": 0.33422502875328064, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 1.2115, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.20474063895568534, | |
| "grad_norm": 0.29944148659706116, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 1.0766, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.20611473720371007, | |
| "grad_norm": 0.36896276473999023, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 1.3203, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.20611473720371007, | |
| "eval_loss": 1.163150668144226, | |
| "eval_runtime": 34.2819, | |
| "eval_samples_per_second": 35.762, | |
| "eval_steps_per_second": 8.955, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.2074888354517348, | |
| "grad_norm": 0.1593729555606842, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.8128, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.20886293369975953, | |
| "grad_norm": 0.13594044744968414, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.7479, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.21023703194778426, | |
| "grad_norm": 0.16041527688503265, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.9017, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.211611130195809, | |
| "grad_norm": 0.15405145287513733, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.8641, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.21298522844383375, | |
| "grad_norm": 0.13940462470054626, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.7811, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.21435932669185848, | |
| "grad_norm": 0.16288061439990997, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.861, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.2157334249398832, | |
| "grad_norm": 0.1333891749382019, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.854, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.21710752318790794, | |
| "grad_norm": 0.17033836245536804, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.9393, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.21848162143593267, | |
| "grad_norm": 0.1751287430524826, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.8876, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.2198557196839574, | |
| "grad_norm": 0.1697334349155426, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.9032, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.22122981793198213, | |
| "grad_norm": 0.15173961222171783, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.846, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.22260391618000688, | |
| "grad_norm": 0.16722245514392853, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.9505, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.2239780144280316, | |
| "grad_norm": 0.15179596841335297, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.91, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.22535211267605634, | |
| "grad_norm": 0.1537434309720993, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.846, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.22672621092408107, | |
| "grad_norm": 0.17167864739894867, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.9822, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.2281003091721058, | |
| "grad_norm": 0.1635286509990692, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.9158, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.22947440742013053, | |
| "grad_norm": 0.1766817271709442, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 1.0387, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.23084850566815526, | |
| "grad_norm": 0.16663429141044617, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.9098, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.23222260391618002, | |
| "grad_norm": 0.1558423638343811, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.8829, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.23359670216420475, | |
| "grad_norm": 0.18819153308868408, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 1.0106, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.23497080041222948, | |
| "grad_norm": 0.17612503468990326, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.951, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.2363448986602542, | |
| "grad_norm": 0.18038241565227509, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.9117, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.23771899690827894, | |
| "grad_norm": 0.16612718999385834, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.8649, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.23909309515630367, | |
| "grad_norm": 0.17079417407512665, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.9076, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.2404671934043284, | |
| "grad_norm": 0.18941156566143036, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.9621, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.24184129165235313, | |
| "grad_norm": 0.20612381398677826, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.9983, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.2432153899003779, | |
| "grad_norm": 0.18256224691867828, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.9754, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.24458948814840262, | |
| "grad_norm": 0.2054128646850586, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.9444, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.24596358639642735, | |
| "grad_norm": 0.24424363672733307, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 1.018, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.24733768464445208, | |
| "grad_norm": 0.3375717103481293, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 1.4213, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2487117828924768, | |
| "grad_norm": 0.27337440848350525, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.8993, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.25008588114050156, | |
| "grad_norm": 1.3832429647445679, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 2.5111, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.25145997938852627, | |
| "grad_norm": 0.3790847659111023, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 1.0207, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.252834077636551, | |
| "grad_norm": 3.289069414138794, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 2.0101, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.2542081758845757, | |
| "grad_norm": 3.439802408218384, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 1.97, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.2555822741326005, | |
| "grad_norm": 3.5293004512786865, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 1.7579, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.25695637238062524, | |
| "grad_norm": 2.723883867263794, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 1.8916, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.25833047062864994, | |
| "grad_norm": 3.4014408588409424, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 1.715, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.2597045688766747, | |
| "grad_norm": 0.6042124032974243, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.8956, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.2610786671246994, | |
| "grad_norm": 3.4375836849212646, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 1.8971, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.26245276537272416, | |
| "grad_norm": 0.7264528274536133, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.969, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.26382686362074886, | |
| "grad_norm": 0.24482984840869904, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.861, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.2652009618687736, | |
| "grad_norm": 0.2977782189846039, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.8754, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.2665750601167983, | |
| "grad_norm": 0.2731274366378784, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.8221, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.2679491583648231, | |
| "grad_norm": 0.26564353704452515, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.905, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.26932325661284784, | |
| "grad_norm": 0.2909500002861023, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 1.0307, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.27069735486087254, | |
| "grad_norm": 0.23745307326316833, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.8961, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.2720714531088973, | |
| "grad_norm": 0.2782082259654999, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.8792, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.273445551356922, | |
| "grad_norm": 0.2878691554069519, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 1.1024, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.27481964960494676, | |
| "grad_norm": 0.4810084402561188, | |
| "learning_rate": 0.0, | |
| "loss": 1.4565, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.27481964960494676, | |
| "eval_loss": 1.154872179031372, | |
| "eval_runtime": 34.3051, | |
| "eval_samples_per_second": 35.738, | |
| "eval_steps_per_second": 8.949, | |
| "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": 2.500272152641536e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |