diff --git "a/LLaVA/text_augmention/4/trainer_state.json" "b/LLaVA/text_augmention/4/trainer_state.json" deleted file mode 100644--- "a/LLaVA/text_augmention/4/trainer_state.json" +++ /dev/null @@ -1,11814 +0,0 @@ -{ - "best_metric": null, - "best_model_checkpoint": null, - "epoch": 4.0, - "eval_steps": 500, - "global_step": 1964, - "is_hyper_param_search": false, - "is_local_process_zero": true, - "is_world_process_zero": true, - "log_history": [ - { - "epoch": 0.0, - "learning_rate": 3.3898305084745766e-07, - "loss": 2.0475, - "step": 1 - }, - { - "epoch": 0.0, - "learning_rate": 6.779661016949153e-07, - "loss": 2.0422, - "step": 2 - }, - { - "epoch": 0.01, - "learning_rate": 1.016949152542373e-06, - "loss": 2.0686, - "step": 3 - }, - { - "epoch": 0.01, - "learning_rate": 1.3559322033898307e-06, - "loss": 2.0872, - "step": 4 - }, - { - "epoch": 0.01, - "learning_rate": 1.6949152542372882e-06, - "loss": 2.124, - "step": 5 - }, - { - "epoch": 0.01, - "learning_rate": 2.033898305084746e-06, - "loss": 2.0267, - "step": 6 - }, - { - "epoch": 0.01, - "learning_rate": 2.372881355932204e-06, - "loss": 2.0424, - "step": 7 - }, - { - "epoch": 0.02, - "learning_rate": 2.7118644067796613e-06, - "loss": 2.0404, - "step": 8 - }, - { - "epoch": 0.02, - "learning_rate": 3.0508474576271192e-06, - "loss": 1.9794, - "step": 9 - }, - { - "epoch": 0.02, - "learning_rate": 3.3898305084745763e-06, - "loss": 1.9899, - "step": 10 - }, - { - "epoch": 0.02, - "learning_rate": 3.7288135593220342e-06, - "loss": 1.8836, - "step": 11 - }, - { - "epoch": 0.02, - "learning_rate": 4.067796610169492e-06, - "loss": 1.8607, - "step": 12 - }, - { - "epoch": 0.03, - "learning_rate": 4.40677966101695e-06, - "loss": 1.8183, - "step": 13 - }, - { - "epoch": 0.03, - "learning_rate": 4.745762711864408e-06, - "loss": 1.8858, - "step": 14 - }, - { - "epoch": 0.03, - "learning_rate": 5.084745762711865e-06, - "loss": 1.774, - "step": 15 - }, - { - "epoch": 0.03, - "learning_rate": 5.423728813559323e-06, - "loss": 1.8471, - "step": 16 - }, - { - "epoch": 0.03, - "learning_rate": 5.7627118644067805e-06, - "loss": 1.719, - "step": 17 - }, - { - "epoch": 0.04, - "learning_rate": 6.1016949152542385e-06, - "loss": 1.7577, - "step": 18 - }, - { - "epoch": 0.04, - "learning_rate": 6.440677966101695e-06, - "loss": 1.7911, - "step": 19 - }, - { - "epoch": 0.04, - "learning_rate": 6.779661016949153e-06, - "loss": 1.7463, - "step": 20 - }, - { - "epoch": 0.04, - "learning_rate": 7.1186440677966106e-06, - "loss": 1.7574, - "step": 21 - }, - { - "epoch": 0.04, - "learning_rate": 7.4576271186440685e-06, - "loss": 1.7467, - "step": 22 - }, - { - "epoch": 0.05, - "learning_rate": 7.796610169491526e-06, - "loss": 1.6981, - "step": 23 - }, - { - "epoch": 0.05, - "learning_rate": 8.135593220338983e-06, - "loss": 1.6691, - "step": 24 - }, - { - "epoch": 0.05, - "learning_rate": 8.47457627118644e-06, - "loss": 1.679, - "step": 25 - }, - { - "epoch": 0.05, - "learning_rate": 8.8135593220339e-06, - "loss": 1.7553, - "step": 26 - }, - { - "epoch": 0.05, - "learning_rate": 9.152542372881356e-06, - "loss": 1.6838, - "step": 27 - }, - { - "epoch": 0.06, - "learning_rate": 9.491525423728815e-06, - "loss": 1.6772, - "step": 28 - }, - { - "epoch": 0.06, - "learning_rate": 9.830508474576272e-06, - "loss": 1.6954, - "step": 29 - }, - { - "epoch": 0.06, - "learning_rate": 1.016949152542373e-05, - "loss": 1.6352, - "step": 30 - }, - { - "epoch": 0.06, - "learning_rate": 1.0508474576271188e-05, - "loss": 1.628, - "step": 31 - }, - { - "epoch": 0.07, - "learning_rate": 1.0847457627118645e-05, - "loss": 1.6512, - "step": 32 - }, - { - "epoch": 0.07, - "learning_rate": 1.1186440677966102e-05, - "loss": 1.6272, - "step": 33 - }, - { - "epoch": 0.07, - "learning_rate": 1.1525423728813561e-05, - "loss": 1.5854, - "step": 34 - }, - { - "epoch": 0.07, - "learning_rate": 1.1864406779661018e-05, - "loss": 1.6363, - "step": 35 - }, - { - "epoch": 0.07, - "learning_rate": 1.2203389830508477e-05, - "loss": 1.6406, - "step": 36 - }, - { - "epoch": 0.08, - "learning_rate": 1.2542372881355932e-05, - "loss": 1.5917, - "step": 37 - }, - { - "epoch": 0.08, - "learning_rate": 1.288135593220339e-05, - "loss": 1.5817, - "step": 38 - }, - { - "epoch": 0.08, - "learning_rate": 1.3220338983050848e-05, - "loss": 1.5786, - "step": 39 - }, - { - "epoch": 0.08, - "learning_rate": 1.3559322033898305e-05, - "loss": 1.5899, - "step": 40 - }, - { - "epoch": 0.08, - "learning_rate": 1.3898305084745764e-05, - "loss": 1.6387, - "step": 41 - }, - { - "epoch": 0.09, - "learning_rate": 1.4237288135593221e-05, - "loss": 1.6945, - "step": 42 - }, - { - "epoch": 0.09, - "learning_rate": 1.4576271186440678e-05, - "loss": 1.5668, - "step": 43 - }, - { - "epoch": 0.09, - "learning_rate": 1.4915254237288137e-05, - "loss": 1.5463, - "step": 44 - }, - { - "epoch": 0.09, - "learning_rate": 1.5254237288135594e-05, - "loss": 1.6098, - "step": 45 - }, - { - "epoch": 0.09, - "learning_rate": 1.5593220338983053e-05, - "loss": 1.5832, - "step": 46 - }, - { - "epoch": 0.1, - "learning_rate": 1.593220338983051e-05, - "loss": 1.5113, - "step": 47 - }, - { - "epoch": 0.1, - "learning_rate": 1.6271186440677967e-05, - "loss": 1.574, - "step": 48 - }, - { - "epoch": 0.1, - "learning_rate": 1.6610169491525424e-05, - "loss": 1.5009, - "step": 49 - }, - { - "epoch": 0.1, - "learning_rate": 1.694915254237288e-05, - "loss": 1.569, - "step": 50 - }, - { - "epoch": 0.1, - "learning_rate": 1.728813559322034e-05, - "loss": 1.6144, - "step": 51 - }, - { - "epoch": 0.11, - "learning_rate": 1.76271186440678e-05, - "loss": 1.5616, - "step": 52 - }, - { - "epoch": 0.11, - "learning_rate": 1.7966101694915256e-05, - "loss": 1.5755, - "step": 53 - }, - { - "epoch": 0.11, - "learning_rate": 1.8305084745762713e-05, - "loss": 1.5196, - "step": 54 - }, - { - "epoch": 0.11, - "learning_rate": 1.864406779661017e-05, - "loss": 1.5124, - "step": 55 - }, - { - "epoch": 0.11, - "learning_rate": 1.898305084745763e-05, - "loss": 1.5619, - "step": 56 - }, - { - "epoch": 0.12, - "learning_rate": 1.9322033898305087e-05, - "loss": 1.5137, - "step": 57 - }, - { - "epoch": 0.12, - "learning_rate": 1.9661016949152545e-05, - "loss": 1.5293, - "step": 58 - }, - { - "epoch": 0.12, - "learning_rate": 2e-05, - "loss": 1.5104, - "step": 59 - }, - { - "epoch": 0.12, - "learning_rate": 1.9999986401854267e-05, - "loss": 1.5144, - "step": 60 - }, - { - "epoch": 0.12, - "learning_rate": 1.999994560745405e-05, - "loss": 1.461, - "step": 61 - }, - { - "epoch": 0.13, - "learning_rate": 1.999987761691029e-05, - "loss": 1.582, - "step": 62 - }, - { - "epoch": 0.13, - "learning_rate": 1.99997824304079e-05, - "loss": 1.4724, - "step": 63 - }, - { - "epoch": 0.13, - "learning_rate": 1.9999660048205748e-05, - "loss": 1.4763, - "step": 64 - }, - { - "epoch": 0.13, - "learning_rate": 1.9999510470636677e-05, - "loss": 1.4531, - "step": 65 - }, - { - "epoch": 0.13, - "learning_rate": 1.9999333698107476e-05, - "loss": 1.5198, - "step": 66 - }, - { - "epoch": 0.14, - "learning_rate": 1.99991297310989e-05, - "loss": 1.4978, - "step": 67 - }, - { - "epoch": 0.14, - "learning_rate": 1.9998898570165663e-05, - "loss": 1.3942, - "step": 68 - }, - { - "epoch": 0.14, - "learning_rate": 1.999864021593644e-05, - "loss": 1.4357, - "step": 69 - }, - { - "epoch": 0.14, - "learning_rate": 1.999835466911386e-05, - "loss": 1.4608, - "step": 70 - }, - { - "epoch": 0.14, - "learning_rate": 1.9998041930474498e-05, - "loss": 1.4911, - "step": 71 - }, - { - "epoch": 0.15, - "learning_rate": 1.99977020008689e-05, - "loss": 1.4844, - "step": 72 - }, - { - "epoch": 0.15, - "learning_rate": 1.999733488122153e-05, - "loss": 1.4292, - "step": 73 - }, - { - "epoch": 0.15, - "learning_rate": 1.999694057253083e-05, - "loss": 1.5228, - "step": 74 - }, - { - "epoch": 0.15, - "learning_rate": 1.9996519075869174e-05, - "loss": 1.5077, - "step": 75 - }, - { - "epoch": 0.15, - "learning_rate": 1.999607039238287e-05, - "loss": 1.4661, - "step": 76 - }, - { - "epoch": 0.16, - "learning_rate": 1.999559452329218e-05, - "loss": 1.498, - "step": 77 - }, - { - "epoch": 0.16, - "learning_rate": 1.999509146989128e-05, - "loss": 1.5297, - "step": 78 - }, - { - "epoch": 0.16, - "learning_rate": 1.9994561233548295e-05, - "loss": 1.4256, - "step": 79 - }, - { - "epoch": 0.16, - "learning_rate": 1.999400381570527e-05, - "loss": 1.4567, - "step": 80 - }, - { - "epoch": 0.16, - "learning_rate": 1.9993419217878175e-05, - "loss": 1.4564, - "step": 81 - }, - { - "epoch": 0.17, - "learning_rate": 1.99928074416569e-05, - "loss": 1.4811, - "step": 82 - }, - { - "epoch": 0.17, - "learning_rate": 1.9992168488705248e-05, - "loss": 1.3717, - "step": 83 - }, - { - "epoch": 0.17, - "learning_rate": 1.9991502360760936e-05, - "loss": 1.4103, - "step": 84 - }, - { - "epoch": 0.17, - "learning_rate": 1.999080905963558e-05, - "loss": 1.4309, - "step": 85 - }, - { - "epoch": 0.18, - "learning_rate": 1.9990088587214712e-05, - "loss": 1.449, - "step": 86 - }, - { - "epoch": 0.18, - "learning_rate": 1.998934094545774e-05, - "loss": 1.3882, - "step": 87 - }, - { - "epoch": 0.18, - "learning_rate": 1.9988566136397977e-05, - "loss": 1.4144, - "step": 88 - }, - { - "epoch": 0.18, - "learning_rate": 1.9987764162142615e-05, - "loss": 1.4518, - "step": 89 - }, - { - "epoch": 0.18, - "learning_rate": 1.9986935024872725e-05, - "loss": 1.427, - "step": 90 - }, - { - "epoch": 0.19, - "learning_rate": 1.9986078726843256e-05, - "loss": 1.375, - "step": 91 - }, - { - "epoch": 0.19, - "learning_rate": 1.998519527038302e-05, - "loss": 1.4086, - "step": 92 - }, - { - "epoch": 0.19, - "learning_rate": 1.998428465789469e-05, - "loss": 1.3717, - "step": 93 - }, - { - "epoch": 0.19, - "learning_rate": 1.9983346891854798e-05, - "loss": 1.376, - "step": 94 - }, - { - "epoch": 0.19, - "learning_rate": 1.9982381974813714e-05, - "loss": 1.3905, - "step": 95 - }, - { - "epoch": 0.2, - "learning_rate": 1.9981389909395658e-05, - "loss": 1.4113, - "step": 96 - }, - { - "epoch": 0.2, - "learning_rate": 1.998037069829868e-05, - "loss": 1.468, - "step": 97 - }, - { - "epoch": 0.2, - "learning_rate": 1.9979324344294654e-05, - "loss": 1.4561, - "step": 98 - }, - { - "epoch": 0.2, - "learning_rate": 1.9978250850229278e-05, - "loss": 1.415, - "step": 99 - }, - { - "epoch": 0.2, - "learning_rate": 1.997715021902205e-05, - "loss": 1.4061, - "step": 100 - }, - { - "epoch": 0.21, - "learning_rate": 1.9976022453666293e-05, - "loss": 1.4115, - "step": 101 - }, - { - "epoch": 0.21, - "learning_rate": 1.99748675572291e-05, - "loss": 1.4231, - "step": 102 - }, - { - "epoch": 0.21, - "learning_rate": 1.997368553285136e-05, - "loss": 1.4678, - "step": 103 - }, - { - "epoch": 0.21, - "learning_rate": 1.9972476383747748e-05, - "loss": 1.4003, - "step": 104 - }, - { - "epoch": 0.21, - "learning_rate": 1.9971240113206698e-05, - "loss": 1.3551, - "step": 105 - }, - { - "epoch": 0.22, - "learning_rate": 1.9969976724590406e-05, - "loss": 1.3579, - "step": 106 - }, - { - "epoch": 0.22, - "learning_rate": 1.9968686221334818e-05, - "loss": 1.3787, - "step": 107 - }, - { - "epoch": 0.22, - "learning_rate": 1.996736860694963e-05, - "loss": 1.3372, - "step": 108 - }, - { - "epoch": 0.22, - "learning_rate": 1.9966023885018263e-05, - "loss": 1.4322, - "step": 109 - }, - { - "epoch": 0.22, - "learning_rate": 1.9964652059197863e-05, - "loss": 1.3522, - "step": 110 - }, - { - "epoch": 0.23, - "learning_rate": 1.9963253133219283e-05, - "loss": 1.358, - "step": 111 - }, - { - "epoch": 0.23, - "learning_rate": 1.9961827110887085e-05, - "loss": 1.357, - "step": 112 - }, - { - "epoch": 0.23, - "learning_rate": 1.996037399607952e-05, - "loss": 1.3703, - "step": 113 - }, - { - "epoch": 0.23, - "learning_rate": 1.9958893792748527e-05, - "loss": 1.3655, - "step": 114 - }, - { - "epoch": 0.23, - "learning_rate": 1.9957386504919702e-05, - "loss": 1.3399, - "step": 115 - }, - { - "epoch": 0.24, - "learning_rate": 1.9955852136692315e-05, - "loss": 1.4453, - "step": 116 - }, - { - "epoch": 0.24, - "learning_rate": 1.9954290692239274e-05, - "loss": 1.3854, - "step": 117 - }, - { - "epoch": 0.24, - "learning_rate": 1.9952702175807135e-05, - "loss": 1.3468, - "step": 118 - }, - { - "epoch": 0.24, - "learning_rate": 1.995108659171607e-05, - "loss": 1.3331, - "step": 119 - }, - { - "epoch": 0.24, - "learning_rate": 1.9949443944359862e-05, - "loss": 1.3231, - "step": 120 - }, - { - "epoch": 0.25, - "learning_rate": 1.9947774238205913e-05, - "loss": 1.2818, - "step": 121 - }, - { - "epoch": 0.25, - "learning_rate": 1.99460774777952e-05, - "loss": 1.3831, - "step": 122 - }, - { - "epoch": 0.25, - "learning_rate": 1.9944353667742286e-05, - "loss": 1.3083, - "step": 123 - }, - { - "epoch": 0.25, - "learning_rate": 1.994260281273529e-05, - "loss": 1.3702, - "step": 124 - }, - { - "epoch": 0.25, - "learning_rate": 1.9940824917535887e-05, - "loss": 1.2673, - "step": 125 - }, - { - "epoch": 0.26, - "learning_rate": 1.9939019986979296e-05, - "loss": 1.287, - "step": 126 - }, - { - "epoch": 0.26, - "learning_rate": 1.993718802597426e-05, - "loss": 1.3455, - "step": 127 - }, - { - "epoch": 0.26, - "learning_rate": 1.993532903950303e-05, - "loss": 1.3123, - "step": 128 - }, - { - "epoch": 0.26, - "learning_rate": 1.9933443032621365e-05, - "loss": 1.2749, - "step": 129 - }, - { - "epoch": 0.26, - "learning_rate": 1.9931530010458498e-05, - "loss": 1.3329, - "step": 130 - }, - { - "epoch": 0.27, - "learning_rate": 1.992958997821714e-05, - "loss": 1.331, - "step": 131 - }, - { - "epoch": 0.27, - "learning_rate": 1.992762294117347e-05, - "loss": 1.3026, - "step": 132 - }, - { - "epoch": 0.27, - "learning_rate": 1.9925628904677084e-05, - "loss": 1.3721, - "step": 133 - }, - { - "epoch": 0.27, - "learning_rate": 1.992360787415103e-05, - "loss": 1.2997, - "step": 134 - }, - { - "epoch": 0.27, - "learning_rate": 1.9921559855091765e-05, - "loss": 1.3598, - "step": 135 - }, - { - "epoch": 0.28, - "learning_rate": 1.9919484853069132e-05, - "loss": 1.2367, - "step": 136 - }, - { - "epoch": 0.28, - "learning_rate": 1.9917382873726375e-05, - "loss": 1.2903, - "step": 137 - }, - { - "epoch": 0.28, - "learning_rate": 1.9915253922780095e-05, - "loss": 1.3336, - "step": 138 - }, - { - "epoch": 0.28, - "learning_rate": 1.9913098006020245e-05, - "loss": 1.2152, - "step": 139 - }, - { - "epoch": 0.29, - "learning_rate": 1.991091512931013e-05, - "loss": 1.2575, - "step": 140 - }, - { - "epoch": 0.29, - "learning_rate": 1.9908705298586356e-05, - "loss": 1.2389, - "step": 141 - }, - { - "epoch": 0.29, - "learning_rate": 1.990646851985884e-05, - "loss": 1.3122, - "step": 142 - }, - { - "epoch": 0.29, - "learning_rate": 1.99042047992108e-05, - "loss": 1.2073, - "step": 143 - }, - { - "epoch": 0.29, - "learning_rate": 1.9901914142798715e-05, - "loss": 1.2668, - "step": 144 - }, - { - "epoch": 0.3, - "learning_rate": 1.9899596556852315e-05, - "loss": 1.2494, - "step": 145 - }, - { - "epoch": 0.3, - "learning_rate": 1.989725204767458e-05, - "loss": 1.2657, - "step": 146 - }, - { - "epoch": 0.3, - "learning_rate": 1.9894880621641705e-05, - "loss": 1.2439, - "step": 147 - }, - { - "epoch": 0.3, - "learning_rate": 1.9892482285203085e-05, - "loss": 1.2707, - "step": 148 - }, - { - "epoch": 0.3, - "learning_rate": 1.9890057044881308e-05, - "loss": 1.2701, - "step": 149 - }, - { - "epoch": 0.31, - "learning_rate": 1.988760490727213e-05, - "loss": 1.2764, - "step": 150 - }, - { - "epoch": 0.31, - "learning_rate": 1.9885125879044456e-05, - "loss": 1.2415, - "step": 151 - }, - { - "epoch": 0.31, - "learning_rate": 1.988261996694032e-05, - "loss": 1.2376, - "step": 152 - }, - { - "epoch": 0.31, - "learning_rate": 1.9880087177774876e-05, - "loss": 1.2685, - "step": 153 - }, - { - "epoch": 0.31, - "learning_rate": 1.9877527518436373e-05, - "loss": 1.2107, - "step": 154 - }, - { - "epoch": 0.32, - "learning_rate": 1.987494099588613e-05, - "loss": 1.3011, - "step": 155 - }, - { - "epoch": 0.32, - "learning_rate": 1.9872327617158532e-05, - "loss": 1.2244, - "step": 156 - }, - { - "epoch": 0.32, - "learning_rate": 1.9869687389361e-05, - "loss": 1.2334, - "step": 157 - }, - { - "epoch": 0.32, - "learning_rate": 1.986702031967398e-05, - "loss": 1.1874, - "step": 158 - }, - { - "epoch": 0.32, - "learning_rate": 1.98643264153509e-05, - "loss": 1.2323, - "step": 159 - }, - { - "epoch": 0.33, - "learning_rate": 1.9861605683718192e-05, - "loss": 1.1937, - "step": 160 - }, - { - "epoch": 0.33, - "learning_rate": 1.985885813217523e-05, - "loss": 1.2774, - "step": 161 - }, - { - "epoch": 0.33, - "learning_rate": 1.985608376819434e-05, - "loss": 1.222, - "step": 162 - }, - { - "epoch": 0.33, - "learning_rate": 1.9853282599320763e-05, - "loss": 1.202, - "step": 163 - }, - { - "epoch": 0.33, - "learning_rate": 1.9850454633172632e-05, - "loss": 1.1815, - "step": 164 - }, - { - "epoch": 0.34, - "learning_rate": 1.9847599877440976e-05, - "loss": 1.2084, - "step": 165 - }, - { - "epoch": 0.34, - "learning_rate": 1.984471833988967e-05, - "loss": 1.1891, - "step": 166 - }, - { - "epoch": 0.34, - "learning_rate": 1.984181002835542e-05, - "loss": 1.2425, - "step": 167 - }, - { - "epoch": 0.34, - "learning_rate": 1.9838874950747767e-05, - "loss": 1.1617, - "step": 168 - }, - { - "epoch": 0.34, - "learning_rate": 1.9835913115049022e-05, - "loss": 1.1206, - "step": 169 - }, - { - "epoch": 0.35, - "learning_rate": 1.9832924529314284e-05, - "loss": 1.2003, - "step": 170 - }, - { - "epoch": 0.35, - "learning_rate": 1.98299092016714e-05, - "loss": 1.2243, - "step": 171 - }, - { - "epoch": 0.35, - "learning_rate": 1.9826867140320936e-05, - "loss": 1.2673, - "step": 172 - }, - { - "epoch": 0.35, - "learning_rate": 1.9823798353536183e-05, - "loss": 1.1423, - "step": 173 - }, - { - "epoch": 0.35, - "learning_rate": 1.982070284966309e-05, - "loss": 1.1209, - "step": 174 - }, - { - "epoch": 0.36, - "learning_rate": 1.9817580637120294e-05, - "loss": 1.1502, - "step": 175 - }, - { - "epoch": 0.36, - "learning_rate": 1.981443172439904e-05, - "loss": 1.1701, - "step": 176 - }, - { - "epoch": 0.36, - "learning_rate": 1.9811256120063208e-05, - "loss": 1.2387, - "step": 177 - }, - { - "epoch": 0.36, - "learning_rate": 1.9808053832749274e-05, - "loss": 1.1376, - "step": 178 - }, - { - "epoch": 0.36, - "learning_rate": 1.9804824871166254e-05, - "loss": 1.1876, - "step": 179 - }, - { - "epoch": 0.37, - "learning_rate": 1.980156924409574e-05, - "loss": 1.112, - "step": 180 - }, - { - "epoch": 0.37, - "learning_rate": 1.9798286960391826e-05, - "loss": 1.1807, - "step": 181 - }, - { - "epoch": 0.37, - "learning_rate": 1.9794978028981106e-05, - "loss": 1.1476, - "step": 182 - }, - { - "epoch": 0.37, - "learning_rate": 1.9791642458862647e-05, - "loss": 1.2169, - "step": 183 - }, - { - "epoch": 0.37, - "learning_rate": 1.978828025910796e-05, - "loss": 1.1944, - "step": 184 - }, - { - "epoch": 0.38, - "learning_rate": 1.9784891438860987e-05, - "loss": 1.1971, - "step": 185 - }, - { - "epoch": 0.38, - "learning_rate": 1.9781476007338058e-05, - "loss": 1.1876, - "step": 186 - }, - { - "epoch": 0.38, - "learning_rate": 1.977803397382788e-05, - "loss": 1.1162, - "step": 187 - }, - { - "epoch": 0.38, - "learning_rate": 1.977456534769151e-05, - "loss": 1.1425, - "step": 188 - }, - { - "epoch": 0.38, - "learning_rate": 1.9771070138362326e-05, - "loss": 1.2565, - "step": 189 - }, - { - "epoch": 0.39, - "learning_rate": 1.9767548355345996e-05, - "loss": 1.1459, - "step": 190 - }, - { - "epoch": 0.39, - "learning_rate": 1.9764000008220464e-05, - "loss": 1.099, - "step": 191 - }, - { - "epoch": 0.39, - "learning_rate": 1.9760425106635925e-05, - "loss": 1.1498, - "step": 192 - }, - { - "epoch": 0.39, - "learning_rate": 1.975682366031478e-05, - "loss": 1.1206, - "step": 193 - }, - { - "epoch": 0.4, - "learning_rate": 1.975319567905163e-05, - "loss": 1.1139, - "step": 194 - }, - { - "epoch": 0.4, - "learning_rate": 1.9749541172713233e-05, - "loss": 1.1086, - "step": 195 - }, - { - "epoch": 0.4, - "learning_rate": 1.97458601512385e-05, - "loss": 1.1021, - "step": 196 - }, - { - "epoch": 0.4, - "learning_rate": 1.9742152624638437e-05, - "loss": 1.1214, - "step": 197 - }, - { - "epoch": 0.4, - "learning_rate": 1.9738418602996145e-05, - "loss": 1.1512, - "step": 198 - }, - { - "epoch": 0.41, - "learning_rate": 1.9734658096466774e-05, - "loss": 1.0703, - "step": 199 - }, - { - "epoch": 0.41, - "learning_rate": 1.9730871115277513e-05, - "loss": 1.1672, - "step": 200 - }, - { - "epoch": 0.41, - "learning_rate": 1.9727057669727544e-05, - "loss": 1.1079, - "step": 201 - }, - { - "epoch": 0.41, - "learning_rate": 1.9723217770188022e-05, - "loss": 1.0522, - "step": 202 - }, - { - "epoch": 0.41, - "learning_rate": 1.9719351427102054e-05, - "loss": 1.1117, - "step": 203 - }, - { - "epoch": 0.42, - "learning_rate": 1.9715458650984657e-05, - "loss": 1.1321, - "step": 204 - }, - { - "epoch": 0.42, - "learning_rate": 1.971153945242274e-05, - "loss": 1.1317, - "step": 205 - }, - { - "epoch": 0.42, - "learning_rate": 1.970759384207507e-05, - "loss": 1.1131, - "step": 206 - }, - { - "epoch": 0.42, - "learning_rate": 1.9703621830672238e-05, - "loss": 1.1306, - "step": 207 - }, - { - "epoch": 0.42, - "learning_rate": 1.9699623429016648e-05, - "loss": 1.101, - "step": 208 - }, - { - "epoch": 0.43, - "learning_rate": 1.9695598647982467e-05, - "loss": 1.0598, - "step": 209 - }, - { - "epoch": 0.43, - "learning_rate": 1.9691547498515608e-05, - "loss": 1.1174, - "step": 210 - }, - { - "epoch": 0.43, - "learning_rate": 1.9687469991633695e-05, - "loss": 1.0875, - "step": 211 - }, - { - "epoch": 0.43, - "learning_rate": 1.9683366138426033e-05, - "loss": 1.1267, - "step": 212 - }, - { - "epoch": 0.43, - "learning_rate": 1.967923595005359e-05, - "loss": 1.0703, - "step": 213 - }, - { - "epoch": 0.44, - "learning_rate": 1.967507943774893e-05, - "loss": 1.122, - "step": 214 - }, - { - "epoch": 0.44, - "learning_rate": 1.9670896612816236e-05, - "loss": 1.166, - "step": 215 - }, - { - "epoch": 0.44, - "learning_rate": 1.966668748663124e-05, - "loss": 1.0819, - "step": 216 - }, - { - "epoch": 0.44, - "learning_rate": 1.9662452070641205e-05, - "loss": 1.082, - "step": 217 - }, - { - "epoch": 0.44, - "learning_rate": 1.9658190376364884e-05, - "loss": 1.0701, - "step": 218 - }, - { - "epoch": 0.45, - "learning_rate": 1.9653902415392517e-05, - "loss": 1.1225, - "step": 219 - }, - { - "epoch": 0.45, - "learning_rate": 1.9649588199385755e-05, - "loss": 1.0628, - "step": 220 - }, - { - "epoch": 0.45, - "learning_rate": 1.964524774007768e-05, - "loss": 1.118, - "step": 221 - }, - { - "epoch": 0.45, - "learning_rate": 1.9640881049272715e-05, - "loss": 1.153, - "step": 222 - }, - { - "epoch": 0.45, - "learning_rate": 1.963648813884665e-05, - "loss": 1.1073, - "step": 223 - }, - { - "epoch": 0.46, - "learning_rate": 1.9632069020746574e-05, - "loss": 0.9986, - "step": 224 - }, - { - "epoch": 0.46, - "learning_rate": 1.9627623706990843e-05, - "loss": 1.0764, - "step": 225 - }, - { - "epoch": 0.46, - "learning_rate": 1.962315220966906e-05, - "loss": 1.0668, - "step": 226 - }, - { - "epoch": 0.46, - "learning_rate": 1.961865454094205e-05, - "loss": 1.0333, - "step": 227 - }, - { - "epoch": 0.46, - "learning_rate": 1.96141307130418e-05, - "loss": 1.0492, - "step": 228 - }, - { - "epoch": 0.47, - "learning_rate": 1.9609580738271438e-05, - "loss": 1.0841, - "step": 229 - }, - { - "epoch": 0.47, - "learning_rate": 1.9605004629005214e-05, - "loss": 1.1166, - "step": 230 - }, - { - "epoch": 0.47, - "learning_rate": 1.9600402397688442e-05, - "loss": 1.0386, - "step": 231 - }, - { - "epoch": 0.47, - "learning_rate": 1.9595774056837495e-05, - "loss": 1.0892, - "step": 232 - }, - { - "epoch": 0.47, - "learning_rate": 1.959111961903973e-05, - "loss": 1.0571, - "step": 233 - }, - { - "epoch": 0.48, - "learning_rate": 1.9586439096953508e-05, - "loss": 1.098, - "step": 234 - }, - { - "epoch": 0.48, - "learning_rate": 1.95817325033081e-05, - "loss": 1.1044, - "step": 235 - }, - { - "epoch": 0.48, - "learning_rate": 1.9576999850903698e-05, - "loss": 1.1167, - "step": 236 - }, - { - "epoch": 0.48, - "learning_rate": 1.9572241152611364e-05, - "loss": 1.0655, - "step": 237 - }, - { - "epoch": 0.48, - "learning_rate": 1.956745642137299e-05, - "loss": 0.9712, - "step": 238 - }, - { - "epoch": 0.49, - "learning_rate": 1.9562645670201278e-05, - "loss": 0.9443, - "step": 239 - }, - { - "epoch": 0.49, - "learning_rate": 1.9557808912179677e-05, - "loss": 1.0327, - "step": 240 - }, - { - "epoch": 0.49, - "learning_rate": 1.9552946160462382e-05, - "loss": 1.0496, - "step": 241 - }, - { - "epoch": 0.49, - "learning_rate": 1.9548057428274267e-05, - "loss": 0.9884, - "step": 242 - }, - { - "epoch": 0.49, - "learning_rate": 1.9543142728910876e-05, - "loss": 1.003, - "step": 243 - }, - { - "epoch": 0.5, - "learning_rate": 1.9538202075738373e-05, - "loss": 1.0142, - "step": 244 - }, - { - "epoch": 0.5, - "learning_rate": 1.9533235482193492e-05, - "loss": 1.0023, - "step": 245 - }, - { - "epoch": 0.5, - "learning_rate": 1.952824296178353e-05, - "loss": 1.0091, - "step": 246 - }, - { - "epoch": 0.5, - "learning_rate": 1.95232245280863e-05, - "loss": 0.947, - "step": 247 - }, - { - "epoch": 0.51, - "learning_rate": 1.9518180194750063e-05, - "loss": 1.0289, - "step": 248 - }, - { - "epoch": 0.51, - "learning_rate": 1.9513109975493553e-05, - "loss": 1.0165, - "step": 249 - }, - { - "epoch": 0.51, - "learning_rate": 1.950801388410587e-05, - "loss": 1.0097, - "step": 250 - }, - { - "epoch": 0.51, - "learning_rate": 1.9502891934446505e-05, - "loss": 1.0517, - "step": 251 - }, - { - "epoch": 0.51, - "learning_rate": 1.949774414044525e-05, - "loss": 1.0441, - "step": 252 - }, - { - "epoch": 0.52, - "learning_rate": 1.9492570516102208e-05, - "loss": 0.995, - "step": 253 - }, - { - "epoch": 0.52, - "learning_rate": 1.948737107548771e-05, - "loss": 1.0072, - "step": 254 - }, - { - "epoch": 0.52, - "learning_rate": 1.9482145832742313e-05, - "loss": 1.0406, - "step": 255 - }, - { - "epoch": 0.52, - "learning_rate": 1.9476894802076735e-05, - "loss": 0.9935, - "step": 256 - }, - { - "epoch": 0.52, - "learning_rate": 1.947161799777183e-05, - "loss": 0.9782, - "step": 257 - }, - { - "epoch": 0.53, - "learning_rate": 1.9466315434178553e-05, - "loss": 0.9694, - "step": 258 - }, - { - "epoch": 0.53, - "learning_rate": 1.9460987125717907e-05, - "loss": 1.0032, - "step": 259 - }, - { - "epoch": 0.53, - "learning_rate": 1.9455633086880916e-05, - "loss": 0.9702, - "step": 260 - }, - { - "epoch": 0.53, - "learning_rate": 1.9450253332228586e-05, - "loss": 0.9764, - "step": 261 - }, - { - "epoch": 0.53, - "learning_rate": 1.9444847876391844e-05, - "loss": 0.9888, - "step": 262 - }, - { - "epoch": 0.54, - "learning_rate": 1.9439416734071535e-05, - "loss": 1.0234, - "step": 263 - }, - { - "epoch": 0.54, - "learning_rate": 1.9433959920038346e-05, - "loss": 0.9688, - "step": 264 - }, - { - "epoch": 0.54, - "learning_rate": 1.942847744913279e-05, - "loss": 0.9638, - "step": 265 - }, - { - "epoch": 0.54, - "learning_rate": 1.942296933626515e-05, - "loss": 1.0383, - "step": 266 - }, - { - "epoch": 0.54, - "learning_rate": 1.941743559641546e-05, - "loss": 1.0097, - "step": 267 - }, - { - "epoch": 0.55, - "learning_rate": 1.941187624463343e-05, - "loss": 0.9479, - "step": 268 - }, - { - "epoch": 0.55, - "learning_rate": 1.940629129603844e-05, - "loss": 1.0841, - "step": 269 - }, - { - "epoch": 0.55, - "learning_rate": 1.940068076581948e-05, - "loss": 0.9246, - "step": 270 - }, - { - "epoch": 0.55, - "learning_rate": 1.939504466923511e-05, - "loss": 0.9655, - "step": 271 - }, - { - "epoch": 0.55, - "learning_rate": 1.9389383021613422e-05, - "loss": 1.038, - "step": 272 - }, - { - "epoch": 0.56, - "learning_rate": 1.9383695838351996e-05, - "loss": 0.9308, - "step": 273 - }, - { - "epoch": 0.56, - "learning_rate": 1.9377983134917868e-05, - "loss": 0.9757, - "step": 274 - }, - { - "epoch": 0.56, - "learning_rate": 1.9372244926847467e-05, - "loss": 0.9758, - "step": 275 - }, - { - "epoch": 0.56, - "learning_rate": 1.936648122974659e-05, - "loss": 0.9944, - "step": 276 - }, - { - "epoch": 0.56, - "learning_rate": 1.936069205929036e-05, - "loss": 0.9349, - "step": 277 - }, - { - "epoch": 0.57, - "learning_rate": 1.9354877431223173e-05, - "loss": 0.9626, - "step": 278 - }, - { - "epoch": 0.57, - "learning_rate": 1.934903736135866e-05, - "loss": 0.941, - "step": 279 - }, - { - "epoch": 0.57, - "learning_rate": 1.9343171865579644e-05, - "loss": 0.9503, - "step": 280 - }, - { - "epoch": 0.57, - "learning_rate": 1.93372809598381e-05, - "loss": 0.9575, - "step": 281 - }, - { - "epoch": 0.57, - "learning_rate": 1.9331364660155105e-05, - "loss": 0.9607, - "step": 282 - }, - { - "epoch": 0.58, - "learning_rate": 1.9325422982620797e-05, - "loss": 0.9667, - "step": 283 - }, - { - "epoch": 0.58, - "learning_rate": 1.9319455943394347e-05, - "loss": 0.8964, - "step": 284 - }, - { - "epoch": 0.58, - "learning_rate": 1.931346355870388e-05, - "loss": 0.97, - "step": 285 - }, - { - "epoch": 0.58, - "learning_rate": 1.930744584484646e-05, - "loss": 0.9952, - "step": 286 - }, - { - "epoch": 0.58, - "learning_rate": 1.9301402818188036e-05, - "loss": 0.9338, - "step": 287 - }, - { - "epoch": 0.59, - "learning_rate": 1.9295334495163407e-05, - "loss": 0.9044, - "step": 288 - }, - { - "epoch": 0.59, - "learning_rate": 1.9289240892276156e-05, - "loss": 0.9817, - "step": 289 - }, - { - "epoch": 0.59, - "learning_rate": 1.928312202609862e-05, - "loss": 0.9446, - "step": 290 - }, - { - "epoch": 0.59, - "learning_rate": 1.927697791327185e-05, - "loss": 0.8932, - "step": 291 - }, - { - "epoch": 0.59, - "learning_rate": 1.9270808570505554e-05, - "loss": 0.9251, - "step": 292 - }, - { - "epoch": 0.6, - "learning_rate": 1.9264614014578057e-05, - "loss": 0.9065, - "step": 293 - }, - { - "epoch": 0.6, - "learning_rate": 1.925839426233625e-05, - "loss": 0.9693, - "step": 294 - }, - { - "epoch": 0.6, - "learning_rate": 1.9252149330695556e-05, - "loss": 0.8916, - "step": 295 - }, - { - "epoch": 0.6, - "learning_rate": 1.924587923663987e-05, - "loss": 0.9133, - "step": 296 - }, - { - "epoch": 0.6, - "learning_rate": 1.9239583997221526e-05, - "loss": 0.8712, - "step": 297 - }, - { - "epoch": 0.61, - "learning_rate": 1.923326362956124e-05, - "loss": 0.9198, - "step": 298 - }, - { - "epoch": 0.61, - "learning_rate": 1.9226918150848067e-05, - "loss": 0.9679, - "step": 299 - }, - { - "epoch": 0.61, - "learning_rate": 1.922054757833936e-05, - "loss": 0.9767, - "step": 300 - }, - { - "epoch": 0.61, - "learning_rate": 1.9214151929360703e-05, - "loss": 0.9264, - "step": 301 - }, - { - "epoch": 0.62, - "learning_rate": 1.9207731221305905e-05, - "loss": 0.9213, - "step": 302 - }, - { - "epoch": 0.62, - "learning_rate": 1.9201285471636894e-05, - "loss": 0.8487, - "step": 303 - }, - { - "epoch": 0.62, - "learning_rate": 1.919481469788373e-05, - "loss": 0.8715, - "step": 304 - }, - { - "epoch": 0.62, - "learning_rate": 1.9188318917644514e-05, - "loss": 0.8739, - "step": 305 - }, - { - "epoch": 0.62, - "learning_rate": 1.9181798148585362e-05, - "loss": 0.8765, - "step": 306 - }, - { - "epoch": 0.63, - "learning_rate": 1.9175252408440342e-05, - "loss": 0.9467, - "step": 307 - }, - { - "epoch": 0.63, - "learning_rate": 1.9168681715011444e-05, - "loss": 0.9219, - "step": 308 - }, - { - "epoch": 0.63, - "learning_rate": 1.916208608616852e-05, - "loss": 0.9096, - "step": 309 - }, - { - "epoch": 0.63, - "learning_rate": 1.915546553984923e-05, - "loss": 0.8879, - "step": 310 - }, - { - "epoch": 0.63, - "learning_rate": 1.9148820094059004e-05, - "loss": 0.9883, - "step": 311 - }, - { - "epoch": 0.64, - "learning_rate": 1.914214976687099e-05, - "loss": 0.9587, - "step": 312 - }, - { - "epoch": 0.64, - "learning_rate": 1.913545457642601e-05, - "loss": 0.8846, - "step": 313 - }, - { - "epoch": 0.64, - "learning_rate": 1.9128734540932494e-05, - "loss": 0.9469, - "step": 314 - }, - { - "epoch": 0.64, - "learning_rate": 1.9121989678666447e-05, - "loss": 0.8724, - "step": 315 - }, - { - "epoch": 0.64, - "learning_rate": 1.9115220007971394e-05, - "loss": 0.9547, - "step": 316 - }, - { - "epoch": 0.65, - "learning_rate": 1.9108425547258328e-05, - "loss": 0.9042, - "step": 317 - }, - { - "epoch": 0.65, - "learning_rate": 1.9101606315005664e-05, - "loss": 0.8872, - "step": 318 - }, - { - "epoch": 0.65, - "learning_rate": 1.909476232975918e-05, - "loss": 0.8879, - "step": 319 - }, - { - "epoch": 0.65, - "learning_rate": 1.9087893610131986e-05, - "loss": 0.831, - "step": 320 - }, - { - "epoch": 0.65, - "learning_rate": 1.9081000174804445e-05, - "loss": 0.8736, - "step": 321 - }, - { - "epoch": 0.66, - "learning_rate": 1.9074082042524146e-05, - "loss": 0.8529, - "step": 322 - }, - { - "epoch": 0.66, - "learning_rate": 1.9067139232105844e-05, - "loss": 0.8313, - "step": 323 - }, - { - "epoch": 0.66, - "learning_rate": 1.906017176243141e-05, - "loss": 0.8372, - "step": 324 - }, - { - "epoch": 0.66, - "learning_rate": 1.9053179652449776e-05, - "loss": 0.7925, - "step": 325 - }, - { - "epoch": 0.66, - "learning_rate": 1.904616292117689e-05, - "loss": 0.7974, - "step": 326 - }, - { - "epoch": 0.67, - "learning_rate": 1.9039121587695653e-05, - "loss": 0.8485, - "step": 327 - }, - { - "epoch": 0.67, - "learning_rate": 1.9032055671155887e-05, - "loss": 0.8272, - "step": 328 - }, - { - "epoch": 0.67, - "learning_rate": 1.9024965190774262e-05, - "loss": 0.8086, - "step": 329 - }, - { - "epoch": 0.67, - "learning_rate": 1.9017850165834258e-05, - "loss": 0.8871, - "step": 330 - }, - { - "epoch": 0.67, - "learning_rate": 1.9010710615686103e-05, - "loss": 0.8806, - "step": 331 - }, - { - "epoch": 0.68, - "learning_rate": 1.900354655974672e-05, - "loss": 0.83, - "step": 332 - }, - { - "epoch": 0.68, - "learning_rate": 1.8996358017499696e-05, - "loss": 0.8042, - "step": 333 - }, - { - "epoch": 0.68, - "learning_rate": 1.8989145008495185e-05, - "loss": 0.8442, - "step": 334 - }, - { - "epoch": 0.68, - "learning_rate": 1.898190755234991e-05, - "loss": 0.7808, - "step": 335 - }, - { - "epoch": 0.68, - "learning_rate": 1.897464566874706e-05, - "loss": 0.8617, - "step": 336 - }, - { - "epoch": 0.69, - "learning_rate": 1.896735937743627e-05, - "loss": 0.8171, - "step": 337 - }, - { - "epoch": 0.69, - "learning_rate": 1.8960048698233544e-05, - "loss": 0.8699, - "step": 338 - }, - { - "epoch": 0.69, - "learning_rate": 1.8952713651021227e-05, - "loss": 0.8833, - "step": 339 - }, - { - "epoch": 0.69, - "learning_rate": 1.8945354255747918e-05, - "loss": 0.8195, - "step": 340 - }, - { - "epoch": 0.69, - "learning_rate": 1.8937970532428446e-05, - "loss": 0.8379, - "step": 341 - }, - { - "epoch": 0.7, - "learning_rate": 1.8930562501143805e-05, - "loss": 0.815, - "step": 342 - }, - { - "epoch": 0.7, - "learning_rate": 1.8923130182041086e-05, - "loss": 0.8459, - "step": 343 - }, - { - "epoch": 0.7, - "learning_rate": 1.8915673595333443e-05, - "loss": 0.8394, - "step": 344 - }, - { - "epoch": 0.7, - "learning_rate": 1.890819276130003e-05, - "loss": 0.8404, - "step": 345 - }, - { - "epoch": 0.7, - "learning_rate": 1.8900687700285937e-05, - "loss": 0.8188, - "step": 346 - }, - { - "epoch": 0.71, - "learning_rate": 1.889315843270215e-05, - "loss": 0.7814, - "step": 347 - }, - { - "epoch": 0.71, - "learning_rate": 1.8885604979025483e-05, - "loss": 0.8415, - "step": 348 - }, - { - "epoch": 0.71, - "learning_rate": 1.8878027359798527e-05, - "loss": 0.8242, - "step": 349 - }, - { - "epoch": 0.71, - "learning_rate": 1.8870425595629597e-05, - "loss": 0.789, - "step": 350 - }, - { - "epoch": 0.71, - "learning_rate": 1.8862799707192675e-05, - "loss": 0.8605, - "step": 351 - }, - { - "epoch": 0.72, - "learning_rate": 1.8855149715227345e-05, - "loss": 0.8504, - "step": 352 - }, - { - "epoch": 0.72, - "learning_rate": 1.8847475640538753e-05, - "loss": 0.8616, - "step": 353 - }, - { - "epoch": 0.72, - "learning_rate": 1.8839777503997535e-05, - "loss": 0.83, - "step": 354 - }, - { - "epoch": 0.72, - "learning_rate": 1.8832055326539762e-05, - "loss": 0.8461, - "step": 355 - }, - { - "epoch": 0.73, - "learning_rate": 1.8824309129166897e-05, - "loss": 0.8269, - "step": 356 - }, - { - "epoch": 0.73, - "learning_rate": 1.881653893294573e-05, - "loss": 0.8349, - "step": 357 - }, - { - "epoch": 0.73, - "learning_rate": 1.8808744759008303e-05, - "loss": 0.8201, - "step": 358 - }, - { - "epoch": 0.73, - "learning_rate": 1.8800926628551884e-05, - "loss": 0.8173, - "step": 359 - }, - { - "epoch": 0.73, - "learning_rate": 1.879308456283889e-05, - "loss": 0.8194, - "step": 360 - }, - { - "epoch": 0.74, - "learning_rate": 1.8785218583196828e-05, - "loss": 0.8422, - "step": 361 - }, - { - "epoch": 0.74, - "learning_rate": 1.8777328711018244e-05, - "loss": 0.839, - "step": 362 - }, - { - "epoch": 0.74, - "learning_rate": 1.8769414967760667e-05, - "loss": 0.8063, - "step": 363 - }, - { - "epoch": 0.74, - "learning_rate": 1.8761477374946548e-05, - "loss": 0.8041, - "step": 364 - }, - { - "epoch": 0.74, - "learning_rate": 1.8753515954163188e-05, - "loss": 0.8466, - "step": 365 - }, - { - "epoch": 0.75, - "learning_rate": 1.87455307270627e-05, - "loss": 0.8675, - "step": 366 - }, - { - "epoch": 0.75, - "learning_rate": 1.873752171536195e-05, - "loss": 0.8377, - "step": 367 - }, - { - "epoch": 0.75, - "learning_rate": 1.8729488940842468e-05, - "loss": 0.8228, - "step": 368 - }, - { - "epoch": 0.75, - "learning_rate": 1.8721432425350426e-05, - "loss": 0.8276, - "step": 369 - }, - { - "epoch": 0.75, - "learning_rate": 1.871335219079656e-05, - "loss": 0.8007, - "step": 370 - }, - { - "epoch": 0.76, - "learning_rate": 1.870524825915611e-05, - "loss": 0.7667, - "step": 371 - }, - { - "epoch": 0.76, - "learning_rate": 1.8697120652468763e-05, - "loss": 0.7707, - "step": 372 - }, - { - "epoch": 0.76, - "learning_rate": 1.86889693928386e-05, - "loss": 0.7662, - "step": 373 - }, - { - "epoch": 0.76, - "learning_rate": 1.8680794502434018e-05, - "loss": 0.8238, - "step": 374 - }, - { - "epoch": 0.76, - "learning_rate": 1.8672596003487693e-05, - "loss": 0.7974, - "step": 375 - }, - { - "epoch": 0.77, - "learning_rate": 1.8664373918296495e-05, - "loss": 0.7971, - "step": 376 - }, - { - "epoch": 0.77, - "learning_rate": 1.8656128269221452e-05, - "loss": 0.7648, - "step": 377 - }, - { - "epoch": 0.77, - "learning_rate": 1.8647859078687673e-05, - "loss": 0.7852, - "step": 378 - }, - { - "epoch": 0.77, - "learning_rate": 1.8639566369184284e-05, - "loss": 0.8115, - "step": 379 - }, - { - "epoch": 0.77, - "learning_rate": 1.8631250163264383e-05, - "loss": 0.7885, - "step": 380 - }, - { - "epoch": 0.78, - "learning_rate": 1.8622910483544965e-05, - "loss": 0.7882, - "step": 381 - }, - { - "epoch": 0.78, - "learning_rate": 1.8614547352706865e-05, - "loss": 0.7347, - "step": 382 - }, - { - "epoch": 0.78, - "learning_rate": 1.8606160793494697e-05, - "loss": 0.7547, - "step": 383 - }, - { - "epoch": 0.78, - "learning_rate": 1.8597750828716798e-05, - "loss": 0.7694, - "step": 384 - }, - { - "epoch": 0.78, - "learning_rate": 1.8589317481245145e-05, - "loss": 0.7718, - "step": 385 - }, - { - "epoch": 0.79, - "learning_rate": 1.858086077401532e-05, - "loss": 0.7638, - "step": 386 - }, - { - "epoch": 0.79, - "learning_rate": 1.8572380730026432e-05, - "loss": 0.7791, - "step": 387 - }, - { - "epoch": 0.79, - "learning_rate": 1.856387737234105e-05, - "loss": 0.7682, - "step": 388 - }, - { - "epoch": 0.79, - "learning_rate": 1.855535072408516e-05, - "loss": 0.787, - "step": 389 - }, - { - "epoch": 0.79, - "learning_rate": 1.854680080844808e-05, - "loss": 0.744, - "step": 390 - }, - { - "epoch": 0.8, - "learning_rate": 1.853822764868241e-05, - "loss": 0.7629, - "step": 391 - }, - { - "epoch": 0.8, - "learning_rate": 1.8529631268103964e-05, - "loss": 0.7451, - "step": 392 - }, - { - "epoch": 0.8, - "learning_rate": 1.852101169009171e-05, - "loss": 0.7589, - "step": 393 - }, - { - "epoch": 0.8, - "learning_rate": 1.8512368938087703e-05, - "loss": 0.71, - "step": 394 - }, - { - "epoch": 0.8, - "learning_rate": 1.8503703035597026e-05, - "loss": 0.6967, - "step": 395 - }, - { - "epoch": 0.81, - "learning_rate": 1.8495014006187718e-05, - "loss": 0.7805, - "step": 396 - }, - { - "epoch": 0.81, - "learning_rate": 1.8486301873490715e-05, - "loss": 0.7716, - "step": 397 - }, - { - "epoch": 0.81, - "learning_rate": 1.8477566661199788e-05, - "loss": 0.7979, - "step": 398 - }, - { - "epoch": 0.81, - "learning_rate": 1.846880839307148e-05, - "loss": 0.7746, - "step": 399 - }, - { - "epoch": 0.81, - "learning_rate": 1.8460027092925027e-05, - "loss": 0.7748, - "step": 400 - }, - { - "epoch": 0.82, - "learning_rate": 1.8451222784642307e-05, - "loss": 0.709, - "step": 401 - }, - { - "epoch": 0.82, - "learning_rate": 1.8442395492167773e-05, - "loss": 0.801, - "step": 402 - }, - { - "epoch": 0.82, - "learning_rate": 1.84335452395084e-05, - "loss": 0.7219, - "step": 403 - }, - { - "epoch": 0.82, - "learning_rate": 1.8424672050733577e-05, - "loss": 0.7784, - "step": 404 - }, - { - "epoch": 0.82, - "learning_rate": 1.8415775949975093e-05, - "loss": 0.7591, - "step": 405 - }, - { - "epoch": 0.83, - "learning_rate": 1.8406856961427042e-05, - "loss": 0.7485, - "step": 406 - }, - { - "epoch": 0.83, - "learning_rate": 1.839791510934577e-05, - "loss": 0.7088, - "step": 407 - }, - { - "epoch": 0.83, - "learning_rate": 1.838895041804979e-05, - "loss": 0.7504, - "step": 408 - }, - { - "epoch": 0.83, - "learning_rate": 1.8379962911919747e-05, - "loss": 0.7271, - "step": 409 - }, - { - "epoch": 0.84, - "learning_rate": 1.837095261539832e-05, - "loss": 0.76, - "step": 410 - }, - { - "epoch": 0.84, - "learning_rate": 1.836191955299017e-05, - "loss": 0.7223, - "step": 411 - }, - { - "epoch": 0.84, - "learning_rate": 1.8352863749261883e-05, - "loss": 0.695, - "step": 412 - }, - { - "epoch": 0.84, - "learning_rate": 1.8343785228841886e-05, - "loss": 0.7105, - "step": 413 - }, - { - "epoch": 0.84, - "learning_rate": 1.8334684016420383e-05, - "loss": 0.8034, - "step": 414 - }, - { - "epoch": 0.85, - "learning_rate": 1.8325560136749302e-05, - "loss": 0.7227, - "step": 415 - }, - { - "epoch": 0.85, - "learning_rate": 1.8316413614642213e-05, - "loss": 0.7179, - "step": 416 - }, - { - "epoch": 0.85, - "learning_rate": 1.8307244474974254e-05, - "loss": 0.7139, - "step": 417 - }, - { - "epoch": 0.85, - "learning_rate": 1.82980527426821e-05, - "loss": 0.7845, - "step": 418 - }, - { - "epoch": 0.85, - "learning_rate": 1.8288838442763838e-05, - "loss": 0.7594, - "step": 419 - }, - { - "epoch": 0.86, - "learning_rate": 1.827960160027896e-05, - "loss": 0.7299, - "step": 420 - }, - { - "epoch": 0.86, - "learning_rate": 1.8270342240348243e-05, - "loss": 0.7318, - "step": 421 - }, - { - "epoch": 0.86, - "learning_rate": 1.8261060388153716e-05, - "loss": 0.7356, - "step": 422 - }, - { - "epoch": 0.86, - "learning_rate": 1.825175606893858e-05, - "loss": 0.668, - "step": 423 - }, - { - "epoch": 0.86, - "learning_rate": 1.8242429308007122e-05, - "loss": 0.7644, - "step": 424 - }, - { - "epoch": 0.87, - "learning_rate": 1.8233080130724677e-05, - "loss": 0.7282, - "step": 425 - }, - { - "epoch": 0.87, - "learning_rate": 1.8223708562517543e-05, - "loss": 0.7626, - "step": 426 - }, - { - "epoch": 0.87, - "learning_rate": 1.8214314628872908e-05, - "loss": 0.722, - "step": 427 - }, - { - "epoch": 0.87, - "learning_rate": 1.8204898355338785e-05, - "loss": 0.6762, - "step": 428 - }, - { - "epoch": 0.87, - "learning_rate": 1.8195459767523953e-05, - "loss": 0.7285, - "step": 429 - }, - { - "epoch": 0.88, - "learning_rate": 1.8185998891097864e-05, - "loss": 0.7727, - "step": 430 - }, - { - "epoch": 0.88, - "learning_rate": 1.8176515751790593e-05, - "loss": 0.7145, - "step": 431 - }, - { - "epoch": 0.88, - "learning_rate": 1.816701037539277e-05, - "loss": 0.7315, - "step": 432 - }, - { - "epoch": 0.88, - "learning_rate": 1.8157482787755485e-05, - "loss": 0.7276, - "step": 433 - }, - { - "epoch": 0.88, - "learning_rate": 1.8147933014790245e-05, - "loss": 0.7078, - "step": 434 - }, - { - "epoch": 0.89, - "learning_rate": 1.8138361082468895e-05, - "loss": 0.7396, - "step": 435 - }, - { - "epoch": 0.89, - "learning_rate": 1.812876701682354e-05, - "loss": 0.7052, - "step": 436 - }, - { - "epoch": 0.89, - "learning_rate": 1.8119150843946475e-05, - "loss": 0.7193, - "step": 437 - }, - { - "epoch": 0.89, - "learning_rate": 1.810951258999013e-05, - "loss": 0.7612, - "step": 438 - }, - { - "epoch": 0.89, - "learning_rate": 1.8099852281166974e-05, - "loss": 0.6896, - "step": 439 - }, - { - "epoch": 0.9, - "learning_rate": 1.8090169943749477e-05, - "loss": 0.724, - "step": 440 - }, - { - "epoch": 0.9, - "learning_rate": 1.8080465604069998e-05, - "loss": 0.6851, - "step": 441 - }, - { - "epoch": 0.9, - "learning_rate": 1.8070739288520736e-05, - "loss": 0.6831, - "step": 442 - }, - { - "epoch": 0.9, - "learning_rate": 1.8060991023553676e-05, - "loss": 0.7354, - "step": 443 - }, - { - "epoch": 0.9, - "learning_rate": 1.8051220835680475e-05, - "loss": 0.7143, - "step": 444 - }, - { - "epoch": 0.91, - "learning_rate": 1.804142875147242e-05, - "loss": 0.6819, - "step": 445 - }, - { - "epoch": 0.91, - "learning_rate": 1.8031614797560352e-05, - "loss": 0.7242, - "step": 446 - }, - { - "epoch": 0.91, - "learning_rate": 1.802177900063459e-05, - "loss": 0.6635, - "step": 447 - }, - { - "epoch": 0.91, - "learning_rate": 1.8011921387444844e-05, - "loss": 0.6626, - "step": 448 - }, - { - "epoch": 0.91, - "learning_rate": 1.8002041984800173e-05, - "loss": 0.6677, - "step": 449 - }, - { - "epoch": 0.92, - "learning_rate": 1.7992140819568886e-05, - "loss": 0.6981, - "step": 450 - }, - { - "epoch": 0.92, - "learning_rate": 1.7982217918678482e-05, - "loss": 0.6492, - "step": 451 - }, - { - "epoch": 0.92, - "learning_rate": 1.797227330911557e-05, - "loss": 0.6833, - "step": 452 - }, - { - "epoch": 0.92, - "learning_rate": 1.7962307017925797e-05, - "loss": 0.6581, - "step": 453 - }, - { - "epoch": 0.92, - "learning_rate": 1.795231907221379e-05, - "loss": 0.676, - "step": 454 - }, - { - "epoch": 0.93, - "learning_rate": 1.7942309499143045e-05, - "loss": 0.6772, - "step": 455 - }, - { - "epoch": 0.93, - "learning_rate": 1.7932278325935894e-05, - "loss": 0.6916, - "step": 456 - }, - { - "epoch": 0.93, - "learning_rate": 1.7922225579873407e-05, - "loss": 0.6917, - "step": 457 - }, - { - "epoch": 0.93, - "learning_rate": 1.7912151288295328e-05, - "loss": 0.6584, - "step": 458 - }, - { - "epoch": 0.93, - "learning_rate": 1.790205547859999e-05, - "loss": 0.664, - "step": 459 - }, - { - "epoch": 0.94, - "learning_rate": 1.7891938178244258e-05, - "loss": 0.6605, - "step": 460 - }, - { - "epoch": 0.94, - "learning_rate": 1.7881799414743428e-05, - "loss": 0.6385, - "step": 461 - }, - { - "epoch": 0.94, - "learning_rate": 1.787163921567118e-05, - "loss": 0.6828, - "step": 462 - }, - { - "epoch": 0.94, - "learning_rate": 1.786145760865949e-05, - "loss": 0.6547, - "step": 463 - }, - { - "epoch": 0.95, - "learning_rate": 1.785125462139855e-05, - "loss": 0.6328, - "step": 464 - }, - { - "epoch": 0.95, - "learning_rate": 1.7841030281636703e-05, - "loss": 0.6375, - "step": 465 - }, - { - "epoch": 0.95, - "learning_rate": 1.7830784617180367e-05, - "loss": 0.6098, - "step": 466 - }, - { - "epoch": 0.95, - "learning_rate": 1.782051765589394e-05, - "loss": 0.707, - "step": 467 - }, - { - "epoch": 0.95, - "learning_rate": 1.7810229425699752e-05, - "loss": 0.6587, - "step": 468 - }, - { - "epoch": 0.96, - "learning_rate": 1.7799919954577977e-05, - "loss": 0.7049, - "step": 469 - }, - { - "epoch": 0.96, - "learning_rate": 1.7789589270566548e-05, - "loss": 0.6766, - "step": 470 - }, - { - "epoch": 0.96, - "learning_rate": 1.7779237401761102e-05, - "loss": 0.631, - "step": 471 - }, - { - "epoch": 0.96, - "learning_rate": 1.7768864376314875e-05, - "loss": 0.7066, - "step": 472 - }, - { - "epoch": 0.96, - "learning_rate": 1.7758470222438656e-05, - "loss": 0.7099, - "step": 473 - }, - { - "epoch": 0.97, - "learning_rate": 1.7748054968400684e-05, - "loss": 0.6522, - "step": 474 - }, - { - "epoch": 0.97, - "learning_rate": 1.773761864252659e-05, - "loss": 0.6697, - "step": 475 - }, - { - "epoch": 0.97, - "learning_rate": 1.772716127319931e-05, - "loss": 0.6838, - "step": 476 - }, - { - "epoch": 0.97, - "learning_rate": 1.771668288885901e-05, - "loss": 0.7187, - "step": 477 - }, - { - "epoch": 0.97, - "learning_rate": 1.770618351800301e-05, - "loss": 0.6507, - "step": 478 - }, - { - "epoch": 0.98, - "learning_rate": 1.7695663189185703e-05, - "loss": 0.6391, - "step": 479 - }, - { - "epoch": 0.98, - "learning_rate": 1.768512193101848e-05, - "loss": 0.6769, - "step": 480 - }, - { - "epoch": 0.98, - "learning_rate": 1.767455977216966e-05, - "loss": 0.6534, - "step": 481 - }, - { - "epoch": 0.98, - "learning_rate": 1.7663976741364394e-05, - "loss": 0.6785, - "step": 482 - }, - { - "epoch": 0.98, - "learning_rate": 1.7653372867384604e-05, - "loss": 0.6531, - "step": 483 - }, - { - "epoch": 0.99, - "learning_rate": 1.7642748179068892e-05, - "loss": 0.6561, - "step": 484 - }, - { - "epoch": 0.99, - "learning_rate": 1.7632102705312472e-05, - "loss": 0.6727, - "step": 485 - }, - { - "epoch": 0.99, - "learning_rate": 1.762143647506708e-05, - "loss": 0.6567, - "step": 486 - }, - { - "epoch": 0.99, - "learning_rate": 1.7610749517340915e-05, - "loss": 0.7389, - "step": 487 - }, - { - "epoch": 0.99, - "learning_rate": 1.7600041861198533e-05, - "loss": 0.6242, - "step": 488 - }, - { - "epoch": 1.0, - "learning_rate": 1.7589313535760787e-05, - "loss": 0.6881, - "step": 489 - }, - { - "epoch": 1.0, - "learning_rate": 1.7578564570204745e-05, - "loss": 0.6379, - "step": 490 - }, - { - "epoch": 1.0, - "learning_rate": 1.7567794993763606e-05, - "loss": 0.652, - "step": 491 - }, - { - "epoch": 1.0, - "learning_rate": 1.755700483572663e-05, - "loss": 0.4741, - "step": 492 - }, - { - "epoch": 1.0, - "learning_rate": 1.754619412543904e-05, - "loss": 0.4143, - "step": 493 - }, - { - "epoch": 1.01, - "learning_rate": 1.7535362892301953e-05, - "loss": 0.4066, - "step": 494 - }, - { - "epoch": 1.01, - "learning_rate": 1.752451116577232e-05, - "loss": 0.4067, - "step": 495 - }, - { - "epoch": 1.01, - "learning_rate": 1.7513638975362804e-05, - "loss": 0.4511, - "step": 496 - }, - { - "epoch": 1.01, - "learning_rate": 1.750274635064173e-05, - "loss": 0.379, - "step": 497 - }, - { - "epoch": 1.01, - "learning_rate": 1.7491833321233004e-05, - "loss": 0.4556, - "step": 498 - }, - { - "epoch": 1.02, - "learning_rate": 1.7480899916816012e-05, - "loss": 0.4033, - "step": 499 - }, - { - "epoch": 1.02, - "learning_rate": 1.7469946167125563e-05, - "loss": 0.413, - "step": 500 - }, - { - "epoch": 1.02, - "learning_rate": 1.7458972101951796e-05, - "loss": 0.3835, - "step": 501 - }, - { - "epoch": 1.02, - "learning_rate": 1.744797775114009e-05, - "loss": 0.4434, - "step": 502 - }, - { - "epoch": 1.02, - "learning_rate": 1.743696314459101e-05, - "loss": 0.42, - "step": 503 - }, - { - "epoch": 1.03, - "learning_rate": 1.74259283122602e-05, - "loss": 0.3905, - "step": 504 - }, - { - "epoch": 1.03, - "learning_rate": 1.7414873284158308e-05, - "loss": 0.4376, - "step": 505 - }, - { - "epoch": 1.03, - "learning_rate": 1.7403798090350913e-05, - "loss": 0.4484, - "step": 506 - }, - { - "epoch": 1.03, - "learning_rate": 1.739270276095844e-05, - "loss": 0.4033, - "step": 507 - }, - { - "epoch": 1.03, - "learning_rate": 1.738158732615606e-05, - "loss": 0.3777, - "step": 508 - }, - { - "epoch": 1.04, - "learning_rate": 1.737045181617364e-05, - "loss": 0.3844, - "step": 509 - }, - { - "epoch": 1.04, - "learning_rate": 1.7359296261295635e-05, - "loss": 0.3808, - "step": 510 - }, - { - "epoch": 1.04, - "learning_rate": 1.734812069186102e-05, - "loss": 0.3667, - "step": 511 - }, - { - "epoch": 1.04, - "learning_rate": 1.7336925138263195e-05, - "loss": 0.4187, - "step": 512 - }, - { - "epoch": 1.04, - "learning_rate": 1.732570963094992e-05, - "loss": 0.3962, - "step": 513 - }, - { - "epoch": 1.05, - "learning_rate": 1.731447420042321e-05, - "loss": 0.3973, - "step": 514 - }, - { - "epoch": 1.05, - "learning_rate": 1.730321887723927e-05, - "loss": 0.3948, - "step": 515 - }, - { - "epoch": 1.05, - "learning_rate": 1.729194369200841e-05, - "loss": 0.3986, - "step": 516 - }, - { - "epoch": 1.05, - "learning_rate": 1.7280648675394948e-05, - "loss": 0.407, - "step": 517 - }, - { - "epoch": 1.05, - "learning_rate": 1.7269333858117137e-05, - "loss": 0.4106, - "step": 518 - }, - { - "epoch": 1.06, - "learning_rate": 1.7257999270947094e-05, - "loss": 0.4081, - "step": 519 - }, - { - "epoch": 1.06, - "learning_rate": 1.7246644944710678e-05, - "loss": 0.3864, - "step": 520 - }, - { - "epoch": 1.06, - "learning_rate": 1.7235270910287462e-05, - "loss": 0.4256, - "step": 521 - }, - { - "epoch": 1.06, - "learning_rate": 1.722387719861059e-05, - "loss": 0.3976, - "step": 522 - }, - { - "epoch": 1.07, - "learning_rate": 1.721246384066674e-05, - "loss": 0.4158, - "step": 523 - }, - { - "epoch": 1.07, - "learning_rate": 1.7201030867496005e-05, - "loss": 0.4193, - "step": 524 - }, - { - "epoch": 1.07, - "learning_rate": 1.718957831019184e-05, - "loss": 0.4395, - "step": 525 - }, - { - "epoch": 1.07, - "learning_rate": 1.7178106199900945e-05, - "loss": 0.3863, - "step": 526 - }, - { - "epoch": 1.07, - "learning_rate": 1.7166614567823214e-05, - "loss": 0.395, - "step": 527 - }, - { - "epoch": 1.08, - "learning_rate": 1.715510344521162e-05, - "loss": 0.428, - "step": 528 - }, - { - "epoch": 1.08, - "learning_rate": 1.7143572863372148e-05, - "loss": 0.4137, - "step": 529 - }, - { - "epoch": 1.08, - "learning_rate": 1.7132022853663706e-05, - "loss": 0.4014, - "step": 530 - }, - { - "epoch": 1.08, - "learning_rate": 1.7120453447498034e-05, - "loss": 0.4183, - "step": 531 - }, - { - "epoch": 1.08, - "learning_rate": 1.7108864676339626e-05, - "loss": 0.4036, - "step": 532 - }, - { - "epoch": 1.09, - "learning_rate": 1.709725657170565e-05, - "loss": 0.3733, - "step": 533 - }, - { - "epoch": 1.09, - "learning_rate": 1.7085629165165835e-05, - "loss": 0.4416, - "step": 534 - }, - { - "epoch": 1.09, - "learning_rate": 1.7073982488342423e-05, - "loss": 0.4209, - "step": 535 - }, - { - "epoch": 1.09, - "learning_rate": 1.7062316572910048e-05, - "loss": 0.4353, - "step": 536 - }, - { - "epoch": 1.09, - "learning_rate": 1.705063145059568e-05, - "loss": 0.3851, - "step": 537 - }, - { - "epoch": 1.1, - "learning_rate": 1.7038927153178517e-05, - "loss": 0.3839, - "step": 538 - }, - { - "epoch": 1.1, - "learning_rate": 1.7027203712489902e-05, - "loss": 0.3858, - "step": 539 - }, - { - "epoch": 1.1, - "learning_rate": 1.7015461160413254e-05, - "loss": 0.4118, - "step": 540 - }, - { - "epoch": 1.1, - "learning_rate": 1.7003699528883955e-05, - "loss": 0.4058, - "step": 541 - }, - { - "epoch": 1.1, - "learning_rate": 1.6991918849889283e-05, - "loss": 0.4174, - "step": 542 - }, - { - "epoch": 1.11, - "learning_rate": 1.6980119155468316e-05, - "loss": 0.4223, - "step": 543 - }, - { - "epoch": 1.11, - "learning_rate": 1.6968300477711842e-05, - "loss": 0.43, - "step": 544 - }, - { - "epoch": 1.11, - "learning_rate": 1.695646284876229e-05, - "loss": 0.3713, - "step": 545 - }, - { - "epoch": 1.11, - "learning_rate": 1.6944606300813614e-05, - "loss": 0.4116, - "step": 546 - }, - { - "epoch": 1.11, - "learning_rate": 1.693273086611123e-05, - "loss": 0.4098, - "step": 547 - }, - { - "epoch": 1.12, - "learning_rate": 1.692083657695192e-05, - "loss": 0.3948, - "step": 548 - }, - { - "epoch": 1.12, - "learning_rate": 1.6908923465683728e-05, - "loss": 0.3893, - "step": 549 - }, - { - "epoch": 1.12, - "learning_rate": 1.6896991564705912e-05, - "loss": 0.3987, - "step": 550 - }, - { - "epoch": 1.12, - "learning_rate": 1.6885040906468807e-05, - "loss": 0.4304, - "step": 551 - }, - { - "epoch": 1.12, - "learning_rate": 1.687307152347378e-05, - "loss": 0.4072, - "step": 552 - }, - { - "epoch": 1.13, - "learning_rate": 1.686108344827311e-05, - "loss": 0.3824, - "step": 553 - }, - { - "epoch": 1.13, - "learning_rate": 1.6849076713469914e-05, - "loss": 0.4114, - "step": 554 - }, - { - "epoch": 1.13, - "learning_rate": 1.683705135171806e-05, - "loss": 0.4138, - "step": 555 - }, - { - "epoch": 1.13, - "learning_rate": 1.682500739572207e-05, - "loss": 0.4236, - "step": 556 - }, - { - "epoch": 1.13, - "learning_rate": 1.681294487823704e-05, - "loss": 0.41, - "step": 557 - }, - { - "epoch": 1.14, - "learning_rate": 1.680086383206855e-05, - "loss": 0.3888, - "step": 558 - }, - { - "epoch": 1.14, - "learning_rate": 1.6788764290072554e-05, - "loss": 0.3719, - "step": 559 - }, - { - "epoch": 1.14, - "learning_rate": 1.6776646285155326e-05, - "loss": 0.3711, - "step": 560 - }, - { - "epoch": 1.14, - "learning_rate": 1.676450985027334e-05, - "loss": 0.4085, - "step": 561 - }, - { - "epoch": 1.14, - "learning_rate": 1.6752355018433205e-05, - "loss": 0.4091, - "step": 562 - }, - { - "epoch": 1.15, - "learning_rate": 1.6740181822691556e-05, - "loss": 0.4123, - "step": 563 - }, - { - "epoch": 1.15, - "learning_rate": 1.6727990296154962e-05, - "loss": 0.4236, - "step": 564 - }, - { - "epoch": 1.15, - "learning_rate": 1.6715780471979862e-05, - "loss": 0.4299, - "step": 565 - }, - { - "epoch": 1.15, - "learning_rate": 1.6703552383372447e-05, - "loss": 0.4287, - "step": 566 - }, - { - "epoch": 1.15, - "learning_rate": 1.6691306063588583e-05, - "loss": 0.3933, - "step": 567 - }, - { - "epoch": 1.16, - "learning_rate": 1.667904154593372e-05, - "loss": 0.3975, - "step": 568 - }, - { - "epoch": 1.16, - "learning_rate": 1.6666758863762796e-05, - "loss": 0.4401, - "step": 569 - }, - { - "epoch": 1.16, - "learning_rate": 1.665445805048015e-05, - "loss": 0.3917, - "step": 570 - }, - { - "epoch": 1.16, - "learning_rate": 1.6642139139539436e-05, - "loss": 0.418, - "step": 571 - }, - { - "epoch": 1.16, - "learning_rate": 1.6629802164443522e-05, - "loss": 0.3824, - "step": 572 - }, - { - "epoch": 1.17, - "learning_rate": 1.66174471587444e-05, - "loss": 0.4081, - "step": 573 - }, - { - "epoch": 1.17, - "learning_rate": 1.660507415604311e-05, - "loss": 0.3873, - "step": 574 - }, - { - "epoch": 1.17, - "learning_rate": 1.659268318998963e-05, - "loss": 0.3857, - "step": 575 - }, - { - "epoch": 1.17, - "learning_rate": 1.658027429428279e-05, - "loss": 0.4197, - "step": 576 - }, - { - "epoch": 1.18, - "learning_rate": 1.6567847502670187e-05, - "loss": 0.4096, - "step": 577 - }, - { - "epoch": 1.18, - "learning_rate": 1.6555402848948085e-05, - "loss": 0.4165, - "step": 578 - }, - { - "epoch": 1.18, - "learning_rate": 1.6542940366961327e-05, - "loss": 0.3448, - "step": 579 - }, - { - "epoch": 1.18, - "learning_rate": 1.6530460090603236e-05, - "loss": 0.4352, - "step": 580 - }, - { - "epoch": 1.18, - "learning_rate": 1.6517962053815547e-05, - "loss": 0.4286, - "step": 581 - }, - { - "epoch": 1.19, - "learning_rate": 1.650544629058828e-05, - "loss": 0.4075, - "step": 582 - }, - { - "epoch": 1.19, - "learning_rate": 1.6492912834959664e-05, - "loss": 0.4105, - "step": 583 - }, - { - "epoch": 1.19, - "learning_rate": 1.6480361721016053e-05, - "loss": 0.3789, - "step": 584 - }, - { - "epoch": 1.19, - "learning_rate": 1.6467792982891828e-05, - "loss": 0.3806, - "step": 585 - }, - { - "epoch": 1.19, - "learning_rate": 1.645520665476929e-05, - "loss": 0.3876, - "step": 586 - }, - { - "epoch": 1.2, - "learning_rate": 1.6442602770878586e-05, - "loss": 0.4095, - "step": 587 - }, - { - "epoch": 1.2, - "learning_rate": 1.6429981365497605e-05, - "loss": 0.417, - "step": 588 - }, - { - "epoch": 1.2, - "learning_rate": 1.641734247295189e-05, - "loss": 0.3941, - "step": 589 - }, - { - "epoch": 1.2, - "learning_rate": 1.640468612761454e-05, - "loss": 0.3977, - "step": 590 - }, - { - "epoch": 1.2, - "learning_rate": 1.639201236390612e-05, - "loss": 0.3921, - "step": 591 - }, - { - "epoch": 1.21, - "learning_rate": 1.6379321216294574e-05, - "loss": 0.3828, - "step": 592 - }, - { - "epoch": 1.21, - "learning_rate": 1.6366612719295106e-05, - "loss": 0.4042, - "step": 593 - }, - { - "epoch": 1.21, - "learning_rate": 1.6353886907470125e-05, - "loss": 0.4176, - "step": 594 - }, - { - "epoch": 1.21, - "learning_rate": 1.6341143815429114e-05, - "loss": 0.4227, - "step": 595 - }, - { - "epoch": 1.21, - "learning_rate": 1.6328383477828556e-05, - "loss": 0.4102, - "step": 596 - }, - { - "epoch": 1.22, - "learning_rate": 1.631560592937184e-05, - "loss": 0.4132, - "step": 597 - }, - { - "epoch": 1.22, - "learning_rate": 1.6302811204809164e-05, - "loss": 0.4208, - "step": 598 - }, - { - "epoch": 1.22, - "learning_rate": 1.6289999338937427e-05, - "loss": 0.4148, - "step": 599 - }, - { - "epoch": 1.22, - "learning_rate": 1.627717036660015e-05, - "loss": 0.416, - "step": 600 - }, - { - "epoch": 1.22, - "learning_rate": 1.626432432268739e-05, - "loss": 0.4171, - "step": 601 - }, - { - "epoch": 1.23, - "learning_rate": 1.625146124213562e-05, - "loss": 0.4235, - "step": 602 - }, - { - "epoch": 1.23, - "learning_rate": 1.623858115992764e-05, - "loss": 0.4142, - "step": 603 - }, - { - "epoch": 1.23, - "learning_rate": 1.6225684111092507e-05, - "loss": 0.3989, - "step": 604 - }, - { - "epoch": 1.23, - "learning_rate": 1.6212770130705403e-05, - "loss": 0.3831, - "step": 605 - }, - { - "epoch": 1.23, - "learning_rate": 1.6199839253887572e-05, - "loss": 0.4487, - "step": 606 - }, - { - "epoch": 1.24, - "learning_rate": 1.6186891515806203e-05, - "loss": 0.4082, - "step": 607 - }, - { - "epoch": 1.24, - "learning_rate": 1.6173926951674336e-05, - "loss": 0.3906, - "step": 608 - }, - { - "epoch": 1.24, - "learning_rate": 1.6160945596750784e-05, - "loss": 0.3706, - "step": 609 - }, - { - "epoch": 1.24, - "learning_rate": 1.6147947486340013e-05, - "loss": 0.3725, - "step": 610 - }, - { - "epoch": 1.24, - "learning_rate": 1.613493265579207e-05, - "loss": 0.4445, - "step": 611 - }, - { - "epoch": 1.25, - "learning_rate": 1.612190114050246e-05, - "loss": 0.45, - "step": 612 - }, - { - "epoch": 1.25, - "learning_rate": 1.610885297591207e-05, - "loss": 0.3583, - "step": 613 - }, - { - "epoch": 1.25, - "learning_rate": 1.609578819750708e-05, - "loss": 0.3921, - "step": 614 - }, - { - "epoch": 1.25, - "learning_rate": 1.6082706840818833e-05, - "loss": 0.407, - "step": 615 - }, - { - "epoch": 1.25, - "learning_rate": 1.6069608941423773e-05, - "loss": 0.3846, - "step": 616 - }, - { - "epoch": 1.26, - "learning_rate": 1.6056494534943324e-05, - "loss": 0.3967, - "step": 617 - }, - { - "epoch": 1.26, - "learning_rate": 1.604336365704381e-05, - "loss": 0.3993, - "step": 618 - }, - { - "epoch": 1.26, - "learning_rate": 1.6030216343436354e-05, - "loss": 0.4293, - "step": 619 - }, - { - "epoch": 1.26, - "learning_rate": 1.6017052629876768e-05, - "loss": 0.4028, - "step": 620 - }, - { - "epoch": 1.26, - "learning_rate": 1.600387255216547e-05, - "loss": 0.3837, - "step": 621 - }, - { - "epoch": 1.27, - "learning_rate": 1.5990676146147387e-05, - "loss": 0.4077, - "step": 622 - }, - { - "epoch": 1.27, - "learning_rate": 1.5977463447711844e-05, - "loss": 0.4141, - "step": 623 - }, - { - "epoch": 1.27, - "learning_rate": 1.596423449279249e-05, - "loss": 0.405, - "step": 624 - }, - { - "epoch": 1.27, - "learning_rate": 1.5950989317367172e-05, - "loss": 0.3987, - "step": 625 - }, - { - "epoch": 1.27, - "learning_rate": 1.5937727957457853e-05, - "loss": 0.389, - "step": 626 - }, - { - "epoch": 1.28, - "learning_rate": 1.5924450449130512e-05, - "loss": 0.3892, - "step": 627 - }, - { - "epoch": 1.28, - "learning_rate": 1.5911156828495056e-05, - "loss": 0.4124, - "step": 628 - }, - { - "epoch": 1.28, - "learning_rate": 1.5897847131705194e-05, - "loss": 0.3971, - "step": 629 - }, - { - "epoch": 1.28, - "learning_rate": 1.5884521394958376e-05, - "loss": 0.3709, - "step": 630 - }, - { - "epoch": 1.29, - "learning_rate": 1.587117965449565e-05, - "loss": 0.4037, - "step": 631 - }, - { - "epoch": 1.29, - "learning_rate": 1.5857821946601616e-05, - "loss": 0.4212, - "step": 632 - }, - { - "epoch": 1.29, - "learning_rate": 1.5844448307604275e-05, - "loss": 0.3914, - "step": 633 - }, - { - "epoch": 1.29, - "learning_rate": 1.5831058773874973e-05, - "loss": 0.4078, - "step": 634 - }, - { - "epoch": 1.29, - "learning_rate": 1.581765338182827e-05, - "loss": 0.377, - "step": 635 - }, - { - "epoch": 1.3, - "learning_rate": 1.5804232167921867e-05, - "loss": 0.4195, - "step": 636 - }, - { - "epoch": 1.3, - "learning_rate": 1.5790795168656485e-05, - "loss": 0.4138, - "step": 637 - }, - { - "epoch": 1.3, - "learning_rate": 1.5777342420575782e-05, - "loss": 0.3912, - "step": 638 - }, - { - "epoch": 1.3, - "learning_rate": 1.5763873960266236e-05, - "loss": 0.4352, - "step": 639 - }, - { - "epoch": 1.3, - "learning_rate": 1.5750389824357073e-05, - "loss": 0.4111, - "step": 640 - }, - { - "epoch": 1.31, - "learning_rate": 1.5736890049520136e-05, - "loss": 0.3796, - "step": 641 - }, - { - "epoch": 1.31, - "learning_rate": 1.5723374672469812e-05, - "loss": 0.4206, - "step": 642 - }, - { - "epoch": 1.31, - "learning_rate": 1.5709843729962905e-05, - "loss": 0.3936, - "step": 643 - }, - { - "epoch": 1.31, - "learning_rate": 1.5696297258798573e-05, - "loss": 0.3982, - "step": 644 - }, - { - "epoch": 1.31, - "learning_rate": 1.5682735295818187e-05, - "loss": 0.4093, - "step": 645 - }, - { - "epoch": 1.32, - "learning_rate": 1.5669157877905253e-05, - "loss": 0.4142, - "step": 646 - }, - { - "epoch": 1.32, - "learning_rate": 1.565556504198532e-05, - "loss": 0.3829, - "step": 647 - }, - { - "epoch": 1.32, - "learning_rate": 1.5641956825025856e-05, - "loss": 0.3993, - "step": 648 - }, - { - "epoch": 1.32, - "learning_rate": 1.5628333264036166e-05, - "loss": 0.3885, - "step": 649 - }, - { - "epoch": 1.32, - "learning_rate": 1.5614694396067285e-05, - "loss": 0.3801, - "step": 650 - }, - { - "epoch": 1.33, - "learning_rate": 1.5601040258211873e-05, - "loss": 0.3981, - "step": 651 - }, - { - "epoch": 1.33, - "learning_rate": 1.558737088760412e-05, - "loss": 0.3494, - "step": 652 - }, - { - "epoch": 1.33, - "learning_rate": 1.5573686321419653e-05, - "loss": 0.4093, - "step": 653 - }, - { - "epoch": 1.33, - "learning_rate": 1.555998659687541e-05, - "loss": 0.4164, - "step": 654 - }, - { - "epoch": 1.33, - "learning_rate": 1.5546271751229558e-05, - "loss": 0.3825, - "step": 655 - }, - { - "epoch": 1.34, - "learning_rate": 1.55325418217814e-05, - "loss": 0.4061, - "step": 656 - }, - { - "epoch": 1.34, - "learning_rate": 1.5518796845871245e-05, - "loss": 0.398, - "step": 657 - }, - { - "epoch": 1.34, - "learning_rate": 1.5505036860880337e-05, - "loss": 0.3714, - "step": 658 - }, - { - "epoch": 1.34, - "learning_rate": 1.549126190423073e-05, - "loss": 0.4003, - "step": 659 - }, - { - "epoch": 1.34, - "learning_rate": 1.5477472013385186e-05, - "loss": 0.3866, - "step": 660 - }, - { - "epoch": 1.35, - "learning_rate": 1.546366722584711e-05, - "loss": 0.3922, - "step": 661 - }, - { - "epoch": 1.35, - "learning_rate": 1.5449847579160398e-05, - "loss": 0.4227, - "step": 662 - }, - { - "epoch": 1.35, - "learning_rate": 1.5436013110909364e-05, - "loss": 0.3706, - "step": 663 - }, - { - "epoch": 1.35, - "learning_rate": 1.5422163858718632e-05, - "loss": 0.3947, - "step": 664 - }, - { - "epoch": 1.35, - "learning_rate": 1.5408299860253034e-05, - "loss": 0.3606, - "step": 665 - }, - { - "epoch": 1.36, - "learning_rate": 1.53944211532175e-05, - "loss": 0.3956, - "step": 666 - }, - { - "epoch": 1.36, - "learning_rate": 1.5380527775356965e-05, - "loss": 0.397, - "step": 667 - }, - { - "epoch": 1.36, - "learning_rate": 1.536661976445627e-05, - "loss": 0.4343, - "step": 668 - }, - { - "epoch": 1.36, - "learning_rate": 1.535269715834004e-05, - "loss": 0.3731, - "step": 669 - }, - { - "epoch": 1.36, - "learning_rate": 1.533875999487261e-05, - "loss": 0.4227, - "step": 670 - }, - { - "epoch": 1.37, - "learning_rate": 1.532480831195789e-05, - "loss": 0.396, - "step": 671 - }, - { - "epoch": 1.37, - "learning_rate": 1.531084214753928e-05, - "loss": 0.368, - "step": 672 - }, - { - "epoch": 1.37, - "learning_rate": 1.5296861539599573e-05, - "loss": 0.3873, - "step": 673 - }, - { - "epoch": 1.37, - "learning_rate": 1.5282866526160837e-05, - "loss": 0.3845, - "step": 674 - }, - { - "epoch": 1.37, - "learning_rate": 1.526885714528432e-05, - "loss": 0.3935, - "step": 675 - }, - { - "epoch": 1.38, - "learning_rate": 1.525483343507034e-05, - "loss": 0.3885, - "step": 676 - }, - { - "epoch": 1.38, - "learning_rate": 1.5240795433658187e-05, - "loss": 0.4284, - "step": 677 - }, - { - "epoch": 1.38, - "learning_rate": 1.5226743179226021e-05, - "loss": 0.4012, - "step": 678 - }, - { - "epoch": 1.38, - "learning_rate": 1.5212676709990762e-05, - "loss": 0.3944, - "step": 679 - }, - { - "epoch": 1.38, - "learning_rate": 1.5198596064207989e-05, - "loss": 0.4413, - "step": 680 - }, - { - "epoch": 1.39, - "learning_rate": 1.5184501280171839e-05, - "loss": 0.4094, - "step": 681 - }, - { - "epoch": 1.39, - "learning_rate": 1.5170392396214897e-05, - "loss": 0.3812, - "step": 682 - }, - { - "epoch": 1.39, - "learning_rate": 1.5156269450708094e-05, - "loss": 0.4115, - "step": 683 - }, - { - "epoch": 1.39, - "learning_rate": 1.5142132482060604e-05, - "loss": 0.3791, - "step": 684 - }, - { - "epoch": 1.4, - "learning_rate": 1.5127981528719735e-05, - "loss": 0.4173, - "step": 685 - }, - { - "epoch": 1.4, - "learning_rate": 1.5113816629170843e-05, - "loss": 0.3955, - "step": 686 - }, - { - "epoch": 1.4, - "learning_rate": 1.5099637821937192e-05, - "loss": 0.3982, - "step": 687 - }, - { - "epoch": 1.4, - "learning_rate": 1.5085445145579884e-05, - "loss": 0.3697, - "step": 688 - }, - { - "epoch": 1.4, - "learning_rate": 1.5071238638697731e-05, - "loss": 0.4247, - "step": 689 - }, - { - "epoch": 1.41, - "learning_rate": 1.505701833992717e-05, - "loss": 0.3731, - "step": 690 - }, - { - "epoch": 1.41, - "learning_rate": 1.5042784287942134e-05, - "loss": 0.4055, - "step": 691 - }, - { - "epoch": 1.41, - "learning_rate": 1.5028536521453969e-05, - "loss": 0.367, - "step": 692 - }, - { - "epoch": 1.41, - "learning_rate": 1.501427507921131e-05, - "loss": 0.3756, - "step": 693 - }, - { - "epoch": 1.41, - "learning_rate": 1.5000000000000002e-05, - "loss": 0.3812, - "step": 694 - }, - { - "epoch": 1.42, - "learning_rate": 1.4985711322642957e-05, - "loss": 0.4041, - "step": 695 - }, - { - "epoch": 1.42, - "learning_rate": 1.4971409086000082e-05, - "loss": 0.3897, - "step": 696 - }, - { - "epoch": 1.42, - "learning_rate": 1.4957093328968158e-05, - "loss": 0.3988, - "step": 697 - }, - { - "epoch": 1.42, - "learning_rate": 1.494276409048073e-05, - "loss": 0.3852, - "step": 698 - }, - { - "epoch": 1.42, - "learning_rate": 1.4928421409508017e-05, - "loss": 0.3719, - "step": 699 - }, - { - "epoch": 1.43, - "learning_rate": 1.4914065325056794e-05, - "loss": 0.3902, - "step": 700 - }, - { - "epoch": 1.43, - "learning_rate": 1.4899695876170283e-05, - "loss": 0.3912, - "step": 701 - }, - { - "epoch": 1.43, - "learning_rate": 1.4885313101928055e-05, - "loss": 0.393, - "step": 702 - }, - { - "epoch": 1.43, - "learning_rate": 1.4870917041445927e-05, - "loss": 0.3956, - "step": 703 - }, - { - "epoch": 1.43, - "learning_rate": 1.4856507733875837e-05, - "loss": 0.385, - "step": 704 - }, - { - "epoch": 1.44, - "learning_rate": 1.4842085218405767e-05, - "loss": 0.3933, - "step": 705 - }, - { - "epoch": 1.44, - "learning_rate": 1.4827649534259605e-05, - "loss": 0.4022, - "step": 706 - }, - { - "epoch": 1.44, - "learning_rate": 1.4813200720697056e-05, - "loss": 0.4203, - "step": 707 - }, - { - "epoch": 1.44, - "learning_rate": 1.4798738817013542e-05, - "loss": 0.3917, - "step": 708 - }, - { - "epoch": 1.44, - "learning_rate": 1.4784263862540073e-05, - "loss": 0.3748, - "step": 709 - }, - { - "epoch": 1.45, - "learning_rate": 1.4769775896643157e-05, - "loss": 0.373, - "step": 710 - }, - { - "epoch": 1.45, - "learning_rate": 1.4755274958724688e-05, - "loss": 0.3999, - "step": 711 - }, - { - "epoch": 1.45, - "learning_rate": 1.474076108822184e-05, - "loss": 0.4174, - "step": 712 - }, - { - "epoch": 1.45, - "learning_rate": 1.472623432460696e-05, - "loss": 0.4, - "step": 713 - }, - { - "epoch": 1.45, - "learning_rate": 1.4711694707387459e-05, - "loss": 0.3999, - "step": 714 - }, - { - "epoch": 1.46, - "learning_rate": 1.4697142276105699e-05, - "loss": 0.393, - "step": 715 - }, - { - "epoch": 1.46, - "learning_rate": 1.4682577070338898e-05, - "loss": 0.397, - "step": 716 - }, - { - "epoch": 1.46, - "learning_rate": 1.4667999129699013e-05, - "loss": 0.3954, - "step": 717 - }, - { - "epoch": 1.46, - "learning_rate": 1.4653408493832639e-05, - "loss": 0.3923, - "step": 718 - }, - { - "epoch": 1.46, - "learning_rate": 1.4638805202420896e-05, - "loss": 0.3803, - "step": 719 - }, - { - "epoch": 1.47, - "learning_rate": 1.4624189295179316e-05, - "loss": 0.3623, - "step": 720 - }, - { - "epoch": 1.47, - "learning_rate": 1.4609560811857746e-05, - "loss": 0.3787, - "step": 721 - }, - { - "epoch": 1.47, - "learning_rate": 1.4594919792240244e-05, - "loss": 0.4042, - "step": 722 - }, - { - "epoch": 1.47, - "learning_rate": 1.4580266276144945e-05, - "loss": 0.3826, - "step": 723 - }, - { - "epoch": 1.47, - "learning_rate": 1.4565600303423981e-05, - "loss": 0.3857, - "step": 724 - }, - { - "epoch": 1.48, - "learning_rate": 1.4550921913963364e-05, - "loss": 0.3936, - "step": 725 - }, - { - "epoch": 1.48, - "learning_rate": 1.453623114768286e-05, - "loss": 0.3788, - "step": 726 - }, - { - "epoch": 1.48, - "learning_rate": 1.4521528044535911e-05, - "loss": 0.3847, - "step": 727 - }, - { - "epoch": 1.48, - "learning_rate": 1.4506812644509505e-05, - "loss": 0.4196, - "step": 728 - }, - { - "epoch": 1.48, - "learning_rate": 1.4492084987624071e-05, - "loss": 0.3831, - "step": 729 - }, - { - "epoch": 1.49, - "learning_rate": 1.447734511393338e-05, - "loss": 0.3732, - "step": 730 - }, - { - "epoch": 1.49, - "learning_rate": 1.446259306352441e-05, - "loss": 0.3744, - "step": 731 - }, - { - "epoch": 1.49, - "learning_rate": 1.4447828876517279e-05, - "loss": 0.3898, - "step": 732 - }, - { - "epoch": 1.49, - "learning_rate": 1.4433052593065093e-05, - "loss": 0.4113, - "step": 733 - }, - { - "epoch": 1.49, - "learning_rate": 1.4418264253353869e-05, - "loss": 0.3735, - "step": 734 - }, - { - "epoch": 1.5, - "learning_rate": 1.44034638976024e-05, - "loss": 0.3859, - "step": 735 - }, - { - "epoch": 1.5, - "learning_rate": 1.4388651566062172e-05, - "loss": 0.4173, - "step": 736 - }, - { - "epoch": 1.5, - "learning_rate": 1.4373827299017226e-05, - "loss": 0.3883, - "step": 737 - }, - { - "epoch": 1.5, - "learning_rate": 1.4358991136784079e-05, - "loss": 0.3986, - "step": 738 - }, - { - "epoch": 1.51, - "learning_rate": 1.4344143119711585e-05, - "loss": 0.3855, - "step": 739 - }, - { - "epoch": 1.51, - "learning_rate": 1.4329283288180841e-05, - "loss": 0.3965, - "step": 740 - }, - { - "epoch": 1.51, - "learning_rate": 1.4314411682605087e-05, - "loss": 0.3877, - "step": 741 - }, - { - "epoch": 1.51, - "learning_rate": 1.4299528343429567e-05, - "loss": 0.3993, - "step": 742 - }, - { - "epoch": 1.51, - "learning_rate": 1.4284633311131448e-05, - "loss": 0.3981, - "step": 743 - }, - { - "epoch": 1.52, - "learning_rate": 1.4269726626219694e-05, - "loss": 0.4045, - "step": 744 - }, - { - "epoch": 1.52, - "learning_rate": 1.4254808329234955e-05, - "loss": 0.4065, - "step": 745 - }, - { - "epoch": 1.52, - "learning_rate": 1.4239878460749474e-05, - "loss": 0.4087, - "step": 746 - }, - { - "epoch": 1.52, - "learning_rate": 1.422493706136695e-05, - "loss": 0.3607, - "step": 747 - }, - { - "epoch": 1.52, - "learning_rate": 1.4209984171722451e-05, - "loss": 0.3888, - "step": 748 - }, - { - "epoch": 1.53, - "learning_rate": 1.419501983248229e-05, - "loss": 0.3616, - "step": 749 - }, - { - "epoch": 1.53, - "learning_rate": 1.4180044084343924e-05, - "loss": 0.3909, - "step": 750 - }, - { - "epoch": 1.53, - "learning_rate": 1.4165056968035831e-05, - "loss": 0.3875, - "step": 751 - }, - { - "epoch": 1.53, - "learning_rate": 1.415005852431741e-05, - "loss": 0.3688, - "step": 752 - }, - { - "epoch": 1.53, - "learning_rate": 1.4135048793978865e-05, - "loss": 0.4015, - "step": 753 - }, - { - "epoch": 1.54, - "learning_rate": 1.4120027817841098e-05, - "loss": 0.3626, - "step": 754 - }, - { - "epoch": 1.54, - "learning_rate": 1.410499563675559e-05, - "loss": 0.3685, - "step": 755 - }, - { - "epoch": 1.54, - "learning_rate": 1.4089952291604298e-05, - "loss": 0.3694, - "step": 756 - }, - { - "epoch": 1.54, - "learning_rate": 1.4074897823299551e-05, - "loss": 0.3743, - "step": 757 - }, - { - "epoch": 1.54, - "learning_rate": 1.4059832272783913e-05, - "loss": 0.3767, - "step": 758 - }, - { - "epoch": 1.55, - "learning_rate": 1.4044755681030094e-05, - "loss": 0.4019, - "step": 759 - }, - { - "epoch": 1.55, - "learning_rate": 1.4029668089040832e-05, - "loss": 0.3945, - "step": 760 - }, - { - "epoch": 1.55, - "learning_rate": 1.4014569537848784e-05, - "loss": 0.4129, - "step": 761 - }, - { - "epoch": 1.55, - "learning_rate": 1.3999460068516408e-05, - "loss": 0.4073, - "step": 762 - }, - { - "epoch": 1.55, - "learning_rate": 1.398433972213586e-05, - "loss": 0.3936, - "step": 763 - }, - { - "epoch": 1.56, - "learning_rate": 1.3969208539828873e-05, - "loss": 0.3609, - "step": 764 - }, - { - "epoch": 1.56, - "learning_rate": 1.3954066562746651e-05, - "loss": 0.3986, - "step": 765 - }, - { - "epoch": 1.56, - "learning_rate": 1.393891383206976e-05, - "loss": 0.3993, - "step": 766 - }, - { - "epoch": 1.56, - "learning_rate": 1.3923750389008e-05, - "loss": 0.3687, - "step": 767 - }, - { - "epoch": 1.56, - "learning_rate": 1.3908576274800321e-05, - "loss": 0.3864, - "step": 768 - }, - { - "epoch": 1.57, - "learning_rate": 1.389339153071468e-05, - "loss": 0.3764, - "step": 769 - }, - { - "epoch": 1.57, - "learning_rate": 1.3878196198047956e-05, - "loss": 0.3867, - "step": 770 - }, - { - "epoch": 1.57, - "learning_rate": 1.386299031812581e-05, - "loss": 0.4092, - "step": 771 - }, - { - "epoch": 1.57, - "learning_rate": 1.3847773932302605e-05, - "loss": 0.3864, - "step": 772 - }, - { - "epoch": 1.57, - "learning_rate": 1.3832547081961261e-05, - "loss": 0.3838, - "step": 773 - }, - { - "epoch": 1.58, - "learning_rate": 1.3817309808513169e-05, - "loss": 0.3843, - "step": 774 - }, - { - "epoch": 1.58, - "learning_rate": 1.3802062153398057e-05, - "loss": 0.3762, - "step": 775 - }, - { - "epoch": 1.58, - "learning_rate": 1.3786804158083898e-05, - "loss": 0.37, - "step": 776 - }, - { - "epoch": 1.58, - "learning_rate": 1.3771535864066774e-05, - "loss": 0.3898, - "step": 777 - }, - { - "epoch": 1.58, - "learning_rate": 1.3756257312870789e-05, - "loss": 0.3948, - "step": 778 - }, - { - "epoch": 1.59, - "learning_rate": 1.3740968546047935e-05, - "loss": 0.3832, - "step": 779 - }, - { - "epoch": 1.59, - "learning_rate": 1.3725669605177983e-05, - "loss": 0.381, - "step": 780 - }, - { - "epoch": 1.59, - "learning_rate": 1.3710360531868378e-05, - "loss": 0.3973, - "step": 781 - }, - { - "epoch": 1.59, - "learning_rate": 1.3695041367754134e-05, - "loss": 0.3872, - "step": 782 - }, - { - "epoch": 1.59, - "learning_rate": 1.3679712154497681e-05, - "loss": 0.3833, - "step": 783 - }, - { - "epoch": 1.6, - "learning_rate": 1.3664372933788804e-05, - "loss": 0.3778, - "step": 784 - }, - { - "epoch": 1.6, - "learning_rate": 1.3649023747344492e-05, - "loss": 0.3878, - "step": 785 - }, - { - "epoch": 1.6, - "learning_rate": 1.3633664636908838e-05, - "loss": 0.3864, - "step": 786 - }, - { - "epoch": 1.6, - "learning_rate": 1.361829564425293e-05, - "loss": 0.3866, - "step": 787 - }, - { - "epoch": 1.6, - "learning_rate": 1.3602916811174724e-05, - "loss": 0.3952, - "step": 788 - }, - { - "epoch": 1.61, - "learning_rate": 1.3587528179498946e-05, - "loss": 0.3562, - "step": 789 - }, - { - "epoch": 1.61, - "learning_rate": 1.3572129791076966e-05, - "loss": 0.3944, - "step": 790 - }, - { - "epoch": 1.61, - "learning_rate": 1.3556721687786692e-05, - "loss": 0.3805, - "step": 791 - }, - { - "epoch": 1.61, - "learning_rate": 1.3541303911532446e-05, - "loss": 0.3591, - "step": 792 - }, - { - "epoch": 1.62, - "learning_rate": 1.3525876504244866e-05, - "loss": 0.3658, - "step": 793 - }, - { - "epoch": 1.62, - "learning_rate": 1.3510439507880778e-05, - "loss": 0.3827, - "step": 794 - }, - { - "epoch": 1.62, - "learning_rate": 1.3494992964423086e-05, - "loss": 0.3582, - "step": 795 - }, - { - "epoch": 1.62, - "learning_rate": 1.347953691588066e-05, - "loss": 0.378, - "step": 796 - }, - { - "epoch": 1.62, - "learning_rate": 1.3464071404288219e-05, - "loss": 0.4212, - "step": 797 - }, - { - "epoch": 1.63, - "learning_rate": 1.3448596471706224e-05, - "loss": 0.4034, - "step": 798 - }, - { - "epoch": 1.63, - "learning_rate": 1.3433112160220747e-05, - "loss": 0.372, - "step": 799 - }, - { - "epoch": 1.63, - "learning_rate": 1.3417618511943374e-05, - "loss": 0.3803, - "step": 800 - }, - { - "epoch": 1.63, - "learning_rate": 1.3402115569011085e-05, - "loss": 0.393, - "step": 801 - }, - { - "epoch": 1.63, - "learning_rate": 1.3386603373586133e-05, - "loss": 0.3908, - "step": 802 - }, - { - "epoch": 1.64, - "learning_rate": 1.3371081967855939e-05, - "loss": 0.3735, - "step": 803 - }, - { - "epoch": 1.64, - "learning_rate": 1.3355551394032968e-05, - "loss": 0.397, - "step": 804 - }, - { - "epoch": 1.64, - "learning_rate": 1.3340011694354624e-05, - "loss": 0.3884, - "step": 805 - }, - { - "epoch": 1.64, - "learning_rate": 1.3324462911083123e-05, - "loss": 0.3541, - "step": 806 - }, - { - "epoch": 1.64, - "learning_rate": 1.3308905086505396e-05, - "loss": 0.3683, - "step": 807 - }, - { - "epoch": 1.65, - "learning_rate": 1.3293338262932944e-05, - "loss": 0.3671, - "step": 808 - }, - { - "epoch": 1.65, - "learning_rate": 1.3277762482701769e-05, - "loss": 0.3778, - "step": 809 - }, - { - "epoch": 1.65, - "learning_rate": 1.326217778817221e-05, - "loss": 0.3664, - "step": 810 - }, - { - "epoch": 1.65, - "learning_rate": 1.324658422172885e-05, - "loss": 0.3856, - "step": 811 - }, - { - "epoch": 1.65, - "learning_rate": 1.323098182578042e-05, - "loss": 0.3744, - "step": 812 - }, - { - "epoch": 1.66, - "learning_rate": 1.3215370642759643e-05, - "loss": 0.3579, - "step": 813 - }, - { - "epoch": 1.66, - "learning_rate": 1.3199750715123144e-05, - "loss": 0.3777, - "step": 814 - }, - { - "epoch": 1.66, - "learning_rate": 1.3184122085351343e-05, - "loss": 0.4052, - "step": 815 - }, - { - "epoch": 1.66, - "learning_rate": 1.3168484795948306e-05, - "loss": 0.3565, - "step": 816 - }, - { - "epoch": 1.66, - "learning_rate": 1.3152838889441672e-05, - "loss": 0.3611, - "step": 817 - }, - { - "epoch": 1.67, - "learning_rate": 1.31371844083825e-05, - "loss": 0.3688, - "step": 818 - }, - { - "epoch": 1.67, - "learning_rate": 1.3121521395345166e-05, - "loss": 0.3787, - "step": 819 - }, - { - "epoch": 1.67, - "learning_rate": 1.3105849892927269e-05, - "loss": 0.3624, - "step": 820 - }, - { - "epoch": 1.67, - "learning_rate": 1.3090169943749475e-05, - "loss": 0.3871, - "step": 821 - }, - { - "epoch": 1.67, - "learning_rate": 1.3074481590455434e-05, - "loss": 0.3817, - "step": 822 - }, - { - "epoch": 1.68, - "learning_rate": 1.305878487571165e-05, - "loss": 0.3441, - "step": 823 - }, - { - "epoch": 1.68, - "learning_rate": 1.3043079842207363e-05, - "loss": 0.3699, - "step": 824 - }, - { - "epoch": 1.68, - "learning_rate": 1.302736653265444e-05, - "loss": 0.3921, - "step": 825 - }, - { - "epoch": 1.68, - "learning_rate": 1.301164498978726e-05, - "loss": 0.4002, - "step": 826 - }, - { - "epoch": 1.68, - "learning_rate": 1.2995915256362583e-05, - "loss": 0.3984, - "step": 827 - }, - { - "epoch": 1.69, - "learning_rate": 1.2980177375159454e-05, - "loss": 0.3585, - "step": 828 - }, - { - "epoch": 1.69, - "learning_rate": 1.2964431388979075e-05, - "loss": 0.3623, - "step": 829 - }, - { - "epoch": 1.69, - "learning_rate": 1.2948677340644687e-05, - "loss": 0.3751, - "step": 830 - }, - { - "epoch": 1.69, - "learning_rate": 1.2932915273001459e-05, - "loss": 0.3892, - "step": 831 - }, - { - "epoch": 1.69, - "learning_rate": 1.2917145228916368e-05, - "loss": 0.387, - "step": 832 - }, - { - "epoch": 1.7, - "learning_rate": 1.2901367251278085e-05, - "loss": 0.3734, - "step": 833 - }, - { - "epoch": 1.7, - "learning_rate": 1.2885581382996866e-05, - "loss": 0.3827, - "step": 834 - }, - { - "epoch": 1.7, - "learning_rate": 1.2869787667004408e-05, - "loss": 0.3798, - "step": 835 - }, - { - "epoch": 1.7, - "learning_rate": 1.2853986146253768e-05, - "loss": 0.3796, - "step": 836 - }, - { - "epoch": 1.7, - "learning_rate": 1.2838176863719218e-05, - "loss": 0.3683, - "step": 837 - }, - { - "epoch": 1.71, - "learning_rate": 1.2822359862396147e-05, - "loss": 0.3736, - "step": 838 - }, - { - "epoch": 1.71, - "learning_rate": 1.2806535185300931e-05, - "loss": 0.3581, - "step": 839 - }, - { - "epoch": 1.71, - "learning_rate": 1.2790702875470826e-05, - "loss": 0.3496, - "step": 840 - }, - { - "epoch": 1.71, - "learning_rate": 1.2774862975963838e-05, - "loss": 0.3652, - "step": 841 - }, - { - "epoch": 1.71, - "learning_rate": 1.2759015529858625e-05, - "loss": 0.3742, - "step": 842 - }, - { - "epoch": 1.72, - "learning_rate": 1.274316058025436e-05, - "loss": 0.3857, - "step": 843 - }, - { - "epoch": 1.72, - "learning_rate": 1.2727298170270623e-05, - "loss": 0.3934, - "step": 844 - }, - { - "epoch": 1.72, - "learning_rate": 1.2711428343047293e-05, - "loss": 0.3516, - "step": 845 - }, - { - "epoch": 1.72, - "learning_rate": 1.2695551141744413e-05, - "loss": 0.3698, - "step": 846 - }, - { - "epoch": 1.73, - "learning_rate": 1.2679666609542084e-05, - "loss": 0.3942, - "step": 847 - }, - { - "epoch": 1.73, - "learning_rate": 1.2663774789640335e-05, - "loss": 0.3929, - "step": 848 - }, - { - "epoch": 1.73, - "learning_rate": 1.2647875725259026e-05, - "loss": 0.367, - "step": 849 - }, - { - "epoch": 1.73, - "learning_rate": 1.2631969459637725e-05, - "loss": 0.3713, - "step": 850 - }, - { - "epoch": 1.73, - "learning_rate": 1.2616056036035564e-05, - "loss": 0.3776, - "step": 851 - }, - { - "epoch": 1.74, - "learning_rate": 1.2600135497731156e-05, - "loss": 0.3811, - "step": 852 - }, - { - "epoch": 1.74, - "learning_rate": 1.2584207888022465e-05, - "loss": 0.3863, - "step": 853 - }, - { - "epoch": 1.74, - "learning_rate": 1.2568273250226681e-05, - "loss": 0.3833, - "step": 854 - }, - { - "epoch": 1.74, - "learning_rate": 1.2552331627680108e-05, - "loss": 0.3864, - "step": 855 - }, - { - "epoch": 1.74, - "learning_rate": 1.253638306373805e-05, - "loss": 0.3967, - "step": 856 - }, - { - "epoch": 1.75, - "learning_rate": 1.2520427601774683e-05, - "loss": 0.3684, - "step": 857 - }, - { - "epoch": 1.75, - "learning_rate": 1.2504465285182947e-05, - "loss": 0.4062, - "step": 858 - }, - { - "epoch": 1.75, - "learning_rate": 1.2488496157374425e-05, - "loss": 0.3873, - "step": 859 - }, - { - "epoch": 1.75, - "learning_rate": 1.2472520261779221e-05, - "loss": 0.3706, - "step": 860 - }, - { - "epoch": 1.75, - "learning_rate": 1.245653764184585e-05, - "loss": 0.3887, - "step": 861 - }, - { - "epoch": 1.76, - "learning_rate": 1.2440548341041108e-05, - "loss": 0.3828, - "step": 862 - }, - { - "epoch": 1.76, - "learning_rate": 1.242455240284996e-05, - "loss": 0.3729, - "step": 863 - }, - { - "epoch": 1.76, - "learning_rate": 1.2408549870775432e-05, - "loss": 0.3903, - "step": 864 - }, - { - "epoch": 1.76, - "learning_rate": 1.2392540788338475e-05, - "loss": 0.373, - "step": 865 - }, - { - "epoch": 1.76, - "learning_rate": 1.2376525199077853e-05, - "loss": 0.3902, - "step": 866 - }, - { - "epoch": 1.77, - "learning_rate": 1.2360503146550033e-05, - "loss": 0.3745, - "step": 867 - }, - { - "epoch": 1.77, - "learning_rate": 1.2344474674329052e-05, - "loss": 0.3598, - "step": 868 - }, - { - "epoch": 1.77, - "learning_rate": 1.2328439826006415e-05, - "loss": 0.3422, - "step": 869 - }, - { - "epoch": 1.77, - "learning_rate": 1.2312398645190965e-05, - "loss": 0.3948, - "step": 870 - }, - { - "epoch": 1.77, - "learning_rate": 1.2296351175508751e-05, - "loss": 0.3528, - "step": 871 - }, - { - "epoch": 1.78, - "learning_rate": 1.2280297460602958e-05, - "loss": 0.3663, - "step": 872 - }, - { - "epoch": 1.78, - "learning_rate": 1.2264237544133721e-05, - "loss": 0.4068, - "step": 873 - }, - { - "epoch": 1.78, - "learning_rate": 1.2248171469778067e-05, - "loss": 0.3545, - "step": 874 - }, - { - "epoch": 1.78, - "learning_rate": 1.223209928122976e-05, - "loss": 0.3866, - "step": 875 - }, - { - "epoch": 1.78, - "learning_rate": 1.2216021022199186e-05, - "loss": 0.3645, - "step": 876 - }, - { - "epoch": 1.79, - "learning_rate": 1.2199936736413246e-05, - "loss": 0.3759, - "step": 877 - }, - { - "epoch": 1.79, - "learning_rate": 1.2183846467615244e-05, - "loss": 0.3694, - "step": 878 - }, - { - "epoch": 1.79, - "learning_rate": 1.2167750259564733e-05, - "loss": 0.3621, - "step": 879 - }, - { - "epoch": 1.79, - "learning_rate": 1.2151648156037435e-05, - "loss": 0.3639, - "step": 880 - }, - { - "epoch": 1.79, - "learning_rate": 1.2135540200825095e-05, - "loss": 0.3568, - "step": 881 - }, - { - "epoch": 1.8, - "learning_rate": 1.2119426437735384e-05, - "loss": 0.393, - "step": 882 - }, - { - "epoch": 1.8, - "learning_rate": 1.2103306910591756e-05, - "loss": 0.3456, - "step": 883 - }, - { - "epoch": 1.8, - "learning_rate": 1.2087181663233354e-05, - "loss": 0.3697, - "step": 884 - }, - { - "epoch": 1.8, - "learning_rate": 1.2071050739514861e-05, - "loss": 0.3852, - "step": 885 - }, - { - "epoch": 1.8, - "learning_rate": 1.2054914183306416e-05, - "loss": 0.3934, - "step": 886 - }, - { - "epoch": 1.81, - "learning_rate": 1.203877203849346e-05, - "loss": 0.3639, - "step": 887 - }, - { - "epoch": 1.81, - "learning_rate": 1.2022624348976647e-05, - "loss": 0.329, - "step": 888 - }, - { - "epoch": 1.81, - "learning_rate": 1.2006471158671702e-05, - "loss": 0.3399, - "step": 889 - }, - { - "epoch": 1.81, - "learning_rate": 1.1990312511509313e-05, - "loss": 0.4173, - "step": 890 - }, - { - "epoch": 1.81, - "learning_rate": 1.1974148451435004e-05, - "loss": 0.3611, - "step": 891 - }, - { - "epoch": 1.82, - "learning_rate": 1.1957979022409028e-05, - "loss": 0.409, - "step": 892 - }, - { - "epoch": 1.82, - "learning_rate": 1.1941804268406232e-05, - "loss": 0.3624, - "step": 893 - }, - { - "epoch": 1.82, - "learning_rate": 1.1925624233415953e-05, - "loss": 0.3551, - "step": 894 - }, - { - "epoch": 1.82, - "learning_rate": 1.1909438961441881e-05, - "loss": 0.3663, - "step": 895 - }, - { - "epoch": 1.82, - "learning_rate": 1.1893248496501955e-05, - "loss": 0.3735, - "step": 896 - }, - { - "epoch": 1.83, - "learning_rate": 1.1877052882628237e-05, - "loss": 0.3779, - "step": 897 - }, - { - "epoch": 1.83, - "learning_rate": 1.186085216386679e-05, - "loss": 0.3533, - "step": 898 - }, - { - "epoch": 1.83, - "learning_rate": 1.184464638427756e-05, - "loss": 0.3793, - "step": 899 - }, - { - "epoch": 1.83, - "learning_rate": 1.182843558793426e-05, - "loss": 0.3478, - "step": 900 - }, - { - "epoch": 1.84, - "learning_rate": 1.1812219818924237e-05, - "loss": 0.366, - "step": 901 - }, - { - "epoch": 1.84, - "learning_rate": 1.1795999121348379e-05, - "loss": 0.3902, - "step": 902 - }, - { - "epoch": 1.84, - "learning_rate": 1.177977353932096e-05, - "loss": 0.3922, - "step": 903 - }, - { - "epoch": 1.84, - "learning_rate": 1.1763543116969549e-05, - "loss": 0.3639, - "step": 904 - }, - { - "epoch": 1.84, - "learning_rate": 1.1747307898434877e-05, - "loss": 0.3491, - "step": 905 - }, - { - "epoch": 1.85, - "learning_rate": 1.1731067927870716e-05, - "loss": 0.3605, - "step": 906 - }, - { - "epoch": 1.85, - "learning_rate": 1.1714823249443762e-05, - "loss": 0.3865, - "step": 907 - }, - { - "epoch": 1.85, - "learning_rate": 1.1698573907333519e-05, - "loss": 0.3613, - "step": 908 - }, - { - "epoch": 1.85, - "learning_rate": 1.168231994573217e-05, - "loss": 0.3835, - "step": 909 - }, - { - "epoch": 1.85, - "learning_rate": 1.166606140884446e-05, - "loss": 0.3543, - "step": 910 - }, - { - "epoch": 1.86, - "learning_rate": 1.1649798340887584e-05, - "loss": 0.3501, - "step": 911 - }, - { - "epoch": 1.86, - "learning_rate": 1.1633530786091051e-05, - "loss": 0.3591, - "step": 912 - }, - { - "epoch": 1.86, - "learning_rate": 1.1617258788696584e-05, - "loss": 0.3732, - "step": 913 - }, - { - "epoch": 1.86, - "learning_rate": 1.1600982392957978e-05, - "loss": 0.3846, - "step": 914 - }, - { - "epoch": 1.86, - "learning_rate": 1.1584701643140987e-05, - "loss": 0.3909, - "step": 915 - }, - { - "epoch": 1.87, - "learning_rate": 1.1568416583523223e-05, - "loss": 0.373, - "step": 916 - }, - { - "epoch": 1.87, - "learning_rate": 1.1552127258394004e-05, - "loss": 0.3837, - "step": 917 - }, - { - "epoch": 1.87, - "learning_rate": 1.1535833712054253e-05, - "loss": 0.3622, - "step": 918 - }, - { - "epoch": 1.87, - "learning_rate": 1.1519535988816376e-05, - "loss": 0.3465, - "step": 919 - }, - { - "epoch": 1.87, - "learning_rate": 1.1503234133004129e-05, - "loss": 0.3845, - "step": 920 - }, - { - "epoch": 1.88, - "learning_rate": 1.1486928188952522e-05, - "loss": 0.3462, - "step": 921 - }, - { - "epoch": 1.88, - "learning_rate": 1.1470618201007677e-05, - "loss": 0.3509, - "step": 922 - }, - { - "epoch": 1.88, - "learning_rate": 1.1454304213526703e-05, - "loss": 0.3684, - "step": 923 - }, - { - "epoch": 1.88, - "learning_rate": 1.1437986270877606e-05, - "loss": 0.3547, - "step": 924 - }, - { - "epoch": 1.88, - "learning_rate": 1.1421664417439132e-05, - "loss": 0.3819, - "step": 925 - }, - { - "epoch": 1.89, - "learning_rate": 1.1405338697600672e-05, - "loss": 0.3699, - "step": 926 - }, - { - "epoch": 1.89, - "learning_rate": 1.1389009155762127e-05, - "loss": 0.358, - "step": 927 - }, - { - "epoch": 1.89, - "learning_rate": 1.13726758363338e-05, - "loss": 0.3707, - "step": 928 - }, - { - "epoch": 1.89, - "learning_rate": 1.1356338783736256e-05, - "loss": 0.3533, - "step": 929 - }, - { - "epoch": 1.89, - "learning_rate": 1.1339998042400229e-05, - "loss": 0.3679, - "step": 930 - }, - { - "epoch": 1.9, - "learning_rate": 1.1323653656766462e-05, - "loss": 0.396, - "step": 931 - }, - { - "epoch": 1.9, - "learning_rate": 1.1307305671285633e-05, - "loss": 0.3636, - "step": 932 - }, - { - "epoch": 1.9, - "learning_rate": 1.12909541304182e-05, - "loss": 0.3684, - "step": 933 - }, - { - "epoch": 1.9, - "learning_rate": 1.1274599078634283e-05, - "loss": 0.359, - "step": 934 - }, - { - "epoch": 1.9, - "learning_rate": 1.125824056041356e-05, - "loss": 0.3592, - "step": 935 - }, - { - "epoch": 1.91, - "learning_rate": 1.1241878620245139e-05, - "loss": 0.3566, - "step": 936 - }, - { - "epoch": 1.91, - "learning_rate": 1.1225513302627422e-05, - "loss": 0.373, - "step": 937 - }, - { - "epoch": 1.91, - "learning_rate": 1.120914465206801e-05, - "loss": 0.3633, - "step": 938 - }, - { - "epoch": 1.91, - "learning_rate": 1.1192772713083557e-05, - "loss": 0.3567, - "step": 939 - }, - { - "epoch": 1.91, - "learning_rate": 1.1176397530199672e-05, - "loss": 0.3853, - "step": 940 - }, - { - "epoch": 1.92, - "learning_rate": 1.1160019147950774e-05, - "loss": 0.3715, - "step": 941 - }, - { - "epoch": 1.92, - "learning_rate": 1.114363761087999e-05, - "loss": 0.3408, - "step": 942 - }, - { - "epoch": 1.92, - "learning_rate": 1.1127252963539025e-05, - "loss": 0.3615, - "step": 943 - }, - { - "epoch": 1.92, - "learning_rate": 1.1110865250488047e-05, - "loss": 0.3581, - "step": 944 - }, - { - "epoch": 1.92, - "learning_rate": 1.1094474516295552e-05, - "loss": 0.3485, - "step": 945 - }, - { - "epoch": 1.93, - "learning_rate": 1.1078080805538267e-05, - "loss": 0.3529, - "step": 946 - }, - { - "epoch": 1.93, - "learning_rate": 1.1061684162801e-05, - "loss": 0.4016, - "step": 947 - }, - { - "epoch": 1.93, - "learning_rate": 1.1045284632676535e-05, - "loss": 0.3785, - "step": 948 - }, - { - "epoch": 1.93, - "learning_rate": 1.102888225976552e-05, - "loss": 0.3705, - "step": 949 - }, - { - "epoch": 1.93, - "learning_rate": 1.1012477088676322e-05, - "loss": 0.3431, - "step": 950 - }, - { - "epoch": 1.94, - "learning_rate": 1.0996069164024923e-05, - "loss": 0.359, - "step": 951 - }, - { - "epoch": 1.94, - "learning_rate": 1.0979658530434794e-05, - "loss": 0.3684, - "step": 952 - }, - { - "epoch": 1.94, - "learning_rate": 1.0963245232536768e-05, - "loss": 0.3671, - "step": 953 - }, - { - "epoch": 1.94, - "learning_rate": 1.0946829314968936e-05, - "loss": 0.3851, - "step": 954 - }, - { - "epoch": 1.95, - "learning_rate": 1.0930410822376503e-05, - "loss": 0.398, - "step": 955 - }, - { - "epoch": 1.95, - "learning_rate": 1.0913989799411675e-05, - "loss": 0.3618, - "step": 956 - }, - { - "epoch": 1.95, - "learning_rate": 1.0897566290733552e-05, - "loss": 0.3413, - "step": 957 - }, - { - "epoch": 1.95, - "learning_rate": 1.0881140341007981e-05, - "loss": 0.3603, - "step": 958 - }, - { - "epoch": 1.95, - "learning_rate": 1.0864711994907457e-05, - "loss": 0.3479, - "step": 959 - }, - { - "epoch": 1.96, - "learning_rate": 1.084828129711099e-05, - "loss": 0.3561, - "step": 960 - }, - { - "epoch": 1.96, - "learning_rate": 1.0831848292303982e-05, - "loss": 0.364, - "step": 961 - }, - { - "epoch": 1.96, - "learning_rate": 1.0815413025178112e-05, - "loss": 0.3456, - "step": 962 - }, - { - "epoch": 1.96, - "learning_rate": 1.0798975540431214e-05, - "loss": 0.3794, - "step": 963 - }, - { - "epoch": 1.96, - "learning_rate": 1.0782535882767144e-05, - "loss": 0.3424, - "step": 964 - }, - { - "epoch": 1.97, - "learning_rate": 1.076609409689568e-05, - "loss": 0.3395, - "step": 965 - }, - { - "epoch": 1.97, - "learning_rate": 1.0749650227532387e-05, - "loss": 0.343, - "step": 966 - }, - { - "epoch": 1.97, - "learning_rate": 1.0733204319398477e-05, - "loss": 0.3418, - "step": 967 - }, - { - "epoch": 1.97, - "learning_rate": 1.0716756417220735e-05, - "loss": 0.413, - "step": 968 - }, - { - "epoch": 1.97, - "learning_rate": 1.0700306565731348e-05, - "loss": 0.3709, - "step": 969 - }, - { - "epoch": 1.98, - "learning_rate": 1.0683854809667814e-05, - "loss": 0.3773, - "step": 970 - }, - { - "epoch": 1.98, - "learning_rate": 1.0667401193772807e-05, - "loss": 0.3813, - "step": 971 - }, - { - "epoch": 1.98, - "learning_rate": 1.0650945762794058e-05, - "loss": 0.353, - "step": 972 - }, - { - "epoch": 1.98, - "learning_rate": 1.0634488561484243e-05, - "loss": 0.3656, - "step": 973 - }, - { - "epoch": 1.98, - "learning_rate": 1.0618029634600843e-05, - "loss": 0.3617, - "step": 974 - }, - { - "epoch": 1.99, - "learning_rate": 1.0601569026906034e-05, - "loss": 0.3341, - "step": 975 - }, - { - "epoch": 1.99, - "learning_rate": 1.0585106783166566e-05, - "loss": 0.3479, - "step": 976 - }, - { - "epoch": 1.99, - "learning_rate": 1.0568642948153635e-05, - "loss": 0.3578, - "step": 977 - }, - { - "epoch": 1.99, - "learning_rate": 1.0552177566642769e-05, - "loss": 0.3602, - "step": 978 - }, - { - "epoch": 1.99, - "learning_rate": 1.05357106834137e-05, - "loss": 0.3449, - "step": 979 - }, - { - "epoch": 2.0, - "learning_rate": 1.0519242343250241e-05, - "loss": 0.3544, - "step": 980 - }, - { - "epoch": 2.0, - "learning_rate": 1.0502772590940171e-05, - "loss": 0.327, - "step": 981 - }, - { - "epoch": 2.0, - "learning_rate": 1.0486301471275111e-05, - "loss": 0.3334, - "step": 982 - }, - { - "epoch": 2.0, - "learning_rate": 1.0469829029050391e-05, - "loss": 0.1874, - "step": 983 - }, - { - "epoch": 2.0, - "learning_rate": 1.0453355309064952e-05, - "loss": 0.1807, - "step": 984 - }, - { - "epoch": 2.01, - "learning_rate": 1.0436880356121201e-05, - "loss": 0.1854, - "step": 985 - }, - { - "epoch": 2.01, - "learning_rate": 1.04204042150249e-05, - "loss": 0.1812, - "step": 986 - }, - { - "epoch": 2.01, - "learning_rate": 1.0403926930585043e-05, - "loss": 0.1779, - "step": 987 - }, - { - "epoch": 2.01, - "learning_rate": 1.038744854761373e-05, - "loss": 0.1778, - "step": 988 - }, - { - "epoch": 2.01, - "learning_rate": 1.0370969110926052e-05, - "loss": 0.1826, - "step": 989 - }, - { - "epoch": 2.02, - "learning_rate": 1.0354488665339975e-05, - "loss": 0.1862, - "step": 990 - }, - { - "epoch": 2.02, - "learning_rate": 1.0338007255676183e-05, - "loss": 0.1829, - "step": 991 - }, - { - "epoch": 2.02, - "learning_rate": 1.0321524926758013e-05, - "loss": 0.1832, - "step": 992 - }, - { - "epoch": 2.02, - "learning_rate": 1.0305041723411278e-05, - "loss": 0.1627, - "step": 993 - }, - { - "epoch": 2.02, - "learning_rate": 1.0288557690464183e-05, - "loss": 0.1918, - "step": 994 - }, - { - "epoch": 2.03, - "learning_rate": 1.0272072872747181e-05, - "loss": 0.1858, - "step": 995 - }, - { - "epoch": 2.03, - "learning_rate": 1.0255587315092865e-05, - "loss": 0.1698, - "step": 996 - }, - { - "epoch": 2.03, - "learning_rate": 1.0239101062335834e-05, - "loss": 0.1799, - "step": 997 - }, - { - "epoch": 2.03, - "learning_rate": 1.0222614159312589e-05, - "loss": 0.1703, - "step": 998 - }, - { - "epoch": 2.03, - "learning_rate": 1.0206126650861387e-05, - "loss": 0.1749, - "step": 999 - }, - { - "epoch": 2.04, - "learning_rate": 1.0189638581822136e-05, - "loss": 0.1901, - "step": 1000 - }, - { - "epoch": 2.04, - "learning_rate": 1.0173149997036272e-05, - "loss": 0.1715, - "step": 1001 - }, - { - "epoch": 2.04, - "learning_rate": 1.0156660941346628e-05, - "loss": 0.1672, - "step": 1002 - }, - { - "epoch": 2.04, - "learning_rate": 1.0140171459597323e-05, - "loss": 0.1662, - "step": 1003 - }, - { - "epoch": 2.04, - "learning_rate": 1.012368159663363e-05, - "loss": 0.1696, - "step": 1004 - }, - { - "epoch": 2.05, - "learning_rate": 1.010719139730186e-05, - "loss": 0.1889, - "step": 1005 - }, - { - "epoch": 2.05, - "learning_rate": 1.0090700906449247e-05, - "loss": 0.1804, - "step": 1006 - }, - { - "epoch": 2.05, - "learning_rate": 1.0074210168923804e-05, - "loss": 0.1671, - "step": 1007 - }, - { - "epoch": 2.05, - "learning_rate": 1.0057719229574217e-05, - "loss": 0.1764, - "step": 1008 - }, - { - "epoch": 2.05, - "learning_rate": 1.0041228133249736e-05, - "loss": 0.1816, - "step": 1009 - }, - { - "epoch": 2.06, - "learning_rate": 1.002473692480002e-05, - "loss": 0.1701, - "step": 1010 - }, - { - "epoch": 2.06, - "learning_rate": 1.0008245649075044e-05, - "loss": 0.1721, - "step": 1011 - }, - { - "epoch": 2.06, - "learning_rate": 9.99175435092496e-06, - "loss": 0.1615, - "step": 1012 - }, - { - "epoch": 2.06, - "learning_rate": 9.975263075199983e-06, - "loss": 0.1874, - "step": 1013 - }, - { - "epoch": 2.07, - "learning_rate": 9.958771866750266e-06, - "loss": 0.1843, - "step": 1014 - }, - { - "epoch": 2.07, - "learning_rate": 9.942280770425783e-06, - "loss": 0.1713, - "step": 1015 - }, - { - "epoch": 2.07, - "learning_rate": 9.925789831076202e-06, - "loss": 0.1689, - "step": 1016 - }, - { - "epoch": 2.07, - "learning_rate": 9.909299093550758e-06, - "loss": 0.1844, - "step": 1017 - }, - { - "epoch": 2.07, - "learning_rate": 9.892808602698142e-06, - "loss": 0.1708, - "step": 1018 - }, - { - "epoch": 2.08, - "learning_rate": 9.876318403366371e-06, - "loss": 0.1671, - "step": 1019 - }, - { - "epoch": 2.08, - "learning_rate": 9.85982854040268e-06, - "loss": 0.1648, - "step": 1020 - }, - { - "epoch": 2.08, - "learning_rate": 9.843339058653374e-06, - "loss": 0.1644, - "step": 1021 - }, - { - "epoch": 2.08, - "learning_rate": 9.82685000296373e-06, - "loss": 0.1805, - "step": 1022 - }, - { - "epoch": 2.08, - "learning_rate": 9.810361418177866e-06, - "loss": 0.1694, - "step": 1023 - }, - { - "epoch": 2.09, - "learning_rate": 9.793873349138618e-06, - "loss": 0.1715, - "step": 1024 - }, - { - "epoch": 2.09, - "learning_rate": 9.777385840687415e-06, - "loss": 0.1718, - "step": 1025 - }, - { - "epoch": 2.09, - "learning_rate": 9.760898937664168e-06, - "loss": 0.1724, - "step": 1026 - }, - { - "epoch": 2.09, - "learning_rate": 9.744412684907139e-06, - "loss": 0.1765, - "step": 1027 - }, - { - "epoch": 2.09, - "learning_rate": 9.727927127252822e-06, - "loss": 0.1728, - "step": 1028 - }, - { - "epoch": 2.1, - "learning_rate": 9.71144230953582e-06, - "loss": 0.1938, - "step": 1029 - }, - { - "epoch": 2.1, - "learning_rate": 9.694958276588722e-06, - "loss": 0.1725, - "step": 1030 - }, - { - "epoch": 2.1, - "learning_rate": 9.678475073241987e-06, - "loss": 0.1647, - "step": 1031 - }, - { - "epoch": 2.1, - "learning_rate": 9.661992744323819e-06, - "loss": 0.1716, - "step": 1032 - }, - { - "epoch": 2.1, - "learning_rate": 9.64551133466003e-06, - "loss": 0.173, - "step": 1033 - }, - { - "epoch": 2.11, - "learning_rate": 9.62903088907395e-06, - "loss": 0.1692, - "step": 1034 - }, - { - "epoch": 2.11, - "learning_rate": 9.612551452386273e-06, - "loss": 0.1691, - "step": 1035 - }, - { - "epoch": 2.11, - "learning_rate": 9.596073069414962e-06, - "loss": 0.1825, - "step": 1036 - }, - { - "epoch": 2.11, - "learning_rate": 9.579595784975102e-06, - "loss": 0.1901, - "step": 1037 - }, - { - "epoch": 2.11, - "learning_rate": 9.5631196438788e-06, - "loss": 0.1874, - "step": 1038 - }, - { - "epoch": 2.12, - "learning_rate": 9.54664469093505e-06, - "loss": 0.1712, - "step": 1039 - }, - { - "epoch": 2.12, - "learning_rate": 9.530170970949612e-06, - "loss": 0.1824, - "step": 1040 - }, - { - "epoch": 2.12, - "learning_rate": 9.513698528724896e-06, - "loss": 0.1756, - "step": 1041 - }, - { - "epoch": 2.12, - "learning_rate": 9.497227409059832e-06, - "loss": 0.1762, - "step": 1042 - }, - { - "epoch": 2.12, - "learning_rate": 9.48075765674976e-06, - "loss": 0.1728, - "step": 1043 - }, - { - "epoch": 2.13, - "learning_rate": 9.464289316586303e-06, - "loss": 0.1899, - "step": 1044 - }, - { - "epoch": 2.13, - "learning_rate": 9.447822433357233e-06, - "loss": 0.1797, - "step": 1045 - }, - { - "epoch": 2.13, - "learning_rate": 9.431357051846367e-06, - "loss": 0.1764, - "step": 1046 - }, - { - "epoch": 2.13, - "learning_rate": 9.414893216833436e-06, - "loss": 0.1742, - "step": 1047 - }, - { - "epoch": 2.13, - "learning_rate": 9.398430973093969e-06, - "loss": 0.1959, - "step": 1048 - }, - { - "epoch": 2.14, - "learning_rate": 9.381970365399162e-06, - "loss": 0.2027, - "step": 1049 - }, - { - "epoch": 2.14, - "learning_rate": 9.365511438515759e-06, - "loss": 0.1665, - "step": 1050 - }, - { - "epoch": 2.14, - "learning_rate": 9.349054237205945e-06, - "loss": 0.178, - "step": 1051 - }, - { - "epoch": 2.14, - "learning_rate": 9.332598806227195e-06, - "loss": 0.1813, - "step": 1052 - }, - { - "epoch": 2.14, - "learning_rate": 9.316145190332188e-06, - "loss": 0.1807, - "step": 1053 - }, - { - "epoch": 2.15, - "learning_rate": 9.299693434268653e-06, - "loss": 0.1717, - "step": 1054 - }, - { - "epoch": 2.15, - "learning_rate": 9.283243582779266e-06, - "loss": 0.1818, - "step": 1055 - }, - { - "epoch": 2.15, - "learning_rate": 9.266795680601526e-06, - "loss": 0.1683, - "step": 1056 - }, - { - "epoch": 2.15, - "learning_rate": 9.250349772467618e-06, - "loss": 0.1815, - "step": 1057 - }, - { - "epoch": 2.15, - "learning_rate": 9.233905903104321e-06, - "loss": 0.1762, - "step": 1058 - }, - { - "epoch": 2.16, - "learning_rate": 9.217464117232859e-06, - "loss": 0.1764, - "step": 1059 - }, - { - "epoch": 2.16, - "learning_rate": 9.20102445956879e-06, - "loss": 0.1902, - "step": 1060 - }, - { - "epoch": 2.16, - "learning_rate": 9.184586974821891e-06, - "loss": 0.1877, - "step": 1061 - }, - { - "epoch": 2.16, - "learning_rate": 9.16815170769602e-06, - "loss": 0.1812, - "step": 1062 - }, - { - "epoch": 2.16, - "learning_rate": 9.15171870288901e-06, - "loss": 0.1942, - "step": 1063 - }, - { - "epoch": 2.17, - "learning_rate": 9.135288005092546e-06, - "loss": 0.1872, - "step": 1064 - }, - { - "epoch": 2.17, - "learning_rate": 9.11885965899202e-06, - "loss": 0.1772, - "step": 1065 - }, - { - "epoch": 2.17, - "learning_rate": 9.102433709266452e-06, - "loss": 0.184, - "step": 1066 - }, - { - "epoch": 2.17, - "learning_rate": 9.086010200588328e-06, - "loss": 0.1839, - "step": 1067 - }, - { - "epoch": 2.18, - "learning_rate": 9.0695891776235e-06, - "loss": 0.1726, - "step": 1068 - }, - { - "epoch": 2.18, - "learning_rate": 9.053170685031065e-06, - "loss": 0.1791, - "step": 1069 - }, - { - "epoch": 2.18, - "learning_rate": 9.036754767463233e-06, - "loss": 0.1771, - "step": 1070 - }, - { - "epoch": 2.18, - "learning_rate": 9.020341469565208e-06, - "loss": 0.1987, - "step": 1071 - }, - { - "epoch": 2.18, - "learning_rate": 9.003930835975082e-06, - "loss": 0.1869, - "step": 1072 - }, - { - "epoch": 2.19, - "learning_rate": 8.987522911323681e-06, - "loss": 0.1793, - "step": 1073 - }, - { - "epoch": 2.19, - "learning_rate": 8.971117740234482e-06, - "loss": 0.1812, - "step": 1074 - }, - { - "epoch": 2.19, - "learning_rate": 8.954715367323468e-06, - "loss": 0.1761, - "step": 1075 - }, - { - "epoch": 2.19, - "learning_rate": 8.938315837199003e-06, - "loss": 0.1838, - "step": 1076 - }, - { - "epoch": 2.19, - "learning_rate": 8.921919194461735e-06, - "loss": 0.1755, - "step": 1077 - }, - { - "epoch": 2.2, - "learning_rate": 8.905525483704448e-06, - "loss": 0.1803, - "step": 1078 - }, - { - "epoch": 2.2, - "learning_rate": 8.889134749511956e-06, - "loss": 0.1833, - "step": 1079 - }, - { - "epoch": 2.2, - "learning_rate": 8.872747036460978e-06, - "loss": 0.1786, - "step": 1080 - }, - { - "epoch": 2.2, - "learning_rate": 8.856362389120012e-06, - "loss": 0.1751, - "step": 1081 - }, - { - "epoch": 2.2, - "learning_rate": 8.83998085204923e-06, - "loss": 0.1819, - "step": 1082 - }, - { - "epoch": 2.21, - "learning_rate": 8.823602469800331e-06, - "loss": 0.1768, - "step": 1083 - }, - { - "epoch": 2.21, - "learning_rate": 8.807227286916444e-06, - "loss": 0.1848, - "step": 1084 - }, - { - "epoch": 2.21, - "learning_rate": 8.790855347931991e-06, - "loss": 0.1825, - "step": 1085 - }, - { - "epoch": 2.21, - "learning_rate": 8.77448669737258e-06, - "loss": 0.1914, - "step": 1086 - }, - { - "epoch": 2.21, - "learning_rate": 8.758121379754865e-06, - "loss": 0.1752, - "step": 1087 - }, - { - "epoch": 2.22, - "learning_rate": 8.741759439586443e-06, - "loss": 0.194, - "step": 1088 - }, - { - "epoch": 2.22, - "learning_rate": 8.725400921365722e-06, - "loss": 0.1822, - "step": 1089 - }, - { - "epoch": 2.22, - "learning_rate": 8.709045869581805e-06, - "loss": 0.1917, - "step": 1090 - }, - { - "epoch": 2.22, - "learning_rate": 8.692694328714369e-06, - "loss": 0.1882, - "step": 1091 - }, - { - "epoch": 2.22, - "learning_rate": 8.67634634323354e-06, - "loss": 0.1992, - "step": 1092 - }, - { - "epoch": 2.23, - "learning_rate": 8.660001957599776e-06, - "loss": 0.18, - "step": 1093 - }, - { - "epoch": 2.23, - "learning_rate": 8.643661216263744e-06, - "loss": 0.1816, - "step": 1094 - }, - { - "epoch": 2.23, - "learning_rate": 8.627324163666202e-06, - "loss": 0.1884, - "step": 1095 - }, - { - "epoch": 2.23, - "learning_rate": 8.610990844237876e-06, - "loss": 0.1945, - "step": 1096 - }, - { - "epoch": 2.23, - "learning_rate": 8.594661302399333e-06, - "loss": 0.1927, - "step": 1097 - }, - { - "epoch": 2.24, - "learning_rate": 8.578335582560872e-06, - "loss": 0.1845, - "step": 1098 - }, - { - "epoch": 2.24, - "learning_rate": 8.562013729122398e-06, - "loss": 0.1768, - "step": 1099 - }, - { - "epoch": 2.24, - "learning_rate": 8.5456957864733e-06, - "loss": 0.1912, - "step": 1100 - }, - { - "epoch": 2.24, - "learning_rate": 8.529381798992326e-06, - "loss": 0.1771, - "step": 1101 - }, - { - "epoch": 2.24, - "learning_rate": 8.513071811047478e-06, - "loss": 0.1928, - "step": 1102 - }, - { - "epoch": 2.25, - "learning_rate": 8.496765866995871e-06, - "loss": 0.2023, - "step": 1103 - }, - { - "epoch": 2.25, - "learning_rate": 8.480464011183631e-06, - "loss": 0.1836, - "step": 1104 - }, - { - "epoch": 2.25, - "learning_rate": 8.464166287945752e-06, - "loss": 0.1837, - "step": 1105 - }, - { - "epoch": 2.25, - "learning_rate": 8.447872741606e-06, - "loss": 0.1823, - "step": 1106 - }, - { - "epoch": 2.25, - "learning_rate": 8.431583416476779e-06, - "loss": 0.1956, - "step": 1107 - }, - { - "epoch": 2.26, - "learning_rate": 8.415298356859014e-06, - "loss": 0.1817, - "step": 1108 - }, - { - "epoch": 2.26, - "learning_rate": 8.399017607042025e-06, - "loss": 0.1784, - "step": 1109 - }, - { - "epoch": 2.26, - "learning_rate": 8.382741211303416e-06, - "loss": 0.1845, - "step": 1110 - }, - { - "epoch": 2.26, - "learning_rate": 8.366469213908949e-06, - "loss": 0.2015, - "step": 1111 - }, - { - "epoch": 2.26, - "learning_rate": 8.35020165911242e-06, - "loss": 0.1881, - "step": 1112 - }, - { - "epoch": 2.27, - "learning_rate": 8.333938591155543e-06, - "loss": 0.1843, - "step": 1113 - }, - { - "epoch": 2.27, - "learning_rate": 8.317680054267834e-06, - "loss": 0.1862, - "step": 1114 - }, - { - "epoch": 2.27, - "learning_rate": 8.301426092666483e-06, - "loss": 0.1851, - "step": 1115 - }, - { - "epoch": 2.27, - "learning_rate": 8.28517675055624e-06, - "loss": 0.1773, - "step": 1116 - }, - { - "epoch": 2.27, - "learning_rate": 8.268932072129288e-06, - "loss": 0.188, - "step": 1117 - }, - { - "epoch": 2.28, - "learning_rate": 8.252692101565123e-06, - "loss": 0.1765, - "step": 1118 - }, - { - "epoch": 2.28, - "learning_rate": 8.236456883030451e-06, - "loss": 0.1954, - "step": 1119 - }, - { - "epoch": 2.28, - "learning_rate": 8.220226460679044e-06, - "loss": 0.1796, - "step": 1120 - }, - { - "epoch": 2.28, - "learning_rate": 8.204000878651626e-06, - "loss": 0.1839, - "step": 1121 - }, - { - "epoch": 2.29, - "learning_rate": 8.187780181075766e-06, - "loss": 0.1789, - "step": 1122 - }, - { - "epoch": 2.29, - "learning_rate": 8.171564412065745e-06, - "loss": 0.1794, - "step": 1123 - }, - { - "epoch": 2.29, - "learning_rate": 8.155353615722442e-06, - "loss": 0.1837, - "step": 1124 - }, - { - "epoch": 2.29, - "learning_rate": 8.139147836133212e-06, - "loss": 0.1869, - "step": 1125 - }, - { - "epoch": 2.29, - "learning_rate": 8.122947117371763e-06, - "loss": 0.182, - "step": 1126 - }, - { - "epoch": 2.3, - "learning_rate": 8.106751503498045e-06, - "loss": 0.1885, - "step": 1127 - }, - { - "epoch": 2.3, - "learning_rate": 8.090561038558124e-06, - "loss": 0.1787, - "step": 1128 - }, - { - "epoch": 2.3, - "learning_rate": 8.074375766584053e-06, - "loss": 0.1876, - "step": 1129 - }, - { - "epoch": 2.3, - "learning_rate": 8.058195731593773e-06, - "loss": 0.1858, - "step": 1130 - }, - { - "epoch": 2.3, - "learning_rate": 8.042020977590975e-06, - "loss": 0.1793, - "step": 1131 - }, - { - "epoch": 2.31, - "learning_rate": 8.025851548565e-06, - "loss": 0.1945, - "step": 1132 - }, - { - "epoch": 2.31, - "learning_rate": 8.009687488490692e-06, - "loss": 0.1852, - "step": 1133 - }, - { - "epoch": 2.31, - "learning_rate": 7.993528841328298e-06, - "loss": 0.1907, - "step": 1134 - }, - { - "epoch": 2.31, - "learning_rate": 7.977375651023354e-06, - "loss": 0.1875, - "step": 1135 - }, - { - "epoch": 2.31, - "learning_rate": 7.961227961506543e-06, - "loss": 0.1913, - "step": 1136 - }, - { - "epoch": 2.32, - "learning_rate": 7.94508581669359e-06, - "loss": 0.18, - "step": 1137 - }, - { - "epoch": 2.32, - "learning_rate": 7.928949260485142e-06, - "loss": 0.2, - "step": 1138 - }, - { - "epoch": 2.32, - "learning_rate": 7.91281833676665e-06, - "loss": 0.1936, - "step": 1139 - }, - { - "epoch": 2.32, - "learning_rate": 7.896693089408245e-06, - "loss": 0.1892, - "step": 1140 - }, - { - "epoch": 2.32, - "learning_rate": 7.880573562264619e-06, - "loss": 0.1908, - "step": 1141 - }, - { - "epoch": 2.33, - "learning_rate": 7.864459799174905e-06, - "loss": 0.2001, - "step": 1142 - }, - { - "epoch": 2.33, - "learning_rate": 7.848351843962567e-06, - "loss": 0.18, - "step": 1143 - }, - { - "epoch": 2.33, - "learning_rate": 7.83224974043527e-06, - "loss": 0.1901, - "step": 1144 - }, - { - "epoch": 2.33, - "learning_rate": 7.81615353238476e-06, - "loss": 0.1799, - "step": 1145 - }, - { - "epoch": 2.33, - "learning_rate": 7.800063263586755e-06, - "loss": 0.1989, - "step": 1146 - }, - { - "epoch": 2.34, - "learning_rate": 7.783978977800818e-06, - "loss": 0.1935, - "step": 1147 - }, - { - "epoch": 2.34, - "learning_rate": 7.767900718770245e-06, - "loss": 0.1832, - "step": 1148 - }, - { - "epoch": 2.34, - "learning_rate": 7.751828530221934e-06, - "loss": 0.1911, - "step": 1149 - }, - { - "epoch": 2.34, - "learning_rate": 7.73576245586628e-06, - "loss": 0.1834, - "step": 1150 - }, - { - "epoch": 2.34, - "learning_rate": 7.719702539397044e-06, - "loss": 0.1862, - "step": 1151 - }, - { - "epoch": 2.35, - "learning_rate": 7.70364882449125e-06, - "loss": 0.1922, - "step": 1152 - }, - { - "epoch": 2.35, - "learning_rate": 7.687601354809042e-06, - "loss": 0.1824, - "step": 1153 - }, - { - "epoch": 2.35, - "learning_rate": 7.671560173993588e-06, - "loss": 0.1984, - "step": 1154 - }, - { - "epoch": 2.35, - "learning_rate": 7.655525325670951e-06, - "loss": 0.1754, - "step": 1155 - }, - { - "epoch": 2.35, - "learning_rate": 7.63949685344997e-06, - "loss": 0.1823, - "step": 1156 - }, - { - "epoch": 2.36, - "learning_rate": 7.62347480092215e-06, - "loss": 0.196, - "step": 1157 - }, - { - "epoch": 2.36, - "learning_rate": 7.6074592116615285e-06, - "loss": 0.1736, - "step": 1158 - }, - { - "epoch": 2.36, - "learning_rate": 7.591450129224569e-06, - "loss": 0.1945, - "step": 1159 - }, - { - "epoch": 2.36, - "learning_rate": 7.575447597150045e-06, - "loss": 0.1787, - "step": 1160 - }, - { - "epoch": 2.36, - "learning_rate": 7.559451658958896e-06, - "loss": 0.1877, - "step": 1161 - }, - { - "epoch": 2.37, - "learning_rate": 7.543462358154153e-06, - "loss": 0.1864, - "step": 1162 - }, - { - "epoch": 2.37, - "learning_rate": 7.527479738220781e-06, - "loss": 0.1821, - "step": 1163 - }, - { - "epoch": 2.37, - "learning_rate": 7.511503842625576e-06, - "loss": 0.1885, - "step": 1164 - }, - { - "epoch": 2.37, - "learning_rate": 7.495534714817054e-06, - "loss": 0.1891, - "step": 1165 - }, - { - "epoch": 2.37, - "learning_rate": 7.479572398225321e-06, - "loss": 0.1922, - "step": 1166 - }, - { - "epoch": 2.38, - "learning_rate": 7.463616936261951e-06, - "loss": 0.1858, - "step": 1167 - }, - { - "epoch": 2.38, - "learning_rate": 7.447668372319896e-06, - "loss": 0.1955, - "step": 1168 - }, - { - "epoch": 2.38, - "learning_rate": 7.431726749773322e-06, - "loss": 0.1789, - "step": 1169 - }, - { - "epoch": 2.38, - "learning_rate": 7.415792111977538e-06, - "loss": 0.1792, - "step": 1170 - }, - { - "epoch": 2.38, - "learning_rate": 7.399864502268847e-06, - "loss": 0.1886, - "step": 1171 - }, - { - "epoch": 2.39, - "learning_rate": 7.383943963964439e-06, - "loss": 0.1846, - "step": 1172 - }, - { - "epoch": 2.39, - "learning_rate": 7.368030540362278e-06, - "loss": 0.1943, - "step": 1173 - }, - { - "epoch": 2.39, - "learning_rate": 7.3521242747409735e-06, - "loss": 0.1976, - "step": 1174 - }, - { - "epoch": 2.39, - "learning_rate": 7.336225210359666e-06, - "loss": 0.1924, - "step": 1175 - }, - { - "epoch": 2.4, - "learning_rate": 7.320333390457923e-06, - "loss": 0.1911, - "step": 1176 - }, - { - "epoch": 2.4, - "learning_rate": 7.304448858255589e-06, - "loss": 0.1906, - "step": 1177 - }, - { - "epoch": 2.4, - "learning_rate": 7.288571656952708e-06, - "loss": 0.182, - "step": 1178 - }, - { - "epoch": 2.4, - "learning_rate": 7.272701829729378e-06, - "loss": 0.2015, - "step": 1179 - }, - { - "epoch": 2.4, - "learning_rate": 7.256839419745643e-06, - "loss": 0.1993, - "step": 1180 - }, - { - "epoch": 2.41, - "learning_rate": 7.240984470141378e-06, - "loss": 0.1833, - "step": 1181 - }, - { - "epoch": 2.41, - "learning_rate": 7.225137024036164e-06, - "loss": 0.1783, - "step": 1182 - }, - { - "epoch": 2.41, - "learning_rate": 7.209297124529178e-06, - "loss": 0.1842, - "step": 1183 - }, - { - "epoch": 2.41, - "learning_rate": 7.193464814699073e-06, - "loss": 0.1944, - "step": 1184 - }, - { - "epoch": 2.41, - "learning_rate": 7.1776401376038565e-06, - "loss": 0.1893, - "step": 1185 - }, - { - "epoch": 2.42, - "learning_rate": 7.161823136280785e-06, - "loss": 0.1885, - "step": 1186 - }, - { - "epoch": 2.42, - "learning_rate": 7.146013853746236e-06, - "loss": 0.1932, - "step": 1187 - }, - { - "epoch": 2.42, - "learning_rate": 7.130212332995596e-06, - "loss": 0.216, - "step": 1188 - }, - { - "epoch": 2.42, - "learning_rate": 7.114418617003137e-06, - "loss": 0.1873, - "step": 1189 - }, - { - "epoch": 2.42, - "learning_rate": 7.098632748721914e-06, - "loss": 0.1916, - "step": 1190 - }, - { - "epoch": 2.43, - "learning_rate": 7.082854771083635e-06, - "loss": 0.1911, - "step": 1191 - }, - { - "epoch": 2.43, - "learning_rate": 7.067084726998548e-06, - "loss": 0.1847, - "step": 1192 - }, - { - "epoch": 2.43, - "learning_rate": 7.0513226593553175e-06, - "loss": 0.181, - "step": 1193 - }, - { - "epoch": 2.43, - "learning_rate": 7.0355686110209266e-06, - "loss": 0.2024, - "step": 1194 - }, - { - "epoch": 2.43, - "learning_rate": 7.019822624840548e-06, - "loss": 0.1821, - "step": 1195 - }, - { - "epoch": 2.44, - "learning_rate": 7.00408474363742e-06, - "loss": 0.1937, - "step": 1196 - }, - { - "epoch": 2.44, - "learning_rate": 6.988355010212742e-06, - "loss": 0.1924, - "step": 1197 - }, - { - "epoch": 2.44, - "learning_rate": 6.972633467345561e-06, - "loss": 0.1974, - "step": 1198 - }, - { - "epoch": 2.44, - "learning_rate": 6.9569201577926395e-06, - "loss": 0.1809, - "step": 1199 - }, - { - "epoch": 2.44, - "learning_rate": 6.941215124288354e-06, - "loss": 0.1954, - "step": 1200 - }, - { - "epoch": 2.45, - "learning_rate": 6.9255184095445695e-06, - "loss": 0.18, - "step": 1201 - }, - { - "epoch": 2.45, - "learning_rate": 6.909830056250527e-06, - "loss": 0.1976, - "step": 1202 - }, - { - "epoch": 2.45, - "learning_rate": 6.894150107072734e-06, - "loss": 0.187, - "step": 1203 - }, - { - "epoch": 2.45, - "learning_rate": 6.878478604654835e-06, - "loss": 0.1742, - "step": 1204 - }, - { - "epoch": 2.45, - "learning_rate": 6.862815591617503e-06, - "loss": 0.1931, - "step": 1205 - }, - { - "epoch": 2.46, - "learning_rate": 6.847161110558329e-06, - "loss": 0.1901, - "step": 1206 - }, - { - "epoch": 2.46, - "learning_rate": 6.831515204051693e-06, - "loss": 0.1965, - "step": 1207 - }, - { - "epoch": 2.46, - "learning_rate": 6.815877914648663e-06, - "loss": 0.1925, - "step": 1208 - }, - { - "epoch": 2.46, - "learning_rate": 6.800249284876859e-06, - "loss": 0.1901, - "step": 1209 - }, - { - "epoch": 2.46, - "learning_rate": 6.7846293572403624e-06, - "loss": 0.1775, - "step": 1210 - }, - { - "epoch": 2.47, - "learning_rate": 6.769018174219584e-06, - "loss": 0.186, - "step": 1211 - }, - { - "epoch": 2.47, - "learning_rate": 6.75341577827115e-06, - "loss": 0.1992, - "step": 1212 - }, - { - "epoch": 2.47, - "learning_rate": 6.737822211827794e-06, - "loss": 0.199, - "step": 1213 - }, - { - "epoch": 2.47, - "learning_rate": 6.722237517298232e-06, - "loss": 0.2039, - "step": 1214 - }, - { - "epoch": 2.47, - "learning_rate": 6.706661737067055e-06, - "loss": 0.1807, - "step": 1215 - }, - { - "epoch": 2.48, - "learning_rate": 6.691094913494611e-06, - "loss": 0.1839, - "step": 1216 - }, - { - "epoch": 2.48, - "learning_rate": 6.6755370889168816e-06, - "loss": 0.1869, - "step": 1217 - }, - { - "epoch": 2.48, - "learning_rate": 6.65998830564538e-06, - "loss": 0.1833, - "step": 1218 - }, - { - "epoch": 2.48, - "learning_rate": 6.644448605967034e-06, - "loss": 0.1808, - "step": 1219 - }, - { - "epoch": 2.48, - "learning_rate": 6.628918032144064e-06, - "loss": 0.1813, - "step": 1220 - }, - { - "epoch": 2.49, - "learning_rate": 6.613396626413869e-06, - "loss": 0.1807, - "step": 1221 - }, - { - "epoch": 2.49, - "learning_rate": 6.5978844309889165e-06, - "loss": 0.1858, - "step": 1222 - }, - { - "epoch": 2.49, - "learning_rate": 6.582381488056627e-06, - "loss": 0.176, - "step": 1223 - }, - { - "epoch": 2.49, - "learning_rate": 6.566887839779257e-06, - "loss": 0.1896, - "step": 1224 - }, - { - "epoch": 2.49, - "learning_rate": 6.55140352829378e-06, - "loss": 0.1888, - "step": 1225 - }, - { - "epoch": 2.5, - "learning_rate": 6.535928595711785e-06, - "loss": 0.191, - "step": 1226 - }, - { - "epoch": 2.5, - "learning_rate": 6.520463084119343e-06, - "loss": 0.2001, - "step": 1227 - }, - { - "epoch": 2.5, - "learning_rate": 6.505007035576918e-06, - "loss": 0.1889, - "step": 1228 - }, - { - "epoch": 2.5, - "learning_rate": 6.489560492119225e-06, - "loss": 0.184, - "step": 1229 - }, - { - "epoch": 2.51, - "learning_rate": 6.474123495755135e-06, - "loss": 0.1889, - "step": 1230 - }, - { - "epoch": 2.51, - "learning_rate": 6.458696088467555e-06, - "loss": 0.1777, - "step": 1231 - }, - { - "epoch": 2.51, - "learning_rate": 6.4432783122133124e-06, - "loss": 0.1865, - "step": 1232 - }, - { - "epoch": 2.51, - "learning_rate": 6.4278702089230375e-06, - "loss": 0.1853, - "step": 1233 - }, - { - "epoch": 2.51, - "learning_rate": 6.412471820501058e-06, - "loss": 0.1961, - "step": 1234 - }, - { - "epoch": 2.52, - "learning_rate": 6.397083188825278e-06, - "loss": 0.1905, - "step": 1235 - }, - { - "epoch": 2.52, - "learning_rate": 6.381704355747074e-06, - "loss": 0.1833, - "step": 1236 - }, - { - "epoch": 2.52, - "learning_rate": 6.366335363091165e-06, - "loss": 0.2025, - "step": 1237 - }, - { - "epoch": 2.52, - "learning_rate": 6.35097625265551e-06, - "loss": 0.1914, - "step": 1238 - }, - { - "epoch": 2.52, - "learning_rate": 6.335627066211196e-06, - "loss": 0.1874, - "step": 1239 - }, - { - "epoch": 2.53, - "learning_rate": 6.320287845502322e-06, - "loss": 0.1811, - "step": 1240 - }, - { - "epoch": 2.53, - "learning_rate": 6.304958632245871e-06, - "loss": 0.1885, - "step": 1241 - }, - { - "epoch": 2.53, - "learning_rate": 6.289639468131623e-06, - "loss": 0.1855, - "step": 1242 - }, - { - "epoch": 2.53, - "learning_rate": 6.274330394822021e-06, - "loss": 0.1877, - "step": 1243 - }, - { - "epoch": 2.53, - "learning_rate": 6.2590314539520695e-06, - "loss": 0.2002, - "step": 1244 - }, - { - "epoch": 2.54, - "learning_rate": 6.243742687129213e-06, - "loss": 0.1865, - "step": 1245 - }, - { - "epoch": 2.54, - "learning_rate": 6.228464135933225e-06, - "loss": 0.1837, - "step": 1246 - }, - { - "epoch": 2.54, - "learning_rate": 6.213195841916104e-06, - "loss": 0.1816, - "step": 1247 - }, - { - "epoch": 2.54, - "learning_rate": 6.197937846601945e-06, - "loss": 0.1885, - "step": 1248 - }, - { - "epoch": 2.54, - "learning_rate": 6.182690191486835e-06, - "loss": 0.1868, - "step": 1249 - }, - { - "epoch": 2.55, - "learning_rate": 6.167452918038744e-06, - "loss": 0.1851, - "step": 1250 - }, - { - "epoch": 2.55, - "learning_rate": 6.152226067697398e-06, - "loss": 0.1776, - "step": 1251 - }, - { - "epoch": 2.55, - "learning_rate": 6.137009681874192e-06, - "loss": 0.1843, - "step": 1252 - }, - { - "epoch": 2.55, - "learning_rate": 6.121803801952049e-06, - "loss": 0.179, - "step": 1253 - }, - { - "epoch": 2.55, - "learning_rate": 6.1066084692853224e-06, - "loss": 0.1889, - "step": 1254 - }, - { - "epoch": 2.56, - "learning_rate": 6.0914237251996815e-06, - "loss": 0.193, - "step": 1255 - }, - { - "epoch": 2.56, - "learning_rate": 6.076249610992002e-06, - "loss": 0.1863, - "step": 1256 - }, - { - "epoch": 2.56, - "learning_rate": 6.061086167930246e-06, - "loss": 0.1848, - "step": 1257 - }, - { - "epoch": 2.56, - "learning_rate": 6.045933437253352e-06, - "loss": 0.1841, - "step": 1258 - }, - { - "epoch": 2.56, - "learning_rate": 6.03079146017113e-06, - "loss": 0.1872, - "step": 1259 - }, - { - "epoch": 2.57, - "learning_rate": 6.015660277864142e-06, - "loss": 0.191, - "step": 1260 - }, - { - "epoch": 2.57, - "learning_rate": 6.000539931483594e-06, - "loss": 0.1868, - "step": 1261 - }, - { - "epoch": 2.57, - "learning_rate": 5.985430462151219e-06, - "loss": 0.1785, - "step": 1262 - }, - { - "epoch": 2.57, - "learning_rate": 5.970331910959169e-06, - "loss": 0.1943, - "step": 1263 - }, - { - "epoch": 2.57, - "learning_rate": 5.955244318969913e-06, - "loss": 0.1764, - "step": 1264 - }, - { - "epoch": 2.58, - "learning_rate": 5.940167727216091e-06, - "loss": 0.1863, - "step": 1265 - }, - { - "epoch": 2.58, - "learning_rate": 5.925102176700453e-06, - "loss": 0.1883, - "step": 1266 - }, - { - "epoch": 2.58, - "learning_rate": 5.910047708395703e-06, - "loss": 0.1797, - "step": 1267 - }, - { - "epoch": 2.58, - "learning_rate": 5.895004363244413e-06, - "loss": 0.1918, - "step": 1268 - }, - { - "epoch": 2.58, - "learning_rate": 5.8799721821589065e-06, - "loss": 0.187, - "step": 1269 - }, - { - "epoch": 2.59, - "learning_rate": 5.864951206021138e-06, - "loss": 0.1893, - "step": 1270 - }, - { - "epoch": 2.59, - "learning_rate": 5.849941475682591e-06, - "loss": 0.1727, - "step": 1271 - }, - { - "epoch": 2.59, - "learning_rate": 5.834943031964172e-06, - "loss": 0.1943, - "step": 1272 - }, - { - "epoch": 2.59, - "learning_rate": 5.819955915656079e-06, - "loss": 0.1823, - "step": 1273 - }, - { - "epoch": 2.59, - "learning_rate": 5.804980167517712e-06, - "loss": 0.1849, - "step": 1274 - }, - { - "epoch": 2.6, - "learning_rate": 5.790015828277553e-06, - "loss": 0.1888, - "step": 1275 - }, - { - "epoch": 2.6, - "learning_rate": 5.775062938633054e-06, - "loss": 0.1922, - "step": 1276 - }, - { - "epoch": 2.6, - "learning_rate": 5.760121539250532e-06, - "loss": 0.1893, - "step": 1277 - }, - { - "epoch": 2.6, - "learning_rate": 5.7451916707650445e-06, - "loss": 0.1911, - "step": 1278 - }, - { - "epoch": 2.6, - "learning_rate": 5.730273373780309e-06, - "loss": 0.1897, - "step": 1279 - }, - { - "epoch": 2.61, - "learning_rate": 5.7153666888685576e-06, - "loss": 0.1921, - "step": 1280 - }, - { - "epoch": 2.61, - "learning_rate": 5.700471656570437e-06, - "loss": 0.1853, - "step": 1281 - }, - { - "epoch": 2.61, - "learning_rate": 5.685588317394919e-06, - "loss": 0.1782, - "step": 1282 - }, - { - "epoch": 2.61, - "learning_rate": 5.670716711819159e-06, - "loss": 0.2019, - "step": 1283 - }, - { - "epoch": 2.62, - "learning_rate": 5.655856880288417e-06, - "loss": 0.1953, - "step": 1284 - }, - { - "epoch": 2.62, - "learning_rate": 5.6410088632159235e-06, - "loss": 0.191, - "step": 1285 - }, - { - "epoch": 2.62, - "learning_rate": 5.626172700982775e-06, - "loss": 0.1852, - "step": 1286 - }, - { - "epoch": 2.62, - "learning_rate": 5.611348433937831e-06, - "loss": 0.1744, - "step": 1287 - }, - { - "epoch": 2.62, - "learning_rate": 5.5965361023976e-06, - "loss": 0.1797, - "step": 1288 - }, - { - "epoch": 2.63, - "learning_rate": 5.581735746646134e-06, - "loss": 0.1845, - "step": 1289 - }, - { - "epoch": 2.63, - "learning_rate": 5.566947406934909e-06, - "loss": 0.1762, - "step": 1290 - }, - { - "epoch": 2.63, - "learning_rate": 5.552171123482724e-06, - "loss": 0.1776, - "step": 1291 - }, - { - "epoch": 2.63, - "learning_rate": 5.537406936475592e-06, - "loss": 0.1839, - "step": 1292 - }, - { - "epoch": 2.63, - "learning_rate": 5.522654886066625e-06, - "loss": 0.1875, - "step": 1293 - }, - { - "epoch": 2.64, - "learning_rate": 5.507915012375928e-06, - "loss": 0.1885, - "step": 1294 - }, - { - "epoch": 2.64, - "learning_rate": 5.493187355490495e-06, - "loss": 0.2033, - "step": 1295 - }, - { - "epoch": 2.64, - "learning_rate": 5.478471955464093e-06, - "loss": 0.1992, - "step": 1296 - }, - { - "epoch": 2.64, - "learning_rate": 5.463768852317146e-06, - "loss": 0.1844, - "step": 1297 - }, - { - "epoch": 2.64, - "learning_rate": 5.449078086036643e-06, - "loss": 0.1796, - "step": 1298 - }, - { - "epoch": 2.65, - "learning_rate": 5.434399696576019e-06, - "loss": 0.1879, - "step": 1299 - }, - { - "epoch": 2.65, - "learning_rate": 5.4197337238550565e-06, - "loss": 0.1794, - "step": 1300 - }, - { - "epoch": 2.65, - "learning_rate": 5.405080207759759e-06, - "loss": 0.1851, - "step": 1301 - }, - { - "epoch": 2.65, - "learning_rate": 5.390439188142254e-06, - "loss": 0.1949, - "step": 1302 - }, - { - "epoch": 2.65, - "learning_rate": 5.375810704820688e-06, - "loss": 0.183, - "step": 1303 - }, - { - "epoch": 2.66, - "learning_rate": 5.361194797579108e-06, - "loss": 0.1952, - "step": 1304 - }, - { - "epoch": 2.66, - "learning_rate": 5.346591506167362e-06, - "loss": 0.1969, - "step": 1305 - }, - { - "epoch": 2.66, - "learning_rate": 5.33200087030099e-06, - "loss": 0.1956, - "step": 1306 - }, - { - "epoch": 2.66, - "learning_rate": 5.3174229296611064e-06, - "loss": 0.1889, - "step": 1307 - }, - { - "epoch": 2.66, - "learning_rate": 5.3028577238943056e-06, - "loss": 0.193, - "step": 1308 - }, - { - "epoch": 2.67, - "learning_rate": 5.288305292612545e-06, - "loss": 0.1948, - "step": 1309 - }, - { - "epoch": 2.67, - "learning_rate": 5.273765675393042e-06, - "loss": 0.1861, - "step": 1310 - }, - { - "epoch": 2.67, - "learning_rate": 5.259238911778159e-06, - "loss": 0.1757, - "step": 1311 - }, - { - "epoch": 2.67, - "learning_rate": 5.244725041275316e-06, - "loss": 0.2063, - "step": 1312 - }, - { - "epoch": 2.67, - "learning_rate": 5.2302241033568476e-06, - "loss": 0.1738, - "step": 1313 - }, - { - "epoch": 2.68, - "learning_rate": 5.215736137459932e-06, - "loss": 0.1927, - "step": 1314 - }, - { - "epoch": 2.68, - "learning_rate": 5.201261182986463e-06, - "loss": 0.1833, - "step": 1315 - }, - { - "epoch": 2.68, - "learning_rate": 5.1867992793029435e-06, - "loss": 0.1855, - "step": 1316 - }, - { - "epoch": 2.68, - "learning_rate": 5.172350465740397e-06, - "loss": 0.1776, - "step": 1317 - }, - { - "epoch": 2.68, - "learning_rate": 5.1579147815942355e-06, - "loss": 0.1773, - "step": 1318 - }, - { - "epoch": 2.69, - "learning_rate": 5.143492266124164e-06, - "loss": 0.1947, - "step": 1319 - }, - { - "epoch": 2.69, - "learning_rate": 5.129082958554081e-06, - "loss": 0.182, - "step": 1320 - }, - { - "epoch": 2.69, - "learning_rate": 5.114686898071947e-06, - "loss": 0.1831, - "step": 1321 - }, - { - "epoch": 2.69, - "learning_rate": 5.10030412382972e-06, - "loss": 0.1866, - "step": 1322 - }, - { - "epoch": 2.69, - "learning_rate": 5.0859346749432085e-06, - "loss": 0.1855, - "step": 1323 - }, - { - "epoch": 2.7, - "learning_rate": 5.071578590491983e-06, - "loss": 0.1894, - "step": 1324 - }, - { - "epoch": 2.7, - "learning_rate": 5.057235909519271e-06, - "loss": 0.1859, - "step": 1325 - }, - { - "epoch": 2.7, - "learning_rate": 5.0429066710318465e-06, - "loss": 0.1712, - "step": 1326 - }, - { - "epoch": 2.7, - "learning_rate": 5.028590913999918e-06, - "loss": 0.18, - "step": 1327 - }, - { - "epoch": 2.7, - "learning_rate": 5.014288677357046e-06, - "loss": 0.1698, - "step": 1328 - }, - { - "epoch": 2.71, - "learning_rate": 5.000000000000003e-06, - "loss": 0.1766, - "step": 1329 - }, - { - "epoch": 2.71, - "learning_rate": 4.985724920788691e-06, - "loss": 0.1744, - "step": 1330 - }, - { - "epoch": 2.71, - "learning_rate": 4.971463478546037e-06, - "loss": 0.1815, - "step": 1331 - }, - { - "epoch": 2.71, - "learning_rate": 4.957215712057867e-06, - "loss": 0.1852, - "step": 1332 - }, - { - "epoch": 2.71, - "learning_rate": 4.942981660072832e-06, - "loss": 0.1966, - "step": 1333 - }, - { - "epoch": 2.72, - "learning_rate": 4.928761361302269e-06, - "loss": 0.1859, - "step": 1334 - }, - { - "epoch": 2.72, - "learning_rate": 4.9145548544201184e-06, - "loss": 0.1854, - "step": 1335 - }, - { - "epoch": 2.72, - "learning_rate": 4.900362178062813e-06, - "loss": 0.1904, - "step": 1336 - }, - { - "epoch": 2.72, - "learning_rate": 4.886183370829159e-06, - "loss": 0.1859, - "step": 1337 - }, - { - "epoch": 2.73, - "learning_rate": 4.8720184712802644e-06, - "loss": 0.1861, - "step": 1338 - }, - { - "epoch": 2.73, - "learning_rate": 4.8578675179394e-06, - "loss": 0.1788, - "step": 1339 - }, - { - "epoch": 2.73, - "learning_rate": 4.843730549291909e-06, - "loss": 0.1911, - "step": 1340 - }, - { - "epoch": 2.73, - "learning_rate": 4.8296076037851056e-06, - "loss": 0.173, - "step": 1341 - }, - { - "epoch": 2.73, - "learning_rate": 4.815498719828164e-06, - "loss": 0.1942, - "step": 1342 - }, - { - "epoch": 2.74, - "learning_rate": 4.801403935792013e-06, - "loss": 0.1786, - "step": 1343 - }, - { - "epoch": 2.74, - "learning_rate": 4.7873232900092414e-06, - "loss": 0.1797, - "step": 1344 - }, - { - "epoch": 2.74, - "learning_rate": 4.7732568207739825e-06, - "loss": 0.1871, - "step": 1345 - }, - { - "epoch": 2.74, - "learning_rate": 4.7592045663418175e-06, - "loss": 0.1804, - "step": 1346 - }, - { - "epoch": 2.74, - "learning_rate": 4.745166564929665e-06, - "loss": 0.1915, - "step": 1347 - }, - { - "epoch": 2.75, - "learning_rate": 4.7311428547156845e-06, - "loss": 0.1737, - "step": 1348 - }, - { - "epoch": 2.75, - "learning_rate": 4.717133473839163e-06, - "loss": 0.1944, - "step": 1349 - }, - { - "epoch": 2.75, - "learning_rate": 4.703138460400427e-06, - "loss": 0.1936, - "step": 1350 - }, - { - "epoch": 2.75, - "learning_rate": 4.6891578524607215e-06, - "loss": 0.1889, - "step": 1351 - }, - { - "epoch": 2.75, - "learning_rate": 4.675191688042118e-06, - "loss": 0.1702, - "step": 1352 - }, - { - "epoch": 2.76, - "learning_rate": 4.661240005127395e-06, - "loss": 0.1837, - "step": 1353 - }, - { - "epoch": 2.76, - "learning_rate": 4.64730284165996e-06, - "loss": 0.188, - "step": 1354 - }, - { - "epoch": 2.76, - "learning_rate": 4.633380235543733e-06, - "loss": 0.1861, - "step": 1355 - }, - { - "epoch": 2.76, - "learning_rate": 4.6194722246430365e-06, - "loss": 0.1862, - "step": 1356 - }, - { - "epoch": 2.76, - "learning_rate": 4.605578846782504e-06, - "loss": 0.1849, - "step": 1357 - }, - { - "epoch": 2.77, - "learning_rate": 4.59170013974697e-06, - "loss": 0.1919, - "step": 1358 - }, - { - "epoch": 2.77, - "learning_rate": 4.577836141281368e-06, - "loss": 0.1809, - "step": 1359 - }, - { - "epoch": 2.77, - "learning_rate": 4.563986889090637e-06, - "loss": 0.1919, - "step": 1360 - }, - { - "epoch": 2.77, - "learning_rate": 4.550152420839604e-06, - "loss": 0.1924, - "step": 1361 - }, - { - "epoch": 2.77, - "learning_rate": 4.536332774152892e-06, - "loss": 0.1821, - "step": 1362 - }, - { - "epoch": 2.78, - "learning_rate": 4.522527986614816e-06, - "loss": 0.1851, - "step": 1363 - }, - { - "epoch": 2.78, - "learning_rate": 4.508738095769278e-06, - "loss": 0.1901, - "step": 1364 - }, - { - "epoch": 2.78, - "learning_rate": 4.494963139119663e-06, - "loss": 0.1911, - "step": 1365 - }, - { - "epoch": 2.78, - "learning_rate": 4.481203154128754e-06, - "loss": 0.1854, - "step": 1366 - }, - { - "epoch": 2.78, - "learning_rate": 4.4674581782186e-06, - "loss": 0.1872, - "step": 1367 - }, - { - "epoch": 2.79, - "learning_rate": 4.4537282487704444e-06, - "loss": 0.1848, - "step": 1368 - }, - { - "epoch": 2.79, - "learning_rate": 4.440013403124597e-06, - "loss": 0.1832, - "step": 1369 - }, - { - "epoch": 2.79, - "learning_rate": 4.42631367858035e-06, - "loss": 0.1807, - "step": 1370 - }, - { - "epoch": 2.79, - "learning_rate": 4.41262911239588e-06, - "loss": 0.1744, - "step": 1371 - }, - { - "epoch": 2.79, - "learning_rate": 4.3989597417881295e-06, - "loss": 0.1924, - "step": 1372 - }, - { - "epoch": 2.8, - "learning_rate": 4.385305603932718e-06, - "loss": 0.1807, - "step": 1373 - }, - { - "epoch": 2.8, - "learning_rate": 4.371666735963836e-06, - "loss": 0.1948, - "step": 1374 - }, - { - "epoch": 2.8, - "learning_rate": 4.358043174974147e-06, - "loss": 0.1925, - "step": 1375 - }, - { - "epoch": 2.8, - "learning_rate": 4.3444349580146836e-06, - "loss": 0.1896, - "step": 1376 - }, - { - "epoch": 2.8, - "learning_rate": 4.330842122094749e-06, - "loss": 0.1937, - "step": 1377 - }, - { - "epoch": 2.81, - "learning_rate": 4.317264704181818e-06, - "loss": 0.2042, - "step": 1378 - }, - { - "epoch": 2.81, - "learning_rate": 4.303702741201431e-06, - "loss": 0.1778, - "step": 1379 - }, - { - "epoch": 2.81, - "learning_rate": 4.290156270037096e-06, - "loss": 0.1737, - "step": 1380 - }, - { - "epoch": 2.81, - "learning_rate": 4.276625327530194e-06, - "loss": 0.1863, - "step": 1381 - }, - { - "epoch": 2.81, - "learning_rate": 4.263109950479864e-06, - "loss": 0.1731, - "step": 1382 - }, - { - "epoch": 2.82, - "learning_rate": 4.249610175642929e-06, - "loss": 0.1845, - "step": 1383 - }, - { - "epoch": 2.82, - "learning_rate": 4.236126039733768e-06, - "loss": 0.1827, - "step": 1384 - }, - { - "epoch": 2.82, - "learning_rate": 4.222657579424225e-06, - "loss": 0.1809, - "step": 1385 - }, - { - "epoch": 2.82, - "learning_rate": 4.20920483134352e-06, - "loss": 0.1828, - "step": 1386 - }, - { - "epoch": 2.82, - "learning_rate": 4.195767832078134e-06, - "loss": 0.1899, - "step": 1387 - }, - { - "epoch": 2.83, - "learning_rate": 4.18234661817173e-06, - "loss": 0.1812, - "step": 1388 - }, - { - "epoch": 2.83, - "learning_rate": 4.168941226125029e-06, - "loss": 0.1909, - "step": 1389 - }, - { - "epoch": 2.83, - "learning_rate": 4.155551692395726e-06, - "loss": 0.1829, - "step": 1390 - }, - { - "epoch": 2.83, - "learning_rate": 4.1421780533983865e-06, - "loss": 0.1965, - "step": 1391 - }, - { - "epoch": 2.84, - "learning_rate": 4.128820345504351e-06, - "loss": 0.1878, - "step": 1392 - }, - { - "epoch": 2.84, - "learning_rate": 4.115478605041628e-06, - "loss": 0.1729, - "step": 1393 - }, - { - "epoch": 2.84, - "learning_rate": 4.1021528682948064e-06, - "loss": 0.1795, - "step": 1394 - }, - { - "epoch": 2.84, - "learning_rate": 4.088843171504947e-06, - "loss": 0.1897, - "step": 1395 - }, - { - "epoch": 2.84, - "learning_rate": 4.07554955086949e-06, - "loss": 0.1849, - "step": 1396 - }, - { - "epoch": 2.85, - "learning_rate": 4.062272042542151e-06, - "loss": 0.1868, - "step": 1397 - }, - { - "epoch": 2.85, - "learning_rate": 4.0490106826328294e-06, - "loss": 0.1928, - "step": 1398 - }, - { - "epoch": 2.85, - "learning_rate": 4.035765507207509e-06, - "loss": 0.1845, - "step": 1399 - }, - { - "epoch": 2.85, - "learning_rate": 4.022536552288158e-06, - "loss": 0.1977, - "step": 1400 - }, - { - "epoch": 2.85, - "learning_rate": 4.009323853852619e-06, - "loss": 0.1873, - "step": 1401 - }, - { - "epoch": 2.86, - "learning_rate": 3.996127447834536e-06, - "loss": 0.1886, - "step": 1402 - }, - { - "epoch": 2.86, - "learning_rate": 3.982947370123236e-06, - "loss": 0.1787, - "step": 1403 - }, - { - "epoch": 2.86, - "learning_rate": 3.9697836565636484e-06, - "loss": 0.1703, - "step": 1404 - }, - { - "epoch": 2.86, - "learning_rate": 3.9566363429561895e-06, - "loss": 0.1903, - "step": 1405 - }, - { - "epoch": 2.86, - "learning_rate": 3.943505465056677e-06, - "loss": 0.1863, - "step": 1406 - }, - { - "epoch": 2.87, - "learning_rate": 3.93039105857623e-06, - "loss": 0.1888, - "step": 1407 - }, - { - "epoch": 2.87, - "learning_rate": 3.917293159181168e-06, - "loss": 0.1778, - "step": 1408 - }, - { - "epoch": 2.87, - "learning_rate": 3.904211802492922e-06, - "loss": 0.1864, - "step": 1409 - }, - { - "epoch": 2.87, - "learning_rate": 3.89114702408793e-06, - "loss": 0.1855, - "step": 1410 - }, - { - "epoch": 2.87, - "learning_rate": 3.878098859497546e-06, - "loss": 0.1952, - "step": 1411 - }, - { - "epoch": 2.88, - "learning_rate": 3.865067344207935e-06, - "loss": 0.185, - "step": 1412 - }, - { - "epoch": 2.88, - "learning_rate": 3.852052513659989e-06, - "loss": 0.1791, - "step": 1413 - }, - { - "epoch": 2.88, - "learning_rate": 3.839054403249221e-06, - "loss": 0.1875, - "step": 1414 - }, - { - "epoch": 2.88, - "learning_rate": 3.826073048325664e-06, - "loss": 0.1794, - "step": 1415 - }, - { - "epoch": 2.88, - "learning_rate": 3.8131084841938014e-06, - "loss": 0.1819, - "step": 1416 - }, - { - "epoch": 2.89, - "learning_rate": 3.800160746112431e-06, - "loss": 0.1863, - "step": 1417 - }, - { - "epoch": 2.89, - "learning_rate": 3.7872298692945995e-06, - "loss": 0.1875, - "step": 1418 - }, - { - "epoch": 2.89, - "learning_rate": 3.774315888907498e-06, - "loss": 0.1784, - "step": 1419 - }, - { - "epoch": 2.89, - "learning_rate": 3.7614188400723606e-06, - "loss": 0.1792, - "step": 1420 - }, - { - "epoch": 2.89, - "learning_rate": 3.748538757864383e-06, - "loss": 0.1889, - "step": 1421 - }, - { - "epoch": 2.9, - "learning_rate": 3.735675677312609e-06, - "loss": 0.1806, - "step": 1422 - }, - { - "epoch": 2.9, - "learning_rate": 3.7228296333998482e-06, - "loss": 0.1829, - "step": 1423 - }, - { - "epoch": 2.9, - "learning_rate": 3.710000661062578e-06, - "loss": 0.1862, - "step": 1424 - }, - { - "epoch": 2.9, - "learning_rate": 3.697188795190837e-06, - "loss": 0.1786, - "step": 1425 - }, - { - "epoch": 2.9, - "learning_rate": 3.684394070628159e-06, - "loss": 0.1786, - "step": 1426 - }, - { - "epoch": 2.91, - "learning_rate": 3.671616522171445e-06, - "loss": 0.1929, - "step": 1427 - }, - { - "epoch": 2.91, - "learning_rate": 3.6588561845708893e-06, - "loss": 0.1922, - "step": 1428 - }, - { - "epoch": 2.91, - "learning_rate": 3.646113092529878e-06, - "loss": 0.1843, - "step": 1429 - }, - { - "epoch": 2.91, - "learning_rate": 3.633387280704895e-06, - "loss": 0.1861, - "step": 1430 - }, - { - "epoch": 2.91, - "learning_rate": 3.620678783705427e-06, - "loss": 0.1733, - "step": 1431 - }, - { - "epoch": 2.92, - "learning_rate": 3.6079876360938814e-06, - "loss": 0.1845, - "step": 1432 - }, - { - "epoch": 2.92, - "learning_rate": 3.595313872385463e-06, - "loss": 0.1746, - "step": 1433 - }, - { - "epoch": 2.92, - "learning_rate": 3.5826575270481147e-06, - "loss": 0.1831, - "step": 1434 - }, - { - "epoch": 2.92, - "learning_rate": 3.570018634502399e-06, - "loss": 0.1791, - "step": 1435 - }, - { - "epoch": 2.92, - "learning_rate": 3.5573972291214154e-06, - "loss": 0.1818, - "step": 1436 - }, - { - "epoch": 2.93, - "learning_rate": 3.5447933452307117e-06, - "loss": 0.1884, - "step": 1437 - }, - { - "epoch": 2.93, - "learning_rate": 3.5322070171081735e-06, - "loss": 0.1831, - "step": 1438 - }, - { - "epoch": 2.93, - "learning_rate": 3.5196382789839477e-06, - "loss": 0.18, - "step": 1439 - }, - { - "epoch": 2.93, - "learning_rate": 3.507087165040344e-06, - "loss": 0.1881, - "step": 1440 - }, - { - "epoch": 2.93, - "learning_rate": 3.4945537094117254e-06, - "loss": 0.1832, - "step": 1441 - }, - { - "epoch": 2.94, - "learning_rate": 3.4820379461844555e-06, - "loss": 0.1842, - "step": 1442 - }, - { - "epoch": 2.94, - "learning_rate": 3.469539909396764e-06, - "loss": 0.1745, - "step": 1443 - }, - { - "epoch": 2.94, - "learning_rate": 3.4570596330386775e-06, - "loss": 0.1783, - "step": 1444 - }, - { - "epoch": 2.94, - "learning_rate": 3.444597151051917e-06, - "loss": 0.1797, - "step": 1445 - }, - { - "epoch": 2.95, - "learning_rate": 3.4321524973298146e-06, - "loss": 0.1795, - "step": 1446 - }, - { - "epoch": 2.95, - "learning_rate": 3.4197257057172116e-06, - "loss": 0.1792, - "step": 1447 - }, - { - "epoch": 2.95, - "learning_rate": 3.4073168100103725e-06, - "loss": 0.1752, - "step": 1448 - }, - { - "epoch": 2.95, - "learning_rate": 3.394925843956892e-06, - "loss": 0.1755, - "step": 1449 - }, - { - "epoch": 2.95, - "learning_rate": 3.382552841255603e-06, - "loss": 0.1722, - "step": 1450 - }, - { - "epoch": 2.96, - "learning_rate": 3.370197835556485e-06, - "loss": 0.1867, - "step": 1451 - }, - { - "epoch": 2.96, - "learning_rate": 3.357860860460569e-06, - "loss": 0.19, - "step": 1452 - }, - { - "epoch": 2.96, - "learning_rate": 3.3455419495198504e-06, - "loss": 0.1809, - "step": 1453 - }, - { - "epoch": 2.96, - "learning_rate": 3.3332411362372063e-06, - "loss": 0.1772, - "step": 1454 - }, - { - "epoch": 2.96, - "learning_rate": 3.320958454066282e-06, - "loss": 0.1799, - "step": 1455 - }, - { - "epoch": 2.97, - "learning_rate": 3.308693936411421e-06, - "loss": 0.1891, - "step": 1456 - }, - { - "epoch": 2.97, - "learning_rate": 3.2964476166275572e-06, - "loss": 0.1851, - "step": 1457 - }, - { - "epoch": 2.97, - "learning_rate": 3.2842195280201394e-06, - "loss": 0.18, - "step": 1458 - }, - { - "epoch": 2.97, - "learning_rate": 3.2720097038450394e-06, - "loss": 0.1831, - "step": 1459 - }, - { - "epoch": 2.97, - "learning_rate": 3.259818177308447e-06, - "loss": 0.1733, - "step": 1460 - }, - { - "epoch": 2.98, - "learning_rate": 3.2476449815667955e-06, - "loss": 0.1935, - "step": 1461 - }, - { - "epoch": 2.98, - "learning_rate": 3.23549014972666e-06, - "loss": 0.1825, - "step": 1462 - }, - { - "epoch": 2.98, - "learning_rate": 3.223353714844678e-06, - "loss": 0.1831, - "step": 1463 - }, - { - "epoch": 2.98, - "learning_rate": 3.2112357099274495e-06, - "loss": 0.1828, - "step": 1464 - }, - { - "epoch": 2.98, - "learning_rate": 3.1991361679314546e-06, - "loss": 0.1793, - "step": 1465 - }, - { - "epoch": 2.99, - "learning_rate": 3.1870551217629608e-06, - "loss": 0.1818, - "step": 1466 - }, - { - "epoch": 2.99, - "learning_rate": 3.1749926042779335e-06, - "loss": 0.1779, - "step": 1467 - }, - { - "epoch": 2.99, - "learning_rate": 3.1629486482819448e-06, - "loss": 0.1875, - "step": 1468 - }, - { - "epoch": 2.99, - "learning_rate": 3.1509232865300886e-06, - "loss": 0.1814, - "step": 1469 - }, - { - "epoch": 2.99, - "learning_rate": 3.138916551726893e-06, - "loss": 0.1913, - "step": 1470 - }, - { - "epoch": 3.0, - "learning_rate": 3.126928476526222e-06, - "loss": 0.1722, - "step": 1471 - }, - { - "epoch": 3.0, - "learning_rate": 3.1149590935311968e-06, - "loss": 0.1809, - "step": 1472 - }, - { - "epoch": 3.0, - "learning_rate": 3.1030084352940936e-06, - "loss": 0.1629, - "step": 1473 - }, - { - "epoch": 3.0, - "learning_rate": 3.091076534316273e-06, - "loss": 0.1008, - "step": 1474 - }, - { - "epoch": 3.0, - "learning_rate": 3.079163423048085e-06, - "loss": 0.0921, - "step": 1475 - }, - { - "epoch": 3.01, - "learning_rate": 3.0672691338887705e-06, - "loss": 0.0962, - "step": 1476 - }, - { - "epoch": 3.01, - "learning_rate": 3.0553936991863876e-06, - "loss": 0.098, - "step": 1477 - }, - { - "epoch": 3.01, - "learning_rate": 3.043537151237712e-06, - "loss": 0.0913, - "step": 1478 - }, - { - "epoch": 3.01, - "learning_rate": 3.0316995222881584e-06, - "loss": 0.0835, - "step": 1479 - }, - { - "epoch": 3.01, - "learning_rate": 3.0198808445316873e-06, - "loss": 0.0851, - "step": 1480 - }, - { - "epoch": 3.02, - "learning_rate": 3.0080811501107187e-06, - "loss": 0.0931, - "step": 1481 - }, - { - "epoch": 3.02, - "learning_rate": 2.996300471116047e-06, - "loss": 0.0875, - "step": 1482 - }, - { - "epoch": 3.02, - "learning_rate": 2.984538839586749e-06, - "loss": 0.0885, - "step": 1483 - }, - { - "epoch": 3.02, - "learning_rate": 2.9727962875101e-06, - "loss": 0.0827, - "step": 1484 - }, - { - "epoch": 3.02, - "learning_rate": 2.961072846821489e-06, - "loss": 0.079, - "step": 1485 - }, - { - "epoch": 3.03, - "learning_rate": 2.9493685494043213e-06, - "loss": 0.078, - "step": 1486 - }, - { - "epoch": 3.03, - "learning_rate": 2.9376834270899523e-06, - "loss": 0.078, - "step": 1487 - }, - { - "epoch": 3.03, - "learning_rate": 2.9260175116575827e-06, - "loss": 0.0814, - "step": 1488 - }, - { - "epoch": 3.03, - "learning_rate": 2.9143708348341693e-06, - "loss": 0.0782, - "step": 1489 - }, - { - "epoch": 3.03, - "learning_rate": 2.9027434282943555e-06, - "loss": 0.0775, - "step": 1490 - }, - { - "epoch": 3.04, - "learning_rate": 2.891135323660373e-06, - "loss": 0.076, - "step": 1491 - }, - { - "epoch": 3.04, - "learning_rate": 2.8795465525019683e-06, - "loss": 0.0763, - "step": 1492 - }, - { - "epoch": 3.04, - "learning_rate": 2.8679771463362973e-06, - "loss": 0.078, - "step": 1493 - }, - { - "epoch": 3.04, - "learning_rate": 2.856427136627854e-06, - "loss": 0.0786, - "step": 1494 - }, - { - "epoch": 3.04, - "learning_rate": 2.8448965547883823e-06, - "loss": 0.0802, - "step": 1495 - }, - { - "epoch": 3.05, - "learning_rate": 2.8333854321767884e-06, - "loss": 0.0821, - "step": 1496 - }, - { - "epoch": 3.05, - "learning_rate": 2.821893800099056e-06, - "loss": 0.0806, - "step": 1497 - }, - { - "epoch": 3.05, - "learning_rate": 2.810421689808164e-06, - "loss": 0.0802, - "step": 1498 - }, - { - "epoch": 3.05, - "learning_rate": 2.798969132503997e-06, - "loss": 0.0774, - "step": 1499 - }, - { - "epoch": 3.05, - "learning_rate": 2.787536159333263e-06, - "loss": 0.0713, - "step": 1500 - }, - { - "epoch": 3.06, - "learning_rate": 2.7761228013894103e-06, - "loss": 0.0849, - "step": 1501 - }, - { - "epoch": 3.06, - "learning_rate": 2.764729089712538e-06, - "loss": 0.08, - "step": 1502 - }, - { - "epoch": 3.06, - "learning_rate": 2.7533550552893197e-06, - "loss": 0.0802, - "step": 1503 - }, - { - "epoch": 3.06, - "learning_rate": 2.7420007290529118e-06, - "loss": 0.083, - "step": 1504 - }, - { - "epoch": 3.07, - "learning_rate": 2.730666141882866e-06, - "loss": 0.0819, - "step": 1505 - }, - { - "epoch": 3.07, - "learning_rate": 2.7193513246050583e-06, - "loss": 0.0792, - "step": 1506 - }, - { - "epoch": 3.07, - "learning_rate": 2.708056307991592e-06, - "loss": 0.0778, - "step": 1507 - }, - { - "epoch": 3.07, - "learning_rate": 2.6967811227607297e-06, - "loss": 0.079, - "step": 1508 - }, - { - "epoch": 3.07, - "learning_rate": 2.6855257995767914e-06, - "loss": 0.0825, - "step": 1509 - }, - { - "epoch": 3.08, - "learning_rate": 2.674290369050082e-06, - "loss": 0.0826, - "step": 1510 - }, - { - "epoch": 3.08, - "learning_rate": 2.6630748617368053e-06, - "loss": 0.083, - "step": 1511 - }, - { - "epoch": 3.08, - "learning_rate": 2.6518793081389815e-06, - "loss": 0.0816, - "step": 1512 - }, - { - "epoch": 3.08, - "learning_rate": 2.6407037387043665e-06, - "loss": 0.0863, - "step": 1513 - }, - { - "epoch": 3.08, - "learning_rate": 2.6295481838263628e-06, - "loss": 0.0838, - "step": 1514 - }, - { - "epoch": 3.09, - "learning_rate": 2.618412673843942e-06, - "loss": 0.0748, - "step": 1515 - }, - { - "epoch": 3.09, - "learning_rate": 2.607297239041564e-06, - "loss": 0.0809, - "step": 1516 - }, - { - "epoch": 3.09, - "learning_rate": 2.5962019096490876e-06, - "loss": 0.0891, - "step": 1517 - }, - { - "epoch": 3.09, - "learning_rate": 2.5851267158416937e-06, - "loss": 0.0828, - "step": 1518 - }, - { - "epoch": 3.09, - "learning_rate": 2.574071687739801e-06, - "loss": 0.0831, - "step": 1519 - }, - { - "epoch": 3.1, - "learning_rate": 2.5630368554089935e-06, - "loss": 0.0842, - "step": 1520 - }, - { - "epoch": 3.1, - "learning_rate": 2.5520222488599135e-06, - "loss": 0.0759, - "step": 1521 - }, - { - "epoch": 3.1, - "learning_rate": 2.5410278980482107e-06, - "loss": 0.0812, - "step": 1522 - }, - { - "epoch": 3.1, - "learning_rate": 2.5300538328744416e-06, - "loss": 0.0785, - "step": 1523 - }, - { - "epoch": 3.1, - "learning_rate": 2.5191000831839897e-06, - "loss": 0.0858, - "step": 1524 - }, - { - "epoch": 3.11, - "learning_rate": 2.5081666787669977e-06, - "loss": 0.0796, - "step": 1525 - }, - { - "epoch": 3.11, - "learning_rate": 2.49725364935827e-06, - "loss": 0.0734, - "step": 1526 - }, - { - "epoch": 3.11, - "learning_rate": 2.486361024637197e-06, - "loss": 0.0793, - "step": 1527 - }, - { - "epoch": 3.11, - "learning_rate": 2.475488834227684e-06, - "loss": 0.079, - "step": 1528 - }, - { - "epoch": 3.11, - "learning_rate": 2.464637107698046e-06, - "loss": 0.0815, - "step": 1529 - }, - { - "epoch": 3.12, - "learning_rate": 2.4538058745609637e-06, - "loss": 0.0834, - "step": 1530 - }, - { - "epoch": 3.12, - "learning_rate": 2.442995164273372e-06, - "loss": 0.0767, - "step": 1531 - }, - { - "epoch": 3.12, - "learning_rate": 2.4322050062363945e-06, - "loss": 0.0764, - "step": 1532 - }, - { - "epoch": 3.12, - "learning_rate": 2.421435429795258e-06, - "loss": 0.0724, - "step": 1533 - }, - { - "epoch": 3.12, - "learning_rate": 2.410686464239217e-06, - "loss": 0.0832, - "step": 1534 - }, - { - "epoch": 3.13, - "learning_rate": 2.3999581388014692e-06, - "loss": 0.0827, - "step": 1535 - }, - { - "epoch": 3.13, - "learning_rate": 2.389250482659089e-06, - "loss": 0.0817, - "step": 1536 - }, - { - "epoch": 3.13, - "learning_rate": 2.378563524932922e-06, - "loss": 0.0841, - "step": 1537 - }, - { - "epoch": 3.13, - "learning_rate": 2.3678972946875324e-06, - "loss": 0.079, - "step": 1538 - }, - { - "epoch": 3.13, - "learning_rate": 2.3572518209311124e-06, - "loss": 0.0804, - "step": 1539 - }, - { - "epoch": 3.14, - "learning_rate": 2.3466271326153977e-06, - "loss": 0.0768, - "step": 1540 - }, - { - "epoch": 3.14, - "learning_rate": 2.336023258635607e-06, - "loss": 0.0795, - "step": 1541 - }, - { - "epoch": 3.14, - "learning_rate": 2.325440227830341e-06, - "loss": 0.076, - "step": 1542 - }, - { - "epoch": 3.14, - "learning_rate": 2.314878068981522e-06, - "loss": 0.0802, - "step": 1543 - }, - { - "epoch": 3.14, - "learning_rate": 2.304336810814305e-06, - "loss": 0.0795, - "step": 1544 - }, - { - "epoch": 3.15, - "learning_rate": 2.2938164819969945e-06, - "loss": 0.0871, - "step": 1545 - }, - { - "epoch": 3.15, - "learning_rate": 2.2833171111409934e-06, - "loss": 0.081, - "step": 1546 - }, - { - "epoch": 3.15, - "learning_rate": 2.2728387268006924e-06, - "loss": 0.0821, - "step": 1547 - }, - { - "epoch": 3.15, - "learning_rate": 2.262381357473412e-06, - "loss": 0.0869, - "step": 1548 - }, - { - "epoch": 3.15, - "learning_rate": 2.2519450315993186e-06, - "loss": 0.0794, - "step": 1549 - }, - { - "epoch": 3.16, - "learning_rate": 2.2415297775613465e-06, - "loss": 0.0868, - "step": 1550 - }, - { - "epoch": 3.16, - "learning_rate": 2.2311356236851267e-06, - "loss": 0.0908, - "step": 1551 - }, - { - "epoch": 3.16, - "learning_rate": 2.2207625982389023e-06, - "loss": 0.0849, - "step": 1552 - }, - { - "epoch": 3.16, - "learning_rate": 2.210410729433454e-06, - "loss": 0.0815, - "step": 1553 - }, - { - "epoch": 3.16, - "learning_rate": 2.2000800454220285e-06, - "loss": 0.0793, - "step": 1554 - }, - { - "epoch": 3.17, - "learning_rate": 2.189770574300253e-06, - "loss": 0.0852, - "step": 1555 - }, - { - "epoch": 3.17, - "learning_rate": 2.1794823441060663e-06, - "loss": 0.079, - "step": 1556 - }, - { - "epoch": 3.17, - "learning_rate": 2.169215382819636e-06, - "loss": 0.0771, - "step": 1557 - }, - { - "epoch": 3.17, - "learning_rate": 2.1589697183632963e-06, - "loss": 0.0774, - "step": 1558 - }, - { - "epoch": 3.18, - "learning_rate": 2.1487453786014513e-06, - "loss": 0.0794, - "step": 1559 - }, - { - "epoch": 3.18, - "learning_rate": 2.1385423913405145e-06, - "loss": 0.0772, - "step": 1560 - }, - { - "epoch": 3.18, - "learning_rate": 2.1283607843288244e-06, - "loss": 0.0866, - "step": 1561 - }, - { - "epoch": 3.18, - "learning_rate": 2.118200585256576e-06, - "loss": 0.0791, - "step": 1562 - }, - { - "epoch": 3.18, - "learning_rate": 2.1080618217557458e-06, - "loss": 0.0869, - "step": 1563 - }, - { - "epoch": 3.19, - "learning_rate": 2.0979445214000103e-06, - "loss": 0.0841, - "step": 1564 - }, - { - "epoch": 3.19, - "learning_rate": 2.087848711704673e-06, - "loss": 0.0766, - "step": 1565 - }, - { - "epoch": 3.19, - "learning_rate": 2.0777744201265935e-06, - "loss": 0.0847, - "step": 1566 - }, - { - "epoch": 3.19, - "learning_rate": 2.0677216740641083e-06, - "loss": 0.078, - "step": 1567 - }, - { - "epoch": 3.19, - "learning_rate": 2.057690500856958e-06, - "loss": 0.0787, - "step": 1568 - }, - { - "epoch": 3.2, - "learning_rate": 2.0476809277862154e-06, - "loss": 0.081, - "step": 1569 - }, - { - "epoch": 3.2, - "learning_rate": 2.037692982074204e-06, - "loss": 0.0733, - "step": 1570 - }, - { - "epoch": 3.2, - "learning_rate": 2.027726690884435e-06, - "loss": 0.082, - "step": 1571 - }, - { - "epoch": 3.2, - "learning_rate": 2.017782081321523e-06, - "loss": 0.0851, - "step": 1572 - }, - { - "epoch": 3.2, - "learning_rate": 2.0078591804311152e-06, - "loss": 0.0801, - "step": 1573 - }, - { - "epoch": 3.21, - "learning_rate": 1.997958015199829e-06, - "loss": 0.0821, - "step": 1574 - }, - { - "epoch": 3.21, - "learning_rate": 1.9880786125551577e-06, - "loss": 0.0813, - "step": 1575 - }, - { - "epoch": 3.21, - "learning_rate": 1.9782209993654155e-06, - "loss": 0.0801, - "step": 1576 - }, - { - "epoch": 3.21, - "learning_rate": 1.96838520243965e-06, - "loss": 0.0886, - "step": 1577 - }, - { - "epoch": 3.21, - "learning_rate": 1.95857124852758e-06, - "loss": 0.0838, - "step": 1578 - }, - { - "epoch": 3.22, - "learning_rate": 1.9487791643195276e-06, - "loss": 0.08, - "step": 1579 - }, - { - "epoch": 3.22, - "learning_rate": 1.939008976446326e-06, - "loss": 0.0803, - "step": 1580 - }, - { - "epoch": 3.22, - "learning_rate": 1.9292607114792637e-06, - "loss": 0.0814, - "step": 1581 - }, - { - "epoch": 3.22, - "learning_rate": 1.9195343959300074e-06, - "loss": 0.0807, - "step": 1582 - }, - { - "epoch": 3.22, - "learning_rate": 1.9098300562505266e-06, - "loss": 0.0879, - "step": 1583 - }, - { - "epoch": 3.23, - "learning_rate": 1.9001477188330263e-06, - "loss": 0.0797, - "step": 1584 - }, - { - "epoch": 3.23, - "learning_rate": 1.8904874100098758e-06, - "loss": 0.076, - "step": 1585 - }, - { - "epoch": 3.23, - "learning_rate": 1.8808491560535292e-06, - "loss": 0.0783, - "step": 1586 - }, - { - "epoch": 3.23, - "learning_rate": 1.8712329831764665e-06, - "loss": 0.0792, - "step": 1587 - }, - { - "epoch": 3.23, - "learning_rate": 1.8616389175311078e-06, - "loss": 0.0806, - "step": 1588 - }, - { - "epoch": 3.24, - "learning_rate": 1.8520669852097573e-06, - "loss": 0.0821, - "step": 1589 - }, - { - "epoch": 3.24, - "learning_rate": 1.8425172122445167e-06, - "loss": 0.0809, - "step": 1590 - }, - { - "epoch": 3.24, - "learning_rate": 1.8329896246072321e-06, - "loss": 0.0842, - "step": 1591 - }, - { - "epoch": 3.24, - "learning_rate": 1.8234842482094094e-06, - "loss": 0.08, - "step": 1592 - }, - { - "epoch": 3.24, - "learning_rate": 1.8140011089021404e-06, - "loss": 0.0781, - "step": 1593 - }, - { - "epoch": 3.25, - "learning_rate": 1.8045402324760519e-06, - "loss": 0.0827, - "step": 1594 - }, - { - "epoch": 3.25, - "learning_rate": 1.7951016446612157e-06, - "loss": 0.0809, - "step": 1595 - }, - { - "epoch": 3.25, - "learning_rate": 1.7856853711270948e-06, - "loss": 0.082, - "step": 1596 - }, - { - "epoch": 3.25, - "learning_rate": 1.7762914374824592e-06, - "loss": 0.0757, - "step": 1597 - }, - { - "epoch": 3.25, - "learning_rate": 1.7669198692753242e-06, - "loss": 0.0766, - "step": 1598 - }, - { - "epoch": 3.26, - "learning_rate": 1.7575706919928814e-06, - "loss": 0.0854, - "step": 1599 - }, - { - "epoch": 3.26, - "learning_rate": 1.7482439310614241e-06, - "loss": 0.0828, - "step": 1600 - }, - { - "epoch": 3.26, - "learning_rate": 1.7389396118462831e-06, - "loss": 0.0787, - "step": 1601 - }, - { - "epoch": 3.26, - "learning_rate": 1.7296577596517572e-06, - "loss": 0.0808, - "step": 1602 - }, - { - "epoch": 3.26, - "learning_rate": 1.7203983997210428e-06, - "loss": 0.0826, - "step": 1603 - }, - { - "epoch": 3.27, - "learning_rate": 1.7111615572361628e-06, - "loss": 0.0772, - "step": 1604 - }, - { - "epoch": 3.27, - "learning_rate": 1.7019472573179041e-06, - "loss": 0.0808, - "step": 1605 - }, - { - "epoch": 3.27, - "learning_rate": 1.6927555250257443e-06, - "loss": 0.0787, - "step": 1606 - }, - { - "epoch": 3.27, - "learning_rate": 1.6835863853577894e-06, - "loss": 0.0783, - "step": 1607 - }, - { - "epoch": 3.27, - "learning_rate": 1.6744398632507009e-06, - "loss": 0.0828, - "step": 1608 - }, - { - "epoch": 3.28, - "learning_rate": 1.6653159835796196e-06, - "loss": 0.0806, - "step": 1609 - }, - { - "epoch": 3.28, - "learning_rate": 1.6562147711581188e-06, - "loss": 0.0832, - "step": 1610 - }, - { - "epoch": 3.28, - "learning_rate": 1.6471362507381183e-06, - "loss": 0.0846, - "step": 1611 - }, - { - "epoch": 3.28, - "learning_rate": 1.638080447009832e-06, - "loss": 0.083, - "step": 1612 - }, - { - "epoch": 3.29, - "learning_rate": 1.629047384601684e-06, - "loss": 0.079, - "step": 1613 - }, - { - "epoch": 3.29, - "learning_rate": 1.6200370880802551e-06, - "loss": 0.0828, - "step": 1614 - }, - { - "epoch": 3.29, - "learning_rate": 1.6110495819502103e-06, - "loss": 0.078, - "step": 1615 - }, - { - "epoch": 3.29, - "learning_rate": 1.602084890654233e-06, - "loss": 0.0794, - "step": 1616 - }, - { - "epoch": 3.29, - "learning_rate": 1.5931430385729584e-06, - "loss": 0.0803, - "step": 1617 - }, - { - "epoch": 3.3, - "learning_rate": 1.5842240500249096e-06, - "loss": 0.0836, - "step": 1618 - }, - { - "epoch": 3.3, - "learning_rate": 1.5753279492664264e-06, - "loss": 0.0831, - "step": 1619 - }, - { - "epoch": 3.3, - "learning_rate": 1.5664547604916048e-06, - "loss": 0.081, - "step": 1620 - }, - { - "epoch": 3.3, - "learning_rate": 1.5576045078322254e-06, - "loss": 0.0846, - "step": 1621 - }, - { - "epoch": 3.3, - "learning_rate": 1.5487772153576962e-06, - "loss": 0.0779, - "step": 1622 - }, - { - "epoch": 3.31, - "learning_rate": 1.539972907074976e-06, - "loss": 0.0851, - "step": 1623 - }, - { - "epoch": 3.31, - "learning_rate": 1.5311916069285238e-06, - "loss": 0.0777, - "step": 1624 - }, - { - "epoch": 3.31, - "learning_rate": 1.5224333388002132e-06, - "loss": 0.0898, - "step": 1625 - }, - { - "epoch": 3.31, - "learning_rate": 1.5136981265092877e-06, - "loss": 0.0792, - "step": 1626 - }, - { - "epoch": 3.31, - "learning_rate": 1.5049859938122856e-06, - "loss": 0.0761, - "step": 1627 - }, - { - "epoch": 3.32, - "learning_rate": 1.4962969644029757e-06, - "loss": 0.0835, - "step": 1628 - }, - { - "epoch": 3.32, - "learning_rate": 1.487631061912298e-06, - "loss": 0.0816, - "step": 1629 - }, - { - "epoch": 3.32, - "learning_rate": 1.4789883099082914e-06, - "loss": 0.0857, - "step": 1630 - }, - { - "epoch": 3.32, - "learning_rate": 1.470368731896038e-06, - "loss": 0.0811, - "step": 1631 - }, - { - "epoch": 3.32, - "learning_rate": 1.4617723513175951e-06, - "loss": 0.0827, - "step": 1632 - }, - { - "epoch": 3.33, - "learning_rate": 1.4531991915519216e-06, - "loss": 0.087, - "step": 1633 - }, - { - "epoch": 3.33, - "learning_rate": 1.4446492759148411e-06, - "loss": 0.0767, - "step": 1634 - }, - { - "epoch": 3.33, - "learning_rate": 1.4361226276589503e-06, - "loss": 0.0797, - "step": 1635 - }, - { - "epoch": 3.33, - "learning_rate": 1.4276192699735702e-06, - "loss": 0.0807, - "step": 1636 - }, - { - "epoch": 3.33, - "learning_rate": 1.4191392259846804e-06, - "loss": 0.0814, - "step": 1637 - }, - { - "epoch": 3.34, - "learning_rate": 1.4106825187548567e-06, - "loss": 0.0824, - "step": 1638 - }, - { - "epoch": 3.34, - "learning_rate": 1.4022491712832032e-06, - "loss": 0.0754, - "step": 1639 - }, - { - "epoch": 3.34, - "learning_rate": 1.393839206505304e-06, - "loss": 0.0815, - "step": 1640 - }, - { - "epoch": 3.34, - "learning_rate": 1.3854526472931384e-06, - "loss": 0.0826, - "step": 1641 - }, - { - "epoch": 3.34, - "learning_rate": 1.377089516455039e-06, - "loss": 0.0817, - "step": 1642 - }, - { - "epoch": 3.35, - "learning_rate": 1.368749836735621e-06, - "loss": 0.0794, - "step": 1643 - }, - { - "epoch": 3.35, - "learning_rate": 1.3604336308157172e-06, - "loss": 0.0819, - "step": 1644 - }, - { - "epoch": 3.35, - "learning_rate": 1.3521409213123282e-06, - "loss": 0.0822, - "step": 1645 - }, - { - "epoch": 3.35, - "learning_rate": 1.3438717307785477e-06, - "loss": 0.076, - "step": 1646 - }, - { - "epoch": 3.35, - "learning_rate": 1.335626081703505e-06, - "loss": 0.0797, - "step": 1647 - }, - { - "epoch": 3.36, - "learning_rate": 1.327403996512312e-06, - "loss": 0.0812, - "step": 1648 - }, - { - "epoch": 3.36, - "learning_rate": 1.319205497565983e-06, - "loss": 0.0846, - "step": 1649 - }, - { - "epoch": 3.36, - "learning_rate": 1.3110306071614021e-06, - "loss": 0.0759, - "step": 1650 - }, - { - "epoch": 3.36, - "learning_rate": 1.3028793475312372e-06, - "loss": 0.0807, - "step": 1651 - }, - { - "epoch": 3.36, - "learning_rate": 1.2947517408438925e-06, - "loss": 0.0863, - "step": 1652 - }, - { - "epoch": 3.37, - "learning_rate": 1.2866478092034418e-06, - "loss": 0.078, - "step": 1653 - }, - { - "epoch": 3.37, - "learning_rate": 1.2785675746495752e-06, - "loss": 0.0779, - "step": 1654 - }, - { - "epoch": 3.37, - "learning_rate": 1.2705110591575343e-06, - "loss": 0.0796, - "step": 1655 - }, - { - "epoch": 3.37, - "learning_rate": 1.2624782846380534e-06, - "loss": 0.0827, - "step": 1656 - }, - { - "epoch": 3.37, - "learning_rate": 1.2544692729373e-06, - "loss": 0.077, - "step": 1657 - }, - { - "epoch": 3.38, - "learning_rate": 1.2464840458368155e-06, - "loss": 0.0807, - "step": 1658 - }, - { - "epoch": 3.38, - "learning_rate": 1.2385226250534566e-06, - "loss": 0.0811, - "step": 1659 - }, - { - "epoch": 3.38, - "learning_rate": 1.2305850322393343e-06, - "loss": 0.079, - "step": 1660 - }, - { - "epoch": 3.38, - "learning_rate": 1.2226712889817571e-06, - "loss": 0.0764, - "step": 1661 - }, - { - "epoch": 3.38, - "learning_rate": 1.2147814168031746e-06, - "loss": 0.0827, - "step": 1662 - }, - { - "epoch": 3.39, - "learning_rate": 1.2069154371611125e-06, - "loss": 0.0808, - "step": 1663 - }, - { - "epoch": 3.39, - "learning_rate": 1.1990733714481185e-06, - "loss": 0.0811, - "step": 1664 - }, - { - "epoch": 3.39, - "learning_rate": 1.1912552409917e-06, - "loss": 0.0871, - "step": 1665 - }, - { - "epoch": 3.39, - "learning_rate": 1.1834610670542722e-06, - "loss": 0.0826, - "step": 1666 - }, - { - "epoch": 3.4, - "learning_rate": 1.1756908708331028e-06, - "loss": 0.075, - "step": 1667 - }, - { - "epoch": 3.4, - "learning_rate": 1.1679446734602396e-06, - "loss": 0.0811, - "step": 1668 - }, - { - "epoch": 3.4, - "learning_rate": 1.1602224960024689e-06, - "loss": 0.0844, - "step": 1669 - }, - { - "epoch": 3.4, - "learning_rate": 1.152524359461249e-06, - "loss": 0.0815, - "step": 1670 - }, - { - "epoch": 3.4, - "learning_rate": 1.1448502847726562e-06, - "loss": 0.0837, - "step": 1671 - }, - { - "epoch": 3.41, - "learning_rate": 1.1372002928073277e-06, - "loss": 0.0843, - "step": 1672 - }, - { - "epoch": 3.41, - "learning_rate": 1.1295744043704048e-06, - "loss": 0.0838, - "step": 1673 - }, - { - "epoch": 3.41, - "learning_rate": 1.1219726402014764e-06, - "loss": 0.0859, - "step": 1674 - }, - { - "epoch": 3.41, - "learning_rate": 1.1143950209745203e-06, - "loss": 0.0794, - "step": 1675 - }, - { - "epoch": 3.41, - "learning_rate": 1.1068415672978516e-06, - "loss": 0.0798, - "step": 1676 - }, - { - "epoch": 3.42, - "learning_rate": 1.0993122997140615e-06, - "loss": 0.0826, - "step": 1677 - }, - { - "epoch": 3.42, - "learning_rate": 1.09180723869997e-06, - "loss": 0.0793, - "step": 1678 - }, - { - "epoch": 3.42, - "learning_rate": 1.0843264046665558e-06, - "loss": 0.0836, - "step": 1679 - }, - { - "epoch": 3.42, - "learning_rate": 1.0768698179589166e-06, - "loss": 0.0878, - "step": 1680 - }, - { - "epoch": 3.42, - "learning_rate": 1.0694374988561995e-06, - "loss": 0.0764, - "step": 1681 - }, - { - "epoch": 3.43, - "learning_rate": 1.0620294675715548e-06, - "loss": 0.084, - "step": 1682 - }, - { - "epoch": 3.43, - "learning_rate": 1.0546457442520852e-06, - "loss": 0.0824, - "step": 1683 - }, - { - "epoch": 3.43, - "learning_rate": 1.0472863489787766e-06, - "loss": 0.0815, - "step": 1684 - }, - { - "epoch": 3.43, - "learning_rate": 1.0399513017664575e-06, - "loss": 0.0774, - "step": 1685 - }, - { - "epoch": 3.43, - "learning_rate": 1.032640622563733e-06, - "loss": 0.0789, - "step": 1686 - }, - { - "epoch": 3.44, - "learning_rate": 1.0253543312529412e-06, - "loss": 0.0869, - "step": 1687 - }, - { - "epoch": 3.44, - "learning_rate": 1.018092447650092e-06, - "loss": 0.085, - "step": 1688 - }, - { - "epoch": 3.44, - "learning_rate": 1.010854991504815e-06, - "loss": 0.0854, - "step": 1689 - }, - { - "epoch": 3.44, - "learning_rate": 1.0036419825003085e-06, - "loss": 0.0859, - "step": 1690 - }, - { - "epoch": 3.44, - "learning_rate": 9.9645344025328e-07, - "loss": 0.0841, - "step": 1691 - }, - { - "epoch": 3.45, - "learning_rate": 9.892893843138996e-07, - "loss": 0.0819, - "step": 1692 - }, - { - "epoch": 3.45, - "learning_rate": 9.821498341657432e-07, - "loss": 0.077, - "step": 1693 - }, - { - "epoch": 3.45, - "learning_rate": 9.750348092257368e-07, - "loss": 0.0839, - "step": 1694 - }, - { - "epoch": 3.45, - "learning_rate": 9.679443288441125e-07, - "loss": 0.0796, - "step": 1695 - }, - { - "epoch": 3.45, - "learning_rate": 9.608784123043502e-07, - "loss": 0.0756, - "step": 1696 - }, - { - "epoch": 3.46, - "learning_rate": 9.538370788231154e-07, - "loss": 0.0827, - "step": 1697 - }, - { - "epoch": 3.46, - "learning_rate": 9.46820347550228e-07, - "loss": 0.0804, - "step": 1698 - }, - { - "epoch": 3.46, - "learning_rate": 9.398282375685919e-07, - "loss": 0.0822, - "step": 1699 - }, - { - "epoch": 3.46, - "learning_rate": 9.328607678941581e-07, - "loss": 0.0749, - "step": 1700 - }, - { - "epoch": 3.46, - "learning_rate": 9.25917957475857e-07, - "loss": 0.0782, - "step": 1701 - }, - { - "epoch": 3.47, - "learning_rate": 9.18999825195559e-07, - "loss": 0.0776, - "step": 1702 - }, - { - "epoch": 3.47, - "learning_rate": 9.121063898680171e-07, - "loss": 0.0785, - "step": 1703 - }, - { - "epoch": 3.47, - "learning_rate": 9.052376702408206e-07, - "loss": 0.0836, - "step": 1704 - }, - { - "epoch": 3.47, - "learning_rate": 8.983936849943397e-07, - "loss": 0.0859, - "step": 1705 - }, - { - "epoch": 3.47, - "learning_rate": 8.915744527416736e-07, - "loss": 0.0829, - "step": 1706 - }, - { - "epoch": 3.48, - "learning_rate": 8.847799920286082e-07, - "loss": 0.0873, - "step": 1707 - }, - { - "epoch": 3.48, - "learning_rate": 8.780103213335556e-07, - "loss": 0.0801, - "step": 1708 - }, - { - "epoch": 3.48, - "learning_rate": 8.712654590675085e-07, - "loss": 0.0816, - "step": 1709 - }, - { - "epoch": 3.48, - "learning_rate": 8.645454235739903e-07, - "loss": 0.0836, - "step": 1710 - }, - { - "epoch": 3.48, - "learning_rate": 8.578502331290095e-07, - "loss": 0.0794, - "step": 1711 - }, - { - "epoch": 3.49, - "learning_rate": 8.511799059409998e-07, - "loss": 0.0842, - "step": 1712 - }, - { - "epoch": 3.49, - "learning_rate": 8.445344601507744e-07, - "loss": 0.0787, - "step": 1713 - }, - { - "epoch": 3.49, - "learning_rate": 8.379139138314829e-07, - "loss": 0.081, - "step": 1714 - }, - { - "epoch": 3.49, - "learning_rate": 8.313182849885548e-07, - "loss": 0.082, - "step": 1715 - }, - { - "epoch": 3.49, - "learning_rate": 8.247475915596581e-07, - "loss": 0.085, - "step": 1716 - }, - { - "epoch": 3.5, - "learning_rate": 8.182018514146406e-07, - "loss": 0.087, - "step": 1717 - }, - { - "epoch": 3.5, - "learning_rate": 8.116810823554877e-07, - "loss": 0.0804, - "step": 1718 - }, - { - "epoch": 3.5, - "learning_rate": 8.051853021162714e-07, - "loss": 0.0858, - "step": 1719 - }, - { - "epoch": 3.5, - "learning_rate": 7.987145283631081e-07, - "loss": 0.0809, - "step": 1720 - }, - { - "epoch": 3.51, - "learning_rate": 7.922687786940997e-07, - "loss": 0.0877, - "step": 1721 - }, - { - "epoch": 3.51, - "learning_rate": 7.858480706392968e-07, - "loss": 0.0804, - "step": 1722 - }, - { - "epoch": 3.51, - "learning_rate": 7.794524216606436e-07, - "loss": 0.0842, - "step": 1723 - }, - { - "epoch": 3.51, - "learning_rate": 7.730818491519343e-07, - "loss": 0.0924, - "step": 1724 - }, - { - "epoch": 3.51, - "learning_rate": 7.66736370438762e-07, - "loss": 0.0818, - "step": 1725 - }, - { - "epoch": 3.52, - "learning_rate": 7.604160027784757e-07, - "loss": 0.0784, - "step": 1726 - }, - { - "epoch": 3.52, - "learning_rate": 7.54120763360131e-07, - "loss": 0.0769, - "step": 1727 - }, - { - "epoch": 3.52, - "learning_rate": 7.478506693044485e-07, - "loss": 0.0823, - "step": 1728 - }, - { - "epoch": 3.52, - "learning_rate": 7.416057376637543e-07, - "loss": 0.0785, - "step": 1729 - }, - { - "epoch": 3.52, - "learning_rate": 7.35385985421948e-07, - "loss": 0.0804, - "step": 1730 - }, - { - "epoch": 3.53, - "learning_rate": 7.291914294944502e-07, - "loss": 0.0824, - "step": 1731 - }, - { - "epoch": 3.53, - "learning_rate": 7.230220867281524e-07, - "loss": 0.0798, - "step": 1732 - }, - { - "epoch": 3.53, - "learning_rate": 7.168779739013821e-07, - "loss": 0.075, - "step": 1733 - }, - { - "epoch": 3.53, - "learning_rate": 7.107591077238474e-07, - "loss": 0.081, - "step": 1734 - }, - { - "epoch": 3.53, - "learning_rate": 7.046655048365947e-07, - "loss": 0.0784, - "step": 1735 - }, - { - "epoch": 3.54, - "learning_rate": 6.985971818119653e-07, - "loss": 0.0856, - "step": 1736 - }, - { - "epoch": 3.54, - "learning_rate": 6.925541551535431e-07, - "loss": 0.0775, - "step": 1737 - }, - { - "epoch": 3.54, - "learning_rate": 6.865364412961229e-07, - "loss": 0.073, - "step": 1738 - }, - { - "epoch": 3.54, - "learning_rate": 6.805440566056554e-07, - "loss": 0.0853, - "step": 1739 - }, - { - "epoch": 3.54, - "learning_rate": 6.745770173792021e-07, - "loss": 0.075, - "step": 1740 - }, - { - "epoch": 3.55, - "learning_rate": 6.686353398448985e-07, - "loss": 0.0772, - "step": 1741 - }, - { - "epoch": 3.55, - "learning_rate": 6.627190401619044e-07, - "loss": 0.0822, - "step": 1742 - }, - { - "epoch": 3.55, - "learning_rate": 6.568281344203576e-07, - "loss": 0.0891, - "step": 1743 - }, - { - "epoch": 3.55, - "learning_rate": 6.509626386413437e-07, - "loss": 0.0795, - "step": 1744 - }, - { - "epoch": 3.55, - "learning_rate": 6.451225687768293e-07, - "loss": 0.0777, - "step": 1745 - }, - { - "epoch": 3.56, - "learning_rate": 6.393079407096414e-07, - "loss": 0.0844, - "step": 1746 - }, - { - "epoch": 3.56, - "learning_rate": 6.335187702534118e-07, - "loss": 0.0823, - "step": 1747 - }, - { - "epoch": 3.56, - "learning_rate": 6.27755073152535e-07, - "loss": 0.0785, - "step": 1748 - }, - { - "epoch": 3.56, - "learning_rate": 6.220168650821334e-07, - "loss": 0.0809, - "step": 1749 - }, - { - "epoch": 3.56, - "learning_rate": 6.163041616480037e-07, - "loss": 0.08, - "step": 1750 - }, - { - "epoch": 3.57, - "learning_rate": 6.106169783865801e-07, - "loss": 0.0788, - "step": 1751 - }, - { - "epoch": 3.57, - "learning_rate": 6.049553307648936e-07, - "loss": 0.0767, - "step": 1752 - }, - { - "epoch": 3.57, - "learning_rate": 5.993192341805221e-07, - "loss": 0.0821, - "step": 1753 - }, - { - "epoch": 3.57, - "learning_rate": 5.937087039615619e-07, - "loss": 0.0778, - "step": 1754 - }, - { - "epoch": 3.57, - "learning_rate": 5.88123755366572e-07, - "loss": 0.0856, - "step": 1755 - }, - { - "epoch": 3.58, - "learning_rate": 5.825644035845435e-07, - "loss": 0.0791, - "step": 1756 - }, - { - "epoch": 3.58, - "learning_rate": 5.77030663734851e-07, - "loss": 0.0838, - "step": 1757 - }, - { - "epoch": 3.58, - "learning_rate": 5.715225508672139e-07, - "loss": 0.0789, - "step": 1758 - }, - { - "epoch": 3.58, - "learning_rate": 5.660400799616572e-07, - "loss": 0.0795, - "step": 1759 - }, - { - "epoch": 3.58, - "learning_rate": 5.605832659284682e-07, - "loss": 0.0798, - "step": 1760 - }, - { - "epoch": 3.59, - "learning_rate": 5.551521236081569e-07, - "loss": 0.0849, - "step": 1761 - }, - { - "epoch": 3.59, - "learning_rate": 5.497466677714181e-07, - "loss": 0.0755, - "step": 1762 - }, - { - "epoch": 3.59, - "learning_rate": 5.443669131190843e-07, - "loss": 0.0797, - "step": 1763 - }, - { - "epoch": 3.59, - "learning_rate": 5.390128742820955e-07, - "loss": 0.0837, - "step": 1764 - }, - { - "epoch": 3.59, - "learning_rate": 5.336845658214496e-07, - "loss": 0.0808, - "step": 1765 - }, - { - "epoch": 3.6, - "learning_rate": 5.283820022281715e-07, - "loss": 0.0797, - "step": 1766 - }, - { - "epoch": 3.6, - "learning_rate": 5.231051979232671e-07, - "loss": 0.0844, - "step": 1767 - }, - { - "epoch": 3.6, - "learning_rate": 5.178541672576898e-07, - "loss": 0.0806, - "step": 1768 - }, - { - "epoch": 3.6, - "learning_rate": 5.126289245122906e-07, - "loss": 0.0858, - "step": 1769 - }, - { - "epoch": 3.6, - "learning_rate": 5.074294838977923e-07, - "loss": 0.0804, - "step": 1770 - }, - { - "epoch": 3.61, - "learning_rate": 5.022558595547489e-07, - "loss": 0.0834, - "step": 1771 - }, - { - "epoch": 3.61, - "learning_rate": 4.971080655534988e-07, - "loss": 0.0816, - "step": 1772 - }, - { - "epoch": 3.61, - "learning_rate": 4.919861158941319e-07, - "loss": 0.0765, - "step": 1773 - }, - { - "epoch": 3.61, - "learning_rate": 4.868900245064511e-07, - "loss": 0.0791, - "step": 1774 - }, - { - "epoch": 3.62, - "learning_rate": 4.818198052499368e-07, - "loss": 0.0832, - "step": 1775 - }, - { - "epoch": 3.62, - "learning_rate": 4.767754719137041e-07, - "loss": 0.0839, - "step": 1776 - }, - { - "epoch": 3.62, - "learning_rate": 4.7175703821646826e-07, - "loss": 0.0754, - "step": 1777 - }, - { - "epoch": 3.62, - "learning_rate": 4.6676451780650947e-07, - "loss": 0.0866, - "step": 1778 - }, - { - "epoch": 3.62, - "learning_rate": 4.6179792426163107e-07, - "loss": 0.0791, - "step": 1779 - }, - { - "epoch": 3.63, - "learning_rate": 4.5685727108912505e-07, - "loss": 0.0866, - "step": 1780 - }, - { - "epoch": 3.63, - "learning_rate": 4.519425717257342e-07, - "loss": 0.0828, - "step": 1781 - }, - { - "epoch": 3.63, - "learning_rate": 4.4705383953762227e-07, - "loss": 0.0849, - "step": 1782 - }, - { - "epoch": 3.63, - "learning_rate": 4.4219108782032504e-07, - "loss": 0.0813, - "step": 1783 - }, - { - "epoch": 3.63, - "learning_rate": 4.3735432979872593e-07, - "loss": 0.0851, - "step": 1784 - }, - { - "epoch": 3.64, - "learning_rate": 4.3254357862701046e-07, - "loss": 0.0753, - "step": 1785 - }, - { - "epoch": 3.64, - "learning_rate": 4.2775884738863736e-07, - "loss": 0.0882, - "step": 1786 - }, - { - "epoch": 3.64, - "learning_rate": 4.230001490963043e-07, - "loss": 0.0824, - "step": 1787 - }, - { - "epoch": 3.64, - "learning_rate": 4.182674966919031e-07, - "loss": 0.0794, - "step": 1788 - }, - { - "epoch": 3.64, - "learning_rate": 4.1356090304649597e-07, - "loss": 0.0808, - "step": 1789 - }, - { - "epoch": 3.65, - "learning_rate": 4.088803809602693e-07, - "loss": 0.0854, - "step": 1790 - }, - { - "epoch": 3.65, - "learning_rate": 4.042259431625084e-07, - "loss": 0.0758, - "step": 1791 - }, - { - "epoch": 3.65, - "learning_rate": 3.995976023115589e-07, - "loss": 0.0815, - "step": 1792 - }, - { - "epoch": 3.65, - "learning_rate": 3.9499537099478955e-07, - "loss": 0.0837, - "step": 1793 - }, - { - "epoch": 3.65, - "learning_rate": 3.9041926172856516e-07, - "loss": 0.0823, - "step": 1794 - }, - { - "epoch": 3.66, - "learning_rate": 3.858692869582037e-07, - "loss": 0.081, - "step": 1795 - }, - { - "epoch": 3.66, - "learning_rate": 3.813454590579502e-07, - "loss": 0.0798, - "step": 1796 - }, - { - "epoch": 3.66, - "learning_rate": 3.7684779033093976e-07, - "loss": 0.0847, - "step": 1797 - }, - { - "epoch": 3.66, - "learning_rate": 3.7237629300916013e-07, - "loss": 0.0818, - "step": 1798 - }, - { - "epoch": 3.66, - "learning_rate": 3.679309792534291e-07, - "loss": 0.0856, - "step": 1799 - }, - { - "epoch": 3.67, - "learning_rate": 3.635118611533517e-07, - "loss": 0.08, - "step": 1800 - }, - { - "epoch": 3.67, - "learning_rate": 3.5911895072728854e-07, - "loss": 0.0771, - "step": 1801 - }, - { - "epoch": 3.67, - "learning_rate": 3.547522599223263e-07, - "loss": 0.0807, - "step": 1802 - }, - { - "epoch": 3.67, - "learning_rate": 3.504118006142454e-07, - "loss": 0.0794, - "step": 1803 - }, - { - "epoch": 3.67, - "learning_rate": 3.4609758460748656e-07, - "loss": 0.0862, - "step": 1804 - }, - { - "epoch": 3.68, - "learning_rate": 3.418096236351165e-07, - "loss": 0.0871, - "step": 1805 - }, - { - "epoch": 3.68, - "learning_rate": 3.37547929358798e-07, - "loss": 0.079, - "step": 1806 - }, - { - "epoch": 3.68, - "learning_rate": 3.333125133687609e-07, - "loss": 0.0845, - "step": 1807 - }, - { - "epoch": 3.68, - "learning_rate": 3.2910338718376455e-07, - "loss": 0.081, - "step": 1808 - }, - { - "epoch": 3.68, - "learning_rate": 3.2492056225107205e-07, - "loss": 0.0908, - "step": 1809 - }, - { - "epoch": 3.69, - "learning_rate": 3.2076404994641484e-07, - "loss": 0.0868, - "step": 1810 - }, - { - "epoch": 3.69, - "learning_rate": 3.166338615739661e-07, - "loss": 0.0853, - "step": 1811 - }, - { - "epoch": 3.69, - "learning_rate": 3.125300083663063e-07, - "loss": 0.0823, - "step": 1812 - }, - { - "epoch": 3.69, - "learning_rate": 3.084525014843931e-07, - "loss": 0.0822, - "step": 1813 - }, - { - "epoch": 3.69, - "learning_rate": 3.044013520175337e-07, - "loss": 0.076, - "step": 1814 - }, - { - "epoch": 3.7, - "learning_rate": 3.0037657098335395e-07, - "loss": 0.0795, - "step": 1815 - }, - { - "epoch": 3.7, - "learning_rate": 2.963781693277645e-07, - "loss": 0.0831, - "step": 1816 - }, - { - "epoch": 3.7, - "learning_rate": 2.9240615792493377e-07, - "loss": 0.091, - "step": 1817 - }, - { - "epoch": 3.7, - "learning_rate": 2.8846054757726174e-07, - "loss": 0.0844, - "step": 1818 - }, - { - "epoch": 3.7, - "learning_rate": 2.845413490153426e-07, - "loss": 0.0901, - "step": 1819 - }, - { - "epoch": 3.71, - "learning_rate": 2.806485728979458e-07, - "loss": 0.0817, - "step": 1820 - }, - { - "epoch": 3.71, - "learning_rate": 2.767822298119771e-07, - "loss": 0.0862, - "step": 1821 - }, - { - "epoch": 3.71, - "learning_rate": 2.7294233027245655e-07, - "loss": 0.0837, - "step": 1822 - }, - { - "epoch": 3.71, - "learning_rate": 2.691288847224871e-07, - "loss": 0.082, - "step": 1823 - }, - { - "epoch": 3.71, - "learning_rate": 2.653419035332261e-07, - "loss": 0.085, - "step": 1824 - }, - { - "epoch": 3.72, - "learning_rate": 2.615813970038583e-07, - "loss": 0.077, - "step": 1825 - }, - { - "epoch": 3.72, - "learning_rate": 2.578473753615651e-07, - "loss": 0.0786, - "step": 1826 - }, - { - "epoch": 3.72, - "learning_rate": 2.541398487615032e-07, - "loss": 0.0786, - "step": 1827 - }, - { - "epoch": 3.72, - "learning_rate": 2.5045882728676805e-07, - "loss": 0.0844, - "step": 1828 - }, - { - "epoch": 3.73, - "learning_rate": 2.4680432094837394e-07, - "loss": 0.0823, - "step": 1829 - }, - { - "epoch": 3.73, - "learning_rate": 2.431763396852227e-07, - "loss": 0.0833, - "step": 1830 - }, - { - "epoch": 3.73, - "learning_rate": 2.3957489336407624e-07, - "loss": 0.0813, - "step": 1831 - }, - { - "epoch": 3.73, - "learning_rate": 2.359999917795375e-07, - "loss": 0.0859, - "step": 1832 - }, - { - "epoch": 3.73, - "learning_rate": 2.3245164465400817e-07, - "loss": 0.0816, - "step": 1833 - }, - { - "epoch": 3.74, - "learning_rate": 2.2892986163767782e-07, - "loss": 0.0798, - "step": 1834 - }, - { - "epoch": 3.74, - "learning_rate": 2.2543465230849158e-07, - "loss": 0.084, - "step": 1835 - }, - { - "epoch": 3.74, - "learning_rate": 2.2196602617212014e-07, - "loss": 0.0846, - "step": 1836 - }, - { - "epoch": 3.74, - "learning_rate": 2.1852399266194312e-07, - "loss": 0.0843, - "step": 1837 - }, - { - "epoch": 3.74, - "learning_rate": 2.1510856113901467e-07, - "loss": 0.0827, - "step": 1838 - }, - { - "epoch": 3.75, - "learning_rate": 2.1171974089204018e-07, - "loss": 0.0838, - "step": 1839 - }, - { - "epoch": 3.75, - "learning_rate": 2.0835754113735618e-07, - "loss": 0.0781, - "step": 1840 - }, - { - "epoch": 3.75, - "learning_rate": 2.0502197101889608e-07, - "loss": 0.0754, - "step": 1841 - }, - { - "epoch": 3.75, - "learning_rate": 2.0171303960817567e-07, - "loss": 0.0849, - "step": 1842 - }, - { - "epoch": 3.75, - "learning_rate": 1.9843075590426087e-07, - "loss": 0.0843, - "step": 1843 - }, - { - "epoch": 3.76, - "learning_rate": 1.9517512883374667e-07, - "loss": 0.0862, - "step": 1844 - }, - { - "epoch": 3.76, - "learning_rate": 1.9194616725073167e-07, - "loss": 0.0769, - "step": 1845 - }, - { - "epoch": 3.76, - "learning_rate": 1.8874387993679244e-07, - "loss": 0.0816, - "step": 1846 - }, - { - "epoch": 3.76, - "learning_rate": 1.8556827560096247e-07, - "loss": 0.0829, - "step": 1847 - }, - { - "epoch": 3.76, - "learning_rate": 1.8241936287970997e-07, - "loss": 0.08, - "step": 1848 - }, - { - "epoch": 3.77, - "learning_rate": 1.7929715033690898e-07, - "loss": 0.0826, - "step": 1849 - }, - { - "epoch": 3.77, - "learning_rate": 1.7620164646381943e-07, - "loss": 0.0814, - "step": 1850 - }, - { - "epoch": 3.77, - "learning_rate": 1.7313285967906378e-07, - "loss": 0.0804, - "step": 1851 - }, - { - "epoch": 3.77, - "learning_rate": 1.7009079832860375e-07, - "loss": 0.083, - "step": 1852 - }, - { - "epoch": 3.77, - "learning_rate": 1.6707547068571805e-07, - "loss": 0.0792, - "step": 1853 - }, - { - "epoch": 3.78, - "learning_rate": 1.6408688495098134e-07, - "loss": 0.0794, - "step": 1854 - }, - { - "epoch": 3.78, - "learning_rate": 1.6112504925223649e-07, - "loss": 0.0825, - "step": 1855 - }, - { - "epoch": 3.78, - "learning_rate": 1.5818997164458005e-07, - "loss": 0.0809, - "step": 1856 - }, - { - "epoch": 3.78, - "learning_rate": 1.552816601103324e-07, - "loss": 0.08, - "step": 1857 - }, - { - "epoch": 3.78, - "learning_rate": 1.5240012255902325e-07, - "loss": 0.0801, - "step": 1858 - }, - { - "epoch": 3.79, - "learning_rate": 1.495453668273672e-07, - "loss": 0.0809, - "step": 1859 - }, - { - "epoch": 3.79, - "learning_rate": 1.4671740067923934e-07, - "loss": 0.0787, - "step": 1860 - }, - { - "epoch": 3.79, - "learning_rate": 1.4391623180566083e-07, - "loss": 0.0834, - "step": 1861 - }, - { - "epoch": 3.79, - "learning_rate": 1.4114186782477e-07, - "loss": 0.0836, - "step": 1862 - }, - { - "epoch": 3.79, - "learning_rate": 1.3839431628181021e-07, - "loss": 0.084, - "step": 1863 - }, - { - "epoch": 3.8, - "learning_rate": 1.3567358464910086e-07, - "loss": 0.0762, - "step": 1864 - }, - { - "epoch": 3.8, - "learning_rate": 1.3297968032602414e-07, - "loss": 0.0833, - "step": 1865 - }, - { - "epoch": 3.8, - "learning_rate": 1.303126106389996e-07, - "loss": 0.0851, - "step": 1866 - }, - { - "epoch": 3.8, - "learning_rate": 1.2767238284146944e-07, - "loss": 0.0875, - "step": 1867 - }, - { - "epoch": 3.8, - "learning_rate": 1.2505900411387328e-07, - "loss": 0.0815, - "step": 1868 - }, - { - "epoch": 3.81, - "learning_rate": 1.2247248156363024e-07, - "loss": 0.0803, - "step": 1869 - }, - { - "epoch": 3.81, - "learning_rate": 1.1991282222512557e-07, - "loss": 0.0868, - "step": 1870 - }, - { - "epoch": 3.81, - "learning_rate": 1.1738003305968193e-07, - "loss": 0.081, - "step": 1871 - }, - { - "epoch": 3.81, - "learning_rate": 1.1487412095554706e-07, - "loss": 0.0809, - "step": 1872 - }, - { - "epoch": 3.81, - "learning_rate": 1.1239509272787164e-07, - "loss": 0.081, - "step": 1873 - }, - { - "epoch": 3.82, - "learning_rate": 1.0994295511869257e-07, - "loss": 0.0858, - "step": 1874 - }, - { - "epoch": 3.82, - "learning_rate": 1.0751771479691753e-07, - "loss": 0.0834, - "step": 1875 - }, - { - "epoch": 3.82, - "learning_rate": 1.0511937835829822e-07, - "loss": 0.0839, - "step": 1876 - }, - { - "epoch": 3.82, - "learning_rate": 1.0274795232542045e-07, - "loss": 0.0809, - "step": 1877 - }, - { - "epoch": 3.82, - "learning_rate": 1.0040344314768524e-07, - "loss": 0.0821, - "step": 1878 - }, - { - "epoch": 3.83, - "learning_rate": 9.80858572012866e-08, - "loss": 0.0786, - "step": 1879 - }, - { - "epoch": 3.83, - "learning_rate": 9.579520078919935e-08, - "loss": 0.0855, - "step": 1880 - }, - { - "epoch": 3.83, - "learning_rate": 9.35314801411602e-08, - "loss": 0.0804, - "step": 1881 - }, - { - "epoch": 3.83, - "learning_rate": 9.129470141364894e-08, - "loss": 0.0755, - "step": 1882 - }, - { - "epoch": 3.84, - "learning_rate": 8.908487068987281e-08, - "loss": 0.082, - "step": 1883 - }, - { - "epoch": 3.84, - "learning_rate": 8.69019939797544e-08, - "loss": 0.0776, - "step": 1884 - }, - { - "epoch": 3.84, - "learning_rate": 8.474607721990713e-08, - "loss": 0.0787, - "step": 1885 - }, - { - "epoch": 3.84, - "learning_rate": 8.26171262736264e-08, - "loss": 0.0837, - "step": 1886 - }, - { - "epoch": 3.84, - "learning_rate": 8.051514693086849e-08, - "loss": 0.076, - "step": 1887 - }, - { - "epoch": 3.85, - "learning_rate": 7.844014490823837e-08, - "loss": 0.0824, - "step": 1888 - }, - { - "epoch": 3.85, - "learning_rate": 7.639212584897082e-08, - "loss": 0.0856, - "step": 1889 - }, - { - "epoch": 3.85, - "learning_rate": 7.437109532291708e-08, - "loss": 0.0774, - "step": 1890 - }, - { - "epoch": 3.85, - "learning_rate": 7.237705882653378e-08, - "loss": 0.0771, - "step": 1891 - }, - { - "epoch": 3.85, - "learning_rate": 7.04100217828596e-08, - "loss": 0.08, - "step": 1892 - }, - { - "epoch": 3.86, - "learning_rate": 6.84699895415053e-08, - "loss": 0.0792, - "step": 1893 - }, - { - "epoch": 3.86, - "learning_rate": 6.655696737863814e-08, - "loss": 0.087, - "step": 1894 - }, - { - "epoch": 3.86, - "learning_rate": 6.467096049697196e-08, - "loss": 0.0776, - "step": 1895 - }, - { - "epoch": 3.86, - "learning_rate": 6.281197402574269e-08, - "loss": 0.0802, - "step": 1896 - }, - { - "epoch": 3.86, - "learning_rate": 6.09800130207061e-08, - "loss": 0.0841, - "step": 1897 - }, - { - "epoch": 3.87, - "learning_rate": 5.917508246411574e-08, - "loss": 0.0902, - "step": 1898 - }, - { - "epoch": 3.87, - "learning_rate": 5.739718726471499e-08, - "loss": 0.0831, - "step": 1899 - }, - { - "epoch": 3.87, - "learning_rate": 5.5646332257717206e-08, - "loss": 0.0796, - "step": 1900 - }, - { - "epoch": 3.87, - "learning_rate": 5.39225222048001e-08, - "loss": 0.0794, - "step": 1901 - }, - { - "epoch": 3.87, - "learning_rate": 5.222576179408689e-08, - "loss": 0.0824, - "step": 1902 - }, - { - "epoch": 3.88, - "learning_rate": 5.0556055640138546e-08, - "loss": 0.0789, - "step": 1903 - }, - { - "epoch": 3.88, - "learning_rate": 4.8913408283934874e-08, - "loss": 0.0758, - "step": 1904 - }, - { - "epoch": 3.88, - "learning_rate": 4.729782419286677e-08, - "loss": 0.0798, - "step": 1905 - }, - { - "epoch": 3.88, - "learning_rate": 4.5709307760726243e-08, - "loss": 0.0819, - "step": 1906 - }, - { - "epoch": 3.88, - "learning_rate": 4.414786330768639e-08, - "loss": 0.0859, - "step": 1907 - }, - { - "epoch": 3.89, - "learning_rate": 4.2613495080299215e-08, - "loss": 0.0838, - "step": 1908 - }, - { - "epoch": 3.89, - "learning_rate": 4.1106207251475626e-08, - "loss": 0.0866, - "step": 1909 - }, - { - "epoch": 3.89, - "learning_rate": 3.962600392047988e-08, - "loss": 0.0876, - "step": 1910 - }, - { - "epoch": 3.89, - "learning_rate": 3.817288911291739e-08, - "loss": 0.0817, - "step": 1911 - }, - { - "epoch": 3.89, - "learning_rate": 3.674686678072026e-08, - "loss": 0.0825, - "step": 1912 - }, - { - "epoch": 3.9, - "learning_rate": 3.534794080214066e-08, - "loss": 0.083, - "step": 1913 - }, - { - "epoch": 3.9, - "learning_rate": 3.397611498173747e-08, - "loss": 0.0812, - "step": 1914 - }, - { - "epoch": 3.9, - "learning_rate": 3.263139305037077e-08, - "loss": 0.0765, - "step": 1915 - }, - { - "epoch": 3.9, - "learning_rate": 3.131377866518293e-08, - "loss": 0.0829, - "step": 1916 - }, - { - "epoch": 3.9, - "learning_rate": 3.002327540959749e-08, - "loss": 0.079, - "step": 1917 - }, - { - "epoch": 3.91, - "learning_rate": 2.875988679330477e-08, - "loss": 0.0836, - "step": 1918 - }, - { - "epoch": 3.91, - "learning_rate": 2.7523616252252972e-08, - "loss": 0.0835, - "step": 1919 - }, - { - "epoch": 3.91, - "learning_rate": 2.631446714864039e-08, - "loss": 0.0784, - "step": 1920 - }, - { - "epoch": 3.91, - "learning_rate": 2.513244277090321e-08, - "loss": 0.0822, - "step": 1921 - }, - { - "epoch": 3.91, - "learning_rate": 2.3977546333709966e-08, - "loss": 0.0825, - "step": 1922 - }, - { - "epoch": 3.92, - "learning_rate": 2.284978097794932e-08, - "loss": 0.0754, - "step": 1923 - }, - { - "epoch": 3.92, - "learning_rate": 2.1749149770725618e-08, - "loss": 0.087, - "step": 1924 - }, - { - "epoch": 3.92, - "learning_rate": 2.0675655705348908e-08, - "loss": 0.0829, - "step": 1925 - }, - { - "epoch": 3.92, - "learning_rate": 1.9629301701323822e-08, - "loss": 0.0852, - "step": 1926 - }, - { - "epoch": 3.92, - "learning_rate": 1.8610090604344043e-08, - "loss": 0.081, - "step": 1927 - }, - { - "epoch": 3.93, - "learning_rate": 1.7618025186287858e-08, - "loss": 0.0807, - "step": 1928 - }, - { - "epoch": 3.93, - "learning_rate": 1.665310814520482e-08, - "loss": 0.0836, - "step": 1929 - }, - { - "epoch": 3.93, - "learning_rate": 1.571534210531023e-08, - "loss": 0.0809, - "step": 1930 - }, - { - "epoch": 3.93, - "learning_rate": 1.480472961698065e-08, - "loss": 0.0877, - "step": 1931 - }, - { - "epoch": 3.93, - "learning_rate": 1.3921273156745074e-08, - "loss": 0.0786, - "step": 1932 - }, - { - "epoch": 3.94, - "learning_rate": 1.3064975127276003e-08, - "loss": 0.0816, - "step": 1933 - }, - { - "epoch": 3.94, - "learning_rate": 1.2235837857387246e-08, - "loss": 0.0775, - "step": 1934 - }, - { - "epoch": 3.94, - "learning_rate": 1.143386360202503e-08, - "loss": 0.0863, - "step": 1935 - }, - { - "epoch": 3.94, - "learning_rate": 1.065905454226135e-08, - "loss": 0.0856, - "step": 1936 - }, - { - "epoch": 3.95, - "learning_rate": 9.911412785290619e-09, - "loss": 0.0828, - "step": 1937 - }, - { - "epoch": 3.95, - "learning_rate": 9.190940364419698e-09, - "loss": 0.084, - "step": 1938 - }, - { - "epoch": 3.95, - "learning_rate": 8.497639239066768e-09, - "loss": 0.0815, - "step": 1939 - }, - { - "epoch": 3.95, - "learning_rate": 7.831511294753569e-09, - "loss": 0.0883, - "step": 1940 - }, - { - "epoch": 3.95, - "learning_rate": 7.192558343102063e-09, - "loss": 0.0831, - "step": 1941 - }, - { - "epoch": 3.96, - "learning_rate": 6.580782121826668e-09, - "loss": 0.0814, - "step": 1942 - }, - { - "epoch": 3.96, - "learning_rate": 5.996184294732033e-09, - "loss": 0.086, - "step": 1943 - }, - { - "epoch": 3.96, - "learning_rate": 5.438766451707489e-09, - "loss": 0.0838, - "step": 1944 - }, - { - "epoch": 3.96, - "learning_rate": 4.908530108722609e-09, - "loss": 0.084, - "step": 1945 - }, - { - "epoch": 3.96, - "learning_rate": 4.405476707822765e-09, - "loss": 0.0811, - "step": 1946 - }, - { - "epoch": 3.97, - "learning_rate": 3.929607617129127e-09, - "loss": 0.0803, - "step": 1947 - }, - { - "epoch": 3.97, - "learning_rate": 3.4809241308264572e-09, - "loss": 0.0838, - "step": 1948 - }, - { - "epoch": 3.97, - "learning_rate": 3.0594274691686522e-09, - "loss": 0.0882, - "step": 1949 - }, - { - "epoch": 3.97, - "learning_rate": 2.6651187784709764e-09, - "loss": 0.0799, - "step": 1950 - }, - { - "epoch": 3.97, - "learning_rate": 2.2979991311045113e-09, - "loss": 0.0792, - "step": 1951 - }, - { - "epoch": 3.98, - "learning_rate": 1.9580695255017048e-09, - "loss": 0.0863, - "step": 1952 - }, - { - "epoch": 3.98, - "learning_rate": 1.6453308861430483e-09, - "loss": 0.0794, - "step": 1953 - }, - { - "epoch": 3.98, - "learning_rate": 1.3597840635615201e-09, - "loss": 0.0803, - "step": 1954 - }, - { - "epoch": 3.98, - "learning_rate": 1.1014298343403617e-09, - "loss": 0.0839, - "step": 1955 - }, - { - "epoch": 3.98, - "learning_rate": 8.702689011041987e-10, - "loss": 0.08, - "step": 1956 - }, - { - "epoch": 3.99, - "learning_rate": 6.663018925279207e-10, - "loss": 0.0813, - "step": 1957 - }, - { - "epoch": 3.99, - "learning_rate": 4.895293633255804e-10, - "loss": 0.0774, - "step": 1958 - }, - { - "epoch": 3.99, - "learning_rate": 3.399517942515029e-10, - "loss": 0.0836, - "step": 1959 - }, - { - "epoch": 3.99, - "learning_rate": 2.1756959210250673e-10, - "loss": 0.0879, - "step": 1960 - }, - { - "epoch": 3.99, - "learning_rate": 1.2238308971235234e-10, - "loss": 0.0876, - "step": 1961 - }, - { - "epoch": 4.0, - "learning_rate": 5.4392545952852704e-11, - "loss": 0.0833, - "step": 1962 - }, - { - "epoch": 4.0, - "learning_rate": 1.359814573387297e-11, - "loss": 0.0784, - "step": 1963 - }, - { - "epoch": 4.0, - "learning_rate": 0.0, - "loss": 0.0745, - "step": 1964 - }, - { - "epoch": 4.0, - "step": 1964, - "total_flos": 207599449374720.0, - "train_loss": 0.43551522325686426, - "train_runtime": 13456.2156, - "train_samples_per_second": 14.004, - "train_steps_per_second": 0.146 - } - ], - "logging_steps": 1.0, - "max_steps": 1964, - "num_input_tokens_seen": 0, - "num_train_epochs": 4, - "save_steps": 50000, - "total_flos": 207599449374720.0, - "train_batch_size": 24, - "trial_name": null, - "trial_params": null -}