| { | |
| "best_metric": 0.8770001309177381, | |
| "best_model_checkpoint": "sumcse/5401.reimple-robert-large", | |
| "epoch": 3.0, | |
| "global_step": 6462, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06, | |
| "eval_avg_sts": 0.8284010538709128, | |
| "eval_sickr_spearman": 0.8087322680313793, | |
| "eval_stsb_spearman": 0.8480698397104464, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "eval_avg_sts": 0.8484297373111971, | |
| "eval_sickr_spearman": 0.8391122278775942, | |
| "eval_stsb_spearman": 0.8577472467448001, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "eval_avg_sts": 0.8522751763001004, | |
| "eval_sickr_spearman": 0.8446321541792678, | |
| "eval_stsb_spearman": 0.859918198420933, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.226245744351595e-06, | |
| "loss": 0.9779, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "eval_avg_sts": 0.8529384795629767, | |
| "eval_sickr_spearman": 0.8425804096191359, | |
| "eval_stsb_spearman": 0.8632965495068176, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "eval_avg_sts": 0.854393833444526, | |
| "eval_sickr_spearman": 0.8428722668204894, | |
| "eval_stsb_spearman": 0.8659154000685627, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "eval_avg_sts": 0.8530765893420603, | |
| "eval_sickr_spearman": 0.8375766255335162, | |
| "eval_stsb_spearman": 0.8685765531506044, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "eval_avg_sts": 0.8598428877676623, | |
| "eval_sickr_spearman": 0.8496016440584608, | |
| "eval_stsb_spearman": 0.8700841314768637, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.452491488703189e-06, | |
| "loss": 0.5021, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "eval_avg_sts": 0.8553352649068404, | |
| "eval_sickr_spearman": 0.8414943783845188, | |
| "eval_stsb_spearman": 0.8691761514291622, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "eval_avg_sts": 0.8611558909552111, | |
| "eval_sickr_spearman": 0.8500423294141554, | |
| "eval_stsb_spearman": 0.8722694524962668, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "eval_avg_sts": 0.8470077761608523, | |
| "eval_sickr_spearman": 0.824938778185323, | |
| "eval_stsb_spearman": 0.8690767741363816, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "eval_avg_sts": 0.8523760853013311, | |
| "eval_sickr_spearman": 0.8380687041677498, | |
| "eval_stsb_spearman": 0.8666834664349123, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.678737233054782e-06, | |
| "loss": 0.4518, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "eval_avg_sts": 0.8575383621060841, | |
| "eval_sickr_spearman": 0.8451627057257532, | |
| "eval_stsb_spearman": 0.869914018486415, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "eval_avg_sts": 0.8567243246034306, | |
| "eval_sickr_spearman": 0.8404439381960673, | |
| "eval_stsb_spearman": 0.8730047110107938, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "eval_avg_sts": 0.855931320312622, | |
| "eval_sickr_spearman": 0.8433319042322368, | |
| "eval_stsb_spearman": 0.8685307363930072, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "eval_avg_sts": 0.8568209870234867, | |
| "eval_sickr_spearman": 0.8408516742161756, | |
| "eval_stsb_spearman": 0.8727902998307979, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.904982977406376e-06, | |
| "loss": 0.4282, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "eval_avg_sts": 0.8540671502188022, | |
| "eval_sickr_spearman": 0.8388130421467362, | |
| "eval_stsb_spearman": 0.869321258290868, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "eval_avg_sts": 0.8558945239823259, | |
| "eval_sickr_spearman": 0.8423186881473997, | |
| "eval_stsb_spearman": 0.869470359817252, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "eval_avg_sts": 0.8555080041232501, | |
| "eval_sickr_spearman": 0.8408939896165424, | |
| "eval_stsb_spearman": 0.870122018629958, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "eval_avg_sts": 0.8568942691339498, | |
| "eval_sickr_spearman": 0.8418429400876551, | |
| "eval_stsb_spearman": 0.8719455981802446, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.13122872175797e-06, | |
| "loss": 0.3707, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "eval_avg_sts": 0.8557911670636686, | |
| "eval_sickr_spearman": 0.8392199616381198, | |
| "eval_stsb_spearman": 0.8723623724892173, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "eval_avg_sts": 0.85588148387946, | |
| "eval_sickr_spearman": 0.83787566717107, | |
| "eval_stsb_spearman": 0.8738873005878499, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "eval_avg_sts": 0.8622210733781358, | |
| "eval_sickr_spearman": 0.8489977970511808, | |
| "eval_stsb_spearman": 0.8754443497050908, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "eval_avg_sts": 0.8624794115602034, | |
| "eval_sickr_spearman": 0.8479586922026686, | |
| "eval_stsb_spearman": 0.8770001309177381, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 5.357474466109565e-06, | |
| "loss": 0.3458, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "eval_avg_sts": 0.8548885349566437, | |
| "eval_sickr_spearman": 0.8383297051729646, | |
| "eval_stsb_spearman": 0.871447364740323, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "eval_avg_sts": 0.8562784353766513, | |
| "eval_sickr_spearman": 0.841656867600684, | |
| "eval_stsb_spearman": 0.8709000031526186, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "eval_avg_sts": 0.8578722811343277, | |
| "eval_sickr_spearman": 0.8429833425291002, | |
| "eval_stsb_spearman": 0.8727612197395551, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "eval_avg_sts": 0.8600916280834878, | |
| "eval_sickr_spearman": 0.8445091945595862, | |
| "eval_stsb_spearman": 0.8756740616073895, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.583720210461158e-06, | |
| "loss": 0.3435, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "eval_avg_sts": 0.8593220407937633, | |
| "eval_sickr_spearman": 0.8451772111183875, | |
| "eval_stsb_spearman": 0.8734668704691391, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "eval_avg_sts": 0.861320493545486, | |
| "eval_sickr_spearman": 0.8473822229232085, | |
| "eval_stsb_spearman": 0.8752587641677635, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "eval_avg_sts": 0.860298885852978, | |
| "eval_sickr_spearman": 0.8464292858706762, | |
| "eval_stsb_spearman": 0.8741684858352797, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "eval_avg_sts": 0.8550809080819418, | |
| "eval_sickr_spearman": 0.8354855435013065, | |
| "eval_stsb_spearman": 0.8746762726625771, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 3.8099659548127517e-06, | |
| "loss": 0.3442, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "eval_avg_sts": 0.8559180414679683, | |
| "eval_sickr_spearman": 0.8384507435485884, | |
| "eval_stsb_spearman": 0.8733853393873484, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "eval_avg_sts": 0.8571707122552362, | |
| "eval_sickr_spearman": 0.8428135045544151, | |
| "eval_stsb_spearman": 0.8715279199560574, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "eval_avg_sts": 0.8560277042936426, | |
| "eval_sickr_spearman": 0.8397806286852069, | |
| "eval_stsb_spearman": 0.8722747799020784, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "eval_avg_sts": 0.8578281901982555, | |
| "eval_sickr_spearman": 0.8433445364119149, | |
| "eval_stsb_spearman": 0.8723118439845962, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 3.036211699164346e-06, | |
| "loss": 0.3131, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "eval_avg_sts": 0.8546342996226579, | |
| "eval_sickr_spearman": 0.8383026636628548, | |
| "eval_stsb_spearman": 0.8709659355824609, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "eval_avg_sts": 0.8559652822309323, | |
| "eval_sickr_spearman": 0.840341391794497, | |
| "eval_stsb_spearman": 0.8715891726673677, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "eval_avg_sts": 0.8546617687967435, | |
| "eval_sickr_spearman": 0.8391962823051108, | |
| "eval_stsb_spearman": 0.8701272552883762, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "eval_avg_sts": 0.858705271470795, | |
| "eval_sickr_spearman": 0.8465492195309672, | |
| "eval_stsb_spearman": 0.8708613234106228, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 2.2624574435159395e-06, | |
| "loss": 0.2846, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "eval_avg_sts": 0.8597192689340263, | |
| "eval_sickr_spearman": 0.8464290457151691, | |
| "eval_stsb_spearman": 0.8730094921528834, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "eval_avg_sts": 0.8606101916960156, | |
| "eval_sickr_spearman": 0.8477548001770637, | |
| "eval_stsb_spearman": 0.8734655832149676, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "eval_avg_sts": 0.859970718351403, | |
| "eval_sickr_spearman": 0.8465689603136584, | |
| "eval_stsb_spearman": 0.8733724763891477, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "eval_avg_sts": 0.8611164416935826, | |
| "eval_sickr_spearman": 0.8490797381102341, | |
| "eval_stsb_spearman": 0.873153145276931, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 1.4887031878675335e-06, | |
| "loss": 0.2804, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "eval_avg_sts": 0.860553080718742, | |
| "eval_sickr_spearman": 0.8488836271230623, | |
| "eval_stsb_spearman": 0.8722225343144216, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "eval_avg_sts": 0.8602837879281291, | |
| "eval_sickr_spearman": 0.8481850627837464, | |
| "eval_stsb_spearman": 0.8723825130725117, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "eval_avg_sts": 0.857967378909783, | |
| "eval_sickr_spearman": 0.8440642824669651, | |
| "eval_stsb_spearman": 0.8718704753526009, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "eval_avg_sts": 0.8601815553282466, | |
| "eval_sickr_spearman": 0.8467775593872041, | |
| "eval_stsb_spearman": 0.873585551269289, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 7.149489322191272e-07, | |
| "loss": 0.2824, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "eval_avg_sts": 0.860910326239195, | |
| "eval_sickr_spearman": 0.8483497614305777, | |
| "eval_stsb_spearman": 0.8734708910478122, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "eval_avg_sts": 0.860010166556528, | |
| "eval_sickr_spearman": 0.8470775616467865, | |
| "eval_stsb_spearman": 0.8729427714662694, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "eval_avg_sts": 0.8590174618570985, | |
| "eval_sickr_spearman": 0.8456035832058536, | |
| "eval_stsb_spearman": 0.8724313405083433, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "eval_avg_sts": 0.8593514591951028, | |
| "eval_sickr_spearman": 0.8458110775640663, | |
| "eval_stsb_spearman": 0.8728918408261394, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 6462, | |
| "train_runtime": 8943.5159, | |
| "train_samples_per_second": 0.723 | |
| } | |
| ], | |
| "max_steps": 6462, | |
| "num_train_epochs": 3, | |
| "total_flos": 169236001393016832, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |