diff --git "a/checkpoint-2976480/trainer_state.json" "b/checkpoint-2976480/trainer_state.json" new file mode 100644--- /dev/null +++ "b/checkpoint-2976480/trainer_state.json" @@ -0,0 +1,35215 @@ +{ + "best_metric": 3.847501277923584, + "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/passive/transformer/3/checkpoints/checkpoint-992160", + "epoch": 1.0250006060157382, + "eval_steps": 10, + "global_step": 2976480, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 4.999998362119627e-05, + "loss": 10.9682, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 4.999161405248948e-05, + "loss": 6.8285, + "step": 512 + }, + { + "epoch": 0.0, + "learning_rate": 4.998322810497896e-05, + "loss": 6.1802, + "step": 1024 + }, + { + "epoch": 0.0, + "learning_rate": 4.997484215746844e-05, + "loss": 5.9721, + "step": 1536 + }, + { + "epoch": 0.0, + "learning_rate": 4.996645620995792e-05, + "loss": 5.8073, + "step": 2048 + }, + { + "epoch": 0.0, + "learning_rate": 4.99580702624474e-05, + "loss": 5.6887, + "step": 2560 + }, + { + "epoch": 0.0, + "learning_rate": 4.994968431493688e-05, + "loss": 5.602, + "step": 3072 + }, + { + "epoch": 0.0, + "learning_rate": 4.994129836742636e-05, + "loss": 5.5256, + "step": 3584 + }, + { + "epoch": 0.0, + "learning_rate": 4.993291241991584e-05, + "loss": 5.444, + "step": 4096 + }, + { + "epoch": 0.0, + "learning_rate": 4.992452647240532e-05, + "loss": 5.3966, + "step": 4608 + }, + { + "epoch": 0.0, + "learning_rate": 4.99161405248948e-05, + "loss": 5.3464, + "step": 5120 + }, + { + "epoch": 0.0, + "learning_rate": 4.990775457738428e-05, + "loss": 5.3064, + "step": 5632 + }, + { + "epoch": 0.0, + "learning_rate": 4.989936862987376e-05, + "loss": 5.2596, + "step": 6144 + }, + { + "epoch": 0.0, + "learning_rate": 4.989099906116697e-05, + "loss": 5.211, + "step": 6656 + }, + { + "epoch": 0.0, + "learning_rate": 4.988261311365645e-05, + "loss": 5.1732, + "step": 7168 + }, + { + "epoch": 0.0, + "learning_rate": 4.987422716614593e-05, + "loss": 5.1263, + "step": 7680 + }, + { + "epoch": 0.0, + "learning_rate": 4.986584121863541e-05, + "loss": 5.1123, + "step": 8192 + }, + { + "epoch": 0.0, + "learning_rate": 4.985745527112489e-05, + "loss": 5.0786, + "step": 8704 + }, + { + "epoch": 0.0, + "learning_rate": 4.984906932361437e-05, + "loss": 5.0487, + "step": 9216 + }, + { + "epoch": 0.0, + "learning_rate": 4.984068337610385e-05, + "loss": 5.0205, + "step": 9728 + }, + { + "epoch": 0.0, + "learning_rate": 4.983229742859333e-05, + "loss": 5.0124, + "step": 10240 + }, + { + "epoch": 0.0, + "learning_rate": 4.9823927859886547e-05, + "loss": 4.9787, + "step": 10752 + }, + { + "epoch": 0.0, + "learning_rate": 4.9815541912376026e-05, + "loss": 4.9548, + "step": 11264 + }, + { + "epoch": 0.0, + "learning_rate": 4.9807155964865506e-05, + "loss": 4.9274, + "step": 11776 + }, + { + "epoch": 0.0, + "learning_rate": 4.9798770017354986e-05, + "loss": 4.9246, + "step": 12288 + }, + { + "epoch": 0.0, + "learning_rate": 4.9790400448648195e-05, + "loss": 4.8876, + "step": 12800 + }, + { + "epoch": 0.0, + "learning_rate": 4.9782014501137675e-05, + "loss": 4.8755, + "step": 13312 + }, + { + "epoch": 0.0, + "learning_rate": 4.9773628553627155e-05, + "loss": 4.8465, + "step": 13824 + }, + { + "epoch": 0.0, + "learning_rate": 4.9765242606116635e-05, + "loss": 4.837, + "step": 14336 + }, + { + "epoch": 0.0, + "learning_rate": 4.9756856658606115e-05, + "loss": 4.8225, + "step": 14848 + }, + { + "epoch": 0.01, + "learning_rate": 4.9748487089899324e-05, + "loss": 4.8048, + "step": 15360 + }, + { + "epoch": 0.01, + "learning_rate": 4.9740101142388804e-05, + "loss": 4.7969, + "step": 15872 + }, + { + "epoch": 0.01, + "learning_rate": 4.9731715194878284e-05, + "loss": 4.7842, + "step": 16384 + }, + { + "epoch": 0.01, + "learning_rate": 4.97233456261715e-05, + "loss": 4.7668, + "step": 16896 + }, + { + "epoch": 0.01, + "learning_rate": 4.971495967866098e-05, + "loss": 4.7669, + "step": 17408 + }, + { + "epoch": 0.01, + "learning_rate": 4.970657373115046e-05, + "loss": 4.7453, + "step": 17920 + }, + { + "epoch": 0.01, + "learning_rate": 4.969818778363994e-05, + "loss": 4.7414, + "step": 18432 + }, + { + "epoch": 0.01, + "learning_rate": 4.968981821493315e-05, + "loss": 4.7015, + "step": 18944 + }, + { + "epoch": 0.01, + "learning_rate": 4.968143226742263e-05, + "loss": 4.7119, + "step": 19456 + }, + { + "epoch": 0.01, + "learning_rate": 4.967304631991211e-05, + "loss": 4.6761, + "step": 19968 + }, + { + "epoch": 0.01, + "learning_rate": 4.966466037240159e-05, + "loss": 4.6873, + "step": 20480 + }, + { + "epoch": 0.01, + "learning_rate": 4.965627442489107e-05, + "loss": 4.6693, + "step": 20992 + }, + { + "epoch": 0.01, + "learning_rate": 4.964788847738054e-05, + "loss": 4.6668, + "step": 21504 + }, + { + "epoch": 0.01, + "learning_rate": 4.963950252987002e-05, + "loss": 4.6434, + "step": 22016 + }, + { + "epoch": 0.01, + "learning_rate": 4.96311165823595e-05, + "loss": 4.6552, + "step": 22528 + }, + { + "epoch": 0.01, + "learning_rate": 4.962274701365272e-05, + "loss": 4.6406, + "step": 23040 + }, + { + "epoch": 0.01, + "learning_rate": 4.96143610661422e-05, + "loss": 4.635, + "step": 23552 + }, + { + "epoch": 0.01, + "learning_rate": 4.960597511863168e-05, + "loss": 4.6289, + "step": 24064 + }, + { + "epoch": 0.01, + "learning_rate": 4.959758917112116e-05, + "loss": 4.5955, + "step": 24576 + }, + { + "epoch": 0.01, + "learning_rate": 4.9589219602414374e-05, + "loss": 4.6008, + "step": 25088 + }, + { + "epoch": 0.01, + "learning_rate": 4.958083365490385e-05, + "loss": 4.601, + "step": 25600 + }, + { + "epoch": 0.01, + "learning_rate": 4.957244770739333e-05, + "loss": 4.5904, + "step": 26112 + }, + { + "epoch": 0.01, + "learning_rate": 4.956406175988281e-05, + "loss": 4.5836, + "step": 26624 + }, + { + "epoch": 0.01, + "learning_rate": 4.9555692191176016e-05, + "loss": 4.5643, + "step": 27136 + }, + { + "epoch": 0.01, + "learning_rate": 4.9547306243665496e-05, + "loss": 4.5609, + "step": 27648 + }, + { + "epoch": 0.01, + "learning_rate": 4.9538920296154976e-05, + "loss": 4.5427, + "step": 28160 + }, + { + "epoch": 0.01, + "learning_rate": 4.9530534348644456e-05, + "loss": 4.5666, + "step": 28672 + }, + { + "epoch": 0.01, + "learning_rate": 4.9522148401133936e-05, + "loss": 4.5208, + "step": 29184 + }, + { + "epoch": 0.01, + "learning_rate": 4.951377883242715e-05, + "loss": 4.5443, + "step": 29696 + }, + { + "epoch": 0.01, + "learning_rate": 4.950539288491663e-05, + "loss": 4.5361, + "step": 30208 + }, + { + "epoch": 0.01, + "learning_rate": 4.949700693740611e-05, + "loss": 4.515, + "step": 30720 + }, + { + "epoch": 0.01, + "learning_rate": 4.948862098989559e-05, + "loss": 4.5151, + "step": 31232 + }, + { + "epoch": 0.01, + "learning_rate": 4.94802514211888e-05, + "loss": 4.5053, + "step": 31744 + }, + { + "epoch": 0.01, + "learning_rate": 4.947186547367828e-05, + "loss": 4.4986, + "step": 32256 + }, + { + "epoch": 0.01, + "learning_rate": 4.946347952616776e-05, + "loss": 4.4909, + "step": 32768 + }, + { + "epoch": 0.01, + "learning_rate": 4.945509357865724e-05, + "loss": 4.502, + "step": 33280 + }, + { + "epoch": 0.01, + "learning_rate": 4.944672400995045e-05, + "loss": 4.4797, + "step": 33792 + }, + { + "epoch": 0.01, + "learning_rate": 4.943833806243993e-05, + "loss": 4.4753, + "step": 34304 + }, + { + "epoch": 0.01, + "learning_rate": 4.942995211492941e-05, + "loss": 4.4667, + "step": 34816 + }, + { + "epoch": 0.01, + "learning_rate": 4.9421582546222625e-05, + "loss": 4.47, + "step": 35328 + }, + { + "epoch": 0.01, + "learning_rate": 4.9413196598712105e-05, + "loss": 4.4777, + "step": 35840 + }, + { + "epoch": 0.01, + "learning_rate": 4.9404810651201585e-05, + "loss": 4.4724, + "step": 36352 + }, + { + "epoch": 0.01, + "learning_rate": 4.9396424703691065e-05, + "loss": 4.4491, + "step": 36864 + }, + { + "epoch": 0.01, + "learning_rate": 4.9388038756180545e-05, + "loss": 4.463, + "step": 37376 + }, + { + "epoch": 0.01, + "learning_rate": 4.9379652808670025e-05, + "loss": 4.4631, + "step": 37888 + }, + { + "epoch": 0.01, + "learning_rate": 4.9371266861159505e-05, + "loss": 4.4346, + "step": 38400 + }, + { + "epoch": 0.01, + "learning_rate": 4.9362897292452714e-05, + "loss": 4.4424, + "step": 38912 + }, + { + "epoch": 0.01, + "learning_rate": 4.9354511344942194e-05, + "loss": 4.4329, + "step": 39424 + }, + { + "epoch": 0.01, + "learning_rate": 4.9346125397431674e-05, + "loss": 4.4289, + "step": 39936 + }, + { + "epoch": 0.01, + "learning_rate": 4.9337739449921154e-05, + "loss": 4.4189, + "step": 40448 + }, + { + "epoch": 0.01, + "learning_rate": 4.9329353502410634e-05, + "loss": 4.4222, + "step": 40960 + }, + { + "epoch": 0.01, + "learning_rate": 4.9320967554900114e-05, + "loss": 4.4249, + "step": 41472 + }, + { + "epoch": 0.01, + "learning_rate": 4.9312581607389594e-05, + "loss": 4.4282, + "step": 41984 + }, + { + "epoch": 0.01, + "learning_rate": 4.930421203868281e-05, + "loss": 4.4054, + "step": 42496 + }, + { + "epoch": 0.01, + "learning_rate": 4.929582609117229e-05, + "loss": 4.3833, + "step": 43008 + }, + { + "epoch": 0.01, + "learning_rate": 4.928744014366177e-05, + "loss": 4.3976, + "step": 43520 + }, + { + "epoch": 0.01, + "learning_rate": 4.927905419615125e-05, + "loss": 4.3984, + "step": 44032 + }, + { + "epoch": 0.01, + "learning_rate": 4.927066824864073e-05, + "loss": 4.3971, + "step": 44544 + }, + { + "epoch": 0.01, + "learning_rate": 4.926229867993394e-05, + "loss": 4.385, + "step": 45056 + }, + { + "epoch": 0.01, + "learning_rate": 4.925392911122715e-05, + "loss": 4.3881, + "step": 45568 + }, + { + "epoch": 0.02, + "learning_rate": 4.924554316371663e-05, + "loss": 4.384, + "step": 46080 + }, + { + "epoch": 0.02, + "learning_rate": 4.923715721620611e-05, + "loss": 4.3802, + "step": 46592 + }, + { + "epoch": 0.02, + "learning_rate": 4.922878764749932e-05, + "loss": 4.3679, + "step": 47104 + }, + { + "epoch": 0.02, + "learning_rate": 4.92204016999888e-05, + "loss": 4.3737, + "step": 47616 + }, + { + "epoch": 0.02, + "learning_rate": 4.921201575247828e-05, + "loss": 4.3608, + "step": 48128 + }, + { + "epoch": 0.02, + "learning_rate": 4.920362980496776e-05, + "loss": 4.3748, + "step": 48640 + }, + { + "epoch": 0.02, + "learning_rate": 4.919524385745724e-05, + "loss": 4.3612, + "step": 49152 + }, + { + "epoch": 0.02, + "learning_rate": 4.918685790994672e-05, + "loss": 4.3415, + "step": 49664 + }, + { + "epoch": 0.02, + "learning_rate": 4.91784719624362e-05, + "loss": 4.3547, + "step": 50176 + }, + { + "epoch": 0.02, + "learning_rate": 4.9170086014925676e-05, + "loss": 4.3551, + "step": 50688 + }, + { + "epoch": 0.02, + "learning_rate": 4.9161700067415156e-05, + "loss": 4.3584, + "step": 51200 + }, + { + "epoch": 0.02, + "learning_rate": 4.9153314119904636e-05, + "loss": 4.3424, + "step": 51712 + }, + { + "epoch": 0.02, + "learning_rate": 4.914494455119785e-05, + "loss": 4.3422, + "step": 52224 + }, + { + "epoch": 0.02, + "learning_rate": 4.9136558603687325e-05, + "loss": 4.3276, + "step": 52736 + }, + { + "epoch": 0.02, + "learning_rate": 4.9128172656176805e-05, + "loss": 4.3364, + "step": 53248 + }, + { + "epoch": 0.02, + "learning_rate": 4.9119786708666285e-05, + "loss": 4.3174, + "step": 53760 + }, + { + "epoch": 0.02, + "learning_rate": 4.91114171399595e-05, + "loss": 4.327, + "step": 54272 + }, + { + "epoch": 0.02, + "learning_rate": 4.910303119244898e-05, + "loss": 4.3231, + "step": 54784 + }, + { + "epoch": 0.02, + "learning_rate": 4.909464524493846e-05, + "loss": 4.3223, + "step": 55296 + }, + { + "epoch": 0.02, + "learning_rate": 4.908625929742794e-05, + "loss": 4.3139, + "step": 55808 + }, + { + "epoch": 0.02, + "learning_rate": 4.907787334991742e-05, + "loss": 4.3161, + "step": 56320 + }, + { + "epoch": 0.02, + "learning_rate": 4.90694874024069e-05, + "loss": 4.3074, + "step": 56832 + }, + { + "epoch": 0.02, + "learning_rate": 4.906110145489638e-05, + "loss": 4.3155, + "step": 57344 + }, + { + "epoch": 0.02, + "learning_rate": 4.905273188618959e-05, + "loss": 4.3062, + "step": 57856 + }, + { + "epoch": 0.02, + "learning_rate": 4.904434593867907e-05, + "loss": 4.3011, + "step": 58368 + }, + { + "epoch": 0.02, + "learning_rate": 4.903595999116855e-05, + "loss": 4.3154, + "step": 58880 + }, + { + "epoch": 0.02, + "learning_rate": 4.902757404365803e-05, + "loss": 4.3056, + "step": 59392 + }, + { + "epoch": 0.02, + "learning_rate": 4.9019220853754975e-05, + "loss": 4.3019, + "step": 59904 + }, + { + "epoch": 0.02, + "learning_rate": 4.9010834906244455e-05, + "loss": 4.2895, + "step": 60416 + }, + { + "epoch": 0.02, + "learning_rate": 4.9002448958733935e-05, + "loss": 4.2959, + "step": 60928 + }, + { + "epoch": 0.02, + "learning_rate": 4.8994063011223415e-05, + "loss": 4.2879, + "step": 61440 + }, + { + "epoch": 0.02, + "learning_rate": 4.8985677063712895e-05, + "loss": 4.2875, + "step": 61952 + }, + { + "epoch": 0.02, + "learning_rate": 4.8977291116202375e-05, + "loss": 4.2958, + "step": 62464 + }, + { + "epoch": 0.02, + "learning_rate": 4.8968905168691855e-05, + "loss": 4.2793, + "step": 62976 + }, + { + "epoch": 0.02, + "learning_rate": 4.8960519221181335e-05, + "loss": 4.2875, + "step": 63488 + }, + { + "epoch": 0.02, + "learning_rate": 4.8952133273670815e-05, + "loss": 4.2677, + "step": 64000 + }, + { + "epoch": 0.02, + "learning_rate": 4.8943763704964024e-05, + "loss": 4.2711, + "step": 64512 + }, + { + "epoch": 0.02, + "learning_rate": 4.8935377757453504e-05, + "loss": 4.277, + "step": 65024 + }, + { + "epoch": 0.02, + "learning_rate": 4.8926991809942984e-05, + "loss": 4.2687, + "step": 65536 + }, + { + "epoch": 0.02, + "learning_rate": 4.8918605862432463e-05, + "loss": 4.26, + "step": 66048 + }, + { + "epoch": 0.02, + "learning_rate": 4.8910219914921943e-05, + "loss": 4.2858, + "step": 66560 + }, + { + "epoch": 0.02, + "learning_rate": 4.8901833967411423e-05, + "loss": 4.2652, + "step": 67072 + }, + { + "epoch": 0.02, + "learning_rate": 4.889346439870464e-05, + "loss": 4.2599, + "step": 67584 + }, + { + "epoch": 0.02, + "learning_rate": 4.888507845119412e-05, + "loss": 4.2603, + "step": 68096 + }, + { + "epoch": 0.02, + "learning_rate": 4.88766925036836e-05, + "loss": 4.2606, + "step": 68608 + }, + { + "epoch": 0.02, + "learning_rate": 4.886830655617308e-05, + "loss": 4.2616, + "step": 69120 + }, + { + "epoch": 0.02, + "learning_rate": 4.885992060866256e-05, + "loss": 4.2584, + "step": 69632 + }, + { + "epoch": 0.02, + "learning_rate": 4.885155103995577e-05, + "loss": 4.2534, + "step": 70144 + }, + { + "epoch": 0.02, + "learning_rate": 4.884316509244525e-05, + "loss": 4.2615, + "step": 70656 + }, + { + "epoch": 0.02, + "learning_rate": 4.883477914493473e-05, + "loss": 4.2583, + "step": 71168 + }, + { + "epoch": 0.02, + "learning_rate": 4.882639319742421e-05, + "loss": 4.2527, + "step": 71680 + }, + { + "epoch": 0.02, + "learning_rate": 4.881802362871742e-05, + "loss": 4.2425, + "step": 72192 + }, + { + "epoch": 0.02, + "learning_rate": 4.88096376812069e-05, + "loss": 4.2498, + "step": 72704 + }, + { + "epoch": 0.02, + "learning_rate": 4.880125173369638e-05, + "loss": 4.2415, + "step": 73216 + }, + { + "epoch": 0.02, + "learning_rate": 4.879286578618586e-05, + "loss": 4.2428, + "step": 73728 + }, + { + "epoch": 0.02, + "learning_rate": 4.878449621747907e-05, + "loss": 4.2393, + "step": 74240 + }, + { + "epoch": 0.02, + "learning_rate": 4.877611026996855e-05, + "loss": 4.2362, + "step": 74752 + }, + { + "epoch": 0.02, + "learning_rate": 4.876772432245803e-05, + "loss": 4.2407, + "step": 75264 + }, + { + "epoch": 0.02, + "learning_rate": 4.875933837494751e-05, + "loss": 4.2284, + "step": 75776 + }, + { + "epoch": 0.02, + "learning_rate": 4.8750952427436986e-05, + "loss": 4.2257, + "step": 76288 + }, + { + "epoch": 0.03, + "eval_loss": 4.201012134552002, + "eval_runtime": 303.7486, + "eval_samples_per_second": 1256.272, + "eval_steps_per_second": 39.259, + "step": 76320 + }, + { + "epoch": 1.0, + "learning_rate": 4.87425828587302e-05, + "loss": 4.2093, + "step": 76800 + }, + { + "epoch": 1.0, + "learning_rate": 4.873419691121968e-05, + "loss": 4.212, + "step": 77312 + }, + { + "epoch": 1.0, + "learning_rate": 4.872581096370916e-05, + "loss": 4.2394, + "step": 77824 + }, + { + "epoch": 1.0, + "learning_rate": 4.8717425016198635e-05, + "loss": 4.2251, + "step": 78336 + }, + { + "epoch": 1.0, + "learning_rate": 4.8709039068688115e-05, + "loss": 4.2166, + "step": 78848 + }, + { + "epoch": 1.0, + "learning_rate": 4.8700653121177595e-05, + "loss": 4.213, + "step": 79360 + }, + { + "epoch": 1.0, + "learning_rate": 4.8692267173667075e-05, + "loss": 4.2111, + "step": 79872 + }, + { + "epoch": 1.0, + "learning_rate": 4.868389760496029e-05, + "loss": 4.1995, + "step": 80384 + }, + { + "epoch": 1.0, + "learning_rate": 4.867551165744977e-05, + "loss": 4.2115, + "step": 80896 + }, + { + "epoch": 1.0, + "learning_rate": 4.866712570993925e-05, + "loss": 4.2041, + "step": 81408 + }, + { + "epoch": 1.0, + "learning_rate": 4.865873976242873e-05, + "loss": 4.2084, + "step": 81920 + }, + { + "epoch": 1.0, + "learning_rate": 4.865035381491821e-05, + "loss": 4.2099, + "step": 82432 + }, + { + "epoch": 1.0, + "learning_rate": 4.8642000625015155e-05, + "loss": 4.1943, + "step": 82944 + }, + { + "epoch": 1.0, + "learning_rate": 4.8633614677504635e-05, + "loss": 4.1905, + "step": 83456 + }, + { + "epoch": 1.0, + "learning_rate": 4.862522872999411e-05, + "loss": 4.1824, + "step": 83968 + }, + { + "epoch": 1.0, + "learning_rate": 4.861684278248359e-05, + "loss": 4.1817, + "step": 84480 + }, + { + "epoch": 1.0, + "learning_rate": 4.860845683497307e-05, + "loss": 4.1913, + "step": 84992 + }, + { + "epoch": 1.0, + "learning_rate": 4.860007088746255e-05, + "loss": 4.1798, + "step": 85504 + }, + { + "epoch": 1.0, + "learning_rate": 4.859168493995203e-05, + "loss": 4.1845, + "step": 86016 + }, + { + "epoch": 1.0, + "learning_rate": 4.858329899244151e-05, + "loss": 4.2032, + "step": 86528 + }, + { + "epoch": 1.0, + "learning_rate": 4.8574929423734724e-05, + "loss": 4.1847, + "step": 87040 + }, + { + "epoch": 1.0, + "learning_rate": 4.8566543476224204e-05, + "loss": 4.1848, + "step": 87552 + }, + { + "epoch": 1.0, + "learning_rate": 4.8558157528713684e-05, + "loss": 4.1765, + "step": 88064 + }, + { + "epoch": 1.0, + "learning_rate": 4.8549771581203164e-05, + "loss": 4.1899, + "step": 88576 + }, + { + "epoch": 1.0, + "learning_rate": 4.8541385633692644e-05, + "loss": 4.1686, + "step": 89088 + }, + { + "epoch": 1.0, + "learning_rate": 4.8532999686182124e-05, + "loss": 4.1726, + "step": 89600 + }, + { + "epoch": 1.0, + "learning_rate": 4.852463011747533e-05, + "loss": 4.1645, + "step": 90112 + }, + { + "epoch": 1.0, + "learning_rate": 4.851624416996481e-05, + "loss": 4.1692, + "step": 90624 + }, + { + "epoch": 1.0, + "learning_rate": 4.850785822245429e-05, + "loss": 4.1626, + "step": 91136 + }, + { + "epoch": 1.01, + "learning_rate": 4.849947227494377e-05, + "loss": 4.1589, + "step": 91648 + }, + { + "epoch": 1.01, + "learning_rate": 4.849110270623698e-05, + "loss": 4.1634, + "step": 92160 + }, + { + "epoch": 1.01, + "learning_rate": 4.848271675872646e-05, + "loss": 4.1716, + "step": 92672 + }, + { + "epoch": 1.01, + "learning_rate": 4.847433081121595e-05, + "loss": 4.161, + "step": 93184 + }, + { + "epoch": 1.01, + "learning_rate": 4.846594486370543e-05, + "loss": 4.1628, + "step": 93696 + }, + { + "epoch": 1.01, + "learning_rate": 4.845757529499864e-05, + "loss": 4.1566, + "step": 94208 + }, + { + "epoch": 1.01, + "learning_rate": 4.844918934748812e-05, + "loss": 4.1642, + "step": 94720 + }, + { + "epoch": 1.01, + "learning_rate": 4.84408033999776e-05, + "loss": 4.1399, + "step": 95232 + }, + { + "epoch": 1.01, + "learning_rate": 4.843241745246708e-05, + "loss": 4.1549, + "step": 95744 + }, + { + "epoch": 1.01, + "learning_rate": 4.842404788376029e-05, + "loss": 4.1328, + "step": 96256 + }, + { + "epoch": 1.01, + "learning_rate": 4.841566193624977e-05, + "loss": 4.154, + "step": 96768 + }, + { + "epoch": 1.01, + "learning_rate": 4.840727598873925e-05, + "loss": 4.1379, + "step": 97280 + }, + { + "epoch": 1.01, + "learning_rate": 4.839889004122873e-05, + "loss": 4.1512, + "step": 97792 + }, + { + "epoch": 1.01, + "learning_rate": 4.8390520472521936e-05, + "loss": 4.1336, + "step": 98304 + }, + { + "epoch": 1.01, + "learning_rate": 4.8382134525011416e-05, + "loss": 4.1464, + "step": 98816 + }, + { + "epoch": 1.01, + "learning_rate": 4.83737485775009e-05, + "loss": 4.145, + "step": 99328 + }, + { + "epoch": 1.01, + "learning_rate": 4.836536262999038e-05, + "loss": 4.1423, + "step": 99840 + }, + { + "epoch": 1.01, + "learning_rate": 4.835699306128359e-05, + "loss": 4.148, + "step": 100352 + }, + { + "epoch": 1.01, + "learning_rate": 4.834860711377307e-05, + "loss": 4.1187, + "step": 100864 + }, + { + "epoch": 1.01, + "learning_rate": 4.834022116626255e-05, + "loss": 4.1267, + "step": 101376 + }, + { + "epoch": 1.01, + "learning_rate": 4.833183521875203e-05, + "loss": 4.1385, + "step": 101888 + }, + { + "epoch": 1.01, + "learning_rate": 4.832346565004524e-05, + "loss": 4.1349, + "step": 102400 + }, + { + "epoch": 1.01, + "learning_rate": 4.831507970253472e-05, + "loss": 4.1285, + "step": 102912 + }, + { + "epoch": 1.01, + "learning_rate": 4.83066937550242e-05, + "loss": 4.1231, + "step": 103424 + }, + { + "epoch": 1.01, + "learning_rate": 4.829830780751368e-05, + "loss": 4.1193, + "step": 103936 + }, + { + "epoch": 1.01, + "learning_rate": 4.828992186000316e-05, + "loss": 4.1113, + "step": 104448 + }, + { + "epoch": 1.01, + "learning_rate": 4.828155229129637e-05, + "loss": 4.1263, + "step": 104960 + }, + { + "epoch": 1.01, + "learning_rate": 4.8273166343785856e-05, + "loss": 4.1052, + "step": 105472 + }, + { + "epoch": 1.01, + "learning_rate": 4.8264780396275336e-05, + "loss": 4.1174, + "step": 105984 + }, + { + "epoch": 1.01, + "learning_rate": 4.8256394448764816e-05, + "loss": 4.127, + "step": 106496 + }, + { + "epoch": 1.01, + "learning_rate": 4.8248024880058025e-05, + "loss": 4.1069, + "step": 107008 + }, + { + "epoch": 1.01, + "learning_rate": 4.8239638932547505e-05, + "loss": 4.1089, + "step": 107520 + }, + { + "epoch": 1.01, + "learning_rate": 4.8231252985036985e-05, + "loss": 4.1078, + "step": 108032 + }, + { + "epoch": 1.01, + "learning_rate": 4.8222867037526465e-05, + "loss": 4.0985, + "step": 108544 + }, + { + "epoch": 1.01, + "learning_rate": 4.8214497468819674e-05, + "loss": 4.0997, + "step": 109056 + }, + { + "epoch": 1.01, + "learning_rate": 4.8206111521309154e-05, + "loss": 4.1168, + "step": 109568 + }, + { + "epoch": 1.01, + "learning_rate": 4.8197725573798634e-05, + "loss": 4.1009, + "step": 110080 + }, + { + "epoch": 1.01, + "learning_rate": 4.8189339626288114e-05, + "loss": 4.0966, + "step": 110592 + }, + { + "epoch": 1.01, + "learning_rate": 4.818097005758132e-05, + "loss": 4.0955, + "step": 111104 + }, + { + "epoch": 1.01, + "learning_rate": 4.817260048887454e-05, + "loss": 4.0962, + "step": 111616 + }, + { + "epoch": 1.01, + "learning_rate": 4.816421454136402e-05, + "loss": 4.1103, + "step": 112128 + }, + { + "epoch": 1.01, + "learning_rate": 4.81558285938535e-05, + "loss": 4.1037, + "step": 112640 + }, + { + "epoch": 1.01, + "learning_rate": 4.814744264634298e-05, + "loss": 4.0957, + "step": 113152 + }, + { + "epoch": 1.01, + "learning_rate": 4.813905669883246e-05, + "loss": 4.1041, + "step": 113664 + }, + { + "epoch": 1.01, + "learning_rate": 4.813067075132194e-05, + "loss": 4.1063, + "step": 114176 + }, + { + "epoch": 1.01, + "learning_rate": 4.812228480381141e-05, + "loss": 4.0869, + "step": 114688 + }, + { + "epoch": 1.01, + "learning_rate": 4.811391523510463e-05, + "loss": 4.0955, + "step": 115200 + }, + { + "epoch": 1.01, + "learning_rate": 4.810552928759411e-05, + "loss": 4.0929, + "step": 115712 + }, + { + "epoch": 1.01, + "learning_rate": 4.809714334008359e-05, + "loss": 4.09, + "step": 116224 + }, + { + "epoch": 1.01, + "learning_rate": 4.808875739257306e-05, + "loss": 4.0846, + "step": 116736 + }, + { + "epoch": 1.01, + "learning_rate": 4.808037144506255e-05, + "loss": 4.0879, + "step": 117248 + }, + { + "epoch": 1.01, + "learning_rate": 4.807198549755203e-05, + "loss": 4.0966, + "step": 117760 + }, + { + "epoch": 1.01, + "learning_rate": 4.806359955004151e-05, + "loss": 4.0952, + "step": 118272 + }, + { + "epoch": 1.01, + "learning_rate": 4.805521360253099e-05, + "loss": 4.0852, + "step": 118784 + }, + { + "epoch": 1.01, + "learning_rate": 4.8046844033824197e-05, + "loss": 4.0617, + "step": 119296 + }, + { + "epoch": 1.01, + "learning_rate": 4.8038458086313677e-05, + "loss": 4.0779, + "step": 119808 + }, + { + "epoch": 1.01, + "learning_rate": 4.8030072138803156e-05, + "loss": 4.084, + "step": 120320 + }, + { + "epoch": 1.01, + "learning_rate": 4.8021686191292636e-05, + "loss": 4.0829, + "step": 120832 + }, + { + "epoch": 1.01, + "learning_rate": 4.8013316622585846e-05, + "loss": 4.0779, + "step": 121344 + }, + { + "epoch": 1.01, + "learning_rate": 4.8004930675075325e-05, + "loss": 4.0743, + "step": 121856 + }, + { + "epoch": 1.02, + "learning_rate": 4.7996561106368535e-05, + "loss": 4.0822, + "step": 122368 + }, + { + "epoch": 1.02, + "learning_rate": 4.7988175158858015e-05, + "loss": 4.0728, + "step": 122880 + }, + { + "epoch": 1.02, + "learning_rate": 4.7979789211347494e-05, + "loss": 4.0667, + "step": 123392 + }, + { + "epoch": 1.02, + "learning_rate": 4.797140326383698e-05, + "loss": 4.0812, + "step": 123904 + }, + { + "epoch": 1.02, + "learning_rate": 4.796301731632646e-05, + "loss": 4.0629, + "step": 124416 + }, + { + "epoch": 1.02, + "learning_rate": 4.795464774761967e-05, + "loss": 4.0842, + "step": 124928 + }, + { + "epoch": 1.02, + "learning_rate": 4.794626180010915e-05, + "loss": 4.0749, + "step": 125440 + }, + { + "epoch": 1.02, + "learning_rate": 4.793787585259863e-05, + "loss": 4.0538, + "step": 125952 + }, + { + "epoch": 1.02, + "learning_rate": 4.792948990508811e-05, + "loss": 4.0688, + "step": 126464 + }, + { + "epoch": 1.02, + "learning_rate": 4.792110395757759e-05, + "loss": 4.066, + "step": 126976 + }, + { + "epoch": 1.02, + "learning_rate": 4.791271801006707e-05, + "loss": 4.0773, + "step": 127488 + }, + { + "epoch": 1.02, + "learning_rate": 4.790433206255655e-05, + "loss": 4.066, + "step": 128000 + }, + { + "epoch": 1.02, + "learning_rate": 4.789596249384976e-05, + "loss": 4.0617, + "step": 128512 + }, + { + "epoch": 1.02, + "learning_rate": 4.788757654633924e-05, + "loss": 4.0532, + "step": 129024 + }, + { + "epoch": 1.02, + "learning_rate": 4.787919059882872e-05, + "loss": 4.0692, + "step": 129536 + }, + { + "epoch": 1.02, + "learning_rate": 4.78708046513182e-05, + "loss": 4.0462, + "step": 130048 + }, + { + "epoch": 1.02, + "learning_rate": 4.7862435082611415e-05, + "loss": 4.0551, + "step": 130560 + }, + { + "epoch": 1.02, + "learning_rate": 4.7854049135100895e-05, + "loss": 4.0587, + "step": 131072 + }, + { + "epoch": 1.02, + "learning_rate": 4.7845663187590375e-05, + "loss": 4.0566, + "step": 131584 + }, + { + "epoch": 1.02, + "learning_rate": 4.7837277240079855e-05, + "loss": 4.0457, + "step": 132096 + }, + { + "epoch": 1.02, + "learning_rate": 4.7828891292569335e-05, + "loss": 4.0567, + "step": 132608 + }, + { + "epoch": 1.02, + "learning_rate": 4.7820505345058815e-05, + "loss": 4.0501, + "step": 133120 + }, + { + "epoch": 1.02, + "learning_rate": 4.7812119397548295e-05, + "loss": 4.0542, + "step": 133632 + }, + { + "epoch": 1.02, + "learning_rate": 4.7803733450037775e-05, + "loss": 4.0505, + "step": 134144 + }, + { + "epoch": 1.02, + "learning_rate": 4.7795363881330984e-05, + "loss": 4.045, + "step": 134656 + }, + { + "epoch": 1.02, + "learning_rate": 4.7786977933820464e-05, + "loss": 4.0711, + "step": 135168 + }, + { + "epoch": 1.02, + "learning_rate": 4.7778591986309944e-05, + "loss": 4.0561, + "step": 135680 + }, + { + "epoch": 1.02, + "learning_rate": 4.777022241760315e-05, + "loss": 4.0541, + "step": 136192 + }, + { + "epoch": 1.02, + "learning_rate": 4.776183647009263e-05, + "loss": 4.0402, + "step": 136704 + }, + { + "epoch": 1.02, + "learning_rate": 4.775345052258212e-05, + "loss": 4.0471, + "step": 137216 + }, + { + "epoch": 1.02, + "learning_rate": 4.77450645750716e-05, + "loss": 4.0475, + "step": 137728 + }, + { + "epoch": 1.02, + "learning_rate": 4.773667862756108e-05, + "loss": 4.0393, + "step": 138240 + }, + { + "epoch": 1.02, + "learning_rate": 4.772829268005055e-05, + "loss": 4.056, + "step": 138752 + }, + { + "epoch": 1.02, + "learning_rate": 4.771990673254003e-05, + "loss": 4.0438, + "step": 139264 + }, + { + "epoch": 1.02, + "learning_rate": 4.771152078502951e-05, + "loss": 4.0422, + "step": 139776 + }, + { + "epoch": 1.02, + "learning_rate": 4.770315121632272e-05, + "loss": 4.0379, + "step": 140288 + }, + { + "epoch": 1.02, + "learning_rate": 4.76947652688122e-05, + "loss": 4.038, + "step": 140800 + }, + { + "epoch": 1.02, + "learning_rate": 4.768637932130168e-05, + "loss": 4.0406, + "step": 141312 + }, + { + "epoch": 1.02, + "learning_rate": 4.76780097525949e-05, + "loss": 4.0395, + "step": 141824 + }, + { + "epoch": 1.02, + "learning_rate": 4.766962380508437e-05, + "loss": 4.028, + "step": 142336 + }, + { + "epoch": 1.02, + "learning_rate": 4.766123785757386e-05, + "loss": 4.0562, + "step": 142848 + }, + { + "epoch": 1.02, + "learning_rate": 4.765285191006334e-05, + "loss": 4.0437, + "step": 143360 + }, + { + "epoch": 1.02, + "learning_rate": 4.764448234135655e-05, + "loss": 4.0334, + "step": 143872 + }, + { + "epoch": 1.02, + "learning_rate": 4.7636096393846026e-05, + "loss": 4.0327, + "step": 144384 + }, + { + "epoch": 1.02, + "learning_rate": 4.7627710446335506e-05, + "loss": 4.0304, + "step": 144896 + }, + { + "epoch": 1.02, + "learning_rate": 4.7619324498824986e-05, + "loss": 4.0427, + "step": 145408 + }, + { + "epoch": 1.02, + "learning_rate": 4.7610938551314466e-05, + "loss": 4.0376, + "step": 145920 + }, + { + "epoch": 1.02, + "learning_rate": 4.7602552603803946e-05, + "loss": 4.0331, + "step": 146432 + }, + { + "epoch": 1.02, + "learning_rate": 4.7594166656293426e-05, + "loss": 4.0437, + "step": 146944 + }, + { + "epoch": 1.02, + "learning_rate": 4.7585780708782906e-05, + "loss": 4.035, + "step": 147456 + }, + { + "epoch": 1.02, + "learning_rate": 4.7577411140076115e-05, + "loss": 4.0394, + "step": 147968 + }, + { + "epoch": 1.02, + "learning_rate": 4.7569025192565595e-05, + "loss": 4.028, + "step": 148480 + }, + { + "epoch": 1.02, + "learning_rate": 4.7560639245055075e-05, + "loss": 4.0362, + "step": 148992 + }, + { + "epoch": 1.02, + "learning_rate": 4.7552253297544555e-05, + "loss": 4.03, + "step": 149504 + }, + { + "epoch": 1.02, + "learning_rate": 4.754388372883777e-05, + "loss": 4.0278, + "step": 150016 + }, + { + "epoch": 1.02, + "learning_rate": 4.753549778132725e-05, + "loss": 4.0289, + "step": 150528 + }, + { + "epoch": 1.02, + "learning_rate": 4.752711183381673e-05, + "loss": 4.0292, + "step": 151040 + }, + { + "epoch": 1.02, + "learning_rate": 4.751872588630621e-05, + "loss": 4.0313, + "step": 151552 + }, + { + "epoch": 1.02, + "learning_rate": 4.751035631759942e-05, + "loss": 4.0182, + "step": 152064 + }, + { + "epoch": 1.02, + "learning_rate": 4.75019703700889e-05, + "loss": 4.0198, + "step": 152576 + }, + { + "epoch": 1.03, + "eval_loss": 4.033823490142822, + "eval_runtime": 303.4895, + "eval_samples_per_second": 1257.345, + "eval_steps_per_second": 39.293, + "step": 152640 + }, + { + "epoch": 0.0, + "learning_rate": 4.749358442257838e-05, + "loss": 4.0055, + "step": 153088 + }, + { + "epoch": 0.0, + "learning_rate": 4.748519847506786e-05, + "loss": 4.0083, + "step": 153600 + }, + { + "epoch": 0.0, + "learning_rate": 4.747681252755734e-05, + "loss": 4.036, + "step": 154112 + }, + { + "epoch": 0.0, + "learning_rate": 4.746842658004682e-05, + "loss": 4.0282, + "step": 154624 + }, + { + "epoch": 0.0, + "learning_rate": 4.74600406325363e-05, + "loss": 4.0193, + "step": 155136 + }, + { + "epoch": 0.0, + "learning_rate": 4.745165468502578e-05, + "loss": 4.0142, + "step": 155648 + }, + { + "epoch": 0.0, + "learning_rate": 4.744326873751526e-05, + "loss": 4.016, + "step": 156160 + }, + { + "epoch": 0.0, + "learning_rate": 4.743488279000474e-05, + "loss": 4.0034, + "step": 156672 + }, + { + "epoch": 0.0, + "learning_rate": 4.742649684249422e-05, + "loss": 4.0211, + "step": 157184 + }, + { + "epoch": 0.0, + "learning_rate": 4.74181108949837e-05, + "loss": 4.014, + "step": 157696 + }, + { + "epoch": 0.0, + "learning_rate": 4.740974132627691e-05, + "loss": 4.0138, + "step": 158208 + }, + { + "epoch": 0.0, + "learning_rate": 4.740135537876639e-05, + "loss": 4.0163, + "step": 158720 + }, + { + "epoch": 0.0, + "learning_rate": 4.739296943125587e-05, + "loss": 4.0035, + "step": 159232 + }, + { + "epoch": 0.0, + "learning_rate": 4.738458348374535e-05, + "loss": 4.0043, + "step": 159744 + }, + { + "epoch": 0.0, + "learning_rate": 4.737619753623483e-05, + "loss": 3.9992, + "step": 160256 + }, + { + "epoch": 0.0, + "learning_rate": 4.736782796752804e-05, + "loss": 3.9966, + "step": 160768 + }, + { + "epoch": 0.0, + "learning_rate": 4.735944202001752e-05, + "loss": 4.0025, + "step": 161280 + }, + { + "epoch": 0.0, + "learning_rate": 4.7351056072507e-05, + "loss": 3.9968, + "step": 161792 + }, + { + "epoch": 0.0, + "learning_rate": 4.734267012499648e-05, + "loss": 4.0015, + "step": 162304 + }, + { + "epoch": 0.0, + "learning_rate": 4.733428417748596e-05, + "loss": 4.026, + "step": 162816 + }, + { + "epoch": 0.0, + "learning_rate": 4.7325898229975444e-05, + "loss": 4.0016, + "step": 163328 + }, + { + "epoch": 0.0, + "learning_rate": 4.731752866126865e-05, + "loss": 4.0044, + "step": 163840 + }, + { + "epoch": 0.0, + "learning_rate": 4.730914271375813e-05, + "loss": 4.0024, + "step": 164352 + }, + { + "epoch": 0.0, + "learning_rate": 4.730075676624761e-05, + "loss": 4.0096, + "step": 164864 + }, + { + "epoch": 0.0, + "learning_rate": 4.729237081873709e-05, + "loss": 3.9884, + "step": 165376 + }, + { + "epoch": 0.0, + "learning_rate": 4.728398487122657e-05, + "loss": 4.0, + "step": 165888 + }, + { + "epoch": 0.0, + "learning_rate": 4.727559892371605e-05, + "loss": 3.9917, + "step": 166400 + }, + { + "epoch": 0.0, + "learning_rate": 4.726721297620553e-05, + "loss": 3.9923, + "step": 166912 + }, + { + "epoch": 0.0, + "learning_rate": 4.725882702869501e-05, + "loss": 3.9871, + "step": 167424 + }, + { + "epoch": 0.01, + "learning_rate": 4.725045745998822e-05, + "loss": 3.9856, + "step": 167936 + }, + { + "epoch": 0.01, + "learning_rate": 4.72420715124777e-05, + "loss": 3.9911, + "step": 168448 + }, + { + "epoch": 0.01, + "learning_rate": 4.723370194377091e-05, + "loss": 4.0011, + "step": 168960 + }, + { + "epoch": 0.01, + "learning_rate": 4.72253159962604e-05, + "loss": 3.9922, + "step": 169472 + }, + { + "epoch": 0.01, + "learning_rate": 4.721693004874988e-05, + "loss": 3.9983, + "step": 169984 + }, + { + "epoch": 0.01, + "learning_rate": 4.720854410123936e-05, + "loss": 3.9828, + "step": 170496 + }, + { + "epoch": 0.01, + "learning_rate": 4.720015815372884e-05, + "loss": 4.0005, + "step": 171008 + }, + { + "epoch": 0.01, + "learning_rate": 4.7191788585022046e-05, + "loss": 3.9725, + "step": 171520 + }, + { + "epoch": 0.01, + "learning_rate": 4.7183402637511526e-05, + "loss": 3.9922, + "step": 172032 + }, + { + "epoch": 0.01, + "learning_rate": 4.7175016690001006e-05, + "loss": 3.9683, + "step": 172544 + }, + { + "epoch": 0.01, + "learning_rate": 4.7166630742490486e-05, + "loss": 3.9884, + "step": 173056 + }, + { + "epoch": 0.01, + "learning_rate": 4.7158244794979966e-05, + "loss": 3.9749, + "step": 173568 + }, + { + "epoch": 0.01, + "learning_rate": 4.7149858847469446e-05, + "loss": 3.9912, + "step": 174080 + }, + { + "epoch": 0.01, + "learning_rate": 4.714147289995892e-05, + "loss": 3.9764, + "step": 174592 + }, + { + "epoch": 0.01, + "learning_rate": 4.7133103331252135e-05, + "loss": 3.9818, + "step": 175104 + }, + { + "epoch": 0.01, + "learning_rate": 4.7124717383741615e-05, + "loss": 3.9854, + "step": 175616 + }, + { + "epoch": 0.01, + "learning_rate": 4.7116331436231095e-05, + "loss": 3.9864, + "step": 176128 + }, + { + "epoch": 0.01, + "learning_rate": 4.7107945488720575e-05, + "loss": 3.9847, + "step": 176640 + }, + { + "epoch": 0.01, + "learning_rate": 4.7099559541210055e-05, + "loss": 3.9662, + "step": 177152 + }, + { + "epoch": 0.01, + "learning_rate": 4.709118997250327e-05, + "loss": 3.9665, + "step": 177664 + }, + { + "epoch": 0.01, + "learning_rate": 4.7082804024992744e-05, + "loss": 3.9839, + "step": 178176 + }, + { + "epoch": 0.01, + "learning_rate": 4.7074418077482224e-05, + "loss": 3.9804, + "step": 178688 + }, + { + "epoch": 0.01, + "learning_rate": 4.7066032129971704e-05, + "loss": 3.9737, + "step": 179200 + }, + { + "epoch": 0.01, + "learning_rate": 4.7057646182461184e-05, + "loss": 3.9727, + "step": 179712 + }, + { + "epoch": 0.01, + "learning_rate": 4.7049260234950664e-05, + "loss": 3.9693, + "step": 180224 + }, + { + "epoch": 0.01, + "learning_rate": 4.704089066624387e-05, + "loss": 3.9563, + "step": 180736 + }, + { + "epoch": 0.01, + "learning_rate": 4.703250471873335e-05, + "loss": 3.9724, + "step": 181248 + }, + { + "epoch": 0.01, + "learning_rate": 4.702411877122283e-05, + "loss": 3.956, + "step": 181760 + }, + { + "epoch": 0.01, + "learning_rate": 4.701574920251605e-05, + "loss": 3.9698, + "step": 182272 + }, + { + "epoch": 0.01, + "learning_rate": 4.700736325500553e-05, + "loss": 3.9783, + "step": 182784 + }, + { + "epoch": 0.01, + "learning_rate": 4.699897730749501e-05, + "loss": 3.9577, + "step": 183296 + }, + { + "epoch": 0.01, + "learning_rate": 4.699059135998449e-05, + "loss": 3.9611, + "step": 183808 + }, + { + "epoch": 0.01, + "learning_rate": 4.69822217912777e-05, + "loss": 3.9615, + "step": 184320 + }, + { + "epoch": 0.01, + "learning_rate": 4.697383584376718e-05, + "loss": 3.9515, + "step": 184832 + }, + { + "epoch": 0.01, + "learning_rate": 4.696544989625666e-05, + "loss": 3.9501, + "step": 185344 + }, + { + "epoch": 0.01, + "learning_rate": 4.695706394874614e-05, + "loss": 3.972, + "step": 185856 + }, + { + "epoch": 0.01, + "learning_rate": 4.694867800123562e-05, + "loss": 3.9586, + "step": 186368 + }, + { + "epoch": 0.01, + "learning_rate": 4.69402920537251e-05, + "loss": 3.9531, + "step": 186880 + }, + { + "epoch": 0.01, + "learning_rate": 4.6931922485018307e-05, + "loss": 3.952, + "step": 187392 + }, + { + "epoch": 0.01, + "learning_rate": 4.6923536537507787e-05, + "loss": 3.9552, + "step": 187904 + }, + { + "epoch": 0.01, + "learning_rate": 4.6915150589997267e-05, + "loss": 3.965, + "step": 188416 + }, + { + "epoch": 0.01, + "learning_rate": 4.690676464248675e-05, + "loss": 3.9638, + "step": 188928 + }, + { + "epoch": 0.01, + "learning_rate": 4.689837869497623e-05, + "loss": 3.9522, + "step": 189440 + }, + { + "epoch": 0.01, + "learning_rate": 4.689000912626944e-05, + "loss": 3.9646, + "step": 189952 + }, + { + "epoch": 0.01, + "learning_rate": 4.688162317875892e-05, + "loss": 3.9616, + "step": 190464 + }, + { + "epoch": 0.01, + "learning_rate": 4.68732372312484e-05, + "loss": 3.9502, + "step": 190976 + }, + { + "epoch": 0.01, + "learning_rate": 4.686485128373788e-05, + "loss": 3.9564, + "step": 191488 + }, + { + "epoch": 0.01, + "learning_rate": 4.685648171503109e-05, + "loss": 3.9596, + "step": 192000 + }, + { + "epoch": 0.01, + "learning_rate": 4.684809576752057e-05, + "loss": 3.952, + "step": 192512 + }, + { + "epoch": 0.01, + "learning_rate": 4.683970982001005e-05, + "loss": 3.9489, + "step": 193024 + }, + { + "epoch": 0.01, + "learning_rate": 4.683132387249953e-05, + "loss": 3.9504, + "step": 193536 + }, + { + "epoch": 0.01, + "learning_rate": 4.682293792498901e-05, + "loss": 3.9592, + "step": 194048 + }, + { + "epoch": 0.01, + "learning_rate": 4.681456835628222e-05, + "loss": 3.9611, + "step": 194560 + }, + { + "epoch": 0.01, + "learning_rate": 4.680618240877171e-05, + "loss": 3.9456, + "step": 195072 + }, + { + "epoch": 0.01, + "learning_rate": 4.679779646126119e-05, + "loss": 3.9334, + "step": 195584 + }, + { + "epoch": 0.01, + "learning_rate": 4.678941051375067e-05, + "loss": 3.9395, + "step": 196096 + }, + { + "epoch": 0.01, + "learning_rate": 4.678102456624015e-05, + "loss": 3.9484, + "step": 196608 + }, + { + "epoch": 0.01, + "learning_rate": 4.6772654997533356e-05, + "loss": 3.9513, + "step": 197120 + }, + { + "epoch": 0.01, + "learning_rate": 4.6764269050022836e-05, + "loss": 3.9432, + "step": 197632 + }, + { + "epoch": 0.01, + "learning_rate": 4.6755883102512316e-05, + "loss": 3.9409, + "step": 198144 + }, + { + "epoch": 0.02, + "learning_rate": 4.6747497155001796e-05, + "loss": 3.955, + "step": 198656 + }, + { + "epoch": 0.02, + "learning_rate": 4.6739127586295005e-05, + "loss": 3.9381, + "step": 199168 + }, + { + "epoch": 0.02, + "learning_rate": 4.6730741638784485e-05, + "loss": 3.9386, + "step": 199680 + }, + { + "epoch": 0.02, + "learning_rate": 4.6722355691273965e-05, + "loss": 3.9489, + "step": 200192 + }, + { + "epoch": 0.02, + "learning_rate": 4.6713969743763445e-05, + "loss": 3.9368, + "step": 200704 + }, + { + "epoch": 0.02, + "learning_rate": 4.670560017505666e-05, + "loss": 3.9556, + "step": 201216 + }, + { + "epoch": 0.02, + "learning_rate": 4.669721422754614e-05, + "loss": 3.9436, + "step": 201728 + }, + { + "epoch": 0.02, + "learning_rate": 4.668882828003562e-05, + "loss": 3.9272, + "step": 202240 + }, + { + "epoch": 0.02, + "learning_rate": 4.66804423325251e-05, + "loss": 3.9413, + "step": 202752 + }, + { + "epoch": 0.02, + "learning_rate": 4.667205638501458e-05, + "loss": 3.9366, + "step": 203264 + }, + { + "epoch": 0.02, + "learning_rate": 4.6663670437504054e-05, + "loss": 3.9531, + "step": 203776 + }, + { + "epoch": 0.02, + "learning_rate": 4.6655284489993534e-05, + "loss": 3.9426, + "step": 204288 + }, + { + "epoch": 0.02, + "learning_rate": 4.6646898542483013e-05, + "loss": 3.9337, + "step": 204800 + }, + { + "epoch": 0.02, + "learning_rate": 4.663852897377623e-05, + "loss": 3.9248, + "step": 205312 + }, + { + "epoch": 0.02, + "learning_rate": 4.66301430262657e-05, + "loss": 3.9436, + "step": 205824 + }, + { + "epoch": 0.02, + "learning_rate": 4.662175707875518e-05, + "loss": 3.9237, + "step": 206336 + }, + { + "epoch": 0.02, + "learning_rate": 4.661337113124466e-05, + "loss": 3.9341, + "step": 206848 + }, + { + "epoch": 0.02, + "learning_rate": 4.660500156253788e-05, + "loss": 3.9346, + "step": 207360 + }, + { + "epoch": 0.02, + "learning_rate": 4.659661561502736e-05, + "loss": 3.933, + "step": 207872 + }, + { + "epoch": 0.02, + "learning_rate": 4.658822966751684e-05, + "loss": 3.922, + "step": 208384 + }, + { + "epoch": 0.02, + "learning_rate": 4.657984372000632e-05, + "loss": 3.9415, + "step": 208896 + }, + { + "epoch": 0.02, + "learning_rate": 4.657147415129953e-05, + "loss": 3.9261, + "step": 209408 + }, + { + "epoch": 0.02, + "learning_rate": 4.656308820378901e-05, + "loss": 3.9305, + "step": 209920 + }, + { + "epoch": 0.02, + "learning_rate": 4.655470225627849e-05, + "loss": 3.9348, + "step": 210432 + }, + { + "epoch": 0.02, + "learning_rate": 4.654631630876797e-05, + "loss": 3.9262, + "step": 210944 + }, + { + "epoch": 0.02, + "learning_rate": 4.6537946740061176e-05, + "loss": 3.9471, + "step": 211456 + }, + { + "epoch": 0.02, + "learning_rate": 4.6529560792550656e-05, + "loss": 3.9367, + "step": 211968 + }, + { + "epoch": 0.02, + "learning_rate": 4.6521174845040136e-05, + "loss": 3.9348, + "step": 212480 + }, + { + "epoch": 0.02, + "learning_rate": 4.6512788897529616e-05, + "loss": 3.9233, + "step": 212992 + }, + { + "epoch": 0.02, + "learning_rate": 4.6504402950019096e-05, + "loss": 3.9292, + "step": 213504 + }, + { + "epoch": 0.02, + "learning_rate": 4.649603338131231e-05, + "loss": 3.9314, + "step": 214016 + }, + { + "epoch": 0.02, + "learning_rate": 4.648764743380179e-05, + "loss": 3.9184, + "step": 214528 + }, + { + "epoch": 0.02, + "learning_rate": 4.647926148629127e-05, + "loss": 3.9345, + "step": 215040 + }, + { + "epoch": 0.02, + "learning_rate": 4.647087553878075e-05, + "loss": 3.9289, + "step": 215552 + }, + { + "epoch": 0.02, + "learning_rate": 4.646250597007396e-05, + "loss": 3.9263, + "step": 216064 + }, + { + "epoch": 0.02, + "learning_rate": 4.645412002256344e-05, + "loss": 3.9265, + "step": 216576 + }, + { + "epoch": 0.02, + "learning_rate": 4.644573407505292e-05, + "loss": 3.9247, + "step": 217088 + }, + { + "epoch": 0.02, + "learning_rate": 4.64373481275424e-05, + "loss": 3.9177, + "step": 217600 + }, + { + "epoch": 0.02, + "learning_rate": 4.642897855883561e-05, + "loss": 3.9297, + "step": 218112 + }, + { + "epoch": 0.02, + "learning_rate": 4.642059261132509e-05, + "loss": 3.9158, + "step": 218624 + }, + { + "epoch": 0.02, + "learning_rate": 4.641220666381457e-05, + "loss": 3.9389, + "step": 219136 + }, + { + "epoch": 0.02, + "learning_rate": 4.640382071630405e-05, + "loss": 3.9346, + "step": 219648 + }, + { + "epoch": 0.02, + "learning_rate": 4.6395451147597266e-05, + "loss": 3.9187, + "step": 220160 + }, + { + "epoch": 0.02, + "learning_rate": 4.6387065200086746e-05, + "loss": 3.9192, + "step": 220672 + }, + { + "epoch": 0.02, + "learning_rate": 4.6378679252576226e-05, + "loss": 3.9207, + "step": 221184 + }, + { + "epoch": 0.02, + "learning_rate": 4.6370293305065705e-05, + "loss": 3.9275, + "step": 221696 + }, + { + "epoch": 0.02, + "learning_rate": 4.6361907357555185e-05, + "loss": 3.9238, + "step": 222208 + }, + { + "epoch": 0.02, + "learning_rate": 4.6353537788848395e-05, + "loss": 3.9199, + "step": 222720 + }, + { + "epoch": 0.02, + "learning_rate": 4.6345151841337874e-05, + "loss": 3.9295, + "step": 223232 + }, + { + "epoch": 0.02, + "learning_rate": 4.6336765893827354e-05, + "loss": 3.9239, + "step": 223744 + }, + { + "epoch": 0.02, + "learning_rate": 4.6328379946316834e-05, + "loss": 3.9293, + "step": 224256 + }, + { + "epoch": 0.02, + "learning_rate": 4.6320010377610043e-05, + "loss": 3.9155, + "step": 224768 + }, + { + "epoch": 0.02, + "learning_rate": 4.6311624430099523e-05, + "loss": 3.9286, + "step": 225280 + }, + { + "epoch": 0.02, + "learning_rate": 4.6303238482589003e-05, + "loss": 3.9208, + "step": 225792 + }, + { + "epoch": 0.02, + "learning_rate": 4.629485253507848e-05, + "loss": 3.9145, + "step": 226304 + }, + { + "epoch": 0.02, + "learning_rate": 4.62864829663717e-05, + "loss": 3.923, + "step": 226816 + }, + { + "epoch": 0.02, + "learning_rate": 4.627809701886118e-05, + "loss": 3.9231, + "step": 227328 + }, + { + "epoch": 0.02, + "learning_rate": 4.626971107135066e-05, + "loss": 3.9192, + "step": 227840 + }, + { + "epoch": 0.02, + "learning_rate": 4.626132512384014e-05, + "loss": 3.9073, + "step": 228352 + }, + { + "epoch": 0.02, + "learning_rate": 4.625295555513335e-05, + "loss": 3.9095, + "step": 228864 + }, + { + "epoch": 0.03, + "eval_loss": 3.958364486694336, + "eval_runtime": 308.0893, + "eval_samples_per_second": 1238.573, + "eval_steps_per_second": 38.706, + "step": 228960 + }, + { + "epoch": 0.0, + "learning_rate": 4.624456960762283e-05, + "loss": 3.9009, + "step": 229376 + }, + { + "epoch": 0.0, + "learning_rate": 4.623618366011231e-05, + "loss": 3.9006, + "step": 229888 + }, + { + "epoch": 0.0, + "learning_rate": 4.622779771260179e-05, + "loss": 3.9274, + "step": 230400 + }, + { + "epoch": 0.0, + "learning_rate": 4.621941176509127e-05, + "loss": 3.922, + "step": 230912 + }, + { + "epoch": 0.0, + "learning_rate": 4.621102581758075e-05, + "loss": 3.9205, + "step": 231424 + }, + { + "epoch": 0.0, + "learning_rate": 4.620263987007023e-05, + "loss": 3.9009, + "step": 231936 + }, + { + "epoch": 0.0, + "learning_rate": 4.619425392255971e-05, + "loss": 3.9157, + "step": 232448 + }, + { + "epoch": 0.0, + "learning_rate": 4.618586797504919e-05, + "loss": 3.8953, + "step": 232960 + }, + { + "epoch": 0.0, + "learning_rate": 4.6177498406342404e-05, + "loss": 3.9188, + "step": 233472 + }, + { + "epoch": 0.0, + "learning_rate": 4.6169112458831884e-05, + "loss": 3.9127, + "step": 233984 + }, + { + "epoch": 0.0, + "learning_rate": 4.6160726511321364e-05, + "loss": 3.9063, + "step": 234496 + }, + { + "epoch": 0.0, + "learning_rate": 4.615234056381084e-05, + "loss": 3.917, + "step": 235008 + }, + { + "epoch": 0.0, + "learning_rate": 4.614398737390778e-05, + "loss": 3.8986, + "step": 235520 + }, + { + "epoch": 0.0, + "learning_rate": 4.613560142639726e-05, + "loss": 3.9049, + "step": 236032 + }, + { + "epoch": 0.0, + "learning_rate": 4.612721547888674e-05, + "loss": 3.8927, + "step": 236544 + }, + { + "epoch": 0.0, + "learning_rate": 4.611882953137622e-05, + "loss": 3.9002, + "step": 237056 + }, + { + "epoch": 0.0, + "learning_rate": 4.61104435838657e-05, + "loss": 3.8979, + "step": 237568 + }, + { + "epoch": 0.0, + "learning_rate": 4.610205763635518e-05, + "loss": 3.8989, + "step": 238080 + }, + { + "epoch": 0.0, + "learning_rate": 4.609367168884466e-05, + "loss": 3.9001, + "step": 238592 + }, + { + "epoch": 0.0, + "learning_rate": 4.608528574133414e-05, + "loss": 3.9214, + "step": 239104 + }, + { + "epoch": 0.0, + "learning_rate": 4.607691617262736e-05, + "loss": 3.9005, + "step": 239616 + }, + { + "epoch": 0.0, + "learning_rate": 4.606853022511684e-05, + "loss": 3.9064, + "step": 240128 + }, + { + "epoch": 0.0, + "learning_rate": 4.606014427760631e-05, + "loss": 3.9019, + "step": 240640 + }, + { + "epoch": 0.0, + "learning_rate": 4.605175833009579e-05, + "loss": 3.9089, + "step": 241152 + }, + { + "epoch": 0.0, + "learning_rate": 4.6043388761389006e-05, + "loss": 3.8895, + "step": 241664 + }, + { + "epoch": 0.0, + "learning_rate": 4.603500281387848e-05, + "loss": 3.9013, + "step": 242176 + }, + { + "epoch": 0.0, + "learning_rate": 4.602661686636796e-05, + "loss": 3.8945, + "step": 242688 + }, + { + "epoch": 0.0, + "learning_rate": 4.601823091885744e-05, + "loss": 3.8923, + "step": 243200 + }, + { + "epoch": 0.0, + "learning_rate": 4.6009861350150655e-05, + "loss": 3.8884, + "step": 243712 + }, + { + "epoch": 0.01, + "learning_rate": 4.600147540264013e-05, + "loss": 3.8947, + "step": 244224 + }, + { + "epoch": 0.01, + "learning_rate": 4.5993089455129615e-05, + "loss": 3.8919, + "step": 244736 + }, + { + "epoch": 0.01, + "learning_rate": 4.598471988642283e-05, + "loss": 3.9031, + "step": 245248 + }, + { + "epoch": 0.01, + "learning_rate": 4.597633393891231e-05, + "loss": 3.8958, + "step": 245760 + }, + { + "epoch": 0.01, + "learning_rate": 4.5967947991401784e-05, + "loss": 3.8984, + "step": 246272 + }, + { + "epoch": 0.01, + "learning_rate": 4.5959562043891264e-05, + "loss": 3.8892, + "step": 246784 + }, + { + "epoch": 0.01, + "learning_rate": 4.5951176096380744e-05, + "loss": 3.9014, + "step": 247296 + }, + { + "epoch": 0.01, + "learning_rate": 4.5942790148870224e-05, + "loss": 3.8851, + "step": 247808 + }, + { + "epoch": 0.01, + "learning_rate": 4.5934404201359704e-05, + "loss": 3.894, + "step": 248320 + }, + { + "epoch": 0.01, + "learning_rate": 4.5926018253849184e-05, + "loss": 3.8727, + "step": 248832 + }, + { + "epoch": 0.01, + "learning_rate": 4.591764868514239e-05, + "loss": 3.8911, + "step": 249344 + }, + { + "epoch": 0.01, + "learning_rate": 4.590926273763187e-05, + "loss": 3.8851, + "step": 249856 + }, + { + "epoch": 0.01, + "learning_rate": 4.590087679012135e-05, + "loss": 3.9006, + "step": 250368 + }, + { + "epoch": 0.01, + "learning_rate": 4.589249084261083e-05, + "loss": 3.8841, + "step": 250880 + }, + { + "epoch": 0.01, + "learning_rate": 4.588412127390405e-05, + "loss": 3.889, + "step": 251392 + }, + { + "epoch": 0.01, + "learning_rate": 4.587573532639353e-05, + "loss": 3.8912, + "step": 251904 + }, + { + "epoch": 0.01, + "learning_rate": 4.586734937888301e-05, + "loss": 3.8971, + "step": 252416 + }, + { + "epoch": 0.01, + "learning_rate": 4.585896343137249e-05, + "loss": 3.8875, + "step": 252928 + }, + { + "epoch": 0.01, + "learning_rate": 4.58505938626657e-05, + "loss": 3.8763, + "step": 253440 + }, + { + "epoch": 0.01, + "learning_rate": 4.584220791515518e-05, + "loss": 3.8701, + "step": 253952 + }, + { + "epoch": 0.01, + "learning_rate": 4.583382196764466e-05, + "loss": 3.8915, + "step": 254464 + }, + { + "epoch": 0.01, + "learning_rate": 4.582543602013414e-05, + "loss": 3.8903, + "step": 254976 + }, + { + "epoch": 0.01, + "learning_rate": 4.581706645142735e-05, + "loss": 3.8832, + "step": 255488 + }, + { + "epoch": 0.01, + "learning_rate": 4.580868050391683e-05, + "loss": 3.8758, + "step": 256000 + }, + { + "epoch": 0.01, + "learning_rate": 4.580029455640631e-05, + "loss": 3.8817, + "step": 256512 + }, + { + "epoch": 0.01, + "learning_rate": 4.579190860889579e-05, + "loss": 3.8644, + "step": 257024 + }, + { + "epoch": 0.01, + "learning_rate": 4.5783539040189e-05, + "loss": 3.8856, + "step": 257536 + }, + { + "epoch": 0.01, + "learning_rate": 4.577515309267848e-05, + "loss": 3.8654, + "step": 258048 + }, + { + "epoch": 0.01, + "learning_rate": 4.576676714516796e-05, + "loss": 3.8802, + "step": 258560 + }, + { + "epoch": 0.01, + "learning_rate": 4.575838119765744e-05, + "loss": 3.8855, + "step": 259072 + }, + { + "epoch": 0.01, + "learning_rate": 4.575001162895065e-05, + "loss": 3.8745, + "step": 259584 + }, + { + "epoch": 0.01, + "learning_rate": 4.574162568144013e-05, + "loss": 3.868, + "step": 260096 + }, + { + "epoch": 0.01, + "learning_rate": 4.573323973392961e-05, + "loss": 3.8754, + "step": 260608 + }, + { + "epoch": 0.01, + "learning_rate": 4.572485378641909e-05, + "loss": 3.8644, + "step": 261120 + }, + { + "epoch": 0.01, + "learning_rate": 4.57164842177123e-05, + "loss": 3.8654, + "step": 261632 + }, + { + "epoch": 0.01, + "learning_rate": 4.570809827020178e-05, + "loss": 3.8805, + "step": 262144 + }, + { + "epoch": 0.01, + "learning_rate": 4.569971232269126e-05, + "loss": 3.8733, + "step": 262656 + }, + { + "epoch": 0.01, + "learning_rate": 4.569132637518074e-05, + "loss": 3.8654, + "step": 263168 + }, + { + "epoch": 0.01, + "learning_rate": 4.5682956806473956e-05, + "loss": 3.8624, + "step": 263680 + }, + { + "epoch": 0.01, + "learning_rate": 4.5674570858963436e-05, + "loss": 3.8703, + "step": 264192 + }, + { + "epoch": 0.01, + "learning_rate": 4.5666184911452916e-05, + "loss": 3.8726, + "step": 264704 + }, + { + "epoch": 0.01, + "learning_rate": 4.5657798963942396e-05, + "loss": 3.878, + "step": 265216 + }, + { + "epoch": 0.01, + "learning_rate": 4.5649429395235605e-05, + "loss": 3.87, + "step": 265728 + }, + { + "epoch": 0.01, + "learning_rate": 4.5641043447725085e-05, + "loss": 3.877, + "step": 266240 + }, + { + "epoch": 0.01, + "learning_rate": 4.5632657500214565e-05, + "loss": 3.8792, + "step": 266752 + }, + { + "epoch": 0.01, + "learning_rate": 4.5624271552704045e-05, + "loss": 3.8647, + "step": 267264 + }, + { + "epoch": 0.01, + "learning_rate": 4.5615901983997254e-05, + "loss": 3.867, + "step": 267776 + }, + { + "epoch": 0.01, + "learning_rate": 4.5607516036486734e-05, + "loss": 3.8757, + "step": 268288 + }, + { + "epoch": 0.01, + "learning_rate": 4.5599130088976214e-05, + "loss": 3.8708, + "step": 268800 + }, + { + "epoch": 0.01, + "learning_rate": 4.5590744141465694e-05, + "loss": 3.8653, + "step": 269312 + }, + { + "epoch": 0.01, + "learning_rate": 4.558237457275891e-05, + "loss": 3.8655, + "step": 269824 + }, + { + "epoch": 0.01, + "learning_rate": 4.557398862524839e-05, + "loss": 3.8727, + "step": 270336 + }, + { + "epoch": 0.01, + "learning_rate": 4.556560267773787e-05, + "loss": 3.8771, + "step": 270848 + }, + { + "epoch": 0.01, + "learning_rate": 4.555721673022735e-05, + "loss": 3.865, + "step": 271360 + }, + { + "epoch": 0.01, + "learning_rate": 4.554884716152056e-05, + "loss": 3.8519, + "step": 271872 + }, + { + "epoch": 0.01, + "learning_rate": 4.554046121401004e-05, + "loss": 3.8522, + "step": 272384 + }, + { + "epoch": 0.01, + "learning_rate": 4.553207526649952e-05, + "loss": 3.8631, + "step": 272896 + }, + { + "epoch": 0.01, + "learning_rate": 4.5523689318989e-05, + "loss": 3.8703, + "step": 273408 + }, + { + "epoch": 0.01, + "learning_rate": 4.551531975028221e-05, + "loss": 3.864, + "step": 273920 + }, + { + "epoch": 0.01, + "learning_rate": 4.550693380277169e-05, + "loss": 3.853, + "step": 274432 + }, + { + "epoch": 0.02, + "learning_rate": 4.549854785526117e-05, + "loss": 3.8691, + "step": 274944 + }, + { + "epoch": 0.02, + "learning_rate": 4.549016190775065e-05, + "loss": 3.8615, + "step": 275456 + }, + { + "epoch": 0.02, + "learning_rate": 4.5481792339043864e-05, + "loss": 3.8576, + "step": 275968 + }, + { + "epoch": 0.02, + "learning_rate": 4.5473406391533343e-05, + "loss": 3.8616, + "step": 276480 + }, + { + "epoch": 0.02, + "learning_rate": 4.5465020444022823e-05, + "loss": 3.8646, + "step": 276992 + }, + { + "epoch": 0.02, + "learning_rate": 4.545665087531603e-05, + "loss": 3.8748, + "step": 277504 + }, + { + "epoch": 0.02, + "learning_rate": 4.544826492780551e-05, + "loss": 3.8585, + "step": 278016 + }, + { + "epoch": 0.02, + "learning_rate": 4.543987898029499e-05, + "loss": 3.8438, + "step": 278528 + }, + { + "epoch": 0.02, + "learning_rate": 4.543149303278447e-05, + "loss": 3.8647, + "step": 279040 + }, + { + "epoch": 0.02, + "learning_rate": 4.542310708527395e-05, + "loss": 3.8557, + "step": 279552 + }, + { + "epoch": 0.02, + "learning_rate": 4.541472113776343e-05, + "loss": 3.8717, + "step": 280064 + }, + { + "epoch": 0.02, + "learning_rate": 4.540633519025291e-05, + "loss": 3.8661, + "step": 280576 + }, + { + "epoch": 0.02, + "learning_rate": 4.5397949242742385e-05, + "loss": 3.8532, + "step": 281088 + }, + { + "epoch": 0.02, + "learning_rate": 4.53895796740356e-05, + "loss": 3.8483, + "step": 281600 + }, + { + "epoch": 0.02, + "learning_rate": 4.538119372652508e-05, + "loss": 3.8652, + "step": 282112 + }, + { + "epoch": 0.02, + "learning_rate": 4.537280777901456e-05, + "loss": 3.844, + "step": 282624 + }, + { + "epoch": 0.02, + "learning_rate": 4.536442183150404e-05, + "loss": 3.8558, + "step": 283136 + }, + { + "epoch": 0.02, + "learning_rate": 4.535605226279726e-05, + "loss": 3.8538, + "step": 283648 + }, + { + "epoch": 0.02, + "learning_rate": 4.534766631528674e-05, + "loss": 3.8582, + "step": 284160 + }, + { + "epoch": 0.02, + "learning_rate": 4.5339296746579946e-05, + "loss": 3.8429, + "step": 284672 + }, + { + "epoch": 0.02, + "learning_rate": 4.5330910799069426e-05, + "loss": 3.8622, + "step": 285184 + }, + { + "epoch": 0.02, + "learning_rate": 4.5322524851558906e-05, + "loss": 3.8472, + "step": 285696 + }, + { + "epoch": 0.02, + "learning_rate": 4.5314138904048386e-05, + "loss": 3.851, + "step": 286208 + }, + { + "epoch": 0.02, + "learning_rate": 4.530575295653786e-05, + "loss": 3.8603, + "step": 286720 + }, + { + "epoch": 0.02, + "learning_rate": 4.529736700902734e-05, + "loss": 3.8482, + "step": 287232 + }, + { + "epoch": 0.02, + "learning_rate": 4.528898106151682e-05, + "loss": 3.8696, + "step": 287744 + }, + { + "epoch": 0.02, + "learning_rate": 4.52805951140063e-05, + "loss": 3.8591, + "step": 288256 + }, + { + "epoch": 0.02, + "learning_rate": 4.5272225545299515e-05, + "loss": 3.8558, + "step": 288768 + }, + { + "epoch": 0.02, + "learning_rate": 4.5263839597788995e-05, + "loss": 3.8464, + "step": 289280 + }, + { + "epoch": 0.02, + "learning_rate": 4.5255453650278475e-05, + "loss": 3.8535, + "step": 289792 + }, + { + "epoch": 0.02, + "learning_rate": 4.5247067702767955e-05, + "loss": 3.851, + "step": 290304 + }, + { + "epoch": 0.02, + "learning_rate": 4.5238698134061164e-05, + "loss": 3.8446, + "step": 290816 + }, + { + "epoch": 0.02, + "learning_rate": 4.5230312186550644e-05, + "loss": 3.8566, + "step": 291328 + }, + { + "epoch": 0.02, + "learning_rate": 4.5221926239040124e-05, + "loss": 3.8533, + "step": 291840 + }, + { + "epoch": 0.02, + "learning_rate": 4.5213540291529604e-05, + "loss": 3.8516, + "step": 292352 + }, + { + "epoch": 0.02, + "learning_rate": 4.520517072282281e-05, + "loss": 3.8483, + "step": 292864 + }, + { + "epoch": 0.02, + "learning_rate": 4.519678477531229e-05, + "loss": 3.8513, + "step": 293376 + }, + { + "epoch": 0.02, + "learning_rate": 4.518839882780177e-05, + "loss": 3.8458, + "step": 293888 + }, + { + "epoch": 0.02, + "learning_rate": 4.518001288029125e-05, + "loss": 3.8545, + "step": 294400 + }, + { + "epoch": 0.02, + "learning_rate": 4.517164331158447e-05, + "loss": 3.8404, + "step": 294912 + }, + { + "epoch": 0.02, + "learning_rate": 4.516325736407395e-05, + "loss": 3.8604, + "step": 295424 + }, + { + "epoch": 0.02, + "learning_rate": 4.515487141656343e-05, + "loss": 3.8547, + "step": 295936 + }, + { + "epoch": 0.02, + "learning_rate": 4.514648546905291e-05, + "loss": 3.8517, + "step": 296448 + }, + { + "epoch": 0.02, + "learning_rate": 4.513811590034612e-05, + "loss": 3.8432, + "step": 296960 + }, + { + "epoch": 0.02, + "learning_rate": 4.51297299528356e-05, + "loss": 3.8477, + "step": 297472 + }, + { + "epoch": 0.02, + "learning_rate": 4.512134400532508e-05, + "loss": 3.8525, + "step": 297984 + }, + { + "epoch": 0.02, + "learning_rate": 4.511295805781456e-05, + "loss": 3.8517, + "step": 298496 + }, + { + "epoch": 0.02, + "learning_rate": 4.5104588489107766e-05, + "loss": 3.8426, + "step": 299008 + }, + { + "epoch": 0.02, + "learning_rate": 4.5096202541597246e-05, + "loss": 3.8537, + "step": 299520 + }, + { + "epoch": 0.02, + "learning_rate": 4.5087816594086726e-05, + "loss": 3.8497, + "step": 300032 + }, + { + "epoch": 0.02, + "learning_rate": 4.5079430646576206e-05, + "loss": 3.8594, + "step": 300544 + }, + { + "epoch": 0.02, + "learning_rate": 4.507106107786942e-05, + "loss": 3.8437, + "step": 301056 + }, + { + "epoch": 0.02, + "learning_rate": 4.50626751303589e-05, + "loss": 3.8558, + "step": 301568 + }, + { + "epoch": 0.02, + "learning_rate": 4.505428918284838e-05, + "loss": 3.85, + "step": 302080 + }, + { + "epoch": 0.02, + "learning_rate": 4.504590323533786e-05, + "loss": 3.8413, + "step": 302592 + }, + { + "epoch": 0.02, + "learning_rate": 4.503753366663107e-05, + "loss": 3.8464, + "step": 303104 + }, + { + "epoch": 0.02, + "learning_rate": 4.502914771912055e-05, + "loss": 3.8553, + "step": 303616 + }, + { + "epoch": 0.02, + "learning_rate": 4.502076177161003e-05, + "loss": 3.8447, + "step": 304128 + }, + { + "epoch": 0.02, + "learning_rate": 4.501237582409951e-05, + "loss": 3.8318, + "step": 304640 + }, + { + "epoch": 0.02, + "learning_rate": 4.500400625539272e-05, + "loss": 3.8455, + "step": 305152 + }, + { + "epoch": 0.03, + "eval_loss": 3.9177167415618896, + "eval_runtime": 304.5017, + "eval_samples_per_second": 1253.166, + "eval_steps_per_second": 39.162, + "step": 305280 + }, + { + "epoch": 1.0, + "learning_rate": 4.49956203078822e-05, + "loss": 3.8347, + "step": 305664 + }, + { + "epoch": 1.0, + "learning_rate": 4.498723436037168e-05, + "loss": 3.829, + "step": 306176 + }, + { + "epoch": 1.0, + "learning_rate": 4.497884841286116e-05, + "loss": 3.8525, + "step": 306688 + }, + { + "epoch": 1.0, + "learning_rate": 4.4970478844154376e-05, + "loss": 3.852, + "step": 307200 + }, + { + "epoch": 1.0, + "learning_rate": 4.4962092896643856e-05, + "loss": 3.8496, + "step": 307712 + }, + { + "epoch": 1.0, + "learning_rate": 4.4953706949133336e-05, + "loss": 3.8359, + "step": 308224 + }, + { + "epoch": 1.0, + "learning_rate": 4.4945321001622816e-05, + "loss": 3.841, + "step": 308736 + }, + { + "epoch": 1.0, + "learning_rate": 4.4936951432916025e-05, + "loss": 3.8221, + "step": 309248 + }, + { + "epoch": 1.0, + "learning_rate": 4.4928565485405505e-05, + "loss": 3.8475, + "step": 309760 + }, + { + "epoch": 1.0, + "learning_rate": 4.4920179537894985e-05, + "loss": 3.8439, + "step": 310272 + }, + { + "epoch": 1.0, + "learning_rate": 4.4911793590384465e-05, + "loss": 3.8342, + "step": 310784 + }, + { + "epoch": 1.0, + "learning_rate": 4.4903424021677674e-05, + "loss": 3.8437, + "step": 311296 + }, + { + "epoch": 1.0, + "learning_rate": 4.489505445297088e-05, + "loss": 3.8301, + "step": 311808 + }, + { + "epoch": 1.0, + "learning_rate": 4.488666850546037e-05, + "loss": 3.8352, + "step": 312320 + }, + { + "epoch": 1.0, + "learning_rate": 4.487828255794985e-05, + "loss": 3.8254, + "step": 312832 + }, + { + "epoch": 1.0, + "learning_rate": 4.486989661043933e-05, + "loss": 3.8265, + "step": 313344 + }, + { + "epoch": 1.0, + "learning_rate": 4.486151066292881e-05, + "loss": 3.8286, + "step": 313856 + }, + { + "epoch": 1.0, + "learning_rate": 4.485312471541829e-05, + "loss": 3.8306, + "step": 314368 + }, + { + "epoch": 1.0, + "learning_rate": 4.484473876790777e-05, + "loss": 3.8324, + "step": 314880 + }, + { + "epoch": 1.0, + "learning_rate": 4.483635282039725e-05, + "loss": 3.85, + "step": 315392 + }, + { + "epoch": 1.0, + "learning_rate": 4.482798325169046e-05, + "loss": 3.8349, + "step": 315904 + }, + { + "epoch": 1.0, + "learning_rate": 4.481959730417994e-05, + "loss": 3.8407, + "step": 316416 + }, + { + "epoch": 1.0, + "learning_rate": 4.481121135666942e-05, + "loss": 3.8338, + "step": 316928 + }, + { + "epoch": 1.0, + "learning_rate": 4.48028254091589e-05, + "loss": 3.8376, + "step": 317440 + }, + { + "epoch": 1.0, + "learning_rate": 4.479445584045211e-05, + "loss": 3.8247, + "step": 317952 + }, + { + "epoch": 1.0, + "learning_rate": 4.478606989294159e-05, + "loss": 3.8328, + "step": 318464 + }, + { + "epoch": 1.0, + "learning_rate": 4.477768394543107e-05, + "loss": 3.8309, + "step": 318976 + }, + { + "epoch": 1.0, + "learning_rate": 4.476931437672428e-05, + "loss": 3.8226, + "step": 319488 + }, + { + "epoch": 1.0, + "learning_rate": 4.476092842921376e-05, + "loss": 3.8198, + "step": 320000 + }, + { + "epoch": 1.0, + "learning_rate": 4.475254248170324e-05, + "loss": 3.8279, + "step": 320512 + }, + { + "epoch": 1.01, + "learning_rate": 4.474415653419272e-05, + "loss": 3.8276, + "step": 321024 + }, + { + "epoch": 1.01, + "learning_rate": 4.473578696548593e-05, + "loss": 3.8368, + "step": 321536 + }, + { + "epoch": 1.01, + "learning_rate": 4.472740101797541e-05, + "loss": 3.8303, + "step": 322048 + }, + { + "epoch": 1.01, + "learning_rate": 4.471901507046489e-05, + "loss": 3.8322, + "step": 322560 + }, + { + "epoch": 1.01, + "learning_rate": 4.471062912295437e-05, + "loss": 3.821, + "step": 323072 + }, + { + "epoch": 1.01, + "learning_rate": 4.470224317544385e-05, + "loss": 3.8353, + "step": 323584 + }, + { + "epoch": 1.01, + "learning_rate": 4.469385722793333e-05, + "loss": 3.8202, + "step": 324096 + }, + { + "epoch": 1.01, + "learning_rate": 4.468547128042281e-05, + "loss": 3.825, + "step": 324608 + }, + { + "epoch": 1.01, + "learning_rate": 4.467710171171602e-05, + "loss": 3.8095, + "step": 325120 + }, + { + "epoch": 1.01, + "learning_rate": 4.46687157642055e-05, + "loss": 3.8239, + "step": 325632 + }, + { + "epoch": 1.01, + "learning_rate": 4.466032981669499e-05, + "loss": 3.8213, + "step": 326144 + }, + { + "epoch": 1.01, + "learning_rate": 4.465194386918447e-05, + "loss": 3.8336, + "step": 326656 + }, + { + "epoch": 1.01, + "learning_rate": 4.464357430047768e-05, + "loss": 3.8191, + "step": 327168 + }, + { + "epoch": 1.01, + "learning_rate": 4.463518835296716e-05, + "loss": 3.8255, + "step": 327680 + }, + { + "epoch": 1.01, + "learning_rate": 4.462680240545664e-05, + "loss": 3.8246, + "step": 328192 + }, + { + "epoch": 1.01, + "learning_rate": 4.461841645794611e-05, + "loss": 3.8304, + "step": 328704 + }, + { + "epoch": 1.01, + "learning_rate": 4.4610046889239326e-05, + "loss": 3.8242, + "step": 329216 + }, + { + "epoch": 1.01, + "learning_rate": 4.4601660941728806e-05, + "loss": 3.8141, + "step": 329728 + }, + { + "epoch": 1.01, + "learning_rate": 4.4593274994218286e-05, + "loss": 3.803, + "step": 330240 + }, + { + "epoch": 1.01, + "learning_rate": 4.458488904670776e-05, + "loss": 3.8292, + "step": 330752 + }, + { + "epoch": 1.01, + "learning_rate": 4.4576519478000975e-05, + "loss": 3.8279, + "step": 331264 + }, + { + "epoch": 1.01, + "learning_rate": 4.4568133530490455e-05, + "loss": 3.818, + "step": 331776 + }, + { + "epoch": 1.01, + "learning_rate": 4.455974758297994e-05, + "loss": 3.8167, + "step": 332288 + }, + { + "epoch": 1.01, + "learning_rate": 4.4551361635469415e-05, + "loss": 3.8174, + "step": 332800 + }, + { + "epoch": 1.01, + "learning_rate": 4.454299206676263e-05, + "loss": 3.7997, + "step": 333312 + }, + { + "epoch": 1.01, + "learning_rate": 4.453460611925211e-05, + "loss": 3.8161, + "step": 333824 + }, + { + "epoch": 1.01, + "learning_rate": 4.4526220171741584e-05, + "loss": 3.8054, + "step": 334336 + }, + { + "epoch": 1.01, + "learning_rate": 4.4517834224231064e-05, + "loss": 3.8161, + "step": 334848 + }, + { + "epoch": 1.01, + "learning_rate": 4.450946465552428e-05, + "loss": 3.8253, + "step": 335360 + }, + { + "epoch": 1.01, + "learning_rate": 4.450107870801376e-05, + "loss": 3.8112, + "step": 335872 + }, + { + "epoch": 1.01, + "learning_rate": 4.449269276050323e-05, + "loss": 3.8054, + "step": 336384 + }, + { + "epoch": 1.01, + "learning_rate": 4.448430681299271e-05, + "loss": 3.8184, + "step": 336896 + }, + { + "epoch": 1.01, + "learning_rate": 4.447593724428593e-05, + "loss": 3.7982, + "step": 337408 + }, + { + "epoch": 1.01, + "learning_rate": 4.446755129677541e-05, + "loss": 3.8067, + "step": 337920 + }, + { + "epoch": 1.01, + "learning_rate": 4.445916534926489e-05, + "loss": 3.8165, + "step": 338432 + }, + { + "epoch": 1.01, + "learning_rate": 4.445077940175437e-05, + "loss": 3.8129, + "step": 338944 + }, + { + "epoch": 1.01, + "learning_rate": 4.4442409833047584e-05, + "loss": 3.8027, + "step": 339456 + }, + { + "epoch": 1.01, + "learning_rate": 4.443402388553706e-05, + "loss": 3.8042, + "step": 339968 + }, + { + "epoch": 1.01, + "learning_rate": 4.442563793802654e-05, + "loss": 3.8049, + "step": 340480 + }, + { + "epoch": 1.01, + "learning_rate": 4.441725199051602e-05, + "loss": 3.8119, + "step": 340992 + }, + { + "epoch": 1.01, + "learning_rate": 4.440888242180923e-05, + "loss": 3.8126, + "step": 341504 + }, + { + "epoch": 1.01, + "learning_rate": 4.4400496474298706e-05, + "loss": 3.8083, + "step": 342016 + }, + { + "epoch": 1.01, + "learning_rate": 4.4392110526788186e-05, + "loss": 3.8171, + "step": 342528 + }, + { + "epoch": 1.01, + "learning_rate": 4.4383724579277666e-05, + "loss": 3.8182, + "step": 343040 + }, + { + "epoch": 1.01, + "learning_rate": 4.437535501057088e-05, + "loss": 3.8067, + "step": 343552 + }, + { + "epoch": 1.01, + "learning_rate": 4.436696906306036e-05, + "loss": 3.8021, + "step": 344064 + }, + { + "epoch": 1.01, + "learning_rate": 4.435858311554984e-05, + "loss": 3.8216, + "step": 344576 + }, + { + "epoch": 1.01, + "learning_rate": 4.435019716803932e-05, + "loss": 3.8096, + "step": 345088 + }, + { + "epoch": 1.01, + "learning_rate": 4.434182759933253e-05, + "loss": 3.8013, + "step": 345600 + }, + { + "epoch": 1.01, + "learning_rate": 4.433344165182201e-05, + "loss": 3.8057, + "step": 346112 + }, + { + "epoch": 1.01, + "learning_rate": 4.432505570431149e-05, + "loss": 3.8112, + "step": 346624 + }, + { + "epoch": 1.01, + "learning_rate": 4.431666975680097e-05, + "loss": 3.8183, + "step": 347136 + }, + { + "epoch": 1.01, + "learning_rate": 4.430830018809418e-05, + "loss": 3.8004, + "step": 347648 + }, + { + "epoch": 1.01, + "learning_rate": 4.429991424058366e-05, + "loss": 3.7915, + "step": 348160 + }, + { + "epoch": 1.01, + "learning_rate": 4.429152829307314e-05, + "loss": 3.7993, + "step": 348672 + }, + { + "epoch": 1.01, + "learning_rate": 4.428314234556262e-05, + "loss": 3.7989, + "step": 349184 + }, + { + "epoch": 1.01, + "learning_rate": 4.4274772776855836e-05, + "loss": 3.8097, + "step": 349696 + }, + { + "epoch": 1.01, + "learning_rate": 4.4266386829345316e-05, + "loss": 3.8058, + "step": 350208 + }, + { + "epoch": 1.01, + "learning_rate": 4.4258000881834796e-05, + "loss": 3.7923, + "step": 350720 + }, + { + "epoch": 1.02, + "learning_rate": 4.4249614934324276e-05, + "loss": 3.8097, + "step": 351232 + }, + { + "epoch": 1.02, + "learning_rate": 4.4241245365617485e-05, + "loss": 3.8002, + "step": 351744 + }, + { + "epoch": 1.02, + "learning_rate": 4.4232859418106965e-05, + "loss": 3.8028, + "step": 352256 + }, + { + "epoch": 1.02, + "learning_rate": 4.4224473470596445e-05, + "loss": 3.7983, + "step": 352768 + }, + { + "epoch": 1.02, + "learning_rate": 4.4216103901889654e-05, + "loss": 3.8038, + "step": 353280 + }, + { + "epoch": 1.02, + "learning_rate": 4.420773433318287e-05, + "loss": 3.8182, + "step": 353792 + }, + { + "epoch": 1.02, + "learning_rate": 4.419934838567235e-05, + "loss": 3.804, + "step": 354304 + }, + { + "epoch": 1.02, + "learning_rate": 4.419096243816182e-05, + "loss": 3.7824, + "step": 354816 + }, + { + "epoch": 1.02, + "learning_rate": 4.418257649065131e-05, + "loss": 3.8053, + "step": 355328 + }, + { + "epoch": 1.02, + "learning_rate": 4.417419054314079e-05, + "loss": 3.8006, + "step": 355840 + }, + { + "epoch": 1.02, + "learning_rate": 4.416580459563027e-05, + "loss": 3.8071, + "step": 356352 + }, + { + "epoch": 1.02, + "learning_rate": 4.415741864811975e-05, + "loss": 3.8087, + "step": 356864 + }, + { + "epoch": 1.02, + "learning_rate": 4.414903270060923e-05, + "loss": 3.796, + "step": 357376 + }, + { + "epoch": 1.02, + "learning_rate": 4.414066313190244e-05, + "loss": 3.7913, + "step": 357888 + }, + { + "epoch": 1.02, + "learning_rate": 4.413227718439192e-05, + "loss": 3.803, + "step": 358400 + }, + { + "epoch": 1.02, + "learning_rate": 4.41238912368814e-05, + "loss": 3.7834, + "step": 358912 + }, + { + "epoch": 1.02, + "learning_rate": 4.411550528937088e-05, + "loss": 3.7995, + "step": 359424 + }, + { + "epoch": 1.02, + "learning_rate": 4.410713572066409e-05, + "loss": 3.7942, + "step": 359936 + }, + { + "epoch": 1.02, + "learning_rate": 4.409874977315357e-05, + "loss": 3.8019, + "step": 360448 + }, + { + "epoch": 1.02, + "learning_rate": 4.4090380204446776e-05, + "loss": 3.7867, + "step": 360960 + }, + { + "epoch": 1.02, + "learning_rate": 4.408199425693626e-05, + "loss": 3.8041, + "step": 361472 + }, + { + "epoch": 1.02, + "learning_rate": 4.407360830942574e-05, + "loss": 3.7893, + "step": 361984 + }, + { + "epoch": 1.02, + "learning_rate": 4.406522236191522e-05, + "loss": 3.7925, + "step": 362496 + }, + { + "epoch": 1.02, + "learning_rate": 4.40568364144047e-05, + "loss": 3.8045, + "step": 363008 + }, + { + "epoch": 1.02, + "learning_rate": 4.404845046689418e-05, + "loss": 3.7946, + "step": 363520 + }, + { + "epoch": 1.02, + "learning_rate": 4.404006451938366e-05, + "loss": 3.8088, + "step": 364032 + }, + { + "epoch": 1.02, + "learning_rate": 4.403167857187314e-05, + "loss": 3.8025, + "step": 364544 + }, + { + "epoch": 1.02, + "learning_rate": 4.402330900316635e-05, + "loss": 3.7943, + "step": 365056 + }, + { + "epoch": 1.02, + "learning_rate": 4.401492305565583e-05, + "loss": 3.798, + "step": 365568 + }, + { + "epoch": 1.02, + "learning_rate": 4.400653710814531e-05, + "loss": 3.7918, + "step": 366080 + }, + { + "epoch": 1.02, + "learning_rate": 4.399815116063479e-05, + "loss": 3.798, + "step": 366592 + }, + { + "epoch": 1.02, + "learning_rate": 4.398976521312427e-05, + "loss": 3.7894, + "step": 367104 + }, + { + "epoch": 1.02, + "learning_rate": 4.398137926561375e-05, + "loss": 3.8021, + "step": 367616 + }, + { + "epoch": 1.02, + "learning_rate": 4.397299331810323e-05, + "loss": 3.7931, + "step": 368128 + }, + { + "epoch": 1.02, + "learning_rate": 4.396460737059271e-05, + "loss": 3.7907, + "step": 368640 + }, + { + "epoch": 1.02, + "learning_rate": 4.395623780188593e-05, + "loss": 3.7907, + "step": 369152 + }, + { + "epoch": 1.02, + "learning_rate": 4.394785185437541e-05, + "loss": 3.7971, + "step": 369664 + }, + { + "epoch": 1.02, + "learning_rate": 4.393946590686489e-05, + "loss": 3.7887, + "step": 370176 + }, + { + "epoch": 1.02, + "learning_rate": 4.393107995935437e-05, + "loss": 3.799, + "step": 370688 + }, + { + "epoch": 1.02, + "learning_rate": 4.3922710390647576e-05, + "loss": 3.7866, + "step": 371200 + }, + { + "epoch": 1.02, + "learning_rate": 4.3914324443137056e-05, + "loss": 3.8041, + "step": 371712 + }, + { + "epoch": 1.02, + "learning_rate": 4.3905938495626536e-05, + "loss": 3.7955, + "step": 372224 + }, + { + "epoch": 1.02, + "learning_rate": 4.3897552548116016e-05, + "loss": 3.7963, + "step": 372736 + }, + { + "epoch": 1.02, + "learning_rate": 4.3889182979409225e-05, + "loss": 3.7914, + "step": 373248 + }, + { + "epoch": 1.02, + "learning_rate": 4.3880797031898705e-05, + "loss": 3.79, + "step": 373760 + }, + { + "epoch": 1.02, + "learning_rate": 4.3872411084388185e-05, + "loss": 3.8022, + "step": 374272 + }, + { + "epoch": 1.02, + "learning_rate": 4.3864041515681394e-05, + "loss": 3.7929, + "step": 374784 + }, + { + "epoch": 1.02, + "learning_rate": 4.385565556817088e-05, + "loss": 3.7853, + "step": 375296 + }, + { + "epoch": 1.02, + "learning_rate": 4.384726962066036e-05, + "loss": 3.8003, + "step": 375808 + }, + { + "epoch": 1.02, + "learning_rate": 4.383888367314984e-05, + "loss": 3.7965, + "step": 376320 + }, + { + "epoch": 1.02, + "learning_rate": 4.383051410444305e-05, + "loss": 3.8073, + "step": 376832 + }, + { + "epoch": 1.02, + "learning_rate": 4.382212815693253e-05, + "loss": 3.7918, + "step": 377344 + }, + { + "epoch": 1.02, + "learning_rate": 4.381374220942201e-05, + "loss": 3.7994, + "step": 377856 + }, + { + "epoch": 1.02, + "learning_rate": 4.380535626191148e-05, + "loss": 3.7939, + "step": 378368 + }, + { + "epoch": 1.02, + "learning_rate": 4.37969866932047e-05, + "loss": 3.7869, + "step": 378880 + }, + { + "epoch": 1.02, + "learning_rate": 4.378860074569418e-05, + "loss": 3.7945, + "step": 379392 + }, + { + "epoch": 1.02, + "learning_rate": 4.378021479818366e-05, + "loss": 3.7967, + "step": 379904 + }, + { + "epoch": 1.02, + "learning_rate": 4.377182885067313e-05, + "loss": 3.7933, + "step": 380416 + }, + { + "epoch": 1.02, + "learning_rate": 4.376344290316262e-05, + "loss": 3.7784, + "step": 380928 + }, + { + "epoch": 1.02, + "learning_rate": 4.37550569556521e-05, + "loss": 3.7916, + "step": 381440 + }, + { + "epoch": 1.03, + "eval_loss": 3.8919215202331543, + "eval_runtime": 304.4828, + "eval_samples_per_second": 1253.243, + "eval_steps_per_second": 39.165, + "step": 381600 + }, + { + "epoch": 0.0, + "learning_rate": 4.3746687386945315e-05, + "loss": 3.7779, + "step": 381952 + }, + { + "epoch": 0.0, + "learning_rate": 4.373830143943479e-05, + "loss": 3.7753, + "step": 382464 + }, + { + "epoch": 0.0, + "learning_rate": 4.372991549192427e-05, + "loss": 3.7974, + "step": 382976 + }, + { + "epoch": 0.0, + "learning_rate": 4.372152954441375e-05, + "loss": 3.7995, + "step": 383488 + }, + { + "epoch": 0.0, + "learning_rate": 4.371315997570696e-05, + "loss": 3.7928, + "step": 384000 + }, + { + "epoch": 0.0, + "learning_rate": 4.370477402819644e-05, + "loss": 3.7843, + "step": 384512 + }, + { + "epoch": 0.0, + "learning_rate": 4.369638808068592e-05, + "loss": 3.7872, + "step": 385024 + }, + { + "epoch": 0.0, + "learning_rate": 4.36880021331754e-05, + "loss": 3.7725, + "step": 385536 + }, + { + "epoch": 0.0, + "learning_rate": 4.367961618566488e-05, + "loss": 3.7918, + "step": 386048 + }, + { + "epoch": 0.0, + "learning_rate": 4.367123023815436e-05, + "loss": 3.7932, + "step": 386560 + }, + { + "epoch": 0.0, + "learning_rate": 4.366284429064384e-05, + "loss": 3.7807, + "step": 387072 + }, + { + "epoch": 0.0, + "learning_rate": 4.365447472193705e-05, + "loss": 3.7895, + "step": 387584 + }, + { + "epoch": 0.0, + "learning_rate": 4.364608877442653e-05, + "loss": 3.7794, + "step": 388096 + }, + { + "epoch": 0.0, + "learning_rate": 4.363770282691601e-05, + "loss": 3.7849, + "step": 388608 + }, + { + "epoch": 0.0, + "learning_rate": 4.362931687940549e-05, + "loss": 3.7729, + "step": 389120 + }, + { + "epoch": 0.0, + "learning_rate": 4.362093093189497e-05, + "loss": 3.7745, + "step": 389632 + }, + { + "epoch": 0.0, + "learning_rate": 4.361254498438445e-05, + "loss": 3.7754, + "step": 390144 + }, + { + "epoch": 0.0, + "learning_rate": 4.360415903687393e-05, + "loss": 3.7797, + "step": 390656 + }, + { + "epoch": 0.0, + "learning_rate": 4.359578946816714e-05, + "loss": 3.7798, + "step": 391168 + }, + { + "epoch": 0.0, + "learning_rate": 4.358740352065662e-05, + "loss": 3.7938, + "step": 391680 + }, + { + "epoch": 0.0, + "learning_rate": 4.35790175731461e-05, + "loss": 3.7886, + "step": 392192 + }, + { + "epoch": 0.0, + "learning_rate": 4.357063162563558e-05, + "loss": 3.7879, + "step": 392704 + }, + { + "epoch": 0.0, + "learning_rate": 4.356226205692879e-05, + "loss": 3.7802, + "step": 393216 + }, + { + "epoch": 0.0, + "learning_rate": 4.355387610941827e-05, + "loss": 3.7878, + "step": 393728 + }, + { + "epoch": 0.0, + "learning_rate": 4.354549016190776e-05, + "loss": 3.7714, + "step": 394240 + }, + { + "epoch": 0.0, + "learning_rate": 4.353710421439724e-05, + "loss": 3.785, + "step": 394752 + }, + { + "epoch": 0.0, + "learning_rate": 4.3528734645690446e-05, + "loss": 3.778, + "step": 395264 + }, + { + "epoch": 0.0, + "learning_rate": 4.3520348698179926e-05, + "loss": 3.7697, + "step": 395776 + }, + { + "epoch": 0.0, + "learning_rate": 4.3511962750669406e-05, + "loss": 3.771, + "step": 396288 + }, + { + "epoch": 0.0, + "learning_rate": 4.3503576803158886e-05, + "loss": 3.7732, + "step": 396800 + }, + { + "epoch": 0.01, + "learning_rate": 4.3495207234452095e-05, + "loss": 3.7791, + "step": 397312 + }, + { + "epoch": 0.01, + "learning_rate": 4.3486837665745304e-05, + "loss": 3.7834, + "step": 397824 + }, + { + "epoch": 0.01, + "learning_rate": 4.3478451718234784e-05, + "loss": 3.785, + "step": 398336 + }, + { + "epoch": 0.01, + "learning_rate": 4.3470065770724264e-05, + "loss": 3.7827, + "step": 398848 + }, + { + "epoch": 0.01, + "learning_rate": 4.3461679823213744e-05, + "loss": 3.771, + "step": 399360 + }, + { + "epoch": 0.01, + "learning_rate": 4.3453293875703224e-05, + "loss": 3.7822, + "step": 399872 + }, + { + "epoch": 0.01, + "learning_rate": 4.344490792819271e-05, + "loss": 3.7746, + "step": 400384 + }, + { + "epoch": 0.01, + "learning_rate": 4.343652198068219e-05, + "loss": 3.769, + "step": 400896 + }, + { + "epoch": 0.01, + "learning_rate": 4.34281524119754e-05, + "loss": 3.7637, + "step": 401408 + }, + { + "epoch": 0.01, + "learning_rate": 4.341976646446488e-05, + "loss": 3.7709, + "step": 401920 + }, + { + "epoch": 0.01, + "learning_rate": 4.341138051695436e-05, + "loss": 3.774, + "step": 402432 + }, + { + "epoch": 0.01, + "learning_rate": 4.340299456944384e-05, + "loss": 3.7805, + "step": 402944 + }, + { + "epoch": 0.01, + "learning_rate": 4.339460862193332e-05, + "loss": 3.7723, + "step": 403456 + }, + { + "epoch": 0.01, + "learning_rate": 4.338622267442279e-05, + "loss": 3.7737, + "step": 403968 + }, + { + "epoch": 0.01, + "learning_rate": 4.337783672691227e-05, + "loss": 3.7779, + "step": 404480 + }, + { + "epoch": 0.01, + "learning_rate": 4.336945077940175e-05, + "loss": 3.7805, + "step": 404992 + }, + { + "epoch": 0.01, + "learning_rate": 4.336108121069497e-05, + "loss": 3.779, + "step": 405504 + }, + { + "epoch": 0.01, + "learning_rate": 4.335269526318445e-05, + "loss": 3.7643, + "step": 406016 + }, + { + "epoch": 0.01, + "learning_rate": 4.334430931567393e-05, + "loss": 3.7491, + "step": 406528 + }, + { + "epoch": 0.01, + "learning_rate": 4.333592336816341e-05, + "loss": 3.781, + "step": 407040 + }, + { + "epoch": 0.01, + "learning_rate": 4.3327553799456624e-05, + "loss": 3.7793, + "step": 407552 + }, + { + "epoch": 0.01, + "learning_rate": 4.33191678519461e-05, + "loss": 3.7724, + "step": 408064 + }, + { + "epoch": 0.01, + "learning_rate": 4.331078190443558e-05, + "loss": 3.7684, + "step": 408576 + }, + { + "epoch": 0.01, + "learning_rate": 4.330239595692506e-05, + "loss": 3.7648, + "step": 409088 + }, + { + "epoch": 0.01, + "learning_rate": 4.3294026388218266e-05, + "loss": 3.7543, + "step": 409600 + }, + { + "epoch": 0.01, + "learning_rate": 4.3285640440707746e-05, + "loss": 3.7643, + "step": 410112 + }, + { + "epoch": 0.01, + "learning_rate": 4.3277254493197226e-05, + "loss": 3.7563, + "step": 410624 + }, + { + "epoch": 0.01, + "learning_rate": 4.3268868545686706e-05, + "loss": 3.7658, + "step": 411136 + }, + { + "epoch": 0.01, + "learning_rate": 4.3260498976979915e-05, + "loss": 3.7767, + "step": 411648 + }, + { + "epoch": 0.01, + "learning_rate": 4.32521130294694e-05, + "loss": 3.7623, + "step": 412160 + }, + { + "epoch": 0.01, + "learning_rate": 4.324374346076261e-05, + "loss": 3.7555, + "step": 412672 + }, + { + "epoch": 0.01, + "learning_rate": 4.32353575132521e-05, + "loss": 3.7713, + "step": 413184 + }, + { + "epoch": 0.01, + "learning_rate": 4.322697156574157e-05, + "loss": 3.7518, + "step": 413696 + }, + { + "epoch": 0.01, + "learning_rate": 4.321858561823105e-05, + "loss": 3.7548, + "step": 414208 + }, + { + "epoch": 0.01, + "learning_rate": 4.321019967072053e-05, + "loss": 3.7706, + "step": 414720 + }, + { + "epoch": 0.01, + "learning_rate": 4.320181372321001e-05, + "loss": 3.7663, + "step": 415232 + }, + { + "epoch": 0.01, + "learning_rate": 4.319342777569949e-05, + "loss": 3.7547, + "step": 415744 + }, + { + "epoch": 0.01, + "learning_rate": 4.31850582069927e-05, + "loss": 3.7555, + "step": 416256 + }, + { + "epoch": 0.01, + "learning_rate": 4.317667225948218e-05, + "loss": 3.7556, + "step": 416768 + }, + { + "epoch": 0.01, + "learning_rate": 4.316828631197166e-05, + "loss": 3.7611, + "step": 417280 + }, + { + "epoch": 0.01, + "learning_rate": 4.315990036446114e-05, + "loss": 3.7704, + "step": 417792 + }, + { + "epoch": 0.01, + "learning_rate": 4.315151441695062e-05, + "loss": 3.7575, + "step": 418304 + }, + { + "epoch": 0.01, + "learning_rate": 4.31431284694401e-05, + "loss": 3.7686, + "step": 418816 + }, + { + "epoch": 0.01, + "learning_rate": 4.313474252192958e-05, + "loss": 3.7699, + "step": 419328 + }, + { + "epoch": 0.01, + "learning_rate": 4.3126372953222796e-05, + "loss": 3.7673, + "step": 419840 + }, + { + "epoch": 0.01, + "learning_rate": 4.3117987005712276e-05, + "loss": 3.751, + "step": 420352 + }, + { + "epoch": 0.01, + "learning_rate": 4.3109601058201756e-05, + "loss": 3.7707, + "step": 420864 + }, + { + "epoch": 0.01, + "learning_rate": 4.3101215110691236e-05, + "loss": 3.7655, + "step": 421376 + }, + { + "epoch": 0.01, + "learning_rate": 4.3092829163180715e-05, + "loss": 3.7507, + "step": 421888 + }, + { + "epoch": 0.01, + "learning_rate": 4.3084459594473925e-05, + "loss": 3.7656, + "step": 422400 + }, + { + "epoch": 0.01, + "learning_rate": 4.3076073646963405e-05, + "loss": 3.7614, + "step": 422912 + }, + { + "epoch": 0.01, + "learning_rate": 4.3067687699452885e-05, + "loss": 3.7705, + "step": 423424 + }, + { + "epoch": 0.01, + "learning_rate": 4.3059301751942364e-05, + "loss": 3.7562, + "step": 423936 + }, + { + "epoch": 0.01, + "learning_rate": 4.3050932183235574e-05, + "loss": 3.7388, + "step": 424448 + }, + { + "epoch": 0.01, + "learning_rate": 4.3042546235725054e-05, + "loss": 3.7548, + "step": 424960 + }, + { + "epoch": 0.01, + "learning_rate": 4.3034160288214533e-05, + "loss": 3.755, + "step": 425472 + }, + { + "epoch": 0.01, + "learning_rate": 4.302577434070402e-05, + "loss": 3.7603, + "step": 425984 + }, + { + "epoch": 0.01, + "learning_rate": 4.30173883931935e-05, + "loss": 3.7588, + "step": 426496 + }, + { + "epoch": 0.01, + "learning_rate": 4.300900244568298e-05, + "loss": 3.7485, + "step": 427008 + }, + { + "epoch": 0.02, + "learning_rate": 4.300061649817245e-05, + "loss": 3.7589, + "step": 427520 + }, + { + "epoch": 0.02, + "learning_rate": 4.299223055066193e-05, + "loss": 3.7581, + "step": 428032 + }, + { + "epoch": 0.02, + "learning_rate": 4.298387736075888e-05, + "loss": 3.7503, + "step": 428544 + }, + { + "epoch": 0.02, + "learning_rate": 4.297549141324836e-05, + "loss": 3.7568, + "step": 429056 + }, + { + "epoch": 0.02, + "learning_rate": 4.296710546573784e-05, + "loss": 3.7599, + "step": 429568 + }, + { + "epoch": 0.02, + "learning_rate": 4.295871951822732e-05, + "loss": 3.7716, + "step": 430080 + }, + { + "epoch": 0.02, + "learning_rate": 4.29503335707168e-05, + "loss": 3.756, + "step": 430592 + }, + { + "epoch": 0.02, + "learning_rate": 4.294196400201001e-05, + "loss": 3.7409, + "step": 431104 + }, + { + "epoch": 0.02, + "learning_rate": 4.293357805449949e-05, + "loss": 3.7574, + "step": 431616 + }, + { + "epoch": 0.02, + "learning_rate": 4.2925192106988974e-05, + "loss": 3.7528, + "step": 432128 + }, + { + "epoch": 0.02, + "learning_rate": 4.2916806159478454e-05, + "loss": 3.7617, + "step": 432640 + }, + { + "epoch": 0.02, + "learning_rate": 4.290842021196793e-05, + "loss": 3.7647, + "step": 433152 + }, + { + "epoch": 0.02, + "learning_rate": 4.290003426445741e-05, + "loss": 3.7466, + "step": 433664 + }, + { + "epoch": 0.02, + "learning_rate": 4.289164831694689e-05, + "loss": 3.7482, + "step": 434176 + }, + { + "epoch": 0.02, + "learning_rate": 4.288326236943637e-05, + "loss": 3.7596, + "step": 434688 + }, + { + "epoch": 0.02, + "learning_rate": 4.2874892800729576e-05, + "loss": 3.74, + "step": 435200 + }, + { + "epoch": 0.02, + "learning_rate": 4.2866506853219056e-05, + "loss": 3.7559, + "step": 435712 + }, + { + "epoch": 0.02, + "learning_rate": 4.2858120905708536e-05, + "loss": 3.7437, + "step": 436224 + }, + { + "epoch": 0.02, + "learning_rate": 4.2849734958198016e-05, + "loss": 3.7596, + "step": 436736 + }, + { + "epoch": 0.02, + "learning_rate": 4.2841349010687496e-05, + "loss": 3.7424, + "step": 437248 + }, + { + "epoch": 0.02, + "learning_rate": 4.283297944198071e-05, + "loss": 3.7575, + "step": 437760 + }, + { + "epoch": 0.02, + "learning_rate": 4.282459349447019e-05, + "loss": 3.7494, + "step": 438272 + }, + { + "epoch": 0.02, + "learning_rate": 4.281620754695967e-05, + "loss": 3.7441, + "step": 438784 + }, + { + "epoch": 0.02, + "learning_rate": 4.280782159944915e-05, + "loss": 3.7587, + "step": 439296 + }, + { + "epoch": 0.02, + "learning_rate": 4.279943565193863e-05, + "loss": 3.7493, + "step": 439808 + }, + { + "epoch": 0.02, + "learning_rate": 4.279104970442811e-05, + "loss": 3.766, + "step": 440320 + }, + { + "epoch": 0.02, + "learning_rate": 4.278268013572132e-05, + "loss": 3.7612, + "step": 440832 + }, + { + "epoch": 0.02, + "learning_rate": 4.27742941882108e-05, + "loss": 3.7474, + "step": 441344 + }, + { + "epoch": 0.02, + "learning_rate": 4.276590824070028e-05, + "loss": 3.7556, + "step": 441856 + }, + { + "epoch": 0.02, + "learning_rate": 4.275752229318976e-05, + "loss": 3.7456, + "step": 442368 + }, + { + "epoch": 0.02, + "learning_rate": 4.274913634567924e-05, + "loss": 3.7524, + "step": 442880 + }, + { + "epoch": 0.02, + "learning_rate": 4.274075039816872e-05, + "loss": 3.7485, + "step": 443392 + }, + { + "epoch": 0.02, + "learning_rate": 4.27323644506582e-05, + "loss": 3.7549, + "step": 443904 + }, + { + "epoch": 0.02, + "learning_rate": 4.272399488195141e-05, + "loss": 3.749, + "step": 444416 + }, + { + "epoch": 0.02, + "learning_rate": 4.2715608934440896e-05, + "loss": 3.7462, + "step": 444928 + }, + { + "epoch": 0.02, + "learning_rate": 4.2707222986930376e-05, + "loss": 3.7516, + "step": 445440 + }, + { + "epoch": 0.02, + "learning_rate": 4.2698837039419856e-05, + "loss": 3.7523, + "step": 445952 + }, + { + "epoch": 0.02, + "learning_rate": 4.2690451091909336e-05, + "loss": 3.7448, + "step": 446464 + }, + { + "epoch": 0.02, + "learning_rate": 4.2682065144398816e-05, + "loss": 3.7536, + "step": 446976 + }, + { + "epoch": 0.02, + "learning_rate": 4.267367919688829e-05, + "loss": 3.7411, + "step": 447488 + }, + { + "epoch": 0.02, + "learning_rate": 4.2665309628181505e-05, + "loss": 3.766, + "step": 448000 + }, + { + "epoch": 0.02, + "learning_rate": 4.2656923680670985e-05, + "loss": 3.7458, + "step": 448512 + }, + { + "epoch": 0.02, + "learning_rate": 4.2648537733160465e-05, + "loss": 3.7577, + "step": 449024 + }, + { + "epoch": 0.02, + "learning_rate": 4.264015178564994e-05, + "loss": 3.7452, + "step": 449536 + }, + { + "epoch": 0.02, + "learning_rate": 4.263176583813942e-05, + "loss": 3.7496, + "step": 450048 + }, + { + "epoch": 0.02, + "learning_rate": 4.26233798906289e-05, + "loss": 3.7569, + "step": 450560 + }, + { + "epoch": 0.02, + "learning_rate": 4.261499394311838e-05, + "loss": 3.7491, + "step": 451072 + }, + { + "epoch": 0.02, + "learning_rate": 4.2606624374411594e-05, + "loss": 3.7444, + "step": 451584 + }, + { + "epoch": 0.02, + "learning_rate": 4.2598238426901074e-05, + "loss": 3.7589, + "step": 452096 + }, + { + "epoch": 0.02, + "learning_rate": 4.2589852479390554e-05, + "loss": 3.7533, + "step": 452608 + }, + { + "epoch": 0.02, + "learning_rate": 4.2581466531880034e-05, + "loss": 3.7618, + "step": 453120 + }, + { + "epoch": 0.02, + "learning_rate": 4.2573080584369514e-05, + "loss": 3.7449, + "step": 453632 + }, + { + "epoch": 0.02, + "learning_rate": 4.2564694636858994e-05, + "loss": 3.7568, + "step": 454144 + }, + { + "epoch": 0.02, + "learning_rate": 4.25563250681522e-05, + "loss": 3.7503, + "step": 454656 + }, + { + "epoch": 0.02, + "learning_rate": 4.254793912064168e-05, + "loss": 3.7463, + "step": 455168 + }, + { + "epoch": 0.02, + "learning_rate": 4.253955317313116e-05, + "loss": 3.7529, + "step": 455680 + }, + { + "epoch": 0.02, + "learning_rate": 4.253116722562064e-05, + "loss": 3.7513, + "step": 456192 + }, + { + "epoch": 0.02, + "learning_rate": 4.252279765691385e-05, + "loss": 3.7569, + "step": 456704 + }, + { + "epoch": 0.02, + "learning_rate": 4.251441170940333e-05, + "loss": 3.7279, + "step": 457216 + }, + { + "epoch": 0.02, + "learning_rate": 4.250602576189281e-05, + "loss": 3.7537, + "step": 457728 + }, + { + "epoch": 0.03, + "eval_loss": 3.875577926635742, + "eval_runtime": 306.0633, + "eval_samples_per_second": 1246.771, + "eval_steps_per_second": 38.963, + "step": 457920 + }, + { + "epoch": 1.0, + "learning_rate": 4.24976398143823e-05, + "loss": 3.743, + "step": 458240 + }, + { + "epoch": 1.0, + "learning_rate": 4.248925386687178e-05, + "loss": 3.7344, + "step": 458752 + }, + { + "epoch": 1.0, + "learning_rate": 4.248086791936126e-05, + "loss": 3.7517, + "step": 459264 + }, + { + "epoch": 1.0, + "learning_rate": 4.247248197185074e-05, + "loss": 3.7597, + "step": 459776 + }, + { + "epoch": 1.0, + "learning_rate": 4.246411240314395e-05, + "loss": 3.7506, + "step": 460288 + }, + { + "epoch": 1.0, + "learning_rate": 4.245572645563343e-05, + "loss": 3.7393, + "step": 460800 + }, + { + "epoch": 1.0, + "learning_rate": 4.244734050812291e-05, + "loss": 3.7457, + "step": 461312 + }, + { + "epoch": 1.0, + "learning_rate": 4.243895456061239e-05, + "loss": 3.7323, + "step": 461824 + }, + { + "epoch": 1.0, + "learning_rate": 4.2430584991905596e-05, + "loss": 3.746, + "step": 462336 + }, + { + "epoch": 1.0, + "learning_rate": 4.2422199044395076e-05, + "loss": 3.7471, + "step": 462848 + }, + { + "epoch": 1.0, + "learning_rate": 4.2413813096884556e-05, + "loss": 3.7453, + "step": 463360 + }, + { + "epoch": 1.0, + "learning_rate": 4.2405427149374036e-05, + "loss": 3.7442, + "step": 463872 + }, + { + "epoch": 1.0, + "learning_rate": 4.2397041201863516e-05, + "loss": 3.7407, + "step": 464384 + }, + { + "epoch": 1.0, + "learning_rate": 4.238867163315673e-05, + "loss": 3.7392, + "step": 464896 + }, + { + "epoch": 1.0, + "learning_rate": 4.238028568564621e-05, + "loss": 3.7311, + "step": 465408 + }, + { + "epoch": 1.0, + "learning_rate": 4.237189973813569e-05, + "loss": 3.7292, + "step": 465920 + }, + { + "epoch": 1.0, + "learning_rate": 4.236351379062517e-05, + "loss": 3.7354, + "step": 466432 + }, + { + "epoch": 1.0, + "learning_rate": 4.235512784311465e-05, + "loss": 3.7366, + "step": 466944 + }, + { + "epoch": 1.0, + "learning_rate": 4.2346741895604125e-05, + "loss": 3.7383, + "step": 467456 + }, + { + "epoch": 1.0, + "learning_rate": 4.233837232689734e-05, + "loss": 3.7522, + "step": 467968 + }, + { + "epoch": 1.0, + "learning_rate": 4.232998637938682e-05, + "loss": 3.7511, + "step": 468480 + }, + { + "epoch": 1.0, + "learning_rate": 4.23216004318763e-05, + "loss": 3.7434, + "step": 468992 + }, + { + "epoch": 1.0, + "learning_rate": 4.2313214484365774e-05, + "loss": 3.7386, + "step": 469504 + }, + { + "epoch": 1.0, + "learning_rate": 4.2304828536855254e-05, + "loss": 3.7459, + "step": 470016 + }, + { + "epoch": 1.0, + "learning_rate": 4.2296442589344734e-05, + "loss": 3.7303, + "step": 470528 + }, + { + "epoch": 1.0, + "learning_rate": 4.228807302063795e-05, + "loss": 3.7466, + "step": 471040 + }, + { + "epoch": 1.0, + "learning_rate": 4.227968707312743e-05, + "loss": 3.7379, + "step": 471552 + }, + { + "epoch": 1.0, + "learning_rate": 4.227130112561691e-05, + "loss": 3.7261, + "step": 472064 + }, + { + "epoch": 1.0, + "learning_rate": 4.226291517810639e-05, + "loss": 3.7324, + "step": 472576 + }, + { + "epoch": 1.0, + "learning_rate": 4.22545456093996e-05, + "loss": 3.7316, + "step": 473088 + }, + { + "epoch": 1.01, + "learning_rate": 4.224615966188908e-05, + "loss": 3.7372, + "step": 473600 + }, + { + "epoch": 1.01, + "learning_rate": 4.223777371437856e-05, + "loss": 3.7454, + "step": 474112 + }, + { + "epoch": 1.01, + "learning_rate": 4.222938776686804e-05, + "loss": 3.7408, + "step": 474624 + }, + { + "epoch": 1.01, + "learning_rate": 4.222101819816125e-05, + "loss": 3.741, + "step": 475136 + }, + { + "epoch": 1.01, + "learning_rate": 4.221263225065073e-05, + "loss": 3.7278, + "step": 475648 + }, + { + "epoch": 1.01, + "learning_rate": 4.220424630314021e-05, + "loss": 3.7462, + "step": 476160 + }, + { + "epoch": 1.01, + "learning_rate": 4.219586035562969e-05, + "loss": 3.7318, + "step": 476672 + }, + { + "epoch": 1.01, + "learning_rate": 4.2187474408119174e-05, + "loss": 3.7238, + "step": 477184 + }, + { + "epoch": 1.01, + "learning_rate": 4.217910483941238e-05, + "loss": 3.7302, + "step": 477696 + }, + { + "epoch": 1.01, + "learning_rate": 4.217071889190186e-05, + "loss": 3.7249, + "step": 478208 + }, + { + "epoch": 1.01, + "learning_rate": 4.216233294439134e-05, + "loss": 3.7346, + "step": 478720 + }, + { + "epoch": 1.01, + "learning_rate": 4.215394699688082e-05, + "loss": 3.7445, + "step": 479232 + }, + { + "epoch": 1.01, + "learning_rate": 4.21455610493703e-05, + "loss": 3.73, + "step": 479744 + }, + { + "epoch": 1.01, + "learning_rate": 4.213717510185978e-05, + "loss": 3.732, + "step": 480256 + }, + { + "epoch": 1.01, + "learning_rate": 4.212880553315299e-05, + "loss": 3.7399, + "step": 480768 + }, + { + "epoch": 1.01, + "learning_rate": 4.212041958564247e-05, + "loss": 3.7422, + "step": 481280 + }, + { + "epoch": 1.01, + "learning_rate": 4.211203363813195e-05, + "loss": 3.739, + "step": 481792 + }, + { + "epoch": 1.01, + "learning_rate": 4.210364769062143e-05, + "loss": 3.7263, + "step": 482304 + }, + { + "epoch": 1.01, + "learning_rate": 4.209527812191464e-05, + "loss": 3.7019, + "step": 482816 + }, + { + "epoch": 1.01, + "learning_rate": 4.208689217440413e-05, + "loss": 3.7448, + "step": 483328 + }, + { + "epoch": 1.01, + "learning_rate": 4.207850622689361e-05, + "loss": 3.7383, + "step": 483840 + }, + { + "epoch": 1.01, + "learning_rate": 4.207012027938309e-05, + "loss": 3.7344, + "step": 484352 + }, + { + "epoch": 1.01, + "learning_rate": 4.20617507106763e-05, + "loss": 3.7272, + "step": 484864 + }, + { + "epoch": 1.01, + "learning_rate": 4.205336476316578e-05, + "loss": 3.7252, + "step": 485376 + }, + { + "epoch": 1.01, + "learning_rate": 4.204497881565526e-05, + "loss": 3.7144, + "step": 485888 + }, + { + "epoch": 1.01, + "learning_rate": 4.203659286814474e-05, + "loss": 3.7243, + "step": 486400 + }, + { + "epoch": 1.01, + "learning_rate": 4.202820692063422e-05, + "loss": 3.7193, + "step": 486912 + }, + { + "epoch": 1.01, + "learning_rate": 4.2019837351927426e-05, + "loss": 3.728, + "step": 487424 + }, + { + "epoch": 1.01, + "learning_rate": 4.2011451404416906e-05, + "loss": 3.7382, + "step": 487936 + }, + { + "epoch": 1.01, + "learning_rate": 4.2003065456906386e-05, + "loss": 3.7194, + "step": 488448 + }, + { + "epoch": 1.01, + "learning_rate": 4.1994679509395866e-05, + "loss": 3.7166, + "step": 488960 + }, + { + "epoch": 1.01, + "learning_rate": 4.1986293561885346e-05, + "loss": 3.7333, + "step": 489472 + }, + { + "epoch": 1.01, + "learning_rate": 4.1977907614374826e-05, + "loss": 3.7168, + "step": 489984 + }, + { + "epoch": 1.01, + "learning_rate": 4.196953804566804e-05, + "loss": 3.7154, + "step": 490496 + }, + { + "epoch": 1.01, + "learning_rate": 4.196115209815752e-05, + "loss": 3.7293, + "step": 491008 + }, + { + "epoch": 1.01, + "learning_rate": 4.1952766150647e-05, + "loss": 3.7255, + "step": 491520 + }, + { + "epoch": 1.01, + "learning_rate": 4.194438020313648e-05, + "loss": 3.7162, + "step": 492032 + }, + { + "epoch": 1.01, + "learning_rate": 4.193601063442969e-05, + "loss": 3.7134, + "step": 492544 + }, + { + "epoch": 1.01, + "learning_rate": 4.192762468691917e-05, + "loss": 3.7169, + "step": 493056 + }, + { + "epoch": 1.01, + "learning_rate": 4.191923873940865e-05, + "loss": 3.7266, + "step": 493568 + }, + { + "epoch": 1.01, + "learning_rate": 4.191085279189813e-05, + "loss": 3.7265, + "step": 494080 + }, + { + "epoch": 1.01, + "learning_rate": 4.1902466844387603e-05, + "loss": 3.7213, + "step": 494592 + }, + { + "epoch": 1.01, + "learning_rate": 4.189409727568082e-05, + "loss": 3.7249, + "step": 495104 + }, + { + "epoch": 1.01, + "learning_rate": 4.18857113281703e-05, + "loss": 3.7308, + "step": 495616 + }, + { + "epoch": 1.01, + "learning_rate": 4.187732538065978e-05, + "loss": 3.7293, + "step": 496128 + }, + { + "epoch": 1.01, + "learning_rate": 4.186893943314926e-05, + "loss": 3.7166, + "step": 496640 + }, + { + "epoch": 1.01, + "learning_rate": 4.186055348563874e-05, + "loss": 3.7309, + "step": 497152 + }, + { + "epoch": 1.01, + "learning_rate": 4.185216753812822e-05, + "loss": 3.7297, + "step": 497664 + }, + { + "epoch": 1.01, + "learning_rate": 4.18437815906177e-05, + "loss": 3.7095, + "step": 498176 + }, + { + "epoch": 1.01, + "learning_rate": 4.183541202191091e-05, + "loss": 3.7272, + "step": 498688 + }, + { + "epoch": 1.01, + "learning_rate": 4.182702607440039e-05, + "loss": 3.7246, + "step": 499200 + }, + { + "epoch": 1.01, + "learning_rate": 4.181864012688987e-05, + "loss": 3.7283, + "step": 499712 + }, + { + "epoch": 1.01, + "learning_rate": 4.181025417937935e-05, + "loss": 3.7195, + "step": 500224 + }, + { + "epoch": 1.01, + "learning_rate": 4.180188461067256e-05, + "loss": 3.7036, + "step": 500736 + }, + { + "epoch": 1.01, + "learning_rate": 4.179349866316204e-05, + "loss": 3.7137, + "step": 501248 + }, + { + "epoch": 1.01, + "learning_rate": 4.178511271565152e-05, + "loss": 3.7171, + "step": 501760 + }, + { + "epoch": 1.01, + "learning_rate": 4.1776726768141e-05, + "loss": 3.7266, + "step": 502272 + }, + { + "epoch": 1.01, + "learning_rate": 4.176835719943421e-05, + "loss": 3.7169, + "step": 502784 + }, + { + "epoch": 1.01, + "learning_rate": 4.175997125192369e-05, + "loss": 3.7127, + "step": 503296 + }, + { + "epoch": 1.02, + "learning_rate": 4.175158530441317e-05, + "loss": 3.7214, + "step": 503808 + }, + { + "epoch": 1.02, + "learning_rate": 4.174319935690265e-05, + "loss": 3.7236, + "step": 504320 + }, + { + "epoch": 1.02, + "learning_rate": 4.173481340939213e-05, + "loss": 3.7091, + "step": 504832 + }, + { + "epoch": 1.02, + "learning_rate": 4.172644384068534e-05, + "loss": 3.7172, + "step": 505344 + }, + { + "epoch": 1.02, + "learning_rate": 4.171805789317482e-05, + "loss": 3.7248, + "step": 505856 + }, + { + "epoch": 1.02, + "learning_rate": 4.17096719456643e-05, + "loss": 3.7312, + "step": 506368 + }, + { + "epoch": 1.02, + "learning_rate": 4.170128599815378e-05, + "loss": 3.7182, + "step": 506880 + }, + { + "epoch": 1.02, + "learning_rate": 4.169290005064326e-05, + "loss": 3.7045, + "step": 507392 + }, + { + "epoch": 1.02, + "learning_rate": 4.168453048193647e-05, + "loss": 3.7178, + "step": 507904 + }, + { + "epoch": 1.02, + "learning_rate": 4.167614453442595e-05, + "loss": 3.7121, + "step": 508416 + }, + { + "epoch": 1.02, + "learning_rate": 4.166775858691544e-05, + "loss": 3.7261, + "step": 508928 + }, + { + "epoch": 1.02, + "learning_rate": 4.165937263940492e-05, + "loss": 3.7278, + "step": 509440 + }, + { + "epoch": 1.02, + "learning_rate": 4.16509866918944e-05, + "loss": 3.7131, + "step": 509952 + }, + { + "epoch": 1.02, + "learning_rate": 4.1642617123187606e-05, + "loss": 3.7091, + "step": 510464 + }, + { + "epoch": 1.02, + "learning_rate": 4.1634231175677086e-05, + "loss": 3.7189, + "step": 510976 + }, + { + "epoch": 1.02, + "learning_rate": 4.1625845228166566e-05, + "loss": 3.7029, + "step": 511488 + }, + { + "epoch": 1.02, + "learning_rate": 4.1617459280656046e-05, + "loss": 3.721, + "step": 512000 + }, + { + "epoch": 1.02, + "learning_rate": 4.1609089711949255e-05, + "loss": 3.7032, + "step": 512512 + }, + { + "epoch": 1.02, + "learning_rate": 4.1600703764438735e-05, + "loss": 3.7233, + "step": 513024 + }, + { + "epoch": 1.02, + "learning_rate": 4.1592317816928215e-05, + "loss": 3.7086, + "step": 513536 + }, + { + "epoch": 1.02, + "learning_rate": 4.1583931869417695e-05, + "loss": 3.7184, + "step": 514048 + }, + { + "epoch": 1.02, + "learning_rate": 4.1575545921907175e-05, + "loss": 3.7127, + "step": 514560 + }, + { + "epoch": 1.02, + "learning_rate": 4.156717635320039e-05, + "loss": 3.7124, + "step": 515072 + }, + { + "epoch": 1.02, + "learning_rate": 4.155879040568987e-05, + "loss": 3.7205, + "step": 515584 + }, + { + "epoch": 1.02, + "learning_rate": 4.155040445817935e-05, + "loss": 3.7096, + "step": 516096 + }, + { + "epoch": 1.02, + "learning_rate": 4.154201851066883e-05, + "loss": 3.7288, + "step": 516608 + }, + { + "epoch": 1.02, + "learning_rate": 4.153364894196204e-05, + "loss": 3.7211, + "step": 517120 + }, + { + "epoch": 1.02, + "learning_rate": 4.152526299445152e-05, + "loss": 3.7141, + "step": 517632 + }, + { + "epoch": 1.02, + "learning_rate": 4.1516877046941e-05, + "loss": 3.719, + "step": 518144 + }, + { + "epoch": 1.02, + "learning_rate": 4.150849109943048e-05, + "loss": 3.7094, + "step": 518656 + }, + { + "epoch": 1.02, + "learning_rate": 4.150010515191996e-05, + "loss": 3.715, + "step": 519168 + }, + { + "epoch": 1.02, + "learning_rate": 4.149173558321317e-05, + "loss": 3.7143, + "step": 519680 + }, + { + "epoch": 1.02, + "learning_rate": 4.148334963570265e-05, + "loss": 3.7125, + "step": 520192 + }, + { + "epoch": 1.02, + "learning_rate": 4.147496368819213e-05, + "loss": 3.7156, + "step": 520704 + }, + { + "epoch": 1.02, + "learning_rate": 4.146657774068161e-05, + "loss": 3.7141, + "step": 521216 + }, + { + "epoch": 1.02, + "learning_rate": 4.145819179317109e-05, + "loss": 3.7142, + "step": 521728 + }, + { + "epoch": 1.02, + "learning_rate": 4.144980584566057e-05, + "loss": 3.7131, + "step": 522240 + }, + { + "epoch": 1.02, + "learning_rate": 4.144141989815005e-05, + "loss": 3.7113, + "step": 522752 + }, + { + "epoch": 1.02, + "learning_rate": 4.1433050329443265e-05, + "loss": 3.7155, + "step": 523264 + }, + { + "epoch": 1.02, + "learning_rate": 4.1424664381932744e-05, + "loss": 3.7018, + "step": 523776 + }, + { + "epoch": 1.02, + "learning_rate": 4.141627843442222e-05, + "loss": 3.7281, + "step": 524288 + }, + { + "epoch": 1.02, + "learning_rate": 4.14078924869117e-05, + "loss": 3.714, + "step": 524800 + }, + { + "epoch": 1.02, + "learning_rate": 4.139950653940118e-05, + "loss": 3.7224, + "step": 525312 + }, + { + "epoch": 1.02, + "learning_rate": 4.139113697069439e-05, + "loss": 3.7026, + "step": 525824 + }, + { + "epoch": 1.02, + "learning_rate": 4.138275102318387e-05, + "loss": 3.7155, + "step": 526336 + }, + { + "epoch": 1.02, + "learning_rate": 4.1374365075673347e-05, + "loss": 3.7186, + "step": 526848 + }, + { + "epoch": 1.02, + "learning_rate": 4.1365979128162827e-05, + "loss": 3.7161, + "step": 527360 + }, + { + "epoch": 1.02, + "learning_rate": 4.135760955945604e-05, + "loss": 3.7031, + "step": 527872 + }, + { + "epoch": 1.02, + "learning_rate": 4.134922361194552e-05, + "loss": 3.7222, + "step": 528384 + }, + { + "epoch": 1.02, + "learning_rate": 4.1340837664435e-05, + "loss": 3.7159, + "step": 528896 + }, + { + "epoch": 1.02, + "learning_rate": 4.133245171692448e-05, + "loss": 3.7285, + "step": 529408 + }, + { + "epoch": 1.02, + "learning_rate": 4.132406576941396e-05, + "loss": 3.7067, + "step": 529920 + }, + { + "epoch": 1.02, + "learning_rate": 4.131569620070717e-05, + "loss": 3.7209, + "step": 530432 + }, + { + "epoch": 1.02, + "learning_rate": 4.130731025319665e-05, + "loss": 3.7138, + "step": 530944 + }, + { + "epoch": 1.02, + "learning_rate": 4.129892430568613e-05, + "loss": 3.7113, + "step": 531456 + }, + { + "epoch": 1.02, + "learning_rate": 4.129053835817561e-05, + "loss": 3.7148, + "step": 531968 + }, + { + "epoch": 1.02, + "learning_rate": 4.128215241066509e-05, + "loss": 3.7165, + "step": 532480 + }, + { + "epoch": 1.02, + "learning_rate": 4.12737828419583e-05, + "loss": 3.7171, + "step": 532992 + }, + { + "epoch": 1.02, + "learning_rate": 4.126539689444778e-05, + "loss": 3.6983, + "step": 533504 + }, + { + "epoch": 1.02, + "learning_rate": 4.125701094693726e-05, + "loss": 3.717, + "step": 534016 + }, + { + "epoch": 1.03, + "eval_loss": 3.8661532402038574, + "eval_runtime": 305.2202, + "eval_samples_per_second": 1250.215, + "eval_steps_per_second": 39.07, + "step": 534240 + }, + { + "epoch": 0.0, + "learning_rate": 4.124862499942675e-05, + "loss": 3.7078, + "step": 534528 + }, + { + "epoch": 0.0, + "learning_rate": 4.124023905191623e-05, + "loss": 3.6982, + "step": 535040 + }, + { + "epoch": 0.0, + "learning_rate": 4.123185310440571e-05, + "loss": 3.7146, + "step": 535552 + }, + { + "epoch": 0.0, + "learning_rate": 4.122346715689519e-05, + "loss": 3.7228, + "step": 536064 + }, + { + "epoch": 0.0, + "learning_rate": 4.121508120938467e-05, + "loss": 3.7156, + "step": 536576 + }, + { + "epoch": 0.0, + "learning_rate": 4.1206711640677876e-05, + "loss": 3.7061, + "step": 537088 + }, + { + "epoch": 0.0, + "learning_rate": 4.1198325693167356e-05, + "loss": 3.7143, + "step": 537600 + }, + { + "epoch": 0.0, + "learning_rate": 4.1189939745656836e-05, + "loss": 3.6986, + "step": 538112 + }, + { + "epoch": 0.0, + "learning_rate": 4.1181553798146316e-05, + "loss": 3.7115, + "step": 538624 + }, + { + "epoch": 0.0, + "learning_rate": 4.1173184229439525e-05, + "loss": 3.711, + "step": 539136 + }, + { + "epoch": 0.0, + "learning_rate": 4.1164798281929005e-05, + "loss": 3.7085, + "step": 539648 + }, + { + "epoch": 0.0, + "learning_rate": 4.1156412334418485e-05, + "loss": 3.7131, + "step": 540160 + }, + { + "epoch": 0.0, + "learning_rate": 4.1148026386907965e-05, + "loss": 3.7006, + "step": 540672 + }, + { + "epoch": 0.0, + "learning_rate": 4.113965681820118e-05, + "loss": 3.7034, + "step": 541184 + }, + { + "epoch": 0.0, + "learning_rate": 4.113127087069066e-05, + "loss": 3.701, + "step": 541696 + }, + { + "epoch": 0.0, + "learning_rate": 4.112288492318014e-05, + "loss": 3.6922, + "step": 542208 + }, + { + "epoch": 0.0, + "learning_rate": 4.111449897566962e-05, + "loss": 3.701, + "step": 542720 + }, + { + "epoch": 0.0, + "learning_rate": 4.110612940696283e-05, + "loss": 3.7026, + "step": 543232 + }, + { + "epoch": 0.0, + "learning_rate": 4.109774345945231e-05, + "loss": 3.7049, + "step": 543744 + }, + { + "epoch": 0.0, + "learning_rate": 4.108935751194179e-05, + "loss": 3.7145, + "step": 544256 + }, + { + "epoch": 0.0, + "learning_rate": 4.108097156443127e-05, + "loss": 3.7167, + "step": 544768 + }, + { + "epoch": 0.0, + "learning_rate": 4.107260199572448e-05, + "loss": 3.7131, + "step": 545280 + }, + { + "epoch": 0.0, + "learning_rate": 4.106421604821396e-05, + "loss": 3.7019, + "step": 545792 + }, + { + "epoch": 0.0, + "learning_rate": 4.105583010070344e-05, + "loss": 3.7089, + "step": 546304 + }, + { + "epoch": 0.0, + "learning_rate": 4.104744415319292e-05, + "loss": 3.6979, + "step": 546816 + }, + { + "epoch": 0.0, + "learning_rate": 4.1039074584486134e-05, + "loss": 3.7098, + "step": 547328 + }, + { + "epoch": 0.0, + "learning_rate": 4.1030688636975614e-05, + "loss": 3.7057, + "step": 547840 + }, + { + "epoch": 0.0, + "learning_rate": 4.1022302689465094e-05, + "loss": 3.6931, + "step": 548352 + }, + { + "epoch": 0.0, + "learning_rate": 4.1013916741954574e-05, + "loss": 3.7015, + "step": 548864 + }, + { + "epoch": 0.0, + "learning_rate": 4.100554717324778e-05, + "loss": 3.6936, + "step": 549376 + }, + { + "epoch": 0.01, + "learning_rate": 4.099716122573726e-05, + "loss": 3.7009, + "step": 549888 + }, + { + "epoch": 0.01, + "learning_rate": 4.098877527822674e-05, + "loss": 3.7133, + "step": 550400 + }, + { + "epoch": 0.01, + "learning_rate": 4.098038933071622e-05, + "loss": 3.7045, + "step": 550912 + }, + { + "epoch": 0.01, + "learning_rate": 4.097201976200943e-05, + "loss": 3.7074, + "step": 551424 + }, + { + "epoch": 0.01, + "learning_rate": 4.096363381449891e-05, + "loss": 3.6921, + "step": 551936 + }, + { + "epoch": 0.01, + "learning_rate": 4.095524786698839e-05, + "loss": 3.7138, + "step": 552448 + }, + { + "epoch": 0.01, + "learning_rate": 4.094686191947787e-05, + "loss": 3.7015, + "step": 552960 + }, + { + "epoch": 0.01, + "learning_rate": 4.093847597196735e-05, + "loss": 3.6903, + "step": 553472 + }, + { + "epoch": 0.01, + "learning_rate": 4.093010640326057e-05, + "loss": 3.6989, + "step": 553984 + }, + { + "epoch": 0.01, + "learning_rate": 4.092172045575005e-05, + "loss": 3.6896, + "step": 554496 + }, + { + "epoch": 0.01, + "learning_rate": 4.091333450823952e-05, + "loss": 3.6989, + "step": 555008 + }, + { + "epoch": 0.01, + "learning_rate": 4.0904948560729e-05, + "loss": 3.705, + "step": 555520 + }, + { + "epoch": 0.01, + "learning_rate": 4.089656261321848e-05, + "loss": 3.6998, + "step": 556032 + }, + { + "epoch": 0.01, + "learning_rate": 4.08881930445117e-05, + "loss": 3.6955, + "step": 556544 + }, + { + "epoch": 0.01, + "learning_rate": 4.087980709700117e-05, + "loss": 3.707, + "step": 557056 + }, + { + "epoch": 0.01, + "learning_rate": 4.087142114949065e-05, + "loss": 3.7056, + "step": 557568 + }, + { + "epoch": 0.01, + "learning_rate": 4.086303520198013e-05, + "loss": 3.707, + "step": 558080 + }, + { + "epoch": 0.01, + "learning_rate": 4.0854665633273346e-05, + "loss": 3.6906, + "step": 558592 + }, + { + "epoch": 0.01, + "learning_rate": 4.0846279685762826e-05, + "loss": 3.6688, + "step": 559104 + }, + { + "epoch": 0.01, + "learning_rate": 4.0837893738252306e-05, + "loss": 3.7144, + "step": 559616 + }, + { + "epoch": 0.01, + "learning_rate": 4.0829507790741786e-05, + "loss": 3.7, + "step": 560128 + }, + { + "epoch": 0.01, + "learning_rate": 4.0821138222034995e-05, + "loss": 3.7025, + "step": 560640 + }, + { + "epoch": 0.01, + "learning_rate": 4.0812752274524475e-05, + "loss": 3.6958, + "step": 561152 + }, + { + "epoch": 0.01, + "learning_rate": 4.0804366327013955e-05, + "loss": 3.6865, + "step": 561664 + }, + { + "epoch": 0.01, + "learning_rate": 4.0795980379503435e-05, + "loss": 3.6845, + "step": 562176 + }, + { + "epoch": 0.01, + "learning_rate": 4.0787610810796644e-05, + "loss": 3.691, + "step": 562688 + }, + { + "epoch": 0.01, + "learning_rate": 4.0779224863286124e-05, + "loss": 3.6839, + "step": 563200 + }, + { + "epoch": 0.01, + "learning_rate": 4.0770838915775604e-05, + "loss": 3.6915, + "step": 563712 + }, + { + "epoch": 0.01, + "learning_rate": 4.0762452968265083e-05, + "loss": 3.709, + "step": 564224 + }, + { + "epoch": 0.01, + "learning_rate": 4.07540833995583e-05, + "loss": 3.6858, + "step": 564736 + }, + { + "epoch": 0.01, + "learning_rate": 4.074569745204778e-05, + "loss": 3.6819, + "step": 565248 + }, + { + "epoch": 0.01, + "learning_rate": 4.073731150453726e-05, + "loss": 3.7057, + "step": 565760 + }, + { + "epoch": 0.01, + "learning_rate": 4.072892555702674e-05, + "loss": 3.6793, + "step": 566272 + }, + { + "epoch": 0.01, + "learning_rate": 4.072055598831995e-05, + "loss": 3.6821, + "step": 566784 + }, + { + "epoch": 0.01, + "learning_rate": 4.071217004080943e-05, + "loss": 3.6968, + "step": 567296 + }, + { + "epoch": 0.01, + "learning_rate": 4.070378409329891e-05, + "loss": 3.6942, + "step": 567808 + }, + { + "epoch": 0.01, + "learning_rate": 4.069539814578839e-05, + "loss": 3.6814, + "step": 568320 + }, + { + "epoch": 0.01, + "learning_rate": 4.06870285770816e-05, + "loss": 3.6889, + "step": 568832 + }, + { + "epoch": 0.01, + "learning_rate": 4.067864262957108e-05, + "loss": 3.6802, + "step": 569344 + }, + { + "epoch": 0.01, + "learning_rate": 4.067025668206056e-05, + "loss": 3.6938, + "step": 569856 + }, + { + "epoch": 0.01, + "learning_rate": 4.066187073455004e-05, + "loss": 3.6949, + "step": 570368 + }, + { + "epoch": 0.01, + "learning_rate": 4.065348478703952e-05, + "loss": 3.6864, + "step": 570880 + }, + { + "epoch": 0.01, + "learning_rate": 4.064511521833273e-05, + "loss": 3.6938, + "step": 571392 + }, + { + "epoch": 0.01, + "learning_rate": 4.063672927082221e-05, + "loss": 3.6954, + "step": 571904 + }, + { + "epoch": 0.01, + "learning_rate": 4.062834332331169e-05, + "loss": 3.7013, + "step": 572416 + }, + { + "epoch": 0.01, + "learning_rate": 4.061995737580117e-05, + "loss": 3.6826, + "step": 572928 + }, + { + "epoch": 0.01, + "learning_rate": 4.061158780709438e-05, + "loss": 3.698, + "step": 573440 + }, + { + "epoch": 0.01, + "learning_rate": 4.060320185958386e-05, + "loss": 3.6979, + "step": 573952 + }, + { + "epoch": 0.01, + "learning_rate": 4.059481591207334e-05, + "loss": 3.6742, + "step": 574464 + }, + { + "epoch": 0.01, + "learning_rate": 4.058642996456282e-05, + "loss": 3.6995, + "step": 574976 + }, + { + "epoch": 0.01, + "learning_rate": 4.057806039585603e-05, + "loss": 3.689, + "step": 575488 + }, + { + "epoch": 0.01, + "learning_rate": 4.056967444834551e-05, + "loss": 3.6975, + "step": 576000 + }, + { + "epoch": 0.01, + "learning_rate": 4.056128850083499e-05, + "loss": 3.6914, + "step": 576512 + }, + { + "epoch": 0.01, + "learning_rate": 4.055290255332447e-05, + "loss": 3.6744, + "step": 577024 + }, + { + "epoch": 0.01, + "learning_rate": 4.054451660581395e-05, + "loss": 3.6742, + "step": 577536 + }, + { + "epoch": 0.01, + "learning_rate": 4.0536147037107167e-05, + "loss": 3.6866, + "step": 578048 + }, + { + "epoch": 0.01, + "learning_rate": 4.0527761089596647e-05, + "loss": 3.6958, + "step": 578560 + }, + { + "epoch": 0.01, + "learning_rate": 4.0519375142086127e-05, + "loss": 3.682, + "step": 579072 + }, + { + "epoch": 0.01, + "learning_rate": 4.0510989194575606e-05, + "loss": 3.6816, + "step": 579584 + }, + { + "epoch": 0.02, + "learning_rate": 4.0502619625868816e-05, + "loss": 3.6885, + "step": 580096 + }, + { + "epoch": 0.02, + "learning_rate": 4.0494233678358296e-05, + "loss": 3.6906, + "step": 580608 + }, + { + "epoch": 0.02, + "learning_rate": 4.0485847730847775e-05, + "loss": 3.6824, + "step": 581120 + }, + { + "epoch": 0.02, + "learning_rate": 4.0477461783337255e-05, + "loss": 3.6834, + "step": 581632 + }, + { + "epoch": 0.02, + "learning_rate": 4.0469092214630465e-05, + "loss": 3.6909, + "step": 582144 + }, + { + "epoch": 0.02, + "learning_rate": 4.0460706267119944e-05, + "loss": 3.695, + "step": 582656 + }, + { + "epoch": 0.02, + "learning_rate": 4.0452320319609424e-05, + "loss": 3.6903, + "step": 583168 + }, + { + "epoch": 0.02, + "learning_rate": 4.0443934372098904e-05, + "loss": 3.6734, + "step": 583680 + }, + { + "epoch": 0.02, + "learning_rate": 4.043556480339212e-05, + "loss": 3.6876, + "step": 584192 + }, + { + "epoch": 0.02, + "learning_rate": 4.04271788558816e-05, + "loss": 3.68, + "step": 584704 + }, + { + "epoch": 0.02, + "learning_rate": 4.041879290837108e-05, + "loss": 3.6958, + "step": 585216 + }, + { + "epoch": 0.02, + "learning_rate": 4.041040696086056e-05, + "loss": 3.6891, + "step": 585728 + }, + { + "epoch": 0.02, + "learning_rate": 4.040203739215377e-05, + "loss": 3.6856, + "step": 586240 + }, + { + "epoch": 0.02, + "learning_rate": 4.039365144464325e-05, + "loss": 3.6808, + "step": 586752 + }, + { + "epoch": 0.02, + "learning_rate": 4.038526549713273e-05, + "loss": 3.6799, + "step": 587264 + }, + { + "epoch": 0.02, + "learning_rate": 4.037687954962221e-05, + "loss": 3.6778, + "step": 587776 + }, + { + "epoch": 0.02, + "learning_rate": 4.036850998091542e-05, + "loss": 3.6855, + "step": 588288 + }, + { + "epoch": 0.02, + "learning_rate": 4.03601240334049e-05, + "loss": 3.6742, + "step": 588800 + }, + { + "epoch": 0.02, + "learning_rate": 4.035173808589438e-05, + "loss": 3.6914, + "step": 589312 + }, + { + "epoch": 0.02, + "learning_rate": 4.034335213838386e-05, + "loss": 3.6794, + "step": 589824 + }, + { + "epoch": 0.02, + "learning_rate": 4.033496619087334e-05, + "loss": 3.6829, + "step": 590336 + }, + { + "epoch": 0.02, + "learning_rate": 4.0326596622166554e-05, + "loss": 3.6799, + "step": 590848 + }, + { + "epoch": 0.02, + "learning_rate": 4.0318210674656034e-05, + "loss": 3.6789, + "step": 591360 + }, + { + "epoch": 0.02, + "learning_rate": 4.0309824727145514e-05, + "loss": 3.6909, + "step": 591872 + }, + { + "epoch": 0.02, + "learning_rate": 4.0301438779634994e-05, + "loss": 3.6791, + "step": 592384 + }, + { + "epoch": 0.02, + "learning_rate": 4.02930692109282e-05, + "loss": 3.6969, + "step": 592896 + }, + { + "epoch": 0.02, + "learning_rate": 4.028468326341768e-05, + "loss": 3.6926, + "step": 593408 + }, + { + "epoch": 0.02, + "learning_rate": 4.027629731590716e-05, + "loss": 3.6818, + "step": 593920 + }, + { + "epoch": 0.02, + "learning_rate": 4.026791136839664e-05, + "loss": 3.6848, + "step": 594432 + }, + { + "epoch": 0.02, + "learning_rate": 4.025954179968985e-05, + "loss": 3.6799, + "step": 594944 + }, + { + "epoch": 0.02, + "learning_rate": 4.025115585217933e-05, + "loss": 3.6821, + "step": 595456 + }, + { + "epoch": 0.02, + "learning_rate": 4.024276990466881e-05, + "loss": 3.6874, + "step": 595968 + }, + { + "epoch": 0.02, + "learning_rate": 4.023438395715829e-05, + "loss": 3.6799, + "step": 596480 + }, + { + "epoch": 0.02, + "learning_rate": 4.022601438845151e-05, + "loss": 3.6842, + "step": 596992 + }, + { + "epoch": 0.02, + "learning_rate": 4.021762844094099e-05, + "loss": 3.6827, + "step": 597504 + }, + { + "epoch": 0.02, + "learning_rate": 4.020924249343047e-05, + "loss": 3.684, + "step": 598016 + }, + { + "epoch": 0.02, + "learning_rate": 4.020085654591995e-05, + "loss": 3.6833, + "step": 598528 + }, + { + "epoch": 0.02, + "learning_rate": 4.0192486977213157e-05, + "loss": 3.6761, + "step": 599040 + }, + { + "epoch": 0.02, + "learning_rate": 4.0184101029702636e-05, + "loss": 3.6833, + "step": 599552 + }, + { + "epoch": 0.02, + "learning_rate": 4.0175715082192116e-05, + "loss": 3.6747, + "step": 600064 + }, + { + "epoch": 0.02, + "learning_rate": 4.0167329134681596e-05, + "loss": 3.6959, + "step": 600576 + }, + { + "epoch": 0.02, + "learning_rate": 4.015894318717107e-05, + "loss": 3.6809, + "step": 601088 + }, + { + "epoch": 0.02, + "learning_rate": 4.0150573618464285e-05, + "loss": 3.6878, + "step": 601600 + }, + { + "epoch": 0.02, + "learning_rate": 4.0142187670953765e-05, + "loss": 3.676, + "step": 602112 + }, + { + "epoch": 0.02, + "learning_rate": 4.0133801723443245e-05, + "loss": 3.6857, + "step": 602624 + }, + { + "epoch": 0.02, + "learning_rate": 4.0125415775932725e-05, + "loss": 3.6858, + "step": 603136 + }, + { + "epoch": 0.02, + "learning_rate": 4.011704620722594e-05, + "loss": 3.6848, + "step": 603648 + }, + { + "epoch": 0.02, + "learning_rate": 4.010866025971542e-05, + "loss": 3.6758, + "step": 604160 + }, + { + "epoch": 0.02, + "learning_rate": 4.0100274312204894e-05, + "loss": 3.6885, + "step": 604672 + }, + { + "epoch": 0.02, + "learning_rate": 4.0091888364694374e-05, + "loss": 3.6842, + "step": 605184 + }, + { + "epoch": 0.02, + "learning_rate": 4.008351879598759e-05, + "loss": 3.7027, + "step": 605696 + }, + { + "epoch": 0.02, + "learning_rate": 4.007513284847707e-05, + "loss": 3.6765, + "step": 606208 + }, + { + "epoch": 0.02, + "learning_rate": 4.006674690096654e-05, + "loss": 3.6909, + "step": 606720 + }, + { + "epoch": 0.02, + "learning_rate": 4.005836095345602e-05, + "loss": 3.6847, + "step": 607232 + }, + { + "epoch": 0.02, + "learning_rate": 4.004999138474924e-05, + "loss": 3.6812, + "step": 607744 + }, + { + "epoch": 0.02, + "learning_rate": 4.004160543723872e-05, + "loss": 3.6809, + "step": 608256 + }, + { + "epoch": 0.02, + "learning_rate": 4.00332194897282e-05, + "loss": 3.6868, + "step": 608768 + }, + { + "epoch": 0.02, + "learning_rate": 4.002483354221768e-05, + "loss": 3.6868, + "step": 609280 + }, + { + "epoch": 0.02, + "learning_rate": 4.0016463973510895e-05, + "loss": 3.6677, + "step": 609792 + }, + { + "epoch": 0.02, + "learning_rate": 4.000807802600037e-05, + "loss": 3.6876, + "step": 610304 + }, + { + "epoch": 0.03, + "eval_loss": 3.858370542526245, + "eval_runtime": 305.1645, + "eval_samples_per_second": 1250.444, + "eval_steps_per_second": 39.077, + "step": 610560 + }, + { + "epoch": 1.0, + "learning_rate": 3.999969207848985e-05, + "loss": 3.6737, + "step": 610816 + }, + { + "epoch": 1.0, + "learning_rate": 3.999130613097933e-05, + "loss": 3.6693, + "step": 611328 + }, + { + "epoch": 1.0, + "learning_rate": 3.9982936562272544e-05, + "loss": 3.6832, + "step": 611840 + }, + { + "epoch": 1.0, + "learning_rate": 3.997455061476202e-05, + "loss": 3.6945, + "step": 612352 + }, + { + "epoch": 1.0, + "learning_rate": 3.99661646672515e-05, + "loss": 3.6859, + "step": 612864 + }, + { + "epoch": 1.0, + "learning_rate": 3.995777871974098e-05, + "loss": 3.6726, + "step": 613376 + }, + { + "epoch": 1.0, + "learning_rate": 3.994939277223046e-05, + "loss": 3.6857, + "step": 613888 + }, + { + "epoch": 1.0, + "learning_rate": 3.994102320352367e-05, + "loss": 3.6689, + "step": 614400 + }, + { + "epoch": 1.0, + "learning_rate": 3.993263725601315e-05, + "loss": 3.6788, + "step": 614912 + }, + { + "epoch": 1.0, + "learning_rate": 3.992425130850263e-05, + "loss": 3.6851, + "step": 615424 + }, + { + "epoch": 1.0, + "learning_rate": 3.991586536099211e-05, + "loss": 3.6775, + "step": 615936 + }, + { + "epoch": 1.0, + "learning_rate": 3.990749579228532e-05, + "loss": 3.6788, + "step": 616448 + }, + { + "epoch": 1.0, + "learning_rate": 3.98991098447748e-05, + "loss": 3.6793, + "step": 616960 + }, + { + "epoch": 1.0, + "learning_rate": 3.989072389726428e-05, + "loss": 3.6691, + "step": 617472 + }, + { + "epoch": 1.0, + "learning_rate": 3.988233794975376e-05, + "loss": 3.6688, + "step": 617984 + }, + { + "epoch": 1.0, + "learning_rate": 3.987396838104697e-05, + "loss": 3.6653, + "step": 618496 + }, + { + "epoch": 1.0, + "learning_rate": 3.986558243353645e-05, + "loss": 3.672, + "step": 619008 + }, + { + "epoch": 1.0, + "learning_rate": 3.985719648602593e-05, + "loss": 3.6749, + "step": 619520 + }, + { + "epoch": 1.0, + "learning_rate": 3.984881053851541e-05, + "loss": 3.6697, + "step": 620032 + }, + { + "epoch": 1.0, + "learning_rate": 3.9840440969808626e-05, + "loss": 3.6835, + "step": 620544 + }, + { + "epoch": 1.0, + "learning_rate": 3.9832055022298106e-05, + "loss": 3.6863, + "step": 621056 + }, + { + "epoch": 1.0, + "learning_rate": 3.9823669074787586e-05, + "loss": 3.6841, + "step": 621568 + }, + { + "epoch": 1.0, + "learning_rate": 3.9815283127277066e-05, + "loss": 3.6739, + "step": 622080 + }, + { + "epoch": 1.0, + "learning_rate": 3.9806913558570275e-05, + "loss": 3.6762, + "step": 622592 + }, + { + "epoch": 1.0, + "learning_rate": 3.9798527611059755e-05, + "loss": 3.6755, + "step": 623104 + }, + { + "epoch": 1.0, + "learning_rate": 3.9790141663549235e-05, + "loss": 3.676, + "step": 623616 + }, + { + "epoch": 1.0, + "learning_rate": 3.9781755716038715e-05, + "loss": 3.6767, + "step": 624128 + }, + { + "epoch": 1.0, + "learning_rate": 3.9773386147331924e-05, + "loss": 3.6684, + "step": 624640 + }, + { + "epoch": 1.0, + "learning_rate": 3.9765000199821404e-05, + "loss": 3.6691, + "step": 625152 + }, + { + "epoch": 1.0, + "learning_rate": 3.9756614252310884e-05, + "loss": 3.6661, + "step": 625664 + }, + { + "epoch": 1.01, + "learning_rate": 3.9748228304800364e-05, + "loss": 3.6688, + "step": 626176 + }, + { + "epoch": 1.01, + "learning_rate": 3.973985873609358e-05, + "loss": 3.6836, + "step": 626688 + }, + { + "epoch": 1.01, + "learning_rate": 3.973147278858306e-05, + "loss": 3.6774, + "step": 627200 + }, + { + "epoch": 1.01, + "learning_rate": 3.972308684107254e-05, + "loss": 3.6781, + "step": 627712 + }, + { + "epoch": 1.01, + "learning_rate": 3.971470089356202e-05, + "loss": 3.6662, + "step": 628224 + }, + { + "epoch": 1.01, + "learning_rate": 3.970633132485523e-05, + "loss": 3.6774, + "step": 628736 + }, + { + "epoch": 1.01, + "learning_rate": 3.969794537734471e-05, + "loss": 3.6734, + "step": 629248 + }, + { + "epoch": 1.01, + "learning_rate": 3.968955942983419e-05, + "loss": 3.6648, + "step": 629760 + }, + { + "epoch": 1.01, + "learning_rate": 3.968117348232367e-05, + "loss": 3.669, + "step": 630272 + }, + { + "epoch": 1.01, + "learning_rate": 3.967280391361688e-05, + "loss": 3.6602, + "step": 630784 + }, + { + "epoch": 1.01, + "learning_rate": 3.966441796610636e-05, + "loss": 3.6676, + "step": 631296 + }, + { + "epoch": 1.01, + "learning_rate": 3.965603201859584e-05, + "loss": 3.6781, + "step": 631808 + }, + { + "epoch": 1.01, + "learning_rate": 3.964764607108532e-05, + "loss": 3.6697, + "step": 632320 + }, + { + "epoch": 1.01, + "learning_rate": 3.9639276502378534e-05, + "loss": 3.6712, + "step": 632832 + }, + { + "epoch": 1.01, + "learning_rate": 3.9630890554868014e-05, + "loss": 3.6791, + "step": 633344 + }, + { + "epoch": 1.01, + "learning_rate": 3.9622504607357494e-05, + "loss": 3.6743, + "step": 633856 + }, + { + "epoch": 1.01, + "learning_rate": 3.9614118659846974e-05, + "loss": 3.675, + "step": 634368 + }, + { + "epoch": 1.01, + "learning_rate": 3.960574909114018e-05, + "loss": 3.6675, + "step": 634880 + }, + { + "epoch": 1.01, + "learning_rate": 3.959736314362966e-05, + "loss": 3.6414, + "step": 635392 + }, + { + "epoch": 1.01, + "learning_rate": 3.958897719611914e-05, + "loss": 3.6832, + "step": 635904 + }, + { + "epoch": 1.01, + "learning_rate": 3.958059124860862e-05, + "loss": 3.6715, + "step": 636416 + }, + { + "epoch": 1.01, + "learning_rate": 3.957222167990183e-05, + "loss": 3.6732, + "step": 636928 + }, + { + "epoch": 1.01, + "learning_rate": 3.956383573239131e-05, + "loss": 3.6706, + "step": 637440 + }, + { + "epoch": 1.01, + "learning_rate": 3.955544978488079e-05, + "loss": 3.6607, + "step": 637952 + }, + { + "epoch": 1.01, + "learning_rate": 3.954706383737027e-05, + "loss": 3.6532, + "step": 638464 + }, + { + "epoch": 1.01, + "learning_rate": 3.953869426866349e-05, + "loss": 3.663, + "step": 638976 + }, + { + "epoch": 1.01, + "learning_rate": 3.953030832115297e-05, + "loss": 3.6583, + "step": 639488 + }, + { + "epoch": 1.01, + "learning_rate": 3.952192237364245e-05, + "loss": 3.6617, + "step": 640000 + }, + { + "epoch": 1.01, + "learning_rate": 3.951353642613193e-05, + "loss": 3.6809, + "step": 640512 + }, + { + "epoch": 1.01, + "learning_rate": 3.950515047862141e-05, + "loss": 3.6598, + "step": 641024 + }, + { + "epoch": 1.01, + "learning_rate": 3.9496780909914616e-05, + "loss": 3.6515, + "step": 641536 + }, + { + "epoch": 1.01, + "learning_rate": 3.9488394962404096e-05, + "loss": 3.6783, + "step": 642048 + }, + { + "epoch": 1.01, + "learning_rate": 3.9480009014893576e-05, + "loss": 3.6485, + "step": 642560 + }, + { + "epoch": 1.01, + "learning_rate": 3.9471639446186785e-05, + "loss": 3.6533, + "step": 643072 + }, + { + "epoch": 1.01, + "learning_rate": 3.9463253498676265e-05, + "loss": 3.6713, + "step": 643584 + }, + { + "epoch": 1.01, + "learning_rate": 3.9454867551165745e-05, + "loss": 3.6665, + "step": 644096 + }, + { + "epoch": 1.01, + "learning_rate": 3.9446481603655225e-05, + "loss": 3.6535, + "step": 644608 + }, + { + "epoch": 1.01, + "learning_rate": 3.943811203494844e-05, + "loss": 3.657, + "step": 645120 + }, + { + "epoch": 1.01, + "learning_rate": 3.942972608743792e-05, + "loss": 3.6545, + "step": 645632 + }, + { + "epoch": 1.01, + "learning_rate": 3.94213401399274e-05, + "loss": 3.6652, + "step": 646144 + }, + { + "epoch": 1.01, + "learning_rate": 3.941295419241688e-05, + "loss": 3.6628, + "step": 646656 + }, + { + "epoch": 1.01, + "learning_rate": 3.940456824490636e-05, + "loss": 3.6607, + "step": 647168 + }, + { + "epoch": 1.01, + "learning_rate": 3.939619867619957e-05, + "loss": 3.6645, + "step": 647680 + }, + { + "epoch": 1.01, + "learning_rate": 3.938781272868905e-05, + "loss": 3.6694, + "step": 648192 + }, + { + "epoch": 1.01, + "learning_rate": 3.937942678117853e-05, + "loss": 3.6746, + "step": 648704 + }, + { + "epoch": 1.01, + "learning_rate": 3.937104083366801e-05, + "loss": 3.6538, + "step": 649216 + }, + { + "epoch": 1.01, + "learning_rate": 3.936267126496122e-05, + "loss": 3.6714, + "step": 649728 + }, + { + "epoch": 1.01, + "learning_rate": 3.93542853174507e-05, + "loss": 3.6683, + "step": 650240 + }, + { + "epoch": 1.01, + "learning_rate": 3.934589936994018e-05, + "loss": 3.6456, + "step": 650752 + }, + { + "epoch": 1.01, + "learning_rate": 3.933751342242966e-05, + "loss": 3.669, + "step": 651264 + }, + { + "epoch": 1.01, + "learning_rate": 3.9329143853722875e-05, + "loss": 3.6655, + "step": 651776 + }, + { + "epoch": 1.01, + "learning_rate": 3.9320757906212355e-05, + "loss": 3.6717, + "step": 652288 + }, + { + "epoch": 1.01, + "learning_rate": 3.9312371958701835e-05, + "loss": 3.661, + "step": 652800 + }, + { + "epoch": 1.01, + "learning_rate": 3.9303986011191315e-05, + "loss": 3.6493, + "step": 653312 + }, + { + "epoch": 1.01, + "learning_rate": 3.9295616442484524e-05, + "loss": 3.6421, + "step": 653824 + }, + { + "epoch": 1.01, + "learning_rate": 3.9287230494974004e-05, + "loss": 3.6577, + "step": 654336 + }, + { + "epoch": 1.01, + "learning_rate": 3.9278844547463484e-05, + "loss": 3.6665, + "step": 654848 + }, + { + "epoch": 1.01, + "learning_rate": 3.9270458599952964e-05, + "loss": 3.6576, + "step": 655360 + }, + { + "epoch": 1.01, + "learning_rate": 3.926208903124617e-05, + "loss": 3.6527, + "step": 655872 + }, + { + "epoch": 1.02, + "learning_rate": 3.925370308373565e-05, + "loss": 3.6571, + "step": 656384 + }, + { + "epoch": 1.02, + "learning_rate": 3.924531713622513e-05, + "loss": 3.6685, + "step": 656896 + }, + { + "epoch": 1.02, + "learning_rate": 3.923693118871461e-05, + "loss": 3.6552, + "step": 657408 + }, + { + "epoch": 1.02, + "learning_rate": 3.922856162000783e-05, + "loss": 3.6571, + "step": 657920 + }, + { + "epoch": 1.02, + "learning_rate": 3.922017567249731e-05, + "loss": 3.6618, + "step": 658432 + }, + { + "epoch": 1.02, + "learning_rate": 3.921178972498679e-05, + "loss": 3.6685, + "step": 658944 + }, + { + "epoch": 1.02, + "learning_rate": 3.920340377747627e-05, + "loss": 3.6623, + "step": 659456 + }, + { + "epoch": 1.02, + "learning_rate": 3.919503420876948e-05, + "loss": 3.6493, + "step": 659968 + }, + { + "epoch": 1.02, + "learning_rate": 3.918664826125896e-05, + "loss": 3.6555, + "step": 660480 + }, + { + "epoch": 1.02, + "learning_rate": 3.917826231374844e-05, + "loss": 3.6546, + "step": 660992 + }, + { + "epoch": 1.02, + "learning_rate": 3.916987636623792e-05, + "loss": 3.6676, + "step": 661504 + }, + { + "epoch": 1.02, + "learning_rate": 3.9161506797531126e-05, + "loss": 3.6621, + "step": 662016 + }, + { + "epoch": 1.02, + "learning_rate": 3.9153120850020606e-05, + "loss": 3.6595, + "step": 662528 + }, + { + "epoch": 1.02, + "learning_rate": 3.9144734902510086e-05, + "loss": 3.6531, + "step": 663040 + }, + { + "epoch": 1.02, + "learning_rate": 3.9136348954999566e-05, + "loss": 3.6518, + "step": 663552 + }, + { + "epoch": 1.02, + "learning_rate": 3.912797938629278e-05, + "loss": 3.6535, + "step": 664064 + }, + { + "epoch": 1.02, + "learning_rate": 3.911959343878226e-05, + "loss": 3.6573, + "step": 664576 + }, + { + "epoch": 1.02, + "learning_rate": 3.911120749127174e-05, + "loss": 3.6436, + "step": 665088 + }, + { + "epoch": 1.02, + "learning_rate": 3.910282154376122e-05, + "loss": 3.6664, + "step": 665600 + }, + { + "epoch": 1.02, + "learning_rate": 3.909445197505443e-05, + "loss": 3.6519, + "step": 666112 + }, + { + "epoch": 1.02, + "learning_rate": 3.908606602754391e-05, + "loss": 3.653, + "step": 666624 + }, + { + "epoch": 1.02, + "learning_rate": 3.907768008003339e-05, + "loss": 3.6516, + "step": 667136 + }, + { + "epoch": 1.02, + "learning_rate": 3.9069294132522864e-05, + "loss": 3.6497, + "step": 667648 + }, + { + "epoch": 1.02, + "learning_rate": 3.906092456381608e-05, + "loss": 3.6674, + "step": 668160 + }, + { + "epoch": 1.02, + "learning_rate": 3.905253861630556e-05, + "loss": 3.6466, + "step": 668672 + }, + { + "epoch": 1.02, + "learning_rate": 3.904415266879504e-05, + "loss": 3.6732, + "step": 669184 + }, + { + "epoch": 1.02, + "learning_rate": 3.903576672128452e-05, + "loss": 3.6652, + "step": 669696 + }, + { + "epoch": 1.02, + "learning_rate": 3.9027397152577736e-05, + "loss": 3.6576, + "step": 670208 + }, + { + "epoch": 1.02, + "learning_rate": 3.9019011205067216e-05, + "loss": 3.6555, + "step": 670720 + }, + { + "epoch": 1.02, + "learning_rate": 3.9010625257556696e-05, + "loss": 3.6574, + "step": 671232 + }, + { + "epoch": 1.02, + "learning_rate": 3.900223931004617e-05, + "loss": 3.6539, + "step": 671744 + }, + { + "epoch": 1.02, + "learning_rate": 3.8993869741339385e-05, + "loss": 3.6604, + "step": 672256 + }, + { + "epoch": 1.02, + "learning_rate": 3.8985483793828865e-05, + "loss": 3.6499, + "step": 672768 + }, + { + "epoch": 1.02, + "learning_rate": 3.897709784631834e-05, + "loss": 3.6607, + "step": 673280 + }, + { + "epoch": 1.02, + "learning_rate": 3.896871189880782e-05, + "loss": 3.653, + "step": 673792 + }, + { + "epoch": 1.02, + "learning_rate": 3.8960342330101034e-05, + "loss": 3.6561, + "step": 674304 + }, + { + "epoch": 1.02, + "learning_rate": 3.8951956382590514e-05, + "loss": 3.6562, + "step": 674816 + }, + { + "epoch": 1.02, + "learning_rate": 3.8943570435079994e-05, + "loss": 3.6488, + "step": 675328 + }, + { + "epoch": 1.02, + "learning_rate": 3.8935184487569473e-05, + "loss": 3.6582, + "step": 675840 + }, + { + "epoch": 1.02, + "learning_rate": 3.892681491886269e-05, + "loss": 3.6477, + "step": 676352 + }, + { + "epoch": 1.02, + "learning_rate": 3.891842897135217e-05, + "loss": 3.6685, + "step": 676864 + }, + { + "epoch": 1.02, + "learning_rate": 3.891004302384164e-05, + "loss": 3.6558, + "step": 677376 + }, + { + "epoch": 1.02, + "learning_rate": 3.890165707633112e-05, + "loss": 3.661, + "step": 677888 + }, + { + "epoch": 1.02, + "learning_rate": 3.889328750762434e-05, + "loss": 3.6479, + "step": 678400 + }, + { + "epoch": 1.02, + "learning_rate": 3.888490156011381e-05, + "loss": 3.6633, + "step": 678912 + }, + { + "epoch": 1.02, + "learning_rate": 3.887651561260329e-05, + "loss": 3.6534, + "step": 679424 + }, + { + "epoch": 1.02, + "learning_rate": 3.886812966509277e-05, + "loss": 3.6574, + "step": 679936 + }, + { + "epoch": 1.02, + "learning_rate": 3.885976009638599e-05, + "loss": 3.6508, + "step": 680448 + }, + { + "epoch": 1.02, + "learning_rate": 3.885137414887546e-05, + "loss": 3.6607, + "step": 680960 + }, + { + "epoch": 1.02, + "learning_rate": 3.884298820136494e-05, + "loss": 3.6557, + "step": 681472 + }, + { + "epoch": 1.02, + "learning_rate": 3.883460225385443e-05, + "loss": 3.6763, + "step": 681984 + }, + { + "epoch": 1.02, + "learning_rate": 3.882623268514764e-05, + "loss": 3.6483, + "step": 682496 + }, + { + "epoch": 1.02, + "learning_rate": 3.8817846737637116e-05, + "loss": 3.6638, + "step": 683008 + }, + { + "epoch": 1.02, + "learning_rate": 3.8809460790126596e-05, + "loss": 3.6595, + "step": 683520 + }, + { + "epoch": 1.02, + "learning_rate": 3.8801074842616076e-05, + "loss": 3.6603, + "step": 684032 + }, + { + "epoch": 1.02, + "learning_rate": 3.8792705273909285e-05, + "loss": 3.6499, + "step": 684544 + }, + { + "epoch": 1.02, + "learning_rate": 3.8784319326398765e-05, + "loss": 3.6606, + "step": 685056 + }, + { + "epoch": 1.02, + "learning_rate": 3.8775933378888245e-05, + "loss": 3.6568, + "step": 685568 + }, + { + "epoch": 1.02, + "learning_rate": 3.8767547431377725e-05, + "loss": 3.6494, + "step": 686080 + }, + { + "epoch": 1.02, + "learning_rate": 3.8759177862670934e-05, + "loss": 3.6572, + "step": 686592 + }, + { + "epoch": 1.03, + "eval_loss": 3.853288173675537, + "eval_runtime": 304.5472, + "eval_samples_per_second": 1252.978, + "eval_steps_per_second": 39.156, + "step": 686880 + }, + { + "epoch": 0.0, + "learning_rate": 3.8750791915160414e-05, + "loss": 3.6494, + "step": 687104 + }, + { + "epoch": 0.0, + "learning_rate": 3.8742405967649894e-05, + "loss": 3.6402, + "step": 687616 + }, + { + "epoch": 0.0, + "learning_rate": 3.873402002013938e-05, + "loss": 3.6549, + "step": 688128 + }, + { + "epoch": 0.0, + "learning_rate": 3.872563407262886e-05, + "loss": 3.6678, + "step": 688640 + }, + { + "epoch": 0.0, + "learning_rate": 3.871726450392207e-05, + "loss": 3.6621, + "step": 689152 + }, + { + "epoch": 0.0, + "learning_rate": 3.870887855641155e-05, + "loss": 3.6455, + "step": 689664 + }, + { + "epoch": 0.0, + "learning_rate": 3.870049260890103e-05, + "loss": 3.6619, + "step": 690176 + }, + { + "epoch": 0.0, + "learning_rate": 3.869210666139051e-05, + "loss": 3.6438, + "step": 690688 + }, + { + "epoch": 0.0, + "learning_rate": 3.868373709268372e-05, + "loss": 3.6505, + "step": 691200 + }, + { + "epoch": 0.0, + "learning_rate": 3.86753511451732e-05, + "loss": 3.6615, + "step": 691712 + }, + { + "epoch": 0.0, + "learning_rate": 3.866696519766268e-05, + "loss": 3.6493, + "step": 692224 + }, + { + "epoch": 0.0, + "learning_rate": 3.865857925015216e-05, + "loss": 3.651, + "step": 692736 + }, + { + "epoch": 0.0, + "learning_rate": 3.865020968144537e-05, + "loss": 3.6552, + "step": 693248 + }, + { + "epoch": 0.0, + "learning_rate": 3.864182373393485e-05, + "loss": 3.6439, + "step": 693760 + }, + { + "epoch": 0.0, + "learning_rate": 3.8633437786424334e-05, + "loss": 3.6423, + "step": 694272 + }, + { + "epoch": 0.0, + "learning_rate": 3.8625051838913814e-05, + "loss": 3.6388, + "step": 694784 + }, + { + "epoch": 0.0, + "learning_rate": 3.8616665891403294e-05, + "loss": 3.6444, + "step": 695296 + }, + { + "epoch": 0.0, + "learning_rate": 3.8608279943892774e-05, + "loss": 3.6481, + "step": 695808 + }, + { + "epoch": 0.0, + "learning_rate": 3.8599893996382254e-05, + "loss": 3.6452, + "step": 696320 + }, + { + "epoch": 0.0, + "learning_rate": 3.8591508048871734e-05, + "loss": 3.6592, + "step": 696832 + }, + { + "epoch": 0.0, + "learning_rate": 3.858313848016494e-05, + "loss": 3.6616, + "step": 697344 + }, + { + "epoch": 0.0, + "learning_rate": 3.857475253265442e-05, + "loss": 3.6599, + "step": 697856 + }, + { + "epoch": 0.0, + "learning_rate": 3.85663665851439e-05, + "loss": 3.6491, + "step": 698368 + }, + { + "epoch": 0.0, + "learning_rate": 3.855798063763338e-05, + "loss": 3.6454, + "step": 698880 + }, + { + "epoch": 0.0, + "learning_rate": 3.854961106892659e-05, + "loss": 3.6526, + "step": 699392 + }, + { + "epoch": 0.0, + "learning_rate": 3.854122512141607e-05, + "loss": 3.6487, + "step": 699904 + }, + { + "epoch": 0.0, + "learning_rate": 3.853283917390555e-05, + "loss": 3.6486, + "step": 700416 + }, + { + "epoch": 0.0, + "learning_rate": 3.852445322639503e-05, + "loss": 3.6448, + "step": 700928 + }, + { + "epoch": 0.0, + "learning_rate": 3.851608365768825e-05, + "loss": 3.6409, + "step": 701440 + }, + { + "epoch": 0.0, + "learning_rate": 3.850769771017773e-05, + "loss": 3.643, + "step": 701952 + }, + { + "epoch": 0.01, + "learning_rate": 3.849931176266721e-05, + "loss": 3.6421, + "step": 702464 + }, + { + "epoch": 0.01, + "learning_rate": 3.849092581515669e-05, + "loss": 3.6538, + "step": 702976 + }, + { + "epoch": 0.01, + "learning_rate": 3.84825562464499e-05, + "loss": 3.6566, + "step": 703488 + }, + { + "epoch": 0.01, + "learning_rate": 3.847417029893938e-05, + "loss": 3.6494, + "step": 704000 + }, + { + "epoch": 0.01, + "learning_rate": 3.846578435142886e-05, + "loss": 3.6442, + "step": 704512 + }, + { + "epoch": 0.01, + "learning_rate": 3.845739840391834e-05, + "loss": 3.6426, + "step": 705024 + }, + { + "epoch": 0.01, + "learning_rate": 3.8449028835211546e-05, + "loss": 3.6534, + "step": 705536 + }, + { + "epoch": 0.01, + "learning_rate": 3.8440642887701026e-05, + "loss": 3.6397, + "step": 706048 + }, + { + "epoch": 0.01, + "learning_rate": 3.8432256940190506e-05, + "loss": 3.6439, + "step": 706560 + }, + { + "epoch": 0.01, + "learning_rate": 3.8423870992679986e-05, + "loss": 3.6328, + "step": 707072 + }, + { + "epoch": 0.01, + "learning_rate": 3.84155014239732e-05, + "loss": 3.6462, + "step": 707584 + }, + { + "epoch": 0.01, + "learning_rate": 3.840711547646268e-05, + "loss": 3.6472, + "step": 708096 + }, + { + "epoch": 0.01, + "learning_rate": 3.839872952895216e-05, + "loss": 3.6448, + "step": 708608 + }, + { + "epoch": 0.01, + "learning_rate": 3.839034358144164e-05, + "loss": 3.6478, + "step": 709120 + }, + { + "epoch": 0.01, + "learning_rate": 3.838197401273485e-05, + "loss": 3.6505, + "step": 709632 + }, + { + "epoch": 0.01, + "learning_rate": 3.837358806522433e-05, + "loss": 3.6505, + "step": 710144 + }, + { + "epoch": 0.01, + "learning_rate": 3.836520211771381e-05, + "loss": 3.6496, + "step": 710656 + }, + { + "epoch": 0.01, + "learning_rate": 3.835681617020329e-05, + "loss": 3.6446, + "step": 711168 + }, + { + "epoch": 0.01, + "learning_rate": 3.83484466014965e-05, + "loss": 3.615, + "step": 711680 + }, + { + "epoch": 0.01, + "learning_rate": 3.834006065398598e-05, + "loss": 3.6538, + "step": 712192 + }, + { + "epoch": 0.01, + "learning_rate": 3.833167470647546e-05, + "loss": 3.6452, + "step": 712704 + }, + { + "epoch": 0.01, + "learning_rate": 3.832328875896494e-05, + "loss": 3.6512, + "step": 713216 + }, + { + "epoch": 0.01, + "learning_rate": 3.8314919190258155e-05, + "loss": 3.6449, + "step": 713728 + }, + { + "epoch": 0.01, + "learning_rate": 3.8306533242747635e-05, + "loss": 3.6356, + "step": 714240 + }, + { + "epoch": 0.01, + "learning_rate": 3.8298147295237115e-05, + "loss": 3.6273, + "step": 714752 + }, + { + "epoch": 0.01, + "learning_rate": 3.8289761347726595e-05, + "loss": 3.6372, + "step": 715264 + }, + { + "epoch": 0.01, + "learning_rate": 3.8281391779019804e-05, + "loss": 3.635, + "step": 715776 + }, + { + "epoch": 0.01, + "learning_rate": 3.8273005831509284e-05, + "loss": 3.6339, + "step": 716288 + }, + { + "epoch": 0.01, + "learning_rate": 3.8264619883998764e-05, + "loss": 3.6555, + "step": 716800 + }, + { + "epoch": 0.01, + "learning_rate": 3.8256233936488244e-05, + "loss": 3.6387, + "step": 717312 + }, + { + "epoch": 0.01, + "learning_rate": 3.824786436778145e-05, + "loss": 3.6209, + "step": 717824 + }, + { + "epoch": 0.01, + "learning_rate": 3.823947842027093e-05, + "loss": 3.6554, + "step": 718336 + }, + { + "epoch": 0.01, + "learning_rate": 3.823109247276041e-05, + "loss": 3.6255, + "step": 718848 + }, + { + "epoch": 0.01, + "learning_rate": 3.822270652524989e-05, + "loss": 3.6254, + "step": 719360 + }, + { + "epoch": 0.01, + "learning_rate": 3.821433695654311e-05, + "loss": 3.6455, + "step": 719872 + }, + { + "epoch": 0.01, + "learning_rate": 3.820595100903259e-05, + "loss": 3.6373, + "step": 720384 + }, + { + "epoch": 0.01, + "learning_rate": 3.819756506152207e-05, + "loss": 3.6311, + "step": 720896 + }, + { + "epoch": 0.01, + "learning_rate": 3.818917911401154e-05, + "loss": 3.632, + "step": 721408 + }, + { + "epoch": 0.01, + "learning_rate": 3.818080954530476e-05, + "loss": 3.6306, + "step": 721920 + }, + { + "epoch": 0.01, + "learning_rate": 3.817242359779424e-05, + "loss": 3.6382, + "step": 722432 + }, + { + "epoch": 0.01, + "learning_rate": 3.816403765028371e-05, + "loss": 3.6386, + "step": 722944 + }, + { + "epoch": 0.01, + "learning_rate": 3.815565170277319e-05, + "loss": 3.638, + "step": 723456 + }, + { + "epoch": 0.01, + "learning_rate": 3.814728213406641e-05, + "loss": 3.6391, + "step": 723968 + }, + { + "epoch": 0.01, + "learning_rate": 3.813889618655589e-05, + "loss": 3.6431, + "step": 724480 + }, + { + "epoch": 0.01, + "learning_rate": 3.813051023904537e-05, + "loss": 3.6514, + "step": 724992 + }, + { + "epoch": 0.01, + "learning_rate": 3.812212429153485e-05, + "loss": 3.6296, + "step": 725504 + }, + { + "epoch": 0.01, + "learning_rate": 3.811375472282806e-05, + "loss": 3.6446, + "step": 726016 + }, + { + "epoch": 0.01, + "learning_rate": 3.810536877531754e-05, + "loss": 3.6464, + "step": 726528 + }, + { + "epoch": 0.01, + "learning_rate": 3.8096982827807016e-05, + "loss": 3.6203, + "step": 727040 + }, + { + "epoch": 0.01, + "learning_rate": 3.8088596880296496e-05, + "loss": 3.6399, + "step": 727552 + }, + { + "epoch": 0.01, + "learning_rate": 3.808022731158971e-05, + "loss": 3.6441, + "step": 728064 + }, + { + "epoch": 0.01, + "learning_rate": 3.8071841364079185e-05, + "loss": 3.6451, + "step": 728576 + }, + { + "epoch": 0.01, + "learning_rate": 3.8063455416568665e-05, + "loss": 3.6427, + "step": 729088 + }, + { + "epoch": 0.01, + "learning_rate": 3.8055069469058145e-05, + "loss": 3.6197, + "step": 729600 + }, + { + "epoch": 0.01, + "learning_rate": 3.804669990035136e-05, + "loss": 3.616, + "step": 730112 + }, + { + "epoch": 0.01, + "learning_rate": 3.8038313952840834e-05, + "loss": 3.6371, + "step": 730624 + }, + { + "epoch": 0.01, + "learning_rate": 3.802992800533032e-05, + "loss": 3.6416, + "step": 731136 + }, + { + "epoch": 0.01, + "learning_rate": 3.80215420578198e-05, + "loss": 3.6296, + "step": 731648 + }, + { + "epoch": 0.01, + "learning_rate": 3.8013172489113016e-05, + "loss": 3.6337, + "step": 732160 + }, + { + "epoch": 0.02, + "learning_rate": 3.800478654160249e-05, + "loss": 3.6328, + "step": 732672 + }, + { + "epoch": 0.02, + "learning_rate": 3.799640059409197e-05, + "loss": 3.6417, + "step": 733184 + }, + { + "epoch": 0.02, + "learning_rate": 3.798801464658145e-05, + "loss": 3.6329, + "step": 733696 + }, + { + "epoch": 0.02, + "learning_rate": 3.797964507787466e-05, + "loss": 3.6339, + "step": 734208 + }, + { + "epoch": 0.02, + "learning_rate": 3.797125913036414e-05, + "loss": 3.635, + "step": 734720 + }, + { + "epoch": 0.02, + "learning_rate": 3.796287318285362e-05, + "loss": 3.643, + "step": 735232 + }, + { + "epoch": 0.02, + "learning_rate": 3.79544872353431e-05, + "loss": 3.6402, + "step": 735744 + }, + { + "epoch": 0.02, + "learning_rate": 3.794611766663631e-05, + "loss": 3.6255, + "step": 736256 + }, + { + "epoch": 0.02, + "learning_rate": 3.793773171912579e-05, + "loss": 3.63, + "step": 736768 + }, + { + "epoch": 0.02, + "learning_rate": 3.7929345771615274e-05, + "loss": 3.6322, + "step": 737280 + }, + { + "epoch": 0.02, + "learning_rate": 3.7920959824104754e-05, + "loss": 3.6433, + "step": 737792 + }, + { + "epoch": 0.02, + "learning_rate": 3.791259025539796e-05, + "loss": 3.6366, + "step": 738304 + }, + { + "epoch": 0.02, + "learning_rate": 3.790420430788744e-05, + "loss": 3.6344, + "step": 738816 + }, + { + "epoch": 0.02, + "learning_rate": 3.789581836037692e-05, + "loss": 3.6298, + "step": 739328 + }, + { + "epoch": 0.02, + "learning_rate": 3.78874324128664e-05, + "loss": 3.6267, + "step": 739840 + }, + { + "epoch": 0.02, + "learning_rate": 3.787906284415961e-05, + "loss": 3.6329, + "step": 740352 + }, + { + "epoch": 0.02, + "learning_rate": 3.787067689664909e-05, + "loss": 3.6268, + "step": 740864 + }, + { + "epoch": 0.02, + "learning_rate": 3.786229094913857e-05, + "loss": 3.6237, + "step": 741376 + }, + { + "epoch": 0.02, + "learning_rate": 3.785390500162805e-05, + "loss": 3.6432, + "step": 741888 + }, + { + "epoch": 0.02, + "learning_rate": 3.784551905411753e-05, + "loss": 3.6274, + "step": 742400 + }, + { + "epoch": 0.02, + "learning_rate": 3.783714948541074e-05, + "loss": 3.6314, + "step": 742912 + }, + { + "epoch": 0.02, + "learning_rate": 3.782876353790023e-05, + "loss": 3.6265, + "step": 743424 + }, + { + "epoch": 0.02, + "learning_rate": 3.782037759038971e-05, + "loss": 3.6236, + "step": 743936 + }, + { + "epoch": 0.02, + "learning_rate": 3.781200802168292e-05, + "loss": 3.6426, + "step": 744448 + }, + { + "epoch": 0.02, + "learning_rate": 3.78036220741724e-05, + "loss": 3.6299, + "step": 744960 + }, + { + "epoch": 0.02, + "learning_rate": 3.779523612666188e-05, + "loss": 3.6459, + "step": 745472 + }, + { + "epoch": 0.02, + "learning_rate": 3.778685017915136e-05, + "loss": 3.6418, + "step": 745984 + }, + { + "epoch": 0.02, + "learning_rate": 3.7778480610444566e-05, + "loss": 3.6346, + "step": 746496 + }, + { + "epoch": 0.02, + "learning_rate": 3.7770094662934046e-05, + "loss": 3.633, + "step": 747008 + }, + { + "epoch": 0.02, + "learning_rate": 3.7761708715423526e-05, + "loss": 3.6346, + "step": 747520 + }, + { + "epoch": 0.02, + "learning_rate": 3.7753322767913006e-05, + "loss": 3.627, + "step": 748032 + }, + { + "epoch": 0.02, + "learning_rate": 3.7744953199206215e-05, + "loss": 3.6388, + "step": 748544 + }, + { + "epoch": 0.02, + "learning_rate": 3.7736567251695695e-05, + "loss": 3.6317, + "step": 749056 + }, + { + "epoch": 0.02, + "learning_rate": 3.772818130418518e-05, + "loss": 3.6338, + "step": 749568 + }, + { + "epoch": 0.02, + "learning_rate": 3.771979535667466e-05, + "loss": 3.6288, + "step": 750080 + }, + { + "epoch": 0.02, + "learning_rate": 3.771142578796787e-05, + "loss": 3.6324, + "step": 750592 + }, + { + "epoch": 0.02, + "learning_rate": 3.770303984045735e-05, + "loss": 3.6329, + "step": 751104 + }, + { + "epoch": 0.02, + "learning_rate": 3.769465389294683e-05, + "loss": 3.6254, + "step": 751616 + }, + { + "epoch": 0.02, + "learning_rate": 3.768626794543631e-05, + "loss": 3.6357, + "step": 752128 + }, + { + "epoch": 0.02, + "learning_rate": 3.767789837672952e-05, + "loss": 3.6172, + "step": 752640 + }, + { + "epoch": 0.02, + "learning_rate": 3.7669512429219e-05, + "loss": 3.651, + "step": 753152 + }, + { + "epoch": 0.02, + "learning_rate": 3.766112648170848e-05, + "loss": 3.6296, + "step": 753664 + }, + { + "epoch": 0.02, + "learning_rate": 3.765274053419796e-05, + "loss": 3.6376, + "step": 754176 + }, + { + "epoch": 0.02, + "learning_rate": 3.764437096549117e-05, + "loss": 3.6259, + "step": 754688 + }, + { + "epoch": 0.02, + "learning_rate": 3.763598501798065e-05, + "loss": 3.641, + "step": 755200 + }, + { + "epoch": 0.02, + "learning_rate": 3.7627599070470135e-05, + "loss": 3.63, + "step": 755712 + }, + { + "epoch": 0.02, + "learning_rate": 3.7619213122959615e-05, + "loss": 3.6344, + "step": 756224 + }, + { + "epoch": 0.02, + "learning_rate": 3.7610843554252824e-05, + "loss": 3.6295, + "step": 756736 + }, + { + "epoch": 0.02, + "learning_rate": 3.7602457606742304e-05, + "loss": 3.6327, + "step": 757248 + }, + { + "epoch": 0.02, + "learning_rate": 3.7594071659231784e-05, + "loss": 3.6343, + "step": 757760 + }, + { + "epoch": 0.02, + "learning_rate": 3.7585685711721264e-05, + "loss": 3.6518, + "step": 758272 + }, + { + "epoch": 0.02, + "learning_rate": 3.757731614301447e-05, + "loss": 3.6237, + "step": 758784 + }, + { + "epoch": 0.02, + "learning_rate": 3.756893019550395e-05, + "loss": 3.6406, + "step": 759296 + }, + { + "epoch": 0.02, + "learning_rate": 3.756054424799343e-05, + "loss": 3.6379, + "step": 759808 + }, + { + "epoch": 0.02, + "learning_rate": 3.755215830048291e-05, + "loss": 3.6336, + "step": 760320 + }, + { + "epoch": 0.02, + "learning_rate": 3.754378873177612e-05, + "loss": 3.6254, + "step": 760832 + }, + { + "epoch": 0.02, + "learning_rate": 3.75354027842656e-05, + "loss": 3.6368, + "step": 761344 + }, + { + "epoch": 0.02, + "learning_rate": 3.752701683675509e-05, + "loss": 3.6326, + "step": 761856 + }, + { + "epoch": 0.02, + "learning_rate": 3.751863088924457e-05, + "loss": 3.6266, + "step": 762368 + }, + { + "epoch": 0.02, + "learning_rate": 3.751026132053778e-05, + "loss": 3.633, + "step": 762880 + }, + { + "epoch": 0.03, + "eval_loss": 3.8511483669281006, + "eval_runtime": 303.7288, + "eval_samples_per_second": 1256.354, + "eval_steps_per_second": 39.262, + "step": 763200 + }, + { + "epoch": 1.0, + "learning_rate": 3.750187537302726e-05, + "loss": 3.6252, + "step": 763392 + }, + { + "epoch": 1.0, + "learning_rate": 3.749348942551674e-05, + "loss": 3.6155, + "step": 763904 + }, + { + "epoch": 1.0, + "learning_rate": 3.748510347800622e-05, + "loss": 3.6333, + "step": 764416 + }, + { + "epoch": 1.0, + "learning_rate": 3.747673390929943e-05, + "loss": 3.6452, + "step": 764928 + }, + { + "epoch": 1.0, + "learning_rate": 3.746834796178891e-05, + "loss": 3.6412, + "step": 765440 + }, + { + "epoch": 1.0, + "learning_rate": 3.745996201427839e-05, + "loss": 3.6235, + "step": 765952 + }, + { + "epoch": 1.0, + "learning_rate": 3.745157606676787e-05, + "loss": 3.6333, + "step": 766464 + }, + { + "epoch": 1.0, + "learning_rate": 3.7443206498061076e-05, + "loss": 3.624, + "step": 766976 + }, + { + "epoch": 1.0, + "learning_rate": 3.7434820550550556e-05, + "loss": 3.6236, + "step": 767488 + }, + { + "epoch": 1.0, + "learning_rate": 3.742645098184377e-05, + "loss": 3.6373, + "step": 768000 + }, + { + "epoch": 1.0, + "learning_rate": 3.741806503433325e-05, + "loss": 3.627, + "step": 768512 + }, + { + "epoch": 1.0, + "learning_rate": 3.740967908682273e-05, + "loss": 3.6312, + "step": 769024 + }, + { + "epoch": 1.0, + "learning_rate": 3.740129313931221e-05, + "loss": 3.6333, + "step": 769536 + }, + { + "epoch": 1.0, + "learning_rate": 3.739290719180169e-05, + "loss": 3.619, + "step": 770048 + }, + { + "epoch": 1.0, + "learning_rate": 3.738452124429117e-05, + "loss": 3.619, + "step": 770560 + }, + { + "epoch": 1.0, + "learning_rate": 3.737613529678065e-05, + "loss": 3.6114, + "step": 771072 + }, + { + "epoch": 1.0, + "learning_rate": 3.736774934927013e-05, + "loss": 3.6257, + "step": 771584 + }, + { + "epoch": 1.0, + "learning_rate": 3.735937978056334e-05, + "loss": 3.6278, + "step": 772096 + }, + { + "epoch": 1.0, + "learning_rate": 3.735099383305282e-05, + "loss": 3.6191, + "step": 772608 + }, + { + "epoch": 1.0, + "learning_rate": 3.73426078855423e-05, + "loss": 3.6333, + "step": 773120 + }, + { + "epoch": 1.0, + "learning_rate": 3.733422193803178e-05, + "loss": 3.6383, + "step": 773632 + }, + { + "epoch": 1.0, + "learning_rate": 3.7325852369324996e-05, + "loss": 3.6337, + "step": 774144 + }, + { + "epoch": 1.0, + "learning_rate": 3.7317466421814476e-05, + "loss": 3.6289, + "step": 774656 + }, + { + "epoch": 1.0, + "learning_rate": 3.7309080474303956e-05, + "loss": 3.6276, + "step": 775168 + }, + { + "epoch": 1.0, + "learning_rate": 3.7300694526793436e-05, + "loss": 3.6292, + "step": 775680 + }, + { + "epoch": 1.0, + "learning_rate": 3.7292324958086645e-05, + "loss": 3.6269, + "step": 776192 + }, + { + "epoch": 1.0, + "learning_rate": 3.7283939010576125e-05, + "loss": 3.6222, + "step": 776704 + }, + { + "epoch": 1.0, + "learning_rate": 3.7275553063065605e-05, + "loss": 3.6205, + "step": 777216 + }, + { + "epoch": 1.0, + "learning_rate": 3.7267167115555085e-05, + "loss": 3.6225, + "step": 777728 + }, + { + "epoch": 1.0, + "learning_rate": 3.7258797546848294e-05, + "loss": 3.6197, + "step": 778240 + }, + { + "epoch": 1.01, + "learning_rate": 3.7250411599337774e-05, + "loss": 3.6213, + "step": 778752 + }, + { + "epoch": 1.01, + "learning_rate": 3.7242025651827254e-05, + "loss": 3.629, + "step": 779264 + }, + { + "epoch": 1.01, + "learning_rate": 3.7233639704316734e-05, + "loss": 3.6298, + "step": 779776 + }, + { + "epoch": 1.01, + "learning_rate": 3.722527013560994e-05, + "loss": 3.6325, + "step": 780288 + }, + { + "epoch": 1.01, + "learning_rate": 3.721688418809943e-05, + "loss": 3.6166, + "step": 780800 + }, + { + "epoch": 1.01, + "learning_rate": 3.720849824058891e-05, + "loss": 3.6236, + "step": 781312 + }, + { + "epoch": 1.01, + "learning_rate": 3.720011229307839e-05, + "loss": 3.6301, + "step": 781824 + }, + { + "epoch": 1.01, + "learning_rate": 3.71917427243716e-05, + "loss": 3.6133, + "step": 782336 + }, + { + "epoch": 1.01, + "learning_rate": 3.718335677686108e-05, + "loss": 3.6221, + "step": 782848 + }, + { + "epoch": 1.01, + "learning_rate": 3.717497082935056e-05, + "loss": 3.6116, + "step": 783360 + }, + { + "epoch": 1.01, + "learning_rate": 3.716660126064377e-05, + "loss": 3.6235, + "step": 783872 + }, + { + "epoch": 1.01, + "learning_rate": 3.715821531313325e-05, + "loss": 3.6211, + "step": 784384 + }, + { + "epoch": 1.01, + "learning_rate": 3.714982936562273e-05, + "loss": 3.6206, + "step": 784896 + }, + { + "epoch": 1.01, + "learning_rate": 3.714144341811221e-05, + "loss": 3.6267, + "step": 785408 + }, + { + "epoch": 1.01, + "learning_rate": 3.713307384940542e-05, + "loss": 3.6275, + "step": 785920 + }, + { + "epoch": 1.01, + "learning_rate": 3.71246879018949e-05, + "loss": 3.628, + "step": 786432 + }, + { + "epoch": 1.01, + "learning_rate": 3.7116301954384384e-05, + "loss": 3.6285, + "step": 786944 + }, + { + "epoch": 1.01, + "learning_rate": 3.7107916006873863e-05, + "loss": 3.6197, + "step": 787456 + }, + { + "epoch": 1.01, + "learning_rate": 3.709954643816707e-05, + "loss": 3.5939, + "step": 787968 + }, + { + "epoch": 1.01, + "learning_rate": 3.709116049065655e-05, + "loss": 3.6326, + "step": 788480 + }, + { + "epoch": 1.01, + "learning_rate": 3.708277454314603e-05, + "loss": 3.6221, + "step": 788992 + }, + { + "epoch": 1.01, + "learning_rate": 3.707438859563551e-05, + "loss": 3.6288, + "step": 789504 + }, + { + "epoch": 1.01, + "learning_rate": 3.706601902692872e-05, + "loss": 3.6214, + "step": 790016 + }, + { + "epoch": 1.01, + "learning_rate": 3.70576330794182e-05, + "loss": 3.6139, + "step": 790528 + }, + { + "epoch": 1.01, + "learning_rate": 3.704924713190768e-05, + "loss": 3.604, + "step": 791040 + }, + { + "epoch": 1.01, + "learning_rate": 3.7040861184397155e-05, + "loss": 3.6137, + "step": 791552 + }, + { + "epoch": 1.01, + "learning_rate": 3.703249161569037e-05, + "loss": 3.6168, + "step": 792064 + }, + { + "epoch": 1.01, + "learning_rate": 3.702410566817985e-05, + "loss": 3.6085, + "step": 792576 + }, + { + "epoch": 1.01, + "learning_rate": 3.701571972066934e-05, + "loss": 3.6322, + "step": 793088 + }, + { + "epoch": 1.01, + "learning_rate": 3.700733377315881e-05, + "loss": 3.619, + "step": 793600 + }, + { + "epoch": 1.01, + "learning_rate": 3.6998964204452026e-05, + "loss": 3.6008, + "step": 794112 + }, + { + "epoch": 1.01, + "learning_rate": 3.6990578256941506e-05, + "loss": 3.6284, + "step": 794624 + }, + { + "epoch": 1.01, + "learning_rate": 3.6982192309430986e-05, + "loss": 3.6046, + "step": 795136 + }, + { + "epoch": 1.01, + "learning_rate": 3.697380636192046e-05, + "loss": 3.605, + "step": 795648 + }, + { + "epoch": 1.01, + "learning_rate": 3.6965436793213675e-05, + "loss": 3.6224, + "step": 796160 + }, + { + "epoch": 1.01, + "learning_rate": 3.6957050845703155e-05, + "loss": 3.6173, + "step": 796672 + }, + { + "epoch": 1.01, + "learning_rate": 3.694866489819263e-05, + "loss": 3.6083, + "step": 797184 + }, + { + "epoch": 1.01, + "learning_rate": 3.694027895068211e-05, + "loss": 3.6081, + "step": 797696 + }, + { + "epoch": 1.01, + "learning_rate": 3.6931909381975324e-05, + "loss": 3.6103, + "step": 798208 + }, + { + "epoch": 1.01, + "learning_rate": 3.6923523434464804e-05, + "loss": 3.6141, + "step": 798720 + }, + { + "epoch": 1.01, + "learning_rate": 3.6915137486954284e-05, + "loss": 3.6217, + "step": 799232 + }, + { + "epoch": 1.01, + "learning_rate": 3.6906751539443764e-05, + "loss": 3.6137, + "step": 799744 + }, + { + "epoch": 1.01, + "learning_rate": 3.689838197073698e-05, + "loss": 3.6169, + "step": 800256 + }, + { + "epoch": 1.01, + "learning_rate": 3.688999602322646e-05, + "loss": 3.6155, + "step": 800768 + }, + { + "epoch": 1.01, + "learning_rate": 3.688161007571593e-05, + "loss": 3.6289, + "step": 801280 + }, + { + "epoch": 1.01, + "learning_rate": 3.687322412820541e-05, + "loss": 3.6092, + "step": 801792 + }, + { + "epoch": 1.01, + "learning_rate": 3.686485455949863e-05, + "loss": 3.6246, + "step": 802304 + }, + { + "epoch": 1.01, + "learning_rate": 3.68564686119881e-05, + "loss": 3.6194, + "step": 802816 + }, + { + "epoch": 1.01, + "learning_rate": 3.684808266447758e-05, + "loss": 3.6038, + "step": 803328 + }, + { + "epoch": 1.01, + "learning_rate": 3.683969671696706e-05, + "loss": 3.6175, + "step": 803840 + }, + { + "epoch": 1.01, + "learning_rate": 3.683132714826028e-05, + "loss": 3.6187, + "step": 804352 + }, + { + "epoch": 1.01, + "learning_rate": 3.682294120074976e-05, + "loss": 3.6221, + "step": 804864 + }, + { + "epoch": 1.01, + "learning_rate": 3.681455525323924e-05, + "loss": 3.6238, + "step": 805376 + }, + { + "epoch": 1.01, + "learning_rate": 3.680616930572872e-05, + "loss": 3.5947, + "step": 805888 + }, + { + "epoch": 1.01, + "learning_rate": 3.679779973702193e-05, + "loss": 3.5986, + "step": 806400 + }, + { + "epoch": 1.01, + "learning_rate": 3.678941378951141e-05, + "loss": 3.6144, + "step": 806912 + }, + { + "epoch": 1.01, + "learning_rate": 3.678102784200089e-05, + "loss": 3.6192, + "step": 807424 + }, + { + "epoch": 1.01, + "learning_rate": 3.677264189449037e-05, + "loss": 3.6061, + "step": 807936 + }, + { + "epoch": 1.01, + "learning_rate": 3.6764272325783576e-05, + "loss": 3.6118, + "step": 808448 + }, + { + "epoch": 1.01, + "learning_rate": 3.6755886378273056e-05, + "loss": 3.6098, + "step": 808960 + }, + { + "epoch": 1.02, + "learning_rate": 3.6747500430762536e-05, + "loss": 3.6184, + "step": 809472 + }, + { + "epoch": 1.02, + "learning_rate": 3.6739114483252016e-05, + "loss": 3.6095, + "step": 809984 + }, + { + "epoch": 1.02, + "learning_rate": 3.673074491454523e-05, + "loss": 3.6103, + "step": 810496 + }, + { + "epoch": 1.02, + "learning_rate": 3.672235896703471e-05, + "loss": 3.6139, + "step": 811008 + }, + { + "epoch": 1.02, + "learning_rate": 3.671397301952419e-05, + "loss": 3.6212, + "step": 811520 + }, + { + "epoch": 1.02, + "learning_rate": 3.670558707201367e-05, + "loss": 3.6194, + "step": 812032 + }, + { + "epoch": 1.02, + "learning_rate": 3.669721750330688e-05, + "loss": 3.6012, + "step": 812544 + }, + { + "epoch": 1.02, + "learning_rate": 3.668883155579636e-05, + "loss": 3.6044, + "step": 813056 + }, + { + "epoch": 1.02, + "learning_rate": 3.668044560828584e-05, + "loss": 3.6133, + "step": 813568 + }, + { + "epoch": 1.02, + "learning_rate": 3.667205966077532e-05, + "loss": 3.6213, + "step": 814080 + }, + { + "epoch": 1.02, + "learning_rate": 3.666369009206853e-05, + "loss": 3.6156, + "step": 814592 + }, + { + "epoch": 1.02, + "learning_rate": 3.665530414455801e-05, + "loss": 3.6127, + "step": 815104 + }, + { + "epoch": 1.02, + "learning_rate": 3.664691819704749e-05, + "loss": 3.6068, + "step": 815616 + }, + { + "epoch": 1.02, + "learning_rate": 3.663853224953697e-05, + "loss": 3.6058, + "step": 816128 + }, + { + "epoch": 1.02, + "learning_rate": 3.6630162680830185e-05, + "loss": 3.6119, + "step": 816640 + }, + { + "epoch": 1.02, + "learning_rate": 3.6621776733319665e-05, + "loss": 3.6041, + "step": 817152 + }, + { + "epoch": 1.02, + "learning_rate": 3.6613390785809145e-05, + "loss": 3.6037, + "step": 817664 + }, + { + "epoch": 1.02, + "learning_rate": 3.6605004838298625e-05, + "loss": 3.6189, + "step": 818176 + }, + { + "epoch": 1.02, + "learning_rate": 3.6596635269591834e-05, + "loss": 3.6068, + "step": 818688 + }, + { + "epoch": 1.02, + "learning_rate": 3.6588249322081314e-05, + "loss": 3.6068, + "step": 819200 + }, + { + "epoch": 1.02, + "learning_rate": 3.6579863374570794e-05, + "loss": 3.6052, + "step": 819712 + }, + { + "epoch": 1.02, + "learning_rate": 3.6571477427060274e-05, + "loss": 3.603, + "step": 820224 + }, + { + "epoch": 1.02, + "learning_rate": 3.656310785835348e-05, + "loss": 3.6183, + "step": 820736 + }, + { + "epoch": 1.02, + "learning_rate": 3.655472191084296e-05, + "loss": 3.612, + "step": 821248 + }, + { + "epoch": 1.02, + "learning_rate": 3.654633596333244e-05, + "loss": 3.6199, + "step": 821760 + }, + { + "epoch": 1.02, + "learning_rate": 3.653795001582192e-05, + "loss": 3.6216, + "step": 822272 + }, + { + "epoch": 1.02, + "learning_rate": 3.652958044711514e-05, + "loss": 3.6154, + "step": 822784 + }, + { + "epoch": 1.02, + "learning_rate": 3.652119449960462e-05, + "loss": 3.6098, + "step": 823296 + }, + { + "epoch": 1.02, + "learning_rate": 3.65128085520941e-05, + "loss": 3.6125, + "step": 823808 + }, + { + "epoch": 1.02, + "learning_rate": 3.650442260458358e-05, + "loss": 3.6092, + "step": 824320 + }, + { + "epoch": 1.02, + "learning_rate": 3.649605303587679e-05, + "loss": 3.6166, + "step": 824832 + }, + { + "epoch": 1.02, + "learning_rate": 3.648766708836627e-05, + "loss": 3.6063, + "step": 825344 + }, + { + "epoch": 1.02, + "learning_rate": 3.647928114085575e-05, + "loss": 3.6167, + "step": 825856 + }, + { + "epoch": 1.02, + "learning_rate": 3.647089519334523e-05, + "loss": 3.6035, + "step": 826368 + }, + { + "epoch": 1.02, + "learning_rate": 3.646252562463844e-05, + "loss": 3.615, + "step": 826880 + }, + { + "epoch": 1.02, + "learning_rate": 3.645413967712792e-05, + "loss": 3.6095, + "step": 827392 + }, + { + "epoch": 1.02, + "learning_rate": 3.64457537296174e-05, + "loss": 3.6068, + "step": 827904 + }, + { + "epoch": 1.02, + "learning_rate": 3.643736778210688e-05, + "loss": 3.611, + "step": 828416 + }, + { + "epoch": 1.02, + "learning_rate": 3.642899821340009e-05, + "loss": 3.5958, + "step": 828928 + }, + { + "epoch": 1.02, + "learning_rate": 3.642061226588957e-05, + "loss": 3.6317, + "step": 829440 + }, + { + "epoch": 1.02, + "learning_rate": 3.641222631837905e-05, + "loss": 3.6079, + "step": 829952 + }, + { + "epoch": 1.02, + "learning_rate": 3.640384037086853e-05, + "loss": 3.6191, + "step": 830464 + }, + { + "epoch": 1.02, + "learning_rate": 3.639547080216174e-05, + "loss": 3.6041, + "step": 830976 + }, + { + "epoch": 1.02, + "learning_rate": 3.638708485465122e-05, + "loss": 3.6155, + "step": 831488 + }, + { + "epoch": 1.02, + "learning_rate": 3.63786989071407e-05, + "loss": 3.6094, + "step": 832000 + }, + { + "epoch": 1.02, + "learning_rate": 3.637031295963018e-05, + "loss": 3.6157, + "step": 832512 + }, + { + "epoch": 1.02, + "learning_rate": 3.636194339092339e-05, + "loss": 3.6064, + "step": 833024 + }, + { + "epoch": 1.02, + "learning_rate": 3.635355744341287e-05, + "loss": 3.611, + "step": 833536 + }, + { + "epoch": 1.02, + "learning_rate": 3.634517149590235e-05, + "loss": 3.6137, + "step": 834048 + }, + { + "epoch": 1.02, + "learning_rate": 3.633678554839183e-05, + "loss": 3.6346, + "step": 834560 + }, + { + "epoch": 1.02, + "learning_rate": 3.6328415979685046e-05, + "loss": 3.6007, + "step": 835072 + }, + { + "epoch": 1.02, + "learning_rate": 3.6320030032174526e-05, + "loss": 3.622, + "step": 835584 + }, + { + "epoch": 1.02, + "learning_rate": 3.6311644084664006e-05, + "loss": 3.6173, + "step": 836096 + }, + { + "epoch": 1.02, + "learning_rate": 3.6303258137153486e-05, + "loss": 3.6094, + "step": 836608 + }, + { + "epoch": 1.02, + "learning_rate": 3.6294904947250424e-05, + "loss": 3.6045, + "step": 837120 + }, + { + "epoch": 1.02, + "learning_rate": 3.6286518999739904e-05, + "loss": 3.614, + "step": 837632 + }, + { + "epoch": 1.02, + "learning_rate": 3.6278133052229384e-05, + "loss": 3.6116, + "step": 838144 + }, + { + "epoch": 1.02, + "learning_rate": 3.6269747104718864e-05, + "loss": 3.6082, + "step": 838656 + }, + { + "epoch": 1.02, + "learning_rate": 3.6261361157208344e-05, + "loss": 3.6086, + "step": 839168 + }, + { + "epoch": 1.03, + "eval_loss": 3.8488876819610596, + "eval_runtime": 304.1493, + "eval_samples_per_second": 1254.617, + "eval_steps_per_second": 39.208, + "step": 839520 + }, + { + "epoch": 0.0, + "learning_rate": 3.6252975209697824e-05, + "loss": 3.6047, + "step": 839680 + }, + { + "epoch": 0.0, + "learning_rate": 3.6244589262187304e-05, + "loss": 3.5969, + "step": 840192 + }, + { + "epoch": 0.0, + "learning_rate": 3.6236203314676784e-05, + "loss": 3.6092, + "step": 840704 + }, + { + "epoch": 0.0, + "learning_rate": 3.6227817367166264e-05, + "loss": 3.6246, + "step": 841216 + }, + { + "epoch": 0.0, + "learning_rate": 3.621944779845948e-05, + "loss": 3.6217, + "step": 841728 + }, + { + "epoch": 0.0, + "learning_rate": 3.621106185094896e-05, + "loss": 3.6023, + "step": 842240 + }, + { + "epoch": 0.0, + "learning_rate": 3.620267590343844e-05, + "loss": 3.6133, + "step": 842752 + }, + { + "epoch": 0.0, + "learning_rate": 3.619428995592792e-05, + "loss": 3.6039, + "step": 843264 + }, + { + "epoch": 0.0, + "learning_rate": 3.618592038722113e-05, + "loss": 3.6018, + "step": 843776 + }, + { + "epoch": 0.0, + "learning_rate": 3.617755081851434e-05, + "loss": 3.6118, + "step": 844288 + }, + { + "epoch": 0.0, + "learning_rate": 3.616916487100382e-05, + "loss": 3.6142, + "step": 844800 + }, + { + "epoch": 0.0, + "learning_rate": 3.61607789234933e-05, + "loss": 3.6115, + "step": 845312 + }, + { + "epoch": 0.0, + "learning_rate": 3.615239297598278e-05, + "loss": 3.6087, + "step": 845824 + }, + { + "epoch": 0.0, + "learning_rate": 3.614400702847226e-05, + "loss": 3.5997, + "step": 846336 + }, + { + "epoch": 0.0, + "learning_rate": 3.613562108096174e-05, + "loss": 3.5992, + "step": 846848 + }, + { + "epoch": 0.0, + "learning_rate": 3.6127251512254953e-05, + "loss": 3.5886, + "step": 847360 + }, + { + "epoch": 0.0, + "learning_rate": 3.6118865564744433e-05, + "loss": 3.6063, + "step": 847872 + }, + { + "epoch": 0.0, + "learning_rate": 3.6110479617233913e-05, + "loss": 3.6062, + "step": 848384 + }, + { + "epoch": 0.0, + "learning_rate": 3.610209366972339e-05, + "loss": 3.5972, + "step": 848896 + }, + { + "epoch": 0.0, + "learning_rate": 3.609370772221287e-05, + "loss": 3.6103, + "step": 849408 + }, + { + "epoch": 0.0, + "learning_rate": 3.608532177470235e-05, + "loss": 3.6176, + "step": 849920 + }, + { + "epoch": 0.0, + "learning_rate": 3.607693582719183e-05, + "loss": 3.6105, + "step": 850432 + }, + { + "epoch": 0.0, + "learning_rate": 3.6068549879681306e-05, + "loss": 3.6098, + "step": 850944 + }, + { + "epoch": 0.0, + "learning_rate": 3.606018031097452e-05, + "loss": 3.6032, + "step": 851456 + }, + { + "epoch": 0.0, + "learning_rate": 3.6051794363464e-05, + "loss": 3.6099, + "step": 851968 + }, + { + "epoch": 0.0, + "learning_rate": 3.6043408415953475e-05, + "loss": 3.6062, + "step": 852480 + }, + { + "epoch": 0.0, + "learning_rate": 3.6035022468442955e-05, + "loss": 3.601, + "step": 852992 + }, + { + "epoch": 0.0, + "learning_rate": 3.602665289973617e-05, + "loss": 3.6008, + "step": 853504 + }, + { + "epoch": 0.0, + "learning_rate": 3.601826695222565e-05, + "loss": 3.6015, + "step": 854016 + }, + { + "epoch": 0.0, + "learning_rate": 3.600988100471513e-05, + "loss": 3.5981, + "step": 854528 + }, + { + "epoch": 0.01, + "learning_rate": 3.600149505720461e-05, + "loss": 3.6061, + "step": 855040 + }, + { + "epoch": 0.01, + "learning_rate": 3.599312548849783e-05, + "loss": 3.6063, + "step": 855552 + }, + { + "epoch": 0.01, + "learning_rate": 3.598473954098731e-05, + "loss": 3.6057, + "step": 856064 + }, + { + "epoch": 0.01, + "learning_rate": 3.597635359347678e-05, + "loss": 3.6162, + "step": 856576 + }, + { + "epoch": 0.01, + "learning_rate": 3.596796764596626e-05, + "loss": 3.5947, + "step": 857088 + }, + { + "epoch": 0.01, + "learning_rate": 3.5959598077259476e-05, + "loss": 3.6012, + "step": 857600 + }, + { + "epoch": 0.01, + "learning_rate": 3.595121212974895e-05, + "loss": 3.6102, + "step": 858112 + }, + { + "epoch": 0.01, + "learning_rate": 3.594282618223843e-05, + "loss": 3.5915, + "step": 858624 + }, + { + "epoch": 0.01, + "learning_rate": 3.5934456613531645e-05, + "loss": 3.6025, + "step": 859136 + }, + { + "epoch": 0.01, + "learning_rate": 3.5926070666021125e-05, + "loss": 3.5921, + "step": 859648 + }, + { + "epoch": 0.01, + "learning_rate": 3.5917684718510605e-05, + "loss": 3.597, + "step": 860160 + }, + { + "epoch": 0.01, + "learning_rate": 3.5909298771000085e-05, + "loss": 3.6013, + "step": 860672 + }, + { + "epoch": 0.01, + "learning_rate": 3.59009292022933e-05, + "loss": 3.6025, + "step": 861184 + }, + { + "epoch": 0.01, + "learning_rate": 3.589254325478278e-05, + "loss": 3.6087, + "step": 861696 + }, + { + "epoch": 0.01, + "learning_rate": 3.5884157307272254e-05, + "loss": 3.6059, + "step": 862208 + }, + { + "epoch": 0.01, + "learning_rate": 3.5875771359761734e-05, + "loss": 3.6073, + "step": 862720 + }, + { + "epoch": 0.01, + "learning_rate": 3.586740179105495e-05, + "loss": 3.6077, + "step": 863232 + }, + { + "epoch": 0.01, + "learning_rate": 3.585901584354442e-05, + "loss": 3.5993, + "step": 863744 + }, + { + "epoch": 0.01, + "learning_rate": 3.58506298960339e-05, + "loss": 3.5752, + "step": 864256 + }, + { + "epoch": 0.01, + "learning_rate": 3.584224394852338e-05, + "loss": 3.6114, + "step": 864768 + }, + { + "epoch": 0.01, + "learning_rate": 3.58338743798166e-05, + "loss": 3.6, + "step": 865280 + }, + { + "epoch": 0.01, + "learning_rate": 3.582548843230608e-05, + "loss": 3.609, + "step": 865792 + }, + { + "epoch": 0.01, + "learning_rate": 3.581710248479556e-05, + "loss": 3.5984, + "step": 866304 + }, + { + "epoch": 0.01, + "learning_rate": 3.580871653728504e-05, + "loss": 3.5919, + "step": 866816 + }, + { + "epoch": 0.01, + "learning_rate": 3.5800346968578254e-05, + "loss": 3.5922, + "step": 867328 + }, + { + "epoch": 0.01, + "learning_rate": 3.579196102106773e-05, + "loss": 3.5873, + "step": 867840 + }, + { + "epoch": 0.01, + "learning_rate": 3.578357507355721e-05, + "loss": 3.5985, + "step": 868352 + }, + { + "epoch": 0.01, + "learning_rate": 3.577518912604669e-05, + "loss": 3.584, + "step": 868864 + }, + { + "epoch": 0.01, + "learning_rate": 3.5766819557339897e-05, + "loss": 3.6166, + "step": 869376 + }, + { + "epoch": 0.01, + "learning_rate": 3.5758433609829377e-05, + "loss": 3.6009, + "step": 869888 + }, + { + "epoch": 0.01, + "learning_rate": 3.5750047662318856e-05, + "loss": 3.5799, + "step": 870400 + }, + { + "epoch": 0.01, + "learning_rate": 3.5741661714808336e-05, + "loss": 3.6084, + "step": 870912 + }, + { + "epoch": 0.01, + "learning_rate": 3.573329214610155e-05, + "loss": 3.5852, + "step": 871424 + }, + { + "epoch": 0.01, + "learning_rate": 3.572490619859103e-05, + "loss": 3.5868, + "step": 871936 + }, + { + "epoch": 0.01, + "learning_rate": 3.571652025108051e-05, + "loss": 3.5982, + "step": 872448 + }, + { + "epoch": 0.01, + "learning_rate": 3.570813430356999e-05, + "loss": 3.5967, + "step": 872960 + }, + { + "epoch": 0.01, + "learning_rate": 3.56997647348632e-05, + "loss": 3.5881, + "step": 873472 + }, + { + "epoch": 0.01, + "learning_rate": 3.569137878735268e-05, + "loss": 3.5914, + "step": 873984 + }, + { + "epoch": 0.01, + "learning_rate": 3.568299283984216e-05, + "loss": 3.5858, + "step": 874496 + }, + { + "epoch": 0.01, + "learning_rate": 3.567460689233164e-05, + "loss": 3.5958, + "step": 875008 + }, + { + "epoch": 0.01, + "learning_rate": 3.566623732362485e-05, + "loss": 3.6011, + "step": 875520 + }, + { + "epoch": 0.01, + "learning_rate": 3.565785137611433e-05, + "loss": 3.5929, + "step": 876032 + }, + { + "epoch": 0.01, + "learning_rate": 3.564946542860381e-05, + "loss": 3.5998, + "step": 876544 + }, + { + "epoch": 0.01, + "learning_rate": 3.564107948109329e-05, + "loss": 3.591, + "step": 877056 + }, + { + "epoch": 0.01, + "learning_rate": 3.5632709912386506e-05, + "loss": 3.6148, + "step": 877568 + }, + { + "epoch": 0.01, + "learning_rate": 3.5624323964875986e-05, + "loss": 3.5853, + "step": 878080 + }, + { + "epoch": 0.01, + "learning_rate": 3.5615938017365466e-05, + "loss": 3.6067, + "step": 878592 + }, + { + "epoch": 0.01, + "learning_rate": 3.5607552069854946e-05, + "loss": 3.5909, + "step": 879104 + }, + { + "epoch": 0.01, + "learning_rate": 3.5599182501148155e-05, + "loss": 3.5915, + "step": 879616 + }, + { + "epoch": 0.01, + "learning_rate": 3.5590796553637635e-05, + "loss": 3.5957, + "step": 880128 + }, + { + "epoch": 0.01, + "learning_rate": 3.5582410606127115e-05, + "loss": 3.6003, + "step": 880640 + }, + { + "epoch": 0.01, + "learning_rate": 3.5574024658616595e-05, + "loss": 3.6025, + "step": 881152 + }, + { + "epoch": 0.01, + "learning_rate": 3.5565655089909804e-05, + "loss": 3.6034, + "step": 881664 + }, + { + "epoch": 0.01, + "learning_rate": 3.5557269142399284e-05, + "loss": 3.577, + "step": 882176 + }, + { + "epoch": 0.01, + "learning_rate": 3.5548883194888764e-05, + "loss": 3.5774, + "step": 882688 + }, + { + "epoch": 0.01, + "learning_rate": 3.5540497247378244e-05, + "loss": 3.5933, + "step": 883200 + }, + { + "epoch": 0.01, + "learning_rate": 3.553212767867146e-05, + "loss": 3.5989, + "step": 883712 + }, + { + "epoch": 0.01, + "learning_rate": 3.552374173116094e-05, + "loss": 3.5883, + "step": 884224 + }, + { + "epoch": 0.01, + "learning_rate": 3.551535578365042e-05, + "loss": 3.5898, + "step": 884736 + }, + { + "epoch": 0.01, + "learning_rate": 3.55069698361399e-05, + "loss": 3.5922, + "step": 885248 + }, + { + "epoch": 0.02, + "learning_rate": 3.549860026743311e-05, + "loss": 3.5946, + "step": 885760 + }, + { + "epoch": 0.02, + "learning_rate": 3.549021431992259e-05, + "loss": 3.5946, + "step": 886272 + }, + { + "epoch": 0.02, + "learning_rate": 3.548182837241207e-05, + "loss": 3.591, + "step": 886784 + }, + { + "epoch": 0.02, + "learning_rate": 3.547344242490155e-05, + "loss": 3.5937, + "step": 887296 + }, + { + "epoch": 0.02, + "learning_rate": 3.546507285619476e-05, + "loss": 3.5997, + "step": 887808 + }, + { + "epoch": 0.02, + "learning_rate": 3.545668690868424e-05, + "loss": 3.6038, + "step": 888320 + }, + { + "epoch": 0.02, + "learning_rate": 3.544830096117372e-05, + "loss": 3.5808, + "step": 888832 + }, + { + "epoch": 0.02, + "learning_rate": 3.54399150136632e-05, + "loss": 3.5848, + "step": 889344 + }, + { + "epoch": 0.02, + "learning_rate": 3.543154544495641e-05, + "loss": 3.5933, + "step": 889856 + }, + { + "epoch": 0.02, + "learning_rate": 3.542315949744589e-05, + "loss": 3.5971, + "step": 890368 + }, + { + "epoch": 0.02, + "learning_rate": 3.541477354993537e-05, + "loss": 3.597, + "step": 890880 + }, + { + "epoch": 0.02, + "learning_rate": 3.540638760242485e-05, + "loss": 3.592, + "step": 891392 + }, + { + "epoch": 0.02, + "learning_rate": 3.539801803371806e-05, + "loss": 3.5888, + "step": 891904 + }, + { + "epoch": 0.02, + "learning_rate": 3.538963208620754e-05, + "loss": 3.5855, + "step": 892416 + }, + { + "epoch": 0.02, + "learning_rate": 3.538124613869702e-05, + "loss": 3.5919, + "step": 892928 + }, + { + "epoch": 0.02, + "learning_rate": 3.53728601911865e-05, + "loss": 3.5867, + "step": 893440 + }, + { + "epoch": 0.02, + "learning_rate": 3.536449062247971e-05, + "loss": 3.5835, + "step": 893952 + }, + { + "epoch": 0.02, + "learning_rate": 3.535610467496919e-05, + "loss": 3.5959, + "step": 894464 + }, + { + "epoch": 0.02, + "learning_rate": 3.534771872745867e-05, + "loss": 3.5905, + "step": 894976 + }, + { + "epoch": 0.02, + "learning_rate": 3.533933277994815e-05, + "loss": 3.5868, + "step": 895488 + }, + { + "epoch": 0.02, + "learning_rate": 3.533096321124136e-05, + "loss": 3.5865, + "step": 896000 + }, + { + "epoch": 0.02, + "learning_rate": 3.532257726373085e-05, + "loss": 3.5813, + "step": 896512 + }, + { + "epoch": 0.02, + "learning_rate": 3.531419131622033e-05, + "loss": 3.6009, + "step": 897024 + }, + { + "epoch": 0.02, + "learning_rate": 3.530580536870981e-05, + "loss": 3.5918, + "step": 897536 + }, + { + "epoch": 0.02, + "learning_rate": 3.5297435800003016e-05, + "loss": 3.5984, + "step": 898048 + }, + { + "epoch": 0.02, + "learning_rate": 3.5289049852492496e-05, + "loss": 3.5956, + "step": 898560 + }, + { + "epoch": 0.02, + "learning_rate": 3.5280663904981976e-05, + "loss": 3.5992, + "step": 899072 + }, + { + "epoch": 0.02, + "learning_rate": 3.5272294336275185e-05, + "loss": 3.5939, + "step": 899584 + }, + { + "epoch": 0.02, + "learning_rate": 3.5263908388764665e-05, + "loss": 3.5874, + "step": 900096 + }, + { + "epoch": 0.02, + "learning_rate": 3.5255522441254145e-05, + "loss": 3.5956, + "step": 900608 + }, + { + "epoch": 0.02, + "learning_rate": 3.5247136493743625e-05, + "loss": 3.5921, + "step": 901120 + }, + { + "epoch": 0.02, + "learning_rate": 3.5238766925036834e-05, + "loss": 3.5876, + "step": 901632 + }, + { + "epoch": 0.02, + "learning_rate": 3.5230380977526314e-05, + "loss": 3.5986, + "step": 902144 + }, + { + "epoch": 0.02, + "learning_rate": 3.52219950300158e-05, + "loss": 3.587, + "step": 902656 + }, + { + "epoch": 0.02, + "learning_rate": 3.521360908250528e-05, + "loss": 3.5934, + "step": 903168 + }, + { + "epoch": 0.02, + "learning_rate": 3.520523951379849e-05, + "loss": 3.5883, + "step": 903680 + }, + { + "epoch": 0.02, + "learning_rate": 3.519685356628797e-05, + "loss": 3.5862, + "step": 904192 + }, + { + "epoch": 0.02, + "learning_rate": 3.518846761877745e-05, + "loss": 3.5902, + "step": 904704 + }, + { + "epoch": 0.02, + "learning_rate": 3.518008167126693e-05, + "loss": 3.5819, + "step": 905216 + }, + { + "epoch": 0.02, + "learning_rate": 3.517171210256014e-05, + "loss": 3.6092, + "step": 905728 + }, + { + "epoch": 0.02, + "learning_rate": 3.516332615504962e-05, + "loss": 3.5915, + "step": 906240 + }, + { + "epoch": 0.02, + "learning_rate": 3.51549402075391e-05, + "loss": 3.5975, + "step": 906752 + }, + { + "epoch": 0.02, + "learning_rate": 3.514655426002858e-05, + "loss": 3.5841, + "step": 907264 + }, + { + "epoch": 0.02, + "learning_rate": 3.513818469132179e-05, + "loss": 3.597, + "step": 907776 + }, + { + "epoch": 0.02, + "learning_rate": 3.512979874381127e-05, + "loss": 3.5914, + "step": 908288 + }, + { + "epoch": 0.02, + "learning_rate": 3.5121412796300754e-05, + "loss": 3.5947, + "step": 908800 + }, + { + "epoch": 0.02, + "learning_rate": 3.5113026848790234e-05, + "loss": 3.5894, + "step": 909312 + }, + { + "epoch": 0.02, + "learning_rate": 3.510465728008344e-05, + "loss": 3.5857, + "step": 909824 + }, + { + "epoch": 0.02, + "learning_rate": 3.509627133257292e-05, + "loss": 3.5927, + "step": 910336 + }, + { + "epoch": 0.02, + "learning_rate": 3.50878853850624e-05, + "loss": 3.6143, + "step": 910848 + }, + { + "epoch": 0.02, + "learning_rate": 3.507949943755188e-05, + "loss": 3.5863, + "step": 911360 + }, + { + "epoch": 0.02, + "learning_rate": 3.507112986884509e-05, + "loss": 3.6019, + "step": 911872 + }, + { + "epoch": 0.02, + "learning_rate": 3.506274392133457e-05, + "loss": 3.5995, + "step": 912384 + }, + { + "epoch": 0.02, + "learning_rate": 3.505435797382405e-05, + "loss": 3.5882, + "step": 912896 + }, + { + "epoch": 0.02, + "learning_rate": 3.504598840511726e-05, + "loss": 3.5932, + "step": 913408 + }, + { + "epoch": 0.02, + "learning_rate": 3.503760245760674e-05, + "loss": 3.5908, + "step": 913920 + }, + { + "epoch": 0.02, + "learning_rate": 3.502921651009622e-05, + "loss": 3.5921, + "step": 914432 + }, + { + "epoch": 0.02, + "learning_rate": 3.502083056258571e-05, + "loss": 3.5883, + "step": 914944 + }, + { + "epoch": 0.02, + "learning_rate": 3.501244461507519e-05, + "loss": 3.5888, + "step": 915456 + }, + { + "epoch": 0.03, + "eval_loss": 3.848485231399536, + "eval_runtime": 304.7796, + "eval_samples_per_second": 1252.023, + "eval_steps_per_second": 39.127, + "step": 915840 + }, + { + "epoch": 0.0, + "learning_rate": 3.500405866756467e-05, + "loss": 3.6004, + "step": 915968 + }, + { + "epoch": 0.0, + "learning_rate": 3.499567272005415e-05, + "loss": 3.5724, + "step": 916480 + }, + { + "epoch": 0.0, + "learning_rate": 3.498728677254363e-05, + "loss": 3.5882, + "step": 916992 + }, + { + "epoch": 0.0, + "learning_rate": 3.49789008250331e-05, + "loss": 3.6064, + "step": 917504 + }, + { + "epoch": 0.0, + "learning_rate": 3.497053125632632e-05, + "loss": 3.5964, + "step": 918016 + }, + { + "epoch": 0.0, + "learning_rate": 3.49621453088158e-05, + "loss": 3.5871, + "step": 918528 + }, + { + "epoch": 0.0, + "learning_rate": 3.495375936130528e-05, + "loss": 3.5979, + "step": 919040 + }, + { + "epoch": 0.0, + "learning_rate": 3.494537341379475e-05, + "loss": 3.5827, + "step": 919552 + }, + { + "epoch": 0.0, + "learning_rate": 3.4937003845087966e-05, + "loss": 3.5803, + "step": 920064 + }, + { + "epoch": 0.0, + "learning_rate": 3.4928634276381175e-05, + "loss": 3.5954, + "step": 920576 + }, + { + "epoch": 0.0, + "learning_rate": 3.492024832887066e-05, + "loss": 3.5923, + "step": 921088 + }, + { + "epoch": 0.0, + "learning_rate": 3.491186238136014e-05, + "loss": 3.5882, + "step": 921600 + }, + { + "epoch": 0.0, + "learning_rate": 3.490347643384962e-05, + "loss": 3.5946, + "step": 922112 + }, + { + "epoch": 0.0, + "learning_rate": 3.48950904863391e-05, + "loss": 3.5757, + "step": 922624 + }, + { + "epoch": 0.0, + "learning_rate": 3.4886704538828575e-05, + "loss": 3.5798, + "step": 923136 + }, + { + "epoch": 0.0, + "learning_rate": 3.487833497012179e-05, + "loss": 3.572, + "step": 923648 + }, + { + "epoch": 0.0, + "learning_rate": 3.486994902261127e-05, + "loss": 3.5873, + "step": 924160 + }, + { + "epoch": 0.0, + "learning_rate": 3.4861563075100744e-05, + "loss": 3.5872, + "step": 924672 + }, + { + "epoch": 0.0, + "learning_rate": 3.4853177127590224e-05, + "loss": 3.5822, + "step": 925184 + }, + { + "epoch": 0.0, + "learning_rate": 3.4844791180079704e-05, + "loss": 3.5834, + "step": 925696 + }, + { + "epoch": 0.0, + "learning_rate": 3.4836405232569184e-05, + "loss": 3.6052, + "step": 926208 + }, + { + "epoch": 0.0, + "learning_rate": 3.4828019285058663e-05, + "loss": 3.5865, + "step": 926720 + }, + { + "epoch": 0.0, + "learning_rate": 3.4819633337548143e-05, + "loss": 3.5903, + "step": 927232 + }, + { + "epoch": 0.0, + "learning_rate": 3.481126376884136e-05, + "loss": 3.5872, + "step": 927744 + }, + { + "epoch": 0.0, + "learning_rate": 3.480287782133084e-05, + "loss": 3.5935, + "step": 928256 + }, + { + "epoch": 0.0, + "learning_rate": 3.479449187382032e-05, + "loss": 3.5851, + "step": 928768 + }, + { + "epoch": 0.0, + "learning_rate": 3.47861059263098e-05, + "loss": 3.5822, + "step": 929280 + }, + { + "epoch": 0.0, + "learning_rate": 3.477773635760301e-05, + "loss": 3.5815, + "step": 929792 + }, + { + "epoch": 0.0, + "learning_rate": 3.476935041009249e-05, + "loss": 3.5831, + "step": 930304 + }, + { + "epoch": 0.0, + "learning_rate": 3.476096446258197e-05, + "loss": 3.575, + "step": 930816 + }, + { + "epoch": 0.01, + "learning_rate": 3.475257851507145e-05, + "loss": 3.5864, + "step": 931328 + }, + { + "epoch": 0.01, + "learning_rate": 3.474420894636466e-05, + "loss": 3.5852, + "step": 931840 + }, + { + "epoch": 0.01, + "learning_rate": 3.473582299885414e-05, + "loss": 3.591, + "step": 932352 + }, + { + "epoch": 0.01, + "learning_rate": 3.472743705134362e-05, + "loss": 3.5932, + "step": 932864 + }, + { + "epoch": 0.01, + "learning_rate": 3.47190511038331e-05, + "loss": 3.5803, + "step": 933376 + }, + { + "epoch": 0.01, + "learning_rate": 3.471068153512631e-05, + "loss": 3.5842, + "step": 933888 + }, + { + "epoch": 0.01, + "learning_rate": 3.470229558761579e-05, + "loss": 3.592, + "step": 934400 + }, + { + "epoch": 0.01, + "learning_rate": 3.469390964010527e-05, + "loss": 3.5705, + "step": 934912 + }, + { + "epoch": 0.01, + "learning_rate": 3.468554007139848e-05, + "loss": 3.5837, + "step": 935424 + }, + { + "epoch": 0.01, + "learning_rate": 3.467717050269169e-05, + "loss": 3.5742, + "step": 935936 + }, + { + "epoch": 0.01, + "learning_rate": 3.466878455518117e-05, + "loss": 3.577, + "step": 936448 + }, + { + "epoch": 0.01, + "learning_rate": 3.466039860767065e-05, + "loss": 3.5844, + "step": 936960 + }, + { + "epoch": 0.01, + "learning_rate": 3.465201266016013e-05, + "loss": 3.5844, + "step": 937472 + }, + { + "epoch": 0.01, + "learning_rate": 3.464362671264961e-05, + "loss": 3.5886, + "step": 937984 + }, + { + "epoch": 0.01, + "learning_rate": 3.463524076513909e-05, + "loss": 3.5827, + "step": 938496 + }, + { + "epoch": 0.01, + "learning_rate": 3.462685481762857e-05, + "loss": 3.5893, + "step": 939008 + }, + { + "epoch": 0.01, + "learning_rate": 3.461846887011805e-05, + "loss": 3.5888, + "step": 939520 + }, + { + "epoch": 0.01, + "learning_rate": 3.461009930141127e-05, + "loss": 3.5821, + "step": 940032 + }, + { + "epoch": 0.01, + "learning_rate": 3.4601713353900747e-05, + "loss": 3.5552, + "step": 940544 + }, + { + "epoch": 0.01, + "learning_rate": 3.4593327406390227e-05, + "loss": 3.5893, + "step": 941056 + }, + { + "epoch": 0.01, + "learning_rate": 3.4584957837683436e-05, + "loss": 3.5841, + "step": 941568 + }, + { + "epoch": 0.01, + "learning_rate": 3.4576571890172916e-05, + "loss": 3.5918, + "step": 942080 + }, + { + "epoch": 0.01, + "learning_rate": 3.4568185942662396e-05, + "loss": 3.5829, + "step": 942592 + }, + { + "epoch": 0.01, + "learning_rate": 3.4559799995151875e-05, + "loss": 3.5759, + "step": 943104 + }, + { + "epoch": 0.01, + "learning_rate": 3.4551430426445085e-05, + "loss": 3.5702, + "step": 943616 + }, + { + "epoch": 0.01, + "learning_rate": 3.4543044478934565e-05, + "loss": 3.5736, + "step": 944128 + }, + { + "epoch": 0.01, + "learning_rate": 3.4534658531424045e-05, + "loss": 3.578, + "step": 944640 + }, + { + "epoch": 0.01, + "learning_rate": 3.4526272583913524e-05, + "loss": 3.5643, + "step": 945152 + }, + { + "epoch": 0.01, + "learning_rate": 3.451790301520674e-05, + "loss": 3.5991, + "step": 945664 + }, + { + "epoch": 0.01, + "learning_rate": 3.450951706769622e-05, + "loss": 3.582, + "step": 946176 + }, + { + "epoch": 0.01, + "learning_rate": 3.45011311201857e-05, + "loss": 3.5629, + "step": 946688 + }, + { + "epoch": 0.01, + "learning_rate": 3.449274517267518e-05, + "loss": 3.5867, + "step": 947200 + }, + { + "epoch": 0.01, + "learning_rate": 3.448437560396839e-05, + "loss": 3.5686, + "step": 947712 + }, + { + "epoch": 0.01, + "learning_rate": 3.447598965645787e-05, + "loss": 3.5687, + "step": 948224 + }, + { + "epoch": 0.01, + "learning_rate": 3.446760370894735e-05, + "loss": 3.5799, + "step": 948736 + }, + { + "epoch": 0.01, + "learning_rate": 3.445921776143683e-05, + "loss": 3.5799, + "step": 949248 + }, + { + "epoch": 0.01, + "learning_rate": 3.445084819273004e-05, + "loss": 3.5651, + "step": 949760 + }, + { + "epoch": 0.01, + "learning_rate": 3.444246224521952e-05, + "loss": 3.5797, + "step": 950272 + }, + { + "epoch": 0.01, + "learning_rate": 3.4434076297709e-05, + "loss": 3.5662, + "step": 950784 + }, + { + "epoch": 0.01, + "learning_rate": 3.442569035019848e-05, + "loss": 3.575, + "step": 951296 + }, + { + "epoch": 0.01, + "learning_rate": 3.4417320781491694e-05, + "loss": 3.5831, + "step": 951808 + }, + { + "epoch": 0.01, + "learning_rate": 3.4408934833981174e-05, + "loss": 3.5768, + "step": 952320 + }, + { + "epoch": 0.01, + "learning_rate": 3.4400548886470654e-05, + "loss": 3.579, + "step": 952832 + }, + { + "epoch": 0.01, + "learning_rate": 3.4392162938960134e-05, + "loss": 3.5764, + "step": 953344 + }, + { + "epoch": 0.01, + "learning_rate": 3.438379337025334e-05, + "loss": 3.5979, + "step": 953856 + }, + { + "epoch": 0.01, + "learning_rate": 3.437540742274282e-05, + "loss": 3.5648, + "step": 954368 + }, + { + "epoch": 0.01, + "learning_rate": 3.43670214752323e-05, + "loss": 3.589, + "step": 954880 + }, + { + "epoch": 0.01, + "learning_rate": 3.435863552772178e-05, + "loss": 3.573, + "step": 955392 + }, + { + "epoch": 0.01, + "learning_rate": 3.435026595901499e-05, + "loss": 3.5705, + "step": 955904 + }, + { + "epoch": 0.01, + "learning_rate": 3.434188001150447e-05, + "loss": 3.5799, + "step": 956416 + }, + { + "epoch": 0.01, + "learning_rate": 3.433349406399395e-05, + "loss": 3.578, + "step": 956928 + }, + { + "epoch": 0.01, + "learning_rate": 3.432510811648343e-05, + "loss": 3.5889, + "step": 957440 + }, + { + "epoch": 0.01, + "learning_rate": 3.431673854777665e-05, + "loss": 3.5857, + "step": 957952 + }, + { + "epoch": 0.01, + "learning_rate": 3.430835260026613e-05, + "loss": 3.5606, + "step": 958464 + }, + { + "epoch": 0.01, + "learning_rate": 3.429996665275561e-05, + "loss": 3.5611, + "step": 958976 + }, + { + "epoch": 0.01, + "learning_rate": 3.429158070524509e-05, + "loss": 3.5706, + "step": 959488 + }, + { + "epoch": 0.01, + "learning_rate": 3.42832111365383e-05, + "loss": 3.5834, + "step": 960000 + }, + { + "epoch": 0.01, + "learning_rate": 3.4274825189027777e-05, + "loss": 3.5715, + "step": 960512 + }, + { + "epoch": 0.01, + "learning_rate": 3.4266439241517257e-05, + "loss": 3.5692, + "step": 961024 + }, + { + "epoch": 0.01, + "learning_rate": 3.4258053294006736e-05, + "loss": 3.5803, + "step": 961536 + }, + { + "epoch": 0.02, + "learning_rate": 3.4249683725299946e-05, + "loss": 3.5756, + "step": 962048 + }, + { + "epoch": 0.02, + "learning_rate": 3.4241297777789426e-05, + "loss": 3.5785, + "step": 962560 + }, + { + "epoch": 0.02, + "learning_rate": 3.4232911830278906e-05, + "loss": 3.5695, + "step": 963072 + }, + { + "epoch": 0.02, + "learning_rate": 3.4224525882768385e-05, + "loss": 3.5764, + "step": 963584 + }, + { + "epoch": 0.02, + "learning_rate": 3.42161563140616e-05, + "loss": 3.5811, + "step": 964096 + }, + { + "epoch": 0.02, + "learning_rate": 3.420777036655108e-05, + "loss": 3.5837, + "step": 964608 + }, + { + "epoch": 0.02, + "learning_rate": 3.419938441904056e-05, + "loss": 3.5634, + "step": 965120 + }, + { + "epoch": 0.02, + "learning_rate": 3.419099847153004e-05, + "loss": 3.566, + "step": 965632 + }, + { + "epoch": 0.02, + "learning_rate": 3.418262890282325e-05, + "loss": 3.5778, + "step": 966144 + }, + { + "epoch": 0.02, + "learning_rate": 3.417424295531273e-05, + "loss": 3.5759, + "step": 966656 + }, + { + "epoch": 0.02, + "learning_rate": 3.416585700780221e-05, + "loss": 3.5822, + "step": 967168 + }, + { + "epoch": 0.02, + "learning_rate": 3.415747106029169e-05, + "loss": 3.5731, + "step": 967680 + }, + { + "epoch": 0.02, + "learning_rate": 3.41491014915849e-05, + "loss": 3.5695, + "step": 968192 + }, + { + "epoch": 0.02, + "learning_rate": 3.414071554407438e-05, + "loss": 3.5698, + "step": 968704 + }, + { + "epoch": 0.02, + "learning_rate": 3.413234597536759e-05, + "loss": 3.5673, + "step": 969216 + }, + { + "epoch": 0.02, + "learning_rate": 3.412396002785707e-05, + "loss": 3.572, + "step": 969728 + }, + { + "epoch": 0.02, + "learning_rate": 3.4115574080346555e-05, + "loss": 3.5684, + "step": 970240 + }, + { + "epoch": 0.02, + "learning_rate": 3.4107188132836035e-05, + "loss": 3.5759, + "step": 970752 + }, + { + "epoch": 0.02, + "learning_rate": 3.4098802185325515e-05, + "loss": 3.5724, + "step": 971264 + }, + { + "epoch": 0.02, + "learning_rate": 3.4090416237814995e-05, + "loss": 3.5677, + "step": 971776 + }, + { + "epoch": 0.02, + "learning_rate": 3.4082030290304475e-05, + "loss": 3.5711, + "step": 972288 + }, + { + "epoch": 0.02, + "learning_rate": 3.407364434279395e-05, + "loss": 3.5592, + "step": 972800 + }, + { + "epoch": 0.02, + "learning_rate": 3.4065274774087164e-05, + "loss": 3.5814, + "step": 973312 + }, + { + "epoch": 0.02, + "learning_rate": 3.4056888826576644e-05, + "loss": 3.5789, + "step": 973824 + }, + { + "epoch": 0.02, + "learning_rate": 3.4048502879066124e-05, + "loss": 3.5787, + "step": 974336 + }, + { + "epoch": 0.02, + "learning_rate": 3.40401169315556e-05, + "loss": 3.5819, + "step": 974848 + }, + { + "epoch": 0.02, + "learning_rate": 3.403174736284881e-05, + "loss": 3.5803, + "step": 975360 + }, + { + "epoch": 0.02, + "learning_rate": 3.402336141533829e-05, + "loss": 3.573, + "step": 975872 + }, + { + "epoch": 0.02, + "learning_rate": 3.401497546782777e-05, + "loss": 3.5678, + "step": 976384 + }, + { + "epoch": 0.02, + "learning_rate": 3.400658952031725e-05, + "loss": 3.5799, + "step": 976896 + }, + { + "epoch": 0.02, + "learning_rate": 3.399821995161047e-05, + "loss": 3.5722, + "step": 977408 + }, + { + "epoch": 0.02, + "learning_rate": 3.398983400409995e-05, + "loss": 3.5708, + "step": 977920 + }, + { + "epoch": 0.02, + "learning_rate": 3.398144805658942e-05, + "loss": 3.578, + "step": 978432 + }, + { + "epoch": 0.02, + "learning_rate": 3.39730621090789e-05, + "loss": 3.569, + "step": 978944 + }, + { + "epoch": 0.02, + "learning_rate": 3.396469254037212e-05, + "loss": 3.5762, + "step": 979456 + }, + { + "epoch": 0.02, + "learning_rate": 3.39563065928616e-05, + "loss": 3.5675, + "step": 979968 + }, + { + "epoch": 0.02, + "learning_rate": 3.394792064535107e-05, + "loss": 3.5695, + "step": 980480 + }, + { + "epoch": 0.02, + "learning_rate": 3.393953469784055e-05, + "loss": 3.5741, + "step": 980992 + }, + { + "epoch": 0.02, + "learning_rate": 3.3931165129133767e-05, + "loss": 3.5668, + "step": 981504 + }, + { + "epoch": 0.02, + "learning_rate": 3.3922779181623246e-05, + "loss": 3.5863, + "step": 982016 + }, + { + "epoch": 0.02, + "learning_rate": 3.3914393234112726e-05, + "loss": 3.5751, + "step": 982528 + }, + { + "epoch": 0.02, + "learning_rate": 3.3906007286602206e-05, + "loss": 3.5803, + "step": 983040 + }, + { + "epoch": 0.02, + "learning_rate": 3.389763771789542e-05, + "loss": 3.5694, + "step": 983552 + }, + { + "epoch": 0.02, + "learning_rate": 3.3889251770384895e-05, + "loss": 3.5743, + "step": 984064 + }, + { + "epoch": 0.02, + "learning_rate": 3.3880865822874375e-05, + "loss": 3.5773, + "step": 984576 + }, + { + "epoch": 0.02, + "learning_rate": 3.3872479875363855e-05, + "loss": 3.5782, + "step": 985088 + }, + { + "epoch": 0.02, + "learning_rate": 3.386411030665707e-05, + "loss": 3.5713, + "step": 985600 + }, + { + "epoch": 0.02, + "learning_rate": 3.3855724359146544e-05, + "loss": 3.5668, + "step": 986112 + }, + { + "epoch": 0.02, + "learning_rate": 3.3847338411636024e-05, + "loss": 3.5762, + "step": 986624 + }, + { + "epoch": 0.02, + "learning_rate": 3.3838952464125504e-05, + "loss": 3.5941, + "step": 987136 + }, + { + "epoch": 0.02, + "learning_rate": 3.3830582895418713e-05, + "loss": 3.574, + "step": 987648 + }, + { + "epoch": 0.02, + "learning_rate": 3.38221969479082e-05, + "loss": 3.5791, + "step": 988160 + }, + { + "epoch": 0.02, + "learning_rate": 3.381381100039768e-05, + "loss": 3.5803, + "step": 988672 + }, + { + "epoch": 0.02, + "learning_rate": 3.3805441431690896e-05, + "loss": 3.5729, + "step": 989184 + }, + { + "epoch": 0.02, + "learning_rate": 3.379705548418037e-05, + "loss": 3.5741, + "step": 989696 + }, + { + "epoch": 0.02, + "learning_rate": 3.378866953666985e-05, + "loss": 3.5721, + "step": 990208 + }, + { + "epoch": 0.02, + "learning_rate": 3.378028358915933e-05, + "loss": 3.5719, + "step": 990720 + }, + { + "epoch": 0.02, + "learning_rate": 3.377189764164881e-05, + "loss": 3.5757, + "step": 991232 + }, + { + "epoch": 0.02, + "learning_rate": 3.376351169413829e-05, + "loss": 3.569, + "step": 991744 + }, + { + "epoch": 0.03, + "eval_loss": 3.847501277923584, + "eval_runtime": 307.3418, + "eval_samples_per_second": 1241.585, + "eval_steps_per_second": 38.8, + "step": 992160 + }, + { + "epoch": 1.0, + "learning_rate": 3.375512574662777e-05, + "loss": 3.5731, + "step": 992256 + }, + { + "epoch": 1.0, + "learning_rate": 3.374673979911725e-05, + "loss": 3.5542, + "step": 992768 + }, + { + "epoch": 1.0, + "learning_rate": 3.373837023041046e-05, + "loss": 3.5706, + "step": 993280 + }, + { + "epoch": 1.0, + "learning_rate": 3.372998428289994e-05, + "loss": 3.592, + "step": 993792 + }, + { + "epoch": 1.0, + "learning_rate": 3.372159833538942e-05, + "loss": 3.5774, + "step": 994304 + }, + { + "epoch": 1.0, + "learning_rate": 3.37132123878789e-05, + "loss": 3.5705, + "step": 994816 + }, + { + "epoch": 1.0, + "learning_rate": 3.3704842819172114e-05, + "loss": 3.5752, + "step": 995328 + }, + { + "epoch": 1.0, + "learning_rate": 3.3696456871661594e-05, + "loss": 3.5701, + "step": 995840 + }, + { + "epoch": 1.0, + "learning_rate": 3.3688070924151074e-05, + "loss": 3.5634, + "step": 996352 + }, + { + "epoch": 1.0, + "learning_rate": 3.367970135544428e-05, + "loss": 3.58, + "step": 996864 + }, + { + "epoch": 1.0, + "learning_rate": 3.367131540793376e-05, + "loss": 3.5711, + "step": 997376 + }, + { + "epoch": 1.0, + "learning_rate": 3.366292946042324e-05, + "loss": 3.5719, + "step": 997888 + }, + { + "epoch": 1.0, + "learning_rate": 3.365454351291272e-05, + "loss": 3.5778, + "step": 998400 + }, + { + "epoch": 1.0, + "learning_rate": 3.36461575654022e-05, + "loss": 3.559, + "step": 998912 + }, + { + "epoch": 1.0, + "learning_rate": 3.363777161789168e-05, + "loss": 3.5593, + "step": 999424 + }, + { + "epoch": 1.0, + "learning_rate": 3.362938567038116e-05, + "loss": 3.5536, + "step": 999936 + }, + { + "epoch": 1.0, + "learning_rate": 3.362099972287064e-05, + "loss": 3.5714, + "step": 1000448 + }, + { + "epoch": 1.0, + "learning_rate": 3.361263015416385e-05, + "loss": 3.5679, + "step": 1000960 + }, + { + "epoch": 1.0, + "learning_rate": 3.360424420665333e-05, + "loss": 3.5676, + "step": 1001472 + }, + { + "epoch": 1.0, + "learning_rate": 3.359585825914282e-05, + "loss": 3.5665, + "step": 1001984 + }, + { + "epoch": 1.0, + "learning_rate": 3.35874723116323e-05, + "loss": 3.5871, + "step": 1002496 + }, + { + "epoch": 1.0, + "learning_rate": 3.357910274292551e-05, + "loss": 3.5692, + "step": 1003008 + }, + { + "epoch": 1.0, + "learning_rate": 3.357071679541499e-05, + "loss": 3.5722, + "step": 1003520 + }, + { + "epoch": 1.0, + "learning_rate": 3.356233084790447e-05, + "loss": 3.5716, + "step": 1004032 + }, + { + "epoch": 1.0, + "learning_rate": 3.355394490039395e-05, + "loss": 3.5795, + "step": 1004544 + }, + { + "epoch": 1.0, + "learning_rate": 3.3545575331687156e-05, + "loss": 3.565, + "step": 1005056 + }, + { + "epoch": 1.0, + "learning_rate": 3.3537189384176636e-05, + "loss": 3.5652, + "step": 1005568 + }, + { + "epoch": 1.0, + "learning_rate": 3.3528803436666116e-05, + "loss": 3.5697, + "step": 1006080 + }, + { + "epoch": 1.0, + "learning_rate": 3.3520417489155596e-05, + "loss": 3.5619, + "step": 1006592 + }, + { + "epoch": 1.0, + "learning_rate": 3.3512047920448805e-05, + "loss": 3.5609, + "step": 1007104 + }, + { + "epoch": 1.01, + "learning_rate": 3.3503661972938285e-05, + "loss": 3.5685, + "step": 1007616 + }, + { + "epoch": 1.01, + "learning_rate": 3.349527602542777e-05, + "loss": 3.565, + "step": 1008128 + }, + { + "epoch": 1.01, + "learning_rate": 3.348690645672098e-05, + "loss": 3.5767, + "step": 1008640 + }, + { + "epoch": 1.01, + "learning_rate": 3.347852050921046e-05, + "loss": 3.5717, + "step": 1009152 + }, + { + "epoch": 1.01, + "learning_rate": 3.347013456169994e-05, + "loss": 3.566, + "step": 1009664 + }, + { + "epoch": 1.01, + "learning_rate": 3.346174861418942e-05, + "loss": 3.5706, + "step": 1010176 + }, + { + "epoch": 1.01, + "learning_rate": 3.345337904548263e-05, + "loss": 3.5714, + "step": 1010688 + }, + { + "epoch": 1.01, + "learning_rate": 3.344499309797211e-05, + "loss": 3.5525, + "step": 1011200 + }, + { + "epoch": 1.01, + "learning_rate": 3.343660715046159e-05, + "loss": 3.5664, + "step": 1011712 + }, + { + "epoch": 1.01, + "learning_rate": 3.34282375817548e-05, + "loss": 3.5574, + "step": 1012224 + }, + { + "epoch": 1.01, + "learning_rate": 3.341985163424428e-05, + "loss": 3.5615, + "step": 1012736 + }, + { + "epoch": 1.01, + "learning_rate": 3.341146568673376e-05, + "loss": 3.5671, + "step": 1013248 + }, + { + "epoch": 1.01, + "learning_rate": 3.340307973922324e-05, + "loss": 3.5694, + "step": 1013760 + }, + { + "epoch": 1.01, + "learning_rate": 3.3394693791712726e-05, + "loss": 3.5685, + "step": 1014272 + }, + { + "epoch": 1.01, + "learning_rate": 3.3386307844202205e-05, + "loss": 3.5685, + "step": 1014784 + }, + { + "epoch": 1.01, + "learning_rate": 3.337792189669168e-05, + "loss": 3.5698, + "step": 1015296 + }, + { + "epoch": 1.01, + "learning_rate": 3.336953594918116e-05, + "loss": 3.5713, + "step": 1015808 + }, + { + "epoch": 1.01, + "learning_rate": 3.3361166380474374e-05, + "loss": 3.5678, + "step": 1016320 + }, + { + "epoch": 1.01, + "learning_rate": 3.335278043296385e-05, + "loss": 3.5422, + "step": 1016832 + }, + { + "epoch": 1.01, + "learning_rate": 3.334439448545333e-05, + "loss": 3.5696, + "step": 1017344 + }, + { + "epoch": 1.01, + "learning_rate": 3.333600853794281e-05, + "loss": 3.5693, + "step": 1017856 + }, + { + "epoch": 1.01, + "learning_rate": 3.3327638969236023e-05, + "loss": 3.5709, + "step": 1018368 + }, + { + "epoch": 1.01, + "learning_rate": 3.33192530217255e-05, + "loss": 3.5649, + "step": 1018880 + }, + { + "epoch": 1.01, + "learning_rate": 3.3310867074214977e-05, + "loss": 3.559, + "step": 1019392 + }, + { + "epoch": 1.01, + "learning_rate": 3.330249750550819e-05, + "loss": 3.554, + "step": 1019904 + }, + { + "epoch": 1.01, + "learning_rate": 3.329411155799768e-05, + "loss": 3.553, + "step": 1020416 + }, + { + "epoch": 1.01, + "learning_rate": 3.328572561048715e-05, + "loss": 3.5651, + "step": 1020928 + }, + { + "epoch": 1.01, + "learning_rate": 3.327733966297663e-05, + "loss": 3.5503, + "step": 1021440 + }, + { + "epoch": 1.01, + "learning_rate": 3.326897009426985e-05, + "loss": 3.5726, + "step": 1021952 + }, + { + "epoch": 1.01, + "learning_rate": 3.326058414675932e-05, + "loss": 3.566, + "step": 1022464 + }, + { + "epoch": 1.01, + "learning_rate": 3.32521981992488e-05, + "loss": 3.5466, + "step": 1022976 + }, + { + "epoch": 1.01, + "learning_rate": 3.324381225173828e-05, + "loss": 3.5682, + "step": 1023488 + }, + { + "epoch": 1.01, + "learning_rate": 3.32354426830315e-05, + "loss": 3.5502, + "step": 1024000 + }, + { + "epoch": 1.01, + "learning_rate": 3.322705673552097e-05, + "loss": 3.5547, + "step": 1024512 + }, + { + "epoch": 1.01, + "learning_rate": 3.321867078801045e-05, + "loss": 3.5566, + "step": 1025024 + }, + { + "epoch": 1.01, + "learning_rate": 3.321028484049993e-05, + "loss": 3.5669, + "step": 1025536 + }, + { + "epoch": 1.01, + "learning_rate": 3.3201915271793146e-05, + "loss": 3.5483, + "step": 1026048 + }, + { + "epoch": 1.01, + "learning_rate": 3.3193529324282626e-05, + "loss": 3.5611, + "step": 1026560 + }, + { + "epoch": 1.01, + "learning_rate": 3.3185143376772106e-05, + "loss": 3.5487, + "step": 1027072 + }, + { + "epoch": 1.01, + "learning_rate": 3.3176757429261586e-05, + "loss": 3.5591, + "step": 1027584 + }, + { + "epoch": 1.01, + "learning_rate": 3.3168387860554795e-05, + "loss": 3.5654, + "step": 1028096 + }, + { + "epoch": 1.01, + "learning_rate": 3.3160001913044275e-05, + "loss": 3.5617, + "step": 1028608 + }, + { + "epoch": 1.01, + "learning_rate": 3.3151615965533755e-05, + "loss": 3.5593, + "step": 1029120 + }, + { + "epoch": 1.01, + "learning_rate": 3.3143230018023235e-05, + "loss": 3.562, + "step": 1029632 + }, + { + "epoch": 1.01, + "learning_rate": 3.3134860449316444e-05, + "loss": 3.5783, + "step": 1030144 + }, + { + "epoch": 1.01, + "learning_rate": 3.3126474501805924e-05, + "loss": 3.5477, + "step": 1030656 + }, + { + "epoch": 1.01, + "learning_rate": 3.3118088554295404e-05, + "loss": 3.5741, + "step": 1031168 + }, + { + "epoch": 1.01, + "learning_rate": 3.3109702606784884e-05, + "loss": 3.5578, + "step": 1031680 + }, + { + "epoch": 1.01, + "learning_rate": 3.31013330380781e-05, + "loss": 3.5557, + "step": 1032192 + }, + { + "epoch": 1.01, + "learning_rate": 3.309294709056758e-05, + "loss": 3.5597, + "step": 1032704 + }, + { + "epoch": 1.01, + "learning_rate": 3.308456114305706e-05, + "loss": 3.5608, + "step": 1033216 + }, + { + "epoch": 1.01, + "learning_rate": 3.307617519554654e-05, + "loss": 3.569, + "step": 1033728 + }, + { + "epoch": 1.01, + "learning_rate": 3.306780562683975e-05, + "loss": 3.5707, + "step": 1034240 + }, + { + "epoch": 1.01, + "learning_rate": 3.305941967932923e-05, + "loss": 3.546, + "step": 1034752 + }, + { + "epoch": 1.01, + "learning_rate": 3.305103373181871e-05, + "loss": 3.5428, + "step": 1035264 + }, + { + "epoch": 1.01, + "learning_rate": 3.304264778430819e-05, + "loss": 3.5522, + "step": 1035776 + }, + { + "epoch": 1.01, + "learning_rate": 3.30342782156014e-05, + "loss": 3.5639, + "step": 1036288 + }, + { + "epoch": 1.01, + "learning_rate": 3.302589226809088e-05, + "loss": 3.5557, + "step": 1036800 + }, + { + "epoch": 1.01, + "learning_rate": 3.301750632058036e-05, + "loss": 3.5512, + "step": 1037312 + }, + { + "epoch": 1.01, + "learning_rate": 3.300912037306984e-05, + "loss": 3.5633, + "step": 1037824 + }, + { + "epoch": 1.02, + "learning_rate": 3.3000750804363053e-05, + "loss": 3.5568, + "step": 1038336 + }, + { + "epoch": 1.02, + "learning_rate": 3.2992364856852533e-05, + "loss": 3.5628, + "step": 1038848 + }, + { + "epoch": 1.02, + "learning_rate": 3.298397890934201e-05, + "loss": 3.5514, + "step": 1039360 + }, + { + "epoch": 1.02, + "learning_rate": 3.297559296183149e-05, + "loss": 3.562, + "step": 1039872 + }, + { + "epoch": 1.02, + "learning_rate": 3.29672233931247e-05, + "loss": 3.5603, + "step": 1040384 + }, + { + "epoch": 1.02, + "learning_rate": 3.295883744561418e-05, + "loss": 3.5698, + "step": 1040896 + }, + { + "epoch": 1.02, + "learning_rate": 3.295045149810366e-05, + "loss": 3.5432, + "step": 1041408 + }, + { + "epoch": 1.02, + "learning_rate": 3.294206555059314e-05, + "loss": 3.5486, + "step": 1041920 + }, + { + "epoch": 1.02, + "learning_rate": 3.293369598188635e-05, + "loss": 3.5645, + "step": 1042432 + }, + { + "epoch": 1.02, + "learning_rate": 3.292531003437583e-05, + "loss": 3.5588, + "step": 1042944 + }, + { + "epoch": 1.02, + "learning_rate": 3.291692408686531e-05, + "loss": 3.5638, + "step": 1043456 + }, + { + "epoch": 1.02, + "learning_rate": 3.290853813935479e-05, + "loss": 3.5568, + "step": 1043968 + }, + { + "epoch": 1.02, + "learning_rate": 3.290016857064801e-05, + "loss": 3.5516, + "step": 1044480 + }, + { + "epoch": 1.02, + "learning_rate": 3.289178262313749e-05, + "loss": 3.5536, + "step": 1044992 + }, + { + "epoch": 1.02, + "learning_rate": 3.2883413054430696e-05, + "loss": 3.5561, + "step": 1045504 + }, + { + "epoch": 1.02, + "learning_rate": 3.2875027106920176e-05, + "loss": 3.5538, + "step": 1046016 + }, + { + "epoch": 1.02, + "learning_rate": 3.2866641159409656e-05, + "loss": 3.5509, + "step": 1046528 + }, + { + "epoch": 1.02, + "learning_rate": 3.2858255211899136e-05, + "loss": 3.5607, + "step": 1047040 + }, + { + "epoch": 1.02, + "learning_rate": 3.2849869264388616e-05, + "loss": 3.5553, + "step": 1047552 + }, + { + "epoch": 1.02, + "learning_rate": 3.2841483316878096e-05, + "loss": 3.5528, + "step": 1048064 + }, + { + "epoch": 1.02, + "learning_rate": 3.2833097369367576e-05, + "loss": 3.5543, + "step": 1048576 + }, + { + "epoch": 1.02, + "learning_rate": 3.2824711421857056e-05, + "loss": 3.542, + "step": 1049088 + }, + { + "epoch": 1.02, + "learning_rate": 3.2816341853150265e-05, + "loss": 3.5675, + "step": 1049600 + }, + { + "epoch": 1.02, + "learning_rate": 3.2807955905639745e-05, + "loss": 3.5598, + "step": 1050112 + }, + { + "epoch": 1.02, + "learning_rate": 3.2799569958129225e-05, + "loss": 3.5599, + "step": 1050624 + }, + { + "epoch": 1.02, + "learning_rate": 3.279118401061871e-05, + "loss": 3.5668, + "step": 1051136 + }, + { + "epoch": 1.02, + "learning_rate": 3.278281444191192e-05, + "loss": 3.5625, + "step": 1051648 + }, + { + "epoch": 1.02, + "learning_rate": 3.27744284944014e-05, + "loss": 3.5541, + "step": 1052160 + }, + { + "epoch": 1.02, + "learning_rate": 3.276604254689088e-05, + "loss": 3.5542, + "step": 1052672 + }, + { + "epoch": 1.02, + "learning_rate": 3.275767297818409e-05, + "loss": 3.5634, + "step": 1053184 + }, + { + "epoch": 1.02, + "learning_rate": 3.274928703067357e-05, + "loss": 3.5561, + "step": 1053696 + }, + { + "epoch": 1.02, + "learning_rate": 3.274090108316305e-05, + "loss": 3.5519, + "step": 1054208 + }, + { + "epoch": 1.02, + "learning_rate": 3.273251513565253e-05, + "loss": 3.5627, + "step": 1054720 + }, + { + "epoch": 1.02, + "learning_rate": 3.272412918814201e-05, + "loss": 3.5571, + "step": 1055232 + }, + { + "epoch": 1.02, + "learning_rate": 3.271574324063149e-05, + "loss": 3.557, + "step": 1055744 + }, + { + "epoch": 1.02, + "learning_rate": 3.270735729312097e-05, + "loss": 3.5518, + "step": 1056256 + }, + { + "epoch": 1.02, + "learning_rate": 3.269897134561045e-05, + "loss": 3.5524, + "step": 1056768 + }, + { + "epoch": 1.02, + "learning_rate": 3.2690601776903665e-05, + "loss": 3.5595, + "step": 1057280 + }, + { + "epoch": 1.02, + "learning_rate": 3.2682215829393145e-05, + "loss": 3.5505, + "step": 1057792 + }, + { + "epoch": 1.02, + "learning_rate": 3.2673829881882625e-05, + "loss": 3.5607, + "step": 1058304 + }, + { + "epoch": 1.02, + "learning_rate": 3.2665443934372105e-05, + "loss": 3.5666, + "step": 1058816 + }, + { + "epoch": 1.02, + "learning_rate": 3.2657074365665314e-05, + "loss": 3.5584, + "step": 1059328 + }, + { + "epoch": 1.02, + "learning_rate": 3.2648688418154794e-05, + "loss": 3.5527, + "step": 1059840 + }, + { + "epoch": 1.02, + "learning_rate": 3.2640302470644274e-05, + "loss": 3.5624, + "step": 1060352 + }, + { + "epoch": 1.02, + "learning_rate": 3.263193290193748e-05, + "loss": 3.5596, + "step": 1060864 + }, + { + "epoch": 1.02, + "learning_rate": 3.262354695442696e-05, + "loss": 3.5592, + "step": 1061376 + }, + { + "epoch": 1.02, + "learning_rate": 3.261516100691644e-05, + "loss": 3.5539, + "step": 1061888 + }, + { + "epoch": 1.02, + "learning_rate": 3.260677505940592e-05, + "loss": 3.5514, + "step": 1062400 + }, + { + "epoch": 1.02, + "learning_rate": 3.259840549069913e-05, + "loss": 3.5646, + "step": 1062912 + }, + { + "epoch": 1.02, + "learning_rate": 3.259001954318862e-05, + "loss": 3.5759, + "step": 1063424 + }, + { + "epoch": 1.02, + "learning_rate": 3.25816335956781e-05, + "loss": 3.5527, + "step": 1063936 + }, + { + "epoch": 1.02, + "learning_rate": 3.257324764816758e-05, + "loss": 3.5633, + "step": 1064448 + }, + { + "epoch": 1.02, + "learning_rate": 3.256487807946079e-05, + "loss": 3.5644, + "step": 1064960 + }, + { + "epoch": 1.02, + "learning_rate": 3.2556508510754e-05, + "loss": 3.5559, + "step": 1065472 + }, + { + "epoch": 1.02, + "learning_rate": 3.254812256324348e-05, + "loss": 3.5588, + "step": 1065984 + }, + { + "epoch": 1.02, + "learning_rate": 3.253973661573296e-05, + "loss": 3.5548, + "step": 1066496 + }, + { + "epoch": 1.02, + "learning_rate": 3.253135066822244e-05, + "loss": 3.5552, + "step": 1067008 + }, + { + "epoch": 1.02, + "learning_rate": 3.252296472071192e-05, + "loss": 3.5635, + "step": 1067520 + }, + { + "epoch": 1.02, + "learning_rate": 3.25145787732014e-05, + "loss": 3.544, + "step": 1068032 + }, + { + "epoch": 1.03, + "eval_loss": 3.8488149642944336, + "eval_runtime": 308.8378, + "eval_samples_per_second": 1235.571, + "eval_steps_per_second": 38.612, + "step": 1068480 + }, + { + "epoch": 0.0, + "learning_rate": 3.250619282569087e-05, + "loss": 3.5656, + "step": 1068544 + }, + { + "epoch": 0.0, + "learning_rate": 3.249780687818036e-05, + "loss": 3.5402, + "step": 1069056 + }, + { + "epoch": 0.0, + "learning_rate": 3.248942093066984e-05, + "loss": 3.5525, + "step": 1069568 + }, + { + "epoch": 0.0, + "learning_rate": 3.248103498315932e-05, + "loss": 3.5726, + "step": 1070080 + }, + { + "epoch": 0.0, + "learning_rate": 3.2472665414452526e-05, + "loss": 3.5661, + "step": 1070592 + }, + { + "epoch": 0.0, + "learning_rate": 3.2464279466942006e-05, + "loss": 3.5529, + "step": 1071104 + }, + { + "epoch": 0.0, + "learning_rate": 3.2455893519431486e-05, + "loss": 3.5576, + "step": 1071616 + }, + { + "epoch": 0.0, + "learning_rate": 3.2447507571920966e-05, + "loss": 3.55, + "step": 1072128 + }, + { + "epoch": 0.0, + "learning_rate": 3.2439138003214175e-05, + "loss": 3.5465, + "step": 1072640 + }, + { + "epoch": 0.0, + "learning_rate": 3.243076843450739e-05, + "loss": 3.5613, + "step": 1073152 + }, + { + "epoch": 0.0, + "learning_rate": 3.242238248699687e-05, + "loss": 3.5566, + "step": 1073664 + }, + { + "epoch": 0.0, + "learning_rate": 3.2413996539486344e-05, + "loss": 3.5573, + "step": 1074176 + }, + { + "epoch": 0.0, + "learning_rate": 3.2405610591975824e-05, + "loss": 3.5605, + "step": 1074688 + }, + { + "epoch": 0.0, + "learning_rate": 3.239722464446531e-05, + "loss": 3.5423, + "step": 1075200 + }, + { + "epoch": 0.0, + "learning_rate": 3.238883869695479e-05, + "loss": 3.5431, + "step": 1075712 + }, + { + "epoch": 0.0, + "learning_rate": 3.2380469128248e-05, + "loss": 3.538, + "step": 1076224 + }, + { + "epoch": 0.0, + "learning_rate": 3.237208318073748e-05, + "loss": 3.5571, + "step": 1076736 + }, + { + "epoch": 0.0, + "learning_rate": 3.236369723322696e-05, + "loss": 3.5466, + "step": 1077248 + }, + { + "epoch": 0.0, + "learning_rate": 3.235531128571644e-05, + "loss": 3.5526, + "step": 1077760 + }, + { + "epoch": 0.0, + "learning_rate": 3.234692533820592e-05, + "loss": 3.5533, + "step": 1078272 + }, + { + "epoch": 0.0, + "learning_rate": 3.23385393906954e-05, + "loss": 3.57, + "step": 1078784 + }, + { + "epoch": 0.0, + "learning_rate": 3.233015344318488e-05, + "loss": 3.5581, + "step": 1079296 + }, + { + "epoch": 0.0, + "learning_rate": 3.232176749567436e-05, + "loss": 3.5535, + "step": 1079808 + }, + { + "epoch": 0.0, + "learning_rate": 3.231339792696757e-05, + "loss": 3.5545, + "step": 1080320 + }, + { + "epoch": 0.0, + "learning_rate": 3.230501197945705e-05, + "loss": 3.562, + "step": 1080832 + }, + { + "epoch": 0.0, + "learning_rate": 3.229662603194653e-05, + "loss": 3.55, + "step": 1081344 + }, + { + "epoch": 0.0, + "learning_rate": 3.228824008443601e-05, + "loss": 3.5464, + "step": 1081856 + }, + { + "epoch": 0.0, + "learning_rate": 3.2279870515729224e-05, + "loss": 3.5556, + "step": 1082368 + }, + { + "epoch": 0.0, + "learning_rate": 3.2271484568218704e-05, + "loss": 3.5428, + "step": 1082880 + }, + { + "epoch": 0.0, + "learning_rate": 3.2263098620708184e-05, + "loss": 3.5465, + "step": 1083392 + }, + { + "epoch": 0.01, + "learning_rate": 3.2254712673197664e-05, + "loss": 3.5491, + "step": 1083904 + }, + { + "epoch": 0.01, + "learning_rate": 3.224634310449087e-05, + "loss": 3.5503, + "step": 1084416 + }, + { + "epoch": 0.01, + "learning_rate": 3.223795715698035e-05, + "loss": 3.5552, + "step": 1084928 + }, + { + "epoch": 0.01, + "learning_rate": 3.222957120946983e-05, + "loss": 3.5567, + "step": 1085440 + }, + { + "epoch": 0.01, + "learning_rate": 3.222118526195931e-05, + "loss": 3.5502, + "step": 1085952 + }, + { + "epoch": 0.01, + "learning_rate": 3.221281569325252e-05, + "loss": 3.5514, + "step": 1086464 + }, + { + "epoch": 0.01, + "learning_rate": 3.2204429745742e-05, + "loss": 3.5562, + "step": 1086976 + }, + { + "epoch": 0.01, + "learning_rate": 3.219604379823148e-05, + "loss": 3.5373, + "step": 1087488 + }, + { + "epoch": 0.01, + "learning_rate": 3.21876742295247e-05, + "loss": 3.5552, + "step": 1088000 + }, + { + "epoch": 0.01, + "learning_rate": 3.217930466081791e-05, + "loss": 3.5364, + "step": 1088512 + }, + { + "epoch": 0.01, + "learning_rate": 3.217091871330739e-05, + "loss": 3.545, + "step": 1089024 + }, + { + "epoch": 0.01, + "learning_rate": 3.216253276579687e-05, + "loss": 3.5479, + "step": 1089536 + }, + { + "epoch": 0.01, + "learning_rate": 3.215414681828635e-05, + "loss": 3.5594, + "step": 1090048 + }, + { + "epoch": 0.01, + "learning_rate": 3.2145760870775827e-05, + "loss": 3.5509, + "step": 1090560 + }, + { + "epoch": 0.01, + "learning_rate": 3.2137374923265307e-05, + "loss": 3.5476, + "step": 1091072 + }, + { + "epoch": 0.01, + "learning_rate": 3.2128988975754787e-05, + "loss": 3.5582, + "step": 1091584 + }, + { + "epoch": 0.01, + "learning_rate": 3.2120603028244266e-05, + "loss": 3.5525, + "step": 1092096 + }, + { + "epoch": 0.01, + "learning_rate": 3.2112233459537476e-05, + "loss": 3.5527, + "step": 1092608 + }, + { + "epoch": 0.01, + "learning_rate": 3.2103847512026956e-05, + "loss": 3.523, + "step": 1093120 + }, + { + "epoch": 0.01, + "learning_rate": 3.2095461564516435e-05, + "loss": 3.5508, + "step": 1093632 + }, + { + "epoch": 0.01, + "learning_rate": 3.2087075617005915e-05, + "loss": 3.5551, + "step": 1094144 + }, + { + "epoch": 0.01, + "learning_rate": 3.207870604829913e-05, + "loss": 3.5551, + "step": 1094656 + }, + { + "epoch": 0.01, + "learning_rate": 3.207032010078861e-05, + "loss": 3.5481, + "step": 1095168 + }, + { + "epoch": 0.01, + "learning_rate": 3.206193415327809e-05, + "loss": 3.5411, + "step": 1095680 + }, + { + "epoch": 0.01, + "learning_rate": 3.205354820576757e-05, + "loss": 3.5452, + "step": 1096192 + }, + { + "epoch": 0.01, + "learning_rate": 3.204517863706078e-05, + "loss": 3.531, + "step": 1096704 + }, + { + "epoch": 0.01, + "learning_rate": 3.203679268955026e-05, + "loss": 3.5538, + "step": 1097216 + }, + { + "epoch": 0.01, + "learning_rate": 3.202840674203974e-05, + "loss": 3.5314, + "step": 1097728 + }, + { + "epoch": 0.01, + "learning_rate": 3.202003717333295e-05, + "loss": 3.5551, + "step": 1098240 + }, + { + "epoch": 0.01, + "learning_rate": 3.201165122582243e-05, + "loss": 3.548, + "step": 1098752 + }, + { + "epoch": 0.01, + "learning_rate": 3.200326527831191e-05, + "loss": 3.5353, + "step": 1099264 + }, + { + "epoch": 0.01, + "learning_rate": 3.199487933080139e-05, + "loss": 3.5531, + "step": 1099776 + }, + { + "epoch": 0.01, + "learning_rate": 3.1986509762094605e-05, + "loss": 3.5332, + "step": 1100288 + }, + { + "epoch": 0.01, + "learning_rate": 3.1978123814584085e-05, + "loss": 3.5365, + "step": 1100800 + }, + { + "epoch": 0.01, + "learning_rate": 3.1969737867073565e-05, + "loss": 3.5391, + "step": 1101312 + }, + { + "epoch": 0.01, + "learning_rate": 3.1961351919563045e-05, + "loss": 3.5523, + "step": 1101824 + }, + { + "epoch": 0.01, + "learning_rate": 3.1952982350856254e-05, + "loss": 3.5326, + "step": 1102336 + }, + { + "epoch": 0.01, + "learning_rate": 3.1944596403345734e-05, + "loss": 3.5447, + "step": 1102848 + }, + { + "epoch": 0.01, + "learning_rate": 3.1936210455835214e-05, + "loss": 3.5325, + "step": 1103360 + }, + { + "epoch": 0.01, + "learning_rate": 3.1927824508324694e-05, + "loss": 3.542, + "step": 1103872 + }, + { + "epoch": 0.01, + "learning_rate": 3.19194549396179e-05, + "loss": 3.5509, + "step": 1104384 + }, + { + "epoch": 0.01, + "learning_rate": 3.191106899210738e-05, + "loss": 3.548, + "step": 1104896 + }, + { + "epoch": 0.01, + "learning_rate": 3.190268304459686e-05, + "loss": 3.5409, + "step": 1105408 + }, + { + "epoch": 0.01, + "learning_rate": 3.189429709708634e-05, + "loss": 3.547, + "step": 1105920 + }, + { + "epoch": 0.01, + "learning_rate": 3.188592752837956e-05, + "loss": 3.5577, + "step": 1106432 + }, + { + "epoch": 0.01, + "learning_rate": 3.187754158086904e-05, + "loss": 3.5361, + "step": 1106944 + }, + { + "epoch": 0.01, + "learning_rate": 3.186915563335852e-05, + "loss": 3.5574, + "step": 1107456 + }, + { + "epoch": 0.01, + "learning_rate": 3.1860769685848e-05, + "loss": 3.5406, + "step": 1107968 + }, + { + "epoch": 0.01, + "learning_rate": 3.185240011714121e-05, + "loss": 3.5419, + "step": 1108480 + }, + { + "epoch": 0.01, + "learning_rate": 3.184401416963069e-05, + "loss": 3.543, + "step": 1108992 + }, + { + "epoch": 0.01, + "learning_rate": 3.183562822212017e-05, + "loss": 3.5434, + "step": 1109504 + }, + { + "epoch": 0.01, + "learning_rate": 3.182724227460965e-05, + "loss": 3.5547, + "step": 1110016 + }, + { + "epoch": 0.01, + "learning_rate": 3.181887270590286e-05, + "loss": 3.5521, + "step": 1110528 + }, + { + "epoch": 0.01, + "learning_rate": 3.1810486758392337e-05, + "loss": 3.5377, + "step": 1111040 + }, + { + "epoch": 0.01, + "learning_rate": 3.1802117189685546e-05, + "loss": 3.5222, + "step": 1111552 + }, + { + "epoch": 0.01, + "learning_rate": 3.1793731242175026e-05, + "loss": 3.5353, + "step": 1112064 + }, + { + "epoch": 0.01, + "learning_rate": 3.178534529466451e-05, + "loss": 3.5477, + "step": 1112576 + }, + { + "epoch": 0.01, + "learning_rate": 3.177695934715399e-05, + "loss": 3.5423, + "step": 1113088 + }, + { + "epoch": 0.01, + "learning_rate": 3.176857339964347e-05, + "loss": 3.5349, + "step": 1113600 + }, + { + "epoch": 0.01, + "learning_rate": 3.176018745213295e-05, + "loss": 3.5466, + "step": 1114112 + }, + { + "epoch": 0.02, + "learning_rate": 3.1751801504622425e-05, + "loss": 3.5437, + "step": 1114624 + }, + { + "epoch": 0.02, + "learning_rate": 3.1743415557111905e-05, + "loss": 3.5444, + "step": 1115136 + }, + { + "epoch": 0.02, + "learning_rate": 3.173504598840512e-05, + "loss": 3.5358, + "step": 1115648 + }, + { + "epoch": 0.02, + "learning_rate": 3.172667641969833e-05, + "loss": 3.5466, + "step": 1116160 + }, + { + "epoch": 0.02, + "learning_rate": 3.171829047218781e-05, + "loss": 3.5393, + "step": 1116672 + }, + { + "epoch": 0.02, + "learning_rate": 3.170990452467729e-05, + "loss": 3.5528, + "step": 1117184 + }, + { + "epoch": 0.02, + "learning_rate": 3.170151857716677e-05, + "loss": 3.532, + "step": 1117696 + }, + { + "epoch": 0.02, + "learning_rate": 3.169313262965625e-05, + "loss": 3.533, + "step": 1118208 + }, + { + "epoch": 0.02, + "learning_rate": 3.168474668214573e-05, + "loss": 3.5499, + "step": 1118720 + }, + { + "epoch": 0.02, + "learning_rate": 3.167636073463521e-05, + "loss": 3.5416, + "step": 1119232 + }, + { + "epoch": 0.02, + "learning_rate": 3.1667991165928426e-05, + "loss": 3.5483, + "step": 1119744 + }, + { + "epoch": 0.02, + "learning_rate": 3.16596052184179e-05, + "loss": 3.5434, + "step": 1120256 + }, + { + "epoch": 0.02, + "learning_rate": 3.165121927090738e-05, + "loss": 3.5358, + "step": 1120768 + }, + { + "epoch": 0.02, + "learning_rate": 3.164283332339686e-05, + "loss": 3.5385, + "step": 1121280 + }, + { + "epoch": 0.02, + "learning_rate": 3.1634480133493804e-05, + "loss": 3.5435, + "step": 1121792 + }, + { + "epoch": 0.02, + "learning_rate": 3.1626094185983284e-05, + "loss": 3.5322, + "step": 1122304 + }, + { + "epoch": 0.02, + "learning_rate": 3.1617708238472764e-05, + "loss": 3.5373, + "step": 1122816 + }, + { + "epoch": 0.02, + "learning_rate": 3.1609322290962244e-05, + "loss": 3.5448, + "step": 1123328 + }, + { + "epoch": 0.02, + "learning_rate": 3.160093634345172e-05, + "loss": 3.5402, + "step": 1123840 + }, + { + "epoch": 0.02, + "learning_rate": 3.1592550395941204e-05, + "loss": 3.5329, + "step": 1124352 + }, + { + "epoch": 0.02, + "learning_rate": 3.1584164448430684e-05, + "loss": 3.5439, + "step": 1124864 + }, + { + "epoch": 0.02, + "learning_rate": 3.1575778500920164e-05, + "loss": 3.5264, + "step": 1125376 + }, + { + "epoch": 0.02, + "learning_rate": 3.156740893221337e-05, + "loss": 3.5482, + "step": 1125888 + }, + { + "epoch": 0.02, + "learning_rate": 3.155902298470285e-05, + "loss": 3.5419, + "step": 1126400 + }, + { + "epoch": 0.02, + "learning_rate": 3.155063703719233e-05, + "loss": 3.5462, + "step": 1126912 + }, + { + "epoch": 0.02, + "learning_rate": 3.154225108968181e-05, + "loss": 3.5506, + "step": 1127424 + }, + { + "epoch": 0.02, + "learning_rate": 3.153388152097502e-05, + "loss": 3.5495, + "step": 1127936 + }, + { + "epoch": 0.02, + "learning_rate": 3.152551195226824e-05, + "loss": 3.5375, + "step": 1128448 + }, + { + "epoch": 0.02, + "learning_rate": 3.151712600475772e-05, + "loss": 3.539, + "step": 1128960 + }, + { + "epoch": 0.02, + "learning_rate": 3.150874005724719e-05, + "loss": 3.5441, + "step": 1129472 + }, + { + "epoch": 0.02, + "learning_rate": 3.150035410973667e-05, + "loss": 3.5413, + "step": 1129984 + }, + { + "epoch": 0.02, + "learning_rate": 3.149196816222616e-05, + "loss": 3.5313, + "step": 1130496 + }, + { + "epoch": 0.02, + "learning_rate": 3.148358221471564e-05, + "loss": 3.5495, + "step": 1131008 + }, + { + "epoch": 0.02, + "learning_rate": 3.147519626720512e-05, + "loss": 3.5395, + "step": 1131520 + }, + { + "epoch": 0.02, + "learning_rate": 3.1466826698498326e-05, + "loss": 3.5442, + "step": 1132032 + }, + { + "epoch": 0.02, + "learning_rate": 3.1458440750987806e-05, + "loss": 3.5342, + "step": 1132544 + }, + { + "epoch": 0.02, + "learning_rate": 3.1450054803477286e-05, + "loss": 3.5346, + "step": 1133056 + }, + { + "epoch": 0.02, + "learning_rate": 3.1441668855966766e-05, + "loss": 3.5467, + "step": 1133568 + }, + { + "epoch": 0.02, + "learning_rate": 3.1433299287259975e-05, + "loss": 3.538, + "step": 1134080 + }, + { + "epoch": 0.02, + "learning_rate": 3.1424913339749455e-05, + "loss": 3.5405, + "step": 1134592 + }, + { + "epoch": 0.02, + "learning_rate": 3.1416527392238935e-05, + "loss": 3.5524, + "step": 1135104 + }, + { + "epoch": 0.02, + "learning_rate": 3.1408141444728415e-05, + "loss": 3.5413, + "step": 1135616 + }, + { + "epoch": 0.02, + "learning_rate": 3.1399771876021624e-05, + "loss": 3.5379, + "step": 1136128 + }, + { + "epoch": 0.02, + "learning_rate": 3.139138592851111e-05, + "loss": 3.5454, + "step": 1136640 + }, + { + "epoch": 0.02, + "learning_rate": 3.138299998100059e-05, + "loss": 3.5457, + "step": 1137152 + }, + { + "epoch": 0.02, + "learning_rate": 3.137461403349007e-05, + "loss": 3.5432, + "step": 1137664 + }, + { + "epoch": 0.02, + "learning_rate": 3.136624446478328e-05, + "loss": 3.5326, + "step": 1138176 + }, + { + "epoch": 0.02, + "learning_rate": 3.135785851727276e-05, + "loss": 3.5376, + "step": 1138688 + }, + { + "epoch": 0.02, + "learning_rate": 3.134947256976224e-05, + "loss": 3.5527, + "step": 1139200 + }, + { + "epoch": 0.02, + "learning_rate": 3.134108662225172e-05, + "loss": 3.5568, + "step": 1139712 + }, + { + "epoch": 0.02, + "learning_rate": 3.133271705354493e-05, + "loss": 3.543, + "step": 1140224 + }, + { + "epoch": 0.02, + "learning_rate": 3.132433110603441e-05, + "loss": 3.5424, + "step": 1140736 + }, + { + "epoch": 0.02, + "learning_rate": 3.131594515852389e-05, + "loss": 3.5492, + "step": 1141248 + }, + { + "epoch": 0.02, + "learning_rate": 3.13075755898171e-05, + "loss": 3.5432, + "step": 1141760 + }, + { + "epoch": 0.02, + "learning_rate": 3.129918964230658e-05, + "loss": 3.5409, + "step": 1142272 + }, + { + "epoch": 0.02, + "learning_rate": 3.1290803694796065e-05, + "loss": 3.5402, + "step": 1142784 + }, + { + "epoch": 0.02, + "learning_rate": 3.1282417747285545e-05, + "loss": 3.5382, + "step": 1143296 + }, + { + "epoch": 0.02, + "learning_rate": 3.1274031799775025e-05, + "loss": 3.5488, + "step": 1143808 + }, + { + "epoch": 0.02, + "learning_rate": 3.1265645852264505e-05, + "loss": 3.5304, + "step": 1144320 + }, + { + "epoch": 0.03, + "eval_loss": 3.850249767303467, + "eval_runtime": 317.4721, + "eval_samples_per_second": 1201.967, + "eval_steps_per_second": 37.562, + "step": 1144800 + }, + { + "epoch": 1.0, + "learning_rate": 3.1257259904753985e-05, + "loss": 3.5395, + "step": 1144832 + }, + { + "epoch": 1.0, + "learning_rate": 3.1248873957243465e-05, + "loss": 3.5278, + "step": 1145344 + }, + { + "epoch": 1.0, + "learning_rate": 3.1240504388536674e-05, + "loss": 3.5347, + "step": 1145856 + }, + { + "epoch": 1.0, + "learning_rate": 3.1232118441026154e-05, + "loss": 3.5568, + "step": 1146368 + }, + { + "epoch": 1.0, + "learning_rate": 3.1223732493515634e-05, + "loss": 3.5527, + "step": 1146880 + }, + { + "epoch": 1.0, + "learning_rate": 3.1215346546005114e-05, + "loss": 3.5377, + "step": 1147392 + }, + { + "epoch": 1.0, + "learning_rate": 3.120697697729832e-05, + "loss": 3.54, + "step": 1147904 + }, + { + "epoch": 1.0, + "learning_rate": 3.11985910297878e-05, + "loss": 3.5364, + "step": 1148416 + }, + { + "epoch": 1.0, + "learning_rate": 3.119020508227728e-05, + "loss": 3.5375, + "step": 1148928 + }, + { + "epoch": 1.0, + "learning_rate": 3.11818355135705e-05, + "loss": 3.5443, + "step": 1149440 + }, + { + "epoch": 1.0, + "learning_rate": 3.117344956605998e-05, + "loss": 3.5383, + "step": 1149952 + }, + { + "epoch": 1.0, + "learning_rate": 3.116506361854946e-05, + "loss": 3.5436, + "step": 1150464 + }, + { + "epoch": 1.0, + "learning_rate": 3.115667767103894e-05, + "loss": 3.5462, + "step": 1150976 + }, + { + "epoch": 1.0, + "learning_rate": 3.114830810233215e-05, + "loss": 3.5288, + "step": 1151488 + }, + { + "epoch": 1.0, + "learning_rate": 3.113992215482163e-05, + "loss": 3.5254, + "step": 1152000 + }, + { + "epoch": 1.0, + "learning_rate": 3.113153620731111e-05, + "loss": 3.5248, + "step": 1152512 + }, + { + "epoch": 1.0, + "learning_rate": 3.112315025980059e-05, + "loss": 3.5403, + "step": 1153024 + }, + { + "epoch": 1.0, + "learning_rate": 3.111476431229007e-05, + "loss": 3.5354, + "step": 1153536 + }, + { + "epoch": 1.0, + "learning_rate": 3.110637836477955e-05, + "loss": 3.5387, + "step": 1154048 + }, + { + "epoch": 1.0, + "learning_rate": 3.109799241726903e-05, + "loss": 3.5365, + "step": 1154560 + }, + { + "epoch": 1.0, + "learning_rate": 3.10896064697585e-05, + "loss": 3.5519, + "step": 1155072 + }, + { + "epoch": 1.0, + "learning_rate": 3.1081236901051716e-05, + "loss": 3.5417, + "step": 1155584 + }, + { + "epoch": 1.0, + "learning_rate": 3.1072850953541196e-05, + "loss": 3.5424, + "step": 1156096 + }, + { + "epoch": 1.0, + "learning_rate": 3.106446500603068e-05, + "loss": 3.5384, + "step": 1156608 + }, + { + "epoch": 1.0, + "learning_rate": 3.1056079058520156e-05, + "loss": 3.5494, + "step": 1157120 + }, + { + "epoch": 1.0, + "learning_rate": 3.104770948981337e-05, + "loss": 3.5326, + "step": 1157632 + }, + { + "epoch": 1.0, + "learning_rate": 3.103932354230285e-05, + "loss": 3.5366, + "step": 1158144 + }, + { + "epoch": 1.0, + "learning_rate": 3.1030937594792325e-05, + "loss": 3.5371, + "step": 1158656 + }, + { + "epoch": 1.0, + "learning_rate": 3.1022551647281805e-05, + "loss": 3.5276, + "step": 1159168 + }, + { + "epoch": 1.0, + "learning_rate": 3.101418207857502e-05, + "loss": 3.5334, + "step": 1159680 + }, + { + "epoch": 1.01, + "learning_rate": 3.10057961310645e-05, + "loss": 3.5314, + "step": 1160192 + }, + { + "epoch": 1.01, + "learning_rate": 3.0997410183553974e-05, + "loss": 3.5363, + "step": 1160704 + }, + { + "epoch": 1.01, + "learning_rate": 3.0989024236043454e-05, + "loss": 3.5406, + "step": 1161216 + }, + { + "epoch": 1.01, + "learning_rate": 3.098065466733667e-05, + "loss": 3.5413, + "step": 1161728 + }, + { + "epoch": 1.01, + "learning_rate": 3.097226871982615e-05, + "loss": 3.5406, + "step": 1162240 + }, + { + "epoch": 1.01, + "learning_rate": 3.096388277231563e-05, + "loss": 3.536, + "step": 1162752 + }, + { + "epoch": 1.01, + "learning_rate": 3.095549682480511e-05, + "loss": 3.5427, + "step": 1163264 + }, + { + "epoch": 1.01, + "learning_rate": 3.0947127256098326e-05, + "loss": 3.5217, + "step": 1163776 + }, + { + "epoch": 1.01, + "learning_rate": 3.09387413085878e-05, + "loss": 3.5423, + "step": 1164288 + }, + { + "epoch": 1.01, + "learning_rate": 3.0930371739881015e-05, + "loss": 3.5176, + "step": 1164800 + }, + { + "epoch": 1.01, + "learning_rate": 3.0921985792370495e-05, + "loss": 3.5325, + "step": 1165312 + }, + { + "epoch": 1.01, + "learning_rate": 3.0913599844859975e-05, + "loss": 3.5322, + "step": 1165824 + }, + { + "epoch": 1.01, + "learning_rate": 3.090521389734945e-05, + "loss": 3.5426, + "step": 1166336 + }, + { + "epoch": 1.01, + "learning_rate": 3.089682794983893e-05, + "loss": 3.536, + "step": 1166848 + }, + { + "epoch": 1.01, + "learning_rate": 3.088844200232841e-05, + "loss": 3.5371, + "step": 1167360 + }, + { + "epoch": 1.01, + "learning_rate": 3.088005605481789e-05, + "loss": 3.5423, + "step": 1167872 + }, + { + "epoch": 1.01, + "learning_rate": 3.0871670107307374e-05, + "loss": 3.5387, + "step": 1168384 + }, + { + "epoch": 1.01, + "learning_rate": 3.0863300538600583e-05, + "loss": 3.536, + "step": 1168896 + }, + { + "epoch": 1.01, + "learning_rate": 3.0854914591090063e-05, + "loss": 3.5129, + "step": 1169408 + }, + { + "epoch": 1.01, + "learning_rate": 3.084652864357954e-05, + "loss": 3.5331, + "step": 1169920 + }, + { + "epoch": 1.01, + "learning_rate": 3.083814269606902e-05, + "loss": 3.5423, + "step": 1170432 + }, + { + "epoch": 1.01, + "learning_rate": 3.082977312736223e-05, + "loss": 3.5389, + "step": 1170944 + }, + { + "epoch": 1.01, + "learning_rate": 3.082138717985171e-05, + "loss": 3.5326, + "step": 1171456 + }, + { + "epoch": 1.01, + "learning_rate": 3.081300123234119e-05, + "loss": 3.5255, + "step": 1171968 + }, + { + "epoch": 1.01, + "learning_rate": 3.080461528483067e-05, + "loss": 3.5283, + "step": 1172480 + }, + { + "epoch": 1.01, + "learning_rate": 3.079624571612388e-05, + "loss": 3.516, + "step": 1172992 + }, + { + "epoch": 1.01, + "learning_rate": 3.078785976861336e-05, + "loss": 3.5404, + "step": 1173504 + }, + { + "epoch": 1.01, + "learning_rate": 3.077947382110284e-05, + "loss": 3.5176, + "step": 1174016 + }, + { + "epoch": 1.01, + "learning_rate": 3.077110425239606e-05, + "loss": 3.5403, + "step": 1174528 + }, + { + "epoch": 1.01, + "learning_rate": 3.076271830488554e-05, + "loss": 3.535, + "step": 1175040 + }, + { + "epoch": 1.01, + "learning_rate": 3.075433235737502e-05, + "loss": 3.5223, + "step": 1175552 + }, + { + "epoch": 1.01, + "learning_rate": 3.07459464098645e-05, + "loss": 3.5326, + "step": 1176064 + }, + { + "epoch": 1.01, + "learning_rate": 3.0737576841157706e-05, + "loss": 3.5176, + "step": 1176576 + }, + { + "epoch": 1.01, + "learning_rate": 3.0729190893647186e-05, + "loss": 3.5245, + "step": 1177088 + }, + { + "epoch": 1.01, + "learning_rate": 3.0720804946136666e-05, + "loss": 3.5187, + "step": 1177600 + }, + { + "epoch": 1.01, + "learning_rate": 3.0712418998626146e-05, + "loss": 3.5419, + "step": 1178112 + }, + { + "epoch": 1.01, + "learning_rate": 3.0704049429919355e-05, + "loss": 3.5221, + "step": 1178624 + }, + { + "epoch": 1.01, + "learning_rate": 3.0695663482408835e-05, + "loss": 3.5214, + "step": 1179136 + }, + { + "epoch": 1.01, + "learning_rate": 3.0687277534898315e-05, + "loss": 3.5198, + "step": 1179648 + }, + { + "epoch": 1.01, + "learning_rate": 3.0678891587387795e-05, + "loss": 3.5276, + "step": 1180160 + }, + { + "epoch": 1.01, + "learning_rate": 3.067052201868101e-05, + "loss": 3.5362, + "step": 1180672 + }, + { + "epoch": 1.01, + "learning_rate": 3.066213607117049e-05, + "loss": 3.5308, + "step": 1181184 + }, + { + "epoch": 1.01, + "learning_rate": 3.065375012365997e-05, + "loss": 3.5228, + "step": 1181696 + }, + { + "epoch": 1.01, + "learning_rate": 3.064536417614945e-05, + "loss": 3.5349, + "step": 1182208 + }, + { + "epoch": 1.01, + "learning_rate": 3.063699460744266e-05, + "loss": 3.544, + "step": 1182720 + }, + { + "epoch": 1.01, + "learning_rate": 3.062860865993214e-05, + "loss": 3.5212, + "step": 1183232 + }, + { + "epoch": 1.01, + "learning_rate": 3.062022271242162e-05, + "loss": 3.5421, + "step": 1183744 + }, + { + "epoch": 1.01, + "learning_rate": 3.06118367649111e-05, + "loss": 3.5275, + "step": 1184256 + }, + { + "epoch": 1.01, + "learning_rate": 3.060346719620431e-05, + "loss": 3.5282, + "step": 1184768 + }, + { + "epoch": 1.01, + "learning_rate": 3.059508124869379e-05, + "loss": 3.5255, + "step": 1185280 + }, + { + "epoch": 1.01, + "learning_rate": 3.058669530118327e-05, + "loss": 3.5311, + "step": 1185792 + }, + { + "epoch": 1.01, + "learning_rate": 3.057830935367275e-05, + "loss": 3.5405, + "step": 1186304 + }, + { + "epoch": 1.01, + "learning_rate": 3.0569939784965964e-05, + "loss": 3.5388, + "step": 1186816 + }, + { + "epoch": 1.01, + "learning_rate": 3.0561553837455444e-05, + "loss": 3.5246, + "step": 1187328 + }, + { + "epoch": 1.01, + "learning_rate": 3.0553184268748654e-05, + "loss": 3.5072, + "step": 1187840 + }, + { + "epoch": 1.01, + "learning_rate": 3.0544798321238133e-05, + "loss": 3.5253, + "step": 1188352 + }, + { + "epoch": 1.01, + "learning_rate": 3.0536412373727613e-05, + "loss": 3.528, + "step": 1188864 + }, + { + "epoch": 1.01, + "learning_rate": 3.0528026426217093e-05, + "loss": 3.532, + "step": 1189376 + }, + { + "epoch": 1.01, + "learning_rate": 3.051964047870657e-05, + "loss": 3.5192, + "step": 1189888 + }, + { + "epoch": 1.01, + "learning_rate": 3.0511254531196053e-05, + "loss": 3.5335, + "step": 1190400 + }, + { + "epoch": 1.02, + "learning_rate": 3.0502868583685533e-05, + "loss": 3.5286, + "step": 1190912 + }, + { + "epoch": 1.02, + "learning_rate": 3.0494482636175013e-05, + "loss": 3.5303, + "step": 1191424 + }, + { + "epoch": 1.02, + "learning_rate": 3.0486113067468226e-05, + "loss": 3.5189, + "step": 1191936 + }, + { + "epoch": 1.02, + "learning_rate": 3.0477727119957706e-05, + "loss": 3.5328, + "step": 1192448 + }, + { + "epoch": 1.02, + "learning_rate": 3.0469341172447186e-05, + "loss": 3.5235, + "step": 1192960 + }, + { + "epoch": 1.02, + "learning_rate": 3.0460955224936666e-05, + "loss": 3.537, + "step": 1193472 + }, + { + "epoch": 1.02, + "learning_rate": 3.0452585656229875e-05, + "loss": 3.5226, + "step": 1193984 + }, + { + "epoch": 1.02, + "learning_rate": 3.0444199708719358e-05, + "loss": 3.5145, + "step": 1194496 + }, + { + "epoch": 1.02, + "learning_rate": 3.0435813761208838e-05, + "loss": 3.5316, + "step": 1195008 + }, + { + "epoch": 1.02, + "learning_rate": 3.0427444192502047e-05, + "loss": 3.5306, + "step": 1195520 + }, + { + "epoch": 1.02, + "learning_rate": 3.0419058244991527e-05, + "loss": 3.5327, + "step": 1196032 + }, + { + "epoch": 1.02, + "learning_rate": 3.0410672297481007e-05, + "loss": 3.5305, + "step": 1196544 + }, + { + "epoch": 1.02, + "learning_rate": 3.0402286349970487e-05, + "loss": 3.5209, + "step": 1197056 + }, + { + "epoch": 1.02, + "learning_rate": 3.03939167812637e-05, + "loss": 3.523, + "step": 1197568 + }, + { + "epoch": 1.02, + "learning_rate": 3.038554721255691e-05, + "loss": 3.5307, + "step": 1198080 + }, + { + "epoch": 1.02, + "learning_rate": 3.037716126504639e-05, + "loss": 3.5167, + "step": 1198592 + }, + { + "epoch": 1.02, + "learning_rate": 3.036877531753587e-05, + "loss": 3.5242, + "step": 1199104 + }, + { + "epoch": 1.02, + "learning_rate": 3.036038937002535e-05, + "loss": 3.5261, + "step": 1199616 + }, + { + "epoch": 1.02, + "learning_rate": 3.035200342251483e-05, + "loss": 3.528, + "step": 1200128 + }, + { + "epoch": 1.02, + "learning_rate": 3.034363385380804e-05, + "loss": 3.5155, + "step": 1200640 + }, + { + "epoch": 1.02, + "learning_rate": 3.033524790629752e-05, + "loss": 3.5304, + "step": 1201152 + }, + { + "epoch": 1.02, + "learning_rate": 3.0326861958787e-05, + "loss": 3.5165, + "step": 1201664 + }, + { + "epoch": 1.02, + "learning_rate": 3.031847601127648e-05, + "loss": 3.5324, + "step": 1202176 + }, + { + "epoch": 1.02, + "learning_rate": 3.031009006376596e-05, + "loss": 3.5286, + "step": 1202688 + }, + { + "epoch": 1.02, + "learning_rate": 3.030170411625544e-05, + "loss": 3.5287, + "step": 1203200 + }, + { + "epoch": 1.02, + "learning_rate": 3.0293334547548653e-05, + "loss": 3.5363, + "step": 1203712 + }, + { + "epoch": 1.02, + "learning_rate": 3.0284948600038133e-05, + "loss": 3.5353, + "step": 1204224 + }, + { + "epoch": 1.02, + "learning_rate": 3.0276562652527613e-05, + "loss": 3.526, + "step": 1204736 + }, + { + "epoch": 1.02, + "learning_rate": 3.0268176705017093e-05, + "loss": 3.5224, + "step": 1205248 + }, + { + "epoch": 1.02, + "learning_rate": 3.0259790757506573e-05, + "loss": 3.5312, + "step": 1205760 + }, + { + "epoch": 1.02, + "learning_rate": 3.0251404809996053e-05, + "loss": 3.5239, + "step": 1206272 + }, + { + "epoch": 1.02, + "learning_rate": 3.024301886248553e-05, + "loss": 3.5212, + "step": 1206784 + }, + { + "epoch": 1.02, + "learning_rate": 3.023463291497501e-05, + "loss": 3.5317, + "step": 1207296 + }, + { + "epoch": 1.02, + "learning_rate": 3.0226263346268225e-05, + "loss": 3.527, + "step": 1207808 + }, + { + "epoch": 1.02, + "learning_rate": 3.0217877398757705e-05, + "loss": 3.5309, + "step": 1208320 + }, + { + "epoch": 1.02, + "learning_rate": 3.020949145124718e-05, + "loss": 3.5212, + "step": 1208832 + }, + { + "epoch": 1.02, + "learning_rate": 3.020110550373666e-05, + "loss": 3.5228, + "step": 1209344 + }, + { + "epoch": 1.02, + "learning_rate": 3.0192735935029874e-05, + "loss": 3.5291, + "step": 1209856 + }, + { + "epoch": 1.02, + "learning_rate": 3.018434998751935e-05, + "loss": 3.5224, + "step": 1210368 + }, + { + "epoch": 1.02, + "learning_rate": 3.017596404000883e-05, + "loss": 3.5266, + "step": 1210880 + }, + { + "epoch": 1.02, + "learning_rate": 3.016757809249831e-05, + "loss": 3.5414, + "step": 1211392 + }, + { + "epoch": 1.02, + "learning_rate": 3.0159208523791527e-05, + "loss": 3.5313, + "step": 1211904 + }, + { + "epoch": 1.02, + "learning_rate": 3.0150822576281003e-05, + "loss": 3.5241, + "step": 1212416 + }, + { + "epoch": 1.02, + "learning_rate": 3.0142436628770483e-05, + "loss": 3.5294, + "step": 1212928 + }, + { + "epoch": 1.02, + "learning_rate": 3.0134050681259963e-05, + "loss": 3.5284, + "step": 1213440 + }, + { + "epoch": 1.02, + "learning_rate": 3.012568111255318e-05, + "loss": 3.5283, + "step": 1213952 + }, + { + "epoch": 1.02, + "learning_rate": 3.0117295165042652e-05, + "loss": 3.5249, + "step": 1214464 + }, + { + "epoch": 1.02, + "learning_rate": 3.0108925596335868e-05, + "loss": 3.5214, + "step": 1214976 + }, + { + "epoch": 1.02, + "learning_rate": 3.0100539648825348e-05, + "loss": 3.5387, + "step": 1215488 + }, + { + "epoch": 1.02, + "learning_rate": 3.0092153701314825e-05, + "loss": 3.538, + "step": 1216000 + }, + { + "epoch": 1.02, + "learning_rate": 3.0083767753804304e-05, + "loss": 3.5317, + "step": 1216512 + }, + { + "epoch": 1.02, + "learning_rate": 3.0075381806293784e-05, + "loss": 3.5268, + "step": 1217024 + }, + { + "epoch": 1.02, + "learning_rate": 3.0067012237587e-05, + "loss": 3.5365, + "step": 1217536 + }, + { + "epoch": 1.02, + "learning_rate": 3.0058626290076473e-05, + "loss": 3.5249, + "step": 1218048 + }, + { + "epoch": 1.02, + "learning_rate": 3.0050240342565953e-05, + "loss": 3.5271, + "step": 1218560 + }, + { + "epoch": 1.02, + "learning_rate": 3.0041854395055437e-05, + "loss": 3.5252, + "step": 1219072 + }, + { + "epoch": 1.02, + "learning_rate": 3.0033468447544917e-05, + "loss": 3.525, + "step": 1219584 + }, + { + "epoch": 1.02, + "learning_rate": 3.0025082500034397e-05, + "loss": 3.5351, + "step": 1220096 + }, + { + "epoch": 1.02, + "learning_rate": 3.0016696552523877e-05, + "loss": 3.5172, + "step": 1220608 + }, + { + "epoch": 1.03, + "learning_rate": 3.0008310605013357e-05, + "loss": 3.5249, + "step": 1221120 + }, + { + "epoch": 1.03, + "eval_loss": 3.8492510318756104, + "eval_runtime": 302.2501, + "eval_samples_per_second": 1262.501, + "eval_steps_per_second": 39.454, + "step": 1221120 + }, + { + "epoch": 0.0, + "learning_rate": 2.9999924657502837e-05, + "loss": 3.5135, + "step": 1221632 + }, + { + "epoch": 0.0, + "learning_rate": 2.9991538709992316e-05, + "loss": 3.5191, + "step": 1222144 + }, + { + "epoch": 0.0, + "learning_rate": 2.9983152762481796e-05, + "loss": 3.5407, + "step": 1222656 + }, + { + "epoch": 0.0, + "learning_rate": 2.997478319377501e-05, + "loss": 3.5407, + "step": 1223168 + }, + { + "epoch": 0.0, + "learning_rate": 2.996639724626449e-05, + "loss": 3.5251, + "step": 1223680 + }, + { + "epoch": 0.0, + "learning_rate": 2.995801129875397e-05, + "loss": 3.5254, + "step": 1224192 + }, + { + "epoch": 0.0, + "learning_rate": 2.994962535124345e-05, + "loss": 3.5228, + "step": 1224704 + }, + { + "epoch": 0.0, + "learning_rate": 2.9941255782536658e-05, + "loss": 3.5171, + "step": 1225216 + }, + { + "epoch": 0.0, + "learning_rate": 2.993288621382987e-05, + "loss": 3.5333, + "step": 1225728 + }, + { + "epoch": 0.0, + "learning_rate": 2.992450026631935e-05, + "loss": 3.5231, + "step": 1226240 + }, + { + "epoch": 0.0, + "learning_rate": 2.991611431880883e-05, + "loss": 3.5304, + "step": 1226752 + }, + { + "epoch": 0.0, + "learning_rate": 2.990772837129831e-05, + "loss": 3.5269, + "step": 1227264 + }, + { + "epoch": 0.0, + "learning_rate": 2.989935880259152e-05, + "loss": 3.5208, + "step": 1227776 + }, + { + "epoch": 0.0, + "learning_rate": 2.9890972855081e-05, + "loss": 3.5124, + "step": 1228288 + }, + { + "epoch": 0.0, + "learning_rate": 2.9882586907570483e-05, + "loss": 3.5082, + "step": 1228800 + }, + { + "epoch": 0.0, + "learning_rate": 2.9874200960059963e-05, + "loss": 3.5218, + "step": 1229312 + }, + { + "epoch": 0.0, + "learning_rate": 2.9865815012549443e-05, + "loss": 3.5199, + "step": 1229824 + }, + { + "epoch": 0.0, + "learning_rate": 2.9857429065038923e-05, + "loss": 3.525, + "step": 1230336 + }, + { + "epoch": 0.0, + "learning_rate": 2.9849043117528402e-05, + "loss": 3.5184, + "step": 1230848 + }, + { + "epoch": 0.0, + "learning_rate": 2.9840657170017882e-05, + "loss": 3.5438, + "step": 1231360 + }, + { + "epoch": 0.0, + "learning_rate": 2.983228760131109e-05, + "loss": 3.5244, + "step": 1231872 + }, + { + "epoch": 0.0, + "learning_rate": 2.9823901653800575e-05, + "loss": 3.5261, + "step": 1232384 + }, + { + "epoch": 0.0, + "learning_rate": 2.9815515706290055e-05, + "loss": 3.5228, + "step": 1232896 + }, + { + "epoch": 0.0, + "learning_rate": 2.9807129758779535e-05, + "loss": 3.5402, + "step": 1233408 + }, + { + "epoch": 0.0, + "learning_rate": 2.9798760190072744e-05, + "loss": 3.5122, + "step": 1233920 + }, + { + "epoch": 0.0, + "learning_rate": 2.9790390621365953e-05, + "loss": 3.5232, + "step": 1234432 + }, + { + "epoch": 0.0, + "learning_rate": 2.9782004673855436e-05, + "loss": 3.5216, + "step": 1234944 + }, + { + "epoch": 0.0, + "learning_rate": 2.9773618726344916e-05, + "loss": 3.5204, + "step": 1235456 + }, + { + "epoch": 0.0, + "learning_rate": 2.9765232778834396e-05, + "loss": 3.5179, + "step": 1235968 + }, + { + "epoch": 0.01, + "learning_rate": 2.9756846831323876e-05, + "loss": 3.5172, + "step": 1236480 + }, + { + "epoch": 0.01, + "learning_rate": 2.9748460883813356e-05, + "loss": 3.5218, + "step": 1236992 + }, + { + "epoch": 0.01, + "learning_rate": 2.9740074936302836e-05, + "loss": 3.5286, + "step": 1237504 + }, + { + "epoch": 0.01, + "learning_rate": 2.9731688988792313e-05, + "loss": 3.523, + "step": 1238016 + }, + { + "epoch": 0.01, + "learning_rate": 2.972331942008553e-05, + "loss": 3.5307, + "step": 1238528 + }, + { + "epoch": 0.01, + "learning_rate": 2.971493347257501e-05, + "loss": 3.5198, + "step": 1239040 + }, + { + "epoch": 0.01, + "learning_rate": 2.970654752506448e-05, + "loss": 3.5291, + "step": 1239552 + }, + { + "epoch": 0.01, + "learning_rate": 2.969816157755396e-05, + "loss": 3.5084, + "step": 1240064 + }, + { + "epoch": 0.01, + "learning_rate": 2.9689792008847177e-05, + "loss": 3.5271, + "step": 1240576 + }, + { + "epoch": 0.01, + "learning_rate": 2.968142244014039e-05, + "loss": 3.5035, + "step": 1241088 + }, + { + "epoch": 0.01, + "learning_rate": 2.967303649262987e-05, + "loss": 3.5183, + "step": 1241600 + }, + { + "epoch": 0.01, + "learning_rate": 2.966465054511935e-05, + "loss": 3.5171, + "step": 1242112 + }, + { + "epoch": 0.01, + "learning_rate": 2.965626459760883e-05, + "loss": 3.5269, + "step": 1242624 + }, + { + "epoch": 0.01, + "learning_rate": 2.964787865009831e-05, + "loss": 3.5171, + "step": 1243136 + }, + { + "epoch": 0.01, + "learning_rate": 2.9639492702587783e-05, + "loss": 3.5264, + "step": 1243648 + }, + { + "epoch": 0.01, + "learning_rate": 2.9631106755077266e-05, + "loss": 3.5225, + "step": 1244160 + }, + { + "epoch": 0.01, + "learning_rate": 2.9622720807566746e-05, + "loss": 3.5237, + "step": 1244672 + }, + { + "epoch": 0.01, + "learning_rate": 2.9614351238859955e-05, + "loss": 3.5239, + "step": 1245184 + }, + { + "epoch": 0.01, + "learning_rate": 2.9605965291349435e-05, + "loss": 3.4985, + "step": 1245696 + }, + { + "epoch": 0.01, + "learning_rate": 2.9597579343838915e-05, + "loss": 3.5202, + "step": 1246208 + }, + { + "epoch": 0.01, + "learning_rate": 2.9589193396328395e-05, + "loss": 3.5259, + "step": 1246720 + }, + { + "epoch": 0.01, + "learning_rate": 2.9580823827621608e-05, + "loss": 3.5237, + "step": 1247232 + }, + { + "epoch": 0.01, + "learning_rate": 2.9572437880111088e-05, + "loss": 3.5216, + "step": 1247744 + }, + { + "epoch": 0.01, + "learning_rate": 2.9564051932600568e-05, + "loss": 3.5112, + "step": 1248256 + }, + { + "epoch": 0.01, + "learning_rate": 2.9555665985090048e-05, + "loss": 3.5169, + "step": 1248768 + }, + { + "epoch": 0.01, + "learning_rate": 2.9547296416383257e-05, + "loss": 3.5025, + "step": 1249280 + }, + { + "epoch": 0.01, + "learning_rate": 2.9538910468872737e-05, + "loss": 3.5281, + "step": 1249792 + }, + { + "epoch": 0.01, + "learning_rate": 2.953052452136222e-05, + "loss": 3.5009, + "step": 1250304 + }, + { + "epoch": 0.01, + "learning_rate": 2.95221385738517e-05, + "loss": 3.5247, + "step": 1250816 + }, + { + "epoch": 0.01, + "learning_rate": 2.951376900514491e-05, + "loss": 3.5189, + "step": 1251328 + }, + { + "epoch": 0.01, + "learning_rate": 2.950538305763439e-05, + "loss": 3.5074, + "step": 1251840 + }, + { + "epoch": 0.01, + "learning_rate": 2.949699711012387e-05, + "loss": 3.5164, + "step": 1252352 + }, + { + "epoch": 0.01, + "learning_rate": 2.948862754141708e-05, + "loss": 3.5091, + "step": 1252864 + }, + { + "epoch": 0.01, + "learning_rate": 2.948024159390656e-05, + "loss": 3.5112, + "step": 1253376 + }, + { + "epoch": 0.01, + "learning_rate": 2.947185564639604e-05, + "loss": 3.5046, + "step": 1253888 + }, + { + "epoch": 0.01, + "learning_rate": 2.946346969888552e-05, + "loss": 3.5263, + "step": 1254400 + }, + { + "epoch": 0.01, + "learning_rate": 2.945510013017873e-05, + "loss": 3.5102, + "step": 1254912 + }, + { + "epoch": 0.01, + "learning_rate": 2.944671418266821e-05, + "loss": 3.5088, + "step": 1255424 + }, + { + "epoch": 0.01, + "learning_rate": 2.943832823515769e-05, + "loss": 3.5075, + "step": 1255936 + }, + { + "epoch": 0.01, + "learning_rate": 2.9429942287647174e-05, + "loss": 3.5137, + "step": 1256448 + }, + { + "epoch": 0.01, + "learning_rate": 2.9421572718940383e-05, + "loss": 3.5197, + "step": 1256960 + }, + { + "epoch": 0.01, + "learning_rate": 2.9413186771429863e-05, + "loss": 3.5196, + "step": 1257472 + }, + { + "epoch": 0.01, + "learning_rate": 2.9404800823919343e-05, + "loss": 3.5092, + "step": 1257984 + }, + { + "epoch": 0.01, + "learning_rate": 2.9396414876408823e-05, + "loss": 3.523, + "step": 1258496 + }, + { + "epoch": 0.01, + "learning_rate": 2.9388045307702035e-05, + "loss": 3.5271, + "step": 1259008 + }, + { + "epoch": 0.01, + "learning_rate": 2.9379659360191515e-05, + "loss": 3.5058, + "step": 1259520 + }, + { + "epoch": 0.01, + "learning_rate": 2.9371273412680995e-05, + "loss": 3.5255, + "step": 1260032 + }, + { + "epoch": 0.01, + "learning_rate": 2.9362887465170475e-05, + "loss": 3.516, + "step": 1260544 + }, + { + "epoch": 0.01, + "learning_rate": 2.9354517896463684e-05, + "loss": 3.5162, + "step": 1261056 + }, + { + "epoch": 0.01, + "learning_rate": 2.9346131948953164e-05, + "loss": 3.5104, + "step": 1261568 + }, + { + "epoch": 0.01, + "learning_rate": 2.9337746001442644e-05, + "loss": 3.516, + "step": 1262080 + }, + { + "epoch": 0.01, + "learning_rate": 2.9329360053932124e-05, + "loss": 3.5253, + "step": 1262592 + }, + { + "epoch": 0.01, + "learning_rate": 2.9321006864029072e-05, + "loss": 3.5298, + "step": 1263104 + }, + { + "epoch": 0.01, + "learning_rate": 2.9312620916518552e-05, + "loss": 3.5068, + "step": 1263616 + }, + { + "epoch": 0.01, + "learning_rate": 2.9304234969008025e-05, + "loss": 3.491, + "step": 1264128 + }, + { + "epoch": 0.01, + "learning_rate": 2.9295849021497505e-05, + "loss": 3.5106, + "step": 1264640 + }, + { + "epoch": 0.01, + "learning_rate": 2.9287463073986985e-05, + "loss": 3.5134, + "step": 1265152 + }, + { + "epoch": 0.01, + "learning_rate": 2.927907712647647e-05, + "loss": 3.5184, + "step": 1265664 + }, + { + "epoch": 0.01, + "learning_rate": 2.927069117896595e-05, + "loss": 3.5075, + "step": 1266176 + }, + { + "epoch": 0.01, + "learning_rate": 2.926230523145543e-05, + "loss": 3.5185, + "step": 1266688 + }, + { + "epoch": 0.02, + "learning_rate": 2.9253935662748638e-05, + "loss": 3.5146, + "step": 1267200 + }, + { + "epoch": 0.02, + "learning_rate": 2.9245549715238118e-05, + "loss": 3.5154, + "step": 1267712 + }, + { + "epoch": 0.02, + "learning_rate": 2.9237163767727598e-05, + "loss": 3.5048, + "step": 1268224 + }, + { + "epoch": 0.02, + "learning_rate": 2.922879419902081e-05, + "loss": 3.5192, + "step": 1268736 + }, + { + "epoch": 0.02, + "learning_rate": 2.922040825151029e-05, + "loss": 3.5105, + "step": 1269248 + }, + { + "epoch": 0.02, + "learning_rate": 2.921202230399977e-05, + "loss": 3.5211, + "step": 1269760 + }, + { + "epoch": 0.02, + "learning_rate": 2.920363635648925e-05, + "loss": 3.5157, + "step": 1270272 + }, + { + "epoch": 0.02, + "learning_rate": 2.919525040897873e-05, + "loss": 3.4994, + "step": 1270784 + }, + { + "epoch": 0.02, + "learning_rate": 2.918686446146821e-05, + "loss": 3.5176, + "step": 1271296 + }, + { + "epoch": 0.02, + "learning_rate": 2.917847851395769e-05, + "loss": 3.5179, + "step": 1271808 + }, + { + "epoch": 0.02, + "learning_rate": 2.917009256644717e-05, + "loss": 3.5206, + "step": 1272320 + }, + { + "epoch": 0.02, + "learning_rate": 2.9161722997740382e-05, + "loss": 3.5124, + "step": 1272832 + }, + { + "epoch": 0.02, + "learning_rate": 2.9153337050229862e-05, + "loss": 3.5077, + "step": 1273344 + }, + { + "epoch": 0.02, + "learning_rate": 2.9144951102719342e-05, + "loss": 3.5082, + "step": 1273856 + }, + { + "epoch": 0.02, + "learning_rate": 2.913658153401255e-05, + "loss": 3.5173, + "step": 1274368 + }, + { + "epoch": 0.02, + "learning_rate": 2.9128211965305764e-05, + "loss": 3.5023, + "step": 1274880 + }, + { + "epoch": 0.02, + "learning_rate": 2.9119826017795244e-05, + "loss": 3.5105, + "step": 1275392 + }, + { + "epoch": 0.02, + "learning_rate": 2.9111440070284724e-05, + "loss": 3.5139, + "step": 1275904 + }, + { + "epoch": 0.02, + "learning_rate": 2.9103054122774204e-05, + "loss": 3.5154, + "step": 1276416 + }, + { + "epoch": 0.02, + "learning_rate": 2.9094668175263684e-05, + "loss": 3.5026, + "step": 1276928 + }, + { + "epoch": 0.02, + "learning_rate": 2.9086282227753164e-05, + "loss": 3.5136, + "step": 1277440 + }, + { + "epoch": 0.02, + "learning_rate": 2.9077896280242644e-05, + "loss": 3.5073, + "step": 1277952 + }, + { + "epoch": 0.02, + "learning_rate": 2.9069526711535856e-05, + "loss": 3.5191, + "step": 1278464 + }, + { + "epoch": 0.02, + "learning_rate": 2.9061140764025336e-05, + "loss": 3.5116, + "step": 1278976 + }, + { + "epoch": 0.02, + "learning_rate": 2.9052754816514816e-05, + "loss": 3.5124, + "step": 1279488 + }, + { + "epoch": 0.02, + "learning_rate": 2.9044368869004296e-05, + "loss": 3.5243, + "step": 1280000 + }, + { + "epoch": 0.02, + "learning_rate": 2.9035982921493776e-05, + "loss": 3.5206, + "step": 1280512 + }, + { + "epoch": 0.02, + "learning_rate": 2.9027613352786985e-05, + "loss": 3.5129, + "step": 1281024 + }, + { + "epoch": 0.02, + "learning_rate": 2.9019227405276468e-05, + "loss": 3.5073, + "step": 1281536 + }, + { + "epoch": 0.02, + "learning_rate": 2.9010841457765948e-05, + "loss": 3.5165, + "step": 1282048 + }, + { + "epoch": 0.02, + "learning_rate": 2.9002455510255428e-05, + "loss": 3.508, + "step": 1282560 + }, + { + "epoch": 0.02, + "learning_rate": 2.8994069562744908e-05, + "loss": 3.5085, + "step": 1283072 + }, + { + "epoch": 0.02, + "learning_rate": 2.8985683615234388e-05, + "loss": 3.5219, + "step": 1283584 + }, + { + "epoch": 0.02, + "learning_rate": 2.897729766772386e-05, + "loss": 3.5085, + "step": 1284096 + }, + { + "epoch": 0.02, + "learning_rate": 2.8968911720213345e-05, + "loss": 3.5184, + "step": 1284608 + }, + { + "epoch": 0.02, + "learning_rate": 2.896054215150656e-05, + "loss": 3.505, + "step": 1285120 + }, + { + "epoch": 0.02, + "learning_rate": 2.8952156203996034e-05, + "loss": 3.5083, + "step": 1285632 + }, + { + "epoch": 0.02, + "learning_rate": 2.8943770256485514e-05, + "loss": 3.5157, + "step": 1286144 + }, + { + "epoch": 0.02, + "learning_rate": 2.8935384308974994e-05, + "loss": 3.5081, + "step": 1286656 + }, + { + "epoch": 0.02, + "learning_rate": 2.892701474026821e-05, + "loss": 3.5071, + "step": 1287168 + }, + { + "epoch": 0.02, + "learning_rate": 2.8918645171561422e-05, + "loss": 3.5292, + "step": 1287680 + }, + { + "epoch": 0.02, + "learning_rate": 2.8910259224050902e-05, + "loss": 3.5165, + "step": 1288192 + }, + { + "epoch": 0.02, + "learning_rate": 2.8901873276540382e-05, + "loss": 3.5143, + "step": 1288704 + }, + { + "epoch": 0.02, + "learning_rate": 2.8893487329029862e-05, + "loss": 3.511, + "step": 1289216 + }, + { + "epoch": 0.02, + "learning_rate": 2.8885101381519335e-05, + "loss": 3.5126, + "step": 1289728 + }, + { + "epoch": 0.02, + "learning_rate": 2.8876715434008815e-05, + "loss": 3.5158, + "step": 1290240 + }, + { + "epoch": 0.02, + "learning_rate": 2.8868329486498298e-05, + "loss": 3.5103, + "step": 1290752 + }, + { + "epoch": 0.02, + "learning_rate": 2.8859959917791507e-05, + "loss": 3.5098, + "step": 1291264 + }, + { + "epoch": 0.02, + "learning_rate": 2.8851573970280987e-05, + "loss": 3.5244, + "step": 1291776 + }, + { + "epoch": 0.02, + "learning_rate": 2.8843188022770467e-05, + "loss": 3.5207, + "step": 1292288 + }, + { + "epoch": 0.02, + "learning_rate": 2.8834802075259947e-05, + "loss": 3.5218, + "step": 1292800 + }, + { + "epoch": 0.02, + "learning_rate": 2.8826416127749427e-05, + "loss": 3.5132, + "step": 1293312 + }, + { + "epoch": 0.02, + "learning_rate": 2.8818030180238907e-05, + "loss": 3.5172, + "step": 1293824 + }, + { + "epoch": 0.02, + "learning_rate": 2.880966061153212e-05, + "loss": 3.5134, + "step": 1294336 + }, + { + "epoch": 0.02, + "learning_rate": 2.88012746640216e-05, + "loss": 3.514, + "step": 1294848 + }, + { + "epoch": 0.02, + "learning_rate": 2.879288871651108e-05, + "loss": 3.5122, + "step": 1295360 + }, + { + "epoch": 0.02, + "learning_rate": 2.878450276900056e-05, + "loss": 3.5126, + "step": 1295872 + }, + { + "epoch": 0.02, + "learning_rate": 2.877611682149004e-05, + "loss": 3.5187, + "step": 1296384 + }, + { + "epoch": 0.02, + "learning_rate": 2.876773087397952e-05, + "loss": 3.507, + "step": 1296896 + }, + { + "epoch": 0.02, + "learning_rate": 2.8759344926469e-05, + "loss": 3.5083, + "step": 1297408 + }, + { + "epoch": 0.03, + "eval_loss": 3.85086989402771, + "eval_runtime": 307.3348, + "eval_samples_per_second": 1241.613, + "eval_steps_per_second": 38.801, + "step": 1297440 + }, + { + "epoch": 1.0, + "learning_rate": 2.8750975357762212e-05, + "loss": 3.5014, + "step": 1297920 + }, + { + "epoch": 1.0, + "learning_rate": 2.8742589410251692e-05, + "loss": 3.507, + "step": 1298432 + }, + { + "epoch": 1.0, + "learning_rate": 2.8734203462741172e-05, + "loss": 3.5288, + "step": 1298944 + }, + { + "epoch": 1.0, + "learning_rate": 2.8725817515230652e-05, + "loss": 3.5214, + "step": 1299456 + }, + { + "epoch": 1.0, + "learning_rate": 2.871744794652386e-05, + "loss": 3.5157, + "step": 1299968 + }, + { + "epoch": 1.0, + "learning_rate": 2.870906199901334e-05, + "loss": 3.5086, + "step": 1300480 + }, + { + "epoch": 1.0, + "learning_rate": 2.8700676051502824e-05, + "loss": 3.5099, + "step": 1300992 + }, + { + "epoch": 1.0, + "learning_rate": 2.8692290103992304e-05, + "loss": 3.5045, + "step": 1301504 + }, + { + "epoch": 1.0, + "learning_rate": 2.8683920535285513e-05, + "loss": 3.5169, + "step": 1302016 + }, + { + "epoch": 1.0, + "learning_rate": 2.8675534587774993e-05, + "loss": 3.5159, + "step": 1302528 + }, + { + "epoch": 1.0, + "learning_rate": 2.8667148640264473e-05, + "loss": 3.5131, + "step": 1303040 + }, + { + "epoch": 1.0, + "learning_rate": 2.8658762692753953e-05, + "loss": 3.5162, + "step": 1303552 + }, + { + "epoch": 1.0, + "learning_rate": 2.8650393124047166e-05, + "loss": 3.5061, + "step": 1304064 + }, + { + "epoch": 1.0, + "learning_rate": 2.8642007176536645e-05, + "loss": 3.5026, + "step": 1304576 + }, + { + "epoch": 1.0, + "learning_rate": 2.8633621229026125e-05, + "loss": 3.4991, + "step": 1305088 + }, + { + "epoch": 1.0, + "learning_rate": 2.8625235281515605e-05, + "loss": 3.5016, + "step": 1305600 + }, + { + "epoch": 1.0, + "learning_rate": 2.8616849334005085e-05, + "loss": 3.5074, + "step": 1306112 + }, + { + "epoch": 1.0, + "learning_rate": 2.8608463386494565e-05, + "loss": 3.5076, + "step": 1306624 + }, + { + "epoch": 1.0, + "learning_rate": 2.8600077438984045e-05, + "loss": 3.5085, + "step": 1307136 + }, + { + "epoch": 1.0, + "learning_rate": 2.8591691491473522e-05, + "loss": 3.5327, + "step": 1307648 + }, + { + "epoch": 1.0, + "learning_rate": 2.8583321922766738e-05, + "loss": 3.5114, + "step": 1308160 + }, + { + "epoch": 1.0, + "learning_rate": 2.8574935975256218e-05, + "loss": 3.5161, + "step": 1308672 + }, + { + "epoch": 1.0, + "learning_rate": 2.8566566406549427e-05, + "loss": 3.507, + "step": 1309184 + }, + { + "epoch": 1.0, + "learning_rate": 2.8558180459038907e-05, + "loss": 3.5212, + "step": 1309696 + }, + { + "epoch": 1.0, + "learning_rate": 2.8549794511528387e-05, + "loss": 3.5042, + "step": 1310208 + }, + { + "epoch": 1.0, + "learning_rate": 2.854140856401787e-05, + "loss": 3.5096, + "step": 1310720 + }, + { + "epoch": 1.0, + "learning_rate": 2.8533022616507343e-05, + "loss": 3.5077, + "step": 1311232 + }, + { + "epoch": 1.0, + "learning_rate": 2.8524636668996823e-05, + "loss": 3.5076, + "step": 1311744 + }, + { + "epoch": 1.0, + "learning_rate": 2.8516250721486303e-05, + "loss": 3.5015, + "step": 1312256 + }, + { + "epoch": 1.01, + "learning_rate": 2.8507864773975783e-05, + "loss": 3.5019, + "step": 1312768 + }, + { + "epoch": 1.01, + "learning_rate": 2.8499495205268996e-05, + "loss": 3.5094, + "step": 1313280 + }, + { + "epoch": 1.01, + "learning_rate": 2.8491109257758476e-05, + "loss": 3.5188, + "step": 1313792 + }, + { + "epoch": 1.01, + "learning_rate": 2.8482723310247955e-05, + "loss": 3.5081, + "step": 1314304 + }, + { + "epoch": 1.01, + "learning_rate": 2.8474353741541165e-05, + "loss": 3.5143, + "step": 1314816 + }, + { + "epoch": 1.01, + "learning_rate": 2.8465967794030645e-05, + "loss": 3.5028, + "step": 1315328 + }, + { + "epoch": 1.01, + "learning_rate": 2.8457581846520124e-05, + "loss": 3.5158, + "step": 1315840 + }, + { + "epoch": 1.01, + "learning_rate": 2.8449195899009608e-05, + "loss": 3.4992, + "step": 1316352 + }, + { + "epoch": 1.01, + "learning_rate": 2.8440809951499088e-05, + "loss": 3.5092, + "step": 1316864 + }, + { + "epoch": 1.01, + "learning_rate": 2.8432440382792297e-05, + "loss": 3.4912, + "step": 1317376 + }, + { + "epoch": 1.01, + "learning_rate": 2.8424054435281777e-05, + "loss": 3.5081, + "step": 1317888 + }, + { + "epoch": 1.01, + "learning_rate": 2.8415668487771257e-05, + "loss": 3.5002, + "step": 1318400 + }, + { + "epoch": 1.01, + "learning_rate": 2.8407282540260737e-05, + "loss": 3.5144, + "step": 1318912 + }, + { + "epoch": 1.01, + "learning_rate": 2.8398896592750217e-05, + "loss": 3.5034, + "step": 1319424 + }, + { + "epoch": 1.01, + "learning_rate": 2.83905106452397e-05, + "loss": 3.5105, + "step": 1319936 + }, + { + "epoch": 1.01, + "learning_rate": 2.838212469772918e-05, + "loss": 3.5111, + "step": 1320448 + }, + { + "epoch": 1.01, + "learning_rate": 2.837373875021866e-05, + "loss": 3.5107, + "step": 1320960 + }, + { + "epoch": 1.01, + "learning_rate": 2.836536918151187e-05, + "loss": 3.5127, + "step": 1321472 + }, + { + "epoch": 1.01, + "learning_rate": 2.835698323400135e-05, + "loss": 3.489, + "step": 1321984 + }, + { + "epoch": 1.01, + "learning_rate": 2.834859728649083e-05, + "loss": 3.5013, + "step": 1322496 + }, + { + "epoch": 1.01, + "learning_rate": 2.834021133898031e-05, + "loss": 3.512, + "step": 1323008 + }, + { + "epoch": 1.01, + "learning_rate": 2.833184177027352e-05, + "loss": 3.5119, + "step": 1323520 + }, + { + "epoch": 1.01, + "learning_rate": 2.8323455822763e-05, + "loss": 3.5053, + "step": 1324032 + }, + { + "epoch": 1.01, + "learning_rate": 2.831506987525248e-05, + "loss": 3.5005, + "step": 1324544 + }, + { + "epoch": 1.01, + "learning_rate": 2.830668392774196e-05, + "loss": 3.5031, + "step": 1325056 + }, + { + "epoch": 1.01, + "learning_rate": 2.829831435903517e-05, + "loss": 3.4949, + "step": 1325568 + }, + { + "epoch": 1.01, + "learning_rate": 2.8289944790328383e-05, + "loss": 3.5079, + "step": 1326080 + }, + { + "epoch": 1.01, + "learning_rate": 2.8281558842817863e-05, + "loss": 3.4941, + "step": 1326592 + }, + { + "epoch": 1.01, + "learning_rate": 2.8273172895307343e-05, + "loss": 3.5057, + "step": 1327104 + }, + { + "epoch": 1.01, + "learning_rate": 2.8264786947796823e-05, + "loss": 3.5117, + "step": 1327616 + }, + { + "epoch": 1.01, + "learning_rate": 2.8256401000286303e-05, + "loss": 3.4958, + "step": 1328128 + }, + { + "epoch": 1.01, + "learning_rate": 2.8248015052775783e-05, + "loss": 3.4994, + "step": 1328640 + }, + { + "epoch": 1.01, + "learning_rate": 2.8239629105265263e-05, + "loss": 3.4967, + "step": 1329152 + }, + { + "epoch": 1.01, + "learning_rate": 2.8231243157754746e-05, + "loss": 3.4942, + "step": 1329664 + }, + { + "epoch": 1.01, + "learning_rate": 2.8222873589047955e-05, + "loss": 3.4918, + "step": 1330176 + }, + { + "epoch": 1.01, + "learning_rate": 2.8214487641537435e-05, + "loss": 3.5121, + "step": 1330688 + }, + { + "epoch": 1.01, + "learning_rate": 2.8206101694026915e-05, + "loss": 3.4993, + "step": 1331200 + }, + { + "epoch": 1.01, + "learning_rate": 2.8197715746516395e-05, + "loss": 3.4958, + "step": 1331712 + }, + { + "epoch": 1.01, + "learning_rate": 2.8189346177809607e-05, + "loss": 3.4964, + "step": 1332224 + }, + { + "epoch": 1.01, + "learning_rate": 2.8180960230299087e-05, + "loss": 3.4977, + "step": 1332736 + }, + { + "epoch": 1.01, + "learning_rate": 2.8172574282788567e-05, + "loss": 3.5068, + "step": 1333248 + }, + { + "epoch": 1.01, + "learning_rate": 2.8164188335278047e-05, + "loss": 3.5022, + "step": 1333760 + }, + { + "epoch": 1.01, + "learning_rate": 2.8155818766571256e-05, + "loss": 3.4983, + "step": 1334272 + }, + { + "epoch": 1.01, + "learning_rate": 2.8147432819060736e-05, + "loss": 3.5093, + "step": 1334784 + }, + { + "epoch": 1.01, + "learning_rate": 2.8139046871550216e-05, + "loss": 3.5122, + "step": 1335296 + }, + { + "epoch": 1.01, + "learning_rate": 2.8130660924039696e-05, + "loss": 3.4952, + "step": 1335808 + }, + { + "epoch": 1.01, + "learning_rate": 2.812229135533291e-05, + "loss": 3.5104, + "step": 1336320 + }, + { + "epoch": 1.01, + "learning_rate": 2.811390540782239e-05, + "loss": 3.5025, + "step": 1336832 + }, + { + "epoch": 1.01, + "learning_rate": 2.810551946031187e-05, + "loss": 3.5028, + "step": 1337344 + }, + { + "epoch": 1.01, + "learning_rate": 2.8097149891605078e-05, + "loss": 3.4993, + "step": 1337856 + }, + { + "epoch": 1.01, + "learning_rate": 2.808876394409456e-05, + "loss": 3.501, + "step": 1338368 + }, + { + "epoch": 1.01, + "learning_rate": 2.808037799658404e-05, + "loss": 3.5103, + "step": 1338880 + }, + { + "epoch": 1.01, + "learning_rate": 2.807199204907352e-05, + "loss": 3.5097, + "step": 1339392 + }, + { + "epoch": 1.01, + "learning_rate": 2.806362248036673e-05, + "loss": 3.4994, + "step": 1339904 + }, + { + "epoch": 1.01, + "learning_rate": 2.805525291165994e-05, + "loss": 3.479, + "step": 1340416 + }, + { + "epoch": 1.01, + "learning_rate": 2.8046866964149422e-05, + "loss": 3.4939, + "step": 1340928 + }, + { + "epoch": 1.01, + "learning_rate": 2.8038481016638902e-05, + "loss": 3.5037, + "step": 1341440 + }, + { + "epoch": 1.01, + "learning_rate": 2.8030095069128382e-05, + "loss": 3.504, + "step": 1341952 + }, + { + "epoch": 1.01, + "learning_rate": 2.8021709121617862e-05, + "loss": 3.499, + "step": 1342464 + }, + { + "epoch": 1.01, + "learning_rate": 2.8013323174107342e-05, + "loss": 3.5007, + "step": 1342976 + }, + { + "epoch": 1.02, + "learning_rate": 2.8004937226596822e-05, + "loss": 3.5048, + "step": 1343488 + }, + { + "epoch": 1.02, + "learning_rate": 2.7996551279086302e-05, + "loss": 3.4988, + "step": 1344000 + }, + { + "epoch": 1.02, + "learning_rate": 2.798818171037951e-05, + "loss": 3.492, + "step": 1344512 + }, + { + "epoch": 1.02, + "learning_rate": 2.7979812141672724e-05, + "loss": 3.5092, + "step": 1345024 + }, + { + "epoch": 1.02, + "learning_rate": 2.7971426194162204e-05, + "loss": 3.4933, + "step": 1345536 + }, + { + "epoch": 1.02, + "learning_rate": 2.7963040246651684e-05, + "loss": 3.5091, + "step": 1346048 + }, + { + "epoch": 1.02, + "learning_rate": 2.7954654299141164e-05, + "loss": 3.5045, + "step": 1346560 + }, + { + "epoch": 1.02, + "learning_rate": 2.7946268351630644e-05, + "loss": 3.4845, + "step": 1347072 + }, + { + "epoch": 1.02, + "learning_rate": 2.7937898782923856e-05, + "loss": 3.5062, + "step": 1347584 + }, + { + "epoch": 1.02, + "learning_rate": 2.7929512835413336e-05, + "loss": 3.4999, + "step": 1348096 + }, + { + "epoch": 1.02, + "learning_rate": 2.7921126887902816e-05, + "loss": 3.5079, + "step": 1348608 + }, + { + "epoch": 1.02, + "learning_rate": 2.7912740940392296e-05, + "loss": 3.5007, + "step": 1349120 + }, + { + "epoch": 1.02, + "learning_rate": 2.790435499288177e-05, + "loss": 3.4962, + "step": 1349632 + }, + { + "epoch": 1.02, + "learning_rate": 2.789596904537125e-05, + "loss": 3.4925, + "step": 1350144 + }, + { + "epoch": 1.02, + "learning_rate": 2.7887583097860732e-05, + "loss": 3.5056, + "step": 1350656 + }, + { + "epoch": 1.02, + "learning_rate": 2.787921352915395e-05, + "loss": 3.4885, + "step": 1351168 + }, + { + "epoch": 1.02, + "learning_rate": 2.787082758164342e-05, + "loss": 3.4929, + "step": 1351680 + }, + { + "epoch": 1.02, + "learning_rate": 2.78624416341329e-05, + "loss": 3.5032, + "step": 1352192 + }, + { + "epoch": 1.02, + "learning_rate": 2.785405568662238e-05, + "loss": 3.5002, + "step": 1352704 + }, + { + "epoch": 1.02, + "learning_rate": 2.784566973911186e-05, + "loss": 3.4861, + "step": 1353216 + }, + { + "epoch": 1.02, + "learning_rate": 2.783728379160134e-05, + "loss": 3.5033, + "step": 1353728 + }, + { + "epoch": 1.02, + "learning_rate": 2.7828897844090825e-05, + "loss": 3.4942, + "step": 1354240 + }, + { + "epoch": 1.02, + "learning_rate": 2.7820528275384034e-05, + "loss": 3.5021, + "step": 1354752 + }, + { + "epoch": 1.02, + "learning_rate": 2.7812142327873514e-05, + "loss": 3.5016, + "step": 1355264 + }, + { + "epoch": 1.02, + "learning_rate": 2.7803756380362994e-05, + "loss": 3.4945, + "step": 1355776 + }, + { + "epoch": 1.02, + "learning_rate": 2.7795370432852474e-05, + "loss": 3.5169, + "step": 1356288 + }, + { + "epoch": 1.02, + "learning_rate": 2.7786984485341954e-05, + "loss": 3.5081, + "step": 1356800 + }, + { + "epoch": 1.02, + "learning_rate": 2.7778614916635166e-05, + "loss": 3.5004, + "step": 1357312 + }, + { + "epoch": 1.02, + "learning_rate": 2.7770228969124646e-05, + "loss": 3.4919, + "step": 1357824 + }, + { + "epoch": 1.02, + "learning_rate": 2.7761843021614126e-05, + "loss": 3.5028, + "step": 1358336 + }, + { + "epoch": 1.02, + "learning_rate": 2.7753457074103606e-05, + "loss": 3.4971, + "step": 1358848 + }, + { + "epoch": 1.02, + "learning_rate": 2.7745071126593086e-05, + "loss": 3.4931, + "step": 1359360 + }, + { + "epoch": 1.02, + "learning_rate": 2.7736701557886295e-05, + "loss": 3.5066, + "step": 1359872 + }, + { + "epoch": 1.02, + "learning_rate": 2.772831561037578e-05, + "loss": 3.5004, + "step": 1360384 + }, + { + "epoch": 1.02, + "learning_rate": 2.7719929662865258e-05, + "loss": 3.4991, + "step": 1360896 + }, + { + "epoch": 1.02, + "learning_rate": 2.7711543715354738e-05, + "loss": 3.497, + "step": 1361408 + }, + { + "epoch": 1.02, + "learning_rate": 2.7703157767844218e-05, + "loss": 3.496, + "step": 1361920 + }, + { + "epoch": 1.02, + "learning_rate": 2.7694771820333698e-05, + "loss": 3.5001, + "step": 1362432 + }, + { + "epoch": 1.02, + "learning_rate": 2.7686385872823178e-05, + "loss": 3.4971, + "step": 1362944 + }, + { + "epoch": 1.02, + "learning_rate": 2.7678016304116387e-05, + "loss": 3.4913, + "step": 1363456 + }, + { + "epoch": 1.02, + "learning_rate": 2.766963035660587e-05, + "loss": 3.5158, + "step": 1363968 + }, + { + "epoch": 1.02, + "learning_rate": 2.766124440909535e-05, + "loss": 3.5088, + "step": 1364480 + }, + { + "epoch": 1.02, + "learning_rate": 2.765285846158483e-05, + "loss": 3.5008, + "step": 1364992 + }, + { + "epoch": 1.02, + "learning_rate": 2.764448889287804e-05, + "loss": 3.4995, + "step": 1365504 + }, + { + "epoch": 1.02, + "learning_rate": 2.763610294536752e-05, + "loss": 3.4949, + "step": 1366016 + }, + { + "epoch": 1.02, + "learning_rate": 2.7627716997857e-05, + "loss": 3.5059, + "step": 1366528 + }, + { + "epoch": 1.02, + "learning_rate": 2.761933105034648e-05, + "loss": 3.501, + "step": 1367040 + }, + { + "epoch": 1.02, + "learning_rate": 2.7610961481639692e-05, + "loss": 3.4956, + "step": 1367552 + }, + { + "epoch": 1.02, + "learning_rate": 2.7602575534129172e-05, + "loss": 3.5125, + "step": 1368064 + }, + { + "epoch": 1.02, + "learning_rate": 2.7594189586618652e-05, + "loss": 3.5015, + "step": 1368576 + }, + { + "epoch": 1.02, + "learning_rate": 2.7585803639108132e-05, + "loss": 3.5115, + "step": 1369088 + }, + { + "epoch": 1.02, + "learning_rate": 2.757741769159761e-05, + "loss": 3.4978, + "step": 1369600 + }, + { + "epoch": 1.02, + "learning_rate": 2.7569031744087088e-05, + "loss": 3.5075, + "step": 1370112 + }, + { + "epoch": 1.02, + "learning_rate": 2.7560662175380304e-05, + "loss": 3.5022, + "step": 1370624 + }, + { + "epoch": 1.02, + "learning_rate": 2.7552276227869784e-05, + "loss": 3.4978, + "step": 1371136 + }, + { + "epoch": 1.02, + "learning_rate": 2.7543890280359257e-05, + "loss": 3.5012, + "step": 1371648 + }, + { + "epoch": 1.02, + "learning_rate": 2.7535504332848737e-05, + "loss": 3.5004, + "step": 1372160 + }, + { + "epoch": 1.02, + "learning_rate": 2.7527134764141953e-05, + "loss": 3.5045, + "step": 1372672 + }, + { + "epoch": 1.02, + "learning_rate": 2.7518748816631433e-05, + "loss": 3.4931, + "step": 1373184 + }, + { + "epoch": 1.02, + "learning_rate": 2.751036286912091e-05, + "loss": 3.4965, + "step": 1373696 + }, + { + "epoch": 1.03, + "eval_loss": 3.8526077270507812, + "eval_runtime": 303.6303, + "eval_samples_per_second": 1256.762, + "eval_steps_per_second": 39.275, + "step": 1373760 + }, + { + "epoch": 0.0, + "learning_rate": 2.750197692161039e-05, + "loss": 3.487, + "step": 1374208 + }, + { + "epoch": 0.0, + "learning_rate": 2.749359097409987e-05, + "loss": 3.4924, + "step": 1374720 + }, + { + "epoch": 0.0, + "learning_rate": 2.748522140539308e-05, + "loss": 3.5139, + "step": 1375232 + }, + { + "epoch": 0.0, + "learning_rate": 2.7476835457882562e-05, + "loss": 3.5097, + "step": 1375744 + }, + { + "epoch": 0.0, + "learning_rate": 2.7468449510372042e-05, + "loss": 3.5016, + "step": 1376256 + }, + { + "epoch": 0.0, + "learning_rate": 2.7460063562861522e-05, + "loss": 3.496, + "step": 1376768 + }, + { + "epoch": 0.0, + "learning_rate": 2.7451677615351002e-05, + "loss": 3.4953, + "step": 1377280 + }, + { + "epoch": 0.0, + "learning_rate": 2.7443291667840482e-05, + "loss": 3.4891, + "step": 1377792 + }, + { + "epoch": 0.0, + "learning_rate": 2.743492209913369e-05, + "loss": 3.5057, + "step": 1378304 + }, + { + "epoch": 0.0, + "learning_rate": 2.742653615162317e-05, + "loss": 3.5017, + "step": 1378816 + }, + { + "epoch": 0.0, + "learning_rate": 2.7418150204112654e-05, + "loss": 3.4984, + "step": 1379328 + }, + { + "epoch": 0.0, + "learning_rate": 2.7409764256602134e-05, + "loss": 3.5006, + "step": 1379840 + }, + { + "epoch": 0.0, + "learning_rate": 2.7401394687895343e-05, + "loss": 3.4926, + "step": 1380352 + }, + { + "epoch": 0.0, + "learning_rate": 2.7393008740384823e-05, + "loss": 3.4879, + "step": 1380864 + }, + { + "epoch": 0.0, + "learning_rate": 2.7384622792874303e-05, + "loss": 3.489, + "step": 1381376 + }, + { + "epoch": 0.0, + "learning_rate": 2.7376236845363783e-05, + "loss": 3.4905, + "step": 1381888 + }, + { + "epoch": 0.0, + "learning_rate": 2.7367850897853263e-05, + "loss": 3.4927, + "step": 1382400 + }, + { + "epoch": 0.0, + "learning_rate": 2.7359464950342743e-05, + "loss": 3.494, + "step": 1382912 + }, + { + "epoch": 0.0, + "learning_rate": 2.7351079002832226e-05, + "loss": 3.495, + "step": 1383424 + }, + { + "epoch": 0.0, + "learning_rate": 2.7342693055321706e-05, + "loss": 3.5222, + "step": 1383936 + }, + { + "epoch": 0.0, + "learning_rate": 2.7334323486614915e-05, + "loss": 3.4977, + "step": 1384448 + }, + { + "epoch": 0.0, + "learning_rate": 2.7325953917908125e-05, + "loss": 3.4972, + "step": 1384960 + }, + { + "epoch": 0.0, + "learning_rate": 2.7317567970397605e-05, + "loss": 3.5001, + "step": 1385472 + }, + { + "epoch": 0.0, + "learning_rate": 2.7309182022887088e-05, + "loss": 3.5057, + "step": 1385984 + }, + { + "epoch": 0.0, + "learning_rate": 2.7300796075376568e-05, + "loss": 3.4889, + "step": 1386496 + }, + { + "epoch": 0.0, + "learning_rate": 2.7292410127866048e-05, + "loss": 3.4997, + "step": 1387008 + }, + { + "epoch": 0.0, + "learning_rate": 2.7284040559159257e-05, + "loss": 3.4954, + "step": 1387520 + }, + { + "epoch": 0.0, + "learning_rate": 2.7275654611648737e-05, + "loss": 3.4937, + "step": 1388032 + }, + { + "epoch": 0.0, + "learning_rate": 2.7267268664138217e-05, + "loss": 3.4871, + "step": 1388544 + }, + { + "epoch": 0.01, + "learning_rate": 2.7258882716627697e-05, + "loss": 3.4906, + "step": 1389056 + }, + { + "epoch": 0.01, + "learning_rate": 2.725049676911718e-05, + "loss": 3.4956, + "step": 1389568 + }, + { + "epoch": 0.01, + "learning_rate": 2.724211082160666e-05, + "loss": 3.5053, + "step": 1390080 + }, + { + "epoch": 0.01, + "learning_rate": 2.723372487409614e-05, + "loss": 3.4965, + "step": 1390592 + }, + { + "epoch": 0.01, + "learning_rate": 2.722533892658562e-05, + "loss": 3.5047, + "step": 1391104 + }, + { + "epoch": 0.01, + "learning_rate": 2.721696935787883e-05, + "loss": 3.4847, + "step": 1391616 + }, + { + "epoch": 0.01, + "learning_rate": 2.720858341036831e-05, + "loss": 3.5073, + "step": 1392128 + }, + { + "epoch": 0.01, + "learning_rate": 2.720019746285779e-05, + "loss": 3.4838, + "step": 1392640 + }, + { + "epoch": 0.01, + "learning_rate": 2.7191811515347266e-05, + "loss": 3.4984, + "step": 1393152 + }, + { + "epoch": 0.01, + "learning_rate": 2.718345832544421e-05, + "loss": 3.4796, + "step": 1393664 + }, + { + "epoch": 0.01, + "learning_rate": 2.717507237793369e-05, + "loss": 3.4949, + "step": 1394176 + }, + { + "epoch": 0.01, + "learning_rate": 2.716668643042317e-05, + "loss": 3.4868, + "step": 1394688 + }, + { + "epoch": 0.01, + "learning_rate": 2.715830048291265e-05, + "loss": 3.5042, + "step": 1395200 + }, + { + "epoch": 0.01, + "learning_rate": 2.7149914535402134e-05, + "loss": 3.4913, + "step": 1395712 + }, + { + "epoch": 0.01, + "learning_rate": 2.7141528587891614e-05, + "loss": 3.4942, + "step": 1396224 + }, + { + "epoch": 0.01, + "learning_rate": 2.7133142640381094e-05, + "loss": 3.4985, + "step": 1396736 + }, + { + "epoch": 0.01, + "learning_rate": 2.7124756692870567e-05, + "loss": 3.5024, + "step": 1397248 + }, + { + "epoch": 0.01, + "learning_rate": 2.7116387124163783e-05, + "loss": 3.4981, + "step": 1397760 + }, + { + "epoch": 0.01, + "learning_rate": 2.7108001176653263e-05, + "loss": 3.481, + "step": 1398272 + }, + { + "epoch": 0.01, + "learning_rate": 2.709961522914274e-05, + "loss": 3.4834, + "step": 1398784 + }, + { + "epoch": 0.01, + "learning_rate": 2.709122928163222e-05, + "loss": 3.5017, + "step": 1399296 + }, + { + "epoch": 0.01, + "learning_rate": 2.7082859712925435e-05, + "loss": 3.5001, + "step": 1399808 + }, + { + "epoch": 0.01, + "learning_rate": 2.7074473765414915e-05, + "loss": 3.4919, + "step": 1400320 + }, + { + "epoch": 0.01, + "learning_rate": 2.7066087817904388e-05, + "loss": 3.4875, + "step": 1400832 + }, + { + "epoch": 0.01, + "learning_rate": 2.705770187039387e-05, + "loss": 3.4912, + "step": 1401344 + }, + { + "epoch": 0.01, + "learning_rate": 2.7049332301687087e-05, + "loss": 3.48, + "step": 1401856 + }, + { + "epoch": 0.01, + "learning_rate": 2.7040962732980297e-05, + "loss": 3.4932, + "step": 1402368 + }, + { + "epoch": 0.01, + "learning_rate": 2.7032576785469776e-05, + "loss": 3.4799, + "step": 1402880 + }, + { + "epoch": 0.01, + "learning_rate": 2.7024190837959256e-05, + "loss": 3.4934, + "step": 1403392 + }, + { + "epoch": 0.01, + "learning_rate": 2.7015804890448736e-05, + "loss": 3.5001, + "step": 1403904 + }, + { + "epoch": 0.01, + "learning_rate": 2.7007418942938213e-05, + "loss": 3.4814, + "step": 1404416 + }, + { + "epoch": 0.01, + "learning_rate": 2.6999032995427693e-05, + "loss": 3.4864, + "step": 1404928 + }, + { + "epoch": 0.01, + "learning_rate": 2.6990647047917173e-05, + "loss": 3.4861, + "step": 1405440 + }, + { + "epoch": 0.01, + "learning_rate": 2.6982261100406653e-05, + "loss": 3.4806, + "step": 1405952 + }, + { + "epoch": 0.01, + "learning_rate": 2.6973907910503598e-05, + "loss": 3.4755, + "step": 1406464 + }, + { + "epoch": 0.01, + "learning_rate": 2.6965521962993078e-05, + "loss": 3.5006, + "step": 1406976 + }, + { + "epoch": 0.01, + "learning_rate": 2.6957136015482558e-05, + "loss": 3.4883, + "step": 1407488 + }, + { + "epoch": 0.01, + "learning_rate": 2.694875006797204e-05, + "loss": 3.4812, + "step": 1408000 + }, + { + "epoch": 0.01, + "learning_rate": 2.6940364120461514e-05, + "loss": 3.4837, + "step": 1408512 + }, + { + "epoch": 0.01, + "learning_rate": 2.6931978172950994e-05, + "loss": 3.488, + "step": 1409024 + }, + { + "epoch": 0.01, + "learning_rate": 2.6923592225440474e-05, + "loss": 3.4923, + "step": 1409536 + }, + { + "epoch": 0.01, + "learning_rate": 2.6915206277929954e-05, + "loss": 3.4915, + "step": 1410048 + }, + { + "epoch": 0.01, + "learning_rate": 2.6906836709223167e-05, + "loss": 3.4839, + "step": 1410560 + }, + { + "epoch": 0.01, + "learning_rate": 2.6898450761712647e-05, + "loss": 3.4988, + "step": 1411072 + }, + { + "epoch": 0.01, + "learning_rate": 2.6890064814202127e-05, + "loss": 3.4947, + "step": 1411584 + }, + { + "epoch": 0.01, + "learning_rate": 2.6881678866691606e-05, + "loss": 3.4847, + "step": 1412096 + }, + { + "epoch": 0.01, + "learning_rate": 2.6873309297984816e-05, + "loss": 3.4949, + "step": 1412608 + }, + { + "epoch": 0.01, + "learning_rate": 2.6864923350474296e-05, + "loss": 3.4946, + "step": 1413120 + }, + { + "epoch": 0.01, + "learning_rate": 2.685653740296378e-05, + "loss": 3.4914, + "step": 1413632 + }, + { + "epoch": 0.01, + "learning_rate": 2.684815145545326e-05, + "loss": 3.4859, + "step": 1414144 + }, + { + "epoch": 0.01, + "learning_rate": 2.6839781886746468e-05, + "loss": 3.49, + "step": 1414656 + }, + { + "epoch": 0.01, + "learning_rate": 2.6831395939235948e-05, + "loss": 3.4955, + "step": 1415168 + }, + { + "epoch": 0.01, + "learning_rate": 2.6823009991725428e-05, + "loss": 3.4962, + "step": 1415680 + }, + { + "epoch": 0.01, + "learning_rate": 2.6814624044214908e-05, + "loss": 3.4855, + "step": 1416192 + }, + { + "epoch": 0.01, + "learning_rate": 2.680625447550812e-05, + "loss": 3.4728, + "step": 1416704 + }, + { + "epoch": 0.01, + "learning_rate": 2.67978685279976e-05, + "loss": 3.4799, + "step": 1417216 + }, + { + "epoch": 0.01, + "learning_rate": 2.678948258048708e-05, + "loss": 3.4899, + "step": 1417728 + }, + { + "epoch": 0.01, + "learning_rate": 2.678109663297656e-05, + "loss": 3.4934, + "step": 1418240 + }, + { + "epoch": 0.01, + "learning_rate": 2.677271068546604e-05, + "loss": 3.4839, + "step": 1418752 + }, + { + "epoch": 0.01, + "learning_rate": 2.676432473795552e-05, + "loss": 3.4841, + "step": 1419264 + }, + { + "epoch": 0.02, + "learning_rate": 2.6755955169248733e-05, + "loss": 3.4945, + "step": 1419776 + }, + { + "epoch": 0.02, + "learning_rate": 2.6747569221738212e-05, + "loss": 3.4831, + "step": 1420288 + }, + { + "epoch": 0.02, + "learning_rate": 2.6739183274227692e-05, + "loss": 3.4808, + "step": 1420800 + }, + { + "epoch": 0.02, + "learning_rate": 2.6730797326717172e-05, + "loss": 3.4919, + "step": 1421312 + }, + { + "epoch": 0.02, + "learning_rate": 2.672242775801038e-05, + "loss": 3.4836, + "step": 1421824 + }, + { + "epoch": 0.02, + "learning_rate": 2.671404181049986e-05, + "loss": 3.4978, + "step": 1422336 + }, + { + "epoch": 0.02, + "learning_rate": 2.670565586298934e-05, + "loss": 3.4887, + "step": 1422848 + }, + { + "epoch": 0.02, + "learning_rate": 2.669726991547882e-05, + "loss": 3.4742, + "step": 1423360 + }, + { + "epoch": 0.02, + "learning_rate": 2.6688883967968305e-05, + "loss": 3.4922, + "step": 1423872 + }, + { + "epoch": 0.02, + "learning_rate": 2.6680498020457785e-05, + "loss": 3.4853, + "step": 1424384 + }, + { + "epoch": 0.02, + "learning_rate": 2.6672112072947265e-05, + "loss": 3.5, + "step": 1424896 + }, + { + "epoch": 0.02, + "learning_rate": 2.6663726125436745e-05, + "loss": 3.4917, + "step": 1425408 + }, + { + "epoch": 0.02, + "learning_rate": 2.6655356556729954e-05, + "loss": 3.4837, + "step": 1425920 + }, + { + "epoch": 0.02, + "learning_rate": 2.6646970609219434e-05, + "loss": 3.4751, + "step": 1426432 + }, + { + "epoch": 0.02, + "learning_rate": 2.6638584661708914e-05, + "loss": 3.4941, + "step": 1426944 + }, + { + "epoch": 0.02, + "learning_rate": 2.6630215093002126e-05, + "loss": 3.4766, + "step": 1427456 + }, + { + "epoch": 0.02, + "learning_rate": 2.6621829145491606e-05, + "loss": 3.4849, + "step": 1427968 + }, + { + "epoch": 0.02, + "learning_rate": 2.6613443197981086e-05, + "loss": 3.487, + "step": 1428480 + }, + { + "epoch": 0.02, + "learning_rate": 2.6605057250470566e-05, + "loss": 3.4867, + "step": 1428992 + }, + { + "epoch": 0.02, + "learning_rate": 2.6596671302960046e-05, + "loss": 3.4719, + "step": 1429504 + }, + { + "epoch": 0.02, + "learning_rate": 2.6588285355449522e-05, + "loss": 3.4957, + "step": 1430016 + }, + { + "epoch": 0.02, + "learning_rate": 2.6579899407939002e-05, + "loss": 3.4799, + "step": 1430528 + }, + { + "epoch": 0.02, + "learning_rate": 2.6571529839232218e-05, + "loss": 3.4865, + "step": 1431040 + }, + { + "epoch": 0.02, + "learning_rate": 2.6563143891721698e-05, + "loss": 3.4908, + "step": 1431552 + }, + { + "epoch": 0.02, + "learning_rate": 2.655475794421117e-05, + "loss": 3.4847, + "step": 1432064 + }, + { + "epoch": 0.02, + "learning_rate": 2.654637199670065e-05, + "loss": 3.5028, + "step": 1432576 + }, + { + "epoch": 0.02, + "learning_rate": 2.6537986049190135e-05, + "loss": 3.4935, + "step": 1433088 + }, + { + "epoch": 0.02, + "learning_rate": 2.6529616480483344e-05, + "loss": 3.4906, + "step": 1433600 + }, + { + "epoch": 0.02, + "learning_rate": 2.6521230532972824e-05, + "loss": 3.4806, + "step": 1434112 + }, + { + "epoch": 0.02, + "learning_rate": 2.6512844585462304e-05, + "loss": 3.4884, + "step": 1434624 + }, + { + "epoch": 0.02, + "learning_rate": 2.6504458637951784e-05, + "loss": 3.4891, + "step": 1435136 + }, + { + "epoch": 0.02, + "learning_rate": 2.6496072690441264e-05, + "loss": 3.4799, + "step": 1435648 + }, + { + "epoch": 0.02, + "learning_rate": 2.6487686742930744e-05, + "loss": 3.4921, + "step": 1436160 + }, + { + "epoch": 0.02, + "learning_rate": 2.6479300795420227e-05, + "loss": 3.4912, + "step": 1436672 + }, + { + "epoch": 0.02, + "learning_rate": 2.6470914847909707e-05, + "loss": 3.4859, + "step": 1437184 + }, + { + "epoch": 0.02, + "learning_rate": 2.6462545279202916e-05, + "loss": 3.4884, + "step": 1437696 + }, + { + "epoch": 0.02, + "learning_rate": 2.6454159331692396e-05, + "loss": 3.4867, + "step": 1438208 + }, + { + "epoch": 0.02, + "learning_rate": 2.6445773384181876e-05, + "loss": 3.4824, + "step": 1438720 + }, + { + "epoch": 0.02, + "learning_rate": 2.6437387436671356e-05, + "loss": 3.4886, + "step": 1439232 + }, + { + "epoch": 0.02, + "learning_rate": 2.642901786796457e-05, + "loss": 3.48, + "step": 1439744 + }, + { + "epoch": 0.02, + "learning_rate": 2.6420631920454048e-05, + "loss": 3.5009, + "step": 1440256 + }, + { + "epoch": 0.02, + "learning_rate": 2.6412245972943528e-05, + "loss": 3.4992, + "step": 1440768 + }, + { + "epoch": 0.02, + "learning_rate": 2.6403860025433008e-05, + "loss": 3.4844, + "step": 1441280 + }, + { + "epoch": 0.02, + "learning_rate": 2.6395490456726217e-05, + "loss": 3.4857, + "step": 1441792 + }, + { + "epoch": 0.02, + "learning_rate": 2.6387104509215697e-05, + "loss": 3.484, + "step": 1442304 + }, + { + "epoch": 0.02, + "learning_rate": 2.637871856170518e-05, + "loss": 3.491, + "step": 1442816 + }, + { + "epoch": 0.02, + "learning_rate": 2.637034899299839e-05, + "loss": 3.4882, + "step": 1443328 + }, + { + "epoch": 0.02, + "learning_rate": 2.63619794242916e-05, + "loss": 3.4865, + "step": 1443840 + }, + { + "epoch": 0.02, + "learning_rate": 2.635359347678108e-05, + "loss": 3.4966, + "step": 1444352 + }, + { + "epoch": 0.02, + "learning_rate": 2.634520752927056e-05, + "loss": 3.4883, + "step": 1444864 + }, + { + "epoch": 0.02, + "learning_rate": 2.6336821581760042e-05, + "loss": 3.5002, + "step": 1445376 + }, + { + "epoch": 0.02, + "learning_rate": 2.6328435634249522e-05, + "loss": 3.4835, + "step": 1445888 + }, + { + "epoch": 0.02, + "learning_rate": 2.6320049686739002e-05, + "loss": 3.4963, + "step": 1446400 + }, + { + "epoch": 0.02, + "learning_rate": 2.631168011803221e-05, + "loss": 3.4919, + "step": 1446912 + }, + { + "epoch": 0.02, + "learning_rate": 2.630329417052169e-05, + "loss": 3.4817, + "step": 1447424 + }, + { + "epoch": 0.02, + "learning_rate": 2.629490822301117e-05, + "loss": 3.4909, + "step": 1447936 + }, + { + "epoch": 0.02, + "learning_rate": 2.628652227550065e-05, + "loss": 3.492, + "step": 1448448 + }, + { + "epoch": 0.02, + "learning_rate": 2.6278136327990134e-05, + "loss": 3.4902, + "step": 1448960 + }, + { + "epoch": 0.02, + "learning_rate": 2.6269750380479614e-05, + "loss": 3.4779, + "step": 1449472 + }, + { + "epoch": 0.02, + "learning_rate": 2.6261364432969094e-05, + "loss": 3.4795, + "step": 1449984 + }, + { + "epoch": 0.03, + "eval_loss": 3.853731632232666, + "eval_runtime": 303.0718, + "eval_samples_per_second": 1259.078, + "eval_steps_per_second": 39.347, + "step": 1450080 + }, + { + "epoch": 1.0, + "learning_rate": 2.6252994864262303e-05, + "loss": 3.4838, + "step": 1450496 + }, + { + "epoch": 1.0, + "learning_rate": 2.6244608916751783e-05, + "loss": 3.4807, + "step": 1451008 + }, + { + "epoch": 1.0, + "learning_rate": 2.6236222969241263e-05, + "loss": 3.4982, + "step": 1451520 + }, + { + "epoch": 1.0, + "learning_rate": 2.6227837021730743e-05, + "loss": 3.4997, + "step": 1452032 + }, + { + "epoch": 1.0, + "learning_rate": 2.6219467453023956e-05, + "loss": 3.4962, + "step": 1452544 + }, + { + "epoch": 1.0, + "learning_rate": 2.6211081505513436e-05, + "loss": 3.4777, + "step": 1453056 + }, + { + "epoch": 1.0, + "learning_rate": 2.6202695558002916e-05, + "loss": 3.4893, + "step": 1453568 + }, + { + "epoch": 1.0, + "learning_rate": 2.6194309610492395e-05, + "loss": 3.4742, + "step": 1454080 + }, + { + "epoch": 1.0, + "learning_rate": 2.6185940041785605e-05, + "loss": 3.4957, + "step": 1454592 + }, + { + "epoch": 1.0, + "learning_rate": 2.6177554094275088e-05, + "loss": 3.4896, + "step": 1455104 + }, + { + "epoch": 1.0, + "learning_rate": 2.6169168146764568e-05, + "loss": 3.4813, + "step": 1455616 + }, + { + "epoch": 1.0, + "learning_rate": 2.6160782199254048e-05, + "loss": 3.4931, + "step": 1456128 + }, + { + "epoch": 1.0, + "learning_rate": 2.6152412630547257e-05, + "loss": 3.4783, + "step": 1456640 + }, + { + "epoch": 1.0, + "learning_rate": 2.6144026683036737e-05, + "loss": 3.479, + "step": 1457152 + }, + { + "epoch": 1.0, + "learning_rate": 2.6135640735526217e-05, + "loss": 3.4739, + "step": 1457664 + }, + { + "epoch": 1.0, + "learning_rate": 2.6127254788015697e-05, + "loss": 3.4833, + "step": 1458176 + }, + { + "epoch": 1.0, + "learning_rate": 2.6118868840505177e-05, + "loss": 3.4794, + "step": 1458688 + }, + { + "epoch": 1.0, + "learning_rate": 2.6110482892994653e-05, + "loss": 3.4806, + "step": 1459200 + }, + { + "epoch": 1.0, + "learning_rate": 2.6102096945484133e-05, + "loss": 3.4841, + "step": 1459712 + }, + { + "epoch": 1.0, + "learning_rate": 2.6093710997973613e-05, + "loss": 3.5047, + "step": 1460224 + }, + { + "epoch": 1.0, + "learning_rate": 2.608534142926683e-05, + "loss": 3.4856, + "step": 1460736 + }, + { + "epoch": 1.0, + "learning_rate": 2.607697186056004e-05, + "loss": 3.4881, + "step": 1461248 + }, + { + "epoch": 1.0, + "learning_rate": 2.606858591304952e-05, + "loss": 3.4875, + "step": 1461760 + }, + { + "epoch": 1.0, + "learning_rate": 2.6060199965539e-05, + "loss": 3.491, + "step": 1462272 + }, + { + "epoch": 1.0, + "learning_rate": 2.6051814018028475e-05, + "loss": 3.4772, + "step": 1462784 + }, + { + "epoch": 1.0, + "learning_rate": 2.6043428070517955e-05, + "loss": 3.4845, + "step": 1463296 + }, + { + "epoch": 1.0, + "learning_rate": 2.603505850181117e-05, + "loss": 3.4801, + "step": 1463808 + }, + { + "epoch": 1.0, + "learning_rate": 2.602667255430065e-05, + "loss": 3.4785, + "step": 1464320 + }, + { + "epoch": 1.0, + "learning_rate": 2.6018286606790127e-05, + "loss": 3.4742, + "step": 1464832 + }, + { + "epoch": 1.01, + "learning_rate": 2.6009900659279607e-05, + "loss": 3.4819, + "step": 1465344 + }, + { + "epoch": 1.01, + "learning_rate": 2.6001514711769087e-05, + "loss": 3.4815, + "step": 1465856 + }, + { + "epoch": 1.01, + "learning_rate": 2.5993128764258567e-05, + "loss": 3.4937, + "step": 1466368 + }, + { + "epoch": 1.01, + "learning_rate": 2.5984742816748047e-05, + "loss": 3.4861, + "step": 1466880 + }, + { + "epoch": 1.01, + "learning_rate": 2.597637324804126e-05, + "loss": 3.489, + "step": 1467392 + }, + { + "epoch": 1.01, + "learning_rate": 2.596798730053074e-05, + "loss": 3.4773, + "step": 1467904 + }, + { + "epoch": 1.01, + "learning_rate": 2.595960135302022e-05, + "loss": 3.4897, + "step": 1468416 + }, + { + "epoch": 1.01, + "learning_rate": 2.59512154055097e-05, + "loss": 3.4759, + "step": 1468928 + }, + { + "epoch": 1.01, + "learning_rate": 2.594284583680291e-05, + "loss": 3.485, + "step": 1469440 + }, + { + "epoch": 1.01, + "learning_rate": 2.5934476268096124e-05, + "loss": 3.4662, + "step": 1469952 + }, + { + "epoch": 1.01, + "learning_rate": 2.59260903205856e-05, + "loss": 3.4794, + "step": 1470464 + }, + { + "epoch": 1.01, + "learning_rate": 2.591770437307508e-05, + "loss": 3.4771, + "step": 1470976 + }, + { + "epoch": 1.01, + "learning_rate": 2.590931842556456e-05, + "loss": 3.4931, + "step": 1471488 + }, + { + "epoch": 1.01, + "learning_rate": 2.590093247805404e-05, + "loss": 3.4809, + "step": 1472000 + }, + { + "epoch": 1.01, + "learning_rate": 2.589254653054352e-05, + "loss": 3.4817, + "step": 1472512 + }, + { + "epoch": 1.01, + "learning_rate": 2.5884160583033e-05, + "loss": 3.4854, + "step": 1473024 + }, + { + "epoch": 1.01, + "learning_rate": 2.587577463552248e-05, + "loss": 3.4911, + "step": 1473536 + }, + { + "epoch": 1.01, + "learning_rate": 2.5867405066815693e-05, + "loss": 3.4821, + "step": 1474048 + }, + { + "epoch": 1.01, + "learning_rate": 2.5859019119305173e-05, + "loss": 3.4717, + "step": 1474560 + }, + { + "epoch": 1.01, + "learning_rate": 2.5850633171794653e-05, + "loss": 3.4688, + "step": 1475072 + }, + { + "epoch": 1.01, + "learning_rate": 2.5842247224284133e-05, + "loss": 3.4884, + "step": 1475584 + }, + { + "epoch": 1.01, + "learning_rate": 2.5833877655577342e-05, + "loss": 3.4899, + "step": 1476096 + }, + { + "epoch": 1.01, + "learning_rate": 2.5825491708066822e-05, + "loss": 3.4803, + "step": 1476608 + }, + { + "epoch": 1.01, + "learning_rate": 2.5817105760556305e-05, + "loss": 3.4749, + "step": 1477120 + }, + { + "epoch": 1.01, + "learning_rate": 2.5808719813045785e-05, + "loss": 3.4824, + "step": 1477632 + }, + { + "epoch": 1.01, + "learning_rate": 2.5800350244338994e-05, + "loss": 3.4661, + "step": 1478144 + }, + { + "epoch": 1.01, + "learning_rate": 2.5791980675632203e-05, + "loss": 3.483, + "step": 1478656 + }, + { + "epoch": 1.01, + "learning_rate": 2.5783594728121683e-05, + "loss": 3.4683, + "step": 1479168 + }, + { + "epoch": 1.01, + "learning_rate": 2.5775208780611167e-05, + "loss": 3.4806, + "step": 1479680 + }, + { + "epoch": 1.01, + "learning_rate": 2.5766822833100647e-05, + "loss": 3.4856, + "step": 1480192 + }, + { + "epoch": 1.01, + "learning_rate": 2.5758436885590127e-05, + "loss": 3.4746, + "step": 1480704 + }, + { + "epoch": 1.01, + "learning_rate": 2.5750050938079607e-05, + "loss": 3.4689, + "step": 1481216 + }, + { + "epoch": 1.01, + "learning_rate": 2.5741664990569087e-05, + "loss": 3.4752, + "step": 1481728 + }, + { + "epoch": 1.01, + "learning_rate": 2.5733279043058566e-05, + "loss": 3.4687, + "step": 1482240 + }, + { + "epoch": 1.01, + "learning_rate": 2.5724925853155508e-05, + "loss": 3.4695, + "step": 1482752 + }, + { + "epoch": 1.01, + "learning_rate": 2.5716539905644988e-05, + "loss": 3.4852, + "step": 1483264 + }, + { + "epoch": 1.01, + "learning_rate": 2.5708153958134468e-05, + "loss": 3.4766, + "step": 1483776 + }, + { + "epoch": 1.01, + "learning_rate": 2.5699768010623948e-05, + "loss": 3.4694, + "step": 1484288 + }, + { + "epoch": 1.01, + "learning_rate": 2.5691382063113428e-05, + "loss": 3.467, + "step": 1484800 + }, + { + "epoch": 1.01, + "learning_rate": 2.5682996115602908e-05, + "loss": 3.4756, + "step": 1485312 + }, + { + "epoch": 1.01, + "learning_rate": 2.5674610168092388e-05, + "loss": 3.476, + "step": 1485824 + }, + { + "epoch": 1.01, + "learning_rate": 2.5666224220581868e-05, + "loss": 3.4829, + "step": 1486336 + }, + { + "epoch": 1.01, + "learning_rate": 2.565785465187508e-05, + "loss": 3.4743, + "step": 1486848 + }, + { + "epoch": 1.01, + "learning_rate": 2.564946870436456e-05, + "loss": 3.4857, + "step": 1487360 + }, + { + "epoch": 1.01, + "learning_rate": 2.564109913565777e-05, + "loss": 3.484, + "step": 1487872 + }, + { + "epoch": 1.01, + "learning_rate": 2.563271318814725e-05, + "loss": 3.4727, + "step": 1488384 + }, + { + "epoch": 1.01, + "learning_rate": 2.562432724063673e-05, + "loss": 3.4782, + "step": 1488896 + }, + { + "epoch": 1.01, + "learning_rate": 2.5615941293126213e-05, + "loss": 3.4821, + "step": 1489408 + }, + { + "epoch": 1.01, + "learning_rate": 2.5607555345615693e-05, + "loss": 3.4822, + "step": 1489920 + }, + { + "epoch": 1.01, + "learning_rate": 2.5599169398105172e-05, + "loss": 3.4745, + "step": 1490432 + }, + { + "epoch": 1.01, + "learning_rate": 2.5590783450594652e-05, + "loss": 3.4772, + "step": 1490944 + }, + { + "epoch": 1.01, + "learning_rate": 2.5582397503084132e-05, + "loss": 3.4833, + "step": 1491456 + }, + { + "epoch": 1.01, + "learning_rate": 2.557402793437734e-05, + "loss": 3.4853, + "step": 1491968 + }, + { + "epoch": 1.01, + "learning_rate": 2.5565658365670554e-05, + "loss": 3.4742, + "step": 1492480 + }, + { + "epoch": 1.01, + "learning_rate": 2.5557272418160034e-05, + "loss": 3.4632, + "step": 1492992 + }, + { + "epoch": 1.01, + "learning_rate": 2.5548886470649514e-05, + "loss": 3.4624, + "step": 1493504 + }, + { + "epoch": 1.01, + "learning_rate": 2.5540500523138994e-05, + "loss": 3.4749, + "step": 1494016 + }, + { + "epoch": 1.01, + "learning_rate": 2.5532114575628474e-05, + "loss": 3.4819, + "step": 1494528 + }, + { + "epoch": 1.01, + "learning_rate": 2.5523728628117954e-05, + "loss": 3.4754, + "step": 1495040 + }, + { + "epoch": 1.01, + "learning_rate": 2.5515342680607434e-05, + "loss": 3.4679, + "step": 1495552 + }, + { + "epoch": 1.02, + "learning_rate": 2.5506973111900646e-05, + "loss": 3.4813, + "step": 1496064 + }, + { + "epoch": 1.02, + "learning_rate": 2.5498587164390126e-05, + "loss": 3.4758, + "step": 1496576 + }, + { + "epoch": 1.02, + "learning_rate": 2.5490201216879606e-05, + "loss": 3.4682, + "step": 1497088 + }, + { + "epoch": 1.02, + "learning_rate": 2.5481831648172815e-05, + "loss": 3.475, + "step": 1497600 + }, + { + "epoch": 1.02, + "learning_rate": 2.5473445700662295e-05, + "loss": 3.4774, + "step": 1498112 + }, + { + "epoch": 1.02, + "learning_rate": 2.5465059753151775e-05, + "loss": 3.4869, + "step": 1498624 + }, + { + "epoch": 1.02, + "learning_rate": 2.545667380564126e-05, + "loss": 3.4732, + "step": 1499136 + }, + { + "epoch": 1.02, + "learning_rate": 2.544828785813073e-05, + "loss": 3.4599, + "step": 1499648 + }, + { + "epoch": 1.02, + "learning_rate": 2.543990191062021e-05, + "loss": 3.4826, + "step": 1500160 + }, + { + "epoch": 1.02, + "learning_rate": 2.543151596310969e-05, + "loss": 3.4728, + "step": 1500672 + }, + { + "epoch": 1.02, + "learning_rate": 2.5423146394402907e-05, + "loss": 3.4855, + "step": 1501184 + }, + { + "epoch": 1.02, + "learning_rate": 2.5414760446892384e-05, + "loss": 3.4826, + "step": 1501696 + }, + { + "epoch": 1.02, + "learning_rate": 2.5406374499381864e-05, + "loss": 3.4718, + "step": 1502208 + }, + { + "epoch": 1.02, + "learning_rate": 2.5397988551871344e-05, + "loss": 3.465, + "step": 1502720 + }, + { + "epoch": 1.02, + "learning_rate": 2.5389602604360824e-05, + "loss": 3.4825, + "step": 1503232 + }, + { + "epoch": 1.02, + "learning_rate": 2.5381233035654033e-05, + "loss": 3.4645, + "step": 1503744 + }, + { + "epoch": 1.02, + "learning_rate": 2.5372847088143513e-05, + "loss": 3.4746, + "step": 1504256 + }, + { + "epoch": 1.02, + "learning_rate": 2.5364461140632996e-05, + "loss": 3.4698, + "step": 1504768 + }, + { + "epoch": 1.02, + "learning_rate": 2.5356075193122476e-05, + "loss": 3.4784, + "step": 1505280 + }, + { + "epoch": 1.02, + "learning_rate": 2.5347689245611956e-05, + "loss": 3.4604, + "step": 1505792 + }, + { + "epoch": 1.02, + "learning_rate": 2.5339303298101436e-05, + "loss": 3.4803, + "step": 1506304 + }, + { + "epoch": 1.02, + "learning_rate": 2.5330917350590916e-05, + "loss": 3.4679, + "step": 1506816 + }, + { + "epoch": 1.02, + "learning_rate": 2.5322547781884125e-05, + "loss": 3.4721, + "step": 1507328 + }, + { + "epoch": 1.02, + "learning_rate": 2.5314161834373605e-05, + "loss": 3.481, + "step": 1507840 + }, + { + "epoch": 1.02, + "learning_rate": 2.5305775886863085e-05, + "loss": 3.4718, + "step": 1508352 + }, + { + "epoch": 1.02, + "learning_rate": 2.529738993935257e-05, + "loss": 3.4899, + "step": 1508864 + }, + { + "epoch": 1.02, + "learning_rate": 2.528900399184205e-05, + "loss": 3.4835, + "step": 1509376 + }, + { + "epoch": 1.02, + "learning_rate": 2.5280634423135257e-05, + "loss": 3.4773, + "step": 1509888 + }, + { + "epoch": 1.02, + "learning_rate": 2.5272248475624737e-05, + "loss": 3.4684, + "step": 1510400 + }, + { + "epoch": 1.02, + "learning_rate": 2.5263862528114217e-05, + "loss": 3.4787, + "step": 1510912 + }, + { + "epoch": 1.02, + "learning_rate": 2.5255476580603697e-05, + "loss": 3.4727, + "step": 1511424 + }, + { + "epoch": 1.02, + "learning_rate": 2.5247090633093177e-05, + "loss": 3.4681, + "step": 1511936 + }, + { + "epoch": 1.02, + "learning_rate": 2.523872106438639e-05, + "loss": 3.4787, + "step": 1512448 + }, + { + "epoch": 1.02, + "learning_rate": 2.523033511687587e-05, + "loss": 3.4781, + "step": 1512960 + }, + { + "epoch": 1.02, + "learning_rate": 2.522194916936535e-05, + "loss": 3.4736, + "step": 1513472 + }, + { + "epoch": 1.02, + "learning_rate": 2.521356322185483e-05, + "loss": 3.4728, + "step": 1513984 + }, + { + "epoch": 1.02, + "learning_rate": 2.520517727434431e-05, + "loss": 3.4769, + "step": 1514496 + }, + { + "epoch": 1.02, + "learning_rate": 2.519679132683379e-05, + "loss": 3.4704, + "step": 1515008 + }, + { + "epoch": 1.02, + "learning_rate": 2.518840537932327e-05, + "loss": 3.4787, + "step": 1515520 + }, + { + "epoch": 1.02, + "learning_rate": 2.5180019431812746e-05, + "loss": 3.4699, + "step": 1516032 + }, + { + "epoch": 1.02, + "learning_rate": 2.517166624190969e-05, + "loss": 3.4838, + "step": 1516544 + }, + { + "epoch": 1.02, + "learning_rate": 2.516328029439917e-05, + "loss": 3.4807, + "step": 1517056 + }, + { + "epoch": 1.02, + "learning_rate": 2.515489434688865e-05, + "loss": 3.4791, + "step": 1517568 + }, + { + "epoch": 1.02, + "learning_rate": 2.514650839937813e-05, + "loss": 3.4728, + "step": 1518080 + }, + { + "epoch": 1.02, + "learning_rate": 2.5138122451867614e-05, + "loss": 3.4716, + "step": 1518592 + }, + { + "epoch": 1.02, + "learning_rate": 2.5129736504357094e-05, + "loss": 3.4782, + "step": 1519104 + }, + { + "epoch": 1.02, + "learning_rate": 2.5121350556846567e-05, + "loss": 3.4768, + "step": 1519616 + }, + { + "epoch": 1.02, + "learning_rate": 2.5112980988139783e-05, + "loss": 3.4685, + "step": 1520128 + }, + { + "epoch": 1.02, + "learning_rate": 2.5104595040629263e-05, + "loss": 3.483, + "step": 1520640 + }, + { + "epoch": 1.02, + "learning_rate": 2.5096209093118743e-05, + "loss": 3.4777, + "step": 1521152 + }, + { + "epoch": 1.02, + "learning_rate": 2.508782314560822e-05, + "loss": 3.4877, + "step": 1521664 + }, + { + "epoch": 1.02, + "learning_rate": 2.50794371980977e-05, + "loss": 3.4759, + "step": 1522176 + }, + { + "epoch": 1.02, + "learning_rate": 2.507105125058718e-05, + "loss": 3.4835, + "step": 1522688 + }, + { + "epoch": 1.02, + "learning_rate": 2.506266530307666e-05, + "loss": 3.4807, + "step": 1523200 + }, + { + "epoch": 1.02, + "learning_rate": 2.505429573436987e-05, + "loss": 3.47, + "step": 1523712 + }, + { + "epoch": 1.02, + "learning_rate": 2.5045909786859352e-05, + "loss": 3.475, + "step": 1524224 + }, + { + "epoch": 1.02, + "learning_rate": 2.5037523839348832e-05, + "loss": 3.4853, + "step": 1524736 + }, + { + "epoch": 1.02, + "learning_rate": 2.5029137891838312e-05, + "loss": 3.474, + "step": 1525248 + }, + { + "epoch": 1.02, + "learning_rate": 2.5020751944327792e-05, + "loss": 3.4648, + "step": 1525760 + }, + { + "epoch": 1.02, + "learning_rate": 2.5012365996817272e-05, + "loss": 3.4755, + "step": 1526272 + }, + { + "epoch": 1.03, + "eval_loss": 3.8546695709228516, + "eval_runtime": 303.2377, + "eval_samples_per_second": 1258.389, + "eval_steps_per_second": 39.326, + "step": 1526400 + }, + { + "epoch": 0.0, + "learning_rate": 2.5003980049306752e-05, + "loss": 3.4659, + "step": 1526784 + }, + { + "epoch": 0.0, + "learning_rate": 2.4995594101796232e-05, + "loss": 3.4678, + "step": 1527296 + }, + { + "epoch": 0.0, + "learning_rate": 2.4987224533089444e-05, + "loss": 3.4828, + "step": 1527808 + }, + { + "epoch": 0.0, + "learning_rate": 2.4978838585578924e-05, + "loss": 3.4883, + "step": 1528320 + }, + { + "epoch": 0.0, + "learning_rate": 2.4970452638068404e-05, + "loss": 3.4842, + "step": 1528832 + }, + { + "epoch": 0.0, + "learning_rate": 2.4962066690557884e-05, + "loss": 3.4724, + "step": 1529344 + }, + { + "epoch": 0.0, + "learning_rate": 2.4953680743047364e-05, + "loss": 3.4736, + "step": 1529856 + }, + { + "epoch": 0.0, + "learning_rate": 2.494529479553684e-05, + "loss": 3.4576, + "step": 1530368 + }, + { + "epoch": 0.0, + "learning_rate": 2.4936925226830053e-05, + "loss": 3.4845, + "step": 1530880 + }, + { + "epoch": 0.0, + "learning_rate": 2.4928539279319537e-05, + "loss": 3.4793, + "step": 1531392 + }, + { + "epoch": 0.0, + "learning_rate": 2.4920153331809016e-05, + "loss": 3.4715, + "step": 1531904 + }, + { + "epoch": 0.0, + "learning_rate": 2.4911767384298493e-05, + "loss": 3.4788, + "step": 1532416 + }, + { + "epoch": 0.0, + "learning_rate": 2.4903381436787973e-05, + "loss": 3.4669, + "step": 1532928 + }, + { + "epoch": 0.0, + "learning_rate": 2.4895011868081186e-05, + "loss": 3.4681, + "step": 1533440 + }, + { + "epoch": 0.0, + "learning_rate": 2.4886625920570662e-05, + "loss": 3.4627, + "step": 1533952 + }, + { + "epoch": 0.0, + "learning_rate": 2.4878239973060145e-05, + "loss": 3.4671, + "step": 1534464 + }, + { + "epoch": 0.0, + "learning_rate": 2.4869854025549625e-05, + "loss": 3.4684, + "step": 1534976 + }, + { + "epoch": 0.0, + "learning_rate": 2.4861468078039105e-05, + "loss": 3.467, + "step": 1535488 + }, + { + "epoch": 0.0, + "learning_rate": 2.4853082130528585e-05, + "loss": 3.4732, + "step": 1536000 + }, + { + "epoch": 0.0, + "learning_rate": 2.4844696183018065e-05, + "loss": 3.4903, + "step": 1536512 + }, + { + "epoch": 0.0, + "learning_rate": 2.4836326614311274e-05, + "loss": 3.4738, + "step": 1537024 + }, + { + "epoch": 0.0, + "learning_rate": 2.4827940666800754e-05, + "loss": 3.4788, + "step": 1537536 + }, + { + "epoch": 0.0, + "learning_rate": 2.4819571098093967e-05, + "loss": 3.4754, + "step": 1538048 + }, + { + "epoch": 0.0, + "learning_rate": 2.4811185150583447e-05, + "loss": 3.479, + "step": 1538560 + }, + { + "epoch": 0.0, + "learning_rate": 2.4802799203072927e-05, + "loss": 3.4701, + "step": 1539072 + }, + { + "epoch": 0.0, + "learning_rate": 2.4794413255562407e-05, + "loss": 3.4737, + "step": 1539584 + }, + { + "epoch": 0.0, + "learning_rate": 2.4786027308051887e-05, + "loss": 3.4731, + "step": 1540096 + }, + { + "epoch": 0.0, + "learning_rate": 2.47776577393451e-05, + "loss": 3.4649, + "step": 1540608 + }, + { + "epoch": 0.0, + "learning_rate": 2.476927179183458e-05, + "loss": 3.4619, + "step": 1541120 + }, + { + "epoch": 0.0, + "learning_rate": 2.476088584432406e-05, + "loss": 3.4716, + "step": 1541632 + }, + { + "epoch": 0.01, + "learning_rate": 2.475249989681354e-05, + "loss": 3.4704, + "step": 1542144 + }, + { + "epoch": 0.01, + "learning_rate": 2.474411394930302e-05, + "loss": 3.479, + "step": 1542656 + }, + { + "epoch": 0.01, + "learning_rate": 2.4735728001792495e-05, + "loss": 3.4742, + "step": 1543168 + }, + { + "epoch": 0.01, + "learning_rate": 2.4727342054281975e-05, + "loss": 3.4789, + "step": 1543680 + }, + { + "epoch": 0.01, + "learning_rate": 2.4718956106771455e-05, + "loss": 3.4606, + "step": 1544192 + }, + { + "epoch": 0.01, + "learning_rate": 2.4710586538064668e-05, + "loss": 3.4803, + "step": 1544704 + }, + { + "epoch": 0.01, + "learning_rate": 2.4702200590554148e-05, + "loss": 3.4658, + "step": 1545216 + }, + { + "epoch": 0.01, + "learning_rate": 2.4693814643043628e-05, + "loss": 3.4712, + "step": 1545728 + }, + { + "epoch": 0.01, + "learning_rate": 2.468544507433684e-05, + "loss": 3.4548, + "step": 1546240 + }, + { + "epoch": 0.01, + "learning_rate": 2.4677059126826317e-05, + "loss": 3.4669, + "step": 1546752 + }, + { + "epoch": 0.01, + "learning_rate": 2.46686731793158e-05, + "loss": 3.4674, + "step": 1547264 + }, + { + "epoch": 0.01, + "learning_rate": 2.466028723180528e-05, + "loss": 3.4798, + "step": 1547776 + }, + { + "epoch": 0.01, + "learning_rate": 2.465190128429476e-05, + "loss": 3.4669, + "step": 1548288 + }, + { + "epoch": 0.01, + "learning_rate": 2.464351533678424e-05, + "loss": 3.4711, + "step": 1548800 + }, + { + "epoch": 0.01, + "learning_rate": 2.463512938927372e-05, + "loss": 3.4726, + "step": 1549312 + }, + { + "epoch": 0.01, + "learning_rate": 2.46267434417632e-05, + "loss": 3.4755, + "step": 1549824 + }, + { + "epoch": 0.01, + "learning_rate": 2.461837387305641e-05, + "loss": 3.471, + "step": 1550336 + }, + { + "epoch": 0.01, + "learning_rate": 2.4609987925545892e-05, + "loss": 3.46, + "step": 1550848 + }, + { + "epoch": 0.01, + "learning_rate": 2.4601601978035372e-05, + "loss": 3.4544, + "step": 1551360 + }, + { + "epoch": 0.01, + "learning_rate": 2.459321603052485e-05, + "loss": 3.4799, + "step": 1551872 + }, + { + "epoch": 0.01, + "learning_rate": 2.458484646181806e-05, + "loss": 3.4752, + "step": 1552384 + }, + { + "epoch": 0.01, + "learning_rate": 2.457647689311127e-05, + "loss": 3.4675, + "step": 1552896 + }, + { + "epoch": 0.01, + "learning_rate": 2.4568090945600754e-05, + "loss": 3.4662, + "step": 1553408 + }, + { + "epoch": 0.01, + "learning_rate": 2.4559704998090234e-05, + "loss": 3.4698, + "step": 1553920 + }, + { + "epoch": 0.01, + "learning_rate": 2.4551319050579714e-05, + "loss": 3.4511, + "step": 1554432 + }, + { + "epoch": 0.01, + "learning_rate": 2.4542933103069194e-05, + "loss": 3.4671, + "step": 1554944 + }, + { + "epoch": 0.01, + "learning_rate": 2.4534547155558674e-05, + "loss": 3.4587, + "step": 1555456 + }, + { + "epoch": 0.01, + "learning_rate": 2.452616120804815e-05, + "loss": 3.4647, + "step": 1555968 + }, + { + "epoch": 0.01, + "learning_rate": 2.451777526053763e-05, + "loss": 3.4763, + "step": 1556480 + }, + { + "epoch": 0.01, + "learning_rate": 2.4509405691830846e-05, + "loss": 3.4616, + "step": 1556992 + }, + { + "epoch": 0.01, + "learning_rate": 2.4501019744320323e-05, + "loss": 3.4559, + "step": 1557504 + }, + { + "epoch": 0.01, + "learning_rate": 2.4492633796809803e-05, + "loss": 3.469, + "step": 1558016 + }, + { + "epoch": 0.01, + "learning_rate": 2.4484247849299283e-05, + "loss": 3.4528, + "step": 1558528 + }, + { + "epoch": 0.01, + "learning_rate": 2.4475894659396224e-05, + "loss": 3.4602, + "step": 1559040 + }, + { + "epoch": 0.01, + "learning_rate": 2.4467508711885708e-05, + "loss": 3.47, + "step": 1559552 + }, + { + "epoch": 0.01, + "learning_rate": 2.4459122764375187e-05, + "loss": 3.4673, + "step": 1560064 + }, + { + "epoch": 0.01, + "learning_rate": 2.4450736816864667e-05, + "loss": 3.4576, + "step": 1560576 + }, + { + "epoch": 0.01, + "learning_rate": 2.4442350869354147e-05, + "loss": 3.4585, + "step": 1561088 + }, + { + "epoch": 0.01, + "learning_rate": 2.4433964921843624e-05, + "loss": 3.4601, + "step": 1561600 + }, + { + "epoch": 0.01, + "learning_rate": 2.4425578974333104e-05, + "loss": 3.4638, + "step": 1562112 + }, + { + "epoch": 0.01, + "learning_rate": 2.4417193026822584e-05, + "loss": 3.4688, + "step": 1562624 + }, + { + "epoch": 0.01, + "learning_rate": 2.4408823458115796e-05, + "loss": 3.4622, + "step": 1563136 + }, + { + "epoch": 0.01, + "learning_rate": 2.4400437510605276e-05, + "loss": 3.4751, + "step": 1563648 + }, + { + "epoch": 0.01, + "learning_rate": 2.4392051563094756e-05, + "loss": 3.4736, + "step": 1564160 + }, + { + "epoch": 0.01, + "learning_rate": 2.4383665615584236e-05, + "loss": 3.4626, + "step": 1564672 + }, + { + "epoch": 0.01, + "learning_rate": 2.4375296046877445e-05, + "loss": 3.4619, + "step": 1565184 + }, + { + "epoch": 0.01, + "learning_rate": 2.4366910099366925e-05, + "loss": 3.4758, + "step": 1565696 + }, + { + "epoch": 0.01, + "learning_rate": 2.435852415185641e-05, + "loss": 3.4681, + "step": 1566208 + }, + { + "epoch": 0.01, + "learning_rate": 2.435013820434589e-05, + "loss": 3.4591, + "step": 1566720 + }, + { + "epoch": 0.01, + "learning_rate": 2.4341768635639098e-05, + "loss": 3.466, + "step": 1567232 + }, + { + "epoch": 0.01, + "learning_rate": 2.4333382688128578e-05, + "loss": 3.4681, + "step": 1567744 + }, + { + "epoch": 0.01, + "learning_rate": 2.4324996740618058e-05, + "loss": 3.4781, + "step": 1568256 + }, + { + "epoch": 0.01, + "learning_rate": 2.431662717191127e-05, + "loss": 3.4605, + "step": 1568768 + }, + { + "epoch": 0.01, + "learning_rate": 2.430824122440075e-05, + "loss": 3.4509, + "step": 1569280 + }, + { + "epoch": 0.01, + "learning_rate": 2.429985527689023e-05, + "loss": 3.4563, + "step": 1569792 + }, + { + "epoch": 0.01, + "learning_rate": 2.429146932937971e-05, + "loss": 3.4597, + "step": 1570304 + }, + { + "epoch": 0.01, + "learning_rate": 2.428308338186919e-05, + "loss": 3.4688, + "step": 1570816 + }, + { + "epoch": 0.01, + "learning_rate": 2.427469743435867e-05, + "loss": 3.4644, + "step": 1571328 + }, + { + "epoch": 0.01, + "learning_rate": 2.426631148684815e-05, + "loss": 3.4578, + "step": 1571840 + }, + { + "epoch": 0.02, + "learning_rate": 2.4257941918141362e-05, + "loss": 3.4675, + "step": 1572352 + }, + { + "epoch": 0.02, + "learning_rate": 2.4249555970630842e-05, + "loss": 3.462, + "step": 1572864 + }, + { + "epoch": 0.02, + "learning_rate": 2.4241170023120322e-05, + "loss": 3.4616, + "step": 1573376 + }, + { + "epoch": 0.02, + "learning_rate": 2.423280045441353e-05, + "loss": 3.4579, + "step": 1573888 + }, + { + "epoch": 0.02, + "learning_rate": 2.422441450690301e-05, + "loss": 3.4673, + "step": 1574400 + }, + { + "epoch": 0.02, + "learning_rate": 2.421602855939249e-05, + "loss": 3.4751, + "step": 1574912 + }, + { + "epoch": 0.02, + "learning_rate": 2.420764261188197e-05, + "loss": 3.4665, + "step": 1575424 + }, + { + "epoch": 0.02, + "learning_rate": 2.419925666437145e-05, + "loss": 3.4445, + "step": 1575936 + }, + { + "epoch": 0.02, + "learning_rate": 2.419087071686093e-05, + "loss": 3.4673, + "step": 1576448 + }, + { + "epoch": 0.02, + "learning_rate": 2.418248476935041e-05, + "loss": 3.4633, + "step": 1576960 + }, + { + "epoch": 0.02, + "learning_rate": 2.4174115200643623e-05, + "loss": 3.4688, + "step": 1577472 + }, + { + "epoch": 0.02, + "learning_rate": 2.41657292531331e-05, + "loss": 3.4709, + "step": 1577984 + }, + { + "epoch": 0.02, + "learning_rate": 2.4157343305622583e-05, + "loss": 3.4614, + "step": 1578496 + }, + { + "epoch": 0.02, + "learning_rate": 2.4148957358112063e-05, + "loss": 3.4544, + "step": 1579008 + }, + { + "epoch": 0.02, + "learning_rate": 2.4140571410601543e-05, + "loss": 3.467, + "step": 1579520 + }, + { + "epoch": 0.02, + "learning_rate": 2.4132201841894752e-05, + "loss": 3.4493, + "step": 1580032 + }, + { + "epoch": 0.02, + "learning_rate": 2.4123815894384232e-05, + "loss": 3.4652, + "step": 1580544 + }, + { + "epoch": 0.02, + "learning_rate": 2.4115429946873712e-05, + "loss": 3.4578, + "step": 1581056 + }, + { + "epoch": 0.02, + "learning_rate": 2.4107043999363192e-05, + "loss": 3.4675, + "step": 1581568 + }, + { + "epoch": 0.02, + "learning_rate": 2.4098658051852672e-05, + "loss": 3.451, + "step": 1582080 + }, + { + "epoch": 0.02, + "learning_rate": 2.4090272104342156e-05, + "loss": 3.4699, + "step": 1582592 + }, + { + "epoch": 0.02, + "learning_rate": 2.4081902535635365e-05, + "loss": 3.4552, + "step": 1583104 + }, + { + "epoch": 0.02, + "learning_rate": 2.4073516588124845e-05, + "loss": 3.4595, + "step": 1583616 + }, + { + "epoch": 0.02, + "learning_rate": 2.4065130640614325e-05, + "loss": 3.469, + "step": 1584128 + }, + { + "epoch": 0.02, + "learning_rate": 2.4056744693103805e-05, + "loss": 3.4662, + "step": 1584640 + }, + { + "epoch": 0.02, + "learning_rate": 2.4048358745593284e-05, + "loss": 3.4752, + "step": 1585152 + }, + { + "epoch": 0.02, + "learning_rate": 2.4039972798082764e-05, + "loss": 3.4711, + "step": 1585664 + }, + { + "epoch": 0.02, + "learning_rate": 2.4031603229375977e-05, + "loss": 3.4607, + "step": 1586176 + }, + { + "epoch": 0.02, + "learning_rate": 2.4023217281865454e-05, + "loss": 3.4629, + "step": 1586688 + }, + { + "epoch": 0.02, + "learning_rate": 2.4014831334354933e-05, + "loss": 3.4599, + "step": 1587200 + }, + { + "epoch": 0.02, + "learning_rate": 2.4006445386844413e-05, + "loss": 3.4655, + "step": 1587712 + }, + { + "epoch": 0.02, + "learning_rate": 2.3998059439333893e-05, + "loss": 3.4574, + "step": 1588224 + }, + { + "epoch": 0.02, + "learning_rate": 2.3989689870627106e-05, + "loss": 3.469, + "step": 1588736 + }, + { + "epoch": 0.02, + "learning_rate": 2.3981303923116586e-05, + "loss": 3.4622, + "step": 1589248 + }, + { + "epoch": 0.02, + "learning_rate": 2.3972917975606066e-05, + "loss": 3.4607, + "step": 1589760 + }, + { + "epoch": 0.02, + "learning_rate": 2.3964532028095546e-05, + "loss": 3.4601, + "step": 1590272 + }, + { + "epoch": 0.02, + "learning_rate": 2.3956146080585026e-05, + "loss": 3.4685, + "step": 1590784 + }, + { + "epoch": 0.02, + "learning_rate": 2.3947760133074506e-05, + "loss": 3.4583, + "step": 1591296 + }, + { + "epoch": 0.02, + "learning_rate": 2.3939374185563986e-05, + "loss": 3.4653, + "step": 1591808 + }, + { + "epoch": 0.02, + "learning_rate": 2.3930988238053466e-05, + "loss": 3.4589, + "step": 1592320 + }, + { + "epoch": 0.02, + "learning_rate": 2.3922635048150407e-05, + "loss": 3.472, + "step": 1592832 + }, + { + "epoch": 0.02, + "learning_rate": 2.3914249100639887e-05, + "loss": 3.467, + "step": 1593344 + }, + { + "epoch": 0.02, + "learning_rate": 2.3905863153129367e-05, + "loss": 3.469, + "step": 1593856 + }, + { + "epoch": 0.02, + "learning_rate": 2.3897477205618847e-05, + "loss": 3.4609, + "step": 1594368 + }, + { + "epoch": 0.02, + "learning_rate": 2.388909125810833e-05, + "loss": 3.4597, + "step": 1594880 + }, + { + "epoch": 0.02, + "learning_rate": 2.388070531059781e-05, + "loss": 3.4712, + "step": 1595392 + }, + { + "epoch": 0.02, + "learning_rate": 2.3872319363087287e-05, + "loss": 3.4631, + "step": 1595904 + }, + { + "epoch": 0.02, + "learning_rate": 2.38639497943805e-05, + "loss": 3.4558, + "step": 1596416 + }, + { + "epoch": 0.02, + "learning_rate": 2.385556384686998e-05, + "loss": 3.4704, + "step": 1596928 + }, + { + "epoch": 0.02, + "learning_rate": 2.384717789935946e-05, + "loss": 3.4673, + "step": 1597440 + }, + { + "epoch": 0.02, + "learning_rate": 2.383879195184894e-05, + "loss": 3.4778, + "step": 1597952 + }, + { + "epoch": 0.02, + "learning_rate": 2.383040600433842e-05, + "loss": 3.4643, + "step": 1598464 + }, + { + "epoch": 0.02, + "learning_rate": 2.38220200568279e-05, + "loss": 3.4702, + "step": 1598976 + }, + { + "epoch": 0.02, + "learning_rate": 2.381363410931738e-05, + "loss": 3.4692, + "step": 1599488 + }, + { + "epoch": 0.02, + "learning_rate": 2.3805264540610588e-05, + "loss": 3.4585, + "step": 1600000 + }, + { + "epoch": 0.02, + "learning_rate": 2.3796878593100068e-05, + "loss": 3.4648, + "step": 1600512 + }, + { + "epoch": 0.02, + "learning_rate": 2.3788492645589548e-05, + "loss": 3.4679, + "step": 1601024 + }, + { + "epoch": 0.02, + "learning_rate": 2.378010669807903e-05, + "loss": 3.465, + "step": 1601536 + }, + { + "epoch": 0.02, + "learning_rate": 2.377173712937224e-05, + "loss": 3.451, + "step": 1602048 + }, + { + "epoch": 0.02, + "learning_rate": 2.376335118186172e-05, + "loss": 3.4649, + "step": 1602560 + }, + { + "epoch": 0.03, + "eval_loss": 3.8561511039733887, + "eval_runtime": 317.9257, + "eval_samples_per_second": 1200.252, + "eval_steps_per_second": 37.509, + "step": 1602720 + }, + { + "epoch": 1.0, + "learning_rate": 2.37549652343512e-05, + "loss": 3.4606, + "step": 1603072 + }, + { + "epoch": 1.0, + "learning_rate": 2.374657928684068e-05, + "loss": 3.4552, + "step": 1603584 + }, + { + "epoch": 1.0, + "learning_rate": 2.373819333933016e-05, + "loss": 3.4719, + "step": 1604096 + }, + { + "epoch": 1.0, + "learning_rate": 2.372980739181964e-05, + "loss": 3.4748, + "step": 1604608 + }, + { + "epoch": 1.0, + "learning_rate": 2.372142144430912e-05, + "loss": 3.4709, + "step": 1605120 + }, + { + "epoch": 1.0, + "learning_rate": 2.3713051875602333e-05, + "loss": 3.4617, + "step": 1605632 + }, + { + "epoch": 1.0, + "learning_rate": 2.3704665928091813e-05, + "loss": 3.4633, + "step": 1606144 + }, + { + "epoch": 1.0, + "learning_rate": 2.369627998058129e-05, + "loss": 3.449, + "step": 1606656 + }, + { + "epoch": 1.0, + "learning_rate": 2.368789403307077e-05, + "loss": 3.4683, + "step": 1607168 + }, + { + "epoch": 1.0, + "learning_rate": 2.367950808556025e-05, + "loss": 3.4693, + "step": 1607680 + }, + { + "epoch": 1.0, + "learning_rate": 2.3671122138049733e-05, + "loss": 3.4584, + "step": 1608192 + }, + { + "epoch": 1.0, + "learning_rate": 2.366275256934294e-05, + "loss": 3.4659, + "step": 1608704 + }, + { + "epoch": 1.0, + "learning_rate": 2.365436662183242e-05, + "loss": 3.454, + "step": 1609216 + }, + { + "epoch": 1.0, + "learning_rate": 2.36459806743219e-05, + "loss": 3.4569, + "step": 1609728 + }, + { + "epoch": 1.0, + "learning_rate": 2.363759472681138e-05, + "loss": 3.4514, + "step": 1610240 + }, + { + "epoch": 1.0, + "learning_rate": 2.362920877930086e-05, + "loss": 3.4559, + "step": 1610752 + }, + { + "epoch": 1.0, + "learning_rate": 2.362082283179034e-05, + "loss": 3.4528, + "step": 1611264 + }, + { + "epoch": 1.0, + "learning_rate": 2.361243688427982e-05, + "loss": 3.4585, + "step": 1611776 + }, + { + "epoch": 1.0, + "learning_rate": 2.36040509367693e-05, + "loss": 3.4599, + "step": 1612288 + }, + { + "epoch": 1.0, + "learning_rate": 2.3595697746866243e-05, + "loss": 3.4758, + "step": 1612800 + }, + { + "epoch": 1.0, + "learning_rate": 2.3587311799355723e-05, + "loss": 3.4685, + "step": 1613312 + }, + { + "epoch": 1.0, + "learning_rate": 2.3578925851845203e-05, + "loss": 3.4652, + "step": 1613824 + }, + { + "epoch": 1.0, + "learning_rate": 2.3570539904334686e-05, + "loss": 3.4608, + "step": 1614336 + }, + { + "epoch": 1.0, + "learning_rate": 2.3562153956824166e-05, + "loss": 3.4667, + "step": 1614848 + }, + { + "epoch": 1.0, + "learning_rate": 2.3553768009313646e-05, + "loss": 3.4551, + "step": 1615360 + }, + { + "epoch": 1.0, + "learning_rate": 2.3545398440606855e-05, + "loss": 3.4675, + "step": 1615872 + }, + { + "epoch": 1.0, + "learning_rate": 2.3537012493096335e-05, + "loss": 3.4605, + "step": 1616384 + }, + { + "epoch": 1.0, + "learning_rate": 2.3528626545585815e-05, + "loss": 3.4511, + "step": 1616896 + }, + { + "epoch": 1.0, + "learning_rate": 2.3520240598075295e-05, + "loss": 3.4524, + "step": 1617408 + }, + { + "epoch": 1.0, + "learning_rate": 2.3511871029368508e-05, + "loss": 3.4554, + "step": 1617920 + }, + { + "epoch": 1.01, + "learning_rate": 2.3503485081857988e-05, + "loss": 3.4613, + "step": 1618432 + }, + { + "epoch": 1.01, + "learning_rate": 2.3495099134347468e-05, + "loss": 3.4667, + "step": 1618944 + }, + { + "epoch": 1.01, + "learning_rate": 2.3486713186836944e-05, + "loss": 3.4652, + "step": 1619456 + }, + { + "epoch": 1.01, + "learning_rate": 2.3478327239326424e-05, + "loss": 3.4674, + "step": 1619968 + }, + { + "epoch": 1.01, + "learning_rate": 2.3469941291815904e-05, + "loss": 3.4519, + "step": 1620480 + }, + { + "epoch": 1.01, + "learning_rate": 2.3461555344305387e-05, + "loss": 3.4633, + "step": 1620992 + }, + { + "epoch": 1.01, + "learning_rate": 2.3453169396794867e-05, + "loss": 3.4574, + "step": 1621504 + }, + { + "epoch": 1.01, + "learning_rate": 2.3444799828088076e-05, + "loss": 3.4524, + "step": 1622016 + }, + { + "epoch": 1.01, + "learning_rate": 2.343643025938129e-05, + "loss": 3.4486, + "step": 1622528 + }, + { + "epoch": 1.01, + "learning_rate": 2.3428044311870765e-05, + "loss": 3.4526, + "step": 1623040 + }, + { + "epoch": 1.01, + "learning_rate": 2.341965836436025e-05, + "loss": 3.4593, + "step": 1623552 + }, + { + "epoch": 1.01, + "learning_rate": 2.341127241684973e-05, + "loss": 3.4645, + "step": 1624064 + }, + { + "epoch": 1.01, + "learning_rate": 2.340288646933921e-05, + "loss": 3.4575, + "step": 1624576 + }, + { + "epoch": 1.01, + "learning_rate": 2.339450052182869e-05, + "loss": 3.4581, + "step": 1625088 + }, + { + "epoch": 1.01, + "learning_rate": 2.338611457431817e-05, + "loss": 3.4635, + "step": 1625600 + }, + { + "epoch": 1.01, + "learning_rate": 2.337772862680765e-05, + "loss": 3.4618, + "step": 1626112 + }, + { + "epoch": 1.01, + "learning_rate": 2.3369359058100858e-05, + "loss": 3.4627, + "step": 1626624 + }, + { + "epoch": 1.01, + "learning_rate": 2.336097311059034e-05, + "loss": 3.4505, + "step": 1627136 + }, + { + "epoch": 1.01, + "learning_rate": 2.335258716307982e-05, + "loss": 3.437, + "step": 1627648 + }, + { + "epoch": 1.01, + "learning_rate": 2.33442012155693e-05, + "loss": 3.468, + "step": 1628160 + }, + { + "epoch": 1.01, + "learning_rate": 2.333583164686251e-05, + "loss": 3.4668, + "step": 1628672 + }, + { + "epoch": 1.01, + "learning_rate": 2.332746207815572e-05, + "loss": 3.4565, + "step": 1629184 + }, + { + "epoch": 1.01, + "learning_rate": 2.3319076130645202e-05, + "loss": 3.4558, + "step": 1629696 + }, + { + "epoch": 1.01, + "learning_rate": 2.3310690183134682e-05, + "loss": 3.4549, + "step": 1630208 + }, + { + "epoch": 1.01, + "learning_rate": 2.3302304235624162e-05, + "loss": 3.4433, + "step": 1630720 + }, + { + "epoch": 1.01, + "learning_rate": 2.3293918288113642e-05, + "loss": 3.4533, + "step": 1631232 + }, + { + "epoch": 1.01, + "learning_rate": 2.3285532340603122e-05, + "loss": 3.4464, + "step": 1631744 + }, + { + "epoch": 1.01, + "learning_rate": 2.32771463930926e-05, + "loss": 3.455, + "step": 1632256 + }, + { + "epoch": 1.01, + "learning_rate": 2.326876044558208e-05, + "loss": 3.4663, + "step": 1632768 + }, + { + "epoch": 1.01, + "learning_rate": 2.3260407255679024e-05, + "loss": 3.4493, + "step": 1633280 + }, + { + "epoch": 1.01, + "learning_rate": 2.3252021308168504e-05, + "loss": 3.4444, + "step": 1633792 + }, + { + "epoch": 1.01, + "learning_rate": 2.3243635360657984e-05, + "loss": 3.4593, + "step": 1634304 + }, + { + "epoch": 1.01, + "learning_rate": 2.3235249413147464e-05, + "loss": 3.4415, + "step": 1634816 + }, + { + "epoch": 1.01, + "learning_rate": 2.3226863465636944e-05, + "loss": 3.4452, + "step": 1635328 + }, + { + "epoch": 1.01, + "learning_rate": 2.3218493896930156e-05, + "loss": 3.4593, + "step": 1635840 + }, + { + "epoch": 1.01, + "learning_rate": 2.3210107949419636e-05, + "loss": 3.4579, + "step": 1636352 + }, + { + "epoch": 1.01, + "learning_rate": 2.3201722001909116e-05, + "loss": 3.4456, + "step": 1636864 + }, + { + "epoch": 1.01, + "learning_rate": 2.3193336054398596e-05, + "loss": 3.4475, + "step": 1637376 + }, + { + "epoch": 1.01, + "learning_rate": 2.3184950106888073e-05, + "loss": 3.4485, + "step": 1637888 + }, + { + "epoch": 1.01, + "learning_rate": 2.3176564159377552e-05, + "loss": 3.45, + "step": 1638400 + }, + { + "epoch": 1.01, + "learning_rate": 2.3168178211867032e-05, + "loss": 3.4605, + "step": 1638912 + }, + { + "epoch": 1.01, + "learning_rate": 2.315980864316025e-05, + "loss": 3.4485, + "step": 1639424 + }, + { + "epoch": 1.01, + "learning_rate": 2.3151422695649725e-05, + "loss": 3.4616, + "step": 1639936 + }, + { + "epoch": 1.01, + "learning_rate": 2.3143053126942937e-05, + "loss": 3.4612, + "step": 1640448 + }, + { + "epoch": 1.01, + "learning_rate": 2.3134667179432417e-05, + "loss": 3.4587, + "step": 1640960 + }, + { + "epoch": 1.01, + "learning_rate": 2.3126281231921894e-05, + "loss": 3.4462, + "step": 1641472 + }, + { + "epoch": 1.01, + "learning_rate": 2.3117895284411374e-05, + "loss": 3.4627, + "step": 1641984 + }, + { + "epoch": 1.01, + "learning_rate": 2.3109509336900857e-05, + "loss": 3.4603, + "step": 1642496 + }, + { + "epoch": 1.01, + "learning_rate": 2.3101123389390337e-05, + "loss": 3.4463, + "step": 1643008 + }, + { + "epoch": 1.01, + "learning_rate": 2.3092737441879817e-05, + "loss": 3.4601, + "step": 1643520 + }, + { + "epoch": 1.01, + "learning_rate": 2.3084351494369297e-05, + "loss": 3.4539, + "step": 1644032 + }, + { + "epoch": 1.01, + "learning_rate": 2.3075981925662506e-05, + "loss": 3.4657, + "step": 1644544 + }, + { + "epoch": 1.01, + "learning_rate": 2.3067595978151986e-05, + "loss": 3.4503, + "step": 1645056 + }, + { + "epoch": 1.01, + "learning_rate": 2.30592264094452e-05, + "loss": 3.4361, + "step": 1645568 + }, + { + "epoch": 1.01, + "learning_rate": 2.305084046193468e-05, + "loss": 3.4486, + "step": 1646080 + }, + { + "epoch": 1.01, + "learning_rate": 2.304245451442416e-05, + "loss": 3.4489, + "step": 1646592 + }, + { + "epoch": 1.01, + "learning_rate": 2.303406856691364e-05, + "loss": 3.4564, + "step": 1647104 + }, + { + "epoch": 1.01, + "learning_rate": 2.302568261940312e-05, + "loss": 3.4523, + "step": 1647616 + }, + { + "epoch": 1.01, + "learning_rate": 2.30172966718926e-05, + "loss": 3.4481, + "step": 1648128 + }, + { + "epoch": 1.02, + "learning_rate": 2.300892710318581e-05, + "loss": 3.4544, + "step": 1648640 + }, + { + "epoch": 1.02, + "learning_rate": 2.300054115567529e-05, + "loss": 3.454, + "step": 1649152 + }, + { + "epoch": 1.02, + "learning_rate": 2.299215520816477e-05, + "loss": 3.4443, + "step": 1649664 + }, + { + "epoch": 1.02, + "learning_rate": 2.298376926065425e-05, + "loss": 3.4517, + "step": 1650176 + }, + { + "epoch": 1.02, + "learning_rate": 2.2975383313143727e-05, + "loss": 3.4586, + "step": 1650688 + }, + { + "epoch": 1.02, + "learning_rate": 2.2966997365633207e-05, + "loss": 3.466, + "step": 1651200 + }, + { + "epoch": 1.02, + "learning_rate": 2.295862779692642e-05, + "loss": 3.4547, + "step": 1651712 + }, + { + "epoch": 1.02, + "learning_rate": 2.29502418494159e-05, + "loss": 3.4353, + "step": 1652224 + }, + { + "epoch": 1.02, + "learning_rate": 2.294185590190538e-05, + "loss": 3.4555, + "step": 1652736 + }, + { + "epoch": 1.02, + "learning_rate": 2.293346995439486e-05, + "loss": 3.4503, + "step": 1653248 + }, + { + "epoch": 1.02, + "learning_rate": 2.292508400688434e-05, + "loss": 3.4589, + "step": 1653760 + }, + { + "epoch": 1.02, + "learning_rate": 2.291669805937382e-05, + "loss": 3.4605, + "step": 1654272 + }, + { + "epoch": 1.02, + "learning_rate": 2.29083121118633e-05, + "loss": 3.4456, + "step": 1654784 + }, + { + "epoch": 1.02, + "learning_rate": 2.289992616435278e-05, + "loss": 3.4462, + "step": 1655296 + }, + { + "epoch": 1.02, + "learning_rate": 2.2891556595645992e-05, + "loss": 3.4581, + "step": 1655808 + }, + { + "epoch": 1.02, + "learning_rate": 2.28831870269392e-05, + "loss": 3.4416, + "step": 1656320 + }, + { + "epoch": 1.02, + "learning_rate": 2.287480107942868e-05, + "loss": 3.4537, + "step": 1656832 + }, + { + "epoch": 1.02, + "learning_rate": 2.286641513191816e-05, + "loss": 3.44, + "step": 1657344 + }, + { + "epoch": 1.02, + "learning_rate": 2.285802918440764e-05, + "loss": 3.4567, + "step": 1657856 + }, + { + "epoch": 1.02, + "learning_rate": 2.284964323689712e-05, + "loss": 3.4406, + "step": 1658368 + }, + { + "epoch": 1.02, + "learning_rate": 2.2841257289386604e-05, + "loss": 3.4562, + "step": 1658880 + }, + { + "epoch": 1.02, + "learning_rate": 2.283287134187608e-05, + "loss": 3.4471, + "step": 1659392 + }, + { + "epoch": 1.02, + "learning_rate": 2.2824501773169293e-05, + "loss": 3.4444, + "step": 1659904 + }, + { + "epoch": 1.02, + "learning_rate": 2.2816115825658773e-05, + "loss": 3.4583, + "step": 1660416 + }, + { + "epoch": 1.02, + "learning_rate": 2.2807729878148253e-05, + "loss": 3.4516, + "step": 1660928 + }, + { + "epoch": 1.02, + "learning_rate": 2.2799343930637733e-05, + "loss": 3.4637, + "step": 1661440 + }, + { + "epoch": 1.02, + "learning_rate": 2.2790957983127213e-05, + "loss": 3.462, + "step": 1661952 + }, + { + "epoch": 1.02, + "learning_rate": 2.2782588414420426e-05, + "loss": 3.4474, + "step": 1662464 + }, + { + "epoch": 1.02, + "learning_rate": 2.2774202466909905e-05, + "loss": 3.4545, + "step": 1662976 + }, + { + "epoch": 1.02, + "learning_rate": 2.2765816519399382e-05, + "loss": 3.4493, + "step": 1663488 + }, + { + "epoch": 1.02, + "learning_rate": 2.2757430571888862e-05, + "loss": 3.4532, + "step": 1664000 + }, + { + "epoch": 1.02, + "learning_rate": 2.2749061003182075e-05, + "loss": 3.4497, + "step": 1664512 + }, + { + "epoch": 1.02, + "learning_rate": 2.2740675055671554e-05, + "loss": 3.4552, + "step": 1665024 + }, + { + "epoch": 1.02, + "learning_rate": 2.2732289108161034e-05, + "loss": 3.451, + "step": 1665536 + }, + { + "epoch": 1.02, + "learning_rate": 2.2723903160650514e-05, + "loss": 3.4472, + "step": 1666048 + }, + { + "epoch": 1.02, + "learning_rate": 2.2715517213139994e-05, + "loss": 3.4506, + "step": 1666560 + }, + { + "epoch": 1.02, + "learning_rate": 2.2707131265629474e-05, + "loss": 3.4563, + "step": 1667072 + }, + { + "epoch": 1.02, + "learning_rate": 2.2698745318118954e-05, + "loss": 3.445, + "step": 1667584 + }, + { + "epoch": 1.02, + "learning_rate": 2.2690359370608434e-05, + "loss": 3.4539, + "step": 1668096 + }, + { + "epoch": 1.02, + "learning_rate": 2.2681989801901647e-05, + "loss": 3.4447, + "step": 1668608 + }, + { + "epoch": 1.02, + "learning_rate": 2.2673620233194856e-05, + "loss": 3.4675, + "step": 1669120 + }, + { + "epoch": 1.02, + "learning_rate": 2.2665234285684336e-05, + "loss": 3.448, + "step": 1669632 + }, + { + "epoch": 1.02, + "learning_rate": 2.2656848338173816e-05, + "loss": 3.4609, + "step": 1670144 + }, + { + "epoch": 1.02, + "learning_rate": 2.2648462390663296e-05, + "loss": 3.45, + "step": 1670656 + }, + { + "epoch": 1.02, + "learning_rate": 2.264007644315278e-05, + "loss": 3.4507, + "step": 1671168 + }, + { + "epoch": 1.02, + "learning_rate": 2.263169049564226e-05, + "loss": 3.4579, + "step": 1671680 + }, + { + "epoch": 1.02, + "learning_rate": 2.2623304548131736e-05, + "loss": 3.4514, + "step": 1672192 + }, + { + "epoch": 1.02, + "learning_rate": 2.2614934979424948e-05, + "loss": 3.4474, + "step": 1672704 + }, + { + "epoch": 1.02, + "learning_rate": 2.2606549031914428e-05, + "loss": 3.4607, + "step": 1673216 + }, + { + "epoch": 1.02, + "learning_rate": 2.2598163084403908e-05, + "loss": 3.4555, + "step": 1673728 + }, + { + "epoch": 1.02, + "learning_rate": 2.2589777136893388e-05, + "loss": 3.4653, + "step": 1674240 + }, + { + "epoch": 1.02, + "learning_rate": 2.2581391189382868e-05, + "loss": 3.4496, + "step": 1674752 + }, + { + "epoch": 1.02, + "learning_rate": 2.2573005241872348e-05, + "loss": 3.4582, + "step": 1675264 + }, + { + "epoch": 1.02, + "learning_rate": 2.256463567316556e-05, + "loss": 3.4551, + "step": 1675776 + }, + { + "epoch": 1.02, + "learning_rate": 2.2556249725655037e-05, + "loss": 3.4501, + "step": 1676288 + }, + { + "epoch": 1.02, + "learning_rate": 2.2547863778144517e-05, + "loss": 3.4547, + "step": 1676800 + }, + { + "epoch": 1.02, + "learning_rate": 2.2539477830633997e-05, + "loss": 3.4561, + "step": 1677312 + }, + { + "epoch": 1.02, + "learning_rate": 2.253110826192721e-05, + "loss": 3.4616, + "step": 1677824 + }, + { + "epoch": 1.02, + "learning_rate": 2.252272231441669e-05, + "loss": 3.4325, + "step": 1678336 + }, + { + "epoch": 1.02, + "learning_rate": 2.251433636690617e-05, + "loss": 3.4587, + "step": 1678848 + }, + { + "epoch": 1.03, + "eval_loss": 3.8573989868164062, + "eval_runtime": 321.5113, + "eval_samples_per_second": 1186.867, + "eval_steps_per_second": 37.09, + "step": 1679040 + }, + { + "epoch": 0.0, + "learning_rate": 2.250595041939565e-05, + "loss": 3.4452, + "step": 1679360 + }, + { + "epoch": 0.0, + "learning_rate": 2.249756447188513e-05, + "loss": 3.4437, + "step": 1679872 + }, + { + "epoch": 0.0, + "learning_rate": 2.248919490317834e-05, + "loss": 3.4571, + "step": 1680384 + }, + { + "epoch": 0.0, + "learning_rate": 2.248080895566782e-05, + "loss": 3.4668, + "step": 1680896 + }, + { + "epoch": 0.0, + "learning_rate": 2.24724230081573e-05, + "loss": 3.4583, + "step": 1681408 + }, + { + "epoch": 0.0, + "learning_rate": 2.246403706064678e-05, + "loss": 3.4484, + "step": 1681920 + }, + { + "epoch": 0.0, + "learning_rate": 2.245565111313626e-05, + "loss": 3.4512, + "step": 1682432 + }, + { + "epoch": 0.0, + "learning_rate": 2.244726516562574e-05, + "loss": 3.4392, + "step": 1682944 + }, + { + "epoch": 0.0, + "learning_rate": 2.243889559691895e-05, + "loss": 3.4559, + "step": 1683456 + }, + { + "epoch": 0.0, + "learning_rate": 2.2430509649408434e-05, + "loss": 3.4552, + "step": 1683968 + }, + { + "epoch": 0.0, + "learning_rate": 2.2422123701897914e-05, + "loss": 3.4524, + "step": 1684480 + }, + { + "epoch": 0.0, + "learning_rate": 2.241373775438739e-05, + "loss": 3.4529, + "step": 1684992 + }, + { + "epoch": 0.0, + "learning_rate": 2.240535180687687e-05, + "loss": 3.4486, + "step": 1685504 + }, + { + "epoch": 0.0, + "learning_rate": 2.2396982238170083e-05, + "loss": 3.4438, + "step": 1686016 + }, + { + "epoch": 0.0, + "learning_rate": 2.2388596290659563e-05, + "loss": 3.4416, + "step": 1686528 + }, + { + "epoch": 0.0, + "learning_rate": 2.2380210343149043e-05, + "loss": 3.4394, + "step": 1687040 + }, + { + "epoch": 0.0, + "learning_rate": 2.2371824395638523e-05, + "loss": 3.4448, + "step": 1687552 + }, + { + "epoch": 0.0, + "learning_rate": 2.2363438448128003e-05, + "loss": 3.4446, + "step": 1688064 + }, + { + "epoch": 0.0, + "learning_rate": 2.2355052500617482e-05, + "loss": 3.4498, + "step": 1688576 + }, + { + "epoch": 0.0, + "learning_rate": 2.2346666553106962e-05, + "loss": 3.4638, + "step": 1689088 + }, + { + "epoch": 0.0, + "learning_rate": 2.233829698440017e-05, + "loss": 3.4604, + "step": 1689600 + }, + { + "epoch": 0.0, + "learning_rate": 2.2329927415693387e-05, + "loss": 3.4511, + "step": 1690112 + }, + { + "epoch": 0.0, + "learning_rate": 2.2321541468182864e-05, + "loss": 3.4513, + "step": 1690624 + }, + { + "epoch": 0.0, + "learning_rate": 2.2313155520672344e-05, + "loss": 3.4562, + "step": 1691136 + }, + { + "epoch": 0.0, + "learning_rate": 2.2304769573161824e-05, + "loss": 3.4441, + "step": 1691648 + }, + { + "epoch": 0.0, + "learning_rate": 2.2296383625651304e-05, + "loss": 3.4599, + "step": 1692160 + }, + { + "epoch": 0.0, + "learning_rate": 2.2288014056944513e-05, + "loss": 3.4481, + "step": 1692672 + }, + { + "epoch": 0.0, + "learning_rate": 2.2279628109433996e-05, + "loss": 3.438, + "step": 1693184 + }, + { + "epoch": 0.0, + "learning_rate": 2.2271242161923476e-05, + "loss": 3.4447, + "step": 1693696 + }, + { + "epoch": 0.0, + "learning_rate": 2.2262856214412956e-05, + "loss": 3.4444, + "step": 1694208 + }, + { + "epoch": 0.01, + "learning_rate": 2.2254470266902436e-05, + "loss": 3.4482, + "step": 1694720 + }, + { + "epoch": 0.01, + "learning_rate": 2.2246084319391916e-05, + "loss": 3.4547, + "step": 1695232 + }, + { + "epoch": 0.01, + "learning_rate": 2.2237698371881393e-05, + "loss": 3.4524, + "step": 1695744 + }, + { + "epoch": 0.01, + "learning_rate": 2.2229312424370873e-05, + "loss": 3.4533, + "step": 1696256 + }, + { + "epoch": 0.01, + "learning_rate": 2.222094285566409e-05, + "loss": 3.4383, + "step": 1696768 + }, + { + "epoch": 0.01, + "learning_rate": 2.221255690815357e-05, + "loss": 3.4574, + "step": 1697280 + }, + { + "epoch": 0.01, + "learning_rate": 2.2204170960643045e-05, + "loss": 3.4449, + "step": 1697792 + }, + { + "epoch": 0.01, + "learning_rate": 2.2195801391936258e-05, + "loss": 3.4395, + "step": 1698304 + }, + { + "epoch": 0.01, + "learning_rate": 2.2187431823229467e-05, + "loss": 3.4442, + "step": 1698816 + }, + { + "epoch": 0.01, + "learning_rate": 2.217904587571895e-05, + "loss": 3.4371, + "step": 1699328 + }, + { + "epoch": 0.01, + "learning_rate": 2.217065992820843e-05, + "loss": 3.4481, + "step": 1699840 + }, + { + "epoch": 0.01, + "learning_rate": 2.216227398069791e-05, + "loss": 3.4566, + "step": 1700352 + }, + { + "epoch": 0.01, + "learning_rate": 2.215388803318739e-05, + "loss": 3.4447, + "step": 1700864 + }, + { + "epoch": 0.01, + "learning_rate": 2.2145502085676866e-05, + "loss": 3.4466, + "step": 1701376 + }, + { + "epoch": 0.01, + "learning_rate": 2.2137116138166346e-05, + "loss": 3.4517, + "step": 1701888 + }, + { + "epoch": 0.01, + "learning_rate": 2.2128730190655826e-05, + "loss": 3.4549, + "step": 1702400 + }, + { + "epoch": 0.01, + "learning_rate": 2.2120360621949042e-05, + "loss": 3.4526, + "step": 1702912 + }, + { + "epoch": 0.01, + "learning_rate": 2.211197467443852e-05, + "loss": 3.4393, + "step": 1703424 + }, + { + "epoch": 0.01, + "learning_rate": 2.2103588726928e-05, + "loss": 3.4179, + "step": 1703936 + }, + { + "epoch": 0.01, + "learning_rate": 2.209520277941748e-05, + "loss": 3.4621, + "step": 1704448 + }, + { + "epoch": 0.01, + "learning_rate": 2.208683321071069e-05, + "loss": 3.4535, + "step": 1704960 + }, + { + "epoch": 0.01, + "learning_rate": 2.2078447263200168e-05, + "loss": 3.4493, + "step": 1705472 + }, + { + "epoch": 0.01, + "learning_rate": 2.207006131568965e-05, + "loss": 3.4407, + "step": 1705984 + }, + { + "epoch": 0.01, + "learning_rate": 2.206167536817913e-05, + "loss": 3.4432, + "step": 1706496 + }, + { + "epoch": 0.01, + "learning_rate": 2.205330579947234e-05, + "loss": 3.4318, + "step": 1707008 + }, + { + "epoch": 0.01, + "learning_rate": 2.204491985196182e-05, + "loss": 3.4399, + "step": 1707520 + }, + { + "epoch": 0.01, + "learning_rate": 2.20365339044513e-05, + "loss": 3.4386, + "step": 1708032 + }, + { + "epoch": 0.01, + "learning_rate": 2.202814795694078e-05, + "loss": 3.4444, + "step": 1708544 + }, + { + "epoch": 0.01, + "learning_rate": 2.201976200943026e-05, + "loss": 3.455, + "step": 1709056 + }, + { + "epoch": 0.01, + "learning_rate": 2.2011376061919743e-05, + "loss": 3.4357, + "step": 1709568 + }, + { + "epoch": 0.01, + "learning_rate": 2.2002990114409223e-05, + "loss": 3.4344, + "step": 1710080 + }, + { + "epoch": 0.01, + "learning_rate": 2.19946041668987e-05, + "loss": 3.4493, + "step": 1710592 + }, + { + "epoch": 0.01, + "learning_rate": 2.1986234598191912e-05, + "loss": 3.4351, + "step": 1711104 + }, + { + "epoch": 0.01, + "learning_rate": 2.1977848650681392e-05, + "loss": 3.4334, + "step": 1711616 + }, + { + "epoch": 0.01, + "learning_rate": 2.1969479081974605e-05, + "loss": 3.4465, + "step": 1712128 + }, + { + "epoch": 0.01, + "learning_rate": 2.1961093134464085e-05, + "loss": 3.4434, + "step": 1712640 + }, + { + "epoch": 0.01, + "learning_rate": 2.1952707186953565e-05, + "loss": 3.4347, + "step": 1713152 + }, + { + "epoch": 0.01, + "learning_rate": 2.1944321239443045e-05, + "loss": 3.4349, + "step": 1713664 + }, + { + "epoch": 0.01, + "learning_rate": 2.193593529193252e-05, + "loss": 3.4365, + "step": 1714176 + }, + { + "epoch": 0.01, + "learning_rate": 2.1927549344422e-05, + "loss": 3.4427, + "step": 1714688 + }, + { + "epoch": 0.01, + "learning_rate": 2.191916339691148e-05, + "loss": 3.4453, + "step": 1715200 + }, + { + "epoch": 0.01, + "learning_rate": 2.1910793828204697e-05, + "loss": 3.441, + "step": 1715712 + }, + { + "epoch": 0.01, + "learning_rate": 2.1902407880694173e-05, + "loss": 3.4439, + "step": 1716224 + }, + { + "epoch": 0.01, + "learning_rate": 2.1894038311987386e-05, + "loss": 3.4502, + "step": 1716736 + }, + { + "epoch": 0.01, + "learning_rate": 2.1885652364476866e-05, + "loss": 3.4499, + "step": 1717248 + }, + { + "epoch": 0.01, + "learning_rate": 2.1877266416966346e-05, + "loss": 3.4364, + "step": 1717760 + }, + { + "epoch": 0.01, + "learning_rate": 2.1868880469455822e-05, + "loss": 3.4497, + "step": 1718272 + }, + { + "epoch": 0.01, + "learning_rate": 2.1860494521945306e-05, + "loss": 3.4497, + "step": 1718784 + }, + { + "epoch": 0.01, + "learning_rate": 2.1852108574434786e-05, + "loss": 3.4309, + "step": 1719296 + }, + { + "epoch": 0.01, + "learning_rate": 2.1843722626924266e-05, + "loss": 3.4498, + "step": 1719808 + }, + { + "epoch": 0.01, + "learning_rate": 2.1835336679413746e-05, + "loss": 3.4439, + "step": 1720320 + }, + { + "epoch": 0.01, + "learning_rate": 2.1826983489510687e-05, + "loss": 3.4514, + "step": 1720832 + }, + { + "epoch": 0.01, + "learning_rate": 2.1818597542000167e-05, + "loss": 3.4405, + "step": 1721344 + }, + { + "epoch": 0.01, + "learning_rate": 2.1810211594489647e-05, + "loss": 3.4276, + "step": 1721856 + }, + { + "epoch": 0.01, + "learning_rate": 2.1801825646979127e-05, + "loss": 3.4362, + "step": 1722368 + }, + { + "epoch": 0.01, + "learning_rate": 2.179345607827234e-05, + "loss": 3.4371, + "step": 1722880 + }, + { + "epoch": 0.01, + "learning_rate": 2.1785070130761816e-05, + "loss": 3.4485, + "step": 1723392 + }, + { + "epoch": 0.01, + "learning_rate": 2.1776684183251296e-05, + "loss": 3.438, + "step": 1723904 + }, + { + "epoch": 0.01, + "learning_rate": 2.1768298235740776e-05, + "loss": 3.4383, + "step": 1724416 + }, + { + "epoch": 0.02, + "learning_rate": 2.175991228823026e-05, + "loss": 3.4417, + "step": 1724928 + }, + { + "epoch": 0.02, + "learning_rate": 2.175152634071974e-05, + "loss": 3.4478, + "step": 1725440 + }, + { + "epoch": 0.02, + "learning_rate": 2.174315677201295e-05, + "loss": 3.4313, + "step": 1725952 + }, + { + "epoch": 0.02, + "learning_rate": 2.173477082450243e-05, + "loss": 3.439, + "step": 1726464 + }, + { + "epoch": 0.02, + "learning_rate": 2.172638487699191e-05, + "loss": 3.4492, + "step": 1726976 + }, + { + "epoch": 0.02, + "learning_rate": 2.171799892948139e-05, + "loss": 3.4522, + "step": 1727488 + }, + { + "epoch": 0.02, + "learning_rate": 2.17096293607746e-05, + "loss": 3.4439, + "step": 1728000 + }, + { + "epoch": 0.02, + "learning_rate": 2.170124341326408e-05, + "loss": 3.4263, + "step": 1728512 + }, + { + "epoch": 0.02, + "learning_rate": 2.169285746575356e-05, + "loss": 3.4428, + "step": 1729024 + }, + { + "epoch": 0.02, + "learning_rate": 2.168447151824304e-05, + "loss": 3.436, + "step": 1729536 + }, + { + "epoch": 0.02, + "learning_rate": 2.167608557073252e-05, + "loss": 3.4484, + "step": 1730048 + }, + { + "epoch": 0.02, + "learning_rate": 2.1667699623221997e-05, + "loss": 3.4487, + "step": 1730560 + }, + { + "epoch": 0.02, + "learning_rate": 2.165931367571148e-05, + "loss": 3.4399, + "step": 1731072 + }, + { + "epoch": 0.02, + "learning_rate": 2.165092772820096e-05, + "loss": 3.4344, + "step": 1731584 + }, + { + "epoch": 0.02, + "learning_rate": 2.1642558159494173e-05, + "loss": 3.4439, + "step": 1732096 + }, + { + "epoch": 0.02, + "learning_rate": 2.1634188590787382e-05, + "loss": 3.4296, + "step": 1732608 + }, + { + "epoch": 0.02, + "learning_rate": 2.1625802643276862e-05, + "loss": 3.4444, + "step": 1733120 + }, + { + "epoch": 0.02, + "learning_rate": 2.1617416695766342e-05, + "loss": 3.4269, + "step": 1733632 + }, + { + "epoch": 0.02, + "learning_rate": 2.1609030748255822e-05, + "loss": 3.4473, + "step": 1734144 + }, + { + "epoch": 0.02, + "learning_rate": 2.1600644800745302e-05, + "loss": 3.4322, + "step": 1734656 + }, + { + "epoch": 0.02, + "learning_rate": 2.1592258853234782e-05, + "loss": 3.4429, + "step": 1735168 + }, + { + "epoch": 0.02, + "learning_rate": 2.1583872905724262e-05, + "loss": 3.4387, + "step": 1735680 + }, + { + "epoch": 0.02, + "learning_rate": 2.157550333701747e-05, + "loss": 3.4387, + "step": 1736192 + }, + { + "epoch": 0.02, + "learning_rate": 2.156711738950695e-05, + "loss": 3.4443, + "step": 1736704 + }, + { + "epoch": 0.02, + "learning_rate": 2.155873144199643e-05, + "loss": 3.4397, + "step": 1737216 + }, + { + "epoch": 0.02, + "learning_rate": 2.1550345494485914e-05, + "loss": 3.4543, + "step": 1737728 + }, + { + "epoch": 0.02, + "learning_rate": 2.1541975925779123e-05, + "loss": 3.4477, + "step": 1738240 + }, + { + "epoch": 0.02, + "learning_rate": 2.1533589978268603e-05, + "loss": 3.4405, + "step": 1738752 + }, + { + "epoch": 0.02, + "learning_rate": 2.1525204030758083e-05, + "loss": 3.4439, + "step": 1739264 + }, + { + "epoch": 0.02, + "learning_rate": 2.1516818083247563e-05, + "loss": 3.436, + "step": 1739776 + }, + { + "epoch": 0.02, + "learning_rate": 2.1508432135737043e-05, + "loss": 3.4411, + "step": 1740288 + }, + { + "epoch": 0.02, + "learning_rate": 2.1500046188226523e-05, + "loss": 3.4418, + "step": 1740800 + }, + { + "epoch": 0.02, + "learning_rate": 2.1491676619519736e-05, + "loss": 3.4392, + "step": 1741312 + }, + { + "epoch": 0.02, + "learning_rate": 2.1483290672009216e-05, + "loss": 3.4423, + "step": 1741824 + }, + { + "epoch": 0.02, + "learning_rate": 2.1474904724498696e-05, + "loss": 3.4392, + "step": 1742336 + }, + { + "epoch": 0.02, + "learning_rate": 2.1466518776988175e-05, + "loss": 3.4423, + "step": 1742848 + }, + { + "epoch": 0.02, + "learning_rate": 2.1458132829477652e-05, + "loss": 3.4429, + "step": 1743360 + }, + { + "epoch": 0.02, + "learning_rate": 2.1449746881967135e-05, + "loss": 3.4392, + "step": 1743872 + }, + { + "epoch": 0.02, + "learning_rate": 2.1441360934456615e-05, + "loss": 3.441, + "step": 1744384 + }, + { + "epoch": 0.02, + "learning_rate": 2.1432991365749828e-05, + "loss": 3.4323, + "step": 1744896 + }, + { + "epoch": 0.02, + "learning_rate": 2.1424621797043037e-05, + "loss": 3.4551, + "step": 1745408 + }, + { + "epoch": 0.02, + "learning_rate": 2.1416235849532517e-05, + "loss": 3.4391, + "step": 1745920 + }, + { + "epoch": 0.02, + "learning_rate": 2.1407849902021997e-05, + "loss": 3.452, + "step": 1746432 + }, + { + "epoch": 0.02, + "learning_rate": 2.1399463954511477e-05, + "loss": 3.4342, + "step": 1746944 + }, + { + "epoch": 0.02, + "learning_rate": 2.1391078007000957e-05, + "loss": 3.4421, + "step": 1747456 + }, + { + "epoch": 0.02, + "learning_rate": 2.1382692059490437e-05, + "loss": 3.4441, + "step": 1747968 + }, + { + "epoch": 0.02, + "learning_rate": 2.1374306111979917e-05, + "loss": 3.4454, + "step": 1748480 + }, + { + "epoch": 0.02, + "learning_rate": 2.1365936543273126e-05, + "loss": 3.4318, + "step": 1748992 + }, + { + "epoch": 0.02, + "learning_rate": 2.1357550595762606e-05, + "loss": 3.4496, + "step": 1749504 + }, + { + "epoch": 0.02, + "learning_rate": 2.134916464825209e-05, + "loss": 3.4425, + "step": 1750016 + }, + { + "epoch": 0.02, + "learning_rate": 2.134077870074157e-05, + "loss": 3.4588, + "step": 1750528 + }, + { + "epoch": 0.02, + "learning_rate": 2.133239275323105e-05, + "loss": 3.4357, + "step": 1751040 + }, + { + "epoch": 0.02, + "learning_rate": 2.132400680572053e-05, + "loss": 3.4497, + "step": 1751552 + }, + { + "epoch": 0.02, + "learning_rate": 2.1315637237013738e-05, + "loss": 3.4447, + "step": 1752064 + }, + { + "epoch": 0.02, + "learning_rate": 2.1307251289503218e-05, + "loss": 3.44, + "step": 1752576 + }, + { + "epoch": 0.02, + "learning_rate": 2.1298865341992698e-05, + "loss": 3.4404, + "step": 1753088 + }, + { + "epoch": 0.02, + "learning_rate": 2.1290479394482178e-05, + "loss": 3.445, + "step": 1753600 + }, + { + "epoch": 0.02, + "learning_rate": 2.128210982577539e-05, + "loss": 3.4478, + "step": 1754112 + }, + { + "epoch": 0.02, + "learning_rate": 2.127372387826487e-05, + "loss": 3.4252, + "step": 1754624 + }, + { + "epoch": 0.02, + "learning_rate": 2.126533793075435e-05, + "loss": 3.4474, + "step": 1755136 + }, + { + "epoch": 0.03, + "eval_loss": 3.8598225116729736, + "eval_runtime": 302.2434, + "eval_samples_per_second": 1262.529, + "eval_steps_per_second": 39.455, + "step": 1755360 + }, + { + "epoch": 1.0, + "learning_rate": 2.125695198324383e-05, + "loss": 3.4368, + "step": 1755648 + }, + { + "epoch": 1.0, + "learning_rate": 2.1248566035733307e-05, + "loss": 3.4329, + "step": 1756160 + }, + { + "epoch": 1.0, + "learning_rate": 2.124018008822279e-05, + "loss": 3.4437, + "step": 1756672 + }, + { + "epoch": 1.0, + "learning_rate": 2.123179414071227e-05, + "loss": 3.4539, + "step": 1757184 + }, + { + "epoch": 1.0, + "learning_rate": 2.122340819320175e-05, + "loss": 3.4477, + "step": 1757696 + }, + { + "epoch": 1.0, + "learning_rate": 2.121503862449496e-05, + "loss": 3.4395, + "step": 1758208 + }, + { + "epoch": 1.0, + "learning_rate": 2.120665267698444e-05, + "loss": 3.441, + "step": 1758720 + }, + { + "epoch": 1.0, + "learning_rate": 2.119826672947392e-05, + "loss": 3.4327, + "step": 1759232 + }, + { + "epoch": 1.0, + "learning_rate": 2.118989716076713e-05, + "loss": 3.4458, + "step": 1759744 + }, + { + "epoch": 1.0, + "learning_rate": 2.118151121325661e-05, + "loss": 3.443, + "step": 1760256 + }, + { + "epoch": 1.0, + "learning_rate": 2.117312526574609e-05, + "loss": 3.44, + "step": 1760768 + }, + { + "epoch": 1.0, + "learning_rate": 2.116473931823557e-05, + "loss": 3.4433, + "step": 1761280 + }, + { + "epoch": 1.0, + "learning_rate": 2.115635337072505e-05, + "loss": 3.4333, + "step": 1761792 + }, + { + "epoch": 1.0, + "learning_rate": 2.114798380201826e-05, + "loss": 3.4323, + "step": 1762304 + }, + { + "epoch": 1.0, + "learning_rate": 2.1139597854507744e-05, + "loss": 3.4362, + "step": 1762816 + }, + { + "epoch": 1.0, + "learning_rate": 2.1131211906997224e-05, + "loss": 3.4278, + "step": 1763328 + }, + { + "epoch": 1.0, + "learning_rate": 2.1122825959486704e-05, + "loss": 3.4364, + "step": 1763840 + }, + { + "epoch": 1.0, + "learning_rate": 2.1114440011976184e-05, + "loss": 3.4336, + "step": 1764352 + }, + { + "epoch": 1.0, + "learning_rate": 2.1106054064465664e-05, + "loss": 3.439, + "step": 1764864 + }, + { + "epoch": 1.0, + "learning_rate": 2.1097684495758873e-05, + "loss": 3.4505, + "step": 1765376 + }, + { + "epoch": 1.0, + "learning_rate": 2.1089298548248353e-05, + "loss": 3.45, + "step": 1765888 + }, + { + "epoch": 1.0, + "learning_rate": 2.1080912600737836e-05, + "loss": 3.4468, + "step": 1766400 + }, + { + "epoch": 1.0, + "learning_rate": 2.1072526653227313e-05, + "loss": 3.437, + "step": 1766912 + }, + { + "epoch": 1.0, + "learning_rate": 2.1064140705716793e-05, + "loss": 3.4433, + "step": 1767424 + }, + { + "epoch": 1.0, + "learning_rate": 2.1055754758206272e-05, + "loss": 3.4357, + "step": 1767936 + }, + { + "epoch": 1.0, + "learning_rate": 2.1047385189499485e-05, + "loss": 3.4449, + "step": 1768448 + }, + { + "epoch": 1.0, + "learning_rate": 2.103899924198896e-05, + "loss": 3.4382, + "step": 1768960 + }, + { + "epoch": 1.0, + "learning_rate": 2.1030613294478445e-05, + "loss": 3.4297, + "step": 1769472 + }, + { + "epoch": 1.0, + "learning_rate": 2.1022227346967925e-05, + "loss": 3.4362, + "step": 1769984 + }, + { + "epoch": 1.0, + "learning_rate": 2.1013857778261137e-05, + "loss": 3.4276, + "step": 1770496 + }, + { + "epoch": 1.01, + "learning_rate": 2.1005471830750614e-05, + "loss": 3.4349, + "step": 1771008 + }, + { + "epoch": 1.01, + "learning_rate": 2.0997085883240094e-05, + "loss": 3.4471, + "step": 1771520 + }, + { + "epoch": 1.01, + "learning_rate": 2.0988699935729574e-05, + "loss": 3.4403, + "step": 1772032 + }, + { + "epoch": 1.01, + "learning_rate": 2.0980330367022786e-05, + "loss": 3.442, + "step": 1772544 + }, + { + "epoch": 1.01, + "learning_rate": 2.0971944419512266e-05, + "loss": 3.4283, + "step": 1773056 + }, + { + "epoch": 1.01, + "learning_rate": 2.0963558472001746e-05, + "loss": 3.445, + "step": 1773568 + }, + { + "epoch": 1.01, + "learning_rate": 2.0955172524491226e-05, + "loss": 3.4365, + "step": 1774080 + }, + { + "epoch": 1.01, + "learning_rate": 2.0946786576980706e-05, + "loss": 3.427, + "step": 1774592 + }, + { + "epoch": 1.01, + "learning_rate": 2.0938417008273915e-05, + "loss": 3.4354, + "step": 1775104 + }, + { + "epoch": 1.01, + "learning_rate": 2.09300310607634e-05, + "loss": 3.4251, + "step": 1775616 + }, + { + "epoch": 1.01, + "learning_rate": 2.092164511325288e-05, + "loss": 3.4356, + "step": 1776128 + }, + { + "epoch": 1.01, + "learning_rate": 2.091325916574236e-05, + "loss": 3.4408, + "step": 1776640 + }, + { + "epoch": 1.01, + "learning_rate": 2.090487321823184e-05, + "loss": 3.4391, + "step": 1777152 + }, + { + "epoch": 1.01, + "learning_rate": 2.089648727072132e-05, + "loss": 3.433, + "step": 1777664 + }, + { + "epoch": 1.01, + "learning_rate": 2.0888117702014527e-05, + "loss": 3.4423, + "step": 1778176 + }, + { + "epoch": 1.01, + "learning_rate": 2.0879731754504007e-05, + "loss": 3.4417, + "step": 1778688 + }, + { + "epoch": 1.01, + "learning_rate": 2.087134580699349e-05, + "loss": 3.4415, + "step": 1779200 + }, + { + "epoch": 1.01, + "learning_rate": 2.0862959859482967e-05, + "loss": 3.4281, + "step": 1779712 + }, + { + "epoch": 1.01, + "learning_rate": 2.0854573911972447e-05, + "loss": 3.4087, + "step": 1780224 + }, + { + "epoch": 1.01, + "learning_rate": 2.084620434326566e-05, + "loss": 3.4528, + "step": 1780736 + }, + { + "epoch": 1.01, + "learning_rate": 2.083781839575514e-05, + "loss": 3.4388, + "step": 1781248 + }, + { + "epoch": 1.01, + "learning_rate": 2.0829432448244616e-05, + "loss": 3.44, + "step": 1781760 + }, + { + "epoch": 1.01, + "learning_rate": 2.08210465007341e-05, + "loss": 3.4337, + "step": 1782272 + }, + { + "epoch": 1.01, + "learning_rate": 2.081266055322358e-05, + "loss": 3.4273, + "step": 1782784 + }, + { + "epoch": 1.01, + "learning_rate": 2.080427460571306e-05, + "loss": 3.4251, + "step": 1783296 + }, + { + "epoch": 1.01, + "learning_rate": 2.079590503700627e-05, + "loss": 3.4306, + "step": 1783808 + }, + { + "epoch": 1.01, + "learning_rate": 2.078751908949575e-05, + "loss": 3.4247, + "step": 1784320 + }, + { + "epoch": 1.01, + "learning_rate": 2.077913314198523e-05, + "loss": 3.4311, + "step": 1784832 + }, + { + "epoch": 1.01, + "learning_rate": 2.077074719447471e-05, + "loss": 3.449, + "step": 1785344 + }, + { + "epoch": 1.01, + "learning_rate": 2.0762361246964192e-05, + "loss": 3.4253, + "step": 1785856 + }, + { + "epoch": 1.01, + "learning_rate": 2.0753975299453672e-05, + "loss": 3.4205, + "step": 1786368 + }, + { + "epoch": 1.01, + "learning_rate": 2.074558935194315e-05, + "loss": 3.4422, + "step": 1786880 + }, + { + "epoch": 1.01, + "learning_rate": 2.073720340443263e-05, + "loss": 3.4196, + "step": 1787392 + }, + { + "epoch": 1.01, + "learning_rate": 2.072883383572584e-05, + "loss": 3.4226, + "step": 1787904 + }, + { + "epoch": 1.01, + "learning_rate": 2.0720464267019053e-05, + "loss": 3.4364, + "step": 1788416 + }, + { + "epoch": 1.01, + "learning_rate": 2.0712078319508533e-05, + "loss": 3.4337, + "step": 1788928 + }, + { + "epoch": 1.01, + "learning_rate": 2.0703692371998013e-05, + "loss": 3.4223, + "step": 1789440 + }, + { + "epoch": 1.01, + "learning_rate": 2.0695306424487493e-05, + "loss": 3.4304, + "step": 1789952 + }, + { + "epoch": 1.01, + "learning_rate": 2.0686920476976973e-05, + "loss": 3.4222, + "step": 1790464 + }, + { + "epoch": 1.01, + "learning_rate": 2.067853452946645e-05, + "loss": 3.4322, + "step": 1790976 + }, + { + "epoch": 1.01, + "learning_rate": 2.0670164960759662e-05, + "loss": 3.4345, + "step": 1791488 + }, + { + "epoch": 1.01, + "learning_rate": 2.0661779013249146e-05, + "loss": 3.4283, + "step": 1792000 + }, + { + "epoch": 1.01, + "learning_rate": 2.0653393065738622e-05, + "loss": 3.4362, + "step": 1792512 + }, + { + "epoch": 1.01, + "learning_rate": 2.0645007118228102e-05, + "loss": 3.4378, + "step": 1793024 + }, + { + "epoch": 1.01, + "learning_rate": 2.0636621170717582e-05, + "loss": 3.4424, + "step": 1793536 + }, + { + "epoch": 1.01, + "learning_rate": 2.0628235223207062e-05, + "loss": 3.4244, + "step": 1794048 + }, + { + "epoch": 1.01, + "learning_rate": 2.0619849275696542e-05, + "loss": 3.4397, + "step": 1794560 + }, + { + "epoch": 1.01, + "learning_rate": 2.0611463328186022e-05, + "loss": 3.4403, + "step": 1795072 + }, + { + "epoch": 1.01, + "learning_rate": 2.0603093759479234e-05, + "loss": 3.4179, + "step": 1795584 + }, + { + "epoch": 1.01, + "learning_rate": 2.0594707811968714e-05, + "loss": 3.4411, + "step": 1796096 + }, + { + "epoch": 1.01, + "learning_rate": 2.0586338243261923e-05, + "loss": 3.4308, + "step": 1796608 + }, + { + "epoch": 1.01, + "learning_rate": 2.0577952295751403e-05, + "loss": 3.4423, + "step": 1797120 + }, + { + "epoch": 1.01, + "learning_rate": 2.0569566348240883e-05, + "loss": 3.435, + "step": 1797632 + }, + { + "epoch": 1.01, + "learning_rate": 2.0561180400730363e-05, + "loss": 3.4187, + "step": 1798144 + }, + { + "epoch": 1.01, + "learning_rate": 2.0552794453219847e-05, + "loss": 3.4186, + "step": 1798656 + }, + { + "epoch": 1.01, + "learning_rate": 2.0544424884513056e-05, + "loss": 3.4279, + "step": 1799168 + }, + { + "epoch": 1.01, + "learning_rate": 2.0536038937002536e-05, + "loss": 3.4391, + "step": 1799680 + }, + { + "epoch": 1.01, + "learning_rate": 2.0527652989492016e-05, + "loss": 3.4275, + "step": 1800192 + }, + { + "epoch": 1.01, + "learning_rate": 2.0519267041981496e-05, + "loss": 3.4288, + "step": 1800704 + }, + { + "epoch": 1.02, + "learning_rate": 2.0510897473274708e-05, + "loss": 3.431, + "step": 1801216 + }, + { + "epoch": 1.02, + "learning_rate": 2.0502511525764188e-05, + "loss": 3.4342, + "step": 1801728 + }, + { + "epoch": 1.02, + "learning_rate": 2.0494125578253668e-05, + "loss": 3.4256, + "step": 1802240 + }, + { + "epoch": 1.02, + "learning_rate": 2.0485739630743148e-05, + "loss": 3.4265, + "step": 1802752 + }, + { + "epoch": 1.02, + "learning_rate": 2.0477353683232625e-05, + "loss": 3.4349, + "step": 1803264 + }, + { + "epoch": 1.02, + "learning_rate": 2.0468984114525837e-05, + "loss": 3.4377, + "step": 1803776 + }, + { + "epoch": 1.02, + "learning_rate": 2.0460598167015317e-05, + "loss": 3.4358, + "step": 1804288 + }, + { + "epoch": 1.02, + "learning_rate": 2.04522122195048e-05, + "loss": 3.4175, + "step": 1804800 + }, + { + "epoch": 1.02, + "learning_rate": 2.0443826271994277e-05, + "loss": 3.433, + "step": 1805312 + }, + { + "epoch": 1.02, + "learning_rate": 2.0435440324483757e-05, + "loss": 3.4225, + "step": 1805824 + }, + { + "epoch": 1.02, + "learning_rate": 2.042707075577697e-05, + "loss": 3.4392, + "step": 1806336 + }, + { + "epoch": 1.02, + "learning_rate": 2.041868480826645e-05, + "loss": 3.4335, + "step": 1806848 + }, + { + "epoch": 1.02, + "learning_rate": 2.041029886075593e-05, + "loss": 3.4315, + "step": 1807360 + }, + { + "epoch": 1.02, + "learning_rate": 2.040191291324541e-05, + "loss": 3.4234, + "step": 1807872 + }, + { + "epoch": 1.02, + "learning_rate": 2.039352696573489e-05, + "loss": 3.4274, + "step": 1808384 + }, + { + "epoch": 1.02, + "learning_rate": 2.0385157397028098e-05, + "loss": 3.4226, + "step": 1808896 + }, + { + "epoch": 1.02, + "learning_rate": 2.0376771449517578e-05, + "loss": 3.432, + "step": 1809408 + }, + { + "epoch": 1.02, + "learning_rate": 2.0368385502007058e-05, + "loss": 3.4201, + "step": 1809920 + }, + { + "epoch": 1.02, + "learning_rate": 2.0359999554496538e-05, + "loss": 3.4377, + "step": 1810432 + }, + { + "epoch": 1.02, + "learning_rate": 2.0351613606986018e-05, + "loss": 3.4235, + "step": 1810944 + }, + { + "epoch": 1.02, + "learning_rate": 2.03432276594755e-05, + "loss": 3.4277, + "step": 1811456 + }, + { + "epoch": 1.02, + "learning_rate": 2.033484171196498e-05, + "loss": 3.4242, + "step": 1811968 + }, + { + "epoch": 1.02, + "learning_rate": 2.032647214325819e-05, + "loss": 3.4248, + "step": 1812480 + }, + { + "epoch": 1.02, + "learning_rate": 2.031808619574767e-05, + "loss": 3.4378, + "step": 1812992 + }, + { + "epoch": 1.02, + "learning_rate": 2.030970024823715e-05, + "loss": 3.4302, + "step": 1813504 + }, + { + "epoch": 1.02, + "learning_rate": 2.030131430072663e-05, + "loss": 3.4433, + "step": 1814016 + }, + { + "epoch": 1.02, + "learning_rate": 2.029292835321611e-05, + "loss": 3.4394, + "step": 1814528 + }, + { + "epoch": 1.02, + "learning_rate": 2.0284558784509323e-05, + "loss": 3.4288, + "step": 1815040 + }, + { + "epoch": 1.02, + "learning_rate": 2.0276172836998803e-05, + "loss": 3.4306, + "step": 1815552 + }, + { + "epoch": 1.02, + "learning_rate": 2.026778688948828e-05, + "loss": 3.4265, + "step": 1816064 + }, + { + "epoch": 1.02, + "learning_rate": 2.025940094197776e-05, + "loss": 3.4273, + "step": 1816576 + }, + { + "epoch": 1.02, + "learning_rate": 2.025101499446724e-05, + "loss": 3.4337, + "step": 1817088 + }, + { + "epoch": 1.02, + "learning_rate": 2.0242645425760455e-05, + "loss": 3.4265, + "step": 1817600 + }, + { + "epoch": 1.02, + "learning_rate": 2.023425947824993e-05, + "loss": 3.4305, + "step": 1818112 + }, + { + "epoch": 1.02, + "learning_rate": 2.022587353073941e-05, + "loss": 3.4299, + "step": 1818624 + }, + { + "epoch": 1.02, + "learning_rate": 2.021748758322889e-05, + "loss": 3.4323, + "step": 1819136 + }, + { + "epoch": 1.02, + "learning_rate": 2.020910163571837e-05, + "loss": 3.4335, + "step": 1819648 + }, + { + "epoch": 1.02, + "learning_rate": 2.020071568820785e-05, + "loss": 3.4266, + "step": 1820160 + }, + { + "epoch": 1.02, + "learning_rate": 2.019232974069733e-05, + "loss": 3.4298, + "step": 1820672 + }, + { + "epoch": 1.02, + "learning_rate": 2.0183943793186815e-05, + "loss": 3.4241, + "step": 1821184 + }, + { + "epoch": 1.02, + "learning_rate": 2.0175590603283753e-05, + "loss": 3.4432, + "step": 1821696 + }, + { + "epoch": 1.02, + "learning_rate": 2.0167204655773233e-05, + "loss": 3.4298, + "step": 1822208 + }, + { + "epoch": 1.02, + "learning_rate": 2.0158818708262713e-05, + "loss": 3.4379, + "step": 1822720 + }, + { + "epoch": 1.02, + "learning_rate": 2.0150432760752193e-05, + "loss": 3.4253, + "step": 1823232 + }, + { + "epoch": 1.02, + "learning_rate": 2.0142046813241676e-05, + "loss": 3.433, + "step": 1823744 + }, + { + "epoch": 1.02, + "learning_rate": 2.0133660865731156e-05, + "loss": 3.43, + "step": 1824256 + }, + { + "epoch": 1.02, + "learning_rate": 2.0125274918220636e-05, + "loss": 3.4329, + "step": 1824768 + }, + { + "epoch": 1.02, + "learning_rate": 2.0116905349513845e-05, + "loss": 3.4246, + "step": 1825280 + }, + { + "epoch": 1.02, + "learning_rate": 2.0108519402003325e-05, + "loss": 3.438, + "step": 1825792 + }, + { + "epoch": 1.02, + "learning_rate": 2.0100133454492805e-05, + "loss": 3.4319, + "step": 1826304 + }, + { + "epoch": 1.02, + "learning_rate": 2.0091747506982285e-05, + "loss": 3.4502, + "step": 1826816 + }, + { + "epoch": 1.02, + "learning_rate": 2.0083361559471765e-05, + "loss": 3.4249, + "step": 1827328 + }, + { + "epoch": 1.02, + "learning_rate": 2.0074975611961245e-05, + "loss": 3.4388, + "step": 1827840 + }, + { + "epoch": 1.02, + "learning_rate": 2.0066606043254457e-05, + "loss": 3.4347, + "step": 1828352 + }, + { + "epoch": 1.02, + "learning_rate": 2.0058220095743934e-05, + "loss": 3.4297, + "step": 1828864 + }, + { + "epoch": 1.02, + "learning_rate": 2.0049834148233414e-05, + "loss": 3.4295, + "step": 1829376 + }, + { + "epoch": 1.02, + "learning_rate": 2.0041448200722894e-05, + "loss": 3.435, + "step": 1829888 + }, + { + "epoch": 1.02, + "learning_rate": 2.003307863201611e-05, + "loss": 3.4377, + "step": 1830400 + }, + { + "epoch": 1.02, + "learning_rate": 2.0024692684505586e-05, + "loss": 3.4146, + "step": 1830912 + }, + { + "epoch": 1.02, + "learning_rate": 2.0016306736995066e-05, + "loss": 3.4365, + "step": 1831424 + }, + { + "epoch": 1.03, + "eval_loss": 3.860889434814453, + "eval_runtime": 302.2632, + "eval_samples_per_second": 1262.446, + "eval_steps_per_second": 39.452, + "step": 1831680 + }, + { + "epoch": 0.0, + "learning_rate": 2.0007920789484546e-05, + "loss": 3.4217, + "step": 1831936 + }, + { + "epoch": 0.0, + "learning_rate": 1.9999534841974026e-05, + "loss": 3.4225, + "step": 1832448 + }, + { + "epoch": 0.0, + "learning_rate": 1.999116527326724e-05, + "loss": 3.4308, + "step": 1832960 + }, + { + "epoch": 0.0, + "learning_rate": 1.998277932575672e-05, + "loss": 3.4446, + "step": 1833472 + }, + { + "epoch": 0.0, + "learning_rate": 1.99743933782462e-05, + "loss": 3.4388, + "step": 1833984 + }, + { + "epoch": 0.0, + "learning_rate": 1.996600743073568e-05, + "loss": 3.4269, + "step": 1834496 + }, + { + "epoch": 0.0, + "learning_rate": 1.995762148322516e-05, + "loss": 3.4347, + "step": 1835008 + }, + { + "epoch": 0.0, + "learning_rate": 1.994923553571464e-05, + "loss": 3.4194, + "step": 1835520 + }, + { + "epoch": 0.0, + "learning_rate": 1.9940849588204115e-05, + "loss": 3.4309, + "step": 1836032 + }, + { + "epoch": 0.0, + "learning_rate": 1.993248001949733e-05, + "loss": 3.4356, + "step": 1836544 + }, + { + "epoch": 0.0, + "learning_rate": 1.992409407198681e-05, + "loss": 3.429, + "step": 1837056 + }, + { + "epoch": 0.0, + "learning_rate": 1.991570812447629e-05, + "loss": 3.4301, + "step": 1837568 + }, + { + "epoch": 0.0, + "learning_rate": 1.9907322176965767e-05, + "loss": 3.4294, + "step": 1838080 + }, + { + "epoch": 0.0, + "learning_rate": 1.989895260825898e-05, + "loss": 3.4192, + "step": 1838592 + }, + { + "epoch": 0.0, + "learning_rate": 1.989056666074846e-05, + "loss": 3.4202, + "step": 1839104 + }, + { + "epoch": 0.0, + "learning_rate": 1.988218071323794e-05, + "loss": 3.4177, + "step": 1839616 + }, + { + "epoch": 0.0, + "learning_rate": 1.987379476572742e-05, + "loss": 3.424, + "step": 1840128 + }, + { + "epoch": 0.0, + "learning_rate": 1.98654088182169e-05, + "loss": 3.4279, + "step": 1840640 + }, + { + "epoch": 0.0, + "learning_rate": 1.985702287070638e-05, + "loss": 3.4242, + "step": 1841152 + }, + { + "epoch": 0.0, + "learning_rate": 1.984863692319586e-05, + "loss": 3.4385, + "step": 1841664 + }, + { + "epoch": 0.0, + "learning_rate": 1.984025097568534e-05, + "loss": 3.4394, + "step": 1842176 + }, + { + "epoch": 0.0, + "learning_rate": 1.9831897785782285e-05, + "loss": 3.4357, + "step": 1842688 + }, + { + "epoch": 0.0, + "learning_rate": 1.9823511838271765e-05, + "loss": 3.4292, + "step": 1843200 + }, + { + "epoch": 0.0, + "learning_rate": 1.981512589076124e-05, + "loss": 3.4309, + "step": 1843712 + }, + { + "epoch": 0.0, + "learning_rate": 1.980673994325072e-05, + "loss": 3.4318, + "step": 1844224 + }, + { + "epoch": 0.0, + "learning_rate": 1.97983539957402e-05, + "loss": 3.4302, + "step": 1844736 + }, + { + "epoch": 0.0, + "learning_rate": 1.978998442703341e-05, + "loss": 3.4277, + "step": 1845248 + }, + { + "epoch": 0.0, + "learning_rate": 1.9781598479522893e-05, + "loss": 3.4221, + "step": 1845760 + }, + { + "epoch": 0.0, + "learning_rate": 1.9773212532012373e-05, + "loss": 3.4241, + "step": 1846272 + }, + { + "epoch": 0.0, + "learning_rate": 1.9764826584501853e-05, + "loss": 3.4207, + "step": 1846784 + }, + { + "epoch": 0.01, + "learning_rate": 1.9756457015795062e-05, + "loss": 3.4219, + "step": 1847296 + }, + { + "epoch": 0.01, + "learning_rate": 1.9748071068284542e-05, + "loss": 3.4368, + "step": 1847808 + }, + { + "epoch": 0.01, + "learning_rate": 1.9739685120774022e-05, + "loss": 3.4329, + "step": 1848320 + }, + { + "epoch": 0.01, + "learning_rate": 1.9731299173263502e-05, + "loss": 3.4324, + "step": 1848832 + }, + { + "epoch": 0.01, + "learning_rate": 1.9722913225752986e-05, + "loss": 3.4205, + "step": 1849344 + }, + { + "epoch": 0.01, + "learning_rate": 1.9714527278242466e-05, + "loss": 3.4301, + "step": 1849856 + }, + { + "epoch": 0.01, + "learning_rate": 1.9706141330731946e-05, + "loss": 3.4272, + "step": 1850368 + }, + { + "epoch": 0.01, + "learning_rate": 1.9697755383221422e-05, + "loss": 3.4192, + "step": 1850880 + }, + { + "epoch": 0.01, + "learning_rate": 1.9689402193318364e-05, + "loss": 3.424, + "step": 1851392 + }, + { + "epoch": 0.01, + "learning_rate": 1.9681016245807847e-05, + "loss": 3.4148, + "step": 1851904 + }, + { + "epoch": 0.01, + "learning_rate": 1.9672630298297327e-05, + "loss": 3.4224, + "step": 1852416 + }, + { + "epoch": 0.01, + "learning_rate": 1.9664244350786807e-05, + "loss": 3.4335, + "step": 1852928 + }, + { + "epoch": 0.01, + "learning_rate": 1.9655858403276287e-05, + "loss": 3.4258, + "step": 1853440 + }, + { + "epoch": 0.01, + "learning_rate": 1.9647472455765767e-05, + "loss": 3.4244, + "step": 1853952 + }, + { + "epoch": 0.01, + "learning_rate": 1.9639102887058976e-05, + "loss": 3.4334, + "step": 1854464 + }, + { + "epoch": 0.01, + "learning_rate": 1.9630716939548456e-05, + "loss": 3.4293, + "step": 1854976 + }, + { + "epoch": 0.01, + "learning_rate": 1.962233099203794e-05, + "loss": 3.4311, + "step": 1855488 + }, + { + "epoch": 0.01, + "learning_rate": 1.961394504452742e-05, + "loss": 3.4229, + "step": 1856000 + }, + { + "epoch": 0.01, + "learning_rate": 1.9605559097016896e-05, + "loss": 3.3955, + "step": 1856512 + }, + { + "epoch": 0.01, + "learning_rate": 1.9597173149506376e-05, + "loss": 3.4394, + "step": 1857024 + }, + { + "epoch": 0.01, + "learning_rate": 1.9588787201995856e-05, + "loss": 3.4276, + "step": 1857536 + }, + { + "epoch": 0.01, + "learning_rate": 1.9580417633289065e-05, + "loss": 3.4323, + "step": 1858048 + }, + { + "epoch": 0.01, + "learning_rate": 1.9572031685778548e-05, + "loss": 3.4247, + "step": 1858560 + }, + { + "epoch": 0.01, + "learning_rate": 1.9563645738268028e-05, + "loss": 3.4182, + "step": 1859072 + }, + { + "epoch": 0.01, + "learning_rate": 1.9555259790757508e-05, + "loss": 3.4116, + "step": 1859584 + }, + { + "epoch": 0.01, + "learning_rate": 1.9546890222050717e-05, + "loss": 3.4205, + "step": 1860096 + }, + { + "epoch": 0.01, + "learning_rate": 1.9538504274540197e-05, + "loss": 3.4153, + "step": 1860608 + }, + { + "epoch": 0.01, + "learning_rate": 1.9530118327029677e-05, + "loss": 3.4199, + "step": 1861120 + }, + { + "epoch": 0.01, + "learning_rate": 1.9521732379519157e-05, + "loss": 3.4354, + "step": 1861632 + }, + { + "epoch": 0.01, + "learning_rate": 1.951334643200864e-05, + "loss": 3.4172, + "step": 1862144 + }, + { + "epoch": 0.01, + "learning_rate": 1.950496048449812e-05, + "loss": 3.4099, + "step": 1862656 + }, + { + "epoch": 0.01, + "learning_rate": 1.94965745369876e-05, + "loss": 3.4341, + "step": 1863168 + }, + { + "epoch": 0.01, + "learning_rate": 1.9488188589477077e-05, + "loss": 3.4064, + "step": 1863680 + }, + { + "epoch": 0.01, + "learning_rate": 1.947981902077029e-05, + "loss": 3.4107, + "step": 1864192 + }, + { + "epoch": 0.01, + "learning_rate": 1.9471449452063502e-05, + "loss": 3.4288, + "step": 1864704 + }, + { + "epoch": 0.01, + "learning_rate": 1.9463063504552982e-05, + "loss": 3.4247, + "step": 1865216 + }, + { + "epoch": 0.01, + "learning_rate": 1.9454677557042462e-05, + "loss": 3.4111, + "step": 1865728 + }, + { + "epoch": 0.01, + "learning_rate": 1.9446291609531942e-05, + "loss": 3.4176, + "step": 1866240 + }, + { + "epoch": 0.01, + "learning_rate": 1.9437905662021422e-05, + "loss": 3.4138, + "step": 1866752 + }, + { + "epoch": 0.01, + "learning_rate": 1.9429519714510898e-05, + "loss": 3.4214, + "step": 1867264 + }, + { + "epoch": 0.01, + "learning_rate": 1.942115014580411e-05, + "loss": 3.4234, + "step": 1867776 + }, + { + "epoch": 0.01, + "learning_rate": 1.9412764198293594e-05, + "loss": 3.4174, + "step": 1868288 + }, + { + "epoch": 0.01, + "learning_rate": 1.9404378250783074e-05, + "loss": 3.4243, + "step": 1868800 + }, + { + "epoch": 0.01, + "learning_rate": 1.939599230327255e-05, + "loss": 3.426, + "step": 1869312 + }, + { + "epoch": 0.01, + "learning_rate": 1.938760635576203e-05, + "loss": 3.4319, + "step": 1869824 + }, + { + "epoch": 0.01, + "learning_rate": 1.937922040825151e-05, + "loss": 3.4151, + "step": 1870336 + }, + { + "epoch": 0.01, + "learning_rate": 1.937083446074099e-05, + "loss": 3.4283, + "step": 1870848 + }, + { + "epoch": 0.01, + "learning_rate": 1.936244851323047e-05, + "loss": 3.4282, + "step": 1871360 + }, + { + "epoch": 0.01, + "learning_rate": 1.9354078944523683e-05, + "loss": 3.4083, + "step": 1871872 + }, + { + "epoch": 0.01, + "learning_rate": 1.9345692997013163e-05, + "loss": 3.4273, + "step": 1872384 + }, + { + "epoch": 0.01, + "learning_rate": 1.9337307049502643e-05, + "loss": 3.424, + "step": 1872896 + }, + { + "epoch": 0.01, + "learning_rate": 1.9328937480795852e-05, + "loss": 3.432, + "step": 1873408 + }, + { + "epoch": 0.01, + "learning_rate": 1.9320551533285332e-05, + "loss": 3.4232, + "step": 1873920 + }, + { + "epoch": 0.01, + "learning_rate": 1.9312165585774812e-05, + "loss": 3.4104, + "step": 1874432 + }, + { + "epoch": 0.01, + "learning_rate": 1.9303779638264295e-05, + "loss": 3.4047, + "step": 1874944 + }, + { + "epoch": 0.01, + "learning_rate": 1.9295410069557504e-05, + "loss": 3.4185, + "step": 1875456 + }, + { + "epoch": 0.01, + "learning_rate": 1.9287024122046984e-05, + "loss": 3.4262, + "step": 1875968 + }, + { + "epoch": 0.01, + "learning_rate": 1.9278638174536464e-05, + "loss": 3.4178, + "step": 1876480 + }, + { + "epoch": 0.01, + "learning_rate": 1.9270252227025944e-05, + "loss": 3.4157, + "step": 1876992 + }, + { + "epoch": 0.02, + "learning_rate": 1.9261882658319157e-05, + "loss": 3.4179, + "step": 1877504 + }, + { + "epoch": 0.02, + "learning_rate": 1.9253496710808637e-05, + "loss": 3.429, + "step": 1878016 + }, + { + "epoch": 0.02, + "learning_rate": 1.9245110763298117e-05, + "loss": 3.4134, + "step": 1878528 + }, + { + "epoch": 0.02, + "learning_rate": 1.9236724815787597e-05, + "loss": 3.417, + "step": 1879040 + }, + { + "epoch": 0.02, + "learning_rate": 1.9228338868277076e-05, + "loss": 3.4236, + "step": 1879552 + }, + { + "epoch": 0.02, + "learning_rate": 1.9219969299570286e-05, + "loss": 3.4298, + "step": 1880064 + }, + { + "epoch": 0.02, + "learning_rate": 1.9211583352059766e-05, + "loss": 3.4235, + "step": 1880576 + }, + { + "epoch": 0.02, + "learning_rate": 1.920319740454925e-05, + "loss": 3.4083, + "step": 1881088 + }, + { + "epoch": 0.02, + "learning_rate": 1.9194811457038725e-05, + "loss": 3.4175, + "step": 1881600 + }, + { + "epoch": 0.02, + "learning_rate": 1.9186441888331938e-05, + "loss": 3.4186, + "step": 1882112 + }, + { + "epoch": 0.02, + "learning_rate": 1.9178055940821418e-05, + "loss": 3.4297, + "step": 1882624 + }, + { + "epoch": 0.02, + "learning_rate": 1.9169669993310898e-05, + "loss": 3.4228, + "step": 1883136 + }, + { + "epoch": 0.02, + "learning_rate": 1.9161284045800378e-05, + "loss": 3.4229, + "step": 1883648 + }, + { + "epoch": 0.02, + "learning_rate": 1.915291447709359e-05, + "loss": 3.4148, + "step": 1884160 + }, + { + "epoch": 0.02, + "learning_rate": 1.914452852958307e-05, + "loss": 3.4138, + "step": 1884672 + }, + { + "epoch": 0.02, + "learning_rate": 1.913614258207255e-05, + "loss": 3.4183, + "step": 1885184 + }, + { + "epoch": 0.02, + "learning_rate": 1.9127756634562027e-05, + "loss": 3.4192, + "step": 1885696 + }, + { + "epoch": 0.02, + "learning_rate": 1.9119370687051507e-05, + "loss": 3.4063, + "step": 1886208 + }, + { + "epoch": 0.02, + "learning_rate": 1.911100111834472e-05, + "loss": 3.4278, + "step": 1886720 + }, + { + "epoch": 0.02, + "learning_rate": 1.91026151708342e-05, + "loss": 3.4134, + "step": 1887232 + }, + { + "epoch": 0.02, + "learning_rate": 1.909422922332368e-05, + "loss": 3.4163, + "step": 1887744 + }, + { + "epoch": 0.02, + "learning_rate": 1.908584327581316e-05, + "loss": 3.4135, + "step": 1888256 + }, + { + "epoch": 0.02, + "learning_rate": 1.907745732830264e-05, + "loss": 3.412, + "step": 1888768 + }, + { + "epoch": 0.02, + "learning_rate": 1.9069087759595848e-05, + "loss": 3.4284, + "step": 1889280 + }, + { + "epoch": 0.02, + "learning_rate": 1.906070181208533e-05, + "loss": 3.4142, + "step": 1889792 + }, + { + "epoch": 0.02, + "learning_rate": 1.905231586457481e-05, + "loss": 3.4362, + "step": 1890304 + }, + { + "epoch": 0.02, + "learning_rate": 1.904392991706429e-05, + "loss": 3.4285, + "step": 1890816 + }, + { + "epoch": 0.02, + "learning_rate": 1.90355603483575e-05, + "loss": 3.4207, + "step": 1891328 + }, + { + "epoch": 0.02, + "learning_rate": 1.902717440084698e-05, + "loss": 3.4192, + "step": 1891840 + }, + { + "epoch": 0.02, + "learning_rate": 1.901878845333646e-05, + "loss": 3.4213, + "step": 1892352 + }, + { + "epoch": 0.02, + "learning_rate": 1.901040250582594e-05, + "loss": 3.4164, + "step": 1892864 + }, + { + "epoch": 0.02, + "learning_rate": 1.900201655831542e-05, + "loss": 3.4258, + "step": 1893376 + }, + { + "epoch": 0.02, + "learning_rate": 1.8993646989608633e-05, + "loss": 3.4138, + "step": 1893888 + }, + { + "epoch": 0.02, + "learning_rate": 1.8985261042098113e-05, + "loss": 3.4223, + "step": 1894400 + }, + { + "epoch": 0.02, + "learning_rate": 1.8976875094587593e-05, + "loss": 3.417, + "step": 1894912 + }, + { + "epoch": 0.02, + "learning_rate": 1.8968489147077073e-05, + "loss": 3.4204, + "step": 1895424 + }, + { + "epoch": 0.02, + "learning_rate": 1.8960103199566553e-05, + "loss": 3.4212, + "step": 1895936 + }, + { + "epoch": 0.02, + "learning_rate": 1.8951717252056033e-05, + "loss": 3.4153, + "step": 1896448 + }, + { + "epoch": 0.02, + "learning_rate": 1.8943331304545513e-05, + "loss": 3.4225, + "step": 1896960 + }, + { + "epoch": 0.02, + "learning_rate": 1.8934945357034992e-05, + "loss": 3.4149, + "step": 1897472 + }, + { + "epoch": 0.02, + "learning_rate": 1.8926592167131934e-05, + "loss": 3.4327, + "step": 1897984 + }, + { + "epoch": 0.02, + "learning_rate": 1.8918206219621414e-05, + "loss": 3.4187, + "step": 1898496 + }, + { + "epoch": 0.02, + "learning_rate": 1.8909820272110894e-05, + "loss": 3.428, + "step": 1899008 + }, + { + "epoch": 0.02, + "learning_rate": 1.8901434324600374e-05, + "loss": 3.414, + "step": 1899520 + }, + { + "epoch": 0.02, + "learning_rate": 1.8893048377089854e-05, + "loss": 3.4287, + "step": 1900032 + }, + { + "epoch": 0.02, + "learning_rate": 1.8884662429579334e-05, + "loss": 3.4185, + "step": 1900544 + }, + { + "epoch": 0.02, + "learning_rate": 1.8876276482068814e-05, + "loss": 3.4225, + "step": 1901056 + }, + { + "epoch": 0.02, + "learning_rate": 1.8867890534558294e-05, + "loss": 3.414, + "step": 1901568 + }, + { + "epoch": 0.02, + "learning_rate": 1.8859520965851503e-05, + "loss": 3.4254, + "step": 1902080 + }, + { + "epoch": 0.02, + "learning_rate": 1.8851135018340986e-05, + "loss": 3.423, + "step": 1902592 + }, + { + "epoch": 0.02, + "learning_rate": 1.8842749070830466e-05, + "loss": 3.4409, + "step": 1903104 + }, + { + "epoch": 0.02, + "learning_rate": 1.8834363123319946e-05, + "loss": 3.4153, + "step": 1903616 + }, + { + "epoch": 0.02, + "learning_rate": 1.8825977175809426e-05, + "loss": 3.426, + "step": 1904128 + }, + { + "epoch": 0.02, + "learning_rate": 1.8817591228298906e-05, + "loss": 3.4241, + "step": 1904640 + }, + { + "epoch": 0.02, + "learning_rate": 1.8809221659592115e-05, + "loss": 3.4234, + "step": 1905152 + }, + { + "epoch": 0.02, + "learning_rate": 1.8800835712081595e-05, + "loss": 3.4144, + "step": 1905664 + }, + { + "epoch": 0.02, + "learning_rate": 1.8792449764571075e-05, + "loss": 3.4247, + "step": 1906176 + }, + { + "epoch": 0.02, + "learning_rate": 1.878406381706056e-05, + "loss": 3.4244, + "step": 1906688 + }, + { + "epoch": 0.02, + "learning_rate": 1.8775694248353768e-05, + "loss": 3.4134, + "step": 1907200 + }, + { + "epoch": 0.02, + "learning_rate": 1.8767308300843247e-05, + "loss": 3.4218, + "step": 1907712 + }, + { + "epoch": 0.03, + "eval_loss": 3.862840414047241, + "eval_runtime": 305.3605, + "eval_samples_per_second": 1249.641, + "eval_steps_per_second": 39.052, + "step": 1908000 + }, + { + "epoch": 1.0, + "learning_rate": 1.8758922353332727e-05, + "loss": 3.4121, + "step": 1908224 + }, + { + "epoch": 1.0, + "learning_rate": 1.8750536405822207e-05, + "loss": 3.4101, + "step": 1908736 + }, + { + "epoch": 1.0, + "learning_rate": 1.874216683711542e-05, + "loss": 3.4204, + "step": 1909248 + }, + { + "epoch": 1.0, + "learning_rate": 1.87337808896049e-05, + "loss": 3.4354, + "step": 1909760 + }, + { + "epoch": 1.0, + "learning_rate": 1.872539494209438e-05, + "loss": 3.4296, + "step": 1910272 + }, + { + "epoch": 1.0, + "learning_rate": 1.8717008994583856e-05, + "loss": 3.4132, + "step": 1910784 + }, + { + "epoch": 1.0, + "learning_rate": 1.8708623047073336e-05, + "loss": 3.4262, + "step": 1911296 + }, + { + "epoch": 1.0, + "learning_rate": 1.8700237099562816e-05, + "loss": 3.4095, + "step": 1911808 + }, + { + "epoch": 1.0, + "learning_rate": 1.8691851152052296e-05, + "loss": 3.4189, + "step": 1912320 + }, + { + "epoch": 1.0, + "learning_rate": 1.868348158334551e-05, + "loss": 3.4294, + "step": 1912832 + }, + { + "epoch": 1.0, + "learning_rate": 1.867509563583499e-05, + "loss": 3.4148, + "step": 1913344 + }, + { + "epoch": 1.0, + "learning_rate": 1.866670968832447e-05, + "loss": 3.4185, + "step": 1913856 + }, + { + "epoch": 1.0, + "learning_rate": 1.865832374081395e-05, + "loss": 3.4232, + "step": 1914368 + }, + { + "epoch": 1.0, + "learning_rate": 1.8649954172107158e-05, + "loss": 3.4104, + "step": 1914880 + }, + { + "epoch": 1.0, + "learning_rate": 1.864156822459664e-05, + "loss": 3.4084, + "step": 1915392 + }, + { + "epoch": 1.0, + "learning_rate": 1.863318227708612e-05, + "loss": 3.4068, + "step": 1915904 + }, + { + "epoch": 1.0, + "learning_rate": 1.86247963295756e-05, + "loss": 3.4122, + "step": 1916416 + }, + { + "epoch": 1.0, + "learning_rate": 1.861641038206508e-05, + "loss": 3.4164, + "step": 1916928 + }, + { + "epoch": 1.0, + "learning_rate": 1.860802443455456e-05, + "loss": 3.4138, + "step": 1917440 + }, + { + "epoch": 1.0, + "learning_rate": 1.8599638487044037e-05, + "loss": 3.4285, + "step": 1917952 + }, + { + "epoch": 1.0, + "learning_rate": 1.859126891833725e-05, + "loss": 3.4289, + "step": 1918464 + }, + { + "epoch": 1.0, + "learning_rate": 1.8582882970826733e-05, + "loss": 3.4259, + "step": 1918976 + }, + { + "epoch": 1.0, + "learning_rate": 1.8574497023316213e-05, + "loss": 3.4196, + "step": 1919488 + }, + { + "epoch": 1.0, + "learning_rate": 1.856611107580569e-05, + "loss": 3.4164, + "step": 1920000 + }, + { + "epoch": 1.0, + "learning_rate": 1.8557741507098902e-05, + "loss": 3.4226, + "step": 1920512 + }, + { + "epoch": 1.0, + "learning_rate": 1.8549355559588382e-05, + "loss": 3.4186, + "step": 1921024 + }, + { + "epoch": 1.0, + "learning_rate": 1.8540969612077862e-05, + "loss": 3.4168, + "step": 1921536 + }, + { + "epoch": 1.0, + "learning_rate": 1.8532583664567342e-05, + "loss": 3.4142, + "step": 1922048 + }, + { + "epoch": 1.0, + "learning_rate": 1.8524197717056822e-05, + "loss": 3.4131, + "step": 1922560 + }, + { + "epoch": 1.0, + "learning_rate": 1.8515828148350035e-05, + "loss": 3.411, + "step": 1923072 + }, + { + "epoch": 1.01, + "learning_rate": 1.850744220083951e-05, + "loss": 3.4128, + "step": 1923584 + }, + { + "epoch": 1.01, + "learning_rate": 1.849905625332899e-05, + "loss": 3.4211, + "step": 1924096 + }, + { + "epoch": 1.01, + "learning_rate": 1.849067030581847e-05, + "loss": 3.4262, + "step": 1924608 + }, + { + "epoch": 1.01, + "learning_rate": 1.848228435830795e-05, + "loss": 3.4188, + "step": 1925120 + }, + { + "epoch": 1.01, + "learning_rate": 1.8473914789601163e-05, + "loss": 3.4144, + "step": 1925632 + }, + { + "epoch": 1.01, + "learning_rate": 1.8465528842090643e-05, + "loss": 3.4109, + "step": 1926144 + }, + { + "epoch": 1.01, + "learning_rate": 1.8457142894580123e-05, + "loss": 3.4233, + "step": 1926656 + }, + { + "epoch": 1.01, + "learning_rate": 1.8448756947069603e-05, + "loss": 3.4097, + "step": 1927168 + }, + { + "epoch": 1.01, + "learning_rate": 1.8440387378362812e-05, + "loss": 3.4143, + "step": 1927680 + }, + { + "epoch": 1.01, + "learning_rate": 1.8432001430852296e-05, + "loss": 3.4028, + "step": 1928192 + }, + { + "epoch": 1.01, + "learning_rate": 1.8423615483341776e-05, + "loss": 3.4162, + "step": 1928704 + }, + { + "epoch": 1.01, + "learning_rate": 1.8415229535831256e-05, + "loss": 3.4163, + "step": 1929216 + }, + { + "epoch": 1.01, + "learning_rate": 1.8406843588320736e-05, + "loss": 3.4159, + "step": 1929728 + }, + { + "epoch": 1.01, + "learning_rate": 1.8398457640810216e-05, + "loss": 3.4173, + "step": 1930240 + }, + { + "epoch": 1.01, + "learning_rate": 1.8390088072103425e-05, + "loss": 3.4204, + "step": 1930752 + }, + { + "epoch": 1.01, + "learning_rate": 1.8381702124592905e-05, + "loss": 3.4207, + "step": 1931264 + }, + { + "epoch": 1.01, + "learning_rate": 1.8373316177082388e-05, + "loss": 3.4197, + "step": 1931776 + }, + { + "epoch": 1.01, + "learning_rate": 1.8364930229571868e-05, + "loss": 3.4151, + "step": 1932288 + }, + { + "epoch": 1.01, + "learning_rate": 1.8356544282061344e-05, + "loss": 3.3863, + "step": 1932800 + }, + { + "epoch": 1.01, + "learning_rate": 1.8348174713354557e-05, + "loss": 3.4281, + "step": 1933312 + }, + { + "epoch": 1.01, + "learning_rate": 1.8339788765844037e-05, + "loss": 3.416, + "step": 1933824 + }, + { + "epoch": 1.01, + "learning_rate": 1.8331402818333517e-05, + "loss": 3.4226, + "step": 1934336 + }, + { + "epoch": 1.01, + "learning_rate": 1.8323016870822997e-05, + "loss": 3.4137, + "step": 1934848 + }, + { + "epoch": 1.01, + "learning_rate": 1.8314630923312477e-05, + "loss": 3.4076, + "step": 1935360 + }, + { + "epoch": 1.01, + "learning_rate": 1.830626135460569e-05, + "loss": 3.4016, + "step": 1935872 + }, + { + "epoch": 1.01, + "learning_rate": 1.8297875407095166e-05, + "loss": 3.4087, + "step": 1936384 + }, + { + "epoch": 1.01, + "learning_rate": 1.8289489459584646e-05, + "loss": 3.4092, + "step": 1936896 + }, + { + "epoch": 1.01, + "learning_rate": 1.8281103512074126e-05, + "loss": 3.4067, + "step": 1937408 + }, + { + "epoch": 1.01, + "learning_rate": 1.8272717564563606e-05, + "loss": 3.4234, + "step": 1937920 + }, + { + "epoch": 1.01, + "learning_rate": 1.8264347995856818e-05, + "loss": 3.4121, + "step": 1938432 + }, + { + "epoch": 1.01, + "learning_rate": 1.8255962048346298e-05, + "loss": 3.3945, + "step": 1938944 + }, + { + "epoch": 1.01, + "learning_rate": 1.8247576100835778e-05, + "loss": 3.4274, + "step": 1939456 + }, + { + "epoch": 1.01, + "learning_rate": 1.8239190153325258e-05, + "loss": 3.3973, + "step": 1939968 + }, + { + "epoch": 1.01, + "learning_rate": 1.8230804205814738e-05, + "loss": 3.3999, + "step": 1940480 + }, + { + "epoch": 1.01, + "learning_rate": 1.822243463710795e-05, + "loss": 3.4186, + "step": 1940992 + }, + { + "epoch": 1.01, + "learning_rate": 1.821404868959743e-05, + "loss": 3.4121, + "step": 1941504 + }, + { + "epoch": 1.01, + "learning_rate": 1.820566274208691e-05, + "loss": 3.4049, + "step": 1942016 + }, + { + "epoch": 1.01, + "learning_rate": 1.819727679457639e-05, + "loss": 3.4041, + "step": 1942528 + }, + { + "epoch": 1.01, + "learning_rate": 1.818889084706587e-05, + "loss": 3.4055, + "step": 1943040 + }, + { + "epoch": 1.01, + "learning_rate": 1.8180504899555347e-05, + "loss": 3.4102, + "step": 1943552 + }, + { + "epoch": 1.01, + "learning_rate": 1.8172118952044827e-05, + "loss": 3.4113, + "step": 1944064 + }, + { + "epoch": 1.01, + "learning_rate": 1.8163749383338043e-05, + "loss": 3.4106, + "step": 1944576 + }, + { + "epoch": 1.01, + "learning_rate": 1.8155363435827523e-05, + "loss": 3.4124, + "step": 1945088 + }, + { + "epoch": 1.01, + "learning_rate": 1.8146977488317e-05, + "loss": 3.4147, + "step": 1945600 + }, + { + "epoch": 1.01, + "learning_rate": 1.813859154080648e-05, + "loss": 3.4249, + "step": 1946112 + }, + { + "epoch": 1.01, + "learning_rate": 1.813020559329596e-05, + "loss": 3.404, + "step": 1946624 + }, + { + "epoch": 1.01, + "learning_rate": 1.812181964578544e-05, + "loss": 3.4178, + "step": 1947136 + }, + { + "epoch": 1.01, + "learning_rate": 1.811343369827492e-05, + "loss": 3.4209, + "step": 1947648 + }, + { + "epoch": 1.01, + "learning_rate": 1.81050477507644e-05, + "loss": 3.3969, + "step": 1948160 + }, + { + "epoch": 1.01, + "learning_rate": 1.809667818205761e-05, + "loss": 3.415, + "step": 1948672 + }, + { + "epoch": 1.01, + "learning_rate": 1.808829223454709e-05, + "loss": 3.4171, + "step": 1949184 + }, + { + "epoch": 1.01, + "learning_rate": 1.80799226658403e-05, + "loss": 3.4204, + "step": 1949696 + }, + { + "epoch": 1.01, + "learning_rate": 1.807153671832978e-05, + "loss": 3.4177, + "step": 1950208 + }, + { + "epoch": 1.01, + "learning_rate": 1.806315077081926e-05, + "loss": 3.3966, + "step": 1950720 + }, + { + "epoch": 1.01, + "learning_rate": 1.8054764823308744e-05, + "loss": 3.3922, + "step": 1951232 + }, + { + "epoch": 1.01, + "learning_rate": 1.8046395254601953e-05, + "loss": 3.4117, + "step": 1951744 + }, + { + "epoch": 1.01, + "learning_rate": 1.8038009307091433e-05, + "loss": 3.4146, + "step": 1952256 + }, + { + "epoch": 1.01, + "learning_rate": 1.8029623359580913e-05, + "loss": 3.4043, + "step": 1952768 + }, + { + "epoch": 1.01, + "learning_rate": 1.8021237412070393e-05, + "loss": 3.4082, + "step": 1953280 + }, + { + "epoch": 1.02, + "learning_rate": 1.8012867843363605e-05, + "loss": 3.4058, + "step": 1953792 + }, + { + "epoch": 1.02, + "learning_rate": 1.8004481895853085e-05, + "loss": 3.4171, + "step": 1954304 + }, + { + "epoch": 1.02, + "learning_rate": 1.7996095948342565e-05, + "loss": 3.4054, + "step": 1954816 + }, + { + "epoch": 1.02, + "learning_rate": 1.7987710000832045e-05, + "loss": 3.4063, + "step": 1955328 + }, + { + "epoch": 1.02, + "learning_rate": 1.7979324053321525e-05, + "loss": 3.4082, + "step": 1955840 + }, + { + "epoch": 1.02, + "learning_rate": 1.7970954484614734e-05, + "loss": 3.4186, + "step": 1956352 + }, + { + "epoch": 1.02, + "learning_rate": 1.7962568537104214e-05, + "loss": 3.4144, + "step": 1956864 + }, + { + "epoch": 1.02, + "learning_rate": 1.7954182589593697e-05, + "loss": 3.4009, + "step": 1957376 + }, + { + "epoch": 1.02, + "learning_rate": 1.7945796642083177e-05, + "loss": 3.4061, + "step": 1957888 + }, + { + "epoch": 1.02, + "learning_rate": 1.7937410694572654e-05, + "loss": 3.4082, + "step": 1958400 + }, + { + "epoch": 1.02, + "learning_rate": 1.7929041125865867e-05, + "loss": 3.4187, + "step": 1958912 + }, + { + "epoch": 1.02, + "learning_rate": 1.7920655178355346e-05, + "loss": 3.4092, + "step": 1959424 + }, + { + "epoch": 1.02, + "learning_rate": 1.7912269230844826e-05, + "loss": 3.4128, + "step": 1959936 + }, + { + "epoch": 1.02, + "learning_rate": 1.7903883283334306e-05, + "loss": 3.4054, + "step": 1960448 + }, + { + "epoch": 1.02, + "learning_rate": 1.7895497335823786e-05, + "loss": 3.4038, + "step": 1960960 + }, + { + "epoch": 1.02, + "learning_rate": 1.7887127767117e-05, + "loss": 3.4105, + "step": 1961472 + }, + { + "epoch": 1.02, + "learning_rate": 1.7878741819606475e-05, + "loss": 3.4036, + "step": 1961984 + }, + { + "epoch": 1.02, + "learning_rate": 1.7870355872095955e-05, + "loss": 3.3992, + "step": 1962496 + }, + { + "epoch": 1.02, + "learning_rate": 1.7861969924585435e-05, + "loss": 3.4164, + "step": 1963008 + }, + { + "epoch": 1.02, + "learning_rate": 1.7853583977074915e-05, + "loss": 3.4052, + "step": 1963520 + }, + { + "epoch": 1.02, + "learning_rate": 1.78451980295644e-05, + "loss": 3.4068, + "step": 1964032 + }, + { + "epoch": 1.02, + "learning_rate": 1.783681208205388e-05, + "loss": 3.4031, + "step": 1964544 + }, + { + "epoch": 1.02, + "learning_rate": 1.7828442513347088e-05, + "loss": 3.3999, + "step": 1965056 + }, + { + "epoch": 1.02, + "learning_rate": 1.7820056565836568e-05, + "loss": 3.4191, + "step": 1965568 + }, + { + "epoch": 1.02, + "learning_rate": 1.7811670618326048e-05, + "loss": 3.4079, + "step": 1966080 + }, + { + "epoch": 1.02, + "learning_rate": 1.7803284670815528e-05, + "loss": 3.4226, + "step": 1966592 + }, + { + "epoch": 1.02, + "learning_rate": 1.779491510210874e-05, + "loss": 3.4178, + "step": 1967104 + }, + { + "epoch": 1.02, + "learning_rate": 1.778652915459822e-05, + "loss": 3.4101, + "step": 1967616 + }, + { + "epoch": 1.02, + "learning_rate": 1.77781432070877e-05, + "loss": 3.4074, + "step": 1968128 + }, + { + "epoch": 1.02, + "learning_rate": 1.776975725957718e-05, + "loss": 3.4139, + "step": 1968640 + }, + { + "epoch": 1.02, + "learning_rate": 1.7761371312066656e-05, + "loss": 3.4038, + "step": 1969152 + }, + { + "epoch": 1.02, + "learning_rate": 1.7752985364556136e-05, + "loss": 3.4157, + "step": 1969664 + }, + { + "epoch": 1.02, + "learning_rate": 1.7744615795849352e-05, + "loss": 3.4102, + "step": 1970176 + }, + { + "epoch": 1.02, + "learning_rate": 1.7736229848338832e-05, + "loss": 3.4107, + "step": 1970688 + }, + { + "epoch": 1.02, + "learning_rate": 1.772784390082831e-05, + "loss": 3.4056, + "step": 1971200 + }, + { + "epoch": 1.02, + "learning_rate": 1.771945795331779e-05, + "loss": 3.4092, + "step": 1971712 + }, + { + "epoch": 1.02, + "learning_rate": 1.771107200580727e-05, + "loss": 3.4125, + "step": 1972224 + }, + { + "epoch": 1.02, + "learning_rate": 1.770268605829675e-05, + "loss": 3.4027, + "step": 1972736 + }, + { + "epoch": 1.02, + "learning_rate": 1.769430011078623e-05, + "loss": 3.4119, + "step": 1973248 + }, + { + "epoch": 1.02, + "learning_rate": 1.7685914163275712e-05, + "loss": 3.3975, + "step": 1973760 + }, + { + "epoch": 1.02, + "learning_rate": 1.7677560973372654e-05, + "loss": 3.4289, + "step": 1974272 + }, + { + "epoch": 1.02, + "learning_rate": 1.766917502586213e-05, + "loss": 3.4047, + "step": 1974784 + }, + { + "epoch": 1.02, + "learning_rate": 1.766078907835161e-05, + "loss": 3.4141, + "step": 1975296 + }, + { + "epoch": 1.02, + "learning_rate": 1.765240313084109e-05, + "loss": 3.4037, + "step": 1975808 + }, + { + "epoch": 1.02, + "learning_rate": 1.7644017183330573e-05, + "loss": 3.4182, + "step": 1976320 + }, + { + "epoch": 1.02, + "learning_rate": 1.7635647614623782e-05, + "loss": 3.4059, + "step": 1976832 + }, + { + "epoch": 1.02, + "learning_rate": 1.7627261667113262e-05, + "loss": 3.4114, + "step": 1977344 + }, + { + "epoch": 1.02, + "learning_rate": 1.7618875719602742e-05, + "loss": 3.4047, + "step": 1977856 + }, + { + "epoch": 1.02, + "learning_rate": 1.7610489772092222e-05, + "loss": 3.4103, + "step": 1978368 + }, + { + "epoch": 1.02, + "learning_rate": 1.7602103824581702e-05, + "loss": 3.4124, + "step": 1978880 + }, + { + "epoch": 1.02, + "learning_rate": 1.7593734255874915e-05, + "loss": 3.4315, + "step": 1979392 + }, + { + "epoch": 1.02, + "learning_rate": 1.7585348308364395e-05, + "loss": 3.4045, + "step": 1979904 + }, + { + "epoch": 1.02, + "learning_rate": 1.7576962360853875e-05, + "loss": 3.4187, + "step": 1980416 + }, + { + "epoch": 1.02, + "learning_rate": 1.7568576413343355e-05, + "loss": 3.416, + "step": 1980928 + }, + { + "epoch": 1.02, + "learning_rate": 1.7560190465832835e-05, + "loss": 3.4119, + "step": 1981440 + }, + { + "epoch": 1.02, + "learning_rate": 1.7551820897126044e-05, + "loss": 3.4026, + "step": 1981952 + }, + { + "epoch": 1.02, + "learning_rate": 1.7543434949615524e-05, + "loss": 3.4145, + "step": 1982464 + }, + { + "epoch": 1.02, + "learning_rate": 1.7535049002105007e-05, + "loss": 3.4126, + "step": 1982976 + }, + { + "epoch": 1.02, + "learning_rate": 1.7526663054594484e-05, + "loss": 3.4037, + "step": 1983488 + }, + { + "epoch": 1.02, + "learning_rate": 1.7518277107083964e-05, + "loss": 3.4091, + "step": 1984000 + }, + { + "epoch": 1.03, + "eval_loss": 3.8629136085510254, + "eval_runtime": 306.1756, + "eval_samples_per_second": 1246.314, + "eval_steps_per_second": 38.948, + "step": 1984320 + }, + { + "epoch": 0.0, + "learning_rate": 1.7509907538377176e-05, + "loss": 3.4035, + "step": 1984512 + }, + { + "epoch": 0.0, + "learning_rate": 1.7501521590866656e-05, + "loss": 3.3972, + "step": 1985024 + }, + { + "epoch": 0.0, + "learning_rate": 1.749315202215987e-05, + "loss": 3.4116, + "step": 1985536 + }, + { + "epoch": 0.0, + "learning_rate": 1.748476607464935e-05, + "loss": 3.4242, + "step": 1986048 + }, + { + "epoch": 0.0, + "learning_rate": 1.747638012713883e-05, + "loss": 3.4203, + "step": 1986560 + }, + { + "epoch": 0.0, + "learning_rate": 1.746799417962831e-05, + "loss": 3.4033, + "step": 1987072 + }, + { + "epoch": 0.0, + "learning_rate": 1.7459608232117785e-05, + "loss": 3.412, + "step": 1987584 + }, + { + "epoch": 0.0, + "learning_rate": 1.7451222284607265e-05, + "loss": 3.4044, + "step": 1988096 + }, + { + "epoch": 0.0, + "learning_rate": 1.7442836337096745e-05, + "loss": 3.4058, + "step": 1988608 + }, + { + "epoch": 0.0, + "learning_rate": 1.7434466768389957e-05, + "loss": 3.4185, + "step": 1989120 + }, + { + "epoch": 0.0, + "learning_rate": 1.7426080820879437e-05, + "loss": 3.4071, + "step": 1989632 + }, + { + "epoch": 0.0, + "learning_rate": 1.7417694873368917e-05, + "loss": 3.4124, + "step": 1990144 + }, + { + "epoch": 0.0, + "learning_rate": 1.7409308925858397e-05, + "loss": 3.4124, + "step": 1990656 + }, + { + "epoch": 0.0, + "learning_rate": 1.7400939357151606e-05, + "loss": 3.3973, + "step": 1991168 + }, + { + "epoch": 0.0, + "learning_rate": 1.739255340964109e-05, + "loss": 3.3994, + "step": 1991680 + }, + { + "epoch": 0.0, + "learning_rate": 1.738416746213057e-05, + "loss": 3.3939, + "step": 1992192 + }, + { + "epoch": 0.0, + "learning_rate": 1.737578151462005e-05, + "loss": 3.4057, + "step": 1992704 + }, + { + "epoch": 0.0, + "learning_rate": 1.736739556710953e-05, + "loss": 3.4101, + "step": 1993216 + }, + { + "epoch": 0.0, + "learning_rate": 1.735900961959901e-05, + "loss": 3.3996, + "step": 1993728 + }, + { + "epoch": 0.0, + "learning_rate": 1.735062367208849e-05, + "loss": 3.4169, + "step": 1994240 + }, + { + "epoch": 0.0, + "learning_rate": 1.73422541033817e-05, + "loss": 3.4191, + "step": 1994752 + }, + { + "epoch": 0.0, + "learning_rate": 1.7333868155871182e-05, + "loss": 3.4144, + "step": 1995264 + }, + { + "epoch": 0.0, + "learning_rate": 1.7325482208360662e-05, + "loss": 3.41, + "step": 1995776 + }, + { + "epoch": 0.0, + "learning_rate": 1.731709626085014e-05, + "loss": 3.4088, + "step": 1996288 + }, + { + "epoch": 0.0, + "learning_rate": 1.7308710313339618e-05, + "loss": 3.4136, + "step": 1996800 + }, + { + "epoch": 0.0, + "learning_rate": 1.730034074463283e-05, + "loss": 3.4103, + "step": 1997312 + }, + { + "epoch": 0.0, + "learning_rate": 1.729195479712231e-05, + "loss": 3.4039, + "step": 1997824 + }, + { + "epoch": 0.0, + "learning_rate": 1.728356884961179e-05, + "loss": 3.4046, + "step": 1998336 + }, + { + "epoch": 0.0, + "learning_rate": 1.727518290210127e-05, + "loss": 3.4051, + "step": 1998848 + }, + { + "epoch": 0.0, + "learning_rate": 1.726679695459075e-05, + "loss": 3.3991, + "step": 1999360 + }, + { + "epoch": 0.01, + "learning_rate": 1.7258427385883963e-05, + "loss": 3.4052, + "step": 1999872 + }, + { + "epoch": 0.01, + "learning_rate": 1.725004143837344e-05, + "loss": 3.4107, + "step": 2000384 + }, + { + "epoch": 0.01, + "learning_rate": 1.724165549086292e-05, + "loss": 3.4137, + "step": 2000896 + }, + { + "epoch": 0.01, + "learning_rate": 1.72332695433524e-05, + "loss": 3.4153, + "step": 2001408 + }, + { + "epoch": 0.01, + "learning_rate": 1.7224883595841883e-05, + "loss": 3.3987, + "step": 2001920 + }, + { + "epoch": 0.01, + "learning_rate": 1.7216497648331363e-05, + "loss": 3.4054, + "step": 2002432 + }, + { + "epoch": 0.01, + "learning_rate": 1.7208111700820843e-05, + "loss": 3.4136, + "step": 2002944 + }, + { + "epoch": 0.01, + "learning_rate": 1.719972575331032e-05, + "loss": 3.3955, + "step": 2003456 + }, + { + "epoch": 0.01, + "learning_rate": 1.7191356184603532e-05, + "loss": 3.4069, + "step": 2003968 + }, + { + "epoch": 0.01, + "learning_rate": 1.7182986615896744e-05, + "loss": 3.394, + "step": 2004480 + }, + { + "epoch": 0.01, + "learning_rate": 1.7174600668386224e-05, + "loss": 3.4079, + "step": 2004992 + }, + { + "epoch": 0.01, + "learning_rate": 1.7166214720875704e-05, + "loss": 3.402, + "step": 2005504 + }, + { + "epoch": 0.01, + "learning_rate": 1.7157828773365184e-05, + "loss": 3.405, + "step": 2006016 + }, + { + "epoch": 0.01, + "learning_rate": 1.7149442825854664e-05, + "loss": 3.4093, + "step": 2006528 + }, + { + "epoch": 0.01, + "learning_rate": 1.7141073257147873e-05, + "loss": 3.4083, + "step": 2007040 + }, + { + "epoch": 0.01, + "learning_rate": 1.7132687309637353e-05, + "loss": 3.4102, + "step": 2007552 + }, + { + "epoch": 0.01, + "learning_rate": 1.7124301362126837e-05, + "loss": 3.4127, + "step": 2008064 + }, + { + "epoch": 0.01, + "learning_rate": 1.7115915414616317e-05, + "loss": 3.4038, + "step": 2008576 + }, + { + "epoch": 0.01, + "learning_rate": 1.7107545845909526e-05, + "loss": 3.3758, + "step": 2009088 + }, + { + "epoch": 0.01, + "learning_rate": 1.7099159898399006e-05, + "loss": 3.4185, + "step": 2009600 + }, + { + "epoch": 0.01, + "learning_rate": 1.7090773950888486e-05, + "loss": 3.406, + "step": 2010112 + }, + { + "epoch": 0.01, + "learning_rate": 1.7082388003377965e-05, + "loss": 3.4133, + "step": 2010624 + }, + { + "epoch": 0.01, + "learning_rate": 1.7074018434671178e-05, + "loss": 3.4015, + "step": 2011136 + }, + { + "epoch": 0.01, + "learning_rate": 1.7065632487160658e-05, + "loss": 3.4, + "step": 2011648 + }, + { + "epoch": 0.01, + "learning_rate": 1.7057246539650138e-05, + "loss": 3.3901, + "step": 2012160 + }, + { + "epoch": 0.01, + "learning_rate": 1.7048860592139618e-05, + "loss": 3.3979, + "step": 2012672 + }, + { + "epoch": 0.01, + "learning_rate": 1.7040474644629094e-05, + "loss": 3.4023, + "step": 2013184 + }, + { + "epoch": 0.01, + "learning_rate": 1.7032088697118574e-05, + "loss": 3.3921, + "step": 2013696 + }, + { + "epoch": 0.01, + "learning_rate": 1.7023702749608054e-05, + "loss": 3.4167, + "step": 2014208 + }, + { + "epoch": 0.01, + "learning_rate": 1.7015316802097538e-05, + "loss": 3.4031, + "step": 2014720 + }, + { + "epoch": 0.01, + "learning_rate": 1.7006947233390747e-05, + "loss": 3.3852, + "step": 2015232 + }, + { + "epoch": 0.01, + "learning_rate": 1.6998561285880227e-05, + "loss": 3.4106, + "step": 2015744 + }, + { + "epoch": 0.01, + "learning_rate": 1.6990175338369707e-05, + "loss": 3.3893, + "step": 2016256 + }, + { + "epoch": 0.01, + "learning_rate": 1.6981789390859187e-05, + "loss": 3.3914, + "step": 2016768 + }, + { + "epoch": 0.01, + "learning_rate": 1.69734198221524e-05, + "loss": 3.4066, + "step": 2017280 + }, + { + "epoch": 0.01, + "learning_rate": 1.696503387464188e-05, + "loss": 3.4023, + "step": 2017792 + }, + { + "epoch": 0.01, + "learning_rate": 1.695664792713136e-05, + "loss": 3.3933, + "step": 2018304 + }, + { + "epoch": 0.01, + "learning_rate": 1.694826197962084e-05, + "loss": 3.3954, + "step": 2018816 + }, + { + "epoch": 0.01, + "learning_rate": 1.693987603211032e-05, + "loss": 3.3961, + "step": 2019328 + }, + { + "epoch": 0.01, + "learning_rate": 1.6931490084599796e-05, + "loss": 3.398, + "step": 2019840 + }, + { + "epoch": 0.01, + "learning_rate": 1.6923104137089275e-05, + "loss": 3.4066, + "step": 2020352 + }, + { + "epoch": 0.01, + "learning_rate": 1.6914718189578755e-05, + "loss": 3.3989, + "step": 2020864 + }, + { + "epoch": 0.01, + "learning_rate": 1.690634862087197e-05, + "loss": 3.4032, + "step": 2021376 + }, + { + "epoch": 0.01, + "learning_rate": 1.689797905216518e-05, + "loss": 3.4001, + "step": 2021888 + }, + { + "epoch": 0.01, + "learning_rate": 1.688959310465466e-05, + "loss": 3.4169, + "step": 2022400 + }, + { + "epoch": 0.01, + "learning_rate": 1.688120715714414e-05, + "loss": 3.3954, + "step": 2022912 + }, + { + "epoch": 0.01, + "learning_rate": 1.687282120963362e-05, + "loss": 3.4081, + "step": 2023424 + }, + { + "epoch": 0.01, + "learning_rate": 1.68644352621231e-05, + "loss": 3.4045, + "step": 2023936 + }, + { + "epoch": 0.01, + "learning_rate": 1.685604931461258e-05, + "loss": 3.3923, + "step": 2024448 + }, + { + "epoch": 0.01, + "learning_rate": 1.684766336710206e-05, + "loss": 3.4045, + "step": 2024960 + }, + { + "epoch": 0.01, + "learning_rate": 1.683927741959154e-05, + "loss": 3.4058, + "step": 2025472 + }, + { + "epoch": 0.01, + "learning_rate": 1.6830924229688482e-05, + "loss": 3.4095, + "step": 2025984 + }, + { + "epoch": 0.01, + "learning_rate": 1.682253828217796e-05, + "loss": 3.4104, + "step": 2026496 + }, + { + "epoch": 0.01, + "learning_rate": 1.6814152334667445e-05, + "loss": 3.3817, + "step": 2027008 + }, + { + "epoch": 0.01, + "learning_rate": 1.680576638715692e-05, + "loss": 3.3824, + "step": 2027520 + }, + { + "epoch": 0.01, + "learning_rate": 1.6797396818450134e-05, + "loss": 3.4004, + "step": 2028032 + }, + { + "epoch": 0.01, + "learning_rate": 1.6789010870939614e-05, + "loss": 3.4048, + "step": 2028544 + }, + { + "epoch": 0.01, + "learning_rate": 1.6780624923429094e-05, + "loss": 3.3935, + "step": 2029056 + }, + { + "epoch": 0.01, + "learning_rate": 1.677223897591857e-05, + "loss": 3.4001, + "step": 2029568 + }, + { + "epoch": 0.01, + "learning_rate": 1.6763869407211786e-05, + "loss": 3.3965, + "step": 2030080 + }, + { + "epoch": 0.02, + "learning_rate": 1.6755483459701266e-05, + "loss": 3.4076, + "step": 2030592 + }, + { + "epoch": 0.02, + "learning_rate": 1.6747097512190743e-05, + "loss": 3.397, + "step": 2031104 + }, + { + "epoch": 0.02, + "learning_rate": 1.6738711564680223e-05, + "loss": 3.3951, + "step": 2031616 + }, + { + "epoch": 0.02, + "learning_rate": 1.6730325617169703e-05, + "loss": 3.4006, + "step": 2032128 + }, + { + "epoch": 0.02, + "learning_rate": 1.6721956048462915e-05, + "loss": 3.4072, + "step": 2032640 + }, + { + "epoch": 0.02, + "learning_rate": 1.6713570100952395e-05, + "loss": 3.407, + "step": 2033152 + }, + { + "epoch": 0.02, + "learning_rate": 1.6705184153441875e-05, + "loss": 3.3877, + "step": 2033664 + }, + { + "epoch": 0.02, + "learning_rate": 1.6696798205931355e-05, + "loss": 3.3954, + "step": 2034176 + }, + { + "epoch": 0.02, + "learning_rate": 1.6688428637224568e-05, + "loss": 3.4008, + "step": 2034688 + }, + { + "epoch": 0.02, + "learning_rate": 1.6680042689714044e-05, + "loss": 3.4081, + "step": 2035200 + }, + { + "epoch": 0.02, + "learning_rate": 1.6671656742203524e-05, + "loss": 3.4006, + "step": 2035712 + }, + { + "epoch": 0.02, + "learning_rate": 1.6663270794693008e-05, + "loss": 3.4026, + "step": 2036224 + }, + { + "epoch": 0.02, + "learning_rate": 1.6654884847182488e-05, + "loss": 3.3942, + "step": 2036736 + }, + { + "epoch": 0.02, + "learning_rate": 1.6646498899671967e-05, + "loss": 3.3929, + "step": 2037248 + }, + { + "epoch": 0.02, + "learning_rate": 1.6638129330965177e-05, + "loss": 3.4017, + "step": 2037760 + }, + { + "epoch": 0.02, + "learning_rate": 1.6629743383454657e-05, + "loss": 3.3936, + "step": 2038272 + }, + { + "epoch": 0.02, + "learning_rate": 1.6621357435944136e-05, + "loss": 3.3908, + "step": 2038784 + }, + { + "epoch": 0.02, + "learning_rate": 1.6612971488433616e-05, + "loss": 3.4064, + "step": 2039296 + }, + { + "epoch": 0.02, + "learning_rate": 1.66045855409231e-05, + "loss": 3.3953, + "step": 2039808 + }, + { + "epoch": 0.02, + "learning_rate": 1.6596199593412576e-05, + "loss": 3.3949, + "step": 2040320 + }, + { + "epoch": 0.02, + "learning_rate": 1.6587813645902056e-05, + "loss": 3.3925, + "step": 2040832 + }, + { + "epoch": 0.02, + "learning_rate": 1.657944407719527e-05, + "loss": 3.3919, + "step": 2041344 + }, + { + "epoch": 0.02, + "learning_rate": 1.657105812968475e-05, + "loss": 3.4061, + "step": 2041856 + }, + { + "epoch": 0.02, + "learning_rate": 1.656267218217423e-05, + "loss": 3.4007, + "step": 2042368 + }, + { + "epoch": 0.02, + "learning_rate": 1.655428623466371e-05, + "loss": 3.4077, + "step": 2042880 + }, + { + "epoch": 0.02, + "learning_rate": 1.654591666595692e-05, + "loss": 3.4083, + "step": 2043392 + }, + { + "epoch": 0.02, + "learning_rate": 1.6537530718446398e-05, + "loss": 3.403, + "step": 2043904 + }, + { + "epoch": 0.02, + "learning_rate": 1.6529144770935878e-05, + "loss": 3.3987, + "step": 2044416 + }, + { + "epoch": 0.02, + "learning_rate": 1.6520758823425358e-05, + "loss": 3.4006, + "step": 2044928 + }, + { + "epoch": 0.02, + "learning_rate": 1.6512372875914838e-05, + "loss": 3.3965, + "step": 2045440 + }, + { + "epoch": 0.02, + "learning_rate": 1.6503986928404318e-05, + "loss": 3.4056, + "step": 2045952 + }, + { + "epoch": 0.02, + "learning_rate": 1.649561735969753e-05, + "loss": 3.3978, + "step": 2046464 + }, + { + "epoch": 0.02, + "learning_rate": 1.648723141218701e-05, + "loss": 3.4036, + "step": 2046976 + }, + { + "epoch": 0.02, + "learning_rate": 1.647884546467649e-05, + "loss": 3.3904, + "step": 2047488 + }, + { + "epoch": 0.02, + "learning_rate": 1.647045951716597e-05, + "loss": 3.4048, + "step": 2048000 + }, + { + "epoch": 0.02, + "learning_rate": 1.646207356965545e-05, + "loss": 3.3996, + "step": 2048512 + }, + { + "epoch": 0.02, + "learning_rate": 1.6453704000948662e-05, + "loss": 3.3963, + "step": 2049024 + }, + { + "epoch": 0.02, + "learning_rate": 1.6445318053438142e-05, + "loss": 3.401, + "step": 2049536 + }, + { + "epoch": 0.02, + "learning_rate": 1.6436932105927622e-05, + "loss": 3.3876, + "step": 2050048 + }, + { + "epoch": 0.02, + "learning_rate": 1.6428546158417102e-05, + "loss": 3.4201, + "step": 2050560 + }, + { + "epoch": 0.02, + "learning_rate": 1.642016021090658e-05, + "loss": 3.3946, + "step": 2051072 + }, + { + "epoch": 0.02, + "learning_rate": 1.641177426339606e-05, + "loss": 3.4087, + "step": 2051584 + }, + { + "epoch": 0.02, + "learning_rate": 1.640338831588554e-05, + "loss": 3.3942, + "step": 2052096 + }, + { + "epoch": 0.02, + "learning_rate": 1.6395002368375022e-05, + "loss": 3.4054, + "step": 2052608 + }, + { + "epoch": 0.02, + "learning_rate": 1.638663279966823e-05, + "loss": 3.3966, + "step": 2053120 + }, + { + "epoch": 0.02, + "learning_rate": 1.6378263230961444e-05, + "loss": 3.4048, + "step": 2053632 + }, + { + "epoch": 0.02, + "learning_rate": 1.6369877283450924e-05, + "loss": 3.3928, + "step": 2054144 + }, + { + "epoch": 0.02, + "learning_rate": 1.6361491335940403e-05, + "loss": 3.399, + "step": 2054656 + }, + { + "epoch": 0.02, + "learning_rate": 1.6353105388429883e-05, + "loss": 3.403, + "step": 2055168 + }, + { + "epoch": 0.02, + "learning_rate": 1.6344735819723096e-05, + "loss": 3.4228, + "step": 2055680 + }, + { + "epoch": 0.02, + "learning_rate": 1.6336349872212576e-05, + "loss": 3.3903, + "step": 2056192 + }, + { + "epoch": 0.02, + "learning_rate": 1.6327963924702052e-05, + "loss": 3.4123, + "step": 2056704 + }, + { + "epoch": 0.02, + "learning_rate": 1.6319577977191532e-05, + "loss": 3.4076, + "step": 2057216 + }, + { + "epoch": 0.02, + "learning_rate": 1.6311192029681012e-05, + "loss": 3.4003, + "step": 2057728 + }, + { + "epoch": 0.02, + "learning_rate": 1.6302822460974225e-05, + "loss": 3.3939, + "step": 2058240 + }, + { + "epoch": 0.02, + "learning_rate": 1.6294436513463705e-05, + "loss": 3.4035, + "step": 2058752 + }, + { + "epoch": 0.02, + "learning_rate": 1.6286050565953185e-05, + "loss": 3.4035, + "step": 2059264 + }, + { + "epoch": 0.02, + "learning_rate": 1.6277664618442665e-05, + "loss": 3.3992, + "step": 2059776 + }, + { + "epoch": 0.02, + "learning_rate": 1.6269295049735874e-05, + "loss": 3.3976, + "step": 2060288 + }, + { + "epoch": 0.03, + "eval_loss": 3.864262580871582, + "eval_runtime": 305.0463, + "eval_samples_per_second": 1250.928, + "eval_steps_per_second": 39.092, + "step": 2060640 + }, + { + "epoch": 1.0, + "learning_rate": 1.6260909102225354e-05, + "loss": 3.3972, + "step": 2060800 + }, + { + "epoch": 1.0, + "learning_rate": 1.6252523154714837e-05, + "loss": 3.3899, + "step": 2061312 + }, + { + "epoch": 1.0, + "learning_rate": 1.6244137207204317e-05, + "loss": 3.4017, + "step": 2061824 + }, + { + "epoch": 1.0, + "learning_rate": 1.6235751259693797e-05, + "loss": 3.4134, + "step": 2062336 + }, + { + "epoch": 1.0, + "learning_rate": 1.6227365312183277e-05, + "loss": 3.4118, + "step": 2062848 + }, + { + "epoch": 1.0, + "learning_rate": 1.6218995743476486e-05, + "loss": 3.3932, + "step": 2063360 + }, + { + "epoch": 1.0, + "learning_rate": 1.6210609795965966e-05, + "loss": 3.4033, + "step": 2063872 + }, + { + "epoch": 1.0, + "learning_rate": 1.6202223848455446e-05, + "loss": 3.3939, + "step": 2064384 + }, + { + "epoch": 1.0, + "learning_rate": 1.6193837900944926e-05, + "loss": 3.3927, + "step": 2064896 + }, + { + "epoch": 1.0, + "learning_rate": 1.618545195343441e-05, + "loss": 3.4043, + "step": 2065408 + }, + { + "epoch": 1.0, + "learning_rate": 1.6177066005923886e-05, + "loss": 3.4042, + "step": 2065920 + }, + { + "epoch": 1.0, + "learning_rate": 1.6168680058413366e-05, + "loss": 3.4025, + "step": 2066432 + }, + { + "epoch": 1.0, + "learning_rate": 1.6160310489706578e-05, + "loss": 3.3979, + "step": 2066944 + }, + { + "epoch": 1.0, + "learning_rate": 1.6151924542196055e-05, + "loss": 3.3905, + "step": 2067456 + }, + { + "epoch": 1.0, + "learning_rate": 1.6143538594685538e-05, + "loss": 3.3898, + "step": 2067968 + }, + { + "epoch": 1.0, + "learning_rate": 1.6135152647175018e-05, + "loss": 3.379, + "step": 2068480 + }, + { + "epoch": 1.0, + "learning_rate": 1.6126766699664498e-05, + "loss": 3.3991, + "step": 2068992 + }, + { + "epoch": 1.0, + "learning_rate": 1.6118380752153978e-05, + "loss": 3.3982, + "step": 2069504 + }, + { + "epoch": 1.0, + "learning_rate": 1.6109994804643458e-05, + "loss": 3.3894, + "step": 2070016 + }, + { + "epoch": 1.0, + "learning_rate": 1.6101608857132938e-05, + "loss": 3.4032, + "step": 2070528 + }, + { + "epoch": 1.0, + "learning_rate": 1.609325566722988e-05, + "loss": 3.4096, + "step": 2071040 + }, + { + "epoch": 1.0, + "learning_rate": 1.608486971971936e-05, + "loss": 3.4036, + "step": 2071552 + }, + { + "epoch": 1.0, + "learning_rate": 1.607648377220884e-05, + "loss": 3.4007, + "step": 2072064 + }, + { + "epoch": 1.0, + "learning_rate": 1.606809782469832e-05, + "loss": 3.3963, + "step": 2072576 + }, + { + "epoch": 1.0, + "learning_rate": 1.60597118771878e-05, + "loss": 3.4054, + "step": 2073088 + }, + { + "epoch": 1.0, + "learning_rate": 1.605134230848101e-05, + "loss": 3.4004, + "step": 2073600 + }, + { + "epoch": 1.0, + "learning_rate": 1.6042956360970492e-05, + "loss": 3.393, + "step": 2074112 + }, + { + "epoch": 1.0, + "learning_rate": 1.6034570413459972e-05, + "loss": 3.3928, + "step": 2074624 + }, + { + "epoch": 1.0, + "learning_rate": 1.6026184465949452e-05, + "loss": 3.3935, + "step": 2075136 + }, + { + "epoch": 1.0, + "learning_rate": 1.6017798518438932e-05, + "loss": 3.3893, + "step": 2075648 + }, + { + "epoch": 1.01, + "learning_rate": 1.600942894973214e-05, + "loss": 3.398, + "step": 2076160 + }, + { + "epoch": 1.01, + "learning_rate": 1.600104300222162e-05, + "loss": 3.4001, + "step": 2076672 + }, + { + "epoch": 1.01, + "learning_rate": 1.59926570547111e-05, + "loss": 3.3982, + "step": 2077184 + }, + { + "epoch": 1.01, + "learning_rate": 1.5984271107200584e-05, + "loss": 3.4087, + "step": 2077696 + }, + { + "epoch": 1.01, + "learning_rate": 1.5975885159690064e-05, + "loss": 3.3891, + "step": 2078208 + }, + { + "epoch": 1.01, + "learning_rate": 1.596749921217954e-05, + "loss": 3.3947, + "step": 2078720 + }, + { + "epoch": 1.01, + "learning_rate": 1.595911326466902e-05, + "loss": 3.4026, + "step": 2079232 + }, + { + "epoch": 1.01, + "learning_rate": 1.59507273171585e-05, + "loss": 3.3858, + "step": 2079744 + }, + { + "epoch": 1.01, + "learning_rate": 1.594235774845171e-05, + "loss": 3.398, + "step": 2080256 + }, + { + "epoch": 1.01, + "learning_rate": 1.5933988179744925e-05, + "loss": 3.3866, + "step": 2080768 + }, + { + "epoch": 1.01, + "learning_rate": 1.5925602232234405e-05, + "loss": 3.3909, + "step": 2081280 + }, + { + "epoch": 1.01, + "learning_rate": 1.5917216284723885e-05, + "loss": 3.3932, + "step": 2081792 + }, + { + "epoch": 1.01, + "learning_rate": 1.5908830337213362e-05, + "loss": 3.398, + "step": 2082304 + }, + { + "epoch": 1.01, + "learning_rate": 1.5900444389702842e-05, + "loss": 3.4016, + "step": 2082816 + }, + { + "epoch": 1.01, + "learning_rate": 1.5892074820996054e-05, + "loss": 3.3969, + "step": 2083328 + }, + { + "epoch": 1.01, + "learning_rate": 1.5883688873485534e-05, + "loss": 3.4019, + "step": 2083840 + }, + { + "epoch": 1.01, + "learning_rate": 1.5875302925975014e-05, + "loss": 3.4008, + "step": 2084352 + }, + { + "epoch": 1.01, + "learning_rate": 1.5866916978464494e-05, + "loss": 3.3926, + "step": 2084864 + }, + { + "epoch": 1.01, + "learning_rate": 1.5858547409757707e-05, + "loss": 3.3683, + "step": 2085376 + }, + { + "epoch": 1.01, + "learning_rate": 1.5850161462247183e-05, + "loss": 3.4063, + "step": 2085888 + }, + { + "epoch": 1.01, + "learning_rate": 1.5841775514736663e-05, + "loss": 3.3946, + "step": 2086400 + }, + { + "epoch": 1.01, + "learning_rate": 1.5833389567226147e-05, + "loss": 3.4055, + "step": 2086912 + }, + { + "epoch": 1.01, + "learning_rate": 1.582501999851936e-05, + "loss": 3.3929, + "step": 2087424 + }, + { + "epoch": 1.01, + "learning_rate": 1.5816634051008836e-05, + "loss": 3.3866, + "step": 2087936 + }, + { + "epoch": 1.01, + "learning_rate": 1.5808248103498316e-05, + "loss": 3.3871, + "step": 2088448 + }, + { + "epoch": 1.01, + "learning_rate": 1.5799862155987796e-05, + "loss": 3.3824, + "step": 2088960 + }, + { + "epoch": 1.01, + "learning_rate": 1.5791476208477276e-05, + "loss": 3.3934, + "step": 2089472 + }, + { + "epoch": 1.01, + "learning_rate": 1.5783090260966756e-05, + "loss": 3.3782, + "step": 2089984 + }, + { + "epoch": 1.01, + "learning_rate": 1.577470431345624e-05, + "loss": 3.4097, + "step": 2090496 + }, + { + "epoch": 1.01, + "learning_rate": 1.5766334744749448e-05, + "loss": 3.3939, + "step": 2091008 + }, + { + "epoch": 1.01, + "learning_rate": 1.5757948797238928e-05, + "loss": 3.3729, + "step": 2091520 + }, + { + "epoch": 1.01, + "learning_rate": 1.5749562849728408e-05, + "loss": 3.4023, + "step": 2092032 + }, + { + "epoch": 1.01, + "learning_rate": 1.5741176902217888e-05, + "loss": 3.3807, + "step": 2092544 + }, + { + "epoch": 1.01, + "learning_rate": 1.5732790954707364e-05, + "loss": 3.3813, + "step": 2093056 + }, + { + "epoch": 1.01, + "learning_rate": 1.572442138600058e-05, + "loss": 3.3908, + "step": 2093568 + }, + { + "epoch": 1.01, + "learning_rate": 1.571603543849006e-05, + "loss": 3.3919, + "step": 2094080 + }, + { + "epoch": 1.01, + "learning_rate": 1.570764949097954e-05, + "loss": 3.3816, + "step": 2094592 + }, + { + "epoch": 1.01, + "learning_rate": 1.5699263543469017e-05, + "loss": 3.3897, + "step": 2095104 + }, + { + "epoch": 1.01, + "learning_rate": 1.5690877595958497e-05, + "loss": 3.3822, + "step": 2095616 + }, + { + "epoch": 1.01, + "learning_rate": 1.5682491648447977e-05, + "loss": 3.3891, + "step": 2096128 + }, + { + "epoch": 1.01, + "learning_rate": 1.5674105700937457e-05, + "loss": 3.3971, + "step": 2096640 + }, + { + "epoch": 1.01, + "learning_rate": 1.566573613223067e-05, + "loss": 3.387, + "step": 2097152 + }, + { + "epoch": 1.01, + "learning_rate": 1.565735018472015e-05, + "loss": 3.3959, + "step": 2097664 + }, + { + "epoch": 1.01, + "learning_rate": 1.564896423720963e-05, + "loss": 3.3867, + "step": 2098176 + }, + { + "epoch": 1.01, + "learning_rate": 1.564057828969911e-05, + "loss": 3.4111, + "step": 2098688 + }, + { + "epoch": 1.01, + "learning_rate": 1.563219234218859e-05, + "loss": 3.3806, + "step": 2099200 + }, + { + "epoch": 1.01, + "learning_rate": 1.562380639467807e-05, + "loss": 3.4028, + "step": 2099712 + }, + { + "epoch": 1.01, + "learning_rate": 1.561542044716755e-05, + "loss": 3.3886, + "step": 2100224 + }, + { + "epoch": 1.01, + "learning_rate": 1.560703449965703e-05, + "loss": 3.3898, + "step": 2100736 + }, + { + "epoch": 1.01, + "learning_rate": 1.559866493095024e-05, + "loss": 3.3928, + "step": 2101248 + }, + { + "epoch": 1.01, + "learning_rate": 1.559029536224345e-05, + "loss": 3.3976, + "step": 2101760 + }, + { + "epoch": 1.01, + "learning_rate": 1.558190941473293e-05, + "loss": 3.3997, + "step": 2102272 + }, + { + "epoch": 1.01, + "learning_rate": 1.557352346722241e-05, + "loss": 3.3986, + "step": 2102784 + }, + { + "epoch": 1.01, + "learning_rate": 1.5565137519711894e-05, + "loss": 3.375, + "step": 2103296 + }, + { + "epoch": 1.01, + "learning_rate": 1.555675157220137e-05, + "loss": 3.3745, + "step": 2103808 + }, + { + "epoch": 1.01, + "learning_rate": 1.5548382003494583e-05, + "loss": 3.3893, + "step": 2104320 + }, + { + "epoch": 1.01, + "learning_rate": 1.5539996055984063e-05, + "loss": 3.3956, + "step": 2104832 + }, + { + "epoch": 1.01, + "learning_rate": 1.5531610108473543e-05, + "loss": 3.3852, + "step": 2105344 + }, + { + "epoch": 1.01, + "learning_rate": 1.552322416096302e-05, + "loss": 3.3877, + "step": 2105856 + }, + { + "epoch": 1.01, + "learning_rate": 1.5514854592256235e-05, + "loss": 3.3883, + "step": 2106368 + }, + { + "epoch": 1.02, + "learning_rate": 1.5506468644745715e-05, + "loss": 3.3912, + "step": 2106880 + }, + { + "epoch": 1.02, + "learning_rate": 1.5498082697235195e-05, + "loss": 3.3924, + "step": 2107392 + }, + { + "epoch": 1.02, + "learning_rate": 1.548969674972467e-05, + "loss": 3.387, + "step": 2107904 + }, + { + "epoch": 1.02, + "learning_rate": 1.548131080221415e-05, + "loss": 3.3918, + "step": 2108416 + }, + { + "epoch": 1.02, + "learning_rate": 1.5472941233507364e-05, + "loss": 3.3981, + "step": 2108928 + }, + { + "epoch": 1.02, + "learning_rate": 1.5464555285996844e-05, + "loss": 3.398, + "step": 2109440 + }, + { + "epoch": 1.02, + "learning_rate": 1.5456169338486324e-05, + "loss": 3.3786, + "step": 2109952 + }, + { + "epoch": 1.02, + "learning_rate": 1.5447783390975804e-05, + "loss": 3.3824, + "step": 2110464 + }, + { + "epoch": 1.02, + "learning_rate": 1.5439413822269016e-05, + "loss": 3.3916, + "step": 2110976 + }, + { + "epoch": 1.02, + "learning_rate": 1.5431027874758493e-05, + "loss": 3.3938, + "step": 2111488 + }, + { + "epoch": 1.02, + "learning_rate": 1.5422641927247973e-05, + "loss": 3.3932, + "step": 2112000 + }, + { + "epoch": 1.02, + "learning_rate": 1.5414255979737456e-05, + "loss": 3.3917, + "step": 2112512 + }, + { + "epoch": 1.02, + "learning_rate": 1.540588641103067e-05, + "loss": 3.3879, + "step": 2113024 + }, + { + "epoch": 1.02, + "learning_rate": 1.5397500463520145e-05, + "loss": 3.3839, + "step": 2113536 + }, + { + "epoch": 1.02, + "learning_rate": 1.5389114516009625e-05, + "loss": 3.3902, + "step": 2114048 + }, + { + "epoch": 1.02, + "learning_rate": 1.5380728568499105e-05, + "loss": 3.3837, + "step": 2114560 + }, + { + "epoch": 1.02, + "learning_rate": 1.5372342620988585e-05, + "loss": 3.381, + "step": 2115072 + }, + { + "epoch": 1.02, + "learning_rate": 1.5363956673478065e-05, + "loss": 3.3931, + "step": 2115584 + }, + { + "epoch": 1.02, + "learning_rate": 1.5355587104771278e-05, + "loss": 3.3867, + "step": 2116096 + }, + { + "epoch": 1.02, + "learning_rate": 1.5347201157260757e-05, + "loss": 3.3836, + "step": 2116608 + }, + { + "epoch": 1.02, + "learning_rate": 1.5338815209750237e-05, + "loss": 3.3846, + "step": 2117120 + }, + { + "epoch": 1.02, + "learning_rate": 1.5330429262239717e-05, + "loss": 3.3786, + "step": 2117632 + }, + { + "epoch": 1.02, + "learning_rate": 1.5322059693532926e-05, + "loss": 3.3977, + "step": 2118144 + }, + { + "epoch": 1.02, + "learning_rate": 1.531367374602241e-05, + "loss": 3.3923, + "step": 2118656 + }, + { + "epoch": 1.02, + "learning_rate": 1.530528779851189e-05, + "loss": 3.3963, + "step": 2119168 + }, + { + "epoch": 1.02, + "learning_rate": 1.529690185100137e-05, + "loss": 3.3963, + "step": 2119680 + }, + { + "epoch": 1.02, + "learning_rate": 1.528853228229458e-05, + "loss": 3.3963, + "step": 2120192 + }, + { + "epoch": 1.02, + "learning_rate": 1.528014633478406e-05, + "loss": 3.3908, + "step": 2120704 + }, + { + "epoch": 1.02, + "learning_rate": 1.527176038727354e-05, + "loss": 3.384, + "step": 2121216 + }, + { + "epoch": 1.02, + "learning_rate": 1.526337443976302e-05, + "loss": 3.3941, + "step": 2121728 + }, + { + "epoch": 1.02, + "learning_rate": 1.5254988492252497e-05, + "loss": 3.3886, + "step": 2122240 + }, + { + "epoch": 1.02, + "learning_rate": 1.5246618923545711e-05, + "loss": 3.3872, + "step": 2122752 + }, + { + "epoch": 1.02, + "learning_rate": 1.5238232976035191e-05, + "loss": 3.3952, + "step": 2123264 + }, + { + "epoch": 1.02, + "learning_rate": 1.5229847028524671e-05, + "loss": 3.3872, + "step": 2123776 + }, + { + "epoch": 1.02, + "learning_rate": 1.522146108101415e-05, + "loss": 3.3906, + "step": 2124288 + }, + { + "epoch": 1.02, + "learning_rate": 1.521307513350363e-05, + "loss": 3.3889, + "step": 2124800 + }, + { + "epoch": 1.02, + "learning_rate": 1.5204705564796842e-05, + "loss": 3.3869, + "step": 2125312 + }, + { + "epoch": 1.02, + "learning_rate": 1.5196319617286323e-05, + "loss": 3.3898, + "step": 2125824 + }, + { + "epoch": 1.02, + "learning_rate": 1.51879336697758e-05, + "loss": 3.3827, + "step": 2126336 + }, + { + "epoch": 1.02, + "learning_rate": 1.517954772226528e-05, + "loss": 3.4076, + "step": 2126848 + }, + { + "epoch": 1.02, + "learning_rate": 1.5171161774754762e-05, + "loss": 3.3887, + "step": 2127360 + }, + { + "epoch": 1.02, + "learning_rate": 1.5162775827244242e-05, + "loss": 3.3986, + "step": 2127872 + }, + { + "epoch": 1.02, + "learning_rate": 1.5154389879733721e-05, + "loss": 3.3835, + "step": 2128384 + }, + { + "epoch": 1.02, + "learning_rate": 1.5146003932223201e-05, + "loss": 3.3949, + "step": 2128896 + }, + { + "epoch": 1.02, + "learning_rate": 1.5137634363516412e-05, + "loss": 3.3902, + "step": 2129408 + }, + { + "epoch": 1.02, + "learning_rate": 1.5129264794809623e-05, + "loss": 3.3956, + "step": 2129920 + }, + { + "epoch": 1.02, + "learning_rate": 1.5120878847299103e-05, + "loss": 3.3835, + "step": 2130432 + }, + { + "epoch": 1.02, + "learning_rate": 1.5112492899788583e-05, + "loss": 3.3857, + "step": 2130944 + }, + { + "epoch": 1.02, + "learning_rate": 1.5104106952278063e-05, + "loss": 3.3923, + "step": 2131456 + }, + { + "epoch": 1.02, + "learning_rate": 1.5095737383571274e-05, + "loss": 3.4121, + "step": 2131968 + }, + { + "epoch": 1.02, + "learning_rate": 1.5087351436060754e-05, + "loss": 3.3862, + "step": 2132480 + }, + { + "epoch": 1.02, + "learning_rate": 1.5078965488550234e-05, + "loss": 3.3979, + "step": 2132992 + }, + { + "epoch": 1.02, + "learning_rate": 1.5070579541039714e-05, + "loss": 3.398, + "step": 2133504 + }, + { + "epoch": 1.02, + "learning_rate": 1.5062193593529195e-05, + "loss": 3.3865, + "step": 2134016 + }, + { + "epoch": 1.02, + "learning_rate": 1.5053824024822404e-05, + "loss": 3.3906, + "step": 2134528 + }, + { + "epoch": 1.02, + "learning_rate": 1.5045438077311886e-05, + "loss": 3.3903, + "step": 2135040 + }, + { + "epoch": 1.02, + "learning_rate": 1.5037052129801366e-05, + "loss": 3.3929, + "step": 2135552 + }, + { + "epoch": 1.02, + "learning_rate": 1.5028666182290846e-05, + "loss": 3.3891, + "step": 2136064 + }, + { + "epoch": 1.02, + "learning_rate": 1.5020280234780326e-05, + "loss": 3.389, + "step": 2136576 + }, + { + "epoch": 1.03, + "eval_loss": 3.865880250930786, + "eval_runtime": 305.1904, + "eval_samples_per_second": 1250.337, + "eval_steps_per_second": 39.074, + "step": 2136960 + }, + { + "epoch": 0.0, + "learning_rate": 1.5011894287269804e-05, + "loss": 3.3959, + "step": 2137088 + }, + { + "epoch": 0.0, + "learning_rate": 1.5003508339759284e-05, + "loss": 3.374, + "step": 2137600 + }, + { + "epoch": 0.0, + "learning_rate": 1.4995138771052497e-05, + "loss": 3.3897, + "step": 2138112 + }, + { + "epoch": 0.0, + "learning_rate": 1.4986752823541975e-05, + "loss": 3.4066, + "step": 2138624 + }, + { + "epoch": 0.0, + "learning_rate": 1.4978366876031455e-05, + "loss": 3.396, + "step": 2139136 + }, + { + "epoch": 0.0, + "learning_rate": 1.4969980928520935e-05, + "loss": 3.3883, + "step": 2139648 + }, + { + "epoch": 0.0, + "learning_rate": 1.4961594981010416e-05, + "loss": 3.3971, + "step": 2140160 + }, + { + "epoch": 0.0, + "learning_rate": 1.4953209033499896e-05, + "loss": 3.3813, + "step": 2140672 + }, + { + "epoch": 0.0, + "learning_rate": 1.4944823085989376e-05, + "loss": 3.3818, + "step": 2141184 + }, + { + "epoch": 0.0, + "learning_rate": 1.4936453517282587e-05, + "loss": 3.3971, + "step": 2141696 + }, + { + "epoch": 0.0, + "learning_rate": 1.4928067569772067e-05, + "loss": 3.3922, + "step": 2142208 + }, + { + "epoch": 0.0, + "learning_rate": 1.4919681622261547e-05, + "loss": 3.3914, + "step": 2142720 + }, + { + "epoch": 0.0, + "learning_rate": 1.4911295674751027e-05, + "loss": 3.3944, + "step": 2143232 + }, + { + "epoch": 0.0, + "learning_rate": 1.4902926106044238e-05, + "loss": 3.3766, + "step": 2143744 + }, + { + "epoch": 0.0, + "learning_rate": 1.4894540158533718e-05, + "loss": 3.3784, + "step": 2144256 + }, + { + "epoch": 0.0, + "learning_rate": 1.4886154211023198e-05, + "loss": 3.3733, + "step": 2144768 + }, + { + "epoch": 0.0, + "learning_rate": 1.487776826351268e-05, + "loss": 3.3887, + "step": 2145280 + }, + { + "epoch": 0.0, + "learning_rate": 1.4869382316002156e-05, + "loss": 3.3896, + "step": 2145792 + }, + { + "epoch": 0.0, + "learning_rate": 1.4860996368491636e-05, + "loss": 3.383, + "step": 2146304 + }, + { + "epoch": 0.0, + "learning_rate": 1.4852610420981117e-05, + "loss": 3.3886, + "step": 2146816 + }, + { + "epoch": 0.0, + "learning_rate": 1.484424085227433e-05, + "loss": 3.4067, + "step": 2147328 + }, + { + "epoch": 0.0, + "learning_rate": 1.4835854904763808e-05, + "loss": 3.3873, + "step": 2147840 + }, + { + "epoch": 0.0, + "learning_rate": 1.4827468957253288e-05, + "loss": 3.3912, + "step": 2148352 + }, + { + "epoch": 0.0, + "learning_rate": 1.4819083009742768e-05, + "loss": 3.3886, + "step": 2148864 + }, + { + "epoch": 0.0, + "learning_rate": 1.4810697062232248e-05, + "loss": 3.3963, + "step": 2149376 + }, + { + "epoch": 0.0, + "learning_rate": 1.4802327493525459e-05, + "loss": 3.3881, + "step": 2149888 + }, + { + "epoch": 0.0, + "learning_rate": 1.4793941546014939e-05, + "loss": 3.3848, + "step": 2150400 + }, + { + "epoch": 0.0, + "learning_rate": 1.4785555598504419e-05, + "loss": 3.3822, + "step": 2150912 + }, + { + "epoch": 0.0, + "learning_rate": 1.4777169650993899e-05, + "loss": 3.3863, + "step": 2151424 + }, + { + "epoch": 0.0, + "learning_rate": 1.476878370348338e-05, + "loss": 3.378, + "step": 2151936 + }, + { + "epoch": 0.01, + "learning_rate": 1.476041413477659e-05, + "loss": 3.3862, + "step": 2152448 + }, + { + "epoch": 0.01, + "learning_rate": 1.4752028187266071e-05, + "loss": 3.3871, + "step": 2152960 + }, + { + "epoch": 0.01, + "learning_rate": 1.4743642239755551e-05, + "loss": 3.3932, + "step": 2153472 + }, + { + "epoch": 0.01, + "learning_rate": 1.4735256292245031e-05, + "loss": 3.3945, + "step": 2153984 + }, + { + "epoch": 0.01, + "learning_rate": 1.4726886723538242e-05, + "loss": 3.3805, + "step": 2154496 + }, + { + "epoch": 0.01, + "learning_rate": 1.4718500776027722e-05, + "loss": 3.3847, + "step": 2155008 + }, + { + "epoch": 0.01, + "learning_rate": 1.4710114828517202e-05, + "loss": 3.3931, + "step": 2155520 + }, + { + "epoch": 0.01, + "learning_rate": 1.4701728881006682e-05, + "loss": 3.3731, + "step": 2156032 + }, + { + "epoch": 0.01, + "learning_rate": 1.4693359312299892e-05, + "loss": 3.3864, + "step": 2156544 + }, + { + "epoch": 0.01, + "learning_rate": 1.4684973364789372e-05, + "loss": 3.3759, + "step": 2157056 + }, + { + "epoch": 0.01, + "learning_rate": 1.4676587417278852e-05, + "loss": 3.3803, + "step": 2157568 + }, + { + "epoch": 0.01, + "learning_rate": 1.4668201469768334e-05, + "loss": 3.3848, + "step": 2158080 + }, + { + "epoch": 0.01, + "learning_rate": 1.465981552225781e-05, + "loss": 3.3876, + "step": 2158592 + }, + { + "epoch": 0.01, + "learning_rate": 1.465142957474729e-05, + "loss": 3.3906, + "step": 2159104 + }, + { + "epoch": 0.01, + "learning_rate": 1.4643060006040505e-05, + "loss": 3.3846, + "step": 2159616 + }, + { + "epoch": 0.01, + "learning_rate": 1.4634674058529985e-05, + "loss": 3.3924, + "step": 2160128 + }, + { + "epoch": 0.01, + "learning_rate": 1.4626288111019463e-05, + "loss": 3.3925, + "step": 2160640 + }, + { + "epoch": 0.01, + "learning_rate": 1.4617902163508943e-05, + "loss": 3.3843, + "step": 2161152 + }, + { + "epoch": 0.01, + "learning_rate": 1.4609516215998423e-05, + "loss": 3.3587, + "step": 2161664 + }, + { + "epoch": 0.01, + "learning_rate": 1.4601146647291635e-05, + "loss": 3.3931, + "step": 2162176 + }, + { + "epoch": 0.01, + "learning_rate": 1.4592760699781114e-05, + "loss": 3.388, + "step": 2162688 + }, + { + "epoch": 0.01, + "learning_rate": 1.4584374752270594e-05, + "loss": 3.3935, + "step": 2163200 + }, + { + "epoch": 0.01, + "learning_rate": 1.4575988804760073e-05, + "loss": 3.3859, + "step": 2163712 + }, + { + "epoch": 0.01, + "learning_rate": 1.4567602857249555e-05, + "loss": 3.3787, + "step": 2164224 + }, + { + "epoch": 0.01, + "learning_rate": 1.4559233288542764e-05, + "loss": 3.3737, + "step": 2164736 + }, + { + "epoch": 0.01, + "learning_rate": 1.4550847341032244e-05, + "loss": 3.3753, + "step": 2165248 + }, + { + "epoch": 0.01, + "learning_rate": 1.4542461393521726e-05, + "loss": 3.3819, + "step": 2165760 + }, + { + "epoch": 0.01, + "learning_rate": 1.4534075446011206e-05, + "loss": 3.3675, + "step": 2166272 + }, + { + "epoch": 0.01, + "learning_rate": 1.4525689498500686e-05, + "loss": 3.4021, + "step": 2166784 + }, + { + "epoch": 0.01, + "learning_rate": 1.4517303550990166e-05, + "loss": 3.3842, + "step": 2167296 + }, + { + "epoch": 0.01, + "learning_rate": 1.4508933982283376e-05, + "loss": 3.3642, + "step": 2167808 + }, + { + "epoch": 0.01, + "learning_rate": 1.4500548034772856e-05, + "loss": 3.3891, + "step": 2168320 + }, + { + "epoch": 0.01, + "learning_rate": 1.4492162087262336e-05, + "loss": 3.3721, + "step": 2168832 + }, + { + "epoch": 0.01, + "learning_rate": 1.4483776139751815e-05, + "loss": 3.3731, + "step": 2169344 + }, + { + "epoch": 0.01, + "learning_rate": 1.4475406571045027e-05, + "loss": 3.3811, + "step": 2169856 + }, + { + "epoch": 0.01, + "learning_rate": 1.4467020623534507e-05, + "loss": 3.3854, + "step": 2170368 + }, + { + "epoch": 0.01, + "learning_rate": 1.4458634676023989e-05, + "loss": 3.3698, + "step": 2170880 + }, + { + "epoch": 0.01, + "learning_rate": 1.4450248728513465e-05, + "loss": 3.3836, + "step": 2171392 + }, + { + "epoch": 0.01, + "learning_rate": 1.4441862781002945e-05, + "loss": 3.3702, + "step": 2171904 + }, + { + "epoch": 0.01, + "learning_rate": 1.4433476833492427e-05, + "loss": 3.3766, + "step": 2172416 + }, + { + "epoch": 0.01, + "learning_rate": 1.442510726478564e-05, + "loss": 3.3874, + "step": 2172928 + }, + { + "epoch": 0.01, + "learning_rate": 1.4416721317275118e-05, + "loss": 3.3807, + "step": 2173440 + }, + { + "epoch": 0.01, + "learning_rate": 1.4408335369764598e-05, + "loss": 3.3823, + "step": 2173952 + }, + { + "epoch": 0.01, + "learning_rate": 1.4399949422254078e-05, + "loss": 3.3785, + "step": 2174464 + }, + { + "epoch": 0.01, + "learning_rate": 1.4391563474743558e-05, + "loss": 3.4017, + "step": 2174976 + }, + { + "epoch": 0.01, + "learning_rate": 1.4383177527233038e-05, + "loss": 3.3685, + "step": 2175488 + }, + { + "epoch": 0.01, + "learning_rate": 1.4374807958526248e-05, + "loss": 3.3936, + "step": 2176000 + }, + { + "epoch": 0.01, + "learning_rate": 1.4366422011015728e-05, + "loss": 3.3776, + "step": 2176512 + }, + { + "epoch": 0.01, + "learning_rate": 1.435803606350521e-05, + "loss": 3.3759, + "step": 2177024 + }, + { + "epoch": 0.01, + "learning_rate": 1.434965011599469e-05, + "loss": 3.386, + "step": 2177536 + }, + { + "epoch": 0.01, + "learning_rate": 1.434126416848417e-05, + "loss": 3.381, + "step": 2178048 + }, + { + "epoch": 0.01, + "learning_rate": 1.433289459977738e-05, + "loss": 3.3953, + "step": 2178560 + }, + { + "epoch": 0.01, + "learning_rate": 1.432450865226686e-05, + "loss": 3.3902, + "step": 2179072 + }, + { + "epoch": 0.01, + "learning_rate": 1.431612270475634e-05, + "loss": 3.3658, + "step": 2179584 + }, + { + "epoch": 0.01, + "learning_rate": 1.430773675724582e-05, + "loss": 3.3653, + "step": 2180096 + }, + { + "epoch": 0.01, + "learning_rate": 1.4299367188539031e-05, + "loss": 3.3748, + "step": 2180608 + }, + { + "epoch": 0.01, + "learning_rate": 1.4290981241028511e-05, + "loss": 3.387, + "step": 2181120 + }, + { + "epoch": 0.01, + "learning_rate": 1.4282595293517991e-05, + "loss": 3.3755, + "step": 2181632 + }, + { + "epoch": 0.01, + "learning_rate": 1.427420934600747e-05, + "loss": 3.374, + "step": 2182144 + }, + { + "epoch": 0.01, + "learning_rate": 1.426582339849695e-05, + "loss": 3.3842, + "step": 2182656 + }, + { + "epoch": 0.02, + "learning_rate": 1.4257453829790164e-05, + "loss": 3.3789, + "step": 2183168 + }, + { + "epoch": 0.02, + "learning_rate": 1.4249067882279644e-05, + "loss": 3.3832, + "step": 2183680 + }, + { + "epoch": 0.02, + "learning_rate": 1.424068193476912e-05, + "loss": 3.3738, + "step": 2184192 + }, + { + "epoch": 0.02, + "learning_rate": 1.4232295987258602e-05, + "loss": 3.3807, + "step": 2184704 + }, + { + "epoch": 0.02, + "learning_rate": 1.4223926418551814e-05, + "loss": 3.3852, + "step": 2185216 + }, + { + "epoch": 0.02, + "learning_rate": 1.4215540471041294e-05, + "loss": 3.3887, + "step": 2185728 + }, + { + "epoch": 0.02, + "learning_rate": 1.4207154523530772e-05, + "loss": 3.369, + "step": 2186240 + }, + { + "epoch": 0.02, + "learning_rate": 1.4198768576020252e-05, + "loss": 3.3718, + "step": 2186752 + }, + { + "epoch": 0.02, + "learning_rate": 1.4190399007313465e-05, + "loss": 3.3837, + "step": 2187264 + }, + { + "epoch": 0.02, + "learning_rate": 1.4182013059802943e-05, + "loss": 3.3805, + "step": 2187776 + }, + { + "epoch": 0.02, + "learning_rate": 1.4173627112292423e-05, + "loss": 3.3864, + "step": 2188288 + }, + { + "epoch": 0.02, + "learning_rate": 1.4165241164781903e-05, + "loss": 3.3791, + "step": 2188800 + }, + { + "epoch": 0.02, + "learning_rate": 1.4156855217271383e-05, + "loss": 3.3745, + "step": 2189312 + }, + { + "epoch": 0.02, + "learning_rate": 1.4148469269760865e-05, + "loss": 3.3774, + "step": 2189824 + }, + { + "epoch": 0.02, + "learning_rate": 1.4140099701054074e-05, + "loss": 3.3769, + "step": 2190336 + }, + { + "epoch": 0.02, + "learning_rate": 1.4131713753543554e-05, + "loss": 3.3754, + "step": 2190848 + }, + { + "epoch": 0.02, + "learning_rate": 1.4123327806033035e-05, + "loss": 3.3765, + "step": 2191360 + }, + { + "epoch": 0.02, + "learning_rate": 1.4114941858522515e-05, + "loss": 3.3805, + "step": 2191872 + }, + { + "epoch": 0.02, + "learning_rate": 1.4106555911011995e-05, + "loss": 3.3772, + "step": 2192384 + }, + { + "epoch": 0.02, + "learning_rate": 1.4098169963501475e-05, + "loss": 3.3728, + "step": 2192896 + }, + { + "epoch": 0.02, + "learning_rate": 1.4089800394794686e-05, + "loss": 3.3773, + "step": 2193408 + }, + { + "epoch": 0.02, + "learning_rate": 1.4081414447284166e-05, + "loss": 3.3654, + "step": 2193920 + }, + { + "epoch": 0.02, + "learning_rate": 1.4073028499773646e-05, + "loss": 3.3857, + "step": 2194432 + }, + { + "epoch": 0.02, + "learning_rate": 1.4064642552263124e-05, + "loss": 3.3851, + "step": 2194944 + }, + { + "epoch": 0.02, + "learning_rate": 1.4056256604752604e-05, + "loss": 3.3829, + "step": 2195456 + }, + { + "epoch": 0.02, + "learning_rate": 1.4047887036045818e-05, + "loss": 3.3895, + "step": 2195968 + }, + { + "epoch": 0.02, + "learning_rate": 1.4039501088535298e-05, + "loss": 3.3857, + "step": 2196480 + }, + { + "epoch": 0.02, + "learning_rate": 1.4031115141024775e-05, + "loss": 3.3782, + "step": 2196992 + }, + { + "epoch": 0.02, + "learning_rate": 1.4022729193514256e-05, + "loss": 3.3762, + "step": 2197504 + }, + { + "epoch": 0.02, + "learning_rate": 1.4014343246003736e-05, + "loss": 3.3847, + "step": 2198016 + }, + { + "epoch": 0.02, + "learning_rate": 1.4005957298493216e-05, + "loss": 3.3774, + "step": 2198528 + }, + { + "epoch": 0.02, + "learning_rate": 1.3997587729786427e-05, + "loss": 3.3801, + "step": 2199040 + }, + { + "epoch": 0.02, + "learning_rate": 1.3989201782275907e-05, + "loss": 3.3857, + "step": 2199552 + }, + { + "epoch": 0.02, + "learning_rate": 1.3980815834765387e-05, + "loss": 3.3747, + "step": 2200064 + }, + { + "epoch": 0.02, + "learning_rate": 1.3972429887254867e-05, + "loss": 3.3825, + "step": 2200576 + }, + { + "epoch": 0.02, + "learning_rate": 1.3964043939744349e-05, + "loss": 3.3752, + "step": 2201088 + }, + { + "epoch": 0.02, + "learning_rate": 1.3955657992233829e-05, + "loss": 3.3796, + "step": 2201600 + }, + { + "epoch": 0.02, + "learning_rate": 1.3947272044723305e-05, + "loss": 3.3788, + "step": 2202112 + }, + { + "epoch": 0.02, + "learning_rate": 1.3938886097212787e-05, + "loss": 3.3751, + "step": 2202624 + }, + { + "epoch": 0.02, + "learning_rate": 1.3930532907309729e-05, + "loss": 3.395, + "step": 2203136 + }, + { + "epoch": 0.02, + "learning_rate": 1.392214695979921e-05, + "loss": 3.3806, + "step": 2203648 + }, + { + "epoch": 0.02, + "learning_rate": 1.391376101228869e-05, + "loss": 3.3862, + "step": 2204160 + }, + { + "epoch": 0.02, + "learning_rate": 1.390537506477817e-05, + "loss": 3.3769, + "step": 2204672 + }, + { + "epoch": 0.02, + "learning_rate": 1.389698911726765e-05, + "loss": 3.3803, + "step": 2205184 + }, + { + "epoch": 0.02, + "learning_rate": 1.3888603169757128e-05, + "loss": 3.3839, + "step": 2205696 + }, + { + "epoch": 0.02, + "learning_rate": 1.3880217222246608e-05, + "loss": 3.3851, + "step": 2206208 + }, + { + "epoch": 0.02, + "learning_rate": 1.3871831274736088e-05, + "loss": 3.3747, + "step": 2206720 + }, + { + "epoch": 0.02, + "learning_rate": 1.38634617060293e-05, + "loss": 3.3724, + "step": 2207232 + }, + { + "epoch": 0.02, + "learning_rate": 1.3855075758518779e-05, + "loss": 3.3843, + "step": 2207744 + }, + { + "epoch": 0.02, + "learning_rate": 1.3846689811008259e-05, + "loss": 3.3992, + "step": 2208256 + }, + { + "epoch": 0.02, + "learning_rate": 1.3838303863497739e-05, + "loss": 3.3817, + "step": 2208768 + }, + { + "epoch": 0.02, + "learning_rate": 1.382991791598722e-05, + "loss": 3.3842, + "step": 2209280 + }, + { + "epoch": 0.02, + "learning_rate": 1.382154834728043e-05, + "loss": 3.3889, + "step": 2209792 + }, + { + "epoch": 0.02, + "learning_rate": 1.3813162399769911e-05, + "loss": 3.38, + "step": 2210304 + }, + { + "epoch": 0.02, + "learning_rate": 1.3804776452259391e-05, + "loss": 3.3809, + "step": 2210816 + }, + { + "epoch": 0.02, + "learning_rate": 1.3796390504748871e-05, + "loss": 3.3792, + "step": 2211328 + }, + { + "epoch": 0.02, + "learning_rate": 1.3788004557238351e-05, + "loss": 3.3825, + "step": 2211840 + }, + { + "epoch": 0.02, + "learning_rate": 1.3779634988531562e-05, + "loss": 3.3818, + "step": 2212352 + }, + { + "epoch": 0.02, + "learning_rate": 1.3771249041021042e-05, + "loss": 3.3754, + "step": 2212864 + }, + { + "epoch": 0.03, + "eval_loss": 3.8660366535186768, + "eval_runtime": 304.4635, + "eval_samples_per_second": 1253.322, + "eval_steps_per_second": 39.167, + "step": 2213280 + }, + { + "epoch": 1.0, + "learning_rate": 1.3762863093510522e-05, + "loss": 3.3799, + "step": 2213376 + }, + { + "epoch": 1.0, + "learning_rate": 1.3754477146000003e-05, + "loss": 3.3634, + "step": 2213888 + }, + { + "epoch": 1.0, + "learning_rate": 1.3746107577293213e-05, + "loss": 3.3785, + "step": 2214400 + }, + { + "epoch": 1.0, + "learning_rate": 1.3737721629782693e-05, + "loss": 3.3996, + "step": 2214912 + }, + { + "epoch": 1.0, + "learning_rate": 1.3729335682272174e-05, + "loss": 3.3856, + "step": 2215424 + }, + { + "epoch": 1.0, + "learning_rate": 1.3720949734761654e-05, + "loss": 3.3787, + "step": 2215936 + }, + { + "epoch": 1.0, + "learning_rate": 1.3712563787251134e-05, + "loss": 3.383, + "step": 2216448 + }, + { + "epoch": 1.0, + "learning_rate": 1.3704177839740612e-05, + "loss": 3.3741, + "step": 2216960 + }, + { + "epoch": 1.0, + "learning_rate": 1.3695808271033825e-05, + "loss": 3.3711, + "step": 2217472 + }, + { + "epoch": 1.0, + "learning_rate": 1.3687422323523305e-05, + "loss": 3.3894, + "step": 2217984 + }, + { + "epoch": 1.0, + "learning_rate": 1.3679036376012783e-05, + "loss": 3.3779, + "step": 2218496 + }, + { + "epoch": 1.0, + "learning_rate": 1.3670650428502263e-05, + "loss": 3.3816, + "step": 2219008 + }, + { + "epoch": 1.0, + "learning_rate": 1.3662280859795475e-05, + "loss": 3.3857, + "step": 2219520 + }, + { + "epoch": 1.0, + "learning_rate": 1.3653894912284957e-05, + "loss": 3.3665, + "step": 2220032 + }, + { + "epoch": 1.0, + "learning_rate": 1.3645508964774434e-05, + "loss": 3.3666, + "step": 2220544 + }, + { + "epoch": 1.0, + "learning_rate": 1.3637123017263914e-05, + "loss": 3.3638, + "step": 2221056 + }, + { + "epoch": 1.0, + "learning_rate": 1.3628737069753395e-05, + "loss": 3.38, + "step": 2221568 + }, + { + "epoch": 1.0, + "learning_rate": 1.3620351122242875e-05, + "loss": 3.3779, + "step": 2222080 + }, + { + "epoch": 1.0, + "learning_rate": 1.3611965174732355e-05, + "loss": 3.3747, + "step": 2222592 + }, + { + "epoch": 1.0, + "learning_rate": 1.3603579227221835e-05, + "loss": 3.3766, + "step": 2223104 + }, + { + "epoch": 1.0, + "learning_rate": 1.3595226037318779e-05, + "loss": 3.3966, + "step": 2223616 + }, + { + "epoch": 1.0, + "learning_rate": 1.3586840089808257e-05, + "loss": 3.3799, + "step": 2224128 + }, + { + "epoch": 1.0, + "learning_rate": 1.3578454142297737e-05, + "loss": 3.38, + "step": 2224640 + }, + { + "epoch": 1.0, + "learning_rate": 1.3570068194787217e-05, + "loss": 3.3802, + "step": 2225152 + }, + { + "epoch": 1.0, + "learning_rate": 1.3561682247276697e-05, + "loss": 3.3897, + "step": 2225664 + }, + { + "epoch": 1.0, + "learning_rate": 1.3553312678569907e-05, + "loss": 3.3746, + "step": 2226176 + }, + { + "epoch": 1.0, + "learning_rate": 1.3544926731059387e-05, + "loss": 3.3749, + "step": 2226688 + }, + { + "epoch": 1.0, + "learning_rate": 1.3536540783548867e-05, + "loss": 3.3796, + "step": 2227200 + }, + { + "epoch": 1.0, + "learning_rate": 1.3528154836038347e-05, + "loss": 3.3728, + "step": 2227712 + }, + { + "epoch": 1.0, + "learning_rate": 1.3519768888527829e-05, + "loss": 3.3691, + "step": 2228224 + }, + { + "epoch": 1.01, + "learning_rate": 1.3511399319821038e-05, + "loss": 3.3776, + "step": 2228736 + }, + { + "epoch": 1.01, + "learning_rate": 1.350301337231052e-05, + "loss": 3.3736, + "step": 2229248 + }, + { + "epoch": 1.01, + "learning_rate": 1.34946274248e-05, + "loss": 3.3865, + "step": 2229760 + }, + { + "epoch": 1.01, + "learning_rate": 1.348624147728948e-05, + "loss": 3.3796, + "step": 2230272 + }, + { + "epoch": 1.01, + "learning_rate": 1.347787190858269e-05, + "loss": 3.3742, + "step": 2230784 + }, + { + "epoch": 1.01, + "learning_rate": 1.346948596107217e-05, + "loss": 3.3777, + "step": 2231296 + }, + { + "epoch": 1.01, + "learning_rate": 1.346110001356165e-05, + "loss": 3.3804, + "step": 2231808 + }, + { + "epoch": 1.01, + "learning_rate": 1.345271406605113e-05, + "loss": 3.3635, + "step": 2232320 + }, + { + "epoch": 1.01, + "learning_rate": 1.3444344497344341e-05, + "loss": 3.3775, + "step": 2232832 + }, + { + "epoch": 1.01, + "learning_rate": 1.3435958549833821e-05, + "loss": 3.3683, + "step": 2233344 + }, + { + "epoch": 1.01, + "learning_rate": 1.3427572602323301e-05, + "loss": 3.3691, + "step": 2233856 + }, + { + "epoch": 1.01, + "learning_rate": 1.3419186654812783e-05, + "loss": 3.3757, + "step": 2234368 + }, + { + "epoch": 1.01, + "learning_rate": 1.3410800707302263e-05, + "loss": 3.3815, + "step": 2234880 + }, + { + "epoch": 1.01, + "learning_rate": 1.3402414759791739e-05, + "loss": 3.3782, + "step": 2235392 + }, + { + "epoch": 1.01, + "learning_rate": 1.3394045191084953e-05, + "loss": 3.3758, + "step": 2235904 + }, + { + "epoch": 1.01, + "learning_rate": 1.3385659243574433e-05, + "loss": 3.3814, + "step": 2236416 + }, + { + "epoch": 1.01, + "learning_rate": 1.3377273296063912e-05, + "loss": 3.381, + "step": 2236928 + }, + { + "epoch": 1.01, + "learning_rate": 1.3368887348553391e-05, + "loss": 3.3766, + "step": 2237440 + }, + { + "epoch": 1.01, + "learning_rate": 1.3360501401042871e-05, + "loss": 3.3534, + "step": 2237952 + }, + { + "epoch": 1.01, + "learning_rate": 1.3352131832336084e-05, + "loss": 3.3807, + "step": 2238464 + }, + { + "epoch": 1.01, + "learning_rate": 1.3343745884825562e-05, + "loss": 3.381, + "step": 2238976 + }, + { + "epoch": 1.01, + "learning_rate": 1.3335359937315042e-05, + "loss": 3.3798, + "step": 2239488 + }, + { + "epoch": 1.01, + "learning_rate": 1.3326973989804522e-05, + "loss": 3.3759, + "step": 2240000 + }, + { + "epoch": 1.01, + "learning_rate": 1.3318588042294004e-05, + "loss": 3.3696, + "step": 2240512 + }, + { + "epoch": 1.01, + "learning_rate": 1.3310218473587213e-05, + "loss": 3.3646, + "step": 2241024 + }, + { + "epoch": 1.01, + "learning_rate": 1.3301832526076693e-05, + "loss": 3.3618, + "step": 2241536 + }, + { + "epoch": 1.01, + "learning_rate": 1.3293446578566174e-05, + "loss": 3.3761, + "step": 2242048 + }, + { + "epoch": 1.01, + "learning_rate": 1.3285060631055654e-05, + "loss": 3.3614, + "step": 2242560 + }, + { + "epoch": 1.01, + "learning_rate": 1.3276674683545134e-05, + "loss": 3.3848, + "step": 2243072 + }, + { + "epoch": 1.01, + "learning_rate": 1.3268288736034614e-05, + "loss": 3.3756, + "step": 2243584 + }, + { + "epoch": 1.01, + "learning_rate": 1.3259919167327825e-05, + "loss": 3.3595, + "step": 2244096 + }, + { + "epoch": 1.01, + "learning_rate": 1.3251533219817305e-05, + "loss": 3.3765, + "step": 2244608 + }, + { + "epoch": 1.01, + "learning_rate": 1.3243147272306785e-05, + "loss": 3.3592, + "step": 2245120 + }, + { + "epoch": 1.01, + "learning_rate": 1.3234761324796267e-05, + "loss": 3.3664, + "step": 2245632 + }, + { + "epoch": 1.01, + "learning_rate": 1.3226391756089476e-05, + "loss": 3.367, + "step": 2246144 + }, + { + "epoch": 1.01, + "learning_rate": 1.3218005808578956e-05, + "loss": 3.3783, + "step": 2246656 + }, + { + "epoch": 1.01, + "learning_rate": 1.3209619861068437e-05, + "loss": 3.3625, + "step": 2247168 + }, + { + "epoch": 1.01, + "learning_rate": 1.3201233913557914e-05, + "loss": 3.3716, + "step": 2247680 + }, + { + "epoch": 1.01, + "learning_rate": 1.3192847966047394e-05, + "loss": 3.3604, + "step": 2248192 + }, + { + "epoch": 1.01, + "learning_rate": 1.3184462018536876e-05, + "loss": 3.3697, + "step": 2248704 + }, + { + "epoch": 1.01, + "learning_rate": 1.3176076071026355e-05, + "loss": 3.3758, + "step": 2249216 + }, + { + "epoch": 1.01, + "learning_rate": 1.3167706502319566e-05, + "loss": 3.3731, + "step": 2249728 + }, + { + "epoch": 1.01, + "learning_rate": 1.3159320554809046e-05, + "loss": 3.3704, + "step": 2250240 + }, + { + "epoch": 1.01, + "learning_rate": 1.3150934607298526e-05, + "loss": 3.3732, + "step": 2250752 + }, + { + "epoch": 1.01, + "learning_rate": 1.3142548659788006e-05, + "loss": 3.389, + "step": 2251264 + }, + { + "epoch": 1.01, + "learning_rate": 1.3134162712277486e-05, + "loss": 3.359, + "step": 2251776 + }, + { + "epoch": 1.01, + "learning_rate": 1.3125776764766968e-05, + "loss": 3.3855, + "step": 2252288 + }, + { + "epoch": 1.01, + "learning_rate": 1.3117390817256448e-05, + "loss": 3.3685, + "step": 2252800 + }, + { + "epoch": 1.01, + "learning_rate": 1.3109004869745924e-05, + "loss": 3.366, + "step": 2253312 + }, + { + "epoch": 1.01, + "learning_rate": 1.3100635301039138e-05, + "loss": 3.3723, + "step": 2253824 + }, + { + "epoch": 1.01, + "learning_rate": 1.3092249353528618e-05, + "loss": 3.3733, + "step": 2254336 + }, + { + "epoch": 1.01, + "learning_rate": 1.3083863406018097e-05, + "loss": 3.3794, + "step": 2254848 + }, + { + "epoch": 1.01, + "learning_rate": 1.307549383731131e-05, + "loss": 3.3832, + "step": 2255360 + }, + { + "epoch": 1.01, + "learning_rate": 1.3067107889800789e-05, + "loss": 3.36, + "step": 2255872 + }, + { + "epoch": 1.01, + "learning_rate": 1.3058721942290269e-05, + "loss": 3.3559, + "step": 2256384 + }, + { + "epoch": 1.01, + "learning_rate": 1.3050335994779747e-05, + "loss": 3.365, + "step": 2256896 + }, + { + "epoch": 1.01, + "learning_rate": 1.304196642607296e-05, + "loss": 3.376, + "step": 2257408 + }, + { + "epoch": 1.01, + "learning_rate": 1.303358047856244e-05, + "loss": 3.3664, + "step": 2257920 + }, + { + "epoch": 1.01, + "learning_rate": 1.3025194531051921e-05, + "loss": 3.3658, + "step": 2258432 + }, + { + "epoch": 1.01, + "learning_rate": 1.3016808583541398e-05, + "loss": 3.375, + "step": 2258944 + }, + { + "epoch": 1.02, + "learning_rate": 1.3008439014834612e-05, + "loss": 3.3702, + "step": 2259456 + }, + { + "epoch": 1.02, + "learning_rate": 1.3000053067324092e-05, + "loss": 3.3761, + "step": 2259968 + }, + { + "epoch": 1.02, + "learning_rate": 1.2991667119813569e-05, + "loss": 3.3616, + "step": 2260480 + }, + { + "epoch": 1.02, + "learning_rate": 1.298328117230305e-05, + "loss": 3.3736, + "step": 2260992 + }, + { + "epoch": 1.02, + "learning_rate": 1.2974911603596263e-05, + "loss": 3.3735, + "step": 2261504 + }, + { + "epoch": 1.02, + "learning_rate": 1.2966525656085743e-05, + "loss": 3.3827, + "step": 2262016 + }, + { + "epoch": 1.02, + "learning_rate": 1.2958139708575221e-05, + "loss": 3.3564, + "step": 2262528 + }, + { + "epoch": 1.02, + "learning_rate": 1.2949753761064701e-05, + "loss": 3.3616, + "step": 2263040 + }, + { + "epoch": 1.02, + "learning_rate": 1.2941384192357913e-05, + "loss": 3.3779, + "step": 2263552 + }, + { + "epoch": 1.02, + "learning_rate": 1.2932998244847393e-05, + "loss": 3.3711, + "step": 2264064 + }, + { + "epoch": 1.02, + "learning_rate": 1.2924612297336872e-05, + "loss": 3.3746, + "step": 2264576 + }, + { + "epoch": 1.02, + "learning_rate": 1.2916226349826352e-05, + "loss": 3.3705, + "step": 2265088 + }, + { + "epoch": 1.02, + "learning_rate": 1.2907856781119564e-05, + "loss": 3.3651, + "step": 2265600 + }, + { + "epoch": 1.02, + "learning_rate": 1.2899470833609042e-05, + "loss": 3.3675, + "step": 2266112 + }, + { + "epoch": 1.02, + "learning_rate": 1.2891084886098522e-05, + "loss": 3.3703, + "step": 2266624 + }, + { + "epoch": 1.02, + "learning_rate": 1.2882698938588002e-05, + "loss": 3.3661, + "step": 2267136 + }, + { + "epoch": 1.02, + "learning_rate": 1.2874312991077484e-05, + "loss": 3.3639, + "step": 2267648 + }, + { + "epoch": 1.02, + "learning_rate": 1.2865927043566964e-05, + "loss": 3.3718, + "step": 2268160 + }, + { + "epoch": 1.02, + "learning_rate": 1.2857557474860175e-05, + "loss": 3.3655, + "step": 2268672 + }, + { + "epoch": 1.02, + "learning_rate": 1.2849171527349655e-05, + "loss": 3.3637, + "step": 2269184 + }, + { + "epoch": 1.02, + "learning_rate": 1.2840785579839135e-05, + "loss": 3.3697, + "step": 2269696 + }, + { + "epoch": 1.02, + "learning_rate": 1.2832399632328615e-05, + "loss": 3.355, + "step": 2270208 + }, + { + "epoch": 1.02, + "learning_rate": 1.2824030063621825e-05, + "loss": 3.3786, + "step": 2270720 + }, + { + "epoch": 1.02, + "learning_rate": 1.2815644116111305e-05, + "loss": 3.3739, + "step": 2271232 + }, + { + "epoch": 1.02, + "learning_rate": 1.2807258168600785e-05, + "loss": 3.3737, + "step": 2271744 + }, + { + "epoch": 1.02, + "learning_rate": 1.2798872221090267e-05, + "loss": 3.3815, + "step": 2272256 + }, + { + "epoch": 1.02, + "learning_rate": 1.2790486273579747e-05, + "loss": 3.3757, + "step": 2272768 + }, + { + "epoch": 1.02, + "learning_rate": 1.2782116704872956e-05, + "loss": 3.3666, + "step": 2273280 + }, + { + "epoch": 1.02, + "learning_rate": 1.2773730757362438e-05, + "loss": 3.3656, + "step": 2273792 + }, + { + "epoch": 1.02, + "learning_rate": 1.2765344809851918e-05, + "loss": 3.374, + "step": 2274304 + }, + { + "epoch": 1.02, + "learning_rate": 1.2756958862341398e-05, + "loss": 3.3666, + "step": 2274816 + }, + { + "epoch": 1.02, + "learning_rate": 1.2748589293634608e-05, + "loss": 3.3679, + "step": 2275328 + }, + { + "epoch": 1.02, + "learning_rate": 1.2740203346124088e-05, + "loss": 3.3787, + "step": 2275840 + }, + { + "epoch": 1.02, + "learning_rate": 1.2731817398613568e-05, + "loss": 3.3678, + "step": 2276352 + }, + { + "epoch": 1.02, + "learning_rate": 1.2723431451103048e-05, + "loss": 3.3722, + "step": 2276864 + }, + { + "epoch": 1.02, + "learning_rate": 1.2715045503592526e-05, + "loss": 3.3682, + "step": 2277376 + }, + { + "epoch": 1.02, + "learning_rate": 1.2706675934885739e-05, + "loss": 3.3666, + "step": 2277888 + }, + { + "epoch": 1.02, + "learning_rate": 1.269828998737522e-05, + "loss": 3.3728, + "step": 2278400 + }, + { + "epoch": 1.02, + "learning_rate": 1.2689904039864697e-05, + "loss": 3.364, + "step": 2278912 + }, + { + "epoch": 1.02, + "learning_rate": 1.2681518092354177e-05, + "loss": 3.3777, + "step": 2279424 + }, + { + "epoch": 1.02, + "learning_rate": 1.2673132144843659e-05, + "loss": 3.3783, + "step": 2279936 + }, + { + "epoch": 1.02, + "learning_rate": 1.2664746197333139e-05, + "loss": 3.3741, + "step": 2280448 + }, + { + "epoch": 1.02, + "learning_rate": 1.2656360249822619e-05, + "loss": 3.3642, + "step": 2280960 + }, + { + "epoch": 1.02, + "learning_rate": 1.2647974302312099e-05, + "loss": 3.3753, + "step": 2281472 + }, + { + "epoch": 1.02, + "learning_rate": 1.263960473360531e-05, + "loss": 3.3724, + "step": 2281984 + }, + { + "epoch": 1.02, + "learning_rate": 1.263123516489852e-05, + "loss": 3.3714, + "step": 2282496 + }, + { + "epoch": 1.02, + "learning_rate": 1.2622849217388e-05, + "loss": 3.363, + "step": 2283008 + }, + { + "epoch": 1.02, + "learning_rate": 1.261446326987748e-05, + "loss": 3.3647, + "step": 2283520 + }, + { + "epoch": 1.02, + "learning_rate": 1.260607732236696e-05, + "loss": 3.3795, + "step": 2284032 + }, + { + "epoch": 1.02, + "learning_rate": 1.2597707753660171e-05, + "loss": 3.39, + "step": 2284544 + }, + { + "epoch": 1.02, + "learning_rate": 1.258932180614965e-05, + "loss": 3.3682, + "step": 2285056 + }, + { + "epoch": 1.02, + "learning_rate": 1.258093585863913e-05, + "loss": 3.3773, + "step": 2285568 + }, + { + "epoch": 1.02, + "learning_rate": 1.257254991112861e-05, + "loss": 3.378, + "step": 2286080 + }, + { + "epoch": 1.02, + "learning_rate": 1.2564180342421822e-05, + "loss": 3.3704, + "step": 2286592 + }, + { + "epoch": 1.02, + "learning_rate": 1.2555794394911301e-05, + "loss": 3.373, + "step": 2287104 + }, + { + "epoch": 1.02, + "learning_rate": 1.2547408447400783e-05, + "loss": 3.3688, + "step": 2287616 + }, + { + "epoch": 1.02, + "learning_rate": 1.2539022499890263e-05, + "loss": 3.3689, + "step": 2288128 + }, + { + "epoch": 1.02, + "learning_rate": 1.2530636552379743e-05, + "loss": 3.3801, + "step": 2288640 + }, + { + "epoch": 1.02, + "learning_rate": 1.2522266983672954e-05, + "loss": 3.3594, + "step": 2289152 + }, + { + "epoch": 1.03, + "eval_loss": 3.867037534713745, + "eval_runtime": 304.0161, + "eval_samples_per_second": 1255.167, + "eval_steps_per_second": 39.225, + "step": 2289600 + }, + { + "epoch": 0.0, + "learning_rate": 1.2513881036162434e-05, + "loss": 3.3787, + "step": 2289664 + }, + { + "epoch": 0.0, + "learning_rate": 1.2505495088651914e-05, + "loss": 3.3567, + "step": 2290176 + }, + { + "epoch": 0.0, + "learning_rate": 1.2497125519945126e-05, + "loss": 3.3681, + "step": 2290688 + }, + { + "epoch": 0.0, + "learning_rate": 1.2488739572434604e-05, + "loss": 3.3853, + "step": 2291200 + }, + { + "epoch": 0.0, + "learning_rate": 1.2480353624924084e-05, + "loss": 3.3796, + "step": 2291712 + }, + { + "epoch": 0.0, + "learning_rate": 1.2471967677413564e-05, + "loss": 3.3696, + "step": 2292224 + }, + { + "epoch": 0.0, + "learning_rate": 1.2463581729903044e-05, + "loss": 3.3697, + "step": 2292736 + }, + { + "epoch": 0.0, + "learning_rate": 1.2455195782392524e-05, + "loss": 3.3639, + "step": 2293248 + }, + { + "epoch": 0.0, + "learning_rate": 1.2446809834882004e-05, + "loss": 3.3612, + "step": 2293760 + }, + { + "epoch": 0.0, + "learning_rate": 1.2438440266175217e-05, + "loss": 3.3768, + "step": 2294272 + }, + { + "epoch": 0.0, + "learning_rate": 1.2430054318664695e-05, + "loss": 3.3711, + "step": 2294784 + }, + { + "epoch": 0.0, + "learning_rate": 1.2421668371154175e-05, + "loss": 3.3731, + "step": 2295296 + }, + { + "epoch": 0.0, + "learning_rate": 1.2413282423643657e-05, + "loss": 3.3752, + "step": 2295808 + }, + { + "epoch": 0.0, + "learning_rate": 1.2404912854936866e-05, + "loss": 3.3593, + "step": 2296320 + }, + { + "epoch": 0.0, + "learning_rate": 1.2396526907426347e-05, + "loss": 3.3562, + "step": 2296832 + }, + { + "epoch": 0.0, + "learning_rate": 1.2388140959915827e-05, + "loss": 3.3525, + "step": 2297344 + }, + { + "epoch": 0.0, + "learning_rate": 1.2379755012405307e-05, + "loss": 3.373, + "step": 2297856 + }, + { + "epoch": 0.0, + "learning_rate": 1.2371369064894786e-05, + "loss": 3.3624, + "step": 2298368 + }, + { + "epoch": 0.0, + "learning_rate": 1.2362983117384267e-05, + "loss": 3.365, + "step": 2298880 + }, + { + "epoch": 0.0, + "learning_rate": 1.2354597169873747e-05, + "loss": 3.3708, + "step": 2299392 + }, + { + "epoch": 0.0, + "learning_rate": 1.2346227601166956e-05, + "loss": 3.3846, + "step": 2299904 + }, + { + "epoch": 0.0, + "learning_rate": 1.2337841653656438e-05, + "loss": 3.3741, + "step": 2300416 + }, + { + "epoch": 0.0, + "learning_rate": 1.2329455706145918e-05, + "loss": 3.3674, + "step": 2300928 + }, + { + "epoch": 0.0, + "learning_rate": 1.2321069758635398e-05, + "loss": 3.3709, + "step": 2301440 + }, + { + "epoch": 0.0, + "learning_rate": 1.2312683811124876e-05, + "loss": 3.3792, + "step": 2301952 + }, + { + "epoch": 0.0, + "learning_rate": 1.2304314242418089e-05, + "loss": 3.3675, + "step": 2302464 + }, + { + "epoch": 0.0, + "learning_rate": 1.2295928294907569e-05, + "loss": 3.3622, + "step": 2302976 + }, + { + "epoch": 0.0, + "learning_rate": 1.2287542347397048e-05, + "loss": 3.3711, + "step": 2303488 + }, + { + "epoch": 0.0, + "learning_rate": 1.2279156399886528e-05, + "loss": 3.3599, + "step": 2304000 + }, + { + "epoch": 0.0, + "learning_rate": 1.2270770452376008e-05, + "loss": 3.3633, + "step": 2304512 + }, + { + "epoch": 0.01, + "learning_rate": 1.2262400883669221e-05, + "loss": 3.365, + "step": 2305024 + }, + { + "epoch": 0.01, + "learning_rate": 1.2254014936158699e-05, + "loss": 3.3694, + "step": 2305536 + }, + { + "epoch": 0.01, + "learning_rate": 1.2245628988648179e-05, + "loss": 3.3699, + "step": 2306048 + }, + { + "epoch": 0.01, + "learning_rate": 1.2237243041137659e-05, + "loss": 3.3732, + "step": 2306560 + }, + { + "epoch": 0.01, + "learning_rate": 1.2228873472430872e-05, + "loss": 3.366, + "step": 2307072 + }, + { + "epoch": 0.01, + "learning_rate": 1.222048752492035e-05, + "loss": 3.3665, + "step": 2307584 + }, + { + "epoch": 0.01, + "learning_rate": 1.221210157740983e-05, + "loss": 3.3732, + "step": 2308096 + }, + { + "epoch": 0.01, + "learning_rate": 1.2203715629899311e-05, + "loss": 3.355, + "step": 2308608 + }, + { + "epoch": 0.01, + "learning_rate": 1.219532968238879e-05, + "loss": 3.3706, + "step": 2309120 + }, + { + "epoch": 0.01, + "learning_rate": 1.2186960113682002e-05, + "loss": 3.3533, + "step": 2309632 + }, + { + "epoch": 0.01, + "learning_rate": 1.2178574166171482e-05, + "loss": 3.3598, + "step": 2310144 + }, + { + "epoch": 0.01, + "learning_rate": 1.2170188218660962e-05, + "loss": 3.362, + "step": 2310656 + }, + { + "epoch": 0.01, + "learning_rate": 1.216180227115044e-05, + "loss": 3.3771, + "step": 2311168 + }, + { + "epoch": 0.01, + "learning_rate": 1.2153416323639922e-05, + "loss": 3.3656, + "step": 2311680 + }, + { + "epoch": 0.01, + "learning_rate": 1.2145030376129402e-05, + "loss": 3.3628, + "step": 2312192 + }, + { + "epoch": 0.01, + "learning_rate": 1.213664442861888e-05, + "loss": 3.3749, + "step": 2312704 + }, + { + "epoch": 0.01, + "learning_rate": 1.2128274859912093e-05, + "loss": 3.3704, + "step": 2313216 + }, + { + "epoch": 0.01, + "learning_rate": 1.2119888912401573e-05, + "loss": 3.3675, + "step": 2313728 + }, + { + "epoch": 0.01, + "learning_rate": 1.2111519343694783e-05, + "loss": 3.3407, + "step": 2314240 + }, + { + "epoch": 0.01, + "learning_rate": 1.2103133396184263e-05, + "loss": 3.3688, + "step": 2314752 + }, + { + "epoch": 0.01, + "learning_rate": 1.2094747448673743e-05, + "loss": 3.3733, + "step": 2315264 + }, + { + "epoch": 0.01, + "learning_rate": 1.2086361501163223e-05, + "loss": 3.3719, + "step": 2315776 + }, + { + "epoch": 0.01, + "learning_rate": 1.2077991932456436e-05, + "loss": 3.3637, + "step": 2316288 + }, + { + "epoch": 0.01, + "learning_rate": 1.2069605984945914e-05, + "loss": 3.3555, + "step": 2316800 + }, + { + "epoch": 0.01, + "learning_rate": 1.2061220037435394e-05, + "loss": 3.3636, + "step": 2317312 + }, + { + "epoch": 0.01, + "learning_rate": 1.2052834089924876e-05, + "loss": 3.3476, + "step": 2317824 + }, + { + "epoch": 0.01, + "learning_rate": 1.2044448142414354e-05, + "loss": 3.3704, + "step": 2318336 + }, + { + "epoch": 0.01, + "learning_rate": 1.2036062194903834e-05, + "loss": 3.3486, + "step": 2318848 + }, + { + "epoch": 0.01, + "learning_rate": 1.2027676247393314e-05, + "loss": 3.372, + "step": 2319360 + }, + { + "epoch": 0.01, + "learning_rate": 1.2019306678686525e-05, + "loss": 3.3667, + "step": 2319872 + }, + { + "epoch": 0.01, + "learning_rate": 1.2010920731176005e-05, + "loss": 3.352, + "step": 2320384 + }, + { + "epoch": 0.01, + "learning_rate": 1.2002534783665484e-05, + "loss": 3.3684, + "step": 2320896 + }, + { + "epoch": 0.01, + "learning_rate": 1.1994148836154966e-05, + "loss": 3.3479, + "step": 2321408 + }, + { + "epoch": 0.01, + "learning_rate": 1.1985779267448175e-05, + "loss": 3.356, + "step": 2321920 + }, + { + "epoch": 0.01, + "learning_rate": 1.1977393319937657e-05, + "loss": 3.3558, + "step": 2322432 + }, + { + "epoch": 0.01, + "learning_rate": 1.1969007372427137e-05, + "loss": 3.369, + "step": 2322944 + }, + { + "epoch": 0.01, + "learning_rate": 1.1960621424916615e-05, + "loss": 3.3511, + "step": 2323456 + }, + { + "epoch": 0.01, + "learning_rate": 1.1952235477406095e-05, + "loss": 3.36, + "step": 2323968 + }, + { + "epoch": 0.01, + "learning_rate": 1.1943865908699308e-05, + "loss": 3.3504, + "step": 2324480 + }, + { + "epoch": 0.01, + "learning_rate": 1.1935479961188788e-05, + "loss": 3.3599, + "step": 2324992 + }, + { + "epoch": 0.01, + "learning_rate": 1.1927094013678267e-05, + "loss": 3.3674, + "step": 2325504 + }, + { + "epoch": 0.01, + "learning_rate": 1.1918708066167747e-05, + "loss": 3.3661, + "step": 2326016 + }, + { + "epoch": 0.01, + "learning_rate": 1.1910322118657227e-05, + "loss": 3.3598, + "step": 2326528 + }, + { + "epoch": 0.01, + "learning_rate": 1.1901952549950438e-05, + "loss": 3.3637, + "step": 2327040 + }, + { + "epoch": 0.01, + "learning_rate": 1.1893566602439918e-05, + "loss": 3.3768, + "step": 2327552 + }, + { + "epoch": 0.01, + "learning_rate": 1.1885180654929398e-05, + "loss": 3.3534, + "step": 2328064 + }, + { + "epoch": 0.01, + "learning_rate": 1.1876794707418878e-05, + "loss": 3.375, + "step": 2328576 + }, + { + "epoch": 0.01, + "learning_rate": 1.1868408759908358e-05, + "loss": 3.3579, + "step": 2329088 + }, + { + "epoch": 0.01, + "learning_rate": 1.1860022812397838e-05, + "loss": 3.3606, + "step": 2329600 + }, + { + "epoch": 0.01, + "learning_rate": 1.1851636864887318e-05, + "loss": 3.3606, + "step": 2330112 + }, + { + "epoch": 0.01, + "learning_rate": 1.184326729618053e-05, + "loss": 3.3622, + "step": 2330624 + }, + { + "epoch": 0.01, + "learning_rate": 1.1834881348670009e-05, + "loss": 3.3718, + "step": 2331136 + }, + { + "epoch": 0.01, + "learning_rate": 1.1826495401159489e-05, + "loss": 3.3709, + "step": 2331648 + }, + { + "epoch": 0.01, + "learning_rate": 1.1818109453648969e-05, + "loss": 3.3572, + "step": 2332160 + }, + { + "epoch": 0.01, + "learning_rate": 1.1809723506138449e-05, + "loss": 3.3412, + "step": 2332672 + }, + { + "epoch": 0.01, + "learning_rate": 1.180135393743166e-05, + "loss": 3.3561, + "step": 2333184 + }, + { + "epoch": 0.01, + "learning_rate": 1.1792967989921141e-05, + "loss": 3.3653, + "step": 2333696 + }, + { + "epoch": 0.01, + "learning_rate": 1.1784582042410621e-05, + "loss": 3.3594, + "step": 2334208 + }, + { + "epoch": 0.01, + "learning_rate": 1.17761960949001e-05, + "loss": 3.3547, + "step": 2334720 + }, + { + "epoch": 0.01, + "learning_rate": 1.1767810147389579e-05, + "loss": 3.3651, + "step": 2335232 + }, + { + "epoch": 0.02, + "learning_rate": 1.1759440578682792e-05, + "loss": 3.3623, + "step": 2335744 + }, + { + "epoch": 0.02, + "learning_rate": 1.175105463117227e-05, + "loss": 3.3633, + "step": 2336256 + }, + { + "epoch": 0.02, + "learning_rate": 1.174266868366175e-05, + "loss": 3.3549, + "step": 2336768 + }, + { + "epoch": 0.02, + "learning_rate": 1.1734282736151231e-05, + "loss": 3.3656, + "step": 2337280 + }, + { + "epoch": 0.02, + "learning_rate": 1.1725913167444442e-05, + "loss": 3.3592, + "step": 2337792 + }, + { + "epoch": 0.02, + "learning_rate": 1.1717527219933922e-05, + "loss": 3.3726, + "step": 2338304 + }, + { + "epoch": 0.02, + "learning_rate": 1.1709141272423402e-05, + "loss": 3.3512, + "step": 2338816 + }, + { + "epoch": 0.02, + "learning_rate": 1.1700755324912882e-05, + "loss": 3.3521, + "step": 2339328 + }, + { + "epoch": 0.02, + "learning_rate": 1.1692385756206093e-05, + "loss": 3.3689, + "step": 2339840 + }, + { + "epoch": 0.02, + "learning_rate": 1.1683999808695573e-05, + "loss": 3.3611, + "step": 2340352 + }, + { + "epoch": 0.02, + "learning_rate": 1.1675613861185053e-05, + "loss": 3.3648, + "step": 2340864 + }, + { + "epoch": 0.02, + "learning_rate": 1.1667227913674533e-05, + "loss": 3.3624, + "step": 2341376 + }, + { + "epoch": 0.02, + "learning_rate": 1.1658841966164013e-05, + "loss": 3.3552, + "step": 2341888 + }, + { + "epoch": 0.02, + "learning_rate": 1.1650472397457224e-05, + "loss": 3.3572, + "step": 2342400 + }, + { + "epoch": 0.02, + "learning_rate": 1.1642086449946703e-05, + "loss": 3.3633, + "step": 2342912 + }, + { + "epoch": 0.02, + "learning_rate": 1.1633700502436185e-05, + "loss": 3.3514, + "step": 2343424 + }, + { + "epoch": 0.02, + "learning_rate": 1.1625314554925663e-05, + "loss": 3.3554, + "step": 2343936 + }, + { + "epoch": 0.02, + "learning_rate": 1.1616944986218876e-05, + "loss": 3.3604, + "step": 2344448 + }, + { + "epoch": 0.02, + "learning_rate": 1.1608559038708356e-05, + "loss": 3.3573, + "step": 2344960 + }, + { + "epoch": 0.02, + "learning_rate": 1.1600173091197834e-05, + "loss": 3.351, + "step": 2345472 + }, + { + "epoch": 0.02, + "learning_rate": 1.1591787143687314e-05, + "loss": 3.3627, + "step": 2345984 + }, + { + "epoch": 0.02, + "learning_rate": 1.1583417574980527e-05, + "loss": 3.3472, + "step": 2346496 + }, + { + "epoch": 0.02, + "learning_rate": 1.1575031627470006e-05, + "loss": 3.3653, + "step": 2347008 + }, + { + "epoch": 0.02, + "learning_rate": 1.1566645679959485e-05, + "loss": 3.3627, + "step": 2347520 + }, + { + "epoch": 0.02, + "learning_rate": 1.1558259732448966e-05, + "loss": 3.3666, + "step": 2348032 + }, + { + "epoch": 0.02, + "learning_rate": 1.1549890163742177e-05, + "loss": 3.3691, + "step": 2348544 + }, + { + "epoch": 0.02, + "learning_rate": 1.1541504216231657e-05, + "loss": 3.3671, + "step": 2349056 + }, + { + "epoch": 0.02, + "learning_rate": 1.1533118268721137e-05, + "loss": 3.3574, + "step": 2349568 + }, + { + "epoch": 0.02, + "learning_rate": 1.1524732321210617e-05, + "loss": 3.3587, + "step": 2350080 + }, + { + "epoch": 0.02, + "learning_rate": 1.1516362752503828e-05, + "loss": 3.364, + "step": 2350592 + }, + { + "epoch": 0.02, + "learning_rate": 1.1507976804993308e-05, + "loss": 3.3573, + "step": 2351104 + }, + { + "epoch": 0.02, + "learning_rate": 1.1499590857482788e-05, + "loss": 3.3523, + "step": 2351616 + }, + { + "epoch": 0.02, + "learning_rate": 1.1491204909972268e-05, + "loss": 3.3695, + "step": 2352128 + }, + { + "epoch": 0.02, + "learning_rate": 1.148281896246175e-05, + "loss": 3.3584, + "step": 2352640 + }, + { + "epoch": 0.02, + "learning_rate": 1.1474433014951228e-05, + "loss": 3.364, + "step": 2353152 + }, + { + "epoch": 0.02, + "learning_rate": 1.1466047067440708e-05, + "loss": 3.3574, + "step": 2353664 + }, + { + "epoch": 0.02, + "learning_rate": 1.1457661119930188e-05, + "loss": 3.3556, + "step": 2354176 + }, + { + "epoch": 0.02, + "learning_rate": 1.1449291551223398e-05, + "loss": 3.3684, + "step": 2354688 + }, + { + "epoch": 0.02, + "learning_rate": 1.1440905603712878e-05, + "loss": 3.3567, + "step": 2355200 + }, + { + "epoch": 0.02, + "learning_rate": 1.143253603500609e-05, + "loss": 3.3619, + "step": 2355712 + }, + { + "epoch": 0.02, + "learning_rate": 1.142415008749557e-05, + "loss": 3.3701, + "step": 2356224 + }, + { + "epoch": 0.02, + "learning_rate": 1.1415764139985049e-05, + "loss": 3.3639, + "step": 2356736 + }, + { + "epoch": 0.02, + "learning_rate": 1.140737819247453e-05, + "loss": 3.358, + "step": 2357248 + }, + { + "epoch": 0.02, + "learning_rate": 1.139899224496401e-05, + "loss": 3.3669, + "step": 2357760 + }, + { + "epoch": 0.02, + "learning_rate": 1.1390606297453489e-05, + "loss": 3.3633, + "step": 2358272 + }, + { + "epoch": 0.02, + "learning_rate": 1.1382236728746701e-05, + "loss": 3.3611, + "step": 2358784 + }, + { + "epoch": 0.02, + "learning_rate": 1.1373850781236181e-05, + "loss": 3.3509, + "step": 2359296 + }, + { + "epoch": 0.02, + "learning_rate": 1.1365464833725661e-05, + "loss": 3.3576, + "step": 2359808 + }, + { + "epoch": 0.02, + "learning_rate": 1.135707888621514e-05, + "loss": 3.3729, + "step": 2360320 + }, + { + "epoch": 0.02, + "learning_rate": 1.1348709317508352e-05, + "loss": 3.3771, + "step": 2360832 + }, + { + "epoch": 0.02, + "learning_rate": 1.1340323369997832e-05, + "loss": 3.3636, + "step": 2361344 + }, + { + "epoch": 0.02, + "learning_rate": 1.1331937422487312e-05, + "loss": 3.3614, + "step": 2361856 + }, + { + "epoch": 0.02, + "learning_rate": 1.1323551474976792e-05, + "loss": 3.3701, + "step": 2362368 + }, + { + "epoch": 0.02, + "learning_rate": 1.1315181906270003e-05, + "loss": 3.3621, + "step": 2362880 + }, + { + "epoch": 0.02, + "learning_rate": 1.1306795958759484e-05, + "loss": 3.3614, + "step": 2363392 + }, + { + "epoch": 0.02, + "learning_rate": 1.1298410011248963e-05, + "loss": 3.3597, + "step": 2363904 + }, + { + "epoch": 0.02, + "learning_rate": 1.1290024063738443e-05, + "loss": 3.3589, + "step": 2364416 + }, + { + "epoch": 0.02, + "learning_rate": 1.1281638116227922e-05, + "loss": 3.3705, + "step": 2364928 + }, + { + "epoch": 0.02, + "learning_rate": 1.1273268547521135e-05, + "loss": 3.3494, + "step": 2365440 + }, + { + "epoch": 0.03, + "eval_loss": 3.868082284927368, + "eval_runtime": 305.0708, + "eval_samples_per_second": 1250.828, + "eval_steps_per_second": 39.089, + "step": 2365920 + }, + { + "epoch": 1.0, + "learning_rate": 1.1264882600010613e-05, + "loss": 3.3601, + "step": 2365952 + }, + { + "epoch": 1.0, + "learning_rate": 1.1256496652500093e-05, + "loss": 3.3507, + "step": 2366464 + }, + { + "epoch": 1.0, + "learning_rate": 1.1248110704989575e-05, + "loss": 3.3557, + "step": 2366976 + }, + { + "epoch": 1.0, + "learning_rate": 1.1239741136282784e-05, + "loss": 3.3756, + "step": 2367488 + }, + { + "epoch": 1.0, + "learning_rate": 1.1231355188772266e-05, + "loss": 3.3745, + "step": 2368000 + }, + { + "epoch": 1.0, + "learning_rate": 1.1222969241261746e-05, + "loss": 3.3579, + "step": 2368512 + }, + { + "epoch": 1.0, + "learning_rate": 1.1214583293751225e-05, + "loss": 3.3601, + "step": 2369024 + }, + { + "epoch": 1.0, + "learning_rate": 1.1206197346240704e-05, + "loss": 3.3567, + "step": 2369536 + }, + { + "epoch": 1.0, + "learning_rate": 1.1197827777533916e-05, + "loss": 3.3575, + "step": 2370048 + }, + { + "epoch": 1.0, + "learning_rate": 1.1189441830023396e-05, + "loss": 3.3655, + "step": 2370560 + }, + { + "epoch": 1.0, + "learning_rate": 1.1181055882512874e-05, + "loss": 3.3595, + "step": 2371072 + }, + { + "epoch": 1.0, + "learning_rate": 1.1172669935002356e-05, + "loss": 3.3627, + "step": 2371584 + }, + { + "epoch": 1.0, + "learning_rate": 1.1164300366295567e-05, + "loss": 3.3687, + "step": 2372096 + }, + { + "epoch": 1.0, + "learning_rate": 1.1155914418785047e-05, + "loss": 3.3515, + "step": 2372608 + }, + { + "epoch": 1.0, + "learning_rate": 1.1147528471274527e-05, + "loss": 3.3453, + "step": 2373120 + }, + { + "epoch": 1.0, + "learning_rate": 1.1139142523764007e-05, + "loss": 3.3445, + "step": 2373632 + }, + { + "epoch": 1.0, + "learning_rate": 1.113077295505722e-05, + "loss": 3.3598, + "step": 2374144 + }, + { + "epoch": 1.0, + "learning_rate": 1.11223870075467e-05, + "loss": 3.3562, + "step": 2374656 + }, + { + "epoch": 1.0, + "learning_rate": 1.1114001060036177e-05, + "loss": 3.3567, + "step": 2375168 + }, + { + "epoch": 1.0, + "learning_rate": 1.1105615112525657e-05, + "loss": 3.3589, + "step": 2375680 + }, + { + "epoch": 1.0, + "learning_rate": 1.1097229165015139e-05, + "loss": 3.3747, + "step": 2376192 + }, + { + "epoch": 1.0, + "learning_rate": 1.1088843217504617e-05, + "loss": 3.3625, + "step": 2376704 + }, + { + "epoch": 1.0, + "learning_rate": 1.1080473648797828e-05, + "loss": 3.3621, + "step": 2377216 + }, + { + "epoch": 1.0, + "learning_rate": 1.107208770128731e-05, + "loss": 3.3581, + "step": 2377728 + }, + { + "epoch": 1.0, + "learning_rate": 1.106370175377679e-05, + "loss": 3.3718, + "step": 2378240 + }, + { + "epoch": 1.0, + "learning_rate": 1.1055315806266268e-05, + "loss": 3.3563, + "step": 2378752 + }, + { + "epoch": 1.0, + "learning_rate": 1.104694623755948e-05, + "loss": 3.3569, + "step": 2379264 + }, + { + "epoch": 1.0, + "learning_rate": 1.103856029004896e-05, + "loss": 3.3605, + "step": 2379776 + }, + { + "epoch": 1.0, + "learning_rate": 1.1030174342538439e-05, + "loss": 3.3515, + "step": 2380288 + }, + { + "epoch": 1.0, + "learning_rate": 1.102178839502792e-05, + "loss": 3.3524, + "step": 2380800 + }, + { + "epoch": 1.01, + "learning_rate": 1.10134024475174e-05, + "loss": 3.3541, + "step": 2381312 + }, + { + "epoch": 1.01, + "learning_rate": 1.1005032878810611e-05, + "loss": 3.3579, + "step": 2381824 + }, + { + "epoch": 1.01, + "learning_rate": 1.0996646931300091e-05, + "loss": 3.3619, + "step": 2382336 + }, + { + "epoch": 1.01, + "learning_rate": 1.0988260983789571e-05, + "loss": 3.3619, + "step": 2382848 + }, + { + "epoch": 1.01, + "learning_rate": 1.0979875036279051e-05, + "loss": 3.3628, + "step": 2383360 + }, + { + "epoch": 1.01, + "learning_rate": 1.0971489088768531e-05, + "loss": 3.3562, + "step": 2383872 + }, + { + "epoch": 1.01, + "learning_rate": 1.0963103141258011e-05, + "loss": 3.3653, + "step": 2384384 + }, + { + "epoch": 1.01, + "learning_rate": 1.095471719374749e-05, + "loss": 3.3451, + "step": 2384896 + }, + { + "epoch": 1.01, + "learning_rate": 1.0946347625040702e-05, + "loss": 3.3645, + "step": 2385408 + }, + { + "epoch": 1.01, + "learning_rate": 1.0937978056333912e-05, + "loss": 3.3416, + "step": 2385920 + }, + { + "epoch": 1.01, + "learning_rate": 1.0929592108823392e-05, + "loss": 3.3532, + "step": 2386432 + }, + { + "epoch": 1.01, + "learning_rate": 1.0921206161312874e-05, + "loss": 3.3533, + "step": 2386944 + }, + { + "epoch": 1.01, + "learning_rate": 1.0912820213802354e-05, + "loss": 3.365, + "step": 2387456 + }, + { + "epoch": 1.01, + "learning_rate": 1.0904434266291832e-05, + "loss": 3.3554, + "step": 2387968 + }, + { + "epoch": 1.01, + "learning_rate": 1.0896064697585045e-05, + "loss": 3.3573, + "step": 2388480 + }, + { + "epoch": 1.01, + "learning_rate": 1.0887678750074525e-05, + "loss": 3.3635, + "step": 2388992 + }, + { + "epoch": 1.01, + "learning_rate": 1.0879292802564003e-05, + "loss": 3.3631, + "step": 2389504 + }, + { + "epoch": 1.01, + "learning_rate": 1.0870906855053483e-05, + "loss": 3.3553, + "step": 2390016 + }, + { + "epoch": 1.01, + "learning_rate": 1.0862537286346695e-05, + "loss": 3.3329, + "step": 2390528 + }, + { + "epoch": 1.01, + "learning_rate": 1.0854151338836175e-05, + "loss": 3.3547, + "step": 2391040 + }, + { + "epoch": 1.01, + "learning_rate": 1.0845765391325655e-05, + "loss": 3.3632, + "step": 2391552 + }, + { + "epoch": 1.01, + "learning_rate": 1.0837379443815135e-05, + "loss": 3.3614, + "step": 2392064 + }, + { + "epoch": 1.01, + "learning_rate": 1.0828993496304615e-05, + "loss": 3.3543, + "step": 2392576 + }, + { + "epoch": 1.01, + "learning_rate": 1.0820623927597828e-05, + "loss": 3.3465, + "step": 2393088 + }, + { + "epoch": 1.01, + "learning_rate": 1.0812237980087306e-05, + "loss": 3.3532, + "step": 2393600 + }, + { + "epoch": 1.01, + "learning_rate": 1.0803852032576786e-05, + "loss": 3.339, + "step": 2394112 + }, + { + "epoch": 1.01, + "learning_rate": 1.0795466085066266e-05, + "loss": 3.3619, + "step": 2394624 + }, + { + "epoch": 1.01, + "learning_rate": 1.0787080137555746e-05, + "loss": 3.3389, + "step": 2395136 + }, + { + "epoch": 1.01, + "learning_rate": 1.0778694190045226e-05, + "loss": 3.3601, + "step": 2395648 + }, + { + "epoch": 1.01, + "learning_rate": 1.0770324621338437e-05, + "loss": 3.3585, + "step": 2396160 + }, + { + "epoch": 1.01, + "learning_rate": 1.0761938673827917e-05, + "loss": 3.3452, + "step": 2396672 + }, + { + "epoch": 1.01, + "learning_rate": 1.0753552726317396e-05, + "loss": 3.3536, + "step": 2397184 + }, + { + "epoch": 1.01, + "learning_rate": 1.0745166778806876e-05, + "loss": 3.3416, + "step": 2397696 + }, + { + "epoch": 1.01, + "learning_rate": 1.0736797210100089e-05, + "loss": 3.3472, + "step": 2398208 + }, + { + "epoch": 1.01, + "learning_rate": 1.0728411262589567e-05, + "loss": 3.3419, + "step": 2398720 + }, + { + "epoch": 1.01, + "learning_rate": 1.0720025315079047e-05, + "loss": 3.3636, + "step": 2399232 + }, + { + "epoch": 1.01, + "learning_rate": 1.0711639367568529e-05, + "loss": 3.347, + "step": 2399744 + }, + { + "epoch": 1.01, + "learning_rate": 1.070326979886174e-05, + "loss": 3.345, + "step": 2400256 + }, + { + "epoch": 1.01, + "learning_rate": 1.069488385135122e-05, + "loss": 3.3442, + "step": 2400768 + }, + { + "epoch": 1.01, + "learning_rate": 1.06864979038407e-05, + "loss": 3.3522, + "step": 2401280 + }, + { + "epoch": 1.01, + "learning_rate": 1.067811195633018e-05, + "loss": 3.3581, + "step": 2401792 + }, + { + "epoch": 1.01, + "learning_rate": 1.066974238762339e-05, + "loss": 3.3539, + "step": 2402304 + }, + { + "epoch": 1.01, + "learning_rate": 1.066135644011287e-05, + "loss": 3.3482, + "step": 2402816 + }, + { + "epoch": 1.01, + "learning_rate": 1.065297049260235e-05, + "loss": 3.3569, + "step": 2403328 + }, + { + "epoch": 1.01, + "learning_rate": 1.064458454509183e-05, + "loss": 3.3674, + "step": 2403840 + }, + { + "epoch": 1.01, + "learning_rate": 1.063619859758131e-05, + "loss": 3.3415, + "step": 2404352 + }, + { + "epoch": 1.01, + "learning_rate": 1.062781265007079e-05, + "loss": 3.3667, + "step": 2404864 + }, + { + "epoch": 1.01, + "learning_rate": 1.061942670256027e-05, + "loss": 3.3493, + "step": 2405376 + }, + { + "epoch": 1.01, + "learning_rate": 1.0611040755049748e-05, + "loss": 3.3499, + "step": 2405888 + }, + { + "epoch": 1.01, + "learning_rate": 1.060267118634296e-05, + "loss": 3.3497, + "step": 2406400 + }, + { + "epoch": 1.01, + "learning_rate": 1.059428523883244e-05, + "loss": 3.3547, + "step": 2406912 + }, + { + "epoch": 1.01, + "learning_rate": 1.058589929132192e-05, + "loss": 3.3628, + "step": 2407424 + }, + { + "epoch": 1.01, + "learning_rate": 1.05775133438114e-05, + "loss": 3.3604, + "step": 2407936 + }, + { + "epoch": 1.01, + "learning_rate": 1.056912739630088e-05, + "loss": 3.3501, + "step": 2408448 + }, + { + "epoch": 1.01, + "learning_rate": 1.0560757827594093e-05, + "loss": 3.3284, + "step": 2408960 + }, + { + "epoch": 1.01, + "learning_rate": 1.0552371880083571e-05, + "loss": 3.3484, + "step": 2409472 + }, + { + "epoch": 1.01, + "learning_rate": 1.0543985932573051e-05, + "loss": 3.3512, + "step": 2409984 + }, + { + "epoch": 1.01, + "learning_rate": 1.0535599985062531e-05, + "loss": 3.3547, + "step": 2410496 + }, + { + "epoch": 1.01, + "learning_rate": 1.0527230416355744e-05, + "loss": 3.3418, + "step": 2411008 + }, + { + "epoch": 1.01, + "learning_rate": 1.0518844468845222e-05, + "loss": 3.3569, + "step": 2411520 + }, + { + "epoch": 1.02, + "learning_rate": 1.0510458521334702e-05, + "loss": 3.3546, + "step": 2412032 + }, + { + "epoch": 1.02, + "learning_rate": 1.0502072573824184e-05, + "loss": 3.3534, + "step": 2412544 + }, + { + "epoch": 1.02, + "learning_rate": 1.0493686626313662e-05, + "loss": 3.3428, + "step": 2413056 + }, + { + "epoch": 1.02, + "learning_rate": 1.0485317057606874e-05, + "loss": 3.3547, + "step": 2413568 + }, + { + "epoch": 1.02, + "learning_rate": 1.0476931110096354e-05, + "loss": 3.3478, + "step": 2414080 + }, + { + "epoch": 1.02, + "learning_rate": 1.0468545162585834e-05, + "loss": 3.3625, + "step": 2414592 + }, + { + "epoch": 1.02, + "learning_rate": 1.0460159215075312e-05, + "loss": 3.349, + "step": 2415104 + }, + { + "epoch": 1.02, + "learning_rate": 1.0451789646368525e-05, + "loss": 3.3403, + "step": 2415616 + }, + { + "epoch": 1.02, + "learning_rate": 1.0443403698858005e-05, + "loss": 3.3581, + "step": 2416128 + }, + { + "epoch": 1.02, + "learning_rate": 1.0435017751347485e-05, + "loss": 3.3526, + "step": 2416640 + }, + { + "epoch": 1.02, + "learning_rate": 1.0426631803836965e-05, + "loss": 3.3528, + "step": 2417152 + }, + { + "epoch": 1.02, + "learning_rate": 1.0418245856326445e-05, + "loss": 3.3553, + "step": 2417664 + }, + { + "epoch": 1.02, + "learning_rate": 1.0409859908815925e-05, + "loss": 3.3454, + "step": 2418176 + }, + { + "epoch": 1.02, + "learning_rate": 1.0401490340109136e-05, + "loss": 3.3465, + "step": 2418688 + }, + { + "epoch": 1.02, + "learning_rate": 1.0393104392598615e-05, + "loss": 3.355, + "step": 2419200 + }, + { + "epoch": 1.02, + "learning_rate": 1.0384718445088095e-05, + "loss": 3.3398, + "step": 2419712 + }, + { + "epoch": 1.02, + "learning_rate": 1.0376332497577575e-05, + "loss": 3.3492, + "step": 2420224 + }, + { + "epoch": 1.02, + "learning_rate": 1.0367962928870786e-05, + "loss": 3.3487, + "step": 2420736 + }, + { + "epoch": 1.02, + "learning_rate": 1.0359576981360266e-05, + "loss": 3.3505, + "step": 2421248 + }, + { + "epoch": 1.02, + "learning_rate": 1.0351191033849748e-05, + "loss": 3.3392, + "step": 2421760 + }, + { + "epoch": 1.02, + "learning_rate": 1.0342805086339226e-05, + "loss": 3.3554, + "step": 2422272 + }, + { + "epoch": 1.02, + "learning_rate": 1.0334435517632437e-05, + "loss": 3.3426, + "step": 2422784 + }, + { + "epoch": 1.02, + "learning_rate": 1.0326049570121918e-05, + "loss": 3.3551, + "step": 2423296 + }, + { + "epoch": 1.02, + "learning_rate": 1.0317663622611398e-05, + "loss": 3.3533, + "step": 2423808 + }, + { + "epoch": 1.02, + "learning_rate": 1.0309277675100877e-05, + "loss": 3.3533, + "step": 2424320 + }, + { + "epoch": 1.02, + "learning_rate": 1.030090810639409e-05, + "loss": 3.3614, + "step": 2424832 + }, + { + "epoch": 1.02, + "learning_rate": 1.0292522158883569e-05, + "loss": 3.3593, + "step": 2425344 + }, + { + "epoch": 1.02, + "learning_rate": 1.0284136211373049e-05, + "loss": 3.352, + "step": 2425856 + }, + { + "epoch": 1.02, + "learning_rate": 1.0275750263862529e-05, + "loss": 3.3457, + "step": 2426368 + }, + { + "epoch": 1.02, + "learning_rate": 1.0267364316352009e-05, + "loss": 3.3572, + "step": 2426880 + }, + { + "epoch": 1.02, + "learning_rate": 1.0258978368841489e-05, + "loss": 3.3491, + "step": 2427392 + }, + { + "epoch": 1.02, + "learning_rate": 1.02506088001347e-05, + "loss": 3.3446, + "step": 2427904 + }, + { + "epoch": 1.02, + "learning_rate": 1.024222285262418e-05, + "loss": 3.3559, + "step": 2428416 + }, + { + "epoch": 1.02, + "learning_rate": 1.023383690511366e-05, + "loss": 3.3514, + "step": 2428928 + }, + { + "epoch": 1.02, + "learning_rate": 1.022545095760314e-05, + "loss": 3.3539, + "step": 2429440 + }, + { + "epoch": 1.02, + "learning_rate": 1.021706501009262e-05, + "loss": 3.345, + "step": 2429952 + }, + { + "epoch": 1.02, + "learning_rate": 1.020869544138583e-05, + "loss": 3.3473, + "step": 2430464 + }, + { + "epoch": 1.02, + "learning_rate": 1.020030949387531e-05, + "loss": 3.3546, + "step": 2430976 + }, + { + "epoch": 1.02, + "learning_rate": 1.019192354636479e-05, + "loss": 3.3464, + "step": 2431488 + }, + { + "epoch": 1.02, + "learning_rate": 1.018353759885427e-05, + "loss": 3.3502, + "step": 2432000 + }, + { + "epoch": 1.02, + "learning_rate": 1.017515165134375e-05, + "loss": 3.362, + "step": 2432512 + }, + { + "epoch": 1.02, + "learning_rate": 1.016676570383323e-05, + "loss": 3.3572, + "step": 2433024 + }, + { + "epoch": 1.02, + "learning_rate": 1.015837975632271e-05, + "loss": 3.3477, + "step": 2433536 + }, + { + "epoch": 1.02, + "learning_rate": 1.014999380881219e-05, + "loss": 3.3535, + "step": 2434048 + }, + { + "epoch": 1.02, + "learning_rate": 1.0141624240105403e-05, + "loss": 3.3528, + "step": 2434560 + }, + { + "epoch": 1.02, + "learning_rate": 1.013323829259488e-05, + "loss": 3.3522, + "step": 2435072 + }, + { + "epoch": 1.02, + "learning_rate": 1.012485234508436e-05, + "loss": 3.3465, + "step": 2435584 + }, + { + "epoch": 1.02, + "learning_rate": 1.011646639757384e-05, + "loss": 3.3446, + "step": 2436096 + }, + { + "epoch": 1.02, + "learning_rate": 1.0108096828867053e-05, + "loss": 3.3628, + "step": 2436608 + }, + { + "epoch": 1.02, + "learning_rate": 1.0099710881356531e-05, + "loss": 3.3643, + "step": 2437120 + }, + { + "epoch": 1.02, + "learning_rate": 1.0091324933846011e-05, + "loss": 3.3556, + "step": 2437632 + }, + { + "epoch": 1.02, + "learning_rate": 1.0082938986335493e-05, + "loss": 3.3505, + "step": 2438144 + }, + { + "epoch": 1.02, + "learning_rate": 1.0074569417628702e-05, + "loss": 3.3626, + "step": 2438656 + }, + { + "epoch": 1.02, + "learning_rate": 1.0066183470118184e-05, + "loss": 3.3485, + "step": 2439168 + }, + { + "epoch": 1.02, + "learning_rate": 1.0057797522607664e-05, + "loss": 3.353, + "step": 2439680 + }, + { + "epoch": 1.02, + "learning_rate": 1.0049411575097144e-05, + "loss": 3.3498, + "step": 2440192 + }, + { + "epoch": 1.02, + "learning_rate": 1.0041025627586622e-05, + "loss": 3.3512, + "step": 2440704 + }, + { + "epoch": 1.02, + "learning_rate": 1.0032656058879834e-05, + "loss": 3.3619, + "step": 2441216 + }, + { + "epoch": 1.02, + "learning_rate": 1.0024270111369314e-05, + "loss": 3.3433, + "step": 2441728 + }, + { + "epoch": 1.03, + "learning_rate": 1.0015884163858794e-05, + "loss": 3.3476, + "step": 2442240 + }, + { + "epoch": 1.03, + "eval_loss": 3.8691704273223877, + "eval_runtime": 304.9817, + "eval_samples_per_second": 1251.193, + "eval_steps_per_second": 39.101, + "step": 2442240 + }, + { + "epoch": 0.0, + "learning_rate": 1.0007498216348274e-05, + "loss": 3.3393, + "step": 2442752 + }, + { + "epoch": 0.0, + "learning_rate": 9.999112268837754e-06, + "loss": 3.3476, + "step": 2443264 + }, + { + "epoch": 0.0, + "learning_rate": 9.990742700130965e-06, + "loss": 3.3657, + "step": 2443776 + }, + { + "epoch": 0.0, + "learning_rate": 9.982356752620445e-06, + "loss": 3.3645, + "step": 2444288 + }, + { + "epoch": 0.0, + "learning_rate": 9.973970805109925e-06, + "loss": 3.3505, + "step": 2444800 + }, + { + "epoch": 0.0, + "learning_rate": 9.965584857599405e-06, + "loss": 3.3517, + "step": 2445312 + }, + { + "epoch": 0.0, + "learning_rate": 9.957198910088885e-06, + "loss": 3.3459, + "step": 2445824 + }, + { + "epoch": 0.0, + "learning_rate": 9.948812962578365e-06, + "loss": 3.342, + "step": 2446336 + }, + { + "epoch": 0.0, + "learning_rate": 9.940443393871576e-06, + "loss": 3.3577, + "step": 2446848 + }, + { + "epoch": 0.0, + "learning_rate": 9.932057446361057e-06, + "loss": 3.3504, + "step": 2447360 + }, + { + "epoch": 0.0, + "learning_rate": 9.923671498850536e-06, + "loss": 3.3554, + "step": 2447872 + }, + { + "epoch": 0.0, + "learning_rate": 9.915285551340015e-06, + "loss": 3.3548, + "step": 2448384 + }, + { + "epoch": 0.0, + "learning_rate": 9.906915982633228e-06, + "loss": 3.3469, + "step": 2448896 + }, + { + "epoch": 0.0, + "learning_rate": 9.898530035122708e-06, + "loss": 3.3368, + "step": 2449408 + }, + { + "epoch": 0.0, + "learning_rate": 9.890144087612186e-06, + "loss": 3.3337, + "step": 2449920 + }, + { + "epoch": 0.0, + "learning_rate": 9.881758140101668e-06, + "loss": 3.349, + "step": 2450432 + }, + { + "epoch": 0.0, + "learning_rate": 9.873372192591148e-06, + "loss": 3.3454, + "step": 2450944 + }, + { + "epoch": 0.0, + "learning_rate": 9.864986245080626e-06, + "loss": 3.35, + "step": 2451456 + }, + { + "epoch": 0.0, + "learning_rate": 9.856600297570106e-06, + "loss": 3.3465, + "step": 2451968 + }, + { + "epoch": 0.0, + "learning_rate": 9.848230728863319e-06, + "loss": 3.3701, + "step": 2452480 + }, + { + "epoch": 0.0, + "learning_rate": 9.839844781352798e-06, + "loss": 3.3512, + "step": 2452992 + }, + { + "epoch": 0.0, + "learning_rate": 9.831458833842277e-06, + "loss": 3.3511, + "step": 2453504 + }, + { + "epoch": 0.0, + "learning_rate": 9.823072886331758e-06, + "loss": 3.3491, + "step": 2454016 + }, + { + "epoch": 0.0, + "learning_rate": 9.814686938821238e-06, + "loss": 3.3657, + "step": 2454528 + }, + { + "epoch": 0.0, + "learning_rate": 9.806317370114449e-06, + "loss": 3.3388, + "step": 2455040 + }, + { + "epoch": 0.0, + "learning_rate": 9.797931422603929e-06, + "loss": 3.3497, + "step": 2455552 + }, + { + "epoch": 0.0, + "learning_rate": 9.789545475093409e-06, + "loss": 3.3485, + "step": 2456064 + }, + { + "epoch": 0.0, + "learning_rate": 9.781159527582889e-06, + "loss": 3.3445, + "step": 2456576 + }, + { + "epoch": 0.0, + "learning_rate": 9.772773580072369e-06, + "loss": 3.3423, + "step": 2457088 + }, + { + "epoch": 0.01, + "learning_rate": 9.76440401136558e-06, + "loss": 3.3444, + "step": 2457600 + }, + { + "epoch": 0.01, + "learning_rate": 9.75601806385506e-06, + "loss": 3.3487, + "step": 2458112 + }, + { + "epoch": 0.01, + "learning_rate": 9.74763211634454e-06, + "loss": 3.353, + "step": 2458624 + }, + { + "epoch": 0.01, + "learning_rate": 9.73924616883402e-06, + "loss": 3.3479, + "step": 2459136 + }, + { + "epoch": 0.01, + "learning_rate": 9.73087660012723e-06, + "loss": 3.3575, + "step": 2459648 + }, + { + "epoch": 0.01, + "learning_rate": 9.722490652616712e-06, + "loss": 3.343, + "step": 2460160 + }, + { + "epoch": 0.01, + "learning_rate": 9.71410470510619e-06, + "loss": 3.3562, + "step": 2460672 + }, + { + "epoch": 0.01, + "learning_rate": 9.70571875759567e-06, + "loss": 3.335, + "step": 2461184 + }, + { + "epoch": 0.01, + "learning_rate": 9.69733281008515e-06, + "loss": 3.3569, + "step": 2461696 + }, + { + "epoch": 0.01, + "learning_rate": 9.688963241378363e-06, + "loss": 3.3314, + "step": 2462208 + }, + { + "epoch": 0.01, + "learning_rate": 9.680577293867841e-06, + "loss": 3.3454, + "step": 2462720 + }, + { + "epoch": 0.01, + "learning_rate": 9.672191346357323e-06, + "loss": 3.3427, + "step": 2463232 + }, + { + "epoch": 0.01, + "learning_rate": 9.663805398846803e-06, + "loss": 3.3538, + "step": 2463744 + }, + { + "epoch": 0.01, + "learning_rate": 9.655435830140012e-06, + "loss": 3.3442, + "step": 2464256 + }, + { + "epoch": 0.01, + "learning_rate": 9.647049882629493e-06, + "loss": 3.3525, + "step": 2464768 + }, + { + "epoch": 0.01, + "learning_rate": 9.638663935118973e-06, + "loss": 3.3501, + "step": 2465280 + }, + { + "epoch": 0.01, + "learning_rate": 9.630277987608453e-06, + "loss": 3.353, + "step": 2465792 + }, + { + "epoch": 0.01, + "learning_rate": 9.621892040097931e-06, + "loss": 3.35, + "step": 2466304 + }, + { + "epoch": 0.01, + "learning_rate": 9.613522471391144e-06, + "loss": 3.3241, + "step": 2466816 + }, + { + "epoch": 0.01, + "learning_rate": 9.605136523880624e-06, + "loss": 3.3491, + "step": 2467328 + }, + { + "epoch": 0.01, + "learning_rate": 9.596750576370104e-06, + "loss": 3.3533, + "step": 2467840 + }, + { + "epoch": 0.01, + "learning_rate": 9.588364628859584e-06, + "loss": 3.3488, + "step": 2468352 + }, + { + "epoch": 0.01, + "learning_rate": 9.579995060152795e-06, + "loss": 3.3468, + "step": 2468864 + }, + { + "epoch": 0.01, + "learning_rate": 9.571609112642276e-06, + "loss": 3.3366, + "step": 2469376 + }, + { + "epoch": 0.01, + "learning_rate": 9.563223165131755e-06, + "loss": 3.3454, + "step": 2469888 + }, + { + "epoch": 0.01, + "learning_rate": 9.554837217621234e-06, + "loss": 3.3305, + "step": 2470400 + }, + { + "epoch": 0.01, + "learning_rate": 9.546451270110714e-06, + "loss": 3.3544, + "step": 2470912 + }, + { + "epoch": 0.01, + "learning_rate": 9.538065322600194e-06, + "loss": 3.3294, + "step": 2471424 + }, + { + "epoch": 0.01, + "learning_rate": 9.529695753893405e-06, + "loss": 3.3537, + "step": 2471936 + }, + { + "epoch": 0.01, + "learning_rate": 9.521309806382885e-06, + "loss": 3.3466, + "step": 2472448 + }, + { + "epoch": 0.01, + "learning_rate": 9.512923858872367e-06, + "loss": 3.3368, + "step": 2472960 + }, + { + "epoch": 0.01, + "learning_rate": 9.504537911361845e-06, + "loss": 3.3431, + "step": 2473472 + }, + { + "epoch": 0.01, + "learning_rate": 9.496151963851325e-06, + "loss": 3.3366, + "step": 2473984 + }, + { + "epoch": 0.01, + "learning_rate": 9.487766016340805e-06, + "loss": 3.3392, + "step": 2474496 + }, + { + "epoch": 0.01, + "learning_rate": 9.479396447634016e-06, + "loss": 3.3319, + "step": 2475008 + }, + { + "epoch": 0.01, + "learning_rate": 9.471010500123496e-06, + "loss": 3.3522, + "step": 2475520 + }, + { + "epoch": 0.01, + "learning_rate": 9.462624552612977e-06, + "loss": 3.3385, + "step": 2476032 + }, + { + "epoch": 0.01, + "learning_rate": 9.454238605102457e-06, + "loss": 3.3357, + "step": 2476544 + }, + { + "epoch": 0.01, + "learning_rate": 9.445852657591936e-06, + "loss": 3.333, + "step": 2477056 + }, + { + "epoch": 0.01, + "learning_rate": 9.437483088885148e-06, + "loss": 3.3407, + "step": 2477568 + }, + { + "epoch": 0.01, + "learning_rate": 9.429097141374628e-06, + "loss": 3.3473, + "step": 2478080 + }, + { + "epoch": 0.01, + "learning_rate": 9.420711193864106e-06, + "loss": 3.3462, + "step": 2478592 + }, + { + "epoch": 0.01, + "learning_rate": 9.412325246353588e-06, + "loss": 3.3381, + "step": 2479104 + }, + { + "epoch": 0.01, + "learning_rate": 9.403955677646799e-06, + "loss": 3.35, + "step": 2479616 + }, + { + "epoch": 0.01, + "learning_rate": 9.395569730136279e-06, + "loss": 3.3546, + "step": 2480128 + }, + { + "epoch": 0.01, + "learning_rate": 9.387183782625759e-06, + "loss": 3.3321, + "step": 2480640 + }, + { + "epoch": 0.01, + "learning_rate": 9.378797835115239e-06, + "loss": 3.3548, + "step": 2481152 + }, + { + "epoch": 0.01, + "learning_rate": 9.370411887604719e-06, + "loss": 3.3439, + "step": 2481664 + }, + { + "epoch": 0.01, + "learning_rate": 9.362025940094197e-06, + "loss": 3.342, + "step": 2482176 + }, + { + "epoch": 0.01, + "learning_rate": 9.35365637138741e-06, + "loss": 3.3388, + "step": 2482688 + }, + { + "epoch": 0.01, + "learning_rate": 9.34527042387689e-06, + "loss": 3.3461, + "step": 2483200 + }, + { + "epoch": 0.01, + "learning_rate": 9.33688447636637e-06, + "loss": 3.3503, + "step": 2483712 + }, + { + "epoch": 0.01, + "learning_rate": 9.32849852885585e-06, + "loss": 3.3573, + "step": 2484224 + }, + { + "epoch": 0.01, + "learning_rate": 9.320112581345329e-06, + "loss": 3.3358, + "step": 2484736 + }, + { + "epoch": 0.01, + "learning_rate": 9.311743012638542e-06, + "loss": 3.3203, + "step": 2485248 + }, + { + "epoch": 0.01, + "learning_rate": 9.303357065128022e-06, + "loss": 3.3397, + "step": 2485760 + }, + { + "epoch": 0.01, + "learning_rate": 9.2949711176175e-06, + "loss": 3.3418, + "step": 2486272 + }, + { + "epoch": 0.01, + "learning_rate": 9.28658517010698e-06, + "loss": 3.3461, + "step": 2486784 + }, + { + "epoch": 0.01, + "learning_rate": 9.278215601400192e-06, + "loss": 3.3368, + "step": 2487296 + }, + { + "epoch": 0.01, + "learning_rate": 9.26982965388967e-06, + "loss": 3.3469, + "step": 2487808 + }, + { + "epoch": 0.02, + "learning_rate": 9.26144370637915e-06, + "loss": 3.344, + "step": 2488320 + }, + { + "epoch": 0.02, + "learning_rate": 9.253057758868632e-06, + "loss": 3.3434, + "step": 2488832 + }, + { + "epoch": 0.02, + "learning_rate": 9.244671811358112e-06, + "loss": 3.3324, + "step": 2489344 + }, + { + "epoch": 0.02, + "learning_rate": 9.236302242651323e-06, + "loss": 3.3467, + "step": 2489856 + }, + { + "epoch": 0.02, + "learning_rate": 9.227916295140803e-06, + "loss": 3.3411, + "step": 2490368 + }, + { + "epoch": 0.02, + "learning_rate": 9.219530347630283e-06, + "loss": 3.3488, + "step": 2490880 + }, + { + "epoch": 0.02, + "learning_rate": 9.211144400119761e-06, + "loss": 3.3431, + "step": 2491392 + }, + { + "epoch": 0.02, + "learning_rate": 9.202774831412974e-06, + "loss": 3.3294, + "step": 2491904 + }, + { + "epoch": 0.02, + "learning_rate": 9.194388883902453e-06, + "loss": 3.3459, + "step": 2492416 + }, + { + "epoch": 0.02, + "learning_rate": 9.186002936391933e-06, + "loss": 3.346, + "step": 2492928 + }, + { + "epoch": 0.02, + "learning_rate": 9.177616988881413e-06, + "loss": 3.347, + "step": 2493440 + }, + { + "epoch": 0.02, + "learning_rate": 9.169231041370893e-06, + "loss": 3.3434, + "step": 2493952 + }, + { + "epoch": 0.02, + "learning_rate": 9.160861472664104e-06, + "loss": 3.3395, + "step": 2494464 + }, + { + "epoch": 0.02, + "learning_rate": 9.152475525153586e-06, + "loss": 3.3335, + "step": 2494976 + }, + { + "epoch": 0.02, + "learning_rate": 9.144089577643064e-06, + "loss": 3.3466, + "step": 2495488 + }, + { + "epoch": 0.02, + "learning_rate": 9.135703630132544e-06, + "loss": 3.3299, + "step": 2496000 + }, + { + "epoch": 0.02, + "learning_rate": 9.127317682622024e-06, + "loss": 3.3371, + "step": 2496512 + }, + { + "epoch": 0.02, + "learning_rate": 9.118931735111504e-06, + "loss": 3.3435, + "step": 2497024 + }, + { + "epoch": 0.02, + "learning_rate": 9.110562166404715e-06, + "loss": 3.3434, + "step": 2497536 + }, + { + "epoch": 0.02, + "learning_rate": 9.102176218894196e-06, + "loss": 3.3297, + "step": 2498048 + }, + { + "epoch": 0.02, + "learning_rate": 9.093790271383676e-06, + "loss": 3.3444, + "step": 2498560 + }, + { + "epoch": 0.02, + "learning_rate": 9.085404323873155e-06, + "loss": 3.3361, + "step": 2499072 + }, + { + "epoch": 0.02, + "learning_rate": 9.077034755166367e-06, + "loss": 3.347, + "step": 2499584 + }, + { + "epoch": 0.02, + "learning_rate": 9.068648807655847e-06, + "loss": 3.3428, + "step": 2500096 + }, + { + "epoch": 0.02, + "learning_rate": 9.060262860145325e-06, + "loss": 3.3415, + "step": 2500608 + }, + { + "epoch": 0.02, + "learning_rate": 9.051876912634805e-06, + "loss": 3.3535, + "step": 2501120 + }, + { + "epoch": 0.02, + "learning_rate": 9.043490965124287e-06, + "loss": 3.3517, + "step": 2501632 + }, + { + "epoch": 0.02, + "learning_rate": 9.035121396417498e-06, + "loss": 3.3416, + "step": 2502144 + }, + { + "epoch": 0.02, + "learning_rate": 9.026735448906978e-06, + "loss": 3.3369, + "step": 2502656 + }, + { + "epoch": 0.02, + "learning_rate": 9.018349501396458e-06, + "loss": 3.3449, + "step": 2503168 + }, + { + "epoch": 0.02, + "learning_rate": 9.009963553885938e-06, + "loss": 3.3371, + "step": 2503680 + }, + { + "epoch": 0.02, + "learning_rate": 9.00159398517915e-06, + "loss": 3.3377, + "step": 2504192 + }, + { + "epoch": 0.02, + "learning_rate": 8.993208037668628e-06, + "loss": 3.3521, + "step": 2504704 + }, + { + "epoch": 0.02, + "learning_rate": 8.984822090158108e-06, + "loss": 3.3392, + "step": 2505216 + }, + { + "epoch": 0.02, + "learning_rate": 8.976436142647588e-06, + "loss": 3.3425, + "step": 2505728 + }, + { + "epoch": 0.02, + "learning_rate": 8.968050195137068e-06, + "loss": 3.3332, + "step": 2506240 + }, + { + "epoch": 0.02, + "learning_rate": 8.959664247626548e-06, + "loss": 3.339, + "step": 2506752 + }, + { + "epoch": 0.02, + "learning_rate": 8.951278300116028e-06, + "loss": 3.3461, + "step": 2507264 + }, + { + "epoch": 0.02, + "learning_rate": 8.942892352605508e-06, + "loss": 3.3377, + "step": 2507776 + }, + { + "epoch": 0.02, + "learning_rate": 8.93453916270245e-06, + "loss": 3.3366, + "step": 2508288 + }, + { + "epoch": 0.02, + "learning_rate": 8.926153215191931e-06, + "loss": 3.3559, + "step": 2508800 + }, + { + "epoch": 0.02, + "learning_rate": 8.917767267681411e-06, + "loss": 3.3474, + "step": 2509312 + }, + { + "epoch": 0.02, + "learning_rate": 8.90938132017089e-06, + "loss": 3.3429, + "step": 2509824 + }, + { + "epoch": 0.02, + "learning_rate": 8.90099537266037e-06, + "loss": 3.3407, + "step": 2510336 + }, + { + "epoch": 0.02, + "learning_rate": 8.892609425149851e-06, + "loss": 3.3408, + "step": 2510848 + }, + { + "epoch": 0.02, + "learning_rate": 8.884239856443062e-06, + "loss": 3.3438, + "step": 2511360 + }, + { + "epoch": 0.02, + "learning_rate": 8.87585390893254e-06, + "loss": 3.3402, + "step": 2511872 + }, + { + "epoch": 0.02, + "learning_rate": 8.867467961422022e-06, + "loss": 3.3379, + "step": 2512384 + }, + { + "epoch": 0.02, + "learning_rate": 8.859082013911502e-06, + "loss": 3.3526, + "step": 2512896 + }, + { + "epoch": 0.02, + "learning_rate": 8.85069606640098e-06, + "loss": 3.3494, + "step": 2513408 + }, + { + "epoch": 0.02, + "learning_rate": 8.842326497694193e-06, + "loss": 3.3509, + "step": 2513920 + }, + { + "epoch": 0.02, + "learning_rate": 8.833940550183672e-06, + "loss": 3.3419, + "step": 2514432 + }, + { + "epoch": 0.02, + "learning_rate": 8.825554602673152e-06, + "loss": 3.3489, + "step": 2514944 + }, + { + "epoch": 0.02, + "learning_rate": 8.817168655162632e-06, + "loss": 3.3438, + "step": 2515456 + }, + { + "epoch": 0.02, + "learning_rate": 8.808799086455843e-06, + "loss": 3.3441, + "step": 2515968 + }, + { + "epoch": 0.02, + "learning_rate": 8.800413138945323e-06, + "loss": 3.3412, + "step": 2516480 + }, + { + "epoch": 0.02, + "learning_rate": 8.792027191434803e-06, + "loss": 3.3418, + "step": 2516992 + }, + { + "epoch": 0.02, + "learning_rate": 8.783641243924283e-06, + "loss": 3.3502, + "step": 2517504 + }, + { + "epoch": 0.02, + "learning_rate": 8.775255296413763e-06, + "loss": 3.3343, + "step": 2518016 + }, + { + "epoch": 0.02, + "learning_rate": 8.766885727706975e-06, + "loss": 3.3376, + "step": 2518528 + }, + { + "epoch": 0.03, + "eval_loss": 3.8688721656799316, + "eval_runtime": 306.463, + "eval_samples_per_second": 1245.145, + "eval_steps_per_second": 38.912, + "step": 2518560 + }, + { + "epoch": 1.0, + "learning_rate": 8.758499780196454e-06, + "loss": 3.3331, + "step": 2519040 + }, + { + "epoch": 1.0, + "learning_rate": 8.750113832685934e-06, + "loss": 3.3387, + "step": 2519552 + }, + { + "epoch": 1.0, + "learning_rate": 8.741727885175414e-06, + "loss": 3.3576, + "step": 2520064 + }, + { + "epoch": 1.0, + "learning_rate": 8.733358316468626e-06, + "loss": 3.3496, + "step": 2520576 + }, + { + "epoch": 1.0, + "learning_rate": 8.724972368958104e-06, + "loss": 3.3479, + "step": 2521088 + }, + { + "epoch": 1.0, + "learning_rate": 8.716586421447586e-06, + "loss": 3.3398, + "step": 2521600 + }, + { + "epoch": 1.0, + "learning_rate": 8.708200473937066e-06, + "loss": 3.3404, + "step": 2522112 + }, + { + "epoch": 1.0, + "learning_rate": 8.699814526426544e-06, + "loss": 3.3336, + "step": 2522624 + }, + { + "epoch": 1.0, + "learning_rate": 8.691444957719757e-06, + "loss": 3.3462, + "step": 2523136 + }, + { + "epoch": 1.0, + "learning_rate": 8.683059010209237e-06, + "loss": 3.3448, + "step": 2523648 + }, + { + "epoch": 1.0, + "learning_rate": 8.674673062698717e-06, + "loss": 3.3432, + "step": 2524160 + }, + { + "epoch": 1.0, + "learning_rate": 8.666287115188197e-06, + "loss": 3.3479, + "step": 2524672 + }, + { + "epoch": 1.0, + "learning_rate": 8.657917546481407e-06, + "loss": 3.3361, + "step": 2525184 + }, + { + "epoch": 1.0, + "learning_rate": 8.649531598970887e-06, + "loss": 3.333, + "step": 2525696 + }, + { + "epoch": 1.0, + "learning_rate": 8.641145651460367e-06, + "loss": 3.3272, + "step": 2526208 + }, + { + "epoch": 1.0, + "learning_rate": 8.632759703949847e-06, + "loss": 3.3318, + "step": 2526720 + }, + { + "epoch": 1.0, + "learning_rate": 8.624373756439327e-06, + "loss": 3.34, + "step": 2527232 + }, + { + "epoch": 1.0, + "learning_rate": 8.615987808928807e-06, + "loss": 3.3391, + "step": 2527744 + }, + { + "epoch": 1.0, + "learning_rate": 8.607601861418287e-06, + "loss": 3.3392, + "step": 2528256 + }, + { + "epoch": 1.0, + "learning_rate": 8.599232292711498e-06, + "loss": 3.3616, + "step": 2528768 + }, + { + "epoch": 1.0, + "learning_rate": 8.590846345200978e-06, + "loss": 3.3407, + "step": 2529280 + }, + { + "epoch": 1.0, + "learning_rate": 8.582460397690458e-06, + "loss": 3.346, + "step": 2529792 + }, + { + "epoch": 1.0, + "learning_rate": 8.574074450179938e-06, + "loss": 3.3373, + "step": 2530304 + }, + { + "epoch": 1.0, + "learning_rate": 8.565688502669418e-06, + "loss": 3.3521, + "step": 2530816 + }, + { + "epoch": 1.0, + "learning_rate": 8.55731893396263e-06, + "loss": 3.3346, + "step": 2531328 + }, + { + "epoch": 1.0, + "learning_rate": 8.548932986452109e-06, + "loss": 3.3386, + "step": 2531840 + }, + { + "epoch": 1.0, + "learning_rate": 8.540547038941588e-06, + "loss": 3.3384, + "step": 2532352 + }, + { + "epoch": 1.0, + "learning_rate": 8.53216109143107e-06, + "loss": 3.3392, + "step": 2532864 + }, + { + "epoch": 1.0, + "learning_rate": 8.523775143920548e-06, + "loss": 3.3337, + "step": 2533376 + }, + { + "epoch": 1.01, + "learning_rate": 8.51540557521376e-06, + "loss": 3.3325, + "step": 2533888 + }, + { + "epoch": 1.01, + "learning_rate": 8.50701962770324e-06, + "loss": 3.3394, + "step": 2534400 + }, + { + "epoch": 1.01, + "learning_rate": 8.49863368019272e-06, + "loss": 3.3476, + "step": 2534912 + }, + { + "epoch": 1.01, + "learning_rate": 8.490247732682199e-06, + "loss": 3.338, + "step": 2535424 + }, + { + "epoch": 1.01, + "learning_rate": 8.481878163975412e-06, + "loss": 3.3437, + "step": 2535936 + }, + { + "epoch": 1.01, + "learning_rate": 8.473492216464891e-06, + "loss": 3.3338, + "step": 2536448 + }, + { + "epoch": 1.01, + "learning_rate": 8.465106268954371e-06, + "loss": 3.346, + "step": 2536960 + }, + { + "epoch": 1.01, + "learning_rate": 8.456720321443851e-06, + "loss": 3.3293, + "step": 2537472 + }, + { + "epoch": 1.01, + "learning_rate": 8.448334373933331e-06, + "loss": 3.3412, + "step": 2537984 + }, + { + "epoch": 1.01, + "learning_rate": 8.439964805226542e-06, + "loss": 3.323, + "step": 2538496 + }, + { + "epoch": 1.01, + "learning_rate": 8.431578857716022e-06, + "loss": 3.3406, + "step": 2539008 + }, + { + "epoch": 1.01, + "learning_rate": 8.423192910205502e-06, + "loss": 3.3296, + "step": 2539520 + }, + { + "epoch": 1.01, + "learning_rate": 8.414806962694982e-06, + "loss": 3.3464, + "step": 2540032 + }, + { + "epoch": 1.01, + "learning_rate": 8.406437393988194e-06, + "loss": 3.3342, + "step": 2540544 + }, + { + "epoch": 1.01, + "learning_rate": 8.398051446477673e-06, + "loss": 3.3403, + "step": 2541056 + }, + { + "epoch": 1.01, + "learning_rate": 8.389665498967153e-06, + "loss": 3.3434, + "step": 2541568 + }, + { + "epoch": 1.01, + "learning_rate": 8.381279551456633e-06, + "loss": 3.3419, + "step": 2542080 + }, + { + "epoch": 1.01, + "learning_rate": 8.372893603946113e-06, + "loss": 3.3415, + "step": 2542592 + }, + { + "epoch": 1.01, + "learning_rate": 8.364524035239323e-06, + "loss": 3.3179, + "step": 2543104 + }, + { + "epoch": 1.01, + "learning_rate": 8.356138087728805e-06, + "loss": 3.3317, + "step": 2543616 + }, + { + "epoch": 1.01, + "learning_rate": 8.347752140218285e-06, + "loss": 3.346, + "step": 2544128 + }, + { + "epoch": 1.01, + "learning_rate": 8.339366192707763e-06, + "loss": 3.3422, + "step": 2544640 + }, + { + "epoch": 1.01, + "learning_rate": 8.330996624000976e-06, + "loss": 3.336, + "step": 2545152 + }, + { + "epoch": 1.01, + "learning_rate": 8.322610676490456e-06, + "loss": 3.3326, + "step": 2545664 + }, + { + "epoch": 1.01, + "learning_rate": 8.314224728979934e-06, + "loss": 3.3338, + "step": 2546176 + }, + { + "epoch": 1.01, + "learning_rate": 8.305838781469414e-06, + "loss": 3.3256, + "step": 2546688 + }, + { + "epoch": 1.01, + "learning_rate": 8.297452833958896e-06, + "loss": 3.3385, + "step": 2547200 + }, + { + "epoch": 1.01, + "learning_rate": 8.289066886448376e-06, + "loss": 3.3229, + "step": 2547712 + }, + { + "epoch": 1.01, + "learning_rate": 8.280697317741586e-06, + "loss": 3.3375, + "step": 2548224 + }, + { + "epoch": 1.01, + "learning_rate": 8.272311370231066e-06, + "loss": 3.3418, + "step": 2548736 + }, + { + "epoch": 1.01, + "learning_rate": 8.263925422720546e-06, + "loss": 3.3268, + "step": 2549248 + }, + { + "epoch": 1.01, + "learning_rate": 8.255539475210024e-06, + "loss": 3.33, + "step": 2549760 + }, + { + "epoch": 1.01, + "learning_rate": 8.247153527699506e-06, + "loss": 3.3294, + "step": 2550272 + }, + { + "epoch": 1.01, + "learning_rate": 8.238783958992717e-06, + "loss": 3.3254, + "step": 2550784 + }, + { + "epoch": 1.01, + "learning_rate": 8.230398011482197e-06, + "loss": 3.3237, + "step": 2551296 + }, + { + "epoch": 1.01, + "learning_rate": 8.222012063971677e-06, + "loss": 3.3435, + "step": 2551808 + }, + { + "epoch": 1.01, + "learning_rate": 8.213626116461157e-06, + "loss": 3.3316, + "step": 2552320 + }, + { + "epoch": 1.01, + "learning_rate": 8.205256547754368e-06, + "loss": 3.3274, + "step": 2552832 + }, + { + "epoch": 1.01, + "learning_rate": 8.19687060024385e-06, + "loss": 3.3284, + "step": 2553344 + }, + { + "epoch": 1.01, + "learning_rate": 8.188484652733328e-06, + "loss": 3.3296, + "step": 2553856 + }, + { + "epoch": 1.01, + "learning_rate": 8.180098705222807e-06, + "loss": 3.3375, + "step": 2554368 + }, + { + "epoch": 1.01, + "learning_rate": 8.17172913651602e-06, + "loss": 3.3351, + "step": 2554880 + }, + { + "epoch": 1.01, + "learning_rate": 8.163343189005498e-06, + "loss": 3.3303, + "step": 2555392 + }, + { + "epoch": 1.01, + "learning_rate": 8.154957241494978e-06, + "loss": 3.3397, + "step": 2555904 + }, + { + "epoch": 1.01, + "learning_rate": 8.14657129398446e-06, + "loss": 3.343, + "step": 2556416 + }, + { + "epoch": 1.01, + "learning_rate": 8.13818534647394e-06, + "loss": 3.3276, + "step": 2556928 + }, + { + "epoch": 1.01, + "learning_rate": 8.129799398963418e-06, + "loss": 3.3408, + "step": 2557440 + }, + { + "epoch": 1.01, + "learning_rate": 8.121413451452898e-06, + "loss": 3.3344, + "step": 2557952 + }, + { + "epoch": 1.01, + "learning_rate": 8.11302750394238e-06, + "loss": 3.336, + "step": 2558464 + }, + { + "epoch": 1.01, + "learning_rate": 8.104657935235589e-06, + "loss": 3.3322, + "step": 2558976 + }, + { + "epoch": 1.01, + "learning_rate": 8.096288366528801e-06, + "loss": 3.3347, + "step": 2559488 + }, + { + "epoch": 1.01, + "learning_rate": 8.087902419018281e-06, + "loss": 3.3423, + "step": 2560000 + }, + { + "epoch": 1.01, + "learning_rate": 8.079516471507761e-06, + "loss": 3.3401, + "step": 2560512 + }, + { + "epoch": 1.01, + "learning_rate": 8.071130523997241e-06, + "loss": 3.334, + "step": 2561024 + }, + { + "epoch": 1.01, + "learning_rate": 8.062744576486721e-06, + "loss": 3.3115, + "step": 2561536 + }, + { + "epoch": 1.01, + "learning_rate": 8.054375007779932e-06, + "loss": 3.3254, + "step": 2562048 + }, + { + "epoch": 1.01, + "learning_rate": 8.045989060269413e-06, + "loss": 3.3371, + "step": 2562560 + }, + { + "epoch": 1.01, + "learning_rate": 8.037603112758892e-06, + "loss": 3.3359, + "step": 2563072 + }, + { + "epoch": 1.01, + "learning_rate": 8.029217165248372e-06, + "loss": 3.3321, + "step": 2563584 + }, + { + "epoch": 1.01, + "learning_rate": 8.020831217737852e-06, + "loss": 3.3324, + "step": 2564096 + }, + { + "epoch": 1.02, + "learning_rate": 8.012461649031062e-06, + "loss": 3.3368, + "step": 2564608 + }, + { + "epoch": 1.02, + "learning_rate": 8.004075701520542e-06, + "loss": 3.3297, + "step": 2565120 + }, + { + "epoch": 1.02, + "learning_rate": 7.995689754010022e-06, + "loss": 3.3218, + "step": 2565632 + }, + { + "epoch": 1.02, + "learning_rate": 7.987303806499504e-06, + "loss": 3.3421, + "step": 2566144 + }, + { + "epoch": 1.02, + "learning_rate": 7.978934237792713e-06, + "loss": 3.327, + "step": 2566656 + }, + { + "epoch": 1.02, + "learning_rate": 7.970548290282195e-06, + "loss": 3.3427, + "step": 2567168 + }, + { + "epoch": 1.02, + "learning_rate": 7.962162342771675e-06, + "loss": 3.3389, + "step": 2567680 + }, + { + "epoch": 1.02, + "learning_rate": 7.953776395261153e-06, + "loss": 3.3177, + "step": 2568192 + }, + { + "epoch": 1.02, + "learning_rate": 7.945406826554365e-06, + "loss": 3.3366, + "step": 2568704 + }, + { + "epoch": 1.02, + "learning_rate": 7.937020879043845e-06, + "loss": 3.3336, + "step": 2569216 + }, + { + "epoch": 1.02, + "learning_rate": 7.928634931533325e-06, + "loss": 3.3399, + "step": 2569728 + }, + { + "epoch": 1.02, + "learning_rate": 7.920248984022804e-06, + "loss": 3.3337, + "step": 2570240 + }, + { + "epoch": 1.02, + "learning_rate": 7.911863036512285e-06, + "loss": 3.3305, + "step": 2570752 + }, + { + "epoch": 1.02, + "learning_rate": 7.903493467805496e-06, + "loss": 3.3256, + "step": 2571264 + }, + { + "epoch": 1.02, + "learning_rate": 7.895107520294976e-06, + "loss": 3.3402, + "step": 2571776 + }, + { + "epoch": 1.02, + "learning_rate": 7.886721572784456e-06, + "loss": 3.321, + "step": 2572288 + }, + { + "epoch": 1.02, + "learning_rate": 7.878335625273936e-06, + "loss": 3.3269, + "step": 2572800 + }, + { + "epoch": 1.02, + "learning_rate": 7.869966056567148e-06, + "loss": 3.3343, + "step": 2573312 + }, + { + "epoch": 1.02, + "learning_rate": 7.861580109056627e-06, + "loss": 3.3308, + "step": 2573824 + }, + { + "epoch": 1.02, + "learning_rate": 7.853194161546107e-06, + "loss": 3.3196, + "step": 2574336 + }, + { + "epoch": 1.02, + "learning_rate": 7.844808214035587e-06, + "loss": 3.3378, + "step": 2574848 + }, + { + "epoch": 1.02, + "learning_rate": 7.836438645328799e-06, + "loss": 3.3278, + "step": 2575360 + }, + { + "epoch": 1.02, + "learning_rate": 7.828052697818277e-06, + "loss": 3.3347, + "step": 2575872 + }, + { + "epoch": 1.02, + "learning_rate": 7.819666750307757e-06, + "loss": 3.3346, + "step": 2576384 + }, + { + "epoch": 1.02, + "learning_rate": 7.811280802797239e-06, + "loss": 3.3289, + "step": 2576896 + }, + { + "epoch": 1.02, + "learning_rate": 7.80291123409045e-06, + "loss": 3.3503, + "step": 2577408 + }, + { + "epoch": 1.02, + "learning_rate": 7.79452528657993e-06, + "loss": 3.3404, + "step": 2577920 + }, + { + "epoch": 1.02, + "learning_rate": 7.78613933906941e-06, + "loss": 3.3341, + "step": 2578432 + }, + { + "epoch": 1.02, + "learning_rate": 7.77775339155889e-06, + "loss": 3.327, + "step": 2578944 + }, + { + "epoch": 1.02, + "learning_rate": 7.769367444048368e-06, + "loss": 3.3364, + "step": 2579456 + }, + { + "epoch": 1.02, + "learning_rate": 7.76098149653785e-06, + "loss": 3.3313, + "step": 2579968 + }, + { + "epoch": 1.02, + "learning_rate": 7.75261192783106e-06, + "loss": 3.329, + "step": 2580480 + }, + { + "epoch": 1.02, + "learning_rate": 7.74422598032054e-06, + "loss": 3.3426, + "step": 2580992 + }, + { + "epoch": 1.02, + "learning_rate": 7.73584003281002e-06, + "loss": 3.3331, + "step": 2581504 + }, + { + "epoch": 1.02, + "learning_rate": 7.7274540852995e-06, + "loss": 3.3299, + "step": 2582016 + }, + { + "epoch": 1.02, + "learning_rate": 7.71906813778898e-06, + "loss": 3.3309, + "step": 2582528 + }, + { + "epoch": 1.02, + "learning_rate": 7.710698569082191e-06, + "loss": 3.3319, + "step": 2583040 + }, + { + "epoch": 1.02, + "learning_rate": 7.702312621571671e-06, + "loss": 3.3353, + "step": 2583552 + }, + { + "epoch": 1.02, + "learning_rate": 7.69392667406115e-06, + "loss": 3.329, + "step": 2584064 + }, + { + "epoch": 1.02, + "learning_rate": 7.68554072655063e-06, + "loss": 3.323, + "step": 2584576 + }, + { + "epoch": 1.02, + "learning_rate": 7.67715477904011e-06, + "loss": 3.3476, + "step": 2585088 + }, + { + "epoch": 1.02, + "learning_rate": 7.66876883152959e-06, + "loss": 3.3411, + "step": 2585600 + }, + { + "epoch": 1.02, + "learning_rate": 7.66038288401907e-06, + "loss": 3.3327, + "step": 2586112 + }, + { + "epoch": 1.02, + "learning_rate": 7.65199693650855e-06, + "loss": 3.3324, + "step": 2586624 + }, + { + "epoch": 1.02, + "learning_rate": 7.643627367801761e-06, + "loss": 3.3275, + "step": 2587136 + }, + { + "epoch": 1.02, + "learning_rate": 7.635257799094974e-06, + "loss": 3.3372, + "step": 2587648 + }, + { + "epoch": 1.02, + "learning_rate": 7.626871851584454e-06, + "loss": 3.3331, + "step": 2588160 + }, + { + "epoch": 1.02, + "learning_rate": 7.618485904073933e-06, + "loss": 3.3278, + "step": 2588672 + }, + { + "epoch": 1.02, + "learning_rate": 7.610099956563413e-06, + "loss": 3.3447, + "step": 2589184 + }, + { + "epoch": 1.02, + "learning_rate": 7.6017303878566245e-06, + "loss": 3.3346, + "step": 2589696 + }, + { + "epoch": 1.02, + "learning_rate": 7.593344440346104e-06, + "loss": 3.343, + "step": 2590208 + }, + { + "epoch": 1.02, + "learning_rate": 7.584958492835584e-06, + "loss": 3.3322, + "step": 2590720 + }, + { + "epoch": 1.02, + "learning_rate": 7.576572545325064e-06, + "loss": 3.3411, + "step": 2591232 + }, + { + "epoch": 1.02, + "learning_rate": 7.568202976618276e-06, + "loss": 3.3342, + "step": 2591744 + }, + { + "epoch": 1.02, + "learning_rate": 7.559817029107754e-06, + "loss": 3.333, + "step": 2592256 + }, + { + "epoch": 1.02, + "learning_rate": 7.551431081597235e-06, + "loss": 3.3325, + "step": 2592768 + }, + { + "epoch": 1.02, + "learning_rate": 7.543045134086715e-06, + "loss": 3.3349, + "step": 2593280 + }, + { + "epoch": 1.02, + "learning_rate": 7.534659186576194e-06, + "loss": 3.3401, + "step": 2593792 + }, + { + "epoch": 1.02, + "learning_rate": 7.526289617869406e-06, + "loss": 3.3255, + "step": 2594304 + }, + { + "epoch": 1.02, + "learning_rate": 7.517903670358887e-06, + "loss": 3.3285, + "step": 2594816 + }, + { + "epoch": 1.03, + "eval_loss": 3.8696951866149902, + "eval_runtime": 305.2864, + "eval_samples_per_second": 1249.944, + "eval_steps_per_second": 39.062, + "step": 2594880 + }, + { + "epoch": 0.0, + "learning_rate": 7.5095177228483666e-06, + "loss": 3.3223, + "step": 2595328 + }, + { + "epoch": 0.0, + "learning_rate": 7.501131775337846e-06, + "loss": 3.3283, + "step": 2595840 + }, + { + "epoch": 0.0, + "learning_rate": 7.492762206631057e-06, + "loss": 3.3471, + "step": 2596352 + }, + { + "epoch": 0.0, + "learning_rate": 7.484376259120537e-06, + "loss": 3.3443, + "step": 2596864 + }, + { + "epoch": 0.0, + "learning_rate": 7.475990311610018e-06, + "loss": 3.3365, + "step": 2597376 + }, + { + "epoch": 0.0, + "learning_rate": 7.467604364099496e-06, + "loss": 3.3317, + "step": 2597888 + }, + { + "epoch": 0.0, + "learning_rate": 7.459218416588977e-06, + "loss": 3.3306, + "step": 2598400 + }, + { + "epoch": 0.0, + "learning_rate": 7.450832469078457e-06, + "loss": 3.3225, + "step": 2598912 + }, + { + "epoch": 0.0, + "learning_rate": 7.442462900371668e-06, + "loss": 3.3401, + "step": 2599424 + }, + { + "epoch": 0.0, + "learning_rate": 7.434076952861148e-06, + "loss": 3.3352, + "step": 2599936 + }, + { + "epoch": 0.0, + "learning_rate": 7.425691005350628e-06, + "loss": 3.332, + "step": 2600448 + }, + { + "epoch": 0.0, + "learning_rate": 7.417305057840109e-06, + "loss": 3.3351, + "step": 2600960 + }, + { + "epoch": 0.0, + "learning_rate": 7.4089354891333185e-06, + "loss": 3.3258, + "step": 2601472 + }, + { + "epoch": 0.0, + "learning_rate": 7.400549541622799e-06, + "loss": 3.3224, + "step": 2601984 + }, + { + "epoch": 0.0, + "learning_rate": 7.392163594112279e-06, + "loss": 3.3228, + "step": 2602496 + }, + { + "epoch": 0.0, + "learning_rate": 7.383777646601758e-06, + "loss": 3.3247, + "step": 2603008 + }, + { + "epoch": 0.0, + "learning_rate": 7.375391699091238e-06, + "loss": 3.3263, + "step": 2603520 + }, + { + "epoch": 0.0, + "learning_rate": 7.367005751580719e-06, + "loss": 3.3286, + "step": 2604032 + }, + { + "epoch": 0.0, + "learning_rate": 7.358636182873931e-06, + "loss": 3.3313, + "step": 2604544 + }, + { + "epoch": 0.0, + "learning_rate": 7.35025023536341e-06, + "loss": 3.356, + "step": 2605056 + }, + { + "epoch": 0.0, + "learning_rate": 7.34186428785289e-06, + "loss": 3.3315, + "step": 2605568 + }, + { + "epoch": 0.0, + "learning_rate": 7.33347834034237e-06, + "loss": 3.3331, + "step": 2606080 + }, + { + "epoch": 0.0, + "learning_rate": 7.3251087716355815e-06, + "loss": 3.334, + "step": 2606592 + }, + { + "epoch": 0.0, + "learning_rate": 7.3167228241250606e-06, + "loss": 3.34, + "step": 2607104 + }, + { + "epoch": 0.0, + "learning_rate": 7.308336876614541e-06, + "loss": 3.3264, + "step": 2607616 + }, + { + "epoch": 0.0, + "learning_rate": 7.299950929104021e-06, + "loss": 3.3323, + "step": 2608128 + }, + { + "epoch": 0.0, + "learning_rate": 7.2915649815935004e-06, + "loss": 3.3287, + "step": 2608640 + }, + { + "epoch": 0.0, + "learning_rate": 7.283195412886712e-06, + "loss": 3.3268, + "step": 2609152 + }, + { + "epoch": 0.0, + "learning_rate": 7.274809465376192e-06, + "loss": 3.3211, + "step": 2609664 + }, + { + "epoch": 0.01, + "learning_rate": 7.266423517865673e-06, + "loss": 3.3238, + "step": 2610176 + }, + { + "epoch": 0.01, + "learning_rate": 7.258037570355151e-06, + "loss": 3.3304, + "step": 2610688 + }, + { + "epoch": 0.01, + "learning_rate": 7.249651622844632e-06, + "loss": 3.3374, + "step": 2611200 + }, + { + "epoch": 0.01, + "learning_rate": 7.2412820541378435e-06, + "loss": 3.3308, + "step": 2611712 + }, + { + "epoch": 0.01, + "learning_rate": 7.232896106627323e-06, + "loss": 3.3381, + "step": 2612224 + }, + { + "epoch": 0.01, + "learning_rate": 7.224510159116803e-06, + "loss": 3.3189, + "step": 2612736 + }, + { + "epoch": 0.01, + "learning_rate": 7.216124211606283e-06, + "loss": 3.342, + "step": 2613248 + }, + { + "epoch": 0.01, + "learning_rate": 7.207738264095763e-06, + "loss": 3.3181, + "step": 2613760 + }, + { + "epoch": 0.01, + "learning_rate": 7.1993523165852425e-06, + "loss": 3.3352, + "step": 2614272 + }, + { + "epoch": 0.01, + "learning_rate": 7.190982747878454e-06, + "loss": 3.3167, + "step": 2614784 + }, + { + "epoch": 0.01, + "learning_rate": 7.182596800367934e-06, + "loss": 3.3297, + "step": 2615296 + }, + { + "epoch": 0.01, + "learning_rate": 7.174210852857413e-06, + "loss": 3.3208, + "step": 2615808 + }, + { + "epoch": 0.01, + "learning_rate": 7.165824905346893e-06, + "loss": 3.3417, + "step": 2616320 + }, + { + "epoch": 0.01, + "learning_rate": 7.157438957836374e-06, + "loss": 3.3276, + "step": 2616832 + }, + { + "epoch": 0.01, + "learning_rate": 7.1490693891295856e-06, + "loss": 3.3275, + "step": 2617344 + }, + { + "epoch": 0.01, + "learning_rate": 7.140683441619065e-06, + "loss": 3.3346, + "step": 2617856 + }, + { + "epoch": 0.01, + "learning_rate": 7.132297494108545e-06, + "loss": 3.3369, + "step": 2618368 + }, + { + "epoch": 0.01, + "learning_rate": 7.123911546598025e-06, + "loss": 3.3294, + "step": 2618880 + }, + { + "epoch": 0.01, + "learning_rate": 7.115525599087504e-06, + "loss": 3.3151, + "step": 2619392 + }, + { + "epoch": 0.01, + "learning_rate": 7.1071396515769845e-06, + "loss": 3.3178, + "step": 2619904 + }, + { + "epoch": 0.01, + "learning_rate": 7.0987537040664644e-06, + "loss": 3.3364, + "step": 2620416 + }, + { + "epoch": 0.01, + "learning_rate": 7.0903677565559436e-06, + "loss": 3.3323, + "step": 2620928 + }, + { + "epoch": 0.01, + "learning_rate": 7.081998187849155e-06, + "loss": 3.3257, + "step": 2621440 + }, + { + "epoch": 0.01, + "learning_rate": 7.073612240338635e-06, + "loss": 3.324, + "step": 2621952 + }, + { + "epoch": 0.01, + "learning_rate": 7.065226292828116e-06, + "loss": 3.3256, + "step": 2622464 + }, + { + "epoch": 0.01, + "learning_rate": 7.056840345317594e-06, + "loss": 3.3153, + "step": 2622976 + }, + { + "epoch": 0.01, + "learning_rate": 7.048470776610807e-06, + "loss": 3.3276, + "step": 2623488 + }, + { + "epoch": 0.01, + "learning_rate": 7.040084829100287e-06, + "loss": 3.3152, + "step": 2624000 + }, + { + "epoch": 0.01, + "learning_rate": 7.031698881589767e-06, + "loss": 3.3302, + "step": 2624512 + }, + { + "epoch": 0.01, + "learning_rate": 7.023312934079246e-06, + "loss": 3.3335, + "step": 2625024 + }, + { + "epoch": 0.01, + "learning_rate": 7.0149269865687265e-06, + "loss": 3.3148, + "step": 2625536 + }, + { + "epoch": 0.01, + "learning_rate": 7.0065410390582065e-06, + "loss": 3.3208, + "step": 2626048 + }, + { + "epoch": 0.01, + "learning_rate": 6.998155091547686e-06, + "loss": 3.3207, + "step": 2626560 + }, + { + "epoch": 0.01, + "learning_rate": 6.989785522840897e-06, + "loss": 3.3136, + "step": 2627072 + }, + { + "epoch": 0.01, + "learning_rate": 6.981399575330377e-06, + "loss": 3.3125, + "step": 2627584 + }, + { + "epoch": 0.01, + "learning_rate": 6.973013627819858e-06, + "loss": 3.3335, + "step": 2628096 + }, + { + "epoch": 0.01, + "learning_rate": 6.964627680309336e-06, + "loss": 3.3239, + "step": 2628608 + }, + { + "epoch": 0.01, + "learning_rate": 6.956258111602548e-06, + "loss": 3.3182, + "step": 2629120 + }, + { + "epoch": 0.01, + "learning_rate": 6.947872164092029e-06, + "loss": 3.3198, + "step": 2629632 + }, + { + "epoch": 0.01, + "learning_rate": 6.939486216581508e-06, + "loss": 3.3232, + "step": 2630144 + }, + { + "epoch": 0.01, + "learning_rate": 6.931100269070988e-06, + "loss": 3.3276, + "step": 2630656 + }, + { + "epoch": 0.01, + "learning_rate": 6.922714321560468e-06, + "loss": 3.3276, + "step": 2631168 + }, + { + "epoch": 0.01, + "learning_rate": 6.91434475285368e-06, + "loss": 3.3196, + "step": 2631680 + }, + { + "epoch": 0.01, + "learning_rate": 6.9059588053431585e-06, + "loss": 3.3321, + "step": 2632192 + }, + { + "epoch": 0.01, + "learning_rate": 6.897572857832639e-06, + "loss": 3.3293, + "step": 2632704 + }, + { + "epoch": 0.01, + "learning_rate": 6.889186910322119e-06, + "loss": 3.3198, + "step": 2633216 + }, + { + "epoch": 0.01, + "learning_rate": 6.880800962811598e-06, + "loss": 3.3303, + "step": 2633728 + }, + { + "epoch": 0.01, + "learning_rate": 6.87243139410481e-06, + "loss": 3.3287, + "step": 2634240 + }, + { + "epoch": 0.01, + "learning_rate": 6.86404544659429e-06, + "loss": 3.3249, + "step": 2634752 + }, + { + "epoch": 0.01, + "learning_rate": 6.855659499083771e-06, + "loss": 3.3249, + "step": 2635264 + }, + { + "epoch": 0.01, + "learning_rate": 6.84727355157325e-06, + "loss": 3.3247, + "step": 2635776 + }, + { + "epoch": 0.01, + "learning_rate": 6.8389039828664615e-06, + "loss": 3.3314, + "step": 2636288 + }, + { + "epoch": 0.01, + "learning_rate": 6.830518035355941e-06, + "loss": 3.3338, + "step": 2636800 + }, + { + "epoch": 0.01, + "learning_rate": 6.822132087845421e-06, + "loss": 3.3201, + "step": 2637312 + }, + { + "epoch": 0.01, + "learning_rate": 6.8137461403349005e-06, + "loss": 3.3083, + "step": 2637824 + }, + { + "epoch": 0.01, + "learning_rate": 6.805376571628112e-06, + "loss": 3.3152, + "step": 2638336 + }, + { + "epoch": 0.01, + "learning_rate": 6.796990624117593e-06, + "loss": 3.3242, + "step": 2638848 + }, + { + "epoch": 0.01, + "learning_rate": 6.788604676607071e-06, + "loss": 3.3296, + "step": 2639360 + }, + { + "epoch": 0.01, + "learning_rate": 6.780218729096552e-06, + "loss": 3.3188, + "step": 2639872 + }, + { + "epoch": 0.01, + "learning_rate": 6.771832781586032e-06, + "loss": 3.3229, + "step": 2640384 + }, + { + "epoch": 0.02, + "learning_rate": 6.763463212879244e-06, + "loss": 3.3311, + "step": 2640896 + }, + { + "epoch": 0.02, + "learning_rate": 6.755077265368723e-06, + "loss": 3.3179, + "step": 2641408 + }, + { + "epoch": 0.02, + "learning_rate": 6.7466913178582035e-06, + "loss": 3.3165, + "step": 2641920 + }, + { + "epoch": 0.02, + "learning_rate": 6.7383053703476834e-06, + "loss": 3.3278, + "step": 2642432 + }, + { + "epoch": 0.02, + "learning_rate": 6.729935801640895e-06, + "loss": 3.3208, + "step": 2642944 + }, + { + "epoch": 0.02, + "learning_rate": 6.721549854130374e-06, + "loss": 3.3325, + "step": 2643456 + }, + { + "epoch": 0.02, + "learning_rate": 6.713163906619854e-06, + "loss": 3.3277, + "step": 2643968 + }, + { + "epoch": 0.02, + "learning_rate": 6.704777959109335e-06, + "loss": 3.3093, + "step": 2644480 + }, + { + "epoch": 0.02, + "learning_rate": 6.696408390402545e-06, + "loss": 3.3281, + "step": 2644992 + }, + { + "epoch": 0.02, + "learning_rate": 6.688022442892025e-06, + "loss": 3.3223, + "step": 2645504 + }, + { + "epoch": 0.02, + "learning_rate": 6.679636495381506e-06, + "loss": 3.333, + "step": 2646016 + }, + { + "epoch": 0.02, + "learning_rate": 6.671250547870986e-06, + "loss": 3.3279, + "step": 2646528 + }, + { + "epoch": 0.02, + "learning_rate": 6.662864600360465e-06, + "loss": 3.3193, + "step": 2647040 + }, + { + "epoch": 0.02, + "learning_rate": 6.654495031653676e-06, + "loss": 3.3119, + "step": 2647552 + }, + { + "epoch": 0.02, + "learning_rate": 6.646109084143156e-06, + "loss": 3.3308, + "step": 2648064 + }, + { + "epoch": 0.02, + "learning_rate": 6.637723136632635e-06, + "loss": 3.3114, + "step": 2648576 + }, + { + "epoch": 0.02, + "learning_rate": 6.629337189122116e-06, + "loss": 3.3217, + "step": 2649088 + }, + { + "epoch": 0.02, + "learning_rate": 6.620951241611596e-06, + "loss": 3.3232, + "step": 2649600 + }, + { + "epoch": 0.02, + "learning_rate": 6.612581672904808e-06, + "loss": 3.3216, + "step": 2650112 + }, + { + "epoch": 0.02, + "learning_rate": 6.604195725394287e-06, + "loss": 3.311, + "step": 2650624 + }, + { + "epoch": 0.02, + "learning_rate": 6.595809777883767e-06, + "loss": 3.3326, + "step": 2651136 + }, + { + "epoch": 0.02, + "learning_rate": 6.587423830373248e-06, + "loss": 3.3177, + "step": 2651648 + }, + { + "epoch": 0.02, + "learning_rate": 6.579054261666459e-06, + "loss": 3.3228, + "step": 2652160 + }, + { + "epoch": 0.02, + "learning_rate": 6.5706683141559384e-06, + "loss": 3.3266, + "step": 2652672 + }, + { + "epoch": 0.02, + "learning_rate": 6.562282366645418e-06, + "loss": 3.3207, + "step": 2653184 + }, + { + "epoch": 0.02, + "learning_rate": 6.553896419134898e-06, + "loss": 3.3383, + "step": 2653696 + }, + { + "epoch": 0.02, + "learning_rate": 6.5455104716243774e-06, + "loss": 3.3279, + "step": 2654208 + }, + { + "epoch": 0.02, + "learning_rate": 6.537140902917589e-06, + "loss": 3.3282, + "step": 2654720 + }, + { + "epoch": 0.02, + "learning_rate": 6.52875495540707e-06, + "loss": 3.3157, + "step": 2655232 + }, + { + "epoch": 0.02, + "learning_rate": 6.520369007896548e-06, + "loss": 3.3244, + "step": 2655744 + }, + { + "epoch": 0.02, + "learning_rate": 6.511983060386029e-06, + "loss": 3.3232, + "step": 2656256 + }, + { + "epoch": 0.02, + "learning_rate": 6.503613491679241e-06, + "loss": 3.3163, + "step": 2656768 + }, + { + "epoch": 0.02, + "learning_rate": 6.4952275441687205e-06, + "loss": 3.3285, + "step": 2657280 + }, + { + "epoch": 0.02, + "learning_rate": 6.4868415966582e-06, + "loss": 3.327, + "step": 2657792 + }, + { + "epoch": 0.02, + "learning_rate": 6.47845564914768e-06, + "loss": 3.3208, + "step": 2658304 + }, + { + "epoch": 0.02, + "learning_rate": 6.47006970163716e-06, + "loss": 3.3254, + "step": 2658816 + }, + { + "epoch": 0.02, + "learning_rate": 6.461700132930372e-06, + "loss": 3.3252, + "step": 2659328 + }, + { + "epoch": 0.02, + "learning_rate": 6.453314185419851e-06, + "loss": 3.3195, + "step": 2659840 + }, + { + "epoch": 0.02, + "learning_rate": 6.444928237909331e-06, + "loss": 3.3222, + "step": 2660352 + }, + { + "epoch": 0.02, + "learning_rate": 6.436542290398812e-06, + "loss": 3.3172, + "step": 2660864 + }, + { + "epoch": 0.02, + "learning_rate": 6.42815634288829e-06, + "loss": 3.3354, + "step": 2661376 + }, + { + "epoch": 0.02, + "learning_rate": 6.419770395377771e-06, + "loss": 3.3372, + "step": 2661888 + }, + { + "epoch": 0.02, + "learning_rate": 6.411384447867251e-06, + "loss": 3.3196, + "step": 2662400 + }, + { + "epoch": 0.02, + "learning_rate": 6.4030148791604626e-06, + "loss": 3.3217, + "step": 2662912 + }, + { + "epoch": 0.02, + "learning_rate": 6.394628931649942e-06, + "loss": 3.3202, + "step": 2663424 + }, + { + "epoch": 0.02, + "learning_rate": 6.386242984139422e-06, + "loss": 3.3268, + "step": 2663936 + }, + { + "epoch": 0.02, + "learning_rate": 6.3778570366289024e-06, + "loss": 3.325, + "step": 2664448 + }, + { + "epoch": 0.02, + "learning_rate": 6.369487467922112e-06, + "loss": 3.3203, + "step": 2664960 + }, + { + "epoch": 0.02, + "learning_rate": 6.361101520411593e-06, + "loss": 3.3323, + "step": 2665472 + }, + { + "epoch": 0.02, + "learning_rate": 6.352715572901073e-06, + "loss": 3.3271, + "step": 2665984 + }, + { + "epoch": 0.02, + "learning_rate": 6.344329625390553e-06, + "loss": 3.3365, + "step": 2666496 + }, + { + "epoch": 0.02, + "learning_rate": 6.335943677880032e-06, + "loss": 3.3219, + "step": 2667008 + }, + { + "epoch": 0.02, + "learning_rate": 6.327574109173244e-06, + "loss": 3.3357, + "step": 2667520 + }, + { + "epoch": 0.02, + "learning_rate": 6.319188161662725e-06, + "loss": 3.3279, + "step": 2668032 + }, + { + "epoch": 0.02, + "learning_rate": 6.310802214152203e-06, + "loss": 3.3193, + "step": 2668544 + }, + { + "epoch": 0.02, + "learning_rate": 6.302416266641684e-06, + "loss": 3.3282, + "step": 2669056 + }, + { + "epoch": 0.02, + "learning_rate": 6.294030319131164e-06, + "loss": 3.3287, + "step": 2669568 + }, + { + "epoch": 0.02, + "learning_rate": 6.285660750424375e-06, + "loss": 3.3278, + "step": 2670080 + }, + { + "epoch": 0.02, + "learning_rate": 6.277274802913854e-06, + "loss": 3.3149, + "step": 2670592 + }, + { + "epoch": 0.02, + "learning_rate": 6.268888855403335e-06, + "loss": 3.3177, + "step": 2671104 + }, + { + "epoch": 0.03, + "eval_loss": 3.8697259426116943, + "eval_runtime": 332.4801, + "eval_samples_per_second": 1147.711, + "eval_steps_per_second": 35.867, + "step": 2671200 + }, + { + "epoch": 1.0, + "learning_rate": 6.260502907892815e-06, + "loss": 3.3208, + "step": 2671616 + }, + { + "epoch": 1.0, + "learning_rate": 6.252116960382294e-06, + "loss": 3.3194, + "step": 2672128 + }, + { + "epoch": 1.0, + "learning_rate": 6.243747391675507e-06, + "loss": 3.3329, + "step": 2672640 + }, + { + "epoch": 1.0, + "learning_rate": 6.235361444164986e-06, + "loss": 3.3339, + "step": 2673152 + }, + { + "epoch": 1.0, + "learning_rate": 6.226975496654466e-06, + "loss": 3.3337, + "step": 2673664 + }, + { + "epoch": 1.0, + "learning_rate": 6.218589549143946e-06, + "loss": 3.3176, + "step": 2674176 + }, + { + "epoch": 1.0, + "learning_rate": 6.210203601633426e-06, + "loss": 3.3269, + "step": 2674688 + }, + { + "epoch": 1.0, + "learning_rate": 6.201817654122905e-06, + "loss": 3.3114, + "step": 2675200 + }, + { + "epoch": 1.0, + "learning_rate": 6.1934480854161165e-06, + "loss": 3.3326, + "step": 2675712 + }, + { + "epoch": 1.0, + "learning_rate": 6.1850621379055964e-06, + "loss": 3.3268, + "step": 2676224 + }, + { + "epoch": 1.0, + "learning_rate": 6.176676190395076e-06, + "loss": 3.3176, + "step": 2676736 + }, + { + "epoch": 1.0, + "learning_rate": 6.168290242884556e-06, + "loss": 3.3319, + "step": 2677248 + }, + { + "epoch": 1.0, + "learning_rate": 6.159920674177768e-06, + "loss": 3.3131, + "step": 2677760 + }, + { + "epoch": 1.0, + "learning_rate": 6.151534726667248e-06, + "loss": 3.3177, + "step": 2678272 + }, + { + "epoch": 1.0, + "learning_rate": 6.143148779156728e-06, + "loss": 3.3104, + "step": 2678784 + }, + { + "epoch": 1.0, + "learning_rate": 6.134762831646208e-06, + "loss": 3.3212, + "step": 2679296 + }, + { + "epoch": 1.0, + "learning_rate": 6.126376884135687e-06, + "loss": 3.3174, + "step": 2679808 + }, + { + "epoch": 1.0, + "learning_rate": 6.117990936625168e-06, + "loss": 3.3182, + "step": 2680320 + }, + { + "epoch": 1.0, + "learning_rate": 6.109604989114647e-06, + "loss": 3.3214, + "step": 2680832 + }, + { + "epoch": 1.0, + "learning_rate": 6.1012354204078585e-06, + "loss": 3.3421, + "step": 2681344 + }, + { + "epoch": 1.0, + "learning_rate": 6.0928494728973385e-06, + "loss": 3.3236, + "step": 2681856 + }, + { + "epoch": 1.0, + "learning_rate": 6.0844635253868184e-06, + "loss": 3.325, + "step": 2682368 + }, + { + "epoch": 1.0, + "learning_rate": 6.076077577876298e-06, + "loss": 3.3241, + "step": 2682880 + }, + { + "epoch": 1.0, + "learning_rate": 6.067691630365778e-06, + "loss": 3.3309, + "step": 2683392 + }, + { + "epoch": 1.0, + "learning_rate": 6.05932206165899e-06, + "loss": 3.3182, + "step": 2683904 + }, + { + "epoch": 1.0, + "learning_rate": 6.050936114148469e-06, + "loss": 3.3234, + "step": 2684416 + }, + { + "epoch": 1.0, + "learning_rate": 6.04255016663795e-06, + "loss": 3.3201, + "step": 2684928 + }, + { + "epoch": 1.0, + "learning_rate": 6.034164219127429e-06, + "loss": 3.3166, + "step": 2685440 + }, + { + "epoch": 1.0, + "learning_rate": 6.02577827161691e-06, + "loss": 3.3121, + "step": 2685952 + }, + { + "epoch": 1.01, + "learning_rate": 6.017408702910121e-06, + "loss": 3.3207, + "step": 2686464 + }, + { + "epoch": 1.01, + "learning_rate": 6.0090227553996005e-06, + "loss": 3.3187, + "step": 2686976 + }, + { + "epoch": 1.01, + "learning_rate": 6.0006368078890805e-06, + "loss": 3.3288, + "step": 2687488 + }, + { + "epoch": 1.01, + "learning_rate": 5.99225086037856e-06, + "loss": 3.3238, + "step": 2688000 + }, + { + "epoch": 1.01, + "learning_rate": 5.983881291671771e-06, + "loss": 3.3268, + "step": 2688512 + }, + { + "epoch": 1.01, + "learning_rate": 5.975495344161251e-06, + "loss": 3.3143, + "step": 2689024 + }, + { + "epoch": 1.01, + "learning_rate": 5.967109396650732e-06, + "loss": 3.3281, + "step": 2689536 + }, + { + "epoch": 1.01, + "learning_rate": 5.958723449140211e-06, + "loss": 3.3147, + "step": 2690048 + }, + { + "epoch": 1.01, + "learning_rate": 5.950337501629692e-06, + "loss": 3.3224, + "step": 2690560 + }, + { + "epoch": 1.01, + "learning_rate": 5.941967932922903e-06, + "loss": 3.3069, + "step": 2691072 + }, + { + "epoch": 1.01, + "learning_rate": 5.933581985412383e-06, + "loss": 3.3193, + "step": 2691584 + }, + { + "epoch": 1.01, + "learning_rate": 5.925196037901863e-06, + "loss": 3.314, + "step": 2692096 + }, + { + "epoch": 1.01, + "learning_rate": 5.916810090391342e-06, + "loss": 3.3326, + "step": 2692608 + }, + { + "epoch": 1.01, + "learning_rate": 5.9084241428808225e-06, + "loss": 3.3189, + "step": 2693120 + }, + { + "epoch": 1.01, + "learning_rate": 5.900054574174033e-06, + "loss": 3.3223, + "step": 2693632 + }, + { + "epoch": 1.01, + "learning_rate": 5.891668626663513e-06, + "loss": 3.324, + "step": 2694144 + }, + { + "epoch": 1.01, + "learning_rate": 5.883282679152993e-06, + "loss": 3.3301, + "step": 2694656 + }, + { + "epoch": 1.01, + "learning_rate": 5.874896731642473e-06, + "loss": 3.3182, + "step": 2695168 + }, + { + "epoch": 1.01, + "learning_rate": 5.866510784131953e-06, + "loss": 3.3113, + "step": 2695680 + }, + { + "epoch": 1.01, + "learning_rate": 5.858141215425165e-06, + "loss": 3.3042, + "step": 2696192 + }, + { + "epoch": 1.01, + "learning_rate": 5.849755267914645e-06, + "loss": 3.3282, + "step": 2696704 + }, + { + "epoch": 1.01, + "learning_rate": 5.841369320404124e-06, + "loss": 3.3281, + "step": 2697216 + }, + { + "epoch": 1.01, + "learning_rate": 5.832983372893605e-06, + "loss": 3.3177, + "step": 2697728 + }, + { + "epoch": 1.01, + "learning_rate": 5.8246138041868154e-06, + "loss": 3.3124, + "step": 2698240 + }, + { + "epoch": 1.01, + "learning_rate": 5.816227856676295e-06, + "loss": 3.3209, + "step": 2698752 + }, + { + "epoch": 1.01, + "learning_rate": 5.807841909165775e-06, + "loss": 3.304, + "step": 2699264 + }, + { + "epoch": 1.01, + "learning_rate": 5.799455961655255e-06, + "loss": 3.3218, + "step": 2699776 + }, + { + "epoch": 1.01, + "learning_rate": 5.791070014144735e-06, + "loss": 3.3064, + "step": 2700288 + }, + { + "epoch": 1.01, + "learning_rate": 5.782700445437947e-06, + "loss": 3.3192, + "step": 2700800 + }, + { + "epoch": 1.01, + "learning_rate": 5.774314497927427e-06, + "loss": 3.3236, + "step": 2701312 + }, + { + "epoch": 1.01, + "learning_rate": 5.765928550416906e-06, + "loss": 3.3134, + "step": 2701824 + }, + { + "epoch": 1.01, + "learning_rate": 5.757542602906387e-06, + "loss": 3.3074, + "step": 2702336 + }, + { + "epoch": 1.01, + "learning_rate": 5.749156655395866e-06, + "loss": 3.3164, + "step": 2702848 + }, + { + "epoch": 1.01, + "learning_rate": 5.7407870866890775e-06, + "loss": 3.3044, + "step": 2703360 + }, + { + "epoch": 1.01, + "learning_rate": 5.7324011391785575e-06, + "loss": 3.3075, + "step": 2703872 + }, + { + "epoch": 1.01, + "learning_rate": 5.724015191668037e-06, + "loss": 3.3215, + "step": 2704384 + }, + { + "epoch": 1.01, + "learning_rate": 5.715629244157517e-06, + "loss": 3.3153, + "step": 2704896 + }, + { + "epoch": 1.01, + "learning_rate": 5.707259675450729e-06, + "loss": 3.3107, + "step": 2705408 + }, + { + "epoch": 1.01, + "learning_rate": 5.698873727940208e-06, + "loss": 3.3055, + "step": 2705920 + }, + { + "epoch": 1.01, + "learning_rate": 5.690487780429688e-06, + "loss": 3.3158, + "step": 2706432 + }, + { + "epoch": 1.01, + "learning_rate": 5.682101832919168e-06, + "loss": 3.3143, + "step": 2706944 + }, + { + "epoch": 1.01, + "learning_rate": 5.673715885408648e-06, + "loss": 3.3191, + "step": 2707456 + }, + { + "epoch": 1.01, + "learning_rate": 5.66534631670186e-06, + "loss": 3.313, + "step": 2707968 + }, + { + "epoch": 1.01, + "learning_rate": 5.65696036919134e-06, + "loss": 3.3243, + "step": 2708480 + }, + { + "epoch": 1.01, + "learning_rate": 5.6485744216808195e-06, + "loss": 3.3232, + "step": 2708992 + }, + { + "epoch": 1.01, + "learning_rate": 5.6401884741702995e-06, + "loss": 3.3123, + "step": 2709504 + }, + { + "epoch": 1.01, + "learning_rate": 5.631802526659779e-06, + "loss": 3.3185, + "step": 2710016 + }, + { + "epoch": 1.01, + "learning_rate": 5.62343295795299e-06, + "loss": 3.3213, + "step": 2710528 + }, + { + "epoch": 1.01, + "learning_rate": 5.61504701044247e-06, + "loss": 3.3206, + "step": 2711040 + }, + { + "epoch": 1.01, + "learning_rate": 5.60666106293195e-06, + "loss": 3.315, + "step": 2711552 + }, + { + "epoch": 1.01, + "learning_rate": 5.59827511542143e-06, + "loss": 3.3165, + "step": 2712064 + }, + { + "epoch": 1.01, + "learning_rate": 5.58988916791091e-06, + "loss": 3.3207, + "step": 2712576 + }, + { + "epoch": 1.01, + "learning_rate": 5.581519599204122e-06, + "loss": 3.324, + "step": 2713088 + }, + { + "epoch": 1.01, + "learning_rate": 5.573133651693601e-06, + "loss": 3.3145, + "step": 2713600 + }, + { + "epoch": 1.01, + "learning_rate": 5.564747704183082e-06, + "loss": 3.3023, + "step": 2714112 + }, + { + "epoch": 1.01, + "learning_rate": 5.556361756672561e-06, + "loss": 3.3025, + "step": 2714624 + }, + { + "epoch": 1.01, + "learning_rate": 5.547992187965772e-06, + "loss": 3.3151, + "step": 2715136 + }, + { + "epoch": 1.01, + "learning_rate": 5.539606240455252e-06, + "loss": 3.3205, + "step": 2715648 + }, + { + "epoch": 1.01, + "learning_rate": 5.531220292944732e-06, + "loss": 3.3158, + "step": 2716160 + }, + { + "epoch": 1.01, + "learning_rate": 5.522834345434212e-06, + "loss": 3.3098, + "step": 2716672 + }, + { + "epoch": 1.02, + "learning_rate": 5.514464776727424e-06, + "loss": 3.3207, + "step": 2717184 + }, + { + "epoch": 1.02, + "learning_rate": 5.506078829216904e-06, + "loss": 3.3147, + "step": 2717696 + }, + { + "epoch": 1.02, + "learning_rate": 5.497692881706383e-06, + "loss": 3.3081, + "step": 2718208 + }, + { + "epoch": 1.02, + "learning_rate": 5.489306934195864e-06, + "loss": 3.3149, + "step": 2718720 + }, + { + "epoch": 1.02, + "learning_rate": 5.4809373654890745e-06, + "loss": 3.3203, + "step": 2719232 + }, + { + "epoch": 1.02, + "learning_rate": 5.4725514179785545e-06, + "loss": 3.3244, + "step": 2719744 + }, + { + "epoch": 1.02, + "learning_rate": 5.4641654704680344e-06, + "loss": 3.3116, + "step": 2720256 + }, + { + "epoch": 1.02, + "learning_rate": 5.455779522957514e-06, + "loss": 3.3013, + "step": 2720768 + }, + { + "epoch": 1.02, + "learning_rate": 5.447409954250725e-06, + "loss": 3.3219, + "step": 2721280 + }, + { + "epoch": 1.02, + "learning_rate": 5.439024006740206e-06, + "loss": 3.3119, + "step": 2721792 + }, + { + "epoch": 1.02, + "learning_rate": 5.430638059229685e-06, + "loss": 3.3243, + "step": 2722304 + }, + { + "epoch": 1.02, + "learning_rate": 5.422252111719165e-06, + "loss": 3.3204, + "step": 2722816 + }, + { + "epoch": 1.02, + "learning_rate": 5.413866164208645e-06, + "loss": 3.3108, + "step": 2723328 + }, + { + "epoch": 1.02, + "learning_rate": 5.405496595501857e-06, + "loss": 3.3048, + "step": 2723840 + }, + { + "epoch": 1.02, + "learning_rate": 5.397110647991337e-06, + "loss": 3.3224, + "step": 2724352 + }, + { + "epoch": 1.02, + "learning_rate": 5.3887247004808165e-06, + "loss": 3.3036, + "step": 2724864 + }, + { + "epoch": 1.02, + "learning_rate": 5.3803387529702965e-06, + "loss": 3.3139, + "step": 2725376 + }, + { + "epoch": 1.02, + "learning_rate": 5.371969184263507e-06, + "loss": 3.3095, + "step": 2725888 + }, + { + "epoch": 1.02, + "learning_rate": 5.363583236752988e-06, + "loss": 3.3161, + "step": 2726400 + }, + { + "epoch": 1.02, + "learning_rate": 5.355197289242467e-06, + "loss": 3.3016, + "step": 2726912 + }, + { + "epoch": 1.02, + "learning_rate": 5.346811341731947e-06, + "loss": 3.3224, + "step": 2727424 + }, + { + "epoch": 1.02, + "learning_rate": 5.338441773025159e-06, + "loss": 3.3088, + "step": 2727936 + }, + { + "epoch": 1.02, + "learning_rate": 5.330055825514639e-06, + "loss": 3.3148, + "step": 2728448 + }, + { + "epoch": 1.02, + "learning_rate": 5.321669878004119e-06, + "loss": 3.3188, + "step": 2728960 + }, + { + "epoch": 1.02, + "learning_rate": 5.313283930493599e-06, + "loss": 3.3129, + "step": 2729472 + }, + { + "epoch": 1.02, + "learning_rate": 5.30491436178681e-06, + "loss": 3.3281, + "step": 2729984 + }, + { + "epoch": 1.02, + "learning_rate": 5.296528414276289e-06, + "loss": 3.3243, + "step": 2730496 + }, + { + "epoch": 1.02, + "learning_rate": 5.28814246676577e-06, + "loss": 3.3165, + "step": 2731008 + }, + { + "epoch": 1.02, + "learning_rate": 5.279756519255249e-06, + "loss": 3.3073, + "step": 2731520 + }, + { + "epoch": 1.02, + "learning_rate": 5.271370571744729e-06, + "loss": 3.3188, + "step": 2732032 + }, + { + "epoch": 1.02, + "learning_rate": 5.262984624234209e-06, + "loss": 3.3138, + "step": 2732544 + }, + { + "epoch": 1.02, + "learning_rate": 5.25461505552742e-06, + "loss": 3.3099, + "step": 2733056 + }, + { + "epoch": 1.02, + "learning_rate": 5.246229108016901e-06, + "loss": 3.3182, + "step": 2733568 + }, + { + "epoch": 1.02, + "learning_rate": 5.23784316050638e-06, + "loss": 3.3167, + "step": 2734080 + }, + { + "epoch": 1.02, + "learning_rate": 5.229457212995861e-06, + "loss": 3.3115, + "step": 2734592 + }, + { + "epoch": 1.02, + "learning_rate": 5.22107126548534e-06, + "loss": 3.3142, + "step": 2735104 + }, + { + "epoch": 1.02, + "learning_rate": 5.212701696778552e-06, + "loss": 3.319, + "step": 2735616 + }, + { + "epoch": 1.02, + "learning_rate": 5.2043157492680314e-06, + "loss": 3.313, + "step": 2736128 + }, + { + "epoch": 1.02, + "learning_rate": 5.195929801757511e-06, + "loss": 3.3161, + "step": 2736640 + }, + { + "epoch": 1.02, + "learning_rate": 5.187543854246991e-06, + "loss": 3.3089, + "step": 2737152 + }, + { + "epoch": 1.02, + "learning_rate": 5.179157906736471e-06, + "loss": 3.3226, + "step": 2737664 + }, + { + "epoch": 1.02, + "learning_rate": 5.170771959225951e-06, + "loss": 3.3227, + "step": 2738176 + }, + { + "epoch": 1.02, + "learning_rate": 5.162386011715431e-06, + "loss": 3.3181, + "step": 2738688 + }, + { + "epoch": 1.02, + "learning_rate": 5.154016443008643e-06, + "loss": 3.3113, + "step": 2739200 + }, + { + "epoch": 1.02, + "learning_rate": 5.145630495498122e-06, + "loss": 3.3121, + "step": 2739712 + }, + { + "epoch": 1.02, + "learning_rate": 5.137244547987602e-06, + "loss": 3.319, + "step": 2740224 + }, + { + "epoch": 1.02, + "learning_rate": 5.128858600477082e-06, + "loss": 3.3172, + "step": 2740736 + }, + { + "epoch": 1.02, + "learning_rate": 5.1204890317702935e-06, + "loss": 3.3068, + "step": 2741248 + }, + { + "epoch": 1.02, + "learning_rate": 5.1121030842597735e-06, + "loss": 3.3208, + "step": 2741760 + }, + { + "epoch": 1.02, + "learning_rate": 5.1037171367492534e-06, + "loss": 3.317, + "step": 2742272 + }, + { + "epoch": 1.02, + "learning_rate": 5.095331189238733e-06, + "loss": 3.3285, + "step": 2742784 + }, + { + "epoch": 1.02, + "learning_rate": 5.086945241728213e-06, + "loss": 3.3173, + "step": 2743296 + }, + { + "epoch": 1.02, + "learning_rate": 5.078575673021425e-06, + "loss": 3.3252, + "step": 2743808 + }, + { + "epoch": 1.02, + "learning_rate": 5.070189725510904e-06, + "loss": 3.3221, + "step": 2744320 + }, + { + "epoch": 1.02, + "learning_rate": 5.061803778000384e-06, + "loss": 3.3104, + "step": 2744832 + }, + { + "epoch": 1.02, + "learning_rate": 5.053417830489864e-06, + "loss": 3.3148, + "step": 2745344 + }, + { + "epoch": 1.02, + "learning_rate": 5.045031882979344e-06, + "loss": 3.3248, + "step": 2745856 + }, + { + "epoch": 1.02, + "learning_rate": 5.036662314272556e-06, + "loss": 3.3159, + "step": 2746368 + }, + { + "epoch": 1.02, + "learning_rate": 5.0282763667620355e-06, + "loss": 3.3051, + "step": 2746880 + }, + { + "epoch": 1.02, + "learning_rate": 5.0198904192515155e-06, + "loss": 3.3149, + "step": 2747392 + }, + { + "epoch": 1.03, + "eval_loss": 3.8697521686553955, + "eval_runtime": 321.4147, + "eval_samples_per_second": 1187.223, + "eval_steps_per_second": 37.102, + "step": 2747520 + }, + { + "epoch": 0.0, + "learning_rate": 5.0115044717409955e-06, + "loss": 3.3068, + "step": 2747904 + }, + { + "epoch": 0.0, + "learning_rate": 5.0031185242304746e-06, + "loss": 3.3095, + "step": 2748416 + }, + { + "epoch": 0.0, + "learning_rate": 4.994748955523686e-06, + "loss": 3.3228, + "step": 2748928 + }, + { + "epoch": 0.0, + "learning_rate": 4.986363008013166e-06, + "loss": 3.3259, + "step": 2749440 + }, + { + "epoch": 0.0, + "learning_rate": 4.977977060502646e-06, + "loss": 3.3247, + "step": 2749952 + }, + { + "epoch": 0.0, + "learning_rate": 4.969591112992126e-06, + "loss": 3.3146, + "step": 2750464 + }, + { + "epoch": 0.0, + "learning_rate": 4.961205165481605e-06, + "loss": 3.3158, + "step": 2750976 + }, + { + "epoch": 0.0, + "learning_rate": 4.952835596774817e-06, + "loss": 3.3003, + "step": 2751488 + }, + { + "epoch": 0.0, + "learning_rate": 4.944449649264297e-06, + "loss": 3.3237, + "step": 2752000 + }, + { + "epoch": 0.0, + "learning_rate": 4.936063701753777e-06, + "loss": 3.3197, + "step": 2752512 + }, + { + "epoch": 0.0, + "learning_rate": 4.927677754243257e-06, + "loss": 3.3094, + "step": 2753024 + }, + { + "epoch": 0.0, + "learning_rate": 4.919308185536468e-06, + "loss": 3.3207, + "step": 2753536 + }, + { + "epoch": 0.0, + "learning_rate": 4.910922238025948e-06, + "loss": 3.3065, + "step": 2754048 + }, + { + "epoch": 0.0, + "learning_rate": 4.902536290515428e-06, + "loss": 3.3107, + "step": 2754560 + }, + { + "epoch": 0.0, + "learning_rate": 4.894150343004908e-06, + "loss": 3.303, + "step": 2755072 + }, + { + "epoch": 0.0, + "learning_rate": 4.885764395494387e-06, + "loss": 3.308, + "step": 2755584 + }, + { + "epoch": 0.0, + "learning_rate": 4.877378447983868e-06, + "loss": 3.3086, + "step": 2756096 + }, + { + "epoch": 0.0, + "learning_rate": 4.868992500473347e-06, + "loss": 3.309, + "step": 2756608 + }, + { + "epoch": 0.0, + "learning_rate": 4.860606552962828e-06, + "loss": 3.3124, + "step": 2757120 + }, + { + "epoch": 0.0, + "learning_rate": 4.8522533630597705e-06, + "loss": 3.3327, + "step": 2757632 + }, + { + "epoch": 0.0, + "learning_rate": 4.8438674155492504e-06, + "loss": 3.3152, + "step": 2758144 + }, + { + "epoch": 0.0, + "learning_rate": 4.83548146803873e-06, + "loss": 3.3187, + "step": 2758656 + }, + { + "epoch": 0.0, + "learning_rate": 4.82709552052821e-06, + "loss": 3.3138, + "step": 2759168 + }, + { + "epoch": 0.0, + "learning_rate": 4.81870957301769e-06, + "loss": 3.3198, + "step": 2759680 + }, + { + "epoch": 0.0, + "learning_rate": 4.810323625507169e-06, + "loss": 3.3137, + "step": 2760192 + }, + { + "epoch": 0.0, + "learning_rate": 4.801954056800381e-06, + "loss": 3.3167, + "step": 2760704 + }, + { + "epoch": 0.0, + "learning_rate": 4.793568109289861e-06, + "loss": 3.316, + "step": 2761216 + }, + { + "epoch": 0.0, + "learning_rate": 4.785182161779341e-06, + "loss": 3.3063, + "step": 2761728 + }, + { + "epoch": 0.0, + "learning_rate": 4.776796214268821e-06, + "loss": 3.302, + "step": 2762240 + }, + { + "epoch": 0.0, + "learning_rate": 4.7684266455620326e-06, + "loss": 3.3132, + "step": 2762752 + }, + { + "epoch": 0.01, + "learning_rate": 4.760040698051512e-06, + "loss": 3.311, + "step": 2763264 + }, + { + "epoch": 0.01, + "learning_rate": 4.7516547505409925e-06, + "loss": 3.3202, + "step": 2763776 + }, + { + "epoch": 0.01, + "learning_rate": 4.7432688030304716e-06, + "loss": 3.3138, + "step": 2764288 + }, + { + "epoch": 0.01, + "learning_rate": 4.734899234323684e-06, + "loss": 3.3195, + "step": 2764800 + }, + { + "epoch": 0.01, + "learning_rate": 4.726513286813163e-06, + "loss": 3.3026, + "step": 2765312 + }, + { + "epoch": 0.01, + "learning_rate": 4.718127339302643e-06, + "loss": 3.3199, + "step": 2765824 + }, + { + "epoch": 0.01, + "learning_rate": 4.709741391792123e-06, + "loss": 3.3069, + "step": 2766336 + }, + { + "epoch": 0.01, + "learning_rate": 4.701355444281603e-06, + "loss": 3.3128, + "step": 2766848 + }, + { + "epoch": 0.01, + "learning_rate": 4.692985875574815e-06, + "loss": 3.2984, + "step": 2767360 + }, + { + "epoch": 0.01, + "learning_rate": 4.684599928064294e-06, + "loss": 3.3097, + "step": 2767872 + }, + { + "epoch": 0.01, + "learning_rate": 4.676213980553775e-06, + "loss": 3.3084, + "step": 2768384 + }, + { + "epoch": 0.01, + "learning_rate": 4.667828033043254e-06, + "loss": 3.3227, + "step": 2768896 + }, + { + "epoch": 0.01, + "learning_rate": 4.659458464336465e-06, + "loss": 3.3102, + "step": 2769408 + }, + { + "epoch": 0.01, + "learning_rate": 4.651072516825945e-06, + "loss": 3.3125, + "step": 2769920 + }, + { + "epoch": 0.01, + "learning_rate": 4.642686569315425e-06, + "loss": 3.3153, + "step": 2770432 + }, + { + "epoch": 0.01, + "learning_rate": 4.634300621804905e-06, + "loss": 3.3179, + "step": 2770944 + }, + { + "epoch": 0.01, + "learning_rate": 4.625914674294385e-06, + "loss": 3.3098, + "step": 2771456 + }, + { + "epoch": 0.01, + "learning_rate": 4.617545105587597e-06, + "loss": 3.3017, + "step": 2771968 + }, + { + "epoch": 0.01, + "learning_rate": 4.609159158077076e-06, + "loss": 3.2959, + "step": 2772480 + }, + { + "epoch": 0.01, + "learning_rate": 4.600773210566557e-06, + "loss": 3.3204, + "step": 2772992 + }, + { + "epoch": 0.01, + "learning_rate": 4.592387263056036e-06, + "loss": 3.3171, + "step": 2773504 + }, + { + "epoch": 0.01, + "learning_rate": 4.584001315545516e-06, + "loss": 3.3096, + "step": 2774016 + }, + { + "epoch": 0.01, + "learning_rate": 4.575631746838727e-06, + "loss": 3.3069, + "step": 2774528 + }, + { + "epoch": 0.01, + "learning_rate": 4.567245799328207e-06, + "loss": 3.3101, + "step": 2775040 + }, + { + "epoch": 0.01, + "learning_rate": 4.558859851817687e-06, + "loss": 3.2948, + "step": 2775552 + }, + { + "epoch": 0.01, + "learning_rate": 4.550473904307167e-06, + "loss": 3.3077, + "step": 2776064 + }, + { + "epoch": 0.01, + "learning_rate": 4.542087956796647e-06, + "loss": 3.3003, + "step": 2776576 + }, + { + "epoch": 0.01, + "learning_rate": 4.533718388089858e-06, + "loss": 3.3092, + "step": 2777088 + }, + { + "epoch": 0.01, + "learning_rate": 4.525332440579339e-06, + "loss": 3.3178, + "step": 2777600 + }, + { + "epoch": 0.01, + "learning_rate": 4.516946493068818e-06, + "loss": 3.3044, + "step": 2778112 + }, + { + "epoch": 0.01, + "learning_rate": 4.508560545558298e-06, + "loss": 3.2975, + "step": 2778624 + }, + { + "epoch": 0.01, + "learning_rate": 4.500174598047778e-06, + "loss": 3.3094, + "step": 2779136 + }, + { + "epoch": 0.01, + "learning_rate": 4.491805029340989e-06, + "loss": 3.2962, + "step": 2779648 + }, + { + "epoch": 0.01, + "learning_rate": 4.4834190818304694e-06, + "loss": 3.3019, + "step": 2780160 + }, + { + "epoch": 0.01, + "learning_rate": 4.4750331343199485e-06, + "loss": 3.3126, + "step": 2780672 + }, + { + "epoch": 0.01, + "learning_rate": 4.466647186809429e-06, + "loss": 3.3083, + "step": 2781184 + }, + { + "epoch": 0.01, + "learning_rate": 4.4582612392989085e-06, + "loss": 3.301, + "step": 2781696 + }, + { + "epoch": 0.01, + "learning_rate": 4.44989167059212e-06, + "loss": 3.3009, + "step": 2782208 + }, + { + "epoch": 0.01, + "learning_rate": 4.4415057230816e-06, + "loss": 3.3043, + "step": 2782720 + }, + { + "epoch": 0.01, + "learning_rate": 4.43311977557108e-06, + "loss": 3.3058, + "step": 2783232 + }, + { + "epoch": 0.01, + "learning_rate": 4.42473382806056e-06, + "loss": 3.3092, + "step": 2783744 + }, + { + "epoch": 0.01, + "learning_rate": 4.416364259353771e-06, + "loss": 3.3027, + "step": 2784256 + }, + { + "epoch": 0.01, + "learning_rate": 4.4079783118432516e-06, + "loss": 3.3165, + "step": 2784768 + }, + { + "epoch": 0.01, + "learning_rate": 4.399592364332731e-06, + "loss": 3.3157, + "step": 2785280 + }, + { + "epoch": 0.01, + "learning_rate": 4.3912064168222115e-06, + "loss": 3.3058, + "step": 2785792 + }, + { + "epoch": 0.01, + "learning_rate": 4.3828204693116906e-06, + "loss": 3.3036, + "step": 2786304 + }, + { + "epoch": 0.01, + "learning_rate": 4.3744345218011705e-06, + "loss": 3.319, + "step": 2786816 + }, + { + "epoch": 0.01, + "learning_rate": 4.3660485742906505e-06, + "loss": 3.3096, + "step": 2787328 + }, + { + "epoch": 0.01, + "learning_rate": 4.357679005583862e-06, + "loss": 3.3051, + "step": 2787840 + }, + { + "epoch": 0.01, + "learning_rate": 4.349309436877074e-06, + "loss": 3.3084, + "step": 2788352 + }, + { + "epoch": 0.01, + "learning_rate": 4.340923489366553e-06, + "loss": 3.3111, + "step": 2788864 + }, + { + "epoch": 0.01, + "learning_rate": 4.332537541856034e-06, + "loss": 3.3199, + "step": 2789376 + }, + { + "epoch": 0.01, + "learning_rate": 4.324151594345513e-06, + "loss": 3.3012, + "step": 2789888 + }, + { + "epoch": 0.01, + "learning_rate": 4.315765646834993e-06, + "loss": 3.2931, + "step": 2790400 + }, + { + "epoch": 0.01, + "learning_rate": 4.307379699324473e-06, + "loss": 3.2999, + "step": 2790912 + }, + { + "epoch": 0.01, + "learning_rate": 4.2990101306176835e-06, + "loss": 3.3008, + "step": 2791424 + }, + { + "epoch": 0.01, + "learning_rate": 4.290624183107164e-06, + "loss": 3.3108, + "step": 2791936 + }, + { + "epoch": 0.01, + "learning_rate": 4.282238235596643e-06, + "loss": 3.3083, + "step": 2792448 + }, + { + "epoch": 0.01, + "learning_rate": 4.273852288086124e-06, + "loss": 3.3009, + "step": 2792960 + }, + { + "epoch": 0.02, + "learning_rate": 4.265482719379335e-06, + "loss": 3.3125, + "step": 2793472 + }, + { + "epoch": 0.02, + "learning_rate": 4.257096771868816e-06, + "loss": 3.3049, + "step": 2793984 + }, + { + "epoch": 0.02, + "learning_rate": 4.248710824358295e-06, + "loss": 3.3049, + "step": 2794496 + }, + { + "epoch": 0.02, + "learning_rate": 4.240324876847775e-06, + "loss": 3.304, + "step": 2795008 + }, + { + "epoch": 0.02, + "learning_rate": 4.231938929337255e-06, + "loss": 3.3123, + "step": 2795520 + }, + { + "epoch": 0.02, + "learning_rate": 4.223569360630466e-06, + "loss": 3.3191, + "step": 2796032 + }, + { + "epoch": 0.02, + "learning_rate": 4.215183413119946e-06, + "loss": 3.31, + "step": 2796544 + }, + { + "epoch": 0.02, + "learning_rate": 4.2067974656094255e-06, + "loss": 3.2873, + "step": 2797056 + }, + { + "epoch": 0.02, + "learning_rate": 4.198427896902637e-06, + "loss": 3.3123, + "step": 2797568 + }, + { + "epoch": 0.02, + "learning_rate": 4.190041949392117e-06, + "loss": 3.3065, + "step": 2798080 + }, + { + "epoch": 0.02, + "learning_rate": 4.181656001881597e-06, + "loss": 3.3103, + "step": 2798592 + }, + { + "epoch": 0.02, + "learning_rate": 4.173270054371077e-06, + "loss": 3.3141, + "step": 2799104 + }, + { + "epoch": 0.02, + "learning_rate": 4.164884106860557e-06, + "loss": 3.3057, + "step": 2799616 + }, + { + "epoch": 0.02, + "learning_rate": 4.156514538153769e-06, + "loss": 3.2996, + "step": 2800128 + }, + { + "epoch": 0.02, + "learning_rate": 4.148128590643248e-06, + "loss": 3.3114, + "step": 2800640 + }, + { + "epoch": 0.02, + "learning_rate": 4.1397426431327285e-06, + "loss": 3.294, + "step": 2801152 + }, + { + "epoch": 0.02, + "learning_rate": 4.131356695622208e-06, + "loss": 3.3059, + "step": 2801664 + }, + { + "epoch": 0.02, + "learning_rate": 4.122987126915419e-06, + "loss": 3.3003, + "step": 2802176 + }, + { + "epoch": 0.02, + "learning_rate": 4.114601179404899e-06, + "loss": 3.3089, + "step": 2802688 + }, + { + "epoch": 0.02, + "learning_rate": 4.106215231894379e-06, + "loss": 3.2949, + "step": 2803200 + }, + { + "epoch": 0.02, + "learning_rate": 4.097829284383859e-06, + "loss": 3.3145, + "step": 2803712 + }, + { + "epoch": 0.02, + "learning_rate": 4.089459715677071e-06, + "loss": 3.2992, + "step": 2804224 + }, + { + "epoch": 0.02, + "learning_rate": 4.081073768166551e-06, + "loss": 3.3043, + "step": 2804736 + }, + { + "epoch": 0.02, + "learning_rate": 4.07268782065603e-06, + "loss": 3.3134, + "step": 2805248 + }, + { + "epoch": 0.02, + "learning_rate": 4.064301873145511e-06, + "loss": 3.3094, + "step": 2805760 + }, + { + "epoch": 0.02, + "learning_rate": 4.0559323044387214e-06, + "loss": 3.3176, + "step": 2806272 + }, + { + "epoch": 0.02, + "learning_rate": 4.047546356928201e-06, + "loss": 3.3143, + "step": 2806784 + }, + { + "epoch": 0.02, + "learning_rate": 4.039160409417681e-06, + "loss": 3.3054, + "step": 2807296 + }, + { + "epoch": 0.02, + "learning_rate": 4.030774461907161e-06, + "loss": 3.3068, + "step": 2807808 + }, + { + "epoch": 0.02, + "learning_rate": 4.022388514396641e-06, + "loss": 3.3033, + "step": 2808320 + }, + { + "epoch": 0.02, + "learning_rate": 4.014018945689853e-06, + "loss": 3.3059, + "step": 2808832 + }, + { + "epoch": 0.02, + "learning_rate": 4.005632998179332e-06, + "loss": 3.2997, + "step": 2809344 + }, + { + "epoch": 0.02, + "learning_rate": 3.997247050668812e-06, + "loss": 3.3118, + "step": 2809856 + }, + { + "epoch": 0.02, + "learning_rate": 3.988861103158292e-06, + "loss": 3.3045, + "step": 2810368 + }, + { + "epoch": 0.02, + "learning_rate": 3.980475155647772e-06, + "loss": 3.3034, + "step": 2810880 + }, + { + "epoch": 0.02, + "learning_rate": 3.972089208137253e-06, + "loss": 3.3031, + "step": 2811392 + }, + { + "epoch": 0.02, + "learning_rate": 3.9637196394304635e-06, + "loss": 3.311, + "step": 2811904 + }, + { + "epoch": 0.02, + "learning_rate": 3.955333691919943e-06, + "loss": 3.3051, + "step": 2812416 + }, + { + "epoch": 0.02, + "learning_rate": 3.946947744409423e-06, + "loss": 3.3086, + "step": 2812928 + }, + { + "epoch": 0.02, + "learning_rate": 3.9385617968989025e-06, + "loss": 3.3021, + "step": 2813440 + }, + { + "epoch": 0.02, + "learning_rate": 3.930175849388383e-06, + "loss": 3.3145, + "step": 2813952 + }, + { + "epoch": 0.02, + "learning_rate": 3.921789901877862e-06, + "loss": 3.3094, + "step": 2814464 + }, + { + "epoch": 0.02, + "learning_rate": 3.913403954367343e-06, + "loss": 3.3105, + "step": 2814976 + }, + { + "epoch": 0.02, + "learning_rate": 3.905034385660554e-06, + "loss": 3.3048, + "step": 2815488 + }, + { + "epoch": 0.02, + "learning_rate": 3.896648438150034e-06, + "loss": 3.3021, + "step": 2816000 + }, + { + "epoch": 0.02, + "learning_rate": 3.888262490639514e-06, + "loss": 3.3122, + "step": 2816512 + }, + { + "epoch": 0.02, + "learning_rate": 3.879876543128994e-06, + "loss": 3.3055, + "step": 2817024 + }, + { + "epoch": 0.02, + "learning_rate": 3.8715069744222055e-06, + "loss": 3.298, + "step": 2817536 + }, + { + "epoch": 0.02, + "learning_rate": 3.863121026911685e-06, + "loss": 3.313, + "step": 2818048 + }, + { + "epoch": 0.02, + "learning_rate": 3.854735079401165e-06, + "loss": 3.3104, + "step": 2818560 + }, + { + "epoch": 0.02, + "learning_rate": 3.8463491318906445e-06, + "loss": 3.322, + "step": 2819072 + }, + { + "epoch": 0.02, + "learning_rate": 3.837963184380125e-06, + "loss": 3.3087, + "step": 2819584 + }, + { + "epoch": 0.02, + "learning_rate": 3.829577236869604e-06, + "loss": 3.3138, + "step": 2820096 + }, + { + "epoch": 0.02, + "learning_rate": 3.821207668162816e-06, + "loss": 3.314, + "step": 2820608 + }, + { + "epoch": 0.02, + "learning_rate": 3.812821720652296e-06, + "loss": 3.3023, + "step": 2821120 + }, + { + "epoch": 0.02, + "learning_rate": 3.8044357731417755e-06, + "loss": 3.3085, + "step": 2821632 + }, + { + "epoch": 0.02, + "learning_rate": 3.796049825631256e-06, + "loss": 3.3121, + "step": 2822144 + }, + { + "epoch": 0.02, + "learning_rate": 3.7876802569244667e-06, + "loss": 3.3113, + "step": 2822656 + }, + { + "epoch": 0.02, + "learning_rate": 3.779294309413947e-06, + "loss": 3.2962, + "step": 2823168 + }, + { + "epoch": 0.02, + "learning_rate": 3.7709083619034266e-06, + "loss": 3.306, + "step": 2823680 + }, + { + "epoch": 0.03, + "eval_loss": 3.870501756668091, + "eval_runtime": 303.0942, + "eval_samples_per_second": 1258.985, + "eval_steps_per_second": 39.344, + "step": 2823840 + }, + { + "epoch": 0.0, + "learning_rate": 3.762522414392907e-06, + "loss": 3.3005, + "step": 2824192 + }, + { + "epoch": 0.0, + "learning_rate": 3.7541364668823865e-06, + "loss": 3.3021, + "step": 2824704 + }, + { + "epoch": 0.0, + "learning_rate": 3.7457668981755977e-06, + "loss": 3.3121, + "step": 2825216 + }, + { + "epoch": 0.0, + "learning_rate": 3.737380950665078e-06, + "loss": 3.3175, + "step": 2825728 + }, + { + "epoch": 0.0, + "learning_rate": 3.7289950031545577e-06, + "loss": 3.3141, + "step": 2826240 + }, + { + "epoch": 0.0, + "learning_rate": 3.720609055644038e-06, + "loss": 3.306, + "step": 2826752 + }, + { + "epoch": 0.0, + "learning_rate": 3.7122231081335176e-06, + "loss": 3.3048, + "step": 2827264 + }, + { + "epoch": 0.0, + "learning_rate": 3.703837160622997e-06, + "loss": 3.2953, + "step": 2827776 + }, + { + "epoch": 0.0, + "learning_rate": 3.6954675919162087e-06, + "loss": 3.3127, + "step": 2828288 + }, + { + "epoch": 0.0, + "learning_rate": 3.6870816444056883e-06, + "loss": 3.3141, + "step": 2828800 + }, + { + "epoch": 0.0, + "learning_rate": 3.6786956968951686e-06, + "loss": 3.3028, + "step": 2829312 + }, + { + "epoch": 0.0, + "learning_rate": 3.670309749384648e-06, + "loss": 3.3105, + "step": 2829824 + }, + { + "epoch": 0.0, + "learning_rate": 3.6619238018741286e-06, + "loss": 3.3003, + "step": 2830336 + }, + { + "epoch": 0.0, + "learning_rate": 3.6535542331673398e-06, + "loss": 3.3022, + "step": 2830848 + }, + { + "epoch": 0.0, + "learning_rate": 3.6451682856568197e-06, + "loss": 3.294, + "step": 2831360 + }, + { + "epoch": 0.0, + "learning_rate": 3.6367823381462997e-06, + "loss": 3.3002, + "step": 2831872 + }, + { + "epoch": 0.0, + "learning_rate": 3.6283963906357792e-06, + "loss": 3.2982, + "step": 2832384 + }, + { + "epoch": 0.0, + "learning_rate": 3.6200104431252596e-06, + "loss": 3.3022, + "step": 2832896 + }, + { + "epoch": 0.0, + "learning_rate": 3.6116408744184704e-06, + "loss": 3.3048, + "step": 2833408 + }, + { + "epoch": 0.0, + "learning_rate": 3.6032549269079508e-06, + "loss": 3.3209, + "step": 2833920 + }, + { + "epoch": 0.0, + "learning_rate": 3.5948689793974303e-06, + "loss": 3.312, + "step": 2834432 + }, + { + "epoch": 0.0, + "learning_rate": 3.5864830318869107e-06, + "loss": 3.3088, + "step": 2834944 + }, + { + "epoch": 0.0, + "learning_rate": 3.57809708437639e-06, + "loss": 3.3047, + "step": 2835456 + }, + { + "epoch": 0.0, + "learning_rate": 3.569727515669602e-06, + "loss": 3.3127, + "step": 2835968 + }, + { + "epoch": 0.0, + "learning_rate": 3.5613415681590814e-06, + "loss": 3.301, + "step": 2836480 + }, + { + "epoch": 0.0, + "learning_rate": 3.5529556206485613e-06, + "loss": 3.3132, + "step": 2836992 + }, + { + "epoch": 0.0, + "learning_rate": 3.5445696731380417e-06, + "loss": 3.3027, + "step": 2837504 + }, + { + "epoch": 0.0, + "learning_rate": 3.5362001044312525e-06, + "loss": 3.2965, + "step": 2838016 + }, + { + "epoch": 0.0, + "learning_rate": 3.527814156920733e-06, + "loss": 3.2979, + "step": 2838528 + }, + { + "epoch": 0.0, + "learning_rate": 3.5194282094102124e-06, + "loss": 3.3022, + "step": 2839040 + }, + { + "epoch": 0.01, + "learning_rate": 3.511042261899693e-06, + "loss": 3.3058, + "step": 2839552 + }, + { + "epoch": 0.01, + "learning_rate": 3.5026726931929036e-06, + "loss": 3.3074, + "step": 2840064 + }, + { + "epoch": 0.01, + "learning_rate": 3.494286745682384e-06, + "loss": 3.3092, + "step": 2840576 + }, + { + "epoch": 0.01, + "learning_rate": 3.4859007981718635e-06, + "loss": 3.3098, + "step": 2841088 + }, + { + "epoch": 0.01, + "learning_rate": 3.477514850661343e-06, + "loss": 3.2953, + "step": 2841600 + }, + { + "epoch": 0.01, + "learning_rate": 3.4691289031508234e-06, + "loss": 3.3089, + "step": 2842112 + }, + { + "epoch": 0.01, + "learning_rate": 3.4607429556403034e-06, + "loss": 3.303, + "step": 2842624 + }, + { + "epoch": 0.01, + "learning_rate": 3.4523570081297833e-06, + "loss": 3.2992, + "step": 2843136 + }, + { + "epoch": 0.01, + "learning_rate": 3.4439874394229945e-06, + "loss": 3.2958, + "step": 2843648 + }, + { + "epoch": 0.01, + "learning_rate": 3.435601491912475e-06, + "loss": 3.2984, + "step": 2844160 + }, + { + "epoch": 0.01, + "learning_rate": 3.4272155444019544e-06, + "loss": 3.3034, + "step": 2844672 + }, + { + "epoch": 0.01, + "learning_rate": 3.418829596891434e-06, + "loss": 3.3113, + "step": 2845184 + }, + { + "epoch": 0.01, + "learning_rate": 3.4104436493809144e-06, + "loss": 3.3021, + "step": 2845696 + }, + { + "epoch": 0.01, + "learning_rate": 3.402057701870394e-06, + "loss": 3.3019, + "step": 2846208 + }, + { + "epoch": 0.01, + "learning_rate": 3.3936881331636055e-06, + "loss": 3.3099, + "step": 2846720 + }, + { + "epoch": 0.01, + "learning_rate": 3.385302185653085e-06, + "loss": 3.3098, + "step": 2847232 + }, + { + "epoch": 0.01, + "learning_rate": 3.3769162381425654e-06, + "loss": 3.3057, + "step": 2847744 + }, + { + "epoch": 0.01, + "learning_rate": 3.368530290632045e-06, + "loss": 3.2943, + "step": 2848256 + }, + { + "epoch": 0.01, + "learning_rate": 3.3601607219252566e-06, + "loss": 3.2821, + "step": 2848768 + }, + { + "epoch": 0.01, + "learning_rate": 3.3517747744147366e-06, + "loss": 3.3138, + "step": 2849280 + }, + { + "epoch": 0.01, + "learning_rate": 3.343388826904216e-06, + "loss": 3.3117, + "step": 2849792 + }, + { + "epoch": 0.01, + "learning_rate": 3.3350028793936965e-06, + "loss": 3.3033, + "step": 2850304 + }, + { + "epoch": 0.01, + "learning_rate": 3.3266333106869073e-06, + "loss": 3.3002, + "step": 2850816 + }, + { + "epoch": 0.01, + "learning_rate": 3.3182473631763876e-06, + "loss": 3.3, + "step": 2851328 + }, + { + "epoch": 0.01, + "learning_rate": 3.309861415665867e-06, + "loss": 3.2909, + "step": 2851840 + }, + { + "epoch": 0.01, + "learning_rate": 3.3014754681553476e-06, + "loss": 3.2981, + "step": 2852352 + }, + { + "epoch": 0.01, + "learning_rate": 3.293089520644827e-06, + "loss": 3.291, + "step": 2852864 + }, + { + "epoch": 0.01, + "learning_rate": 3.2847199519380387e-06, + "loss": 3.3014, + "step": 2853376 + }, + { + "epoch": 0.01, + "learning_rate": 3.2763340044275183e-06, + "loss": 3.3092, + "step": 2853888 + }, + { + "epoch": 0.01, + "learning_rate": 3.2679480569169982e-06, + "loss": 3.2937, + "step": 2854400 + }, + { + "epoch": 0.01, + "learning_rate": 3.259562109406478e-06, + "loss": 3.2874, + "step": 2854912 + }, + { + "epoch": 0.01, + "learning_rate": 3.251176161895958e-06, + "loss": 3.3045, + "step": 2855424 + }, + { + "epoch": 0.01, + "learning_rate": 3.242790214385438e-06, + "loss": 3.2869, + "step": 2855936 + }, + { + "epoch": 0.01, + "learning_rate": 3.2344206456786493e-06, + "loss": 3.2908, + "step": 2856448 + }, + { + "epoch": 0.01, + "learning_rate": 3.2260346981681297e-06, + "loss": 3.3053, + "step": 2856960 + }, + { + "epoch": 0.01, + "learning_rate": 3.217648750657609e-06, + "loss": 3.3035, + "step": 2857472 + }, + { + "epoch": 0.01, + "learning_rate": 3.2092628031470887e-06, + "loss": 3.2923, + "step": 2857984 + }, + { + "epoch": 0.01, + "learning_rate": 3.200876855636569e-06, + "loss": 3.2925, + "step": 2858496 + }, + { + "epoch": 0.01, + "learning_rate": 3.1924909081260486e-06, + "loss": 3.2948, + "step": 2859008 + }, + { + "epoch": 0.01, + "learning_rate": 3.184104960615529e-06, + "loss": 3.2958, + "step": 2859520 + }, + { + "epoch": 0.01, + "learning_rate": 3.1757190131050086e-06, + "loss": 3.3036, + "step": 2860032 + }, + { + "epoch": 0.01, + "learning_rate": 3.16734944439822e-06, + "loss": 3.2936, + "step": 2860544 + }, + { + "epoch": 0.01, + "learning_rate": 3.1589634968876997e-06, + "loss": 3.3041, + "step": 2861056 + }, + { + "epoch": 0.01, + "learning_rate": 3.1505775493771797e-06, + "loss": 3.3073, + "step": 2861568 + }, + { + "epoch": 0.01, + "learning_rate": 3.14219160186666e-06, + "loss": 3.305, + "step": 2862080 + }, + { + "epoch": 0.01, + "learning_rate": 3.1338056543561396e-06, + "loss": 3.293, + "step": 2862592 + }, + { + "epoch": 0.01, + "learning_rate": 3.1254360856493512e-06, + "loss": 3.308, + "step": 2863104 + }, + { + "epoch": 0.01, + "learning_rate": 3.1170501381388308e-06, + "loss": 3.3055, + "step": 2863616 + }, + { + "epoch": 0.01, + "learning_rate": 3.1086641906283107e-06, + "loss": 3.2916, + "step": 2864128 + }, + { + "epoch": 0.01, + "learning_rate": 3.1002782431177907e-06, + "loss": 3.3086, + "step": 2864640 + }, + { + "epoch": 0.01, + "learning_rate": 3.091908674411002e-06, + "loss": 3.2998, + "step": 2865152 + }, + { + "epoch": 0.01, + "learning_rate": 3.083522726900482e-06, + "loss": 3.3105, + "step": 2865664 + }, + { + "epoch": 0.01, + "learning_rate": 3.075136779389962e-06, + "loss": 3.2956, + "step": 2866176 + }, + { + "epoch": 0.01, + "learning_rate": 3.0667508318794418e-06, + "loss": 3.2828, + "step": 2866688 + }, + { + "epoch": 0.01, + "learning_rate": 3.0583648843689217e-06, + "loss": 3.2948, + "step": 2867200 + }, + { + "epoch": 0.01, + "learning_rate": 3.049995315662133e-06, + "loss": 3.294, + "step": 2867712 + }, + { + "epoch": 0.01, + "learning_rate": 3.041609368151613e-06, + "loss": 3.301, + "step": 2868224 + }, + { + "epoch": 0.01, + "learning_rate": 3.033223420641093e-06, + "loss": 3.2988, + "step": 2868736 + }, + { + "epoch": 0.01, + "learning_rate": 3.024837473130573e-06, + "loss": 3.2953, + "step": 2869248 + }, + { + "epoch": 0.02, + "learning_rate": 3.016467904423784e-06, + "loss": 3.3014, + "step": 2869760 + }, + { + "epoch": 0.02, + "learning_rate": 3.008081956913264e-06, + "loss": 3.2982, + "step": 2870272 + }, + { + "epoch": 0.02, + "learning_rate": 2.9996960094027435e-06, + "loss": 3.2906, + "step": 2870784 + }, + { + "epoch": 0.02, + "learning_rate": 2.9913100618922235e-06, + "loss": 3.297, + "step": 2871296 + }, + { + "epoch": 0.02, + "learning_rate": 2.9829241143817034e-06, + "loss": 3.3047, + "step": 2871808 + }, + { + "epoch": 0.02, + "learning_rate": 2.974554545674915e-06, + "loss": 3.3111, + "step": 2872320 + }, + { + "epoch": 0.02, + "learning_rate": 2.966168598164395e-06, + "loss": 3.2979, + "step": 2872832 + }, + { + "epoch": 0.02, + "learning_rate": 2.957782650653875e-06, + "loss": 3.283, + "step": 2873344 + }, + { + "epoch": 0.02, + "learning_rate": 2.949396703143355e-06, + "loss": 3.302, + "step": 2873856 + }, + { + "epoch": 0.02, + "learning_rate": 2.941027134436566e-06, + "loss": 3.2959, + "step": 2874368 + }, + { + "epoch": 0.02, + "learning_rate": 2.932641186926046e-06, + "loss": 3.3032, + "step": 2874880 + }, + { + "epoch": 0.02, + "learning_rate": 2.9242552394155256e-06, + "loss": 3.3058, + "step": 2875392 + }, + { + "epoch": 0.02, + "learning_rate": 2.9158692919050056e-06, + "loss": 3.293, + "step": 2875904 + }, + { + "epoch": 0.02, + "learning_rate": 2.9074997231982168e-06, + "loss": 3.2933, + "step": 2876416 + }, + { + "epoch": 0.02, + "learning_rate": 2.8991137756876967e-06, + "loss": 3.3041, + "step": 2876928 + }, + { + "epoch": 0.02, + "learning_rate": 2.8907278281771767e-06, + "loss": 3.2879, + "step": 2877440 + }, + { + "epoch": 0.02, + "learning_rate": 2.8823418806666567e-06, + "loss": 3.2996, + "step": 2877952 + }, + { + "epoch": 0.02, + "learning_rate": 2.8739723119598683e-06, + "loss": 3.287, + "step": 2878464 + }, + { + "epoch": 0.02, + "learning_rate": 2.8655863644493482e-06, + "loss": 3.302, + "step": 2878976 + }, + { + "epoch": 0.02, + "learning_rate": 2.857200416938828e-06, + "loss": 3.2874, + "step": 2879488 + }, + { + "epoch": 0.02, + "learning_rate": 2.8488144694283077e-06, + "loss": 3.3045, + "step": 2880000 + }, + { + "epoch": 0.02, + "learning_rate": 2.8404449007215194e-06, + "loss": 3.295, + "step": 2880512 + }, + { + "epoch": 0.02, + "learning_rate": 2.832058953210999e-06, + "loss": 3.2925, + "step": 2881024 + }, + { + "epoch": 0.02, + "learning_rate": 2.823673005700479e-06, + "loss": 3.3038, + "step": 2881536 + }, + { + "epoch": 0.02, + "learning_rate": 2.815287058189959e-06, + "loss": 3.2994, + "step": 2882048 + }, + { + "epoch": 0.02, + "learning_rate": 2.80691748948317e-06, + "loss": 3.3103, + "step": 2882560 + }, + { + "epoch": 0.02, + "learning_rate": 2.79853154197265e-06, + "loss": 3.308, + "step": 2883072 + }, + { + "epoch": 0.02, + "learning_rate": 2.79014559446213e-06, + "loss": 3.2942, + "step": 2883584 + }, + { + "epoch": 0.02, + "learning_rate": 2.78175964695161e-06, + "loss": 3.3029, + "step": 2884096 + }, + { + "epoch": 0.02, + "learning_rate": 2.77337369944109e-06, + "loss": 3.2945, + "step": 2884608 + }, + { + "epoch": 0.02, + "learning_rate": 2.7650041307343015e-06, + "loss": 3.299, + "step": 2885120 + }, + { + "epoch": 0.02, + "learning_rate": 2.756618183223781e-06, + "loss": 3.2967, + "step": 2885632 + }, + { + "epoch": 0.02, + "learning_rate": 2.748232235713261e-06, + "loss": 3.3015, + "step": 2886144 + }, + { + "epoch": 0.02, + "learning_rate": 2.739846288202741e-06, + "loss": 3.2989, + "step": 2886656 + }, + { + "epoch": 0.02, + "learning_rate": 2.731460340692221e-06, + "loss": 3.2945, + "step": 2887168 + }, + { + "epoch": 0.02, + "learning_rate": 2.723074393181701e-06, + "loss": 3.2981, + "step": 2887680 + }, + { + "epoch": 0.02, + "learning_rate": 2.7146884456711804e-06, + "loss": 3.3041, + "step": 2888192 + }, + { + "epoch": 0.02, + "learning_rate": 2.706318876964392e-06, + "loss": 3.2946, + "step": 2888704 + }, + { + "epoch": 0.02, + "learning_rate": 2.6979329294538715e-06, + "loss": 3.3028, + "step": 2889216 + }, + { + "epoch": 0.02, + "learning_rate": 2.6895469819433515e-06, + "loss": 3.2908, + "step": 2889728 + }, + { + "epoch": 0.02, + "learning_rate": 2.681161034432832e-06, + "loss": 3.3123, + "step": 2890240 + }, + { + "epoch": 0.02, + "learning_rate": 2.672775086922312e-06, + "loss": 3.2955, + "step": 2890752 + }, + { + "epoch": 0.02, + "learning_rate": 2.6643891394117914e-06, + "loss": 3.3071, + "step": 2891264 + }, + { + "epoch": 0.02, + "learning_rate": 2.656019570705003e-06, + "loss": 3.2954, + "step": 2891776 + }, + { + "epoch": 0.02, + "learning_rate": 2.6476336231944825e-06, + "loss": 3.2985, + "step": 2892288 + }, + { + "epoch": 0.02, + "learning_rate": 2.6392476756839625e-06, + "loss": 3.3038, + "step": 2892800 + }, + { + "epoch": 0.02, + "learning_rate": 2.6308617281734424e-06, + "loss": 3.2978, + "step": 2893312 + }, + { + "epoch": 0.02, + "learning_rate": 2.6224921594666537e-06, + "loss": 3.2923, + "step": 2893824 + }, + { + "epoch": 0.02, + "learning_rate": 2.6141062119561336e-06, + "loss": 3.3057, + "step": 2894336 + }, + { + "epoch": 0.02, + "learning_rate": 2.6057202644456136e-06, + "loss": 3.3022, + "step": 2894848 + }, + { + "epoch": 0.02, + "learning_rate": 2.5973343169350935e-06, + "loss": 3.3123, + "step": 2895360 + }, + { + "epoch": 0.02, + "learning_rate": 2.5889483694245735e-06, + "loss": 3.2968, + "step": 2895872 + }, + { + "epoch": 0.02, + "learning_rate": 2.5805624219140534e-06, + "loss": 3.3059, + "step": 2896384 + }, + { + "epoch": 0.02, + "learning_rate": 2.5721928532072647e-06, + "loss": 3.3037, + "step": 2896896 + }, + { + "epoch": 0.02, + "learning_rate": 2.5638069056967446e-06, + "loss": 3.2973, + "step": 2897408 + }, + { + "epoch": 0.02, + "learning_rate": 2.5554209581862246e-06, + "loss": 3.2997, + "step": 2897920 + }, + { + "epoch": 0.02, + "learning_rate": 2.5470350106757045e-06, + "loss": 3.3026, + "step": 2898432 + }, + { + "epoch": 0.02, + "learning_rate": 2.5386654419689157e-06, + "loss": 3.3087, + "step": 2898944 + }, + { + "epoch": 0.02, + "learning_rate": 2.5302794944583957e-06, + "loss": 3.2789, + "step": 2899456 + }, + { + "epoch": 0.02, + "learning_rate": 2.5218935469478756e-06, + "loss": 3.3021, + "step": 2899968 + }, + { + "epoch": 0.03, + "eval_loss": 3.869551420211792, + "eval_runtime": 302.8824, + "eval_samples_per_second": 1259.865, + "eval_steps_per_second": 39.372, + "step": 2900160 + }, + { + "epoch": 1.0, + "learning_rate": 2.513507599437355e-06, + "loss": 3.2965, + "step": 2900480 + }, + { + "epoch": 1.0, + "learning_rate": 2.505121651926835e-06, + "loss": 3.2931, + "step": 2900992 + }, + { + "epoch": 1.0, + "learning_rate": 2.4967520832200468e-06, + "loss": 3.3041, + "step": 2901504 + }, + { + "epoch": 1.0, + "learning_rate": 2.4883661357095267e-06, + "loss": 3.3135, + "step": 2902016 + }, + { + "epoch": 1.0, + "learning_rate": 2.4799801881990067e-06, + "loss": 3.3065, + "step": 2902528 + }, + { + "epoch": 1.0, + "learning_rate": 2.4715942406884866e-06, + "loss": 3.2952, + "step": 2903040 + }, + { + "epoch": 1.0, + "learning_rate": 2.4632082931779666e-06, + "loss": 3.2991, + "step": 2903552 + }, + { + "epoch": 1.0, + "learning_rate": 2.454822345667446e-06, + "loss": 3.2884, + "step": 2904064 + }, + { + "epoch": 1.0, + "learning_rate": 2.4464527769606578e-06, + "loss": 3.3023, + "step": 2904576 + }, + { + "epoch": 1.0, + "learning_rate": 2.4380668294501373e-06, + "loss": 3.3022, + "step": 2905088 + }, + { + "epoch": 1.0, + "learning_rate": 2.4296808819396173e-06, + "loss": 3.3001, + "step": 2905600 + }, + { + "epoch": 1.0, + "learning_rate": 2.421294934429097e-06, + "loss": 3.3002, + "step": 2906112 + }, + { + "epoch": 1.0, + "learning_rate": 2.412908986918577e-06, + "loss": 3.2965, + "step": 2906624 + }, + { + "epoch": 1.0, + "learning_rate": 2.4045394182117884e-06, + "loss": 3.2931, + "step": 2907136 + }, + { + "epoch": 1.0, + "learning_rate": 2.3961534707012683e-06, + "loss": 3.2878, + "step": 2907648 + }, + { + "epoch": 1.0, + "learning_rate": 2.3877675231907483e-06, + "loss": 3.2884, + "step": 2908160 + }, + { + "epoch": 1.0, + "learning_rate": 2.3793815756802282e-06, + "loss": 3.2939, + "step": 2908672 + }, + { + "epoch": 1.0, + "learning_rate": 2.370995628169708e-06, + "loss": 3.2928, + "step": 2909184 + }, + { + "epoch": 1.0, + "learning_rate": 2.362609680659188e-06, + "loss": 3.2965, + "step": 2909696 + }, + { + "epoch": 1.0, + "learning_rate": 2.3542401119523994e-06, + "loss": 3.3119, + "step": 2910208 + }, + { + "epoch": 1.0, + "learning_rate": 2.3458541644418793e-06, + "loss": 3.309, + "step": 2910720 + }, + { + "epoch": 1.0, + "learning_rate": 2.3374682169313593e-06, + "loss": 3.3003, + "step": 2911232 + }, + { + "epoch": 1.0, + "learning_rate": 2.329082269420839e-06, + "loss": 3.2968, + "step": 2911744 + }, + { + "epoch": 1.0, + "learning_rate": 2.3206963219103188e-06, + "loss": 3.3033, + "step": 2912256 + }, + { + "epoch": 1.0, + "learning_rate": 2.31232675320353e-06, + "loss": 3.2943, + "step": 2912768 + }, + { + "epoch": 1.0, + "learning_rate": 2.30394080569301e-06, + "loss": 3.3072, + "step": 2913280 + }, + { + "epoch": 1.0, + "learning_rate": 2.2955548581824903e-06, + "loss": 3.296, + "step": 2913792 + }, + { + "epoch": 1.0, + "learning_rate": 2.2871689106719703e-06, + "loss": 3.2848, + "step": 2914304 + }, + { + "epoch": 1.0, + "learning_rate": 2.2787829631614502e-06, + "loss": 3.2931, + "step": 2914816 + }, + { + "epoch": 1.0, + "learning_rate": 2.2704133944546614e-06, + "loss": 3.2914, + "step": 2915328 + }, + { + "epoch": 1.01, + "learning_rate": 2.2620274469441414e-06, + "loss": 3.2956, + "step": 2915840 + }, + { + "epoch": 1.01, + "learning_rate": 2.253641499433621e-06, + "loss": 3.3037, + "step": 2916352 + }, + { + "epoch": 1.01, + "learning_rate": 2.245255551923101e-06, + "loss": 3.3, + "step": 2916864 + }, + { + "epoch": 1.01, + "learning_rate": 2.236885983216312e-06, + "loss": 3.3021, + "step": 2917376 + }, + { + "epoch": 1.01, + "learning_rate": 2.228500035705792e-06, + "loss": 3.2861, + "step": 2917888 + }, + { + "epoch": 1.01, + "learning_rate": 2.220114088195272e-06, + "loss": 3.3052, + "step": 2918400 + }, + { + "epoch": 1.01, + "learning_rate": 2.211728140684752e-06, + "loss": 3.2953, + "step": 2918912 + }, + { + "epoch": 1.01, + "learning_rate": 2.203342193174232e-06, + "loss": 3.288, + "step": 2919424 + }, + { + "epoch": 1.01, + "learning_rate": 2.1949726244674436e-06, + "loss": 3.294, + "step": 2919936 + }, + { + "epoch": 1.01, + "learning_rate": 2.1865866769569235e-06, + "loss": 3.2851, + "step": 2920448 + }, + { + "epoch": 1.01, + "learning_rate": 2.178200729446403e-06, + "loss": 3.2977, + "step": 2920960 + }, + { + "epoch": 1.01, + "learning_rate": 2.169814781935883e-06, + "loss": 3.3051, + "step": 2921472 + }, + { + "epoch": 1.01, + "learning_rate": 2.161428834425363e-06, + "loss": 3.2929, + "step": 2921984 + }, + { + "epoch": 1.01, + "learning_rate": 2.153042886914843e-06, + "loss": 3.2925, + "step": 2922496 + }, + { + "epoch": 1.01, + "learning_rate": 2.144673318208054e-06, + "loss": 3.3027, + "step": 2923008 + }, + { + "epoch": 1.01, + "learning_rate": 2.136287370697534e-06, + "loss": 3.3033, + "step": 2923520 + }, + { + "epoch": 1.01, + "learning_rate": 2.127901423187014e-06, + "loss": 3.2995, + "step": 2924032 + }, + { + "epoch": 1.01, + "learning_rate": 2.1195154756764936e-06, + "loss": 3.2901, + "step": 2924544 + }, + { + "epoch": 1.01, + "learning_rate": 2.111145906969705e-06, + "loss": 3.2657, + "step": 2925056 + }, + { + "epoch": 1.01, + "learning_rate": 2.102759959459185e-06, + "loss": 3.3105, + "step": 2925568 + }, + { + "epoch": 1.01, + "learning_rate": 2.094374011948665e-06, + "loss": 3.3015, + "step": 2926080 + }, + { + "epoch": 1.01, + "learning_rate": 2.085988064438145e-06, + "loss": 3.2975, + "step": 2926592 + }, + { + "epoch": 1.01, + "learning_rate": 2.0776184957313563e-06, + "loss": 3.2886, + "step": 2927104 + }, + { + "epoch": 1.01, + "learning_rate": 2.0692325482208362e-06, + "loss": 3.2905, + "step": 2927616 + }, + { + "epoch": 1.01, + "learning_rate": 2.060846600710316e-06, + "loss": 3.283, + "step": 2928128 + }, + { + "epoch": 1.01, + "learning_rate": 2.0524606531997957e-06, + "loss": 3.287, + "step": 2928640 + }, + { + "epoch": 1.01, + "learning_rate": 2.0440747056892757e-06, + "loss": 3.2867, + "step": 2929152 + }, + { + "epoch": 1.01, + "learning_rate": 2.0356887581787556e-06, + "loss": 3.2935, + "step": 2929664 + }, + { + "epoch": 1.01, + "learning_rate": 2.027319189471967e-06, + "loss": 3.3047, + "step": 2930176 + }, + { + "epoch": 1.01, + "learning_rate": 2.018933241961447e-06, + "loss": 3.2831, + "step": 2930688 + }, + { + "epoch": 1.01, + "learning_rate": 2.0105472944509268e-06, + "loss": 3.2835, + "step": 2931200 + }, + { + "epoch": 1.01, + "learning_rate": 2.0021613469404067e-06, + "loss": 3.2965, + "step": 2931712 + }, + { + "epoch": 1.01, + "learning_rate": 1.9937917782336184e-06, + "loss": 3.2837, + "step": 2932224 + }, + { + "epoch": 1.01, + "learning_rate": 1.9854058307230983e-06, + "loss": 3.2823, + "step": 2932736 + }, + { + "epoch": 1.01, + "learning_rate": 1.977019883212578e-06, + "loss": 3.2958, + "step": 2933248 + }, + { + "epoch": 1.01, + "learning_rate": 1.968633935702058e-06, + "loss": 3.2933, + "step": 2933760 + }, + { + "epoch": 1.01, + "learning_rate": 1.9602479881915378e-06, + "loss": 3.2843, + "step": 2934272 + }, + { + "epoch": 1.01, + "learning_rate": 1.951878419484749e-06, + "loss": 3.285, + "step": 2934784 + }, + { + "epoch": 1.01, + "learning_rate": 1.943492471974229e-06, + "loss": 3.286, + "step": 2935296 + }, + { + "epoch": 1.01, + "learning_rate": 1.935106524463709e-06, + "loss": 3.2909, + "step": 2935808 + }, + { + "epoch": 1.01, + "learning_rate": 1.926720576953189e-06, + "loss": 3.2947, + "step": 2936320 + }, + { + "epoch": 1.01, + "learning_rate": 1.9183510082464e-06, + "loss": 3.2896, + "step": 2936832 + }, + { + "epoch": 1.01, + "learning_rate": 1.9099650607358804e-06, + "loss": 3.2944, + "step": 2937344 + }, + { + "epoch": 1.01, + "learning_rate": 1.9015791132253598e-06, + "loss": 3.2994, + "step": 2937856 + }, + { + "epoch": 1.01, + "learning_rate": 1.8931931657148397e-06, + "loss": 3.2989, + "step": 2938368 + }, + { + "epoch": 1.01, + "learning_rate": 1.8848072182043199e-06, + "loss": 3.2877, + "step": 2938880 + }, + { + "epoch": 1.01, + "learning_rate": 1.8764212706937998e-06, + "loss": 3.2974, + "step": 2939392 + }, + { + "epoch": 1.01, + "learning_rate": 1.8680353231832798e-06, + "loss": 3.2991, + "step": 2939904 + }, + { + "epoch": 1.01, + "learning_rate": 1.859665754476491e-06, + "loss": 3.2811, + "step": 2940416 + }, + { + "epoch": 1.01, + "learning_rate": 1.851279806965971e-06, + "loss": 3.2995, + "step": 2940928 + }, + { + "epoch": 1.01, + "learning_rate": 1.8428938594554507e-06, + "loss": 3.2941, + "step": 2941440 + }, + { + "epoch": 1.01, + "learning_rate": 1.8345079119449307e-06, + "loss": 3.3024, + "step": 2941952 + }, + { + "epoch": 1.01, + "learning_rate": 1.8261219644344106e-06, + "loss": 3.2905, + "step": 2942464 + }, + { + "epoch": 1.01, + "learning_rate": 1.8177523957276218e-06, + "loss": 3.2743, + "step": 2942976 + }, + { + "epoch": 1.01, + "learning_rate": 1.8093664482171018e-06, + "loss": 3.2846, + "step": 2943488 + }, + { + "epoch": 1.01, + "learning_rate": 1.8009805007065817e-06, + "loss": 3.2864, + "step": 2944000 + }, + { + "epoch": 1.01, + "learning_rate": 1.7925945531960617e-06, + "loss": 3.2965, + "step": 2944512 + }, + { + "epoch": 1.01, + "learning_rate": 1.7842249844892731e-06, + "loss": 3.2892, + "step": 2945024 + }, + { + "epoch": 1.01, + "learning_rate": 1.7758390369787527e-06, + "loss": 3.289, + "step": 2945536 + }, + { + "epoch": 1.02, + "learning_rate": 1.7674530894682326e-06, + "loss": 3.2917, + "step": 2946048 + }, + { + "epoch": 1.02, + "learning_rate": 1.7590671419577126e-06, + "loss": 3.2968, + "step": 2946560 + }, + { + "epoch": 1.02, + "learning_rate": 1.7506811944471925e-06, + "loss": 3.2787, + "step": 2947072 + }, + { + "epoch": 1.02, + "learning_rate": 1.742311625740404e-06, + "loss": 3.2895, + "step": 2947584 + }, + { + "epoch": 1.02, + "learning_rate": 1.733925678229884e-06, + "loss": 3.2982, + "step": 2948096 + }, + { + "epoch": 1.02, + "learning_rate": 1.7255397307193639e-06, + "loss": 3.3029, + "step": 2948608 + }, + { + "epoch": 1.02, + "learning_rate": 1.7171537832088434e-06, + "loss": 3.2923, + "step": 2949120 + }, + { + "epoch": 1.02, + "learning_rate": 1.7087678356983233e-06, + "loss": 3.2772, + "step": 2949632 + }, + { + "epoch": 1.02, + "learning_rate": 1.7003982669915348e-06, + "loss": 3.2937, + "step": 2950144 + }, + { + "epoch": 1.02, + "learning_rate": 1.6920123194810147e-06, + "loss": 3.2862, + "step": 2950656 + }, + { + "epoch": 1.02, + "learning_rate": 1.6836263719704947e-06, + "loss": 3.297, + "step": 2951168 + }, + { + "epoch": 1.02, + "learning_rate": 1.6752404244599746e-06, + "loss": 3.2991, + "step": 2951680 + }, + { + "epoch": 1.02, + "learning_rate": 1.6668544769494546e-06, + "loss": 3.2889, + "step": 2952192 + }, + { + "epoch": 1.02, + "learning_rate": 1.6584849082426658e-06, + "loss": 3.2852, + "step": 2952704 + }, + { + "epoch": 1.02, + "learning_rate": 1.6500989607321458e-06, + "loss": 3.2936, + "step": 2953216 + }, + { + "epoch": 1.02, + "learning_rate": 1.6417130132216255e-06, + "loss": 3.2814, + "step": 2953728 + }, + { + "epoch": 1.02, + "learning_rate": 1.6333270657111055e-06, + "loss": 3.2938, + "step": 2954240 + }, + { + "epoch": 1.02, + "learning_rate": 1.6249574970043167e-06, + "loss": 3.2777, + "step": 2954752 + }, + { + "epoch": 1.02, + "learning_rate": 1.6165715494937966e-06, + "loss": 3.2957, + "step": 2955264 + }, + { + "epoch": 1.02, + "learning_rate": 1.6081856019832766e-06, + "loss": 3.2818, + "step": 2955776 + }, + { + "epoch": 1.02, + "learning_rate": 1.5997996544727565e-06, + "loss": 3.2918, + "step": 2956288 + }, + { + "epoch": 1.02, + "learning_rate": 1.5914137069622365e-06, + "loss": 3.2872, + "step": 2956800 + }, + { + "epoch": 1.02, + "learning_rate": 1.583044138255448e-06, + "loss": 3.2869, + "step": 2957312 + }, + { + "epoch": 1.02, + "learning_rate": 1.5746581907449279e-06, + "loss": 3.2935, + "step": 2957824 + }, + { + "epoch": 1.02, + "learning_rate": 1.5662722432344074e-06, + "loss": 3.2898, + "step": 2958336 + }, + { + "epoch": 1.02, + "learning_rate": 1.5578862957238874e-06, + "loss": 3.3045, + "step": 2958848 + }, + { + "epoch": 1.02, + "learning_rate": 1.549516727017099e-06, + "loss": 3.2991, + "step": 2959360 + }, + { + "epoch": 1.02, + "learning_rate": 1.5411307795065788e-06, + "loss": 3.2892, + "step": 2959872 + }, + { + "epoch": 1.02, + "learning_rate": 1.5327448319960587e-06, + "loss": 3.2935, + "step": 2960384 + }, + { + "epoch": 1.02, + "learning_rate": 1.5243588844855385e-06, + "loss": 3.2855, + "step": 2960896 + }, + { + "epoch": 1.02, + "learning_rate": 1.5159729369750184e-06, + "loss": 3.29, + "step": 2961408 + }, + { + "epoch": 1.02, + "learning_rate": 1.5076033682682298e-06, + "loss": 3.2898, + "step": 2961920 + }, + { + "epoch": 1.02, + "learning_rate": 1.4992174207577098e-06, + "loss": 3.2909, + "step": 2962432 + }, + { + "epoch": 1.02, + "learning_rate": 1.4908314732471897e-06, + "loss": 3.2924, + "step": 2962944 + }, + { + "epoch": 1.02, + "learning_rate": 1.4824455257366695e-06, + "loss": 3.2901, + "step": 2963456 + }, + { + "epoch": 1.02, + "learning_rate": 1.4740595782261494e-06, + "loss": 3.2912, + "step": 2963968 + }, + { + "epoch": 1.02, + "learning_rate": 1.4656736307156292e-06, + "loss": 3.2942, + "step": 2964480 + }, + { + "epoch": 1.02, + "learning_rate": 1.4573040620088408e-06, + "loss": 3.2919, + "step": 2964992 + }, + { + "epoch": 1.02, + "learning_rate": 1.4489181144983206e-06, + "loss": 3.2906, + "step": 2965504 + }, + { + "epoch": 1.02, + "learning_rate": 1.4405321669878005e-06, + "loss": 3.2831, + "step": 2966016 + }, + { + "epoch": 1.02, + "learning_rate": 1.4321462194772805e-06, + "loss": 3.3035, + "step": 2966528 + }, + { + "epoch": 1.02, + "learning_rate": 1.4237602719667602e-06, + "loss": 3.2896, + "step": 2967040 + }, + { + "epoch": 1.02, + "learning_rate": 1.4153743244562402e-06, + "loss": 3.3024, + "step": 2967552 + }, + { + "epoch": 1.02, + "learning_rate": 1.4070047557494516e-06, + "loss": 3.2844, + "step": 2968064 + }, + { + "epoch": 1.02, + "learning_rate": 1.3986188082389316e-06, + "loss": 3.2923, + "step": 2968576 + }, + { + "epoch": 1.02, + "learning_rate": 1.3902328607284113e-06, + "loss": 3.2929, + "step": 2969088 + }, + { + "epoch": 1.02, + "learning_rate": 1.3818469132178913e-06, + "loss": 3.2939, + "step": 2969600 + }, + { + "epoch": 1.02, + "learning_rate": 1.3734773445111025e-06, + "loss": 3.2809, + "step": 2970112 + }, + { + "epoch": 1.02, + "learning_rate": 1.3650913970005824e-06, + "loss": 3.3, + "step": 2970624 + }, + { + "epoch": 1.02, + "learning_rate": 1.3567054494900624e-06, + "loss": 3.2924, + "step": 2971136 + }, + { + "epoch": 1.02, + "learning_rate": 1.3483195019795423e-06, + "loss": 3.3074, + "step": 2971648 + }, + { + "epoch": 1.02, + "learning_rate": 1.3399335544690223e-06, + "loss": 3.2866, + "step": 2972160 + }, + { + "epoch": 1.02, + "learning_rate": 1.331547606958502e-06, + "loss": 3.3014, + "step": 2972672 + }, + { + "epoch": 1.02, + "learning_rate": 1.3231780382517135e-06, + "loss": 3.2962, + "step": 2973184 + }, + { + "epoch": 1.02, + "learning_rate": 1.3147920907411932e-06, + "loss": 3.291, + "step": 2973696 + }, + { + "epoch": 1.02, + "learning_rate": 1.3064061432306734e-06, + "loss": 3.2908, + "step": 2974208 + }, + { + "epoch": 1.02, + "learning_rate": 1.2980201957201531e-06, + "loss": 3.2961, + "step": 2974720 + }, + { + "epoch": 1.02, + "learning_rate": 1.2896506270133645e-06, + "loss": 3.2998, + "step": 2975232 + }, + { + "epoch": 1.02, + "learning_rate": 1.2812646795028443e-06, + "loss": 3.2766, + "step": 2975744 + }, + { + "epoch": 1.02, + "learning_rate": 1.2728787319923242e-06, + "loss": 3.2969, + "step": 2976256 + }, + { + "epoch": 1.03, + "eval_loss": 3.868901014328003, + "eval_runtime": 304.3176, + "eval_samples_per_second": 1253.923, + "eval_steps_per_second": 39.186, + "step": 2976480 + } + ], + "logging_steps": 512, + "max_steps": 3052726, + "num_train_epochs": 9223372036854775807, + "save_steps": 10, + "total_flos": 2.071853443103109e+18, + "trial_name": null, + "trial_params": null +}