| { | |
| "best_metric": 0.00332257361151278, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.8629989212513485, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.004314994606256742, | |
| "grad_norm": 17.087665557861328, | |
| "learning_rate": 1e-05, | |
| "loss": 1.8617, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.004314994606256742, | |
| "eval_loss": 1.3578087091445923, | |
| "eval_runtime": 43.1351, | |
| "eval_samples_per_second": 9.041, | |
| "eval_steps_per_second": 2.272, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.008629989212513484, | |
| "grad_norm": 18.310680389404297, | |
| "learning_rate": 2e-05, | |
| "loss": 2.15, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.012944983818770227, | |
| "grad_norm": 11.765509605407715, | |
| "learning_rate": 3e-05, | |
| "loss": 1.918, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.017259978425026967, | |
| "grad_norm": 10.882978439331055, | |
| "learning_rate": 4e-05, | |
| "loss": 1.534, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.021574973031283712, | |
| "grad_norm": 10.474682807922363, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0628, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.025889967637540454, | |
| "grad_norm": 7.007547855377197, | |
| "learning_rate": 6e-05, | |
| "loss": 0.6529, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.030204962243797196, | |
| "grad_norm": 7.29658842086792, | |
| "learning_rate": 7e-05, | |
| "loss": 0.3454, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.034519956850053934, | |
| "grad_norm": 7.774313449859619, | |
| "learning_rate": 8e-05, | |
| "loss": 0.172, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.038834951456310676, | |
| "grad_norm": 11.443821907043457, | |
| "learning_rate": 9e-05, | |
| "loss": 0.1285, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.043149946062567425, | |
| "grad_norm": 3.870835542678833, | |
| "learning_rate": 0.0001, | |
| "loss": 0.105, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04746494066882417, | |
| "grad_norm": 1.3818610906600952, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.0662, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.05177993527508091, | |
| "grad_norm": 0.9391865730285645, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.0341, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.05609492988133765, | |
| "grad_norm": 1.3468683958053589, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.0455, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.06040992448759439, | |
| "grad_norm": 1.2473664283752441, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.0308, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.06472491909385113, | |
| "grad_norm": 0.7611331939697266, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.0273, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.06903991370010787, | |
| "grad_norm": 0.6905964612960815, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.0319, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.07335490830636461, | |
| "grad_norm": 0.4306313991546631, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.0312, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.07766990291262135, | |
| "grad_norm": 0.8068592548370361, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.0349, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.08198489751887811, | |
| "grad_norm": 1.119532823562622, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.0342, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.08629989212513485, | |
| "grad_norm": 0.9314156770706177, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.0281, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.09061488673139159, | |
| "grad_norm": 0.5867143273353577, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.0223, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.09492988133764833, | |
| "grad_norm": 0.7694201469421387, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.0233, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.09924487594390508, | |
| "grad_norm": 0.7751960754394531, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.0172, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.10355987055016182, | |
| "grad_norm": 0.907068133354187, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.0256, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.10787486515641856, | |
| "grad_norm": 0.8207224011421204, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.0326, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.1121898597626753, | |
| "grad_norm": 0.7676880359649658, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.0237, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.11650485436893204, | |
| "grad_norm": 1.1785573959350586, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.0349, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.12081984897518878, | |
| "grad_norm": 3.5154619216918945, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.0276, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.12513484358144553, | |
| "grad_norm": 0.45314833521842957, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.0196, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.12944983818770225, | |
| "grad_norm": 1.771968126296997, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.0488, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.133764832793959, | |
| "grad_norm": 0.781960666179657, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.0287, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.13807982740021574, | |
| "grad_norm": 5.066781997680664, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.0237, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.1423948220064725, | |
| "grad_norm": 1.486770749092102, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.0369, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.14670981661272922, | |
| "grad_norm": 0.8950640559196472, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.0209, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.15102481121898598, | |
| "grad_norm": 1.024499535560608, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.0264, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.1553398058252427, | |
| "grad_norm": 0.7848332524299622, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.0324, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.15965480043149946, | |
| "grad_norm": 0.42584487795829773, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.0179, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.16396979503775622, | |
| "grad_norm": 0.4489155411720276, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.0241, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.16828478964401294, | |
| "grad_norm": 0.8728815317153931, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.0219, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.1725997842502697, | |
| "grad_norm": 1.0815844535827637, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.0389, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.17691477885652643, | |
| "grad_norm": 0.3527948558330536, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.0188, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.18122977346278318, | |
| "grad_norm": 0.5202154517173767, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.0176, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.1855447680690399, | |
| "grad_norm": 0.6290869116783142, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.0288, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.18985976267529667, | |
| "grad_norm": 0.5054512619972229, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.018, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.1941747572815534, | |
| "grad_norm": 0.5555866956710815, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.0173, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.19848975188781015, | |
| "grad_norm": 0.9120898842811584, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.0316, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.20280474649406688, | |
| "grad_norm": 0.43381747603416443, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.0195, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.20711974110032363, | |
| "grad_norm": 0.5131010413169861, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.0214, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.21143473570658036, | |
| "grad_norm": 0.5879638195037842, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.0177, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.21574973031283712, | |
| "grad_norm": 0.6071250438690186, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.0214, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.21574973031283712, | |
| "eval_loss": 0.00678399158641696, | |
| "eval_runtime": 44.1541, | |
| "eval_samples_per_second": 8.833, | |
| "eval_steps_per_second": 2.219, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.22006472491909385, | |
| "grad_norm": 1.976332187652588, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.0361, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.2243797195253506, | |
| "grad_norm": 0.2932753264904022, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.0171, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.22869471413160733, | |
| "grad_norm": 0.31595897674560547, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.0195, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.23300970873786409, | |
| "grad_norm": 1.3101752996444702, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.0292, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.2373247033441208, | |
| "grad_norm": 0.7268949747085571, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.0143, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.24163969795037757, | |
| "grad_norm": 4.275149822235107, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.0417, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.2459546925566343, | |
| "grad_norm": 0.3323039710521698, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.0145, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.25026968716289105, | |
| "grad_norm": 0.3134349584579468, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.0159, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.2545846817691478, | |
| "grad_norm": 1.0623679161071777, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.0273, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.2588996763754045, | |
| "grad_norm": 1.1334818601608276, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.019, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.2632146709816613, | |
| "grad_norm": 0.45052817463874817, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.0226, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.267529665587918, | |
| "grad_norm": 0.6137033104896545, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.0212, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.27184466019417475, | |
| "grad_norm": 0.7195796370506287, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.0213, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.2761596548004315, | |
| "grad_norm": 0.5169023871421814, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.0222, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.28047464940668826, | |
| "grad_norm": 0.36940544843673706, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.0206, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.284789644012945, | |
| "grad_norm": 0.6449036598205566, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.0438, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.2891046386192017, | |
| "grad_norm": 1.0550373792648315, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.0297, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.29341963322545844, | |
| "grad_norm": 0.3580387830734253, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.0208, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.2977346278317152, | |
| "grad_norm": 0.4854778051376343, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.0218, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.30204962243797195, | |
| "grad_norm": 0.4402463436126709, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.0157, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.3063646170442287, | |
| "grad_norm": 0.3080613315105438, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.015, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.3106796116504854, | |
| "grad_norm": 0.3834276497364044, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.0122, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.3149946062567422, | |
| "grad_norm": 0.9351459741592407, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.0232, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.3193096008629989, | |
| "grad_norm": 0.8108393549919128, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.0276, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.32362459546925565, | |
| "grad_norm": 0.43026992678642273, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.0136, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.32793959007551243, | |
| "grad_norm": 0.4602266550064087, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.0175, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.33225458468176916, | |
| "grad_norm": 11.124612808227539, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.0284, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3365695792880259, | |
| "grad_norm": 6.074794769287109, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.028, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.3408845738942826, | |
| "grad_norm": 1.4700261354446411, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.0224, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.3451995685005394, | |
| "grad_norm": 17.13897132873535, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.0538, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.34951456310679613, | |
| "grad_norm": 0.4576086103916168, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.021, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.35382955771305286, | |
| "grad_norm": 0.3490538001060486, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.0213, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.3581445523193096, | |
| "grad_norm": 7.836808204650879, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.0245, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.36245954692556637, | |
| "grad_norm": 16.858186721801758, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.2472, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.3667745415318231, | |
| "grad_norm": 13.203956604003906, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.0769, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.3710895361380798, | |
| "grad_norm": 8.34621810913086, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.057, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.37540453074433655, | |
| "grad_norm": 0.9256253242492676, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.0226, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.37971952535059333, | |
| "grad_norm": 0.4759206473827362, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.0165, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.38403451995685006, | |
| "grad_norm": 0.8075918555259705, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.0152, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.3883495145631068, | |
| "grad_norm": 0.435330867767334, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.0122, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.3926645091693635, | |
| "grad_norm": 0.4027554988861084, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.0163, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.3969795037756203, | |
| "grad_norm": 0.7867634296417236, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.0268, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.40129449838187703, | |
| "grad_norm": 0.49145984649658203, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.0255, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.40560949298813376, | |
| "grad_norm": 0.4385581612586975, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.0172, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.4099244875943905, | |
| "grad_norm": 0.42079663276672363, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.0172, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.41423948220064727, | |
| "grad_norm": 0.37281689047813416, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.0163, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.418554476806904, | |
| "grad_norm": 0.5622067451477051, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.0149, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.4228694714131607, | |
| "grad_norm": 0.7449488639831543, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.015, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.42718446601941745, | |
| "grad_norm": 0.26745229959487915, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.0088, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.43149946062567424, | |
| "grad_norm": 0.23704928159713745, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.0098, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.43149946062567424, | |
| "eval_loss": 0.004985797218978405, | |
| "eval_runtime": 44.1566, | |
| "eval_samples_per_second": 8.832, | |
| "eval_steps_per_second": 2.219, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.43581445523193096, | |
| "grad_norm": 0.9159976243972778, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.0251, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.4401294498381877, | |
| "grad_norm": 0.8912190198898315, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.0209, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.4444444444444444, | |
| "grad_norm": 0.54429692029953, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.015, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.4487594390507012, | |
| "grad_norm": 0.24080169200897217, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.0134, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.45307443365695793, | |
| "grad_norm": 12.340768814086914, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0413, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.45738942826321466, | |
| "grad_norm": 0.1942518800497055, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.0107, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.4617044228694714, | |
| "grad_norm": 0.5876197814941406, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.0147, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.46601941747572817, | |
| "grad_norm": 0.1846633106470108, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.009, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.4703344120819849, | |
| "grad_norm": 0.22596366703510284, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.0127, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.4746494066882416, | |
| "grad_norm": 0.24374543130397797, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.0102, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.47896440129449835, | |
| "grad_norm": 0.29142627120018005, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.0142, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.48327939590075514, | |
| "grad_norm": 0.3160979449748993, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.0128, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.48759439050701187, | |
| "grad_norm": 0.5761608481407166, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.0311, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.4919093851132686, | |
| "grad_norm": 0.416759729385376, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.0173, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.4962243797195254, | |
| "grad_norm": 0.43046465516090393, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.0197, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.5005393743257821, | |
| "grad_norm": 0.4779050946235657, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.0172, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.5048543689320388, | |
| "grad_norm": 0.3801040053367615, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.0165, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.5091693635382956, | |
| "grad_norm": 0.24152928590774536, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.019, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.5134843581445523, | |
| "grad_norm": 0.295428067445755, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.0168, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.517799352750809, | |
| "grad_norm": 0.360342413187027, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.0206, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.5221143473570659, | |
| "grad_norm": 0.36108753085136414, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.0158, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.5264293419633226, | |
| "grad_norm": 0.41004496812820435, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.0209, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.5307443365695793, | |
| "grad_norm": 0.3360210657119751, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.0164, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.535059331175836, | |
| "grad_norm": 0.3026689291000366, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.0154, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.5393743257820928, | |
| "grad_norm": 0.35539960861206055, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.0164, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.5436893203883495, | |
| "grad_norm": 0.3963693678379059, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.0163, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.5480043149946062, | |
| "grad_norm": 0.2587314248085022, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.0126, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.552319309600863, | |
| "grad_norm": 0.4688221514225006, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.0128, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.5566343042071198, | |
| "grad_norm": 0.3882474899291992, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.0161, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.5609492988133765, | |
| "grad_norm": 0.4484683573246002, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.0144, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.5652642934196332, | |
| "grad_norm": 0.4020631015300751, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.0204, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.56957928802589, | |
| "grad_norm": 0.41258952021598816, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.0171, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.5738942826321467, | |
| "grad_norm": 0.32843923568725586, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.0171, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.5782092772384034, | |
| "grad_norm": 0.42162683606147766, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.013, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.5825242718446602, | |
| "grad_norm": 0.36203646659851074, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.0146, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.5868392664509169, | |
| "grad_norm": 0.22073198854923248, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.0081, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.5911542610571737, | |
| "grad_norm": 1.1045697927474976, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.0159, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.5954692556634305, | |
| "grad_norm": 0.31469646096229553, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.0133, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.5997842502696872, | |
| "grad_norm": 0.5070643424987793, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.0121, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.6040992448759439, | |
| "grad_norm": 0.6250373721122742, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.0152, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.6084142394822006, | |
| "grad_norm": 0.33579790592193604, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.0101, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.6127292340884574, | |
| "grad_norm": 0.3963364064693451, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.0139, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.6170442286947141, | |
| "grad_norm": 0.39243972301483154, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.0191, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.6213592233009708, | |
| "grad_norm": 0.6678121089935303, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.0151, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.6256742179072277, | |
| "grad_norm": 0.30046969652175903, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.0132, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.6299892125134844, | |
| "grad_norm": 0.42828816175460815, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.0127, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.6343042071197411, | |
| "grad_norm": 0.6971274614334106, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.0192, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.6386192017259978, | |
| "grad_norm": 0.4674164056777954, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.0105, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.6429341963322546, | |
| "grad_norm": 0.4478580355644226, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.0127, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.6472491909385113, | |
| "grad_norm": 0.3241710960865021, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.0166, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6472491909385113, | |
| "eval_loss": 0.003608308732509613, | |
| "eval_runtime": 43.5706, | |
| "eval_samples_per_second": 8.951, | |
| "eval_steps_per_second": 2.249, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.651564185544768, | |
| "grad_norm": 0.8593759536743164, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.0119, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.6558791801510249, | |
| "grad_norm": 0.2231779545545578, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.0073, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.6601941747572816, | |
| "grad_norm": 0.32959336042404175, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.0116, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.6645091693635383, | |
| "grad_norm": 0.2642689049243927, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.0164, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.668824163969795, | |
| "grad_norm": 0.3499794602394104, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.0164, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.6731391585760518, | |
| "grad_norm": 0.22071419656276703, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.0069, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.6774541531823085, | |
| "grad_norm": 0.29923346638679504, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.0074, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.6817691477885652, | |
| "grad_norm": 0.21840479969978333, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.0198, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.686084142394822, | |
| "grad_norm": 0.22816699743270874, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.0082, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.6903991370010788, | |
| "grad_norm": 0.20753133296966553, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.0115, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.6947141316073355, | |
| "grad_norm": 0.28496307134628296, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.0129, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.6990291262135923, | |
| "grad_norm": 0.25898557901382446, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.0131, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.703344120819849, | |
| "grad_norm": 0.1916905790567398, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.0125, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.7076591154261057, | |
| "grad_norm": 0.389384001493454, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.0139, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.7119741100323624, | |
| "grad_norm": 0.35845088958740234, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.0134, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.7162891046386192, | |
| "grad_norm": 0.7304808497428894, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.0205, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.7206040992448759, | |
| "grad_norm": 0.38205623626708984, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.015, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.7249190938511327, | |
| "grad_norm": 0.3484541177749634, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.0176, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.7292340884573895, | |
| "grad_norm": 0.43734773993492126, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.0136, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.7335490830636462, | |
| "grad_norm": 0.2967865765094757, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.0122, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.7378640776699029, | |
| "grad_norm": 0.3538125157356262, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.0146, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.7421790722761596, | |
| "grad_norm": 0.26428455114364624, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.0112, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.7464940668824164, | |
| "grad_norm": 0.4113234579563141, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.0122, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.7508090614886731, | |
| "grad_norm": 0.2964576482772827, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.0146, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.7551240560949298, | |
| "grad_norm": 0.32202237844467163, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.0111, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.7594390507011867, | |
| "grad_norm": 0.6488569974899292, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.0231, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.7637540453074434, | |
| "grad_norm": 0.3091130256652832, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.0134, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.7680690399137001, | |
| "grad_norm": 0.4929616451263428, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.0195, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.7723840345199569, | |
| "grad_norm": 0.3596843183040619, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.013, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.7766990291262136, | |
| "grad_norm": 0.2965730130672455, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.0122, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.7810140237324703, | |
| "grad_norm": 0.3788065016269684, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.0154, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.785329018338727, | |
| "grad_norm": 0.5955777168273926, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.0219, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.7896440129449838, | |
| "grad_norm": 0.42128297686576843, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.0151, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.7939590075512406, | |
| "grad_norm": 0.412579208612442, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.0179, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.7982740021574973, | |
| "grad_norm": 0.4108341634273529, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.0122, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.8025889967637541, | |
| "grad_norm": 0.39559030532836914, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.0168, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.8069039913700108, | |
| "grad_norm": 0.44083407521247864, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.017, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.8112189859762675, | |
| "grad_norm": 0.3507978618144989, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.0109, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.8155339805825242, | |
| "grad_norm": 0.43479645252227783, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.0133, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.819848975188781, | |
| "grad_norm": 0.4022519886493683, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.0144, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.8241639697950378, | |
| "grad_norm": 0.37215185165405273, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.0122, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.8284789644012945, | |
| "grad_norm": 0.46314895153045654, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.0191, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.8327939590075513, | |
| "grad_norm": 0.29280897974967957, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.0098, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.837108953613808, | |
| "grad_norm": 0.32389187812805176, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.0109, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.8414239482200647, | |
| "grad_norm": 0.8599640727043152, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.021, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.8457389428263214, | |
| "grad_norm": 0.482661634683609, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.0183, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.8500539374325782, | |
| "grad_norm": 0.338364839553833, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.0105, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.8543689320388349, | |
| "grad_norm": 0.2569652497768402, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.0047, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.8586839266450917, | |
| "grad_norm": 0.3186763525009155, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.0096, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.8629989212513485, | |
| "grad_norm": 0.47406089305877686, | |
| "learning_rate": 0.0, | |
| "loss": 0.0128, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.8629989212513485, | |
| "eval_loss": 0.00332257361151278, | |
| "eval_runtime": 43.5521, | |
| "eval_samples_per_second": 8.955, | |
| "eval_steps_per_second": 2.25, | |
| "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": 3.1446181247503565e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |