| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.9843184559710494, | |
| "eval_steps": 500, | |
| "global_step": 102, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.019300361881785282, | |
| "grad_norm": 20.085893630981445, | |
| "learning_rate": 0.0, | |
| "loss": 1.5371, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.038600723763570564, | |
| "grad_norm": 24.923002243041992, | |
| "learning_rate": 5e-06, | |
| "loss": 1.5356, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.05790108564535585, | |
| "grad_norm": 21.43843650817871, | |
| "learning_rate": 7.924812503605782e-06, | |
| "loss": 1.4658, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.07720144752714113, | |
| "grad_norm": 10.800460815429688, | |
| "learning_rate": 1e-05, | |
| "loss": 1.1598, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.09650180940892641, | |
| "grad_norm": 6.396805286407471, | |
| "learning_rate": 1e-05, | |
| "loss": 1.0447, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.1158021712907117, | |
| "grad_norm": 7.415082931518555, | |
| "learning_rate": 1e-05, | |
| "loss": 1.0807, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.13510253317249699, | |
| "grad_norm": 6.602322578430176, | |
| "learning_rate": 1e-05, | |
| "loss": 1.0522, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.15440289505428226, | |
| "grad_norm": 5.840353012084961, | |
| "learning_rate": 1e-05, | |
| "loss": 0.9866, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.17370325693606756, | |
| "grad_norm": 5.387495040893555, | |
| "learning_rate": 1e-05, | |
| "loss": 1.0037, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.19300361881785283, | |
| "grad_norm": 4.899038791656494, | |
| "learning_rate": 1e-05, | |
| "loss": 0.9326, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.21230398069963813, | |
| "grad_norm": 4.857559680938721, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8738, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.2316043425814234, | |
| "grad_norm": 7.552295684814453, | |
| "learning_rate": 1e-05, | |
| "loss": 0.9318, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.25090470446320867, | |
| "grad_norm": 72.52302551269531, | |
| "learning_rate": 1e-05, | |
| "loss": 0.9038, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.27020506634499397, | |
| "grad_norm": 6.182501316070557, | |
| "learning_rate": 1e-05, | |
| "loss": 0.9649, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.28950542822677927, | |
| "grad_norm": 14.671534538269043, | |
| "learning_rate": 1e-05, | |
| "loss": 0.9451, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.3088057901085645, | |
| "grad_norm": 7.690091133117676, | |
| "learning_rate": 1e-05, | |
| "loss": 0.9173, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.3281061519903498, | |
| "grad_norm": 5.256872653961182, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8687, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.3474065138721351, | |
| "grad_norm": 5.03083610534668, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8548, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.36670687575392036, | |
| "grad_norm": 6.011650085449219, | |
| "learning_rate": 1e-05, | |
| "loss": 0.9021, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.38600723763570566, | |
| "grad_norm": 4.863091945648193, | |
| "learning_rate": 1e-05, | |
| "loss": 0.863, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.40530759951749096, | |
| "grad_norm": 6.599653244018555, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8321, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.42460796139927626, | |
| "grad_norm": 11.472197532653809, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8921, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.4439083232810615, | |
| "grad_norm": 5.892179489135742, | |
| "learning_rate": 1e-05, | |
| "loss": 0.9132, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.4632086851628468, | |
| "grad_norm": 5.178156852722168, | |
| "learning_rate": 1e-05, | |
| "loss": 0.891, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.4825090470446321, | |
| "grad_norm": 6.018004894256592, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8839, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.5018094089264173, | |
| "grad_norm": 4.832773208618164, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8527, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.5211097708082026, | |
| "grad_norm": 4.564591884613037, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8524, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.5404101326899879, | |
| "grad_norm": 4.709479808807373, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8517, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.5597104945717732, | |
| "grad_norm": 5.215322971343994, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8427, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.5790108564535585, | |
| "grad_norm": 4.680248737335205, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8777, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.5983112183353438, | |
| "grad_norm": 5.544800281524658, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8687, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.617611580217129, | |
| "grad_norm": 6.865499496459961, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8827, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.6369119420989143, | |
| "grad_norm": 4.069559097290039, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8862, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.6562123039806996, | |
| "grad_norm": 5.174818992614746, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8618, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.6755126658624849, | |
| "grad_norm": 12.915410041809082, | |
| "learning_rate": 1e-05, | |
| "loss": 0.9145, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.6948130277442702, | |
| "grad_norm": 3.8403801918029785, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8393, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.7141133896260555, | |
| "grad_norm": 5.092712879180908, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8467, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.7334137515078407, | |
| "grad_norm": 3.652506113052368, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8165, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.752714113389626, | |
| "grad_norm": 4.251276969909668, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8276, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.7720144752714113, | |
| "grad_norm": 4.063849449157715, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8141, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.7913148371531966, | |
| "grad_norm": 13.772127151489258, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8257, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.8106151990349819, | |
| "grad_norm": 3.2057080268859863, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8811, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.8299155609167672, | |
| "grad_norm": 3.1989691257476807, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8089, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.8492159227985525, | |
| "grad_norm": 3.2385919094085693, | |
| "learning_rate": 1e-05, | |
| "loss": 0.868, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.8685162846803377, | |
| "grad_norm": 3.792576789855957, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8358, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.887816646562123, | |
| "grad_norm": 4.059788703918457, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8842, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.9071170084439083, | |
| "grad_norm": 5.195308685302734, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8498, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.9264173703256936, | |
| "grad_norm": 5.914234638214111, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8596, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.9457177322074789, | |
| "grad_norm": 4.043056964874268, | |
| "learning_rate": 1e-05, | |
| "loss": 0.9006, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.9650180940892642, | |
| "grad_norm": 3.690598726272583, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8672, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.9843184559710495, | |
| "grad_norm": 3.392371416091919, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8641, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 1.0193003618817853, | |
| "grad_norm": 2.909470558166504, | |
| "learning_rate": 1e-05, | |
| "loss": 1.1104, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 1.0386007237635706, | |
| "grad_norm": 3.598850965499878, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7121, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 1.057901085645356, | |
| "grad_norm": 2.9036705493927, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7087, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 1.0772014475271412, | |
| "grad_norm": 4.788527011871338, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7226, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 1.0965018094089265, | |
| "grad_norm": 4.0117573738098145, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7155, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 1.1158021712907118, | |
| "grad_norm": 5.838222980499268, | |
| "learning_rate": 1e-05, | |
| "loss": 0.714, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 1.135102533172497, | |
| "grad_norm": 4.493499279022217, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6922, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 1.1544028950542822, | |
| "grad_norm": 8.510027885437012, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7162, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 1.1737032569360675, | |
| "grad_norm": 4.332089900970459, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6684, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.1930036188178528, | |
| "grad_norm": 6.2718634605407715, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6902, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 1.212303980699638, | |
| "grad_norm": 4.732446670532227, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6815, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 1.2316043425814234, | |
| "grad_norm": 3.6639814376831055, | |
| "learning_rate": 1e-05, | |
| "loss": 0.679, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 1.2509047044632087, | |
| "grad_norm": 3.1242263317108154, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6972, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 1.270205066344994, | |
| "grad_norm": 3.297794818878174, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6767, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 1.2895054282267793, | |
| "grad_norm": 3.250779151916504, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6771, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 1.3088057901085646, | |
| "grad_norm": 2.957329273223877, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6964, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 1.3281061519903499, | |
| "grad_norm": 4.555089473724365, | |
| "learning_rate": 1e-05, | |
| "loss": 0.679, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.3474065138721352, | |
| "grad_norm": 8.270596504211426, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7211, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 1.3667068757539202, | |
| "grad_norm": 4.109578609466553, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6777, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.3860072376357055, | |
| "grad_norm": 3.5171566009521484, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7172, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 1.4053075995174908, | |
| "grad_norm": 3.425497055053711, | |
| "learning_rate": 1e-05, | |
| "loss": 0.693, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.4246079613992761, | |
| "grad_norm": 4.035482883453369, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7502, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 1.4439083232810614, | |
| "grad_norm": 7.8259406089782715, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7002, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.4632086851628467, | |
| "grad_norm": 6.383781433105469, | |
| "learning_rate": 1e-05, | |
| "loss": 0.733, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.482509047044632, | |
| "grad_norm": 4.586795806884766, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6888, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.5018094089264173, | |
| "grad_norm": 3.248703718185425, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7034, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.5211097708082026, | |
| "grad_norm": 5.46513032913208, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7176, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.540410132689988, | |
| "grad_norm": 6.007761001586914, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6983, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.5597104945717732, | |
| "grad_norm": 3.7299880981445312, | |
| "learning_rate": 1e-05, | |
| "loss": 0.72, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.5790108564535585, | |
| "grad_norm": 3.501554250717163, | |
| "learning_rate": 1e-05, | |
| "loss": 0.727, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.5983112183353438, | |
| "grad_norm": 3.446808099746704, | |
| "learning_rate": 1e-05, | |
| "loss": 0.717, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.6176115802171291, | |
| "grad_norm": 3.2995755672454834, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6923, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.6369119420989144, | |
| "grad_norm": 7.1533098220825195, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6898, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.6562123039806997, | |
| "grad_norm": 3.07770037651062, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6799, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.675512665862485, | |
| "grad_norm": 3.425074338912964, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7234, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.6948130277442703, | |
| "grad_norm": 3.45398211479187, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7196, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.7141133896260556, | |
| "grad_norm": 4.230635643005371, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7123, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.7334137515078407, | |
| "grad_norm": 3.4547479152679443, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7317, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.752714113389626, | |
| "grad_norm": 3.5444135665893555, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7076, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.7720144752714113, | |
| "grad_norm": 3.361970901489258, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7349, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.7913148371531966, | |
| "grad_norm": 4.442152976989746, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6986, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.810615199034982, | |
| "grad_norm": 4.054971694946289, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7092, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.8299155609167672, | |
| "grad_norm": 2.9335885047912598, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7208, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.8492159227985525, | |
| "grad_norm": 4.029760360717773, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6943, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.8685162846803376, | |
| "grad_norm": 4.110034942626953, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7137, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.887816646562123, | |
| "grad_norm": 4.746143341064453, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7243, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.9071170084439082, | |
| "grad_norm": 2.9600183963775635, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7038, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.9264173703256935, | |
| "grad_norm": 2.582796096801758, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6802, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.9457177322074788, | |
| "grad_norm": 4.347536087036133, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.965018094089264, | |
| "grad_norm": 3.207463502883911, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7254, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.9843184559710494, | |
| "grad_norm": 3.204589605331421, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7292, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.9843184559710494, | |
| "step": 102, | |
| "total_flos": 5.52152723399639e+16, | |
| "train_loss": 0.8226426015881931, | |
| "train_runtime": 11731.6659, | |
| "train_samples_per_second": 1.13, | |
| "train_steps_per_second": 0.009 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 102, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": false, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.52152723399639e+16, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |