|
{ |
|
"best_metric": 1.3693137168884277, |
|
"best_model_checkpoint": "miner_id_24/checkpoint-200", |
|
"epoch": 0.4158004158004158, |
|
"eval_steps": 50, |
|
"global_step": 200, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.002079002079002079, |
|
"grad_norm": 29.60595703125, |
|
"learning_rate": 1e-05, |
|
"loss": 6.7852, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.002079002079002079, |
|
"eval_loss": 1.588921070098877, |
|
"eval_runtime": 16.4824, |
|
"eval_samples_per_second": 49.204, |
|
"eval_steps_per_second": 12.316, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.004158004158004158, |
|
"grad_norm": 25.005966186523438, |
|
"learning_rate": 2e-05, |
|
"loss": 6.8426, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.006237006237006237, |
|
"grad_norm": 27.206989288330078, |
|
"learning_rate": 3e-05, |
|
"loss": 7.2415, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.008316008316008316, |
|
"grad_norm": 30.3399658203125, |
|
"learning_rate": 4e-05, |
|
"loss": 6.8651, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.010395010395010396, |
|
"grad_norm": 27.907804489135742, |
|
"learning_rate": 5e-05, |
|
"loss": 6.755, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.012474012474012475, |
|
"grad_norm": 26.569618225097656, |
|
"learning_rate": 6e-05, |
|
"loss": 6.4593, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.014553014553014554, |
|
"grad_norm": 24.82715606689453, |
|
"learning_rate": 7e-05, |
|
"loss": 6.8985, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.016632016632016633, |
|
"grad_norm": 48.51078796386719, |
|
"learning_rate": 8e-05, |
|
"loss": 6.5583, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.018711018711018712, |
|
"grad_norm": 24.294008255004883, |
|
"learning_rate": 9e-05, |
|
"loss": 6.2934, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.02079002079002079, |
|
"grad_norm": 26.923362731933594, |
|
"learning_rate": 0.0001, |
|
"loss": 6.5698, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.02286902286902287, |
|
"grad_norm": 28.07476234436035, |
|
"learning_rate": 9.999316524962345e-05, |
|
"loss": 6.2198, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.02494802494802495, |
|
"grad_norm": 24.96772003173828, |
|
"learning_rate": 9.997266286704631e-05, |
|
"loss": 6.1341, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.02702702702702703, |
|
"grad_norm": 26.545703887939453, |
|
"learning_rate": 9.993849845741524e-05, |
|
"loss": 6.188, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.029106029106029108, |
|
"grad_norm": 26.985395431518555, |
|
"learning_rate": 9.989068136093873e-05, |
|
"loss": 5.6819, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.031185031185031187, |
|
"grad_norm": 25.727678298950195, |
|
"learning_rate": 9.98292246503335e-05, |
|
"loss": 5.7254, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.033264033264033266, |
|
"grad_norm": 28.322805404663086, |
|
"learning_rate": 9.975414512725057e-05, |
|
"loss": 5.6859, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.035343035343035345, |
|
"grad_norm": 30.68333625793457, |
|
"learning_rate": 9.966546331768191e-05, |
|
"loss": 5.5553, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.037422037422037424, |
|
"grad_norm": 28.860702514648438, |
|
"learning_rate": 9.956320346634876e-05, |
|
"loss": 5.595, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.0395010395010395, |
|
"grad_norm": 26.24447250366211, |
|
"learning_rate": 9.944739353007344e-05, |
|
"loss": 5.6486, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.04158004158004158, |
|
"grad_norm": 32.81097412109375, |
|
"learning_rate": 9.931806517013612e-05, |
|
"loss": 5.2332, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.04365904365904366, |
|
"grad_norm": 35.39635467529297, |
|
"learning_rate": 9.917525374361912e-05, |
|
"loss": 5.7035, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.04573804573804574, |
|
"grad_norm": 35.23345947265625, |
|
"learning_rate": 9.901899829374047e-05, |
|
"loss": 4.9686, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.04781704781704782, |
|
"grad_norm": 40.80004119873047, |
|
"learning_rate": 9.884934153917997e-05, |
|
"loss": 5.7427, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.0498960498960499, |
|
"grad_norm": 32.2061882019043, |
|
"learning_rate": 9.86663298624003e-05, |
|
"loss": 5.4809, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.05197505197505198, |
|
"grad_norm": 31.03780174255371, |
|
"learning_rate": 9.847001329696653e-05, |
|
"loss": 5.668, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.05405405405405406, |
|
"grad_norm": 27.15180206298828, |
|
"learning_rate": 9.826044551386744e-05, |
|
"loss": 5.355, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.056133056133056136, |
|
"grad_norm": 29.899612426757812, |
|
"learning_rate": 9.803768380684242e-05, |
|
"loss": 5.2276, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.058212058212058215, |
|
"grad_norm": 29.565778732299805, |
|
"learning_rate": 9.780178907671789e-05, |
|
"loss": 5.4713, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.060291060291060294, |
|
"grad_norm": 32.85282897949219, |
|
"learning_rate": 9.755282581475769e-05, |
|
"loss": 5.3408, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.062370062370062374, |
|
"grad_norm": 32.41181182861328, |
|
"learning_rate": 9.729086208503174e-05, |
|
"loss": 5.4768, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.06444906444906445, |
|
"grad_norm": 29.074604034423828, |
|
"learning_rate": 9.701596950580806e-05, |
|
"loss": 5.3076, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.06652806652806653, |
|
"grad_norm": 27.18608856201172, |
|
"learning_rate": 9.672822322997305e-05, |
|
"loss": 5.08, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.06860706860706861, |
|
"grad_norm": 27.19757080078125, |
|
"learning_rate": 9.642770192448536e-05, |
|
"loss": 5.7476, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.07068607068607069, |
|
"grad_norm": 36.73359298706055, |
|
"learning_rate": 9.611448774886924e-05, |
|
"loss": 5.6883, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.07276507276507277, |
|
"grad_norm": 30.20871925354004, |
|
"learning_rate": 9.578866633275288e-05, |
|
"loss": 5.085, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.07484407484407485, |
|
"grad_norm": 26.26553726196289, |
|
"learning_rate": 9.545032675245813e-05, |
|
"loss": 5.7193, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.07692307692307693, |
|
"grad_norm": 28.0688419342041, |
|
"learning_rate": 9.509956150664796e-05, |
|
"loss": 5.3497, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 0.079002079002079, |
|
"grad_norm": 26.292394638061523, |
|
"learning_rate": 9.473646649103818e-05, |
|
"loss": 5.5423, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 0.08108108108108109, |
|
"grad_norm": 24.79701042175293, |
|
"learning_rate": 9.43611409721806e-05, |
|
"loss": 5.6192, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 0.08316008316008316, |
|
"grad_norm": 34.27083969116211, |
|
"learning_rate": 9.397368756032445e-05, |
|
"loss": 5.8226, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.08523908523908524, |
|
"grad_norm": 27.875831604003906, |
|
"learning_rate": 9.357421218136386e-05, |
|
"loss": 5.3377, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 0.08731808731808732, |
|
"grad_norm": 26.744335174560547, |
|
"learning_rate": 9.316282404787871e-05, |
|
"loss": 5.4491, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 0.0893970893970894, |
|
"grad_norm": 54.93026351928711, |
|
"learning_rate": 9.273963562927695e-05, |
|
"loss": 5.5718, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 0.09147609147609148, |
|
"grad_norm": 23.76496696472168, |
|
"learning_rate": 9.230476262104677e-05, |
|
"loss": 5.9858, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 0.09355509355509356, |
|
"grad_norm": 25.368408203125, |
|
"learning_rate": 9.185832391312644e-05, |
|
"loss": 5.8774, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.09563409563409564, |
|
"grad_norm": 27.482393264770508, |
|
"learning_rate": 9.140044155740101e-05, |
|
"loss": 6.3263, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 0.09771309771309772, |
|
"grad_norm": 31.6037654876709, |
|
"learning_rate": 9.093124073433463e-05, |
|
"loss": 6.2651, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 0.0997920997920998, |
|
"grad_norm": 26.929533004760742, |
|
"learning_rate": 9.045084971874738e-05, |
|
"loss": 6.3211, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 0.10187110187110188, |
|
"grad_norm": 24.196531295776367, |
|
"learning_rate": 8.995939984474624e-05, |
|
"loss": 6.4937, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 0.10395010395010396, |
|
"grad_norm": 25.627565383911133, |
|
"learning_rate": 8.945702546981969e-05, |
|
"loss": 5.5208, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.10395010395010396, |
|
"eval_loss": 1.4520061016082764, |
|
"eval_runtime": 16.5108, |
|
"eval_samples_per_second": 49.119, |
|
"eval_steps_per_second": 12.295, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.10602910602910603, |
|
"grad_norm": 30.188610076904297, |
|
"learning_rate": 8.894386393810563e-05, |
|
"loss": 6.7074, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 0.10810810810810811, |
|
"grad_norm": 28.066347122192383, |
|
"learning_rate": 8.842005554284296e-05, |
|
"loss": 6.2885, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 0.1101871101871102, |
|
"grad_norm": 30.150691986083984, |
|
"learning_rate": 8.788574348801675e-05, |
|
"loss": 6.4745, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 0.11226611226611227, |
|
"grad_norm": 27.11690330505371, |
|
"learning_rate": 8.73410738492077e-05, |
|
"loss": 6.6094, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 0.11434511434511435, |
|
"grad_norm": 29.3702392578125, |
|
"learning_rate": 8.678619553365659e-05, |
|
"loss": 6.366, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.11642411642411643, |
|
"grad_norm": 27.218273162841797, |
|
"learning_rate": 8.622126023955446e-05, |
|
"loss": 6.4681, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 0.11850311850311851, |
|
"grad_norm": 26.06937026977539, |
|
"learning_rate": 8.564642241456986e-05, |
|
"loss": 6.2893, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 0.12058212058212059, |
|
"grad_norm": 24.98716926574707, |
|
"learning_rate": 8.506183921362443e-05, |
|
"loss": 6.3051, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 0.12266112266112267, |
|
"grad_norm": 24.424068450927734, |
|
"learning_rate": 8.44676704559283e-05, |
|
"loss": 6.3688, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 0.12474012474012475, |
|
"grad_norm": 22.242055892944336, |
|
"learning_rate": 8.386407858128706e-05, |
|
"loss": 6.1212, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.12681912681912683, |
|
"grad_norm": 25.234708786010742, |
|
"learning_rate": 8.32512286056924e-05, |
|
"loss": 5.9599, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 0.1288981288981289, |
|
"grad_norm": 24.131328582763672, |
|
"learning_rate": 8.262928807620843e-05, |
|
"loss": 5.6572, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 0.13097713097713098, |
|
"grad_norm": 20.404661178588867, |
|
"learning_rate": 8.199842702516583e-05, |
|
"loss": 5.9329, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 0.13305613305613306, |
|
"grad_norm": 22.83519744873047, |
|
"learning_rate": 8.135881792367686e-05, |
|
"loss": 5.6288, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 0.13513513513513514, |
|
"grad_norm": 24.050945281982422, |
|
"learning_rate": 8.07106356344834e-05, |
|
"loss": 5.4463, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.13721413721413722, |
|
"grad_norm": 21.087100982666016, |
|
"learning_rate": 8.005405736415126e-05, |
|
"loss": 5.3214, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 0.1392931392931393, |
|
"grad_norm": 22.5960693359375, |
|
"learning_rate": 7.938926261462366e-05, |
|
"loss": 4.9233, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 0.14137214137214138, |
|
"grad_norm": 23.592405319213867, |
|
"learning_rate": 7.871643313414718e-05, |
|
"loss": 5.3317, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 0.14345114345114346, |
|
"grad_norm": 18.197025299072266, |
|
"learning_rate": 7.803575286758364e-05, |
|
"loss": 5.4483, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 0.14553014553014554, |
|
"grad_norm": 26.235504150390625, |
|
"learning_rate": 7.734740790612136e-05, |
|
"loss": 5.3065, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.14760914760914762, |
|
"grad_norm": 21.71451759338379, |
|
"learning_rate": 7.66515864363997e-05, |
|
"loss": 4.974, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 0.1496881496881497, |
|
"grad_norm": 29.15395736694336, |
|
"learning_rate": 7.594847868906076e-05, |
|
"loss": 5.3881, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 0.15176715176715178, |
|
"grad_norm": 22.060916900634766, |
|
"learning_rate": 7.52382768867422e-05, |
|
"loss": 5.297, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 0.15384615384615385, |
|
"grad_norm": 22.163185119628906, |
|
"learning_rate": 7.452117519152542e-05, |
|
"loss": 5.5406, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 0.15592515592515593, |
|
"grad_norm": 17.246747970581055, |
|
"learning_rate": 7.379736965185368e-05, |
|
"loss": 5.0765, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.158004158004158, |
|
"grad_norm": 16.834630966186523, |
|
"learning_rate": 7.30670581489344e-05, |
|
"loss": 4.6125, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 0.1600831600831601, |
|
"grad_norm": 18.680545806884766, |
|
"learning_rate": 7.233044034264034e-05, |
|
"loss": 5.0347, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 0.16216216216216217, |
|
"grad_norm": 17.589662551879883, |
|
"learning_rate": 7.158771761692464e-05, |
|
"loss": 5.1071, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 0.16424116424116425, |
|
"grad_norm": 16.554645538330078, |
|
"learning_rate": 7.083909302476453e-05, |
|
"loss": 4.8337, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 0.16632016632016633, |
|
"grad_norm": 19.671749114990234, |
|
"learning_rate": 7.008477123264848e-05, |
|
"loss": 4.9519, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.1683991683991684, |
|
"grad_norm": 21.42405891418457, |
|
"learning_rate": 6.932495846462261e-05, |
|
"loss": 5.3256, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 0.1704781704781705, |
|
"grad_norm": 15.732622146606445, |
|
"learning_rate": 6.855986244591104e-05, |
|
"loss": 5.0526, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 0.17255717255717257, |
|
"grad_norm": 17.768220901489258, |
|
"learning_rate": 6.778969234612584e-05, |
|
"loss": 4.8137, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 0.17463617463617465, |
|
"grad_norm": 18.31966781616211, |
|
"learning_rate": 6.701465872208216e-05, |
|
"loss": 5.3164, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 0.17671517671517672, |
|
"grad_norm": 18.258047103881836, |
|
"learning_rate": 6.623497346023418e-05, |
|
"loss": 5.3665, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.1787941787941788, |
|
"grad_norm": 18.64510154724121, |
|
"learning_rate": 6.545084971874738e-05, |
|
"loss": 5.5286, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 0.18087318087318088, |
|
"grad_norm": 17.050338745117188, |
|
"learning_rate": 6.466250186922325e-05, |
|
"loss": 4.9467, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 0.18295218295218296, |
|
"grad_norm": 18.383134841918945, |
|
"learning_rate": 6.387014543809223e-05, |
|
"loss": 5.1102, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 0.18503118503118504, |
|
"grad_norm": 24.1225643157959, |
|
"learning_rate": 6.307399704769099e-05, |
|
"loss": 5.6518, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 0.18711018711018712, |
|
"grad_norm": 23.475433349609375, |
|
"learning_rate": 6.227427435703997e-05, |
|
"loss": 5.5404, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.1891891891891892, |
|
"grad_norm": 24.808155059814453, |
|
"learning_rate": 6.147119600233758e-05, |
|
"loss": 5.7452, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 0.19126819126819128, |
|
"grad_norm": 32.905982971191406, |
|
"learning_rate": 6.066498153718735e-05, |
|
"loss": 5.891, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 0.19334719334719336, |
|
"grad_norm": 25.104476928710938, |
|
"learning_rate": 5.985585137257401e-05, |
|
"loss": 6.0769, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 0.19542619542619544, |
|
"grad_norm": 27.174358367919922, |
|
"learning_rate": 5.90440267166055e-05, |
|
"loss": 5.5983, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 0.19750519750519752, |
|
"grad_norm": 26.053733825683594, |
|
"learning_rate": 5.8229729514036705e-05, |
|
"loss": 6.0037, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.1995841995841996, |
|
"grad_norm": 27.530136108398438, |
|
"learning_rate": 5.74131823855921e-05, |
|
"loss": 6.5557, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 0.20166320166320167, |
|
"grad_norm": 26.11812973022461, |
|
"learning_rate": 5.6594608567103456e-05, |
|
"loss": 6.1534, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 0.20374220374220375, |
|
"grad_norm": 25.85369110107422, |
|
"learning_rate": 5.577423184847932e-05, |
|
"loss": 6.6225, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 0.20582120582120583, |
|
"grad_norm": 30.74239730834961, |
|
"learning_rate": 5.495227651252315e-05, |
|
"loss": 6.4503, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 0.2079002079002079, |
|
"grad_norm": 26.553922653198242, |
|
"learning_rate": 5.4128967273616625e-05, |
|
"loss": 5.3192, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.2079002079002079, |
|
"eval_loss": 1.4002317190170288, |
|
"eval_runtime": 16.5443, |
|
"eval_samples_per_second": 49.02, |
|
"eval_steps_per_second": 12.27, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.20997920997921, |
|
"grad_norm": 33.29289627075195, |
|
"learning_rate": 5.330452921628497e-05, |
|
"loss": 6.0499, |
|
"step": 101 |
|
}, |
|
{ |
|
"epoch": 0.21205821205821207, |
|
"grad_norm": 56.924251556396484, |
|
"learning_rate": 5.247918773366112e-05, |
|
"loss": 6.2076, |
|
"step": 102 |
|
}, |
|
{ |
|
"epoch": 0.21413721413721415, |
|
"grad_norm": 33.388427734375, |
|
"learning_rate": 5.165316846586541e-05, |
|
"loss": 6.3714, |
|
"step": 103 |
|
}, |
|
{ |
|
"epoch": 0.21621621621621623, |
|
"grad_norm": 31.86151885986328, |
|
"learning_rate": 5.0826697238317935e-05, |
|
"loss": 6.2998, |
|
"step": 104 |
|
}, |
|
{ |
|
"epoch": 0.2182952182952183, |
|
"grad_norm": 45.58109664916992, |
|
"learning_rate": 5e-05, |
|
"loss": 6.1464, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.2203742203742204, |
|
"grad_norm": 36.673439025878906, |
|
"learning_rate": 4.917330276168208e-05, |
|
"loss": 6.4994, |
|
"step": 106 |
|
}, |
|
{ |
|
"epoch": 0.22245322245322247, |
|
"grad_norm": 46.388973236083984, |
|
"learning_rate": 4.834683153413459e-05, |
|
"loss": 5.9618, |
|
"step": 107 |
|
}, |
|
{ |
|
"epoch": 0.22453222453222454, |
|
"grad_norm": 35.77516174316406, |
|
"learning_rate": 4.7520812266338885e-05, |
|
"loss": 6.0713, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 0.22661122661122662, |
|
"grad_norm": 42.96281051635742, |
|
"learning_rate": 4.669547078371504e-05, |
|
"loss": 5.863, |
|
"step": 109 |
|
}, |
|
{ |
|
"epoch": 0.2286902286902287, |
|
"grad_norm": 37.156272888183594, |
|
"learning_rate": 4.5871032726383386e-05, |
|
"loss": 5.7125, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.23076923076923078, |
|
"grad_norm": 42.857574462890625, |
|
"learning_rate": 4.504772348747687e-05, |
|
"loss": 5.9667, |
|
"step": 111 |
|
}, |
|
{ |
|
"epoch": 0.23284823284823286, |
|
"grad_norm": 44.69389724731445, |
|
"learning_rate": 4.4225768151520694e-05, |
|
"loss": 5.9149, |
|
"step": 112 |
|
}, |
|
{ |
|
"epoch": 0.23492723492723494, |
|
"grad_norm": 48.71627426147461, |
|
"learning_rate": 4.3405391432896555e-05, |
|
"loss": 5.4195, |
|
"step": 113 |
|
}, |
|
{ |
|
"epoch": 0.23700623700623702, |
|
"grad_norm": 32.36886978149414, |
|
"learning_rate": 4.2586817614407895e-05, |
|
"loss": 5.2663, |
|
"step": 114 |
|
}, |
|
{ |
|
"epoch": 0.2390852390852391, |
|
"grad_norm": 31.383926391601562, |
|
"learning_rate": 4.17702704859633e-05, |
|
"loss": 5.3309, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.24116424116424118, |
|
"grad_norm": 29.009599685668945, |
|
"learning_rate": 4.095597328339452e-05, |
|
"loss": 5.1069, |
|
"step": 116 |
|
}, |
|
{ |
|
"epoch": 0.24324324324324326, |
|
"grad_norm": 35.83613967895508, |
|
"learning_rate": 4.0144148627425993e-05, |
|
"loss": 5.4003, |
|
"step": 117 |
|
}, |
|
{ |
|
"epoch": 0.24532224532224534, |
|
"grad_norm": 32.853843688964844, |
|
"learning_rate": 3.933501846281267e-05, |
|
"loss": 5.1017, |
|
"step": 118 |
|
}, |
|
{ |
|
"epoch": 0.24740124740124741, |
|
"grad_norm": 33.55854797363281, |
|
"learning_rate": 3.852880399766243e-05, |
|
"loss": 5.3952, |
|
"step": 119 |
|
}, |
|
{ |
|
"epoch": 0.2494802494802495, |
|
"grad_norm": 36.20463180541992, |
|
"learning_rate": 3.772572564296005e-05, |
|
"loss": 5.0288, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.2515592515592516, |
|
"grad_norm": 27.810020446777344, |
|
"learning_rate": 3.6926002952309016e-05, |
|
"loss": 5.2437, |
|
"step": 121 |
|
}, |
|
{ |
|
"epoch": 0.25363825363825365, |
|
"grad_norm": 28.593204498291016, |
|
"learning_rate": 3.612985456190778e-05, |
|
"loss": 4.9836, |
|
"step": 122 |
|
}, |
|
{ |
|
"epoch": 0.25571725571725573, |
|
"grad_norm": 29.15188980102539, |
|
"learning_rate": 3.533749813077677e-05, |
|
"loss": 5.4529, |
|
"step": 123 |
|
}, |
|
{ |
|
"epoch": 0.2577962577962578, |
|
"grad_norm": 27.521888732910156, |
|
"learning_rate": 3.4549150281252636e-05, |
|
"loss": 5.1731, |
|
"step": 124 |
|
}, |
|
{ |
|
"epoch": 0.2598752598752599, |
|
"grad_norm": 38.01824188232422, |
|
"learning_rate": 3.3765026539765834e-05, |
|
"loss": 5.1737, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.26195426195426197, |
|
"grad_norm": 27.05405044555664, |
|
"learning_rate": 3.298534127791785e-05, |
|
"loss": 5.4588, |
|
"step": 126 |
|
}, |
|
{ |
|
"epoch": 0.26403326403326405, |
|
"grad_norm": 32.03307342529297, |
|
"learning_rate": 3.221030765387417e-05, |
|
"loss": 5.1365, |
|
"step": 127 |
|
}, |
|
{ |
|
"epoch": 0.2661122661122661, |
|
"grad_norm": 26.123497009277344, |
|
"learning_rate": 3.144013755408895e-05, |
|
"loss": 4.7003, |
|
"step": 128 |
|
}, |
|
{ |
|
"epoch": 0.2681912681912682, |
|
"grad_norm": 26.265361785888672, |
|
"learning_rate": 3.0675041535377405e-05, |
|
"loss": 4.8316, |
|
"step": 129 |
|
}, |
|
{ |
|
"epoch": 0.2702702702702703, |
|
"grad_norm": 27.146703720092773, |
|
"learning_rate": 2.991522876735154e-05, |
|
"loss": 5.1508, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.27234927234927236, |
|
"grad_norm": 26.394529342651367, |
|
"learning_rate": 2.916090697523549e-05, |
|
"loss": 5.0902, |
|
"step": 131 |
|
}, |
|
{ |
|
"epoch": 0.27442827442827444, |
|
"grad_norm": 30.819000244140625, |
|
"learning_rate": 2.8412282383075363e-05, |
|
"loss": 4.9659, |
|
"step": 132 |
|
}, |
|
{ |
|
"epoch": 0.2765072765072765, |
|
"grad_norm": 36.68458557128906, |
|
"learning_rate": 2.766955965735968e-05, |
|
"loss": 5.0427, |
|
"step": 133 |
|
}, |
|
{ |
|
"epoch": 0.2785862785862786, |
|
"grad_norm": 32.3036003112793, |
|
"learning_rate": 2.693294185106562e-05, |
|
"loss": 5.4486, |
|
"step": 134 |
|
}, |
|
{ |
|
"epoch": 0.2806652806652807, |
|
"grad_norm": 27.863311767578125, |
|
"learning_rate": 2.6202630348146324e-05, |
|
"loss": 5.2872, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 0.28274428274428276, |
|
"grad_norm": 29.791475296020508, |
|
"learning_rate": 2.547882480847461e-05, |
|
"loss": 5.1583, |
|
"step": 136 |
|
}, |
|
{ |
|
"epoch": 0.28482328482328484, |
|
"grad_norm": 28.367292404174805, |
|
"learning_rate": 2.476172311325783e-05, |
|
"loss": 4.9147, |
|
"step": 137 |
|
}, |
|
{ |
|
"epoch": 0.2869022869022869, |
|
"grad_norm": 28.982662200927734, |
|
"learning_rate": 2.405152131093926e-05, |
|
"loss": 5.4158, |
|
"step": 138 |
|
}, |
|
{ |
|
"epoch": 0.288981288981289, |
|
"grad_norm": 30.221160888671875, |
|
"learning_rate": 2.3348413563600325e-05, |
|
"loss": 4.9438, |
|
"step": 139 |
|
}, |
|
{ |
|
"epoch": 0.2910602910602911, |
|
"grad_norm": 26.62199592590332, |
|
"learning_rate": 2.2652592093878666e-05, |
|
"loss": 5.0532, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.29313929313929316, |
|
"grad_norm": 26.665769577026367, |
|
"learning_rate": 2.196424713241637e-05, |
|
"loss": 5.1416, |
|
"step": 141 |
|
}, |
|
{ |
|
"epoch": 0.29521829521829523, |
|
"grad_norm": 25.941627502441406, |
|
"learning_rate": 2.128356686585282e-05, |
|
"loss": 5.7407, |
|
"step": 142 |
|
}, |
|
{ |
|
"epoch": 0.2972972972972973, |
|
"grad_norm": 29.765132904052734, |
|
"learning_rate": 2.061073738537635e-05, |
|
"loss": 5.7022, |
|
"step": 143 |
|
}, |
|
{ |
|
"epoch": 0.2993762993762994, |
|
"grad_norm": 28.01789093017578, |
|
"learning_rate": 1.9945942635848748e-05, |
|
"loss": 5.4899, |
|
"step": 144 |
|
}, |
|
{ |
|
"epoch": 0.30145530145530147, |
|
"grad_norm": 29.412342071533203, |
|
"learning_rate": 1.928936436551661e-05, |
|
"loss": 6.1604, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 0.30353430353430355, |
|
"grad_norm": 30.55994415283203, |
|
"learning_rate": 1.8641182076323148e-05, |
|
"loss": 6.4566, |
|
"step": 146 |
|
}, |
|
{ |
|
"epoch": 0.30561330561330563, |
|
"grad_norm": 28.54621124267578, |
|
"learning_rate": 1.800157297483417e-05, |
|
"loss": 5.8721, |
|
"step": 147 |
|
}, |
|
{ |
|
"epoch": 0.3076923076923077, |
|
"grad_norm": 29.459003448486328, |
|
"learning_rate": 1.7370711923791567e-05, |
|
"loss": 5.9285, |
|
"step": 148 |
|
}, |
|
{ |
|
"epoch": 0.3097713097713098, |
|
"grad_norm": 29.382354736328125, |
|
"learning_rate": 1.6748771394307585e-05, |
|
"loss": 6.6228, |
|
"step": 149 |
|
}, |
|
{ |
|
"epoch": 0.31185031185031187, |
|
"grad_norm": 35.232913970947266, |
|
"learning_rate": 1.6135921418712956e-05, |
|
"loss": 5.4351, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.31185031185031187, |
|
"eval_loss": 1.3784149885177612, |
|
"eval_runtime": 16.5592, |
|
"eval_samples_per_second": 48.976, |
|
"eval_steps_per_second": 12.259, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.31392931392931395, |
|
"grad_norm": 36.84860610961914, |
|
"learning_rate": 1.553232954407171e-05, |
|
"loss": 6.2217, |
|
"step": 151 |
|
}, |
|
{ |
|
"epoch": 0.316008316008316, |
|
"grad_norm": 50.49912643432617, |
|
"learning_rate": 1.4938160786375572e-05, |
|
"loss": 5.8683, |
|
"step": 152 |
|
}, |
|
{ |
|
"epoch": 0.3180873180873181, |
|
"grad_norm": 57.88787078857422, |
|
"learning_rate": 1.435357758543015e-05, |
|
"loss": 6.1006, |
|
"step": 153 |
|
}, |
|
{ |
|
"epoch": 0.3201663201663202, |
|
"grad_norm": 42.55754470825195, |
|
"learning_rate": 1.3778739760445552e-05, |
|
"loss": 6.06, |
|
"step": 154 |
|
}, |
|
{ |
|
"epoch": 0.32224532224532226, |
|
"grad_norm": 41.394317626953125, |
|
"learning_rate": 1.3213804466343421e-05, |
|
"loss": 6.0936, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 0.32432432432432434, |
|
"grad_norm": 32.322364807128906, |
|
"learning_rate": 1.2658926150792322e-05, |
|
"loss": 6.4178, |
|
"step": 156 |
|
}, |
|
{ |
|
"epoch": 0.3264033264033264, |
|
"grad_norm": 41.43376159667969, |
|
"learning_rate": 1.2114256511983274e-05, |
|
"loss": 6.0143, |
|
"step": 157 |
|
}, |
|
{ |
|
"epoch": 0.3284823284823285, |
|
"grad_norm": 37.7407341003418, |
|
"learning_rate": 1.157994445715706e-05, |
|
"loss": 5.7424, |
|
"step": 158 |
|
}, |
|
{ |
|
"epoch": 0.3305613305613306, |
|
"grad_norm": 29.474939346313477, |
|
"learning_rate": 1.1056136061894384e-05, |
|
"loss": 5.8458, |
|
"step": 159 |
|
}, |
|
{ |
|
"epoch": 0.33264033264033266, |
|
"grad_norm": 25.122840881347656, |
|
"learning_rate": 1.0542974530180327e-05, |
|
"loss": 6.1322, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.33471933471933474, |
|
"grad_norm": 27.80811882019043, |
|
"learning_rate": 1.0040600155253765e-05, |
|
"loss": 5.5908, |
|
"step": 161 |
|
}, |
|
{ |
|
"epoch": 0.3367983367983368, |
|
"grad_norm": 28.213783264160156, |
|
"learning_rate": 9.549150281252633e-06, |
|
"loss": 5.8159, |
|
"step": 162 |
|
}, |
|
{ |
|
"epoch": 0.3388773388773389, |
|
"grad_norm": 27.08670425415039, |
|
"learning_rate": 9.068759265665384e-06, |
|
"loss": 5.3409, |
|
"step": 163 |
|
}, |
|
{ |
|
"epoch": 0.340956340956341, |
|
"grad_norm": 24.957714080810547, |
|
"learning_rate": 8.599558442598998e-06, |
|
"loss": 5.1652, |
|
"step": 164 |
|
}, |
|
{ |
|
"epoch": 0.34303534303534305, |
|
"grad_norm": 24.598251342773438, |
|
"learning_rate": 8.141676086873572e-06, |
|
"loss": 5.4767, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 0.34511434511434513, |
|
"grad_norm": 25.575883865356445, |
|
"learning_rate": 7.695237378953223e-06, |
|
"loss": 5.4212, |
|
"step": 166 |
|
}, |
|
{ |
|
"epoch": 0.3471933471933472, |
|
"grad_norm": 25.705076217651367, |
|
"learning_rate": 7.260364370723044e-06, |
|
"loss": 5.0383, |
|
"step": 167 |
|
}, |
|
{ |
|
"epoch": 0.3492723492723493, |
|
"grad_norm": 26.083740234375, |
|
"learning_rate": 6.837175952121306e-06, |
|
"loss": 5.3445, |
|
"step": 168 |
|
}, |
|
{ |
|
"epoch": 0.35135135135135137, |
|
"grad_norm": 25.030664443969727, |
|
"learning_rate": 6.425787818636131e-06, |
|
"loss": 5.1529, |
|
"step": 169 |
|
}, |
|
{ |
|
"epoch": 0.35343035343035345, |
|
"grad_norm": 24.932870864868164, |
|
"learning_rate": 6.026312439675552e-06, |
|
"loss": 4.6416, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.35550935550935553, |
|
"grad_norm": 27.457372665405273, |
|
"learning_rate": 5.6388590278194096e-06, |
|
"loss": 5.1129, |
|
"step": 171 |
|
}, |
|
{ |
|
"epoch": 0.3575883575883576, |
|
"grad_norm": 27.58135414123535, |
|
"learning_rate": 5.263533508961827e-06, |
|
"loss": 4.8054, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 0.3596673596673597, |
|
"grad_norm": 25.648788452148438, |
|
"learning_rate": 4.900438493352055e-06, |
|
"loss": 4.8878, |
|
"step": 173 |
|
}, |
|
{ |
|
"epoch": 0.36174636174636177, |
|
"grad_norm": 24.551259994506836, |
|
"learning_rate": 4.549673247541875e-06, |
|
"loss": 4.9167, |
|
"step": 174 |
|
}, |
|
{ |
|
"epoch": 0.36382536382536385, |
|
"grad_norm": 32.0989875793457, |
|
"learning_rate": 4.2113336672471245e-06, |
|
"loss": 4.7836, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.3659043659043659, |
|
"grad_norm": 31.729732513427734, |
|
"learning_rate": 3.885512251130763e-06, |
|
"loss": 5.5952, |
|
"step": 176 |
|
}, |
|
{ |
|
"epoch": 0.367983367983368, |
|
"grad_norm": 28.27291488647461, |
|
"learning_rate": 3.5722980755146517e-06, |
|
"loss": 5.1343, |
|
"step": 177 |
|
}, |
|
{ |
|
"epoch": 0.3700623700623701, |
|
"grad_norm": 27.010332107543945, |
|
"learning_rate": 3.271776770026963e-06, |
|
"loss": 5.1042, |
|
"step": 178 |
|
}, |
|
{ |
|
"epoch": 0.37214137214137216, |
|
"grad_norm": 26.788328170776367, |
|
"learning_rate": 2.9840304941919415e-06, |
|
"loss": 4.9696, |
|
"step": 179 |
|
}, |
|
{ |
|
"epoch": 0.37422037422037424, |
|
"grad_norm": 23.08694839477539, |
|
"learning_rate": 2.7091379149682685e-06, |
|
"loss": 4.9501, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.3762993762993763, |
|
"grad_norm": 20.737171173095703, |
|
"learning_rate": 2.4471741852423237e-06, |
|
"loss": 5.0214, |
|
"step": 181 |
|
}, |
|
{ |
|
"epoch": 0.3783783783783784, |
|
"grad_norm": 25.595474243164062, |
|
"learning_rate": 2.1982109232821178e-06, |
|
"loss": 4.8622, |
|
"step": 182 |
|
}, |
|
{ |
|
"epoch": 0.3804573804573805, |
|
"grad_norm": 25.813657760620117, |
|
"learning_rate": 1.962316193157593e-06, |
|
"loss": 5.1404, |
|
"step": 183 |
|
}, |
|
{ |
|
"epoch": 0.38253638253638256, |
|
"grad_norm": 31.10436248779297, |
|
"learning_rate": 1.7395544861325718e-06, |
|
"loss": 5.0456, |
|
"step": 184 |
|
}, |
|
{ |
|
"epoch": 0.38461538461538464, |
|
"grad_norm": 30.64323616027832, |
|
"learning_rate": 1.5299867030334814e-06, |
|
"loss": 4.9777, |
|
"step": 185 |
|
}, |
|
{ |
|
"epoch": 0.3866943866943867, |
|
"grad_norm": 28.30755615234375, |
|
"learning_rate": 1.333670137599713e-06, |
|
"loss": 5.0554, |
|
"step": 186 |
|
}, |
|
{ |
|
"epoch": 0.3887733887733888, |
|
"grad_norm": 29.582347869873047, |
|
"learning_rate": 1.1506584608200367e-06, |
|
"loss": 5.2584, |
|
"step": 187 |
|
}, |
|
{ |
|
"epoch": 0.3908523908523909, |
|
"grad_norm": 25.09845542907715, |
|
"learning_rate": 9.810017062595322e-07, |
|
"loss": 4.6825, |
|
"step": 188 |
|
}, |
|
{ |
|
"epoch": 0.39293139293139295, |
|
"grad_norm": 25.186092376708984, |
|
"learning_rate": 8.247462563808817e-07, |
|
"loss": 5.3549, |
|
"step": 189 |
|
}, |
|
{ |
|
"epoch": 0.39501039501039503, |
|
"grad_norm": 27.82665252685547, |
|
"learning_rate": 6.819348298638839e-07, |
|
"loss": 5.6513, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.3970893970893971, |
|
"grad_norm": 26.3640193939209, |
|
"learning_rate": 5.526064699265753e-07, |
|
"loss": 5.1095, |
|
"step": 191 |
|
}, |
|
{ |
|
"epoch": 0.3991683991683992, |
|
"grad_norm": 25.616100311279297, |
|
"learning_rate": 4.367965336512403e-07, |
|
"loss": 5.5777, |
|
"step": 192 |
|
}, |
|
{ |
|
"epoch": 0.40124740124740127, |
|
"grad_norm": 26.20846176147461, |
|
"learning_rate": 3.3453668231809286e-07, |
|
"loss": 5.6722, |
|
"step": 193 |
|
}, |
|
{ |
|
"epoch": 0.40332640332640335, |
|
"grad_norm": 26.799989700317383, |
|
"learning_rate": 2.458548727494292e-07, |
|
"loss": 5.9049, |
|
"step": 194 |
|
}, |
|
{ |
|
"epoch": 0.40540540540540543, |
|
"grad_norm": 24.2194766998291, |
|
"learning_rate": 1.7077534966650766e-07, |
|
"loss": 5.6891, |
|
"step": 195 |
|
}, |
|
{ |
|
"epoch": 0.4074844074844075, |
|
"grad_norm": 26.87697982788086, |
|
"learning_rate": 1.0931863906127327e-07, |
|
"loss": 6.2231, |
|
"step": 196 |
|
}, |
|
{ |
|
"epoch": 0.4095634095634096, |
|
"grad_norm": 24.078659057617188, |
|
"learning_rate": 6.150154258476315e-08, |
|
"loss": 6.0784, |
|
"step": 197 |
|
}, |
|
{ |
|
"epoch": 0.41164241164241167, |
|
"grad_norm": 33.70384216308594, |
|
"learning_rate": 2.7337132953697554e-08, |
|
"loss": 6.1821, |
|
"step": 198 |
|
}, |
|
{ |
|
"epoch": 0.41372141372141374, |
|
"grad_norm": 39.264137268066406, |
|
"learning_rate": 6.834750376549792e-09, |
|
"loss": 6.438, |
|
"step": 199 |
|
}, |
|
{ |
|
"epoch": 0.4158004158004158, |
|
"grad_norm": 29.41855239868164, |
|
"learning_rate": 0.0, |
|
"loss": 5.075, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.4158004158004158, |
|
"eval_loss": 1.3693137168884277, |
|
"eval_runtime": 16.5656, |
|
"eval_samples_per_second": 48.957, |
|
"eval_steps_per_second": 12.254, |
|
"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": 2.98048718635008e+16, |
|
"train_batch_size": 8, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|