| { | |
| "best_metric": 0.8748746375681541, | |
| "best_model_checkpoint": "v3dsyn8105.san-roberta-large", | |
| "epoch": 3.0, | |
| "global_step": 6126, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06, | |
| "eval_avg_sts": 0.831522642680213, | |
| "eval_sickr_spearman": 0.819481868688736, | |
| "eval_stsb_spearman": 0.8435634166716901, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "eval_avg_sts": 0.8394633788185508, | |
| "eval_sickr_spearman": 0.8260827829593759, | |
| "eval_stsb_spearman": 0.8528439746777257, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "eval_avg_sts": 0.8387226540359579, | |
| "eval_sickr_spearman": 0.8118421377561928, | |
| "eval_stsb_spearman": 0.865603170315723, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.183806725432584e-06, | |
| "loss": 0.4868, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "eval_avg_sts": 0.8340637818765272, | |
| "eval_sickr_spearman": 0.8097645044323858, | |
| "eval_stsb_spearman": 0.8583630593206686, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "eval_avg_sts": 0.8467054456929737, | |
| "eval_sickr_spearman": 0.8272759235502001, | |
| "eval_stsb_spearman": 0.8661349678357472, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "eval_avg_sts": 0.8413222265544698, | |
| "eval_sickr_spearman": 0.816061670017531, | |
| "eval_stsb_spearman": 0.8665827830914085, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "eval_avg_sts": 0.8498089066273828, | |
| "eval_sickr_spearman": 0.8340720362170844, | |
| "eval_stsb_spearman": 0.8655457770376812, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.367613450865165e-06, | |
| "loss": 0.1238, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "eval_avg_sts": 0.8486804289849034, | |
| "eval_sickr_spearman": 0.8251996831283348, | |
| "eval_stsb_spearman": 0.8721611748414719, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "eval_avg_sts": 0.8467220562876994, | |
| "eval_sickr_spearman": 0.8293082155142495, | |
| "eval_stsb_spearman": 0.8641358970611491, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "eval_avg_sts": 0.8450297472977634, | |
| "eval_sickr_spearman": 0.8284341455302786, | |
| "eval_stsb_spearman": 0.8616253490652481, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "eval_avg_sts": 0.8399309524387217, | |
| "eval_sickr_spearman": 0.817097460720044, | |
| "eval_stsb_spearman": 0.8627644441573996, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.551420176297748e-06, | |
| "loss": 0.1051, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "eval_avg_sts": 0.8488287267471869, | |
| "eval_sickr_spearman": 0.8298005823350916, | |
| "eval_stsb_spearman": 0.8678568711592822, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_avg_sts": 0.8500693337499818, | |
| "eval_sickr_spearman": 0.832001703620696, | |
| "eval_stsb_spearman": 0.8681369638792675, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "eval_avg_sts": 0.8539883178184261, | |
| "eval_sickr_spearman": 0.8361804094458117, | |
| "eval_stsb_spearman": 0.8717962261910405, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "eval_avg_sts": 0.8506565279083476, | |
| "eval_sickr_spearman": 0.8323338386871405, | |
| "eval_stsb_spearman": 0.8689792171295547, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.735226901730331e-06, | |
| "loss": 0.0964, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "eval_avg_sts": 0.8566417773406937, | |
| "eval_sickr_spearman": 0.8427579325700513, | |
| "eval_stsb_spearman": 0.8705256221113361, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "eval_avg_sts": 0.8506860271225665, | |
| "eval_sickr_spearman": 0.8358071597565362, | |
| "eval_stsb_spearman": 0.8655648944885967, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "eval_avg_sts": 0.8526145284643478, | |
| "eval_sickr_spearman": 0.8352194992304408, | |
| "eval_stsb_spearman": 0.8700095576982549, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "eval_avg_sts": 0.8512789094461894, | |
| "eval_sickr_spearman": 0.8339942258327547, | |
| "eval_stsb_spearman": 0.8685635930596243, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 5.919033627162912e-06, | |
| "loss": 0.0724, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "eval_avg_sts": 0.8542737971692321, | |
| "eval_sickr_spearman": 0.8376304683982283, | |
| "eval_stsb_spearman": 0.870917125940236, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "eval_avg_sts": 0.853215393568972, | |
| "eval_sickr_spearman": 0.8382630380041683, | |
| "eval_stsb_spearman": 0.8681677491337757, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "eval_avg_sts": 0.8512616991878239, | |
| "eval_sickr_spearman": 0.8376803727126226, | |
| "eval_stsb_spearman": 0.8648430256630253, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "eval_avg_sts": 0.8445713819963026, | |
| "eval_sickr_spearman": 0.8245212438205214, | |
| "eval_stsb_spearman": 0.8646215201720836, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.102840352595495e-06, | |
| "loss": 0.0672, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "eval_avg_sts": 0.8547468779877831, | |
| "eval_sickr_spearman": 0.840675015825086, | |
| "eval_stsb_spearman": 0.8688187401504804, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "eval_avg_sts": 0.8480351385120373, | |
| "eval_sickr_spearman": 0.828842553985807, | |
| "eval_stsb_spearman": 0.8672277230382677, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "eval_avg_sts": 0.851759527195628, | |
| "eval_sickr_spearman": 0.8318612606800909, | |
| "eval_stsb_spearman": 0.871657793711165, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "eval_avg_sts": 0.852688164945469, | |
| "eval_sickr_spearman": 0.8305016923227837, | |
| "eval_stsb_spearman": 0.8748746375681541, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.286647078028077e-06, | |
| "loss": 0.0671, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "eval_avg_sts": 0.8476247717318004, | |
| "eval_sickr_spearman": 0.825738544055369, | |
| "eval_stsb_spearman": 0.8695109994082318, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "eval_avg_sts": 0.8488707460006111, | |
| "eval_sickr_spearman": 0.8274017650359679, | |
| "eval_stsb_spearman": 0.8703397269652542, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "eval_avg_sts": 0.8474621405059352, | |
| "eval_sickr_spearman": 0.8264578578305062, | |
| "eval_stsb_spearman": 0.8684664231813641, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "eval_avg_sts": 0.8514977368080117, | |
| "eval_sickr_spearman": 0.8317304719908749, | |
| "eval_stsb_spearman": 0.8712650016251485, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 3.47045380346066e-06, | |
| "loss": 0.062, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "eval_avg_sts": 0.8495823034187644, | |
| "eval_sickr_spearman": 0.8280114718376232, | |
| "eval_stsb_spearman": 0.8711531349999054, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "eval_avg_sts": 0.8472080808839515, | |
| "eval_sickr_spearman": 0.8245909849798095, | |
| "eval_stsb_spearman": 0.8698251767880936, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "eval_avg_sts": 0.8502145621787748, | |
| "eval_sickr_spearman": 0.8296834344786841, | |
| "eval_stsb_spearman": 0.8707456898788655, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "eval_avg_sts": 0.8491882134989206, | |
| "eval_sickr_spearman": 0.8280905310305905, | |
| "eval_stsb_spearman": 0.8702858959672507, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 2.654260528893242e-06, | |
| "loss": 0.0507, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "eval_avg_sts": 0.8489413736886298, | |
| "eval_sickr_spearman": 0.8264288470452376, | |
| "eval_stsb_spearman": 0.8714539003320219, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "eval_avg_sts": 0.8403852153317499, | |
| "eval_sickr_spearman": 0.8150179061521793, | |
| "eval_stsb_spearman": 0.8657525245113207, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "eval_avg_sts": 0.8448990255323547, | |
| "eval_sickr_spearman": 0.8213330834003639, | |
| "eval_stsb_spearman": 0.8684649676643454, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "eval_avg_sts": 0.846744031918401, | |
| "eval_sickr_spearman": 0.8236191236733109, | |
| "eval_stsb_spearman": 0.8698689401634909, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 1.8380672543258246e-06, | |
| "loss": 0.0444, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "eval_avg_sts": 0.8446342020355491, | |
| "eval_sickr_spearman": 0.8218428374799265, | |
| "eval_stsb_spearman": 0.8674255665911716, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "eval_avg_sts": 0.8466793063787386, | |
| "eval_sickr_spearman": 0.8251428143042321, | |
| "eval_stsb_spearman": 0.8682157984532451, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "eval_avg_sts": 0.8472692519799312, | |
| "eval_sickr_spearman": 0.8273726581884964, | |
| "eval_stsb_spearman": 0.8671658457713661, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "eval_avg_sts": 0.8463050384868719, | |
| "eval_sickr_spearman": 0.8245047211216268, | |
| "eval_stsb_spearman": 0.868105355852117, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 1.021873979758407e-06, | |
| "loss": 0.0437, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "eval_avg_sts": 0.845129637101806, | |
| "eval_sickr_spearman": 0.8231749320472111, | |
| "eval_stsb_spearman": 0.8670843421564008, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "eval_avg_sts": 0.844980005172939, | |
| "eval_sickr_spearman": 0.8226033139089961, | |
| "eval_stsb_spearman": 0.8673566964368821, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "eval_avg_sts": 0.8454315705537849, | |
| "eval_sickr_spearman": 0.8231090814071395, | |
| "eval_stsb_spearman": 0.8677540597004305, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "eval_avg_sts": 0.8447869475256539, | |
| "eval_sickr_spearman": 0.8219029724189271, | |
| "eval_stsb_spearman": 0.8676709226323807, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 2.0568070519098922e-07, | |
| "loss": 0.0443, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "eval_avg_sts": 0.8453072401863975, | |
| "eval_sickr_spearman": 0.8228684455889346, | |
| "eval_stsb_spearman": 0.8677460347838605, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_avg_sts": 0.8456332694526848, | |
| "eval_sickr_spearman": 0.8233002451908632, | |
| "eval_stsb_spearman": 0.8679662937145064, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 6126, | |
| "train_runtime": 4896.9547, | |
| "train_samples_per_second": 1.251 | |
| } | |
| ], | |
| "max_steps": 6126, | |
| "num_train_epochs": 3, | |
| "total_flos": 213963637016494080, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |