diff --git "a/trainer_state.json" "b/trainer_state.json" deleted file mode 100644--- "a/trainer_state.json" +++ /dev/null @@ -1,11913 +0,0 @@ -{ - "best_metric": 42.0728, - "best_model_checkpoint": "/newdisk1/lkw/chatcell/t5_model/v22_small_50_0_100%/checkpoint-185000", - "epoch": 49.29942916450441, - "eval_steps": 5000, - "global_step": 190000, - "is_hyper_param_search": false, - "is_local_process_zero": true, - "is_world_process_zero": true, - "log_history": [ - { - "epoch": 0.03, - "learning_rate": 3.997924234561495e-05, - "loss": 7.3867, - "step": 100 - }, - { - "epoch": 0.05, - "learning_rate": 3.995848469122989e-05, - "loss": 5.0855, - "step": 200 - }, - { - "epoch": 0.08, - "learning_rate": 3.993772703684484e-05, - "loss": 5.0196, - "step": 300 - }, - { - "epoch": 0.1, - "learning_rate": 3.991696938245979e-05, - "loss": 4.9903, - "step": 400 - }, - { - "epoch": 0.13, - "learning_rate": 3.9896211728074735e-05, - "loss": 4.9698, - "step": 500 - }, - { - "epoch": 0.16, - "learning_rate": 3.987545407368968e-05, - "loss": 4.9409, - "step": 600 - }, - { - "epoch": 0.18, - "learning_rate": 3.9854696419304625e-05, - "loss": 4.924, - "step": 700 - }, - { - "epoch": 0.21, - "learning_rate": 3.983393876491956e-05, - "loss": 4.9034, - "step": 800 - }, - { - "epoch": 0.23, - "learning_rate": 3.981318111053451e-05, - "loss": 4.8898, - "step": 900 - }, - { - "epoch": 0.26, - "learning_rate": 3.979242345614946e-05, - "loss": 4.8703, - "step": 1000 - }, - { - "epoch": 0.29, - "learning_rate": 3.9771665801764405e-05, - "loss": 4.8548, - "step": 1100 - }, - { - "epoch": 0.31, - "learning_rate": 3.975090814737935e-05, - "loss": 4.8347, - "step": 1200 - }, - { - "epoch": 0.34, - "learning_rate": 3.9730150492994295e-05, - "loss": 4.8244, - "step": 1300 - }, - { - "epoch": 0.36, - "learning_rate": 3.970939283860924e-05, - "loss": 4.8042, - "step": 1400 - }, - { - "epoch": 0.39, - "learning_rate": 3.9688635184224185e-05, - "loss": 4.7933, - "step": 1500 - }, - { - "epoch": 0.42, - "learning_rate": 3.966787752983913e-05, - "loss": 4.7756, - "step": 1600 - }, - { - "epoch": 0.44, - "learning_rate": 3.964711987545408e-05, - "loss": 4.7667, - "step": 1700 - }, - { - "epoch": 0.47, - "learning_rate": 3.9626362221069026e-05, - "loss": 4.7527, - "step": 1800 - }, - { - "epoch": 0.49, - "learning_rate": 3.960560456668397e-05, - "loss": 4.7391, - "step": 1900 - }, - { - "epoch": 0.52, - "learning_rate": 3.9584846912298916e-05, - "loss": 4.73, - "step": 2000 - }, - { - "epoch": 0.54, - "learning_rate": 3.956408925791386e-05, - "loss": 4.7012, - "step": 2100 - }, - { - "epoch": 0.57, - "learning_rate": 3.9543331603528806e-05, - "loss": 4.6942, - "step": 2200 - }, - { - "epoch": 0.6, - "learning_rate": 3.952257394914375e-05, - "loss": 4.6838, - "step": 2300 - }, - { - "epoch": 0.62, - "learning_rate": 3.9501816294758696e-05, - "loss": 4.669, - "step": 2400 - }, - { - "epoch": 0.65, - "learning_rate": 3.948105864037364e-05, - "loss": 4.6536, - "step": 2500 - }, - { - "epoch": 0.67, - "learning_rate": 3.9460300985988586e-05, - "loss": 4.6464, - "step": 2600 - }, - { - "epoch": 0.7, - "learning_rate": 3.943954333160353e-05, - "loss": 4.6331, - "step": 2700 - }, - { - "epoch": 0.73, - "learning_rate": 3.9418785677218476e-05, - "loss": 4.6069, - "step": 2800 - }, - { - "epoch": 0.75, - "learning_rate": 3.939802802283342e-05, - "loss": 4.6039, - "step": 2900 - }, - { - "epoch": 0.78, - "learning_rate": 3.937727036844837e-05, - "loss": 4.5902, - "step": 3000 - }, - { - "epoch": 0.8, - "learning_rate": 3.935651271406332e-05, - "loss": 4.5851, - "step": 3100 - }, - { - "epoch": 0.83, - "learning_rate": 3.933575505967826e-05, - "loss": 4.5493, - "step": 3200 - }, - { - "epoch": 0.86, - "learning_rate": 3.931499740529321e-05, - "loss": 4.5515, - "step": 3300 - }, - { - "epoch": 0.88, - "learning_rate": 3.929423975090815e-05, - "loss": 4.5391, - "step": 3400 - }, - { - "epoch": 0.91, - "learning_rate": 3.92734820965231e-05, - "loss": 4.5266, - "step": 3500 - }, - { - "epoch": 0.93, - "learning_rate": 3.925272444213804e-05, - "loss": 4.518, - "step": 3600 - }, - { - "epoch": 0.96, - "learning_rate": 3.923196678775299e-05, - "loss": 4.5006, - "step": 3700 - }, - { - "epoch": 0.99, - "learning_rate": 3.921120913336793e-05, - "loss": 4.4839, - "step": 3800 - }, - { - "epoch": 1.01, - "learning_rate": 3.919045147898288e-05, - "loss": 4.4753, - "step": 3900 - }, - { - "epoch": 1.04, - "learning_rate": 3.916969382459782e-05, - "loss": 4.4482, - "step": 4000 - }, - { - "epoch": 1.06, - "learning_rate": 3.914893617021277e-05, - "loss": 4.446, - "step": 4100 - }, - { - "epoch": 1.09, - "learning_rate": 3.912817851582771e-05, - "loss": 4.4486, - "step": 4200 - }, - { - "epoch": 1.12, - "learning_rate": 3.9107420861442664e-05, - "loss": 4.4225, - "step": 4300 - }, - { - "epoch": 1.14, - "learning_rate": 3.908666320705761e-05, - "loss": 4.4129, - "step": 4400 - }, - { - "epoch": 1.17, - "learning_rate": 3.9065905552672554e-05, - "loss": 4.401, - "step": 4500 - }, - { - "epoch": 1.19, - "learning_rate": 3.90451478982875e-05, - "loss": 4.3917, - "step": 4600 - }, - { - "epoch": 1.22, - "learning_rate": 3.9024390243902444e-05, - "loss": 4.372, - "step": 4700 - }, - { - "epoch": 1.25, - "learning_rate": 3.900363258951739e-05, - "loss": 4.3572, - "step": 4800 - }, - { - "epoch": 1.27, - "learning_rate": 3.8982874935132334e-05, - "loss": 4.35, - "step": 4900 - }, - { - "epoch": 1.3, - "learning_rate": 3.896211728074728e-05, - "loss": 4.3412, - "step": 5000 - }, - { - "epoch": 1.3, - "eval_gen_len": 12.5597, - "eval_loss": 4.288102149963379, - "eval_rouge1": 13.4415, - "eval_rouge2": 1.5683, - "eval_rougeL": 12.4048, - "eval_rougeLsum": 12.4255, - "eval_runtime": 81.4031, - "eval_samples_per_second": 47.345, - "eval_steps_per_second": 5.921, - "step": 5000 - }, - { - "epoch": 1.32, - "learning_rate": 3.8941359626362224e-05, - "loss": 4.3273, - "step": 5100 - }, - { - "epoch": 1.35, - "learning_rate": 3.892060197197717e-05, - "loss": 4.3152, - "step": 5200 - }, - { - "epoch": 1.38, - "learning_rate": 3.8899844317592114e-05, - "loss": 4.3043, - "step": 5300 - }, - { - "epoch": 1.4, - "learning_rate": 3.887908666320706e-05, - "loss": 4.287, - "step": 5400 - }, - { - "epoch": 1.43, - "learning_rate": 3.8858329008822004e-05, - "loss": 4.2814, - "step": 5500 - }, - { - "epoch": 1.45, - "learning_rate": 3.8837571354436955e-05, - "loss": 4.267, - "step": 5600 - }, - { - "epoch": 1.48, - "learning_rate": 3.88168137000519e-05, - "loss": 4.2597, - "step": 5700 - }, - { - "epoch": 1.5, - "learning_rate": 3.8796056045666845e-05, - "loss": 4.2513, - "step": 5800 - }, - { - "epoch": 1.53, - "learning_rate": 3.877529839128179e-05, - "loss": 4.2313, - "step": 5900 - }, - { - "epoch": 1.56, - "learning_rate": 3.8754540736896735e-05, - "loss": 4.2101, - "step": 6000 - }, - { - "epoch": 1.58, - "learning_rate": 3.873378308251168e-05, - "loss": 4.2142, - "step": 6100 - }, - { - "epoch": 1.61, - "learning_rate": 3.8713025428126625e-05, - "loss": 4.2015, - "step": 6200 - }, - { - "epoch": 1.63, - "learning_rate": 3.869226777374157e-05, - "loss": 4.1821, - "step": 6300 - }, - { - "epoch": 1.66, - "learning_rate": 3.8671510119356515e-05, - "loss": 4.1763, - "step": 6400 - }, - { - "epoch": 1.69, - "learning_rate": 3.865075246497146e-05, - "loss": 4.1673, - "step": 6500 - }, - { - "epoch": 1.71, - "learning_rate": 3.8629994810586405e-05, - "loss": 4.1591, - "step": 6600 - }, - { - "epoch": 1.74, - "learning_rate": 3.860923715620135e-05, - "loss": 4.1596, - "step": 6700 - }, - { - "epoch": 1.76, - "learning_rate": 3.8588479501816295e-05, - "loss": 4.1356, - "step": 6800 - }, - { - "epoch": 1.79, - "learning_rate": 3.856772184743125e-05, - "loss": 4.1433, - "step": 6900 - }, - { - "epoch": 1.82, - "learning_rate": 3.854696419304619e-05, - "loss": 4.1157, - "step": 7000 - }, - { - "epoch": 1.84, - "learning_rate": 3.852620653866114e-05, - "loss": 4.0836, - "step": 7100 - }, - { - "epoch": 1.87, - "learning_rate": 3.850544888427608e-05, - "loss": 4.1102, - "step": 7200 - }, - { - "epoch": 1.89, - "learning_rate": 3.8484691229891027e-05, - "loss": 4.0818, - "step": 7300 - }, - { - "epoch": 1.92, - "learning_rate": 3.846393357550597e-05, - "loss": 4.0663, - "step": 7400 - }, - { - "epoch": 1.95, - "learning_rate": 3.8443175921120917e-05, - "loss": 4.07, - "step": 7500 - }, - { - "epoch": 1.97, - "learning_rate": 3.842241826673586e-05, - "loss": 4.0689, - "step": 7600 - }, - { - "epoch": 2.0, - "learning_rate": 3.8401660612350806e-05, - "loss": 4.0442, - "step": 7700 - }, - { - "epoch": 2.02, - "learning_rate": 3.838090295796575e-05, - "loss": 4.0421, - "step": 7800 - }, - { - "epoch": 2.05, - "learning_rate": 3.8360145303580696e-05, - "loss": 4.0381, - "step": 7900 - }, - { - "epoch": 2.08, - "learning_rate": 3.833938764919564e-05, - "loss": 4.0242, - "step": 8000 - }, - { - "epoch": 2.1, - "learning_rate": 3.8318629994810586e-05, - "loss": 4.0184, - "step": 8100 - }, - { - "epoch": 2.13, - "learning_rate": 3.829787234042554e-05, - "loss": 4.0159, - "step": 8200 - }, - { - "epoch": 2.15, - "learning_rate": 3.827711468604048e-05, - "loss": 4.0, - "step": 8300 - }, - { - "epoch": 2.18, - "learning_rate": 3.825635703165543e-05, - "loss": 3.9939, - "step": 8400 - }, - { - "epoch": 2.21, - "learning_rate": 3.823559937727037e-05, - "loss": 3.9851, - "step": 8500 - }, - { - "epoch": 2.23, - "learning_rate": 3.821484172288532e-05, - "loss": 3.9711, - "step": 8600 - }, - { - "epoch": 2.26, - "learning_rate": 3.819408406850026e-05, - "loss": 3.9732, - "step": 8700 - }, - { - "epoch": 2.28, - "learning_rate": 3.817332641411521e-05, - "loss": 3.9666, - "step": 8800 - }, - { - "epoch": 2.31, - "learning_rate": 3.815256875973015e-05, - "loss": 3.9528, - "step": 8900 - }, - { - "epoch": 2.34, - "learning_rate": 3.81318111053451e-05, - "loss": 3.9496, - "step": 9000 - }, - { - "epoch": 2.36, - "learning_rate": 3.811105345096004e-05, - "loss": 3.9426, - "step": 9100 - }, - { - "epoch": 2.39, - "learning_rate": 3.809029579657499e-05, - "loss": 3.9496, - "step": 9200 - }, - { - "epoch": 2.41, - "learning_rate": 3.806953814218993e-05, - "loss": 3.9376, - "step": 9300 - }, - { - "epoch": 2.44, - "learning_rate": 3.804878048780488e-05, - "loss": 3.9404, - "step": 9400 - }, - { - "epoch": 2.46, - "learning_rate": 3.802802283341983e-05, - "loss": 3.9204, - "step": 9500 - }, - { - "epoch": 2.49, - "learning_rate": 3.8007265179034774e-05, - "loss": 3.8957, - "step": 9600 - }, - { - "epoch": 2.52, - "learning_rate": 3.798650752464972e-05, - "loss": 3.9224, - "step": 9700 - }, - { - "epoch": 2.54, - "learning_rate": 3.7965749870264664e-05, - "loss": 3.9076, - "step": 9800 - }, - { - "epoch": 2.57, - "learning_rate": 3.794499221587961e-05, - "loss": 3.897, - "step": 9900 - }, - { - "epoch": 2.59, - "learning_rate": 3.7924234561494554e-05, - "loss": 3.8995, - "step": 10000 - }, - { - "epoch": 2.59, - "eval_gen_len": 12.2688, - "eval_loss": 3.854327917098999, - "eval_rouge1": 24.0743, - "eval_rouge2": 1.9282, - "eval_rougeL": 20.7914, - "eval_rougeLsum": 20.7945, - "eval_runtime": 81.1697, - "eval_samples_per_second": 47.481, - "eval_steps_per_second": 5.938, - "step": 10000 - }, - { - "epoch": 2.62, - "learning_rate": 3.79034769071095e-05, - "loss": 3.895, - "step": 10100 - }, - { - "epoch": 2.65, - "learning_rate": 3.7882719252724444e-05, - "loss": 3.8924, - "step": 10200 - }, - { - "epoch": 2.67, - "learning_rate": 3.786196159833939e-05, - "loss": 3.894, - "step": 10300 - }, - { - "epoch": 2.7, - "learning_rate": 3.7841203943954334e-05, - "loss": 3.8961, - "step": 10400 - }, - { - "epoch": 2.72, - "learning_rate": 3.782044628956928e-05, - "loss": 3.8752, - "step": 10500 - }, - { - "epoch": 2.75, - "learning_rate": 3.7799688635184224e-05, - "loss": 3.8819, - "step": 10600 - }, - { - "epoch": 2.78, - "learning_rate": 3.777893098079917e-05, - "loss": 3.8756, - "step": 10700 - }, - { - "epoch": 2.8, - "learning_rate": 3.775817332641412e-05, - "loss": 3.8762, - "step": 10800 - }, - { - "epoch": 2.83, - "learning_rate": 3.7737415672029066e-05, - "loss": 3.849, - "step": 10900 - }, - { - "epoch": 2.85, - "learning_rate": 3.771665801764401e-05, - "loss": 3.8517, - "step": 11000 - }, - { - "epoch": 2.88, - "learning_rate": 3.7695900363258956e-05, - "loss": 3.8561, - "step": 11100 - }, - { - "epoch": 2.91, - "learning_rate": 3.76751427088739e-05, - "loss": 3.8511, - "step": 11200 - }, - { - "epoch": 2.93, - "learning_rate": 3.7654385054488846e-05, - "loss": 3.8213, - "step": 11300 - }, - { - "epoch": 2.96, - "learning_rate": 3.763362740010379e-05, - "loss": 3.8539, - "step": 11400 - }, - { - "epoch": 2.98, - "learning_rate": 3.7612869745718736e-05, - "loss": 3.844, - "step": 11500 - }, - { - "epoch": 3.01, - "learning_rate": 3.759211209133368e-05, - "loss": 3.8349, - "step": 11600 - }, - { - "epoch": 3.04, - "learning_rate": 3.7571354436948626e-05, - "loss": 3.8363, - "step": 11700 - }, - { - "epoch": 3.06, - "learning_rate": 3.755059678256357e-05, - "loss": 3.8444, - "step": 11800 - }, - { - "epoch": 3.09, - "learning_rate": 3.7529839128178515e-05, - "loss": 3.8336, - "step": 11900 - }, - { - "epoch": 3.11, - "learning_rate": 3.750908147379346e-05, - "loss": 3.8238, - "step": 12000 - }, - { - "epoch": 3.14, - "learning_rate": 3.748832381940841e-05, - "loss": 3.7921, - "step": 12100 - }, - { - "epoch": 3.17, - "learning_rate": 3.746756616502336e-05, - "loss": 3.8307, - "step": 12200 - }, - { - "epoch": 3.19, - "learning_rate": 3.74468085106383e-05, - "loss": 3.8104, - "step": 12300 - }, - { - "epoch": 3.22, - "learning_rate": 3.742605085625325e-05, - "loss": 3.8083, - "step": 12400 - }, - { - "epoch": 3.24, - "learning_rate": 3.740529320186819e-05, - "loss": 3.8045, - "step": 12500 - }, - { - "epoch": 3.27, - "learning_rate": 3.738453554748314e-05, - "loss": 3.814, - "step": 12600 - }, - { - "epoch": 3.3, - "learning_rate": 3.736377789309808e-05, - "loss": 3.816, - "step": 12700 - }, - { - "epoch": 3.32, - "learning_rate": 3.734302023871303e-05, - "loss": 3.8151, - "step": 12800 - }, - { - "epoch": 3.35, - "learning_rate": 3.732226258432797e-05, - "loss": 3.8171, - "step": 12900 - }, - { - "epoch": 3.37, - "learning_rate": 3.730150492994292e-05, - "loss": 3.7895, - "step": 13000 - }, - { - "epoch": 3.4, - "learning_rate": 3.728074727555786e-05, - "loss": 3.7782, - "step": 13100 - }, - { - "epoch": 3.43, - "learning_rate": 3.725998962117281e-05, - "loss": 3.7925, - "step": 13200 - }, - { - "epoch": 3.45, - "learning_rate": 3.723923196678775e-05, - "loss": 3.7955, - "step": 13300 - }, - { - "epoch": 3.48, - "learning_rate": 3.7218474312402704e-05, - "loss": 3.7964, - "step": 13400 - }, - { - "epoch": 3.5, - "learning_rate": 3.719771665801765e-05, - "loss": 3.7809, - "step": 13500 - }, - { - "epoch": 3.53, - "learning_rate": 3.7176959003632593e-05, - "loss": 3.816, - "step": 13600 - }, - { - "epoch": 3.55, - "learning_rate": 3.715620134924754e-05, - "loss": 3.795, - "step": 13700 - }, - { - "epoch": 3.58, - "learning_rate": 3.713544369486248e-05, - "loss": 3.7816, - "step": 13800 - }, - { - "epoch": 3.61, - "learning_rate": 3.711468604047743e-05, - "loss": 3.784, - "step": 13900 - }, - { - "epoch": 3.63, - "learning_rate": 3.709392838609237e-05, - "loss": 3.7755, - "step": 14000 - }, - { - "epoch": 3.66, - "learning_rate": 3.7073170731707325e-05, - "loss": 3.7809, - "step": 14100 - }, - { - "epoch": 3.68, - "learning_rate": 3.705241307732227e-05, - "loss": 3.7854, - "step": 14200 - }, - { - "epoch": 3.71, - "learning_rate": 3.703165542293721e-05, - "loss": 3.7811, - "step": 14300 - }, - { - "epoch": 3.74, - "learning_rate": 3.701089776855215e-05, - "loss": 3.7859, - "step": 14400 - }, - { - "epoch": 3.76, - "learning_rate": 3.69901401141671e-05, - "loss": 3.771, - "step": 14500 - }, - { - "epoch": 3.79, - "learning_rate": 3.696938245978204e-05, - "loss": 3.7613, - "step": 14600 - }, - { - "epoch": 3.81, - "learning_rate": 3.6948624805396995e-05, - "loss": 3.7879, - "step": 14700 - }, - { - "epoch": 3.84, - "learning_rate": 3.692786715101194e-05, - "loss": 3.775, - "step": 14800 - }, - { - "epoch": 3.87, - "learning_rate": 3.6907109496626885e-05, - "loss": 3.7578, - "step": 14900 - }, - { - "epoch": 3.89, - "learning_rate": 3.688635184224183e-05, - "loss": 3.7673, - "step": 15000 - }, - { - "epoch": 3.89, - "eval_gen_len": 12.2688, - "eval_loss": 3.7353405952453613, - "eval_rouge1": 25.2952, - "eval_rouge2": 2.0011, - "eval_rougeL": 21.2449, - "eval_rougeLsum": 21.2222, - "eval_runtime": 81.3221, - "eval_samples_per_second": 47.392, - "eval_steps_per_second": 5.927, - "step": 15000 - }, - { - "epoch": 3.92, - "learning_rate": 3.6865594187856775e-05, - "loss": 3.7752, - "step": 15100 - }, - { - "epoch": 3.94, - "learning_rate": 3.684483653347172e-05, - "loss": 3.7749, - "step": 15200 - }, - { - "epoch": 3.97, - "learning_rate": 3.6824078879086665e-05, - "loss": 3.7379, - "step": 15300 - }, - { - "epoch": 4.0, - "learning_rate": 3.6803321224701616e-05, - "loss": 3.756, - "step": 15400 - }, - { - "epoch": 4.02, - "learning_rate": 3.678256357031656e-05, - "loss": 3.7516, - "step": 15500 - }, - { - "epoch": 4.05, - "learning_rate": 3.6761805915931506e-05, - "loss": 3.7543, - "step": 15600 - }, - { - "epoch": 4.07, - "learning_rate": 3.674104826154645e-05, - "loss": 3.7594, - "step": 15700 - }, - { - "epoch": 4.1, - "learning_rate": 3.6720290607161396e-05, - "loss": 3.7529, - "step": 15800 - }, - { - "epoch": 4.13, - "learning_rate": 3.6699532952776334e-05, - "loss": 3.7571, - "step": 15900 - }, - { - "epoch": 4.15, - "learning_rate": 3.667877529839128e-05, - "loss": 3.7617, - "step": 16000 - }, - { - "epoch": 4.18, - "learning_rate": 3.665801764400623e-05, - "loss": 3.7688, - "step": 16100 - }, - { - "epoch": 4.2, - "learning_rate": 3.6637259989621176e-05, - "loss": 3.7514, - "step": 16200 - }, - { - "epoch": 4.23, - "learning_rate": 3.661650233523612e-05, - "loss": 3.7574, - "step": 16300 - }, - { - "epoch": 4.26, - "learning_rate": 3.6595744680851066e-05, - "loss": 3.766, - "step": 16400 - }, - { - "epoch": 4.28, - "learning_rate": 3.657498702646601e-05, - "loss": 3.7406, - "step": 16500 - }, - { - "epoch": 4.31, - "learning_rate": 3.6554229372080956e-05, - "loss": 3.7226, - "step": 16600 - }, - { - "epoch": 4.33, - "learning_rate": 3.653347171769591e-05, - "loss": 3.7476, - "step": 16700 - }, - { - "epoch": 4.36, - "learning_rate": 3.651271406331085e-05, - "loss": 3.7391, - "step": 16800 - }, - { - "epoch": 4.39, - "learning_rate": 3.64919564089258e-05, - "loss": 3.7396, - "step": 16900 - }, - { - "epoch": 4.41, - "learning_rate": 3.647119875454074e-05, - "loss": 3.7518, - "step": 17000 - }, - { - "epoch": 4.44, - "learning_rate": 3.645044110015569e-05, - "loss": 3.7414, - "step": 17100 - }, - { - "epoch": 4.46, - "learning_rate": 3.642968344577063e-05, - "loss": 3.747, - "step": 17200 - }, - { - "epoch": 4.49, - "learning_rate": 3.640892579138558e-05, - "loss": 3.7615, - "step": 17300 - }, - { - "epoch": 4.51, - "learning_rate": 3.638816813700052e-05, - "loss": 3.7379, - "step": 17400 - }, - { - "epoch": 4.54, - "learning_rate": 3.636741048261547e-05, - "loss": 3.7485, - "step": 17500 - }, - { - "epoch": 4.57, - "learning_rate": 3.634665282823041e-05, - "loss": 3.7474, - "step": 17600 - }, - { - "epoch": 4.59, - "learning_rate": 3.632589517384536e-05, - "loss": 3.7209, - "step": 17700 - }, - { - "epoch": 4.62, - "learning_rate": 3.63051375194603e-05, - "loss": 3.7424, - "step": 17800 - }, - { - "epoch": 4.64, - "learning_rate": 3.628437986507525e-05, - "loss": 3.7427, - "step": 17900 - }, - { - "epoch": 4.67, - "learning_rate": 3.62636222106902e-05, - "loss": 3.7466, - "step": 18000 - }, - { - "epoch": 4.7, - "learning_rate": 3.6242864556305144e-05, - "loss": 3.7372, - "step": 18100 - }, - { - "epoch": 4.72, - "learning_rate": 3.622210690192009e-05, - "loss": 3.7325, - "step": 18200 - }, - { - "epoch": 4.75, - "learning_rate": 3.6201349247535034e-05, - "loss": 3.7315, - "step": 18300 - }, - { - "epoch": 4.77, - "learning_rate": 3.618059159314998e-05, - "loss": 3.7248, - "step": 18400 - }, - { - "epoch": 4.8, - "learning_rate": 3.6159833938764924e-05, - "loss": 3.7315, - "step": 18500 - }, - { - "epoch": 4.83, - "learning_rate": 3.613907628437987e-05, - "loss": 3.7164, - "step": 18600 - }, - { - "epoch": 4.85, - "learning_rate": 3.6118318629994814e-05, - "loss": 3.7248, - "step": 18700 - }, - { - "epoch": 4.88, - "learning_rate": 3.609756097560976e-05, - "loss": 3.7286, - "step": 18800 - }, - { - "epoch": 4.9, - "learning_rate": 3.6076803321224704e-05, - "loss": 3.7321, - "step": 18900 - }, - { - "epoch": 4.93, - "learning_rate": 3.605604566683965e-05, - "loss": 3.7238, - "step": 19000 - }, - { - "epoch": 4.96, - "learning_rate": 3.6035288012454594e-05, - "loss": 3.7334, - "step": 19100 - }, - { - "epoch": 4.98, - "learning_rate": 3.601453035806954e-05, - "loss": 3.7116, - "step": 19200 - }, - { - "epoch": 5.01, - "learning_rate": 3.599377270368449e-05, - "loss": 3.7395, - "step": 19300 - }, - { - "epoch": 5.03, - "learning_rate": 3.5973015049299435e-05, - "loss": 3.7334, - "step": 19400 - }, - { - "epoch": 5.06, - "learning_rate": 3.595225739491438e-05, - "loss": 3.7274, - "step": 19500 - }, - { - "epoch": 5.09, - "learning_rate": 3.5931499740529325e-05, - "loss": 3.7216, - "step": 19600 - }, - { - "epoch": 5.11, - "learning_rate": 3.591074208614427e-05, - "loss": 3.7238, - "step": 19700 - }, - { - "epoch": 5.14, - "learning_rate": 3.5889984431759215e-05, - "loss": 3.7158, - "step": 19800 - }, - { - "epoch": 5.16, - "learning_rate": 3.586922677737416e-05, - "loss": 3.7302, - "step": 19900 - }, - { - "epoch": 5.19, - "learning_rate": 3.5848469122989105e-05, - "loss": 3.7137, - "step": 20000 - }, - { - "epoch": 5.19, - "eval_gen_len": 12.2688, - "eval_loss": 3.690828561782837, - "eval_rouge1": 26.1164, - "eval_rouge2": 2.0984, - "eval_rougeL": 22.1244, - "eval_rougeLsum": 22.1089, - "eval_runtime": 81.0762, - "eval_samples_per_second": 47.536, - "eval_steps_per_second": 5.945, - "step": 20000 - }, - { - "epoch": 5.22, - "learning_rate": 3.582771146860405e-05, - "loss": 3.7286, - "step": 20100 - }, - { - "epoch": 5.24, - "learning_rate": 3.5806953814218995e-05, - "loss": 3.7333, - "step": 20200 - }, - { - "epoch": 5.27, - "learning_rate": 3.578619615983394e-05, - "loss": 3.7041, - "step": 20300 - }, - { - "epoch": 5.29, - "learning_rate": 3.5765438505448885e-05, - "loss": 3.6972, - "step": 20400 - }, - { - "epoch": 5.32, - "learning_rate": 3.574468085106383e-05, - "loss": 3.7198, - "step": 20500 - }, - { - "epoch": 5.35, - "learning_rate": 3.572392319667878e-05, - "loss": 3.7212, - "step": 20600 - }, - { - "epoch": 5.37, - "learning_rate": 3.570316554229373e-05, - "loss": 3.7162, - "step": 20700 - }, - { - "epoch": 5.4, - "learning_rate": 3.568240788790867e-05, - "loss": 3.7239, - "step": 20800 - }, - { - "epoch": 5.42, - "learning_rate": 3.566165023352362e-05, - "loss": 3.7266, - "step": 20900 - }, - { - "epoch": 5.45, - "learning_rate": 3.564089257913856e-05, - "loss": 3.7136, - "step": 21000 - }, - { - "epoch": 5.47, - "learning_rate": 3.562013492475351e-05, - "loss": 3.7118, - "step": 21100 - }, - { - "epoch": 5.5, - "learning_rate": 3.559937727036845e-05, - "loss": 3.7149, - "step": 21200 - }, - { - "epoch": 5.53, - "learning_rate": 3.55786196159834e-05, - "loss": 3.7064, - "step": 21300 - }, - { - "epoch": 5.55, - "learning_rate": 3.555786196159834e-05, - "loss": 3.6967, - "step": 21400 - }, - { - "epoch": 5.58, - "learning_rate": 3.5537104307213287e-05, - "loss": 3.7106, - "step": 21500 - }, - { - "epoch": 5.6, - "learning_rate": 3.551634665282823e-05, - "loss": 3.6849, - "step": 21600 - }, - { - "epoch": 5.63, - "learning_rate": 3.5495588998443177e-05, - "loss": 3.6902, - "step": 21700 - }, - { - "epoch": 5.66, - "learning_rate": 3.547483134405812e-05, - "loss": 3.6975, - "step": 21800 - }, - { - "epoch": 5.68, - "learning_rate": 3.545407368967307e-05, - "loss": 3.7142, - "step": 21900 - }, - { - "epoch": 5.71, - "learning_rate": 3.543331603528802e-05, - "loss": 3.6887, - "step": 22000 - }, - { - "epoch": 5.73, - "learning_rate": 3.541255838090296e-05, - "loss": 3.6966, - "step": 22100 - }, - { - "epoch": 5.76, - "learning_rate": 3.539180072651791e-05, - "loss": 3.7046, - "step": 22200 - }, - { - "epoch": 5.79, - "learning_rate": 3.537104307213285e-05, - "loss": 3.7046, - "step": 22300 - }, - { - "epoch": 5.81, - "learning_rate": 3.53502854177478e-05, - "loss": 3.7108, - "step": 22400 - }, - { - "epoch": 5.84, - "learning_rate": 3.532952776336274e-05, - "loss": 3.6994, - "step": 22500 - }, - { - "epoch": 5.86, - "learning_rate": 3.530877010897769e-05, - "loss": 3.6976, - "step": 22600 - }, - { - "epoch": 5.89, - "learning_rate": 3.528801245459263e-05, - "loss": 3.7163, - "step": 22700 - }, - { - "epoch": 5.92, - "learning_rate": 3.526725480020758e-05, - "loss": 3.7022, - "step": 22800 - }, - { - "epoch": 5.94, - "learning_rate": 3.524649714582252e-05, - "loss": 3.6893, - "step": 22900 - }, - { - "epoch": 5.97, - "learning_rate": 3.522573949143747e-05, - "loss": 3.6954, - "step": 23000 - }, - { - "epoch": 5.99, - "learning_rate": 3.520498183705241e-05, - "loss": 3.7143, - "step": 23100 - }, - { - "epoch": 6.02, - "learning_rate": 3.5184224182667365e-05, - "loss": 3.6649, - "step": 23200 - }, - { - "epoch": 6.05, - "learning_rate": 3.516346652828231e-05, - "loss": 3.6953, - "step": 23300 - }, - { - "epoch": 6.07, - "learning_rate": 3.5142708873897255e-05, - "loss": 3.7025, - "step": 23400 - }, - { - "epoch": 6.1, - "learning_rate": 3.51219512195122e-05, - "loss": 3.7151, - "step": 23500 - }, - { - "epoch": 6.12, - "learning_rate": 3.5101193565127144e-05, - "loss": 3.6985, - "step": 23600 - }, - { - "epoch": 6.15, - "learning_rate": 3.508043591074209e-05, - "loss": 3.7094, - "step": 23700 - }, - { - "epoch": 6.18, - "learning_rate": 3.5059678256357034e-05, - "loss": 3.6963, - "step": 23800 - }, - { - "epoch": 6.2, - "learning_rate": 3.503892060197198e-05, - "loss": 3.7083, - "step": 23900 - }, - { - "epoch": 6.23, - "learning_rate": 3.5018162947586924e-05, - "loss": 3.6809, - "step": 24000 - }, - { - "epoch": 6.25, - "learning_rate": 3.499740529320187e-05, - "loss": 3.6915, - "step": 24100 - }, - { - "epoch": 6.28, - "learning_rate": 3.4976647638816814e-05, - "loss": 3.6886, - "step": 24200 - }, - { - "epoch": 6.31, - "learning_rate": 3.495588998443176e-05, - "loss": 3.7067, - "step": 24300 - }, - { - "epoch": 6.33, - "learning_rate": 3.4935132330046704e-05, - "loss": 3.6912, - "step": 24400 - }, - { - "epoch": 6.36, - "learning_rate": 3.4914374675661656e-05, - "loss": 3.7084, - "step": 24500 - }, - { - "epoch": 6.38, - "learning_rate": 3.48936170212766e-05, - "loss": 3.6849, - "step": 24600 - }, - { - "epoch": 6.41, - "learning_rate": 3.4872859366891546e-05, - "loss": 3.7204, - "step": 24700 - }, - { - "epoch": 6.43, - "learning_rate": 3.485210171250649e-05, - "loss": 3.7057, - "step": 24800 - }, - { - "epoch": 6.46, - "learning_rate": 3.4831344058121436e-05, - "loss": 3.7103, - "step": 24900 - }, - { - "epoch": 6.49, - "learning_rate": 3.481058640373638e-05, - "loss": 3.6905, - "step": 25000 - }, - { - "epoch": 6.49, - "eval_gen_len": 12.2701, - "eval_loss": 3.667330503463745, - "eval_rouge1": 26.7248, - "eval_rouge2": 2.0984, - "eval_rougeL": 22.7285, - "eval_rougeLsum": 22.7508, - "eval_runtime": 81.3332, - "eval_samples_per_second": 47.385, - "eval_steps_per_second": 5.926, - "step": 25000 - }, - { - "epoch": 6.51, - "learning_rate": 3.4789828749351326e-05, - "loss": 3.6919, - "step": 25100 - }, - { - "epoch": 6.54, - "learning_rate": 3.476907109496627e-05, - "loss": 3.7111, - "step": 25200 - }, - { - "epoch": 6.56, - "learning_rate": 3.4748313440581216e-05, - "loss": 3.6901, - "step": 25300 - }, - { - "epoch": 6.59, - "learning_rate": 3.472755578619616e-05, - "loss": 3.6921, - "step": 25400 - }, - { - "epoch": 6.62, - "learning_rate": 3.4706798131811106e-05, - "loss": 3.6951, - "step": 25500 - }, - { - "epoch": 6.64, - "learning_rate": 3.468604047742605e-05, - "loss": 3.7004, - "step": 25600 - }, - { - "epoch": 6.67, - "learning_rate": 3.4665282823040996e-05, - "loss": 3.6868, - "step": 25700 - }, - { - "epoch": 6.69, - "learning_rate": 3.464452516865594e-05, - "loss": 3.6954, - "step": 25800 - }, - { - "epoch": 6.72, - "learning_rate": 3.462376751427089e-05, - "loss": 3.7049, - "step": 25900 - }, - { - "epoch": 6.75, - "learning_rate": 3.460300985988584e-05, - "loss": 3.6947, - "step": 26000 - }, - { - "epoch": 6.77, - "learning_rate": 3.458225220550078e-05, - "loss": 3.6963, - "step": 26100 - }, - { - "epoch": 6.8, - "learning_rate": 3.456149455111573e-05, - "loss": 3.6925, - "step": 26200 - }, - { - "epoch": 6.82, - "learning_rate": 3.454073689673067e-05, - "loss": 3.6845, - "step": 26300 - }, - { - "epoch": 6.85, - "learning_rate": 3.451997924234562e-05, - "loss": 3.6786, - "step": 26400 - }, - { - "epoch": 6.88, - "learning_rate": 3.449922158796056e-05, - "loss": 3.6842, - "step": 26500 - }, - { - "epoch": 6.9, - "learning_rate": 3.447846393357551e-05, - "loss": 3.6805, - "step": 26600 - }, - { - "epoch": 6.93, - "learning_rate": 3.445770627919045e-05, - "loss": 3.7005, - "step": 26700 - }, - { - "epoch": 6.95, - "learning_rate": 3.44369486248054e-05, - "loss": 3.683, - "step": 26800 - }, - { - "epoch": 6.98, - "learning_rate": 3.441619097042034e-05, - "loss": 3.6775, - "step": 26900 - }, - { - "epoch": 7.01, - "learning_rate": 3.439543331603529e-05, - "loss": 3.6823, - "step": 27000 - }, - { - "epoch": 7.03, - "learning_rate": 3.437467566165023e-05, - "loss": 3.6702, - "step": 27100 - }, - { - "epoch": 7.06, - "learning_rate": 3.4353918007265184e-05, - "loss": 3.6796, - "step": 27200 - }, - { - "epoch": 7.08, - "learning_rate": 3.433316035288013e-05, - "loss": 3.6707, - "step": 27300 - }, - { - "epoch": 7.11, - "learning_rate": 3.4312402698495074e-05, - "loss": 3.6907, - "step": 27400 - }, - { - "epoch": 7.14, - "learning_rate": 3.429164504411002e-05, - "loss": 3.6784, - "step": 27500 - }, - { - "epoch": 7.16, - "learning_rate": 3.4270887389724963e-05, - "loss": 3.6915, - "step": 27600 - }, - { - "epoch": 7.19, - "learning_rate": 3.425012973533991e-05, - "loss": 3.6872, - "step": 27700 - }, - { - "epoch": 7.21, - "learning_rate": 3.422937208095486e-05, - "loss": 3.694, - "step": 27800 - }, - { - "epoch": 7.24, - "learning_rate": 3.42086144265698e-05, - "loss": 3.6703, - "step": 27900 - }, - { - "epoch": 7.27, - "learning_rate": 3.418785677218474e-05, - "loss": 3.6963, - "step": 28000 - }, - { - "epoch": 7.29, - "learning_rate": 3.416709911779969e-05, - "loss": 3.6891, - "step": 28100 - }, - { - "epoch": 7.32, - "learning_rate": 3.414634146341463e-05, - "loss": 3.7015, - "step": 28200 - }, - { - "epoch": 7.34, - "learning_rate": 3.412558380902958e-05, - "loss": 3.689, - "step": 28300 - }, - { - "epoch": 7.37, - "learning_rate": 3.410482615464452e-05, - "loss": 3.6655, - "step": 28400 - }, - { - "epoch": 7.39, - "learning_rate": 3.4084068500259475e-05, - "loss": 3.6964, - "step": 28500 - }, - { - "epoch": 7.42, - "learning_rate": 3.406331084587442e-05, - "loss": 3.6859, - "step": 28600 - }, - { - "epoch": 7.45, - "learning_rate": 3.4042553191489365e-05, - "loss": 3.6751, - "step": 28700 - }, - { - "epoch": 7.47, - "learning_rate": 3.402179553710431e-05, - "loss": 3.6822, - "step": 28800 - }, - { - "epoch": 7.5, - "learning_rate": 3.4001037882719255e-05, - "loss": 3.6945, - "step": 28900 - }, - { - "epoch": 7.52, - "learning_rate": 3.39802802283342e-05, - "loss": 3.6647, - "step": 29000 - }, - { - "epoch": 7.55, - "learning_rate": 3.395952257394915e-05, - "loss": 3.6802, - "step": 29100 - }, - { - "epoch": 7.58, - "learning_rate": 3.3938764919564097e-05, - "loss": 3.6815, - "step": 29200 - }, - { - "epoch": 7.6, - "learning_rate": 3.391800726517904e-05, - "loss": 3.6923, - "step": 29300 - }, - { - "epoch": 7.63, - "learning_rate": 3.389724961079398e-05, - "loss": 3.6765, - "step": 29400 - }, - { - "epoch": 7.65, - "learning_rate": 3.3876491956408925e-05, - "loss": 3.6841, - "step": 29500 - }, - { - "epoch": 7.68, - "learning_rate": 3.385573430202387e-05, - "loss": 3.669, - "step": 29600 - }, - { - "epoch": 7.71, - "learning_rate": 3.3834976647638815e-05, - "loss": 3.6691, - "step": 29700 - }, - { - "epoch": 7.73, - "learning_rate": 3.3814218993253766e-05, - "loss": 3.6699, - "step": 29800 - }, - { - "epoch": 7.76, - "learning_rate": 3.379346133886871e-05, - "loss": 3.6903, - "step": 29900 - }, - { - "epoch": 7.78, - "learning_rate": 3.3772703684483656e-05, - "loss": 3.6904, - "step": 30000 - }, - { - "epoch": 7.78, - "eval_gen_len": 12.2865, - "eval_loss": 3.6548309326171875, - "eval_rouge1": 27.2987, - "eval_rouge2": 2.0984, - "eval_rougeL": 23.3145, - "eval_rougeLsum": 23.3265, - "eval_runtime": 80.6803, - "eval_samples_per_second": 47.769, - "eval_steps_per_second": 5.974, - "step": 30000 - }, - { - "epoch": 7.81, - "learning_rate": 3.37519460300986e-05, - "loss": 3.69, - "step": 30100 - }, - { - "epoch": 7.84, - "learning_rate": 3.3731188375713546e-05, - "loss": 3.6648, - "step": 30200 - }, - { - "epoch": 7.86, - "learning_rate": 3.371043072132849e-05, - "loss": 3.6907, - "step": 30300 - }, - { - "epoch": 7.89, - "learning_rate": 3.368967306694344e-05, - "loss": 3.6636, - "step": 30400 - }, - { - "epoch": 7.91, - "learning_rate": 3.366891541255839e-05, - "loss": 3.6756, - "step": 30500 - }, - { - "epoch": 7.94, - "learning_rate": 3.364815775817333e-05, - "loss": 3.6653, - "step": 30600 - }, - { - "epoch": 7.97, - "learning_rate": 3.362740010378828e-05, - "loss": 3.6831, - "step": 30700 - }, - { - "epoch": 7.99, - "learning_rate": 3.360664244940322e-05, - "loss": 3.6724, - "step": 30800 - }, - { - "epoch": 8.02, - "learning_rate": 3.358588479501816e-05, - "loss": 3.6916, - "step": 30900 - }, - { - "epoch": 8.04, - "learning_rate": 3.3565127140633106e-05, - "loss": 3.6814, - "step": 31000 - }, - { - "epoch": 8.07, - "learning_rate": 3.354436948624806e-05, - "loss": 3.6791, - "step": 31100 - }, - { - "epoch": 8.1, - "learning_rate": 3.3523611831863e-05, - "loss": 3.6805, - "step": 31200 - }, - { - "epoch": 8.12, - "learning_rate": 3.350285417747795e-05, - "loss": 3.6936, - "step": 31300 - }, - { - "epoch": 8.15, - "learning_rate": 3.348209652309289e-05, - "loss": 3.6695, - "step": 31400 - }, - { - "epoch": 8.17, - "learning_rate": 3.346133886870784e-05, - "loss": 3.66, - "step": 31500 - }, - { - "epoch": 8.2, - "learning_rate": 3.344058121432278e-05, - "loss": 3.6621, - "step": 31600 - }, - { - "epoch": 8.23, - "learning_rate": 3.3419823559937734e-05, - "loss": 3.6892, - "step": 31700 - }, - { - "epoch": 8.25, - "learning_rate": 3.339906590555268e-05, - "loss": 3.6252, - "step": 31800 - }, - { - "epoch": 8.28, - "learning_rate": 3.3378308251167624e-05, - "loss": 3.6804, - "step": 31900 - }, - { - "epoch": 8.3, - "learning_rate": 3.335755059678257e-05, - "loss": 3.6879, - "step": 32000 - }, - { - "epoch": 8.33, - "learning_rate": 3.3336792942397514e-05, - "loss": 3.6683, - "step": 32100 - }, - { - "epoch": 8.35, - "learning_rate": 3.331603528801246e-05, - "loss": 3.6801, - "step": 32200 - }, - { - "epoch": 8.38, - "learning_rate": 3.3295277633627404e-05, - "loss": 3.6568, - "step": 32300 - }, - { - "epoch": 8.41, - "learning_rate": 3.327451997924235e-05, - "loss": 3.6621, - "step": 32400 - }, - { - "epoch": 8.43, - "learning_rate": 3.3253762324857294e-05, - "loss": 3.6686, - "step": 32500 - }, - { - "epoch": 8.46, - "learning_rate": 3.323300467047224e-05, - "loss": 3.6942, - "step": 32600 - }, - { - "epoch": 8.48, - "learning_rate": 3.3212247016087184e-05, - "loss": 3.6566, - "step": 32700 - }, - { - "epoch": 8.51, - "learning_rate": 3.319148936170213e-05, - "loss": 3.678, - "step": 32800 - }, - { - "epoch": 8.54, - "learning_rate": 3.3170731707317074e-05, - "loss": 3.6526, - "step": 32900 - }, - { - "epoch": 8.56, - "learning_rate": 3.3149974052932026e-05, - "loss": 3.6796, - "step": 33000 - }, - { - "epoch": 8.59, - "learning_rate": 3.312921639854697e-05, - "loss": 3.671, - "step": 33100 - }, - { - "epoch": 8.61, - "learning_rate": 3.3108458744161916e-05, - "loss": 3.6556, - "step": 33200 - }, - { - "epoch": 8.64, - "learning_rate": 3.308770108977686e-05, - "loss": 3.6819, - "step": 33300 - }, - { - "epoch": 8.67, - "learning_rate": 3.3066943435391806e-05, - "loss": 3.6574, - "step": 33400 - }, - { - "epoch": 8.69, - "learning_rate": 3.304618578100675e-05, - "loss": 3.6426, - "step": 33500 - }, - { - "epoch": 8.72, - "learning_rate": 3.3025428126621695e-05, - "loss": 3.6796, - "step": 33600 - }, - { - "epoch": 8.74, - "learning_rate": 3.300467047223664e-05, - "loss": 3.6892, - "step": 33700 - }, - { - "epoch": 8.77, - "learning_rate": 3.2983912817851585e-05, - "loss": 3.6702, - "step": 33800 - }, - { - "epoch": 8.8, - "learning_rate": 3.296315516346653e-05, - "loss": 3.6809, - "step": 33900 - }, - { - "epoch": 8.82, - "learning_rate": 3.2942397509081475e-05, - "loss": 3.6658, - "step": 34000 - }, - { - "epoch": 8.85, - "learning_rate": 3.292163985469642e-05, - "loss": 3.6747, - "step": 34100 - }, - { - "epoch": 8.87, - "learning_rate": 3.2900882200311365e-05, - "loss": 3.6703, - "step": 34200 - }, - { - "epoch": 8.9, - "learning_rate": 3.288012454592632e-05, - "loss": 3.6684, - "step": 34300 - }, - { - "epoch": 8.93, - "learning_rate": 3.285936689154126e-05, - "loss": 3.6565, - "step": 34400 - }, - { - "epoch": 8.95, - "learning_rate": 3.283860923715621e-05, - "loss": 3.6783, - "step": 34500 - }, - { - "epoch": 8.98, - "learning_rate": 3.281785158277115e-05, - "loss": 3.6727, - "step": 34600 - }, - { - "epoch": 9.0, - "learning_rate": 3.27970939283861e-05, - "loss": 3.6891, - "step": 34700 - }, - { - "epoch": 9.03, - "learning_rate": 3.277633627400104e-05, - "loss": 3.6667, - "step": 34800 - }, - { - "epoch": 9.06, - "learning_rate": 3.275557861961599e-05, - "loss": 3.6713, - "step": 34900 - }, - { - "epoch": 9.08, - "learning_rate": 3.273482096523093e-05, - "loss": 3.6643, - "step": 35000 - }, - { - "epoch": 9.08, - "eval_gen_len": 12.2763, - "eval_loss": 3.646332263946533, - "eval_rouge1": 27.3301, - "eval_rouge2": 2.0984, - "eval_rougeL": 23.3261, - "eval_rougeLsum": 23.3581, - "eval_runtime": 76.595, - "eval_samples_per_second": 50.317, - "eval_steps_per_second": 6.293, - "step": 35000 - }, - { - "epoch": 9.11, - "learning_rate": 3.271406331084588e-05, - "loss": 3.6574, - "step": 35100 - }, - { - "epoch": 9.13, - "learning_rate": 3.269330565646082e-05, - "loss": 3.6598, - "step": 35200 - }, - { - "epoch": 9.16, - "learning_rate": 3.267254800207577e-05, - "loss": 3.6643, - "step": 35300 - }, - { - "epoch": 9.19, - "learning_rate": 3.265179034769071e-05, - "loss": 3.6675, - "step": 35400 - }, - { - "epoch": 9.21, - "learning_rate": 3.263103269330566e-05, - "loss": 3.6809, - "step": 35500 - }, - { - "epoch": 9.24, - "learning_rate": 3.26102750389206e-05, - "loss": 3.6862, - "step": 35600 - }, - { - "epoch": 9.26, - "learning_rate": 3.258951738453555e-05, - "loss": 3.662, - "step": 35700 - }, - { - "epoch": 9.29, - "learning_rate": 3.25687597301505e-05, - "loss": 3.671, - "step": 35800 - }, - { - "epoch": 9.31, - "learning_rate": 3.254800207576544e-05, - "loss": 3.6771, - "step": 35900 - }, - { - "epoch": 9.34, - "learning_rate": 3.252724442138039e-05, - "loss": 3.6686, - "step": 36000 - }, - { - "epoch": 9.37, - "learning_rate": 3.250648676699533e-05, - "loss": 3.6508, - "step": 36100 - }, - { - "epoch": 9.39, - "learning_rate": 3.248572911261028e-05, - "loss": 3.6553, - "step": 36200 - }, - { - "epoch": 9.42, - "learning_rate": 3.246497145822522e-05, - "loss": 3.665, - "step": 36300 - }, - { - "epoch": 9.44, - "learning_rate": 3.244421380384017e-05, - "loss": 3.6832, - "step": 36400 - }, - { - "epoch": 9.47, - "learning_rate": 3.242345614945511e-05, - "loss": 3.6666, - "step": 36500 - }, - { - "epoch": 9.5, - "learning_rate": 3.240269849507006e-05, - "loss": 3.6608, - "step": 36600 - }, - { - "epoch": 9.52, - "learning_rate": 3.2381940840685e-05, - "loss": 3.6624, - "step": 36700 - }, - { - "epoch": 9.55, - "learning_rate": 3.236118318629995e-05, - "loss": 3.6643, - "step": 36800 - }, - { - "epoch": 9.57, - "learning_rate": 3.234042553191489e-05, - "loss": 3.6563, - "step": 36900 - }, - { - "epoch": 9.6, - "learning_rate": 3.2319667877529845e-05, - "loss": 3.6471, - "step": 37000 - }, - { - "epoch": 9.63, - "learning_rate": 3.229891022314479e-05, - "loss": 3.659, - "step": 37100 - }, - { - "epoch": 9.65, - "learning_rate": 3.2278152568759735e-05, - "loss": 3.6758, - "step": 37200 - }, - { - "epoch": 9.68, - "learning_rate": 3.225739491437468e-05, - "loss": 3.6391, - "step": 37300 - }, - { - "epoch": 9.7, - "learning_rate": 3.2236637259989625e-05, - "loss": 3.6669, - "step": 37400 - }, - { - "epoch": 9.73, - "learning_rate": 3.221587960560457e-05, - "loss": 3.6713, - "step": 37500 - }, - { - "epoch": 9.76, - "learning_rate": 3.2195121951219514e-05, - "loss": 3.6773, - "step": 37600 - }, - { - "epoch": 9.78, - "learning_rate": 3.217436429683446e-05, - "loss": 3.643, - "step": 37700 - }, - { - "epoch": 9.81, - "learning_rate": 3.2153606642449404e-05, - "loss": 3.6637, - "step": 37800 - }, - { - "epoch": 9.83, - "learning_rate": 3.213284898806435e-05, - "loss": 3.6277, - "step": 37900 - }, - { - "epoch": 9.86, - "learning_rate": 3.2112091333679294e-05, - "loss": 3.676, - "step": 38000 - }, - { - "epoch": 9.89, - "learning_rate": 3.209133367929424e-05, - "loss": 3.6501, - "step": 38100 - }, - { - "epoch": 9.91, - "learning_rate": 3.2070576024909184e-05, - "loss": 3.6605, - "step": 38200 - }, - { - "epoch": 9.94, - "learning_rate": 3.2049818370524136e-05, - "loss": 3.6792, - "step": 38300 - }, - { - "epoch": 9.96, - "learning_rate": 3.202906071613908e-05, - "loss": 3.6747, - "step": 38400 - }, - { - "epoch": 9.99, - "learning_rate": 3.2008303061754026e-05, - "loss": 3.6834, - "step": 38500 - }, - { - "epoch": 10.02, - "learning_rate": 3.198754540736897e-05, - "loss": 3.6586, - "step": 38600 - }, - { - "epoch": 10.04, - "learning_rate": 3.1966787752983916e-05, - "loss": 3.6534, - "step": 38700 - }, - { - "epoch": 10.07, - "learning_rate": 3.194603009859886e-05, - "loss": 3.6691, - "step": 38800 - }, - { - "epoch": 10.09, - "learning_rate": 3.1925272444213806e-05, - "loss": 3.655, - "step": 38900 - }, - { - "epoch": 10.12, - "learning_rate": 3.190451478982875e-05, - "loss": 3.6692, - "step": 39000 - }, - { - "epoch": 10.15, - "learning_rate": 3.1883757135443696e-05, - "loss": 3.6568, - "step": 39100 - }, - { - "epoch": 10.17, - "learning_rate": 3.186299948105864e-05, - "loss": 3.6839, - "step": 39200 - }, - { - "epoch": 10.2, - "learning_rate": 3.1842241826673586e-05, - "loss": 3.6715, - "step": 39300 - }, - { - "epoch": 10.22, - "learning_rate": 3.182148417228853e-05, - "loss": 3.6353, - "step": 39400 - }, - { - "epoch": 10.25, - "learning_rate": 3.1800726517903476e-05, - "loss": 3.6357, - "step": 39500 - }, - { - "epoch": 10.28, - "learning_rate": 3.177996886351843e-05, - "loss": 3.6646, - "step": 39600 - }, - { - "epoch": 10.3, - "learning_rate": 3.175921120913337e-05, - "loss": 3.6664, - "step": 39700 - }, - { - "epoch": 10.33, - "learning_rate": 3.173845355474832e-05, - "loss": 3.6654, - "step": 39800 - }, - { - "epoch": 10.35, - "learning_rate": 3.171769590036326e-05, - "loss": 3.6732, - "step": 39900 - }, - { - "epoch": 10.38, - "learning_rate": 3.169693824597821e-05, - "loss": 3.6677, - "step": 40000 - }, - { - "epoch": 10.38, - "eval_gen_len": 12.2836, - "eval_loss": 3.640160322189331, - "eval_rouge1": 27.5354, - "eval_rouge2": 2.095, - "eval_rougeL": 23.5496, - "eval_rougeLsum": 23.563, - "eval_runtime": 75.9134, - "eval_samples_per_second": 50.768, - "eval_steps_per_second": 6.349, - "step": 40000 - }, - { - "epoch": 10.4, - "learning_rate": 3.167618059159315e-05, - "loss": 3.6737, - "step": 40100 - }, - { - "epoch": 10.43, - "learning_rate": 3.16554229372081e-05, - "loss": 3.6693, - "step": 40200 - }, - { - "epoch": 10.46, - "learning_rate": 3.163466528282304e-05, - "loss": 3.6675, - "step": 40300 - }, - { - "epoch": 10.48, - "learning_rate": 3.161390762843799e-05, - "loss": 3.6627, - "step": 40400 - }, - { - "epoch": 10.51, - "learning_rate": 3.159314997405293e-05, - "loss": 3.6404, - "step": 40500 - }, - { - "epoch": 10.53, - "learning_rate": 3.157239231966788e-05, - "loss": 3.6688, - "step": 40600 - }, - { - "epoch": 10.56, - "learning_rate": 3.155163466528282e-05, - "loss": 3.6564, - "step": 40700 - }, - { - "epoch": 10.59, - "learning_rate": 3.153087701089777e-05, - "loss": 3.6103, - "step": 40800 - }, - { - "epoch": 10.61, - "learning_rate": 3.151011935651272e-05, - "loss": 3.6527, - "step": 40900 - }, - { - "epoch": 10.64, - "learning_rate": 3.1489361702127664e-05, - "loss": 3.6814, - "step": 41000 - }, - { - "epoch": 10.66, - "learning_rate": 3.146860404774261e-05, - "loss": 3.6341, - "step": 41100 - }, - { - "epoch": 10.69, - "learning_rate": 3.1447846393357554e-05, - "loss": 3.6594, - "step": 41200 - }, - { - "epoch": 10.72, - "learning_rate": 3.14270887389725e-05, - "loss": 3.6601, - "step": 41300 - }, - { - "epoch": 10.74, - "learning_rate": 3.1406331084587444e-05, - "loss": 3.6622, - "step": 41400 - }, - { - "epoch": 10.77, - "learning_rate": 3.138557343020239e-05, - "loss": 3.627, - "step": 41500 - }, - { - "epoch": 10.79, - "learning_rate": 3.1364815775817334e-05, - "loss": 3.6415, - "step": 41600 - }, - { - "epoch": 10.82, - "learning_rate": 3.134405812143228e-05, - "loss": 3.6322, - "step": 41700 - }, - { - "epoch": 10.85, - "learning_rate": 3.1323300467047223e-05, - "loss": 3.6771, - "step": 41800 - }, - { - "epoch": 10.87, - "learning_rate": 3.130254281266217e-05, - "loss": 3.6656, - "step": 41900 - }, - { - "epoch": 10.9, - "learning_rate": 3.1281785158277113e-05, - "loss": 3.6416, - "step": 42000 - }, - { - "epoch": 10.92, - "learning_rate": 3.126102750389206e-05, - "loss": 3.6719, - "step": 42100 - }, - { - "epoch": 10.95, - "learning_rate": 3.124026984950701e-05, - "loss": 3.6718, - "step": 42200 - }, - { - "epoch": 10.98, - "learning_rate": 3.1219512195121955e-05, - "loss": 3.6597, - "step": 42300 - }, - { - "epoch": 11.0, - "learning_rate": 3.11987545407369e-05, - "loss": 3.6492, - "step": 42400 - }, - { - "epoch": 11.03, - "learning_rate": 3.1177996886351845e-05, - "loss": 3.6484, - "step": 42500 - }, - { - "epoch": 11.05, - "learning_rate": 3.115723923196679e-05, - "loss": 3.6498, - "step": 42600 - }, - { - "epoch": 11.08, - "learning_rate": 3.1136481577581735e-05, - "loss": 3.6788, - "step": 42700 - }, - { - "epoch": 11.11, - "learning_rate": 3.111572392319669e-05, - "loss": 3.6653, - "step": 42800 - }, - { - "epoch": 11.13, - "learning_rate": 3.1094966268811625e-05, - "loss": 3.6476, - "step": 42900 - }, - { - "epoch": 11.16, - "learning_rate": 3.107420861442657e-05, - "loss": 3.671, - "step": 43000 - }, - { - "epoch": 11.18, - "learning_rate": 3.1053450960041515e-05, - "loss": 3.6251, - "step": 43100 - }, - { - "epoch": 11.21, - "learning_rate": 3.103269330565646e-05, - "loss": 3.6515, - "step": 43200 - }, - { - "epoch": 11.24, - "learning_rate": 3.1011935651271405e-05, - "loss": 3.6637, - "step": 43300 - }, - { - "epoch": 11.26, - "learning_rate": 3.099117799688635e-05, - "loss": 3.6481, - "step": 43400 - }, - { - "epoch": 11.29, - "learning_rate": 3.09704203425013e-05, - "loss": 3.6647, - "step": 43500 - }, - { - "epoch": 11.31, - "learning_rate": 3.0949662688116246e-05, - "loss": 3.6511, - "step": 43600 - }, - { - "epoch": 11.34, - "learning_rate": 3.092890503373119e-05, - "loss": 3.6438, - "step": 43700 - }, - { - "epoch": 11.36, - "learning_rate": 3.0908147379346136e-05, - "loss": 3.6742, - "step": 43800 - }, - { - "epoch": 11.39, - "learning_rate": 3.088738972496108e-05, - "loss": 3.6609, - "step": 43900 - }, - { - "epoch": 11.42, - "learning_rate": 3.0866632070576026e-05, - "loss": 3.6299, - "step": 44000 - }, - { - "epoch": 11.44, - "learning_rate": 3.084587441619098e-05, - "loss": 3.6595, - "step": 44100 - }, - { - "epoch": 11.47, - "learning_rate": 3.082511676180592e-05, - "loss": 3.6559, - "step": 44200 - }, - { - "epoch": 11.49, - "learning_rate": 3.080435910742087e-05, - "loss": 3.6838, - "step": 44300 - }, - { - "epoch": 11.52, - "learning_rate": 3.078360145303581e-05, - "loss": 3.667, - "step": 44400 - }, - { - "epoch": 11.55, - "learning_rate": 3.076284379865075e-05, - "loss": 3.6619, - "step": 44500 - }, - { - "epoch": 11.57, - "learning_rate": 3.0742086144265696e-05, - "loss": 3.6601, - "step": 44600 - }, - { - "epoch": 11.6, - "learning_rate": 3.072132848988064e-05, - "loss": 3.6608, - "step": 44700 - }, - { - "epoch": 11.62, - "learning_rate": 3.070057083549559e-05, - "loss": 3.654, - "step": 44800 - }, - { - "epoch": 11.65, - "learning_rate": 3.067981318111054e-05, - "loss": 3.6641, - "step": 44900 - }, - { - "epoch": 11.68, - "learning_rate": 3.065905552672548e-05, - "loss": 3.6526, - "step": 45000 - }, - { - "epoch": 11.68, - "eval_gen_len": 12.2807, - "eval_loss": 3.6362619400024414, - "eval_rouge1": 28.1357, - "eval_rouge2": 2.3033, - "eval_rougeL": 24.161, - "eval_rougeLsum": 24.1576, - "eval_runtime": 76.1206, - "eval_samples_per_second": 50.63, - "eval_steps_per_second": 6.332, - "step": 45000 - }, - { - "epoch": 11.7, - "learning_rate": 3.063829787234043e-05, - "loss": 3.6808, - "step": 45100 - }, - { - "epoch": 11.73, - "learning_rate": 3.061754021795537e-05, - "loss": 3.6443, - "step": 45200 - }, - { - "epoch": 11.75, - "learning_rate": 3.059678256357032e-05, - "loss": 3.6565, - "step": 45300 - }, - { - "epoch": 11.78, - "learning_rate": 3.057602490918527e-05, - "loss": 3.6611, - "step": 45400 - }, - { - "epoch": 11.81, - "learning_rate": 3.0555267254800214e-05, - "loss": 3.6532, - "step": 45500 - }, - { - "epoch": 11.83, - "learning_rate": 3.053450960041516e-05, - "loss": 3.655, - "step": 45600 - }, - { - "epoch": 11.86, - "learning_rate": 3.05137519460301e-05, - "loss": 3.6631, - "step": 45700 - }, - { - "epoch": 11.88, - "learning_rate": 3.0492994291645046e-05, - "loss": 3.6478, - "step": 45800 - }, - { - "epoch": 11.91, - "learning_rate": 3.047223663725999e-05, - "loss": 3.65, - "step": 45900 - }, - { - "epoch": 11.94, - "learning_rate": 3.0451478982874936e-05, - "loss": 3.6404, - "step": 46000 - }, - { - "epoch": 11.96, - "learning_rate": 3.0430721328489884e-05, - "loss": 3.6551, - "step": 46100 - }, - { - "epoch": 11.99, - "learning_rate": 3.040996367410483e-05, - "loss": 3.6547, - "step": 46200 - }, - { - "epoch": 12.01, - "learning_rate": 3.0389206019719774e-05, - "loss": 3.6557, - "step": 46300 - }, - { - "epoch": 12.04, - "learning_rate": 3.036844836533472e-05, - "loss": 3.6672, - "step": 46400 - }, - { - "epoch": 12.07, - "learning_rate": 3.0347690710949664e-05, - "loss": 3.6625, - "step": 46500 - }, - { - "epoch": 12.09, - "learning_rate": 3.032693305656461e-05, - "loss": 3.6332, - "step": 46600 - }, - { - "epoch": 12.12, - "learning_rate": 3.0306175402179554e-05, - "loss": 3.6408, - "step": 46700 - }, - { - "epoch": 12.14, - "learning_rate": 3.0285417747794502e-05, - "loss": 3.6518, - "step": 46800 - }, - { - "epoch": 12.17, - "learning_rate": 3.0264660093409447e-05, - "loss": 3.6416, - "step": 46900 - }, - { - "epoch": 12.2, - "learning_rate": 3.0243902439024392e-05, - "loss": 3.6405, - "step": 47000 - }, - { - "epoch": 12.22, - "learning_rate": 3.0223144784639337e-05, - "loss": 3.659, - "step": 47100 - }, - { - "epoch": 12.25, - "learning_rate": 3.0202387130254282e-05, - "loss": 3.6342, - "step": 47200 - }, - { - "epoch": 12.27, - "learning_rate": 3.0181629475869227e-05, - "loss": 3.6718, - "step": 47300 - }, - { - "epoch": 12.3, - "learning_rate": 3.0160871821484176e-05, - "loss": 3.6391, - "step": 47400 - }, - { - "epoch": 12.32, - "learning_rate": 3.014011416709912e-05, - "loss": 3.6564, - "step": 47500 - }, - { - "epoch": 12.35, - "learning_rate": 3.0119356512714066e-05, - "loss": 3.6481, - "step": 47600 - }, - { - "epoch": 12.38, - "learning_rate": 3.009859885832901e-05, - "loss": 3.6446, - "step": 47700 - }, - { - "epoch": 12.4, - "learning_rate": 3.0077841203943955e-05, - "loss": 3.6356, - "step": 47800 - }, - { - "epoch": 12.43, - "learning_rate": 3.00570835495589e-05, - "loss": 3.6563, - "step": 47900 - }, - { - "epoch": 12.45, - "learning_rate": 3.0036325895173845e-05, - "loss": 3.65, - "step": 48000 - }, - { - "epoch": 12.48, - "learning_rate": 3.0015568240788794e-05, - "loss": 3.6357, - "step": 48100 - }, - { - "epoch": 12.51, - "learning_rate": 2.999481058640374e-05, - "loss": 3.6648, - "step": 48200 - }, - { - "epoch": 12.53, - "learning_rate": 2.9974052932018684e-05, - "loss": 3.6636, - "step": 48300 - }, - { - "epoch": 12.56, - "learning_rate": 2.995329527763363e-05, - "loss": 3.6416, - "step": 48400 - }, - { - "epoch": 12.58, - "learning_rate": 2.9932537623248574e-05, - "loss": 3.638, - "step": 48500 - }, - { - "epoch": 12.61, - "learning_rate": 2.991177996886352e-05, - "loss": 3.6637, - "step": 48600 - }, - { - "epoch": 12.64, - "learning_rate": 2.9891022314478467e-05, - "loss": 3.6412, - "step": 48700 - }, - { - "epoch": 12.66, - "learning_rate": 2.9870264660093412e-05, - "loss": 3.6471, - "step": 48800 - }, - { - "epoch": 12.69, - "learning_rate": 2.9849507005708357e-05, - "loss": 3.6729, - "step": 48900 - }, - { - "epoch": 12.71, - "learning_rate": 2.9828749351323302e-05, - "loss": 3.6692, - "step": 49000 - }, - { - "epoch": 12.74, - "learning_rate": 2.9807991696938247e-05, - "loss": 3.6498, - "step": 49100 - }, - { - "epoch": 12.77, - "learning_rate": 2.9787234042553192e-05, - "loss": 3.6471, - "step": 49200 - }, - { - "epoch": 12.79, - "learning_rate": 2.9766476388168137e-05, - "loss": 3.6535, - "step": 49300 - }, - { - "epoch": 12.82, - "learning_rate": 2.9745718733783085e-05, - "loss": 3.6526, - "step": 49400 - }, - { - "epoch": 12.84, - "learning_rate": 2.972496107939803e-05, - "loss": 3.6555, - "step": 49500 - }, - { - "epoch": 12.87, - "learning_rate": 2.9704203425012975e-05, - "loss": 3.6368, - "step": 49600 - }, - { - "epoch": 12.9, - "learning_rate": 2.968344577062792e-05, - "loss": 3.6546, - "step": 49700 - }, - { - "epoch": 12.92, - "learning_rate": 2.9662688116242865e-05, - "loss": 3.6621, - "step": 49800 - }, - { - "epoch": 12.95, - "learning_rate": 2.964193046185781e-05, - "loss": 3.6332, - "step": 49900 - }, - { - "epoch": 12.97, - "learning_rate": 2.962117280747276e-05, - "loss": 3.652, - "step": 50000 - }, - { - "epoch": 12.97, - "eval_gen_len": 12.2789, - "eval_loss": 3.6313188076019287, - "eval_rouge1": 31.2625, - "eval_rouge2": 3.8392, - "eval_rougeL": 27.2814, - "eval_rougeLsum": 27.2856, - "eval_runtime": 76.4786, - "eval_samples_per_second": 50.393, - "eval_steps_per_second": 6.302, - "step": 50000 - }, - { - "epoch": 13.0, - "learning_rate": 2.9600415153087703e-05, - "loss": 3.6524, - "step": 50100 - }, - { - "epoch": 13.03, - "learning_rate": 2.9579657498702648e-05, - "loss": 3.666, - "step": 50200 - }, - { - "epoch": 13.05, - "learning_rate": 2.9558899844317593e-05, - "loss": 3.6342, - "step": 50300 - }, - { - "epoch": 13.08, - "learning_rate": 2.9538142189932538e-05, - "loss": 3.6535, - "step": 50400 - }, - { - "epoch": 13.1, - "learning_rate": 2.9517384535547483e-05, - "loss": 3.6558, - "step": 50500 - }, - { - "epoch": 13.13, - "learning_rate": 2.9496626881162428e-05, - "loss": 3.6452, - "step": 50600 - }, - { - "epoch": 13.16, - "learning_rate": 2.947586922677738e-05, - "loss": 3.6442, - "step": 50700 - }, - { - "epoch": 13.18, - "learning_rate": 2.945511157239232e-05, - "loss": 3.6576, - "step": 50800 - }, - { - "epoch": 13.21, - "learning_rate": 2.9434353918007266e-05, - "loss": 3.6663, - "step": 50900 - }, - { - "epoch": 13.23, - "learning_rate": 2.941359626362221e-05, - "loss": 3.6328, - "step": 51000 - }, - { - "epoch": 13.26, - "learning_rate": 2.9392838609237156e-05, - "loss": 3.6412, - "step": 51100 - }, - { - "epoch": 13.28, - "learning_rate": 2.93720809548521e-05, - "loss": 3.6603, - "step": 51200 - }, - { - "epoch": 13.31, - "learning_rate": 2.9351323300467053e-05, - "loss": 3.6423, - "step": 51300 - }, - { - "epoch": 13.34, - "learning_rate": 2.9330565646081998e-05, - "loss": 3.6527, - "step": 51400 - }, - { - "epoch": 13.36, - "learning_rate": 2.9309807991696943e-05, - "loss": 3.6655, - "step": 51500 - }, - { - "epoch": 13.39, - "learning_rate": 2.9289050337311885e-05, - "loss": 3.6603, - "step": 51600 - }, - { - "epoch": 13.41, - "learning_rate": 2.926829268292683e-05, - "loss": 3.6243, - "step": 51700 - }, - { - "epoch": 13.44, - "learning_rate": 2.9247535028541774e-05, - "loss": 3.6392, - "step": 51800 - }, - { - "epoch": 13.47, - "learning_rate": 2.922677737415672e-05, - "loss": 3.6298, - "step": 51900 - }, - { - "epoch": 13.49, - "learning_rate": 2.920601971977167e-05, - "loss": 3.6615, - "step": 52000 - }, - { - "epoch": 13.52, - "learning_rate": 2.9185262065386616e-05, - "loss": 3.6398, - "step": 52100 - }, - { - "epoch": 13.54, - "learning_rate": 2.916450441100156e-05, - "loss": 3.6652, - "step": 52200 - }, - { - "epoch": 13.57, - "learning_rate": 2.9143746756616506e-05, - "loss": 3.6702, - "step": 52300 - }, - { - "epoch": 13.6, - "learning_rate": 2.9122989102231448e-05, - "loss": 3.6696, - "step": 52400 - }, - { - "epoch": 13.62, - "learning_rate": 2.9102231447846393e-05, - "loss": 3.654, - "step": 52500 - }, - { - "epoch": 13.65, - "learning_rate": 2.9081473793461344e-05, - "loss": 3.6507, - "step": 52600 - }, - { - "epoch": 13.67, - "learning_rate": 2.906071613907629e-05, - "loss": 3.6341, - "step": 52700 - }, - { - "epoch": 13.7, - "learning_rate": 2.9039958484691234e-05, - "loss": 3.625, - "step": 52800 - }, - { - "epoch": 13.73, - "learning_rate": 2.901920083030618e-05, - "loss": 3.651, - "step": 52900 - }, - { - "epoch": 13.75, - "learning_rate": 2.8998443175921124e-05, - "loss": 3.6476, - "step": 53000 - }, - { - "epoch": 13.78, - "learning_rate": 2.8977685521536066e-05, - "loss": 3.6234, - "step": 53100 - }, - { - "epoch": 13.8, - "learning_rate": 2.895692786715101e-05, - "loss": 3.6568, - "step": 53200 - }, - { - "epoch": 13.83, - "learning_rate": 2.8936170212765963e-05, - "loss": 3.6294, - "step": 53300 - }, - { - "epoch": 13.86, - "learning_rate": 2.8915412558380908e-05, - "loss": 3.6513, - "step": 53400 - }, - { - "epoch": 13.88, - "learning_rate": 2.8894654903995852e-05, - "loss": 3.6455, - "step": 53500 - }, - { - "epoch": 13.91, - "learning_rate": 2.8873897249610797e-05, - "loss": 3.6467, - "step": 53600 - }, - { - "epoch": 13.93, - "learning_rate": 2.8853139595225742e-05, - "loss": 3.6241, - "step": 53700 - }, - { - "epoch": 13.96, - "learning_rate": 2.8832381940840687e-05, - "loss": 3.6388, - "step": 53800 - }, - { - "epoch": 13.99, - "learning_rate": 2.8811624286455636e-05, - "loss": 3.6207, - "step": 53900 - }, - { - "epoch": 14.01, - "learning_rate": 2.879086663207058e-05, - "loss": 3.665, - "step": 54000 - }, - { - "epoch": 14.04, - "learning_rate": 2.8770108977685526e-05, - "loss": 3.6486, - "step": 54100 - }, - { - "epoch": 14.06, - "learning_rate": 2.874935132330047e-05, - "loss": 3.6488, - "step": 54200 - }, - { - "epoch": 14.09, - "learning_rate": 2.8728593668915416e-05, - "loss": 3.6435, - "step": 54300 - }, - { - "epoch": 14.12, - "learning_rate": 2.870783601453036e-05, - "loss": 3.6244, - "step": 54400 - }, - { - "epoch": 14.14, - "learning_rate": 2.8687078360145306e-05, - "loss": 3.6598, - "step": 54500 - }, - { - "epoch": 14.17, - "learning_rate": 2.8666320705760254e-05, - "loss": 3.6124, - "step": 54600 - }, - { - "epoch": 14.19, - "learning_rate": 2.86455630513752e-05, - "loss": 3.6635, - "step": 54700 - }, - { - "epoch": 14.22, - "learning_rate": 2.8624805396990144e-05, - "loss": 3.6364, - "step": 54800 - }, - { - "epoch": 14.24, - "learning_rate": 2.860404774260509e-05, - "loss": 3.6522, - "step": 54900 - }, - { - "epoch": 14.27, - "learning_rate": 2.8583290088220034e-05, - "loss": 3.6544, - "step": 55000 - }, - { - "epoch": 14.27, - "eval_gen_len": 12.2802, - "eval_loss": 3.628124713897705, - "eval_rouge1": 31.4632, - "eval_rouge2": 4.1625, - "eval_rougeL": 27.459, - "eval_rougeLsum": 27.4835, - "eval_runtime": 76.195, - "eval_samples_per_second": 50.581, - "eval_steps_per_second": 6.326, - "step": 55000 - }, - { - "epoch": 14.3, - "learning_rate": 2.856253243383498e-05, - "loss": 3.6345, - "step": 55100 - }, - { - "epoch": 14.32, - "learning_rate": 2.8541774779449927e-05, - "loss": 3.6543, - "step": 55200 - }, - { - "epoch": 14.35, - "learning_rate": 2.8521017125064872e-05, - "loss": 3.6405, - "step": 55300 - }, - { - "epoch": 14.37, - "learning_rate": 2.8500259470679817e-05, - "loss": 3.6421, - "step": 55400 - }, - { - "epoch": 14.4, - "learning_rate": 2.8479501816294762e-05, - "loss": 3.6653, - "step": 55500 - }, - { - "epoch": 14.43, - "learning_rate": 2.8458744161909707e-05, - "loss": 3.6695, - "step": 55600 - }, - { - "epoch": 14.45, - "learning_rate": 2.8437986507524652e-05, - "loss": 3.6273, - "step": 55700 - }, - { - "epoch": 14.48, - "learning_rate": 2.8417228853139597e-05, - "loss": 3.6402, - "step": 55800 - }, - { - "epoch": 14.5, - "learning_rate": 2.8396471198754545e-05, - "loss": 3.6508, - "step": 55900 - }, - { - "epoch": 14.53, - "learning_rate": 2.837571354436949e-05, - "loss": 3.6514, - "step": 56000 - }, - { - "epoch": 14.56, - "learning_rate": 2.8354955889984435e-05, - "loss": 3.6448, - "step": 56100 - }, - { - "epoch": 14.58, - "learning_rate": 2.833419823559938e-05, - "loss": 3.6388, - "step": 56200 - }, - { - "epoch": 14.61, - "learning_rate": 2.8313440581214325e-05, - "loss": 3.632, - "step": 56300 - }, - { - "epoch": 14.63, - "learning_rate": 2.829268292682927e-05, - "loss": 3.6273, - "step": 56400 - }, - { - "epoch": 14.66, - "learning_rate": 2.8271925272444215e-05, - "loss": 3.6467, - "step": 56500 - }, - { - "epoch": 14.69, - "learning_rate": 2.8251167618059163e-05, - "loss": 3.6213, - "step": 56600 - }, - { - "epoch": 14.71, - "learning_rate": 2.823040996367411e-05, - "loss": 3.6341, - "step": 56700 - }, - { - "epoch": 14.74, - "learning_rate": 2.8209652309289053e-05, - "loss": 3.6443, - "step": 56800 - }, - { - "epoch": 14.76, - "learning_rate": 2.8188894654904e-05, - "loss": 3.6283, - "step": 56900 - }, - { - "epoch": 14.79, - "learning_rate": 2.8168137000518943e-05, - "loss": 3.6681, - "step": 57000 - }, - { - "epoch": 14.82, - "learning_rate": 2.8147379346133888e-05, - "loss": 3.6549, - "step": 57100 - }, - { - "epoch": 14.84, - "learning_rate": 2.8126621691748837e-05, - "loss": 3.6512, - "step": 57200 - }, - { - "epoch": 14.87, - "learning_rate": 2.810586403736378e-05, - "loss": 3.6574, - "step": 57300 - }, - { - "epoch": 14.89, - "learning_rate": 2.8085106382978727e-05, - "loss": 3.6357, - "step": 57400 - }, - { - "epoch": 14.92, - "learning_rate": 2.806434872859367e-05, - "loss": 3.6537, - "step": 57500 - }, - { - "epoch": 14.95, - "learning_rate": 2.8043591074208617e-05, - "loss": 3.6414, - "step": 57600 - }, - { - "epoch": 14.97, - "learning_rate": 2.802283341982356e-05, - "loss": 3.6496, - "step": 57700 - }, - { - "epoch": 15.0, - "learning_rate": 2.8002075765438506e-05, - "loss": 3.6444, - "step": 57800 - }, - { - "epoch": 15.02, - "learning_rate": 2.7981318111053455e-05, - "loss": 3.652, - "step": 57900 - }, - { - "epoch": 15.05, - "learning_rate": 2.79605604566684e-05, - "loss": 3.6259, - "step": 58000 - }, - { - "epoch": 15.08, - "learning_rate": 2.7939802802283345e-05, - "loss": 3.6478, - "step": 58100 - }, - { - "epoch": 15.1, - "learning_rate": 2.791904514789829e-05, - "loss": 3.6543, - "step": 58200 - }, - { - "epoch": 15.13, - "learning_rate": 2.7898287493513235e-05, - "loss": 3.6468, - "step": 58300 - }, - { - "epoch": 15.15, - "learning_rate": 2.787752983912818e-05, - "loss": 3.6576, - "step": 58400 - }, - { - "epoch": 15.18, - "learning_rate": 2.7856772184743128e-05, - "loss": 3.6458, - "step": 58500 - }, - { - "epoch": 15.2, - "learning_rate": 2.7836014530358073e-05, - "loss": 3.6502, - "step": 58600 - }, - { - "epoch": 15.23, - "learning_rate": 2.7815256875973018e-05, - "loss": 3.6211, - "step": 58700 - }, - { - "epoch": 15.26, - "learning_rate": 2.7794499221587963e-05, - "loss": 3.645, - "step": 58800 - }, - { - "epoch": 15.28, - "learning_rate": 2.7773741567202908e-05, - "loss": 3.6441, - "step": 58900 - }, - { - "epoch": 15.31, - "learning_rate": 2.7752983912817853e-05, - "loss": 3.6321, - "step": 59000 - }, - { - "epoch": 15.33, - "learning_rate": 2.7732226258432798e-05, - "loss": 3.6391, - "step": 59100 - }, - { - "epoch": 15.36, - "learning_rate": 2.7711468604047746e-05, - "loss": 3.6461, - "step": 59200 - }, - { - "epoch": 15.39, - "learning_rate": 2.769071094966269e-05, - "loss": 3.619, - "step": 59300 - }, - { - "epoch": 15.41, - "learning_rate": 2.7669953295277636e-05, - "loss": 3.6244, - "step": 59400 - }, - { - "epoch": 15.44, - "learning_rate": 2.764919564089258e-05, - "loss": 3.664, - "step": 59500 - }, - { - "epoch": 15.46, - "learning_rate": 2.7628437986507526e-05, - "loss": 3.6522, - "step": 59600 - }, - { - "epoch": 15.49, - "learning_rate": 2.760768033212247e-05, - "loss": 3.6308, - "step": 59700 - }, - { - "epoch": 15.52, - "learning_rate": 2.758692267773742e-05, - "loss": 3.6505, - "step": 59800 - }, - { - "epoch": 15.54, - "learning_rate": 2.7566165023352364e-05, - "loss": 3.6457, - "step": 59900 - }, - { - "epoch": 15.57, - "learning_rate": 2.754540736896731e-05, - "loss": 3.6381, - "step": 60000 - }, - { - "epoch": 15.57, - "eval_gen_len": 12.2794, - "eval_loss": 3.6244375705718994, - "eval_rouge1": 34.0589, - "eval_rouge2": 5.2472, - "eval_rougeL": 30.0323, - "eval_rougeLsum": 30.0873, - "eval_runtime": 76.347, - "eval_samples_per_second": 50.48, - "eval_steps_per_second": 6.313, - "step": 60000 - }, - { - "epoch": 15.59, - "learning_rate": 2.7524649714582254e-05, - "loss": 3.6259, - "step": 60100 - }, - { - "epoch": 15.62, - "learning_rate": 2.75038920601972e-05, - "loss": 3.6488, - "step": 60200 - }, - { - "epoch": 15.65, - "learning_rate": 2.7483134405812144e-05, - "loss": 3.6415, - "step": 60300 - }, - { - "epoch": 15.67, - "learning_rate": 2.746237675142709e-05, - "loss": 3.6595, - "step": 60400 - }, - { - "epoch": 15.7, - "learning_rate": 2.7441619097042038e-05, - "loss": 3.636, - "step": 60500 - }, - { - "epoch": 15.72, - "learning_rate": 2.7420861442656982e-05, - "loss": 3.6107, - "step": 60600 - }, - { - "epoch": 15.75, - "learning_rate": 2.7400103788271927e-05, - "loss": 3.6465, - "step": 60700 - }, - { - "epoch": 15.78, - "learning_rate": 2.7379346133886872e-05, - "loss": 3.6097, - "step": 60800 - }, - { - "epoch": 15.8, - "learning_rate": 2.7358588479501817e-05, - "loss": 3.6584, - "step": 60900 - }, - { - "epoch": 15.83, - "learning_rate": 2.7337830825116762e-05, - "loss": 3.6508, - "step": 61000 - }, - { - "epoch": 15.85, - "learning_rate": 2.731707317073171e-05, - "loss": 3.6474, - "step": 61100 - }, - { - "epoch": 15.88, - "learning_rate": 2.7296315516346656e-05, - "loss": 3.5899, - "step": 61200 - }, - { - "epoch": 15.91, - "learning_rate": 2.72755578619616e-05, - "loss": 3.624, - "step": 61300 - }, - { - "epoch": 15.93, - "learning_rate": 2.7254800207576546e-05, - "loss": 3.6318, - "step": 61400 - }, - { - "epoch": 15.96, - "learning_rate": 2.723404255319149e-05, - "loss": 3.6407, - "step": 61500 - }, - { - "epoch": 15.98, - "learning_rate": 2.7213284898806436e-05, - "loss": 3.6605, - "step": 61600 - }, - { - "epoch": 16.01, - "learning_rate": 2.719252724442138e-05, - "loss": 3.6452, - "step": 61700 - }, - { - "epoch": 16.04, - "learning_rate": 2.717176959003633e-05, - "loss": 3.646, - "step": 61800 - }, - { - "epoch": 16.06, - "learning_rate": 2.7151011935651274e-05, - "loss": 3.6475, - "step": 61900 - }, - { - "epoch": 16.09, - "learning_rate": 2.713025428126622e-05, - "loss": 3.6483, - "step": 62000 - }, - { - "epoch": 16.11, - "learning_rate": 2.7109496626881164e-05, - "loss": 3.6051, - "step": 62100 - }, - { - "epoch": 16.14, - "learning_rate": 2.708873897249611e-05, - "loss": 3.6369, - "step": 62200 - }, - { - "epoch": 16.17, - "learning_rate": 2.7067981318111054e-05, - "loss": 3.6364, - "step": 62300 - }, - { - "epoch": 16.19, - "learning_rate": 2.7047223663726002e-05, - "loss": 3.6398, - "step": 62400 - }, - { - "epoch": 16.22, - "learning_rate": 2.7026466009340947e-05, - "loss": 3.6348, - "step": 62500 - }, - { - "epoch": 16.24, - "learning_rate": 2.7005708354955892e-05, - "loss": 3.6484, - "step": 62600 - }, - { - "epoch": 16.27, - "learning_rate": 2.6984950700570837e-05, - "loss": 3.6168, - "step": 62700 - }, - { - "epoch": 16.29, - "learning_rate": 2.6964193046185782e-05, - "loss": 3.6328, - "step": 62800 - }, - { - "epoch": 16.32, - "learning_rate": 2.6943435391800727e-05, - "loss": 3.6456, - "step": 62900 - }, - { - "epoch": 16.35, - "learning_rate": 2.6922677737415672e-05, - "loss": 3.6292, - "step": 63000 - }, - { - "epoch": 16.37, - "learning_rate": 2.690192008303062e-05, - "loss": 3.6523, - "step": 63100 - }, - { - "epoch": 16.4, - "learning_rate": 2.6881162428645565e-05, - "loss": 3.6296, - "step": 63200 - }, - { - "epoch": 16.42, - "learning_rate": 2.686040477426051e-05, - "loss": 3.6355, - "step": 63300 - }, - { - "epoch": 16.45, - "learning_rate": 2.6839647119875455e-05, - "loss": 3.627, - "step": 63400 - }, - { - "epoch": 16.48, - "learning_rate": 2.68188894654904e-05, - "loss": 3.6334, - "step": 63500 - }, - { - "epoch": 16.5, - "learning_rate": 2.6798131811105345e-05, - "loss": 3.6365, - "step": 63600 - }, - { - "epoch": 16.53, - "learning_rate": 2.6777374156720293e-05, - "loss": 3.6399, - "step": 63700 - }, - { - "epoch": 16.55, - "learning_rate": 2.675661650233524e-05, - "loss": 3.6597, - "step": 63800 - }, - { - "epoch": 16.58, - "learning_rate": 2.6735858847950183e-05, - "loss": 3.6349, - "step": 63900 - }, - { - "epoch": 16.61, - "learning_rate": 2.671510119356513e-05, - "loss": 3.6331, - "step": 64000 - }, - { - "epoch": 16.63, - "learning_rate": 2.6694343539180073e-05, - "loss": 3.6382, - "step": 64100 - }, - { - "epoch": 16.66, - "learning_rate": 2.6673585884795018e-05, - "loss": 3.6401, - "step": 64200 - }, - { - "epoch": 16.68, - "learning_rate": 2.6652828230409963e-05, - "loss": 3.6546, - "step": 64300 - }, - { - "epoch": 16.71, - "learning_rate": 2.663207057602491e-05, - "loss": 3.6492, - "step": 64400 - }, - { - "epoch": 16.74, - "learning_rate": 2.6611312921639857e-05, - "loss": 3.6681, - "step": 64500 - }, - { - "epoch": 16.76, - "learning_rate": 2.65905552672548e-05, - "loss": 3.6566, - "step": 64600 - }, - { - "epoch": 16.79, - "learning_rate": 2.6569797612869746e-05, - "loss": 3.6528, - "step": 64700 - }, - { - "epoch": 16.81, - "learning_rate": 2.654903995848469e-05, - "loss": 3.6492, - "step": 64800 - }, - { - "epoch": 16.84, - "learning_rate": 2.6528282304099636e-05, - "loss": 3.6372, - "step": 64900 - }, - { - "epoch": 16.87, - "learning_rate": 2.6507524649714588e-05, - "loss": 3.63, - "step": 65000 - }, - { - "epoch": 16.87, - "eval_gen_len": 12.2807, - "eval_loss": 3.6208901405334473, - "eval_rouge1": 35.2903, - "eval_rouge2": 5.4605, - "eval_rougeL": 31.2651, - "eval_rougeLsum": 31.3001, - "eval_runtime": 76.0126, - "eval_samples_per_second": 50.702, - "eval_steps_per_second": 6.341, - "step": 65000 - }, - { - "epoch": 16.89, - "learning_rate": 2.648676699532953e-05, - "loss": 3.6417, - "step": 65100 - }, - { - "epoch": 16.92, - "learning_rate": 2.6466009340944475e-05, - "loss": 3.6349, - "step": 65200 - }, - { - "epoch": 16.94, - "learning_rate": 2.644525168655942e-05, - "loss": 3.6443, - "step": 65300 - }, - { - "epoch": 16.97, - "learning_rate": 2.6424494032174365e-05, - "loss": 3.639, - "step": 65400 - }, - { - "epoch": 17.0, - "learning_rate": 2.640373637778931e-05, - "loss": 3.6413, - "step": 65500 - }, - { - "epoch": 17.02, - "learning_rate": 2.6382978723404255e-05, - "loss": 3.646, - "step": 65600 - }, - { - "epoch": 17.05, - "learning_rate": 2.6362221069019206e-05, - "loss": 3.6391, - "step": 65700 - }, - { - "epoch": 17.07, - "learning_rate": 2.634146341463415e-05, - "loss": 3.6542, - "step": 65800 - }, - { - "epoch": 17.1, - "learning_rate": 2.6320705760249093e-05, - "loss": 3.6401, - "step": 65900 - }, - { - "epoch": 17.13, - "learning_rate": 2.6299948105864038e-05, - "loss": 3.6508, - "step": 66000 - }, - { - "epoch": 17.15, - "learning_rate": 2.6279190451478983e-05, - "loss": 3.6418, - "step": 66100 - }, - { - "epoch": 17.18, - "learning_rate": 2.6258432797093928e-05, - "loss": 3.6355, - "step": 66200 - }, - { - "epoch": 17.2, - "learning_rate": 2.6237675142708873e-05, - "loss": 3.628, - "step": 66300 - }, - { - "epoch": 17.23, - "learning_rate": 2.6216917488323825e-05, - "loss": 3.6268, - "step": 66400 - }, - { - "epoch": 17.25, - "learning_rate": 2.619615983393877e-05, - "loss": 3.6455, - "step": 66500 - }, - { - "epoch": 17.28, - "learning_rate": 2.6175402179553714e-05, - "loss": 3.6476, - "step": 66600 - }, - { - "epoch": 17.31, - "learning_rate": 2.6154644525168656e-05, - "loss": 3.62, - "step": 66700 - }, - { - "epoch": 17.33, - "learning_rate": 2.61338868707836e-05, - "loss": 3.6338, - "step": 66800 - }, - { - "epoch": 17.36, - "learning_rate": 2.6113129216398546e-05, - "loss": 3.6408, - "step": 66900 - }, - { - "epoch": 17.38, - "learning_rate": 2.6092371562013498e-05, - "loss": 3.6332, - "step": 67000 - }, - { - "epoch": 17.41, - "learning_rate": 2.6071613907628443e-05, - "loss": 3.6388, - "step": 67100 - }, - { - "epoch": 17.44, - "learning_rate": 2.6050856253243388e-05, - "loss": 3.6309, - "step": 67200 - }, - { - "epoch": 17.46, - "learning_rate": 2.6030098598858333e-05, - "loss": 3.6142, - "step": 67300 - }, - { - "epoch": 17.49, - "learning_rate": 2.6009340944473274e-05, - "loss": 3.6209, - "step": 67400 - }, - { - "epoch": 17.51, - "learning_rate": 2.598858329008822e-05, - "loss": 3.6563, - "step": 67500 - }, - { - "epoch": 17.54, - "learning_rate": 2.5967825635703164e-05, - "loss": 3.6453, - "step": 67600 - }, - { - "epoch": 17.57, - "learning_rate": 2.5947067981318116e-05, - "loss": 3.6369, - "step": 67700 - }, - { - "epoch": 17.59, - "learning_rate": 2.592631032693306e-05, - "loss": 3.6374, - "step": 67800 - }, - { - "epoch": 17.62, - "learning_rate": 2.5905552672548006e-05, - "loss": 3.6093, - "step": 67900 - }, - { - "epoch": 17.64, - "learning_rate": 2.588479501816295e-05, - "loss": 3.6268, - "step": 68000 - }, - { - "epoch": 17.67, - "learning_rate": 2.5864037363777896e-05, - "loss": 3.6367, - "step": 68100 - }, - { - "epoch": 17.7, - "learning_rate": 2.5843279709392837e-05, - "loss": 3.6321, - "step": 68200 - }, - { - "epoch": 17.72, - "learning_rate": 2.582252205500779e-05, - "loss": 3.6407, - "step": 68300 - }, - { - "epoch": 17.75, - "learning_rate": 2.5801764400622734e-05, - "loss": 3.643, - "step": 68400 - }, - { - "epoch": 17.77, - "learning_rate": 2.578100674623768e-05, - "loss": 3.6505, - "step": 68500 - }, - { - "epoch": 17.8, - "learning_rate": 2.5760249091852624e-05, - "loss": 3.6214, - "step": 68600 - }, - { - "epoch": 17.83, - "learning_rate": 2.573949143746757e-05, - "loss": 3.6302, - "step": 68700 - }, - { - "epoch": 17.85, - "learning_rate": 2.5718733783082514e-05, - "loss": 3.6455, - "step": 68800 - }, - { - "epoch": 17.88, - "learning_rate": 2.5697976128697455e-05, - "loss": 3.6369, - "step": 68900 - }, - { - "epoch": 17.9, - "learning_rate": 2.5677218474312407e-05, - "loss": 3.6254, - "step": 69000 - }, - { - "epoch": 17.93, - "learning_rate": 2.5656460819927352e-05, - "loss": 3.6399, - "step": 69100 - }, - { - "epoch": 17.96, - "learning_rate": 2.5635703165542297e-05, - "loss": 3.6317, - "step": 69200 - }, - { - "epoch": 17.98, - "learning_rate": 2.5614945511157242e-05, - "loss": 3.6274, - "step": 69300 - }, - { - "epoch": 18.01, - "learning_rate": 2.5594187856772187e-05, - "loss": 3.6413, - "step": 69400 - }, - { - "epoch": 18.03, - "learning_rate": 2.5573430202387132e-05, - "loss": 3.5785, - "step": 69500 - }, - { - "epoch": 18.06, - "learning_rate": 2.555267254800208e-05, - "loss": 3.6545, - "step": 69600 - }, - { - "epoch": 18.09, - "learning_rate": 2.5531914893617025e-05, - "loss": 3.6421, - "step": 69700 - }, - { - "epoch": 18.11, - "learning_rate": 2.551115723923197e-05, - "loss": 3.6563, - "step": 69800 - }, - { - "epoch": 18.14, - "learning_rate": 2.5490399584846915e-05, - "loss": 3.6307, - "step": 69900 - }, - { - "epoch": 18.16, - "learning_rate": 2.546964193046186e-05, - "loss": 3.6466, - "step": 70000 - }, - { - "epoch": 18.16, - "eval_gen_len": 12.2802, - "eval_loss": 3.6190104484558105, - "eval_rouge1": 35.9674, - "eval_rouge2": 5.5141, - "eval_rougeL": 31.9434, - "eval_rougeLsum": 31.9335, - "eval_runtime": 75.2606, - "eval_samples_per_second": 51.209, - "eval_steps_per_second": 6.404, - "step": 70000 - }, - { - "epoch": 18.19, - "learning_rate": 2.5448884276076805e-05, - "loss": 3.6478, - "step": 70100 - }, - { - "epoch": 18.21, - "learning_rate": 2.542812662169175e-05, - "loss": 3.6251, - "step": 70200 - }, - { - "epoch": 18.24, - "learning_rate": 2.54073689673067e-05, - "loss": 3.6281, - "step": 70300 - }, - { - "epoch": 18.27, - "learning_rate": 2.5386611312921644e-05, - "loss": 3.6251, - "step": 70400 - }, - { - "epoch": 18.29, - "learning_rate": 2.536585365853659e-05, - "loss": 3.6252, - "step": 70500 - }, - { - "epoch": 18.32, - "learning_rate": 2.5345096004151533e-05, - "loss": 3.6214, - "step": 70600 - }, - { - "epoch": 18.34, - "learning_rate": 2.532433834976648e-05, - "loss": 3.6205, - "step": 70700 - }, - { - "epoch": 18.37, - "learning_rate": 2.5303580695381423e-05, - "loss": 3.6499, - "step": 70800 - }, - { - "epoch": 18.4, - "learning_rate": 2.5282823040996372e-05, - "loss": 3.6373, - "step": 70900 - }, - { - "epoch": 18.42, - "learning_rate": 2.5262065386611317e-05, - "loss": 3.6431, - "step": 71000 - }, - { - "epoch": 18.45, - "learning_rate": 2.5241307732226262e-05, - "loss": 3.6412, - "step": 71100 - }, - { - "epoch": 18.47, - "learning_rate": 2.5220550077841207e-05, - "loss": 3.6491, - "step": 71200 - }, - { - "epoch": 18.5, - "learning_rate": 2.519979242345615e-05, - "loss": 3.6163, - "step": 71300 - }, - { - "epoch": 18.53, - "learning_rate": 2.5179034769071097e-05, - "loss": 3.6328, - "step": 71400 - }, - { - "epoch": 18.55, - "learning_rate": 2.515827711468604e-05, - "loss": 3.623, - "step": 71500 - }, - { - "epoch": 18.58, - "learning_rate": 2.513751946030099e-05, - "loss": 3.6301, - "step": 71600 - }, - { - "epoch": 18.6, - "learning_rate": 2.5116761805915935e-05, - "loss": 3.6226, - "step": 71700 - }, - { - "epoch": 18.63, - "learning_rate": 2.509600415153088e-05, - "loss": 3.6297, - "step": 71800 - }, - { - "epoch": 18.66, - "learning_rate": 2.5075246497145825e-05, - "loss": 3.6523, - "step": 71900 - }, - { - "epoch": 18.68, - "learning_rate": 2.505448884276077e-05, - "loss": 3.6711, - "step": 72000 - }, - { - "epoch": 18.71, - "learning_rate": 2.5033731188375715e-05, - "loss": 3.6309, - "step": 72100 - }, - { - "epoch": 18.73, - "learning_rate": 2.5012973533990663e-05, - "loss": 3.6492, - "step": 72200 - }, - { - "epoch": 18.76, - "learning_rate": 2.4992215879605608e-05, - "loss": 3.6338, - "step": 72300 - }, - { - "epoch": 18.79, - "learning_rate": 2.4971458225220553e-05, - "loss": 3.6364, - "step": 72400 - }, - { - "epoch": 18.81, - "learning_rate": 2.4950700570835498e-05, - "loss": 3.634, - "step": 72500 - }, - { - "epoch": 18.84, - "learning_rate": 2.4929942916450443e-05, - "loss": 3.6298, - "step": 72600 - }, - { - "epoch": 18.86, - "learning_rate": 2.4909185262065388e-05, - "loss": 3.6512, - "step": 72700 - }, - { - "epoch": 18.89, - "learning_rate": 2.4888427607680333e-05, - "loss": 3.6493, - "step": 72800 - }, - { - "epoch": 18.92, - "learning_rate": 2.486766995329528e-05, - "loss": 3.6528, - "step": 72900 - }, - { - "epoch": 18.94, - "learning_rate": 2.4846912298910226e-05, - "loss": 3.6366, - "step": 73000 - }, - { - "epoch": 18.97, - "learning_rate": 2.482615464452517e-05, - "loss": 3.6066, - "step": 73100 - }, - { - "epoch": 18.99, - "learning_rate": 2.4805396990140116e-05, - "loss": 3.6273, - "step": 73200 - }, - { - "epoch": 19.02, - "learning_rate": 2.478463933575506e-05, - "loss": 3.6446, - "step": 73300 - }, - { - "epoch": 19.05, - "learning_rate": 2.4763881681370006e-05, - "loss": 3.6293, - "step": 73400 - }, - { - "epoch": 19.07, - "learning_rate": 2.4743124026984954e-05, - "loss": 3.6043, - "step": 73500 - }, - { - "epoch": 19.1, - "learning_rate": 2.47223663725999e-05, - "loss": 3.6257, - "step": 73600 - }, - { - "epoch": 19.12, - "learning_rate": 2.4701608718214844e-05, - "loss": 3.6326, - "step": 73700 - }, - { - "epoch": 19.15, - "learning_rate": 2.468085106382979e-05, - "loss": 3.6104, - "step": 73800 - }, - { - "epoch": 19.17, - "learning_rate": 2.4660093409444734e-05, - "loss": 3.6372, - "step": 73900 - }, - { - "epoch": 19.2, - "learning_rate": 2.463933575505968e-05, - "loss": 3.6068, - "step": 74000 - }, - { - "epoch": 19.23, - "learning_rate": 2.4618578100674624e-05, - "loss": 3.6266, - "step": 74100 - }, - { - "epoch": 19.25, - "learning_rate": 2.4597820446289573e-05, - "loss": 3.63, - "step": 74200 - }, - { - "epoch": 19.28, - "learning_rate": 2.4577062791904518e-05, - "loss": 3.6421, - "step": 74300 - }, - { - "epoch": 19.3, - "learning_rate": 2.4556305137519463e-05, - "loss": 3.6375, - "step": 74400 - }, - { - "epoch": 19.33, - "learning_rate": 2.4535547483134408e-05, - "loss": 3.6254, - "step": 74500 - }, - { - "epoch": 19.36, - "learning_rate": 2.4514789828749353e-05, - "loss": 3.6355, - "step": 74600 - }, - { - "epoch": 19.38, - "learning_rate": 2.4494032174364297e-05, - "loss": 3.6511, - "step": 74700 - }, - { - "epoch": 19.41, - "learning_rate": 2.4473274519979246e-05, - "loss": 3.6639, - "step": 74800 - }, - { - "epoch": 19.43, - "learning_rate": 2.445251686559419e-05, - "loss": 3.6405, - "step": 74900 - }, - { - "epoch": 19.46, - "learning_rate": 2.4431759211209136e-05, - "loss": 3.6307, - "step": 75000 - }, - { - "epoch": 19.46, - "eval_gen_len": 12.2797, - "eval_loss": 3.616529703140259, - "eval_rouge1": 37.1937, - "eval_rouge2": 6.2828, - "eval_rougeL": 33.1867, - "eval_rougeLsum": 33.1809, - "eval_runtime": 75.7672, - "eval_samples_per_second": 50.866, - "eval_steps_per_second": 6.362, - "step": 75000 - }, - { - "epoch": 19.49, - "learning_rate": 2.441100155682408e-05, - "loss": 3.6128, - "step": 75100 - }, - { - "epoch": 19.51, - "learning_rate": 2.4390243902439026e-05, - "loss": 3.6349, - "step": 75200 - }, - { - "epoch": 19.54, - "learning_rate": 2.436948624805397e-05, - "loss": 3.6422, - "step": 75300 - }, - { - "epoch": 19.56, - "learning_rate": 2.4348728593668916e-05, - "loss": 3.6391, - "step": 75400 - }, - { - "epoch": 19.59, - "learning_rate": 2.4327970939283864e-05, - "loss": 3.6343, - "step": 75500 - }, - { - "epoch": 19.62, - "learning_rate": 2.430721328489881e-05, - "loss": 3.6291, - "step": 75600 - }, - { - "epoch": 19.64, - "learning_rate": 2.4286455630513754e-05, - "loss": 3.6087, - "step": 75700 - }, - { - "epoch": 19.67, - "learning_rate": 2.42656979761287e-05, - "loss": 3.6324, - "step": 75800 - }, - { - "epoch": 19.69, - "learning_rate": 2.4244940321743644e-05, - "loss": 3.648, - "step": 75900 - }, - { - "epoch": 19.72, - "learning_rate": 2.422418266735859e-05, - "loss": 3.6319, - "step": 76000 - }, - { - "epoch": 19.75, - "learning_rate": 2.4203425012973534e-05, - "loss": 3.6331, - "step": 76100 - }, - { - "epoch": 19.77, - "learning_rate": 2.4182667358588482e-05, - "loss": 3.6391, - "step": 76200 - }, - { - "epoch": 19.8, - "learning_rate": 2.4161909704203427e-05, - "loss": 3.6544, - "step": 76300 - }, - { - "epoch": 19.82, - "learning_rate": 2.4141152049818372e-05, - "loss": 3.6293, - "step": 76400 - }, - { - "epoch": 19.85, - "learning_rate": 2.4120394395433317e-05, - "loss": 3.6295, - "step": 76500 - }, - { - "epoch": 19.88, - "learning_rate": 2.4099636741048262e-05, - "loss": 3.6216, - "step": 76600 - }, - { - "epoch": 19.9, - "learning_rate": 2.4078879086663207e-05, - "loss": 3.6345, - "step": 76700 - }, - { - "epoch": 19.93, - "learning_rate": 2.4058121432278155e-05, - "loss": 3.6347, - "step": 76800 - }, - { - "epoch": 19.95, - "learning_rate": 2.40373637778931e-05, - "loss": 3.6191, - "step": 76900 - }, - { - "epoch": 19.98, - "learning_rate": 2.4016606123508045e-05, - "loss": 3.6214, - "step": 77000 - }, - { - "epoch": 20.01, - "learning_rate": 2.399584846912299e-05, - "loss": 3.6456, - "step": 77100 - }, - { - "epoch": 20.03, - "learning_rate": 2.3975090814737935e-05, - "loss": 3.6474, - "step": 77200 - }, - { - "epoch": 20.06, - "learning_rate": 2.395433316035288e-05, - "loss": 3.6347, - "step": 77300 - }, - { - "epoch": 20.08, - "learning_rate": 2.3933575505967825e-05, - "loss": 3.626, - "step": 77400 - }, - { - "epoch": 20.11, - "learning_rate": 2.3912817851582774e-05, - "loss": 3.6323, - "step": 77500 - }, - { - "epoch": 20.13, - "learning_rate": 2.389206019719772e-05, - "loss": 3.6282, - "step": 77600 - }, - { - "epoch": 20.16, - "learning_rate": 2.3871302542812663e-05, - "loss": 3.6296, - "step": 77700 - }, - { - "epoch": 20.19, - "learning_rate": 2.385054488842761e-05, - "loss": 3.6191, - "step": 77800 - }, - { - "epoch": 20.21, - "learning_rate": 2.3829787234042553e-05, - "loss": 3.6191, - "step": 77900 - }, - { - "epoch": 20.24, - "learning_rate": 2.38090295796575e-05, - "loss": 3.6518, - "step": 78000 - }, - { - "epoch": 20.26, - "learning_rate": 2.3788271925272447e-05, - "loss": 3.6158, - "step": 78100 - }, - { - "epoch": 20.29, - "learning_rate": 2.3767514270887392e-05, - "loss": 3.5855, - "step": 78200 - }, - { - "epoch": 20.32, - "learning_rate": 2.3746756616502337e-05, - "loss": 3.6417, - "step": 78300 - }, - { - "epoch": 20.34, - "learning_rate": 2.372599896211728e-05, - "loss": 3.5877, - "step": 78400 - }, - { - "epoch": 20.37, - "learning_rate": 2.3705241307732227e-05, - "loss": 3.6276, - "step": 78500 - }, - { - "epoch": 20.39, - "learning_rate": 2.368448365334717e-05, - "loss": 3.6331, - "step": 78600 - }, - { - "epoch": 20.42, - "learning_rate": 2.3663725998962117e-05, - "loss": 3.6065, - "step": 78700 - }, - { - "epoch": 20.45, - "learning_rate": 2.3642968344577065e-05, - "loss": 3.6643, - "step": 78800 - }, - { - "epoch": 20.47, - "learning_rate": 2.362221069019201e-05, - "loss": 3.6337, - "step": 78900 - }, - { - "epoch": 20.5, - "learning_rate": 2.3601453035806955e-05, - "loss": 3.6463, - "step": 79000 - }, - { - "epoch": 20.52, - "learning_rate": 2.35806953814219e-05, - "loss": 3.646, - "step": 79100 - }, - { - "epoch": 20.55, - "learning_rate": 2.3559937727036845e-05, - "loss": 3.6284, - "step": 79200 - }, - { - "epoch": 20.58, - "learning_rate": 2.353918007265179e-05, - "loss": 3.6241, - "step": 79300 - }, - { - "epoch": 20.6, - "learning_rate": 2.3518422418266738e-05, - "loss": 3.6432, - "step": 79400 - }, - { - "epoch": 20.63, - "learning_rate": 2.3497664763881683e-05, - "loss": 3.5956, - "step": 79500 - }, - { - "epoch": 20.65, - "learning_rate": 2.3476907109496628e-05, - "loss": 3.6143, - "step": 79600 - }, - { - "epoch": 20.68, - "learning_rate": 2.3456149455111573e-05, - "loss": 3.6383, - "step": 79700 - }, - { - "epoch": 20.71, - "learning_rate": 2.3435391800726518e-05, - "loss": 3.6151, - "step": 79800 - }, - { - "epoch": 20.73, - "learning_rate": 2.3414634146341463e-05, - "loss": 3.6192, - "step": 79900 - }, - { - "epoch": 20.76, - "learning_rate": 2.3393876491956408e-05, - "loss": 3.6267, - "step": 80000 - }, - { - "epoch": 20.76, - "eval_gen_len": 12.2774, - "eval_loss": 3.6140060424804688, - "eval_rouge1": 37.7966, - "eval_rouge2": 6.6257, - "eval_rougeL": 33.7751, - "eval_rougeLsum": 33.8284, - "eval_runtime": 75.5711, - "eval_samples_per_second": 50.998, - "eval_steps_per_second": 6.378, - "step": 80000 - }, - { - "epoch": 20.78, - "learning_rate": 2.337311883757136e-05, - "loss": 3.6422, - "step": 80100 - }, - { - "epoch": 20.81, - "learning_rate": 2.33523611831863e-05, - "loss": 3.6305, - "step": 80200 - }, - { - "epoch": 20.84, - "learning_rate": 2.3331603528801246e-05, - "loss": 3.6263, - "step": 80300 - }, - { - "epoch": 20.86, - "learning_rate": 2.331084587441619e-05, - "loss": 3.6409, - "step": 80400 - }, - { - "epoch": 20.89, - "learning_rate": 2.3290088220031136e-05, - "loss": 3.6349, - "step": 80500 - }, - { - "epoch": 20.91, - "learning_rate": 2.326933056564608e-05, - "loss": 3.6293, - "step": 80600 - }, - { - "epoch": 20.94, - "learning_rate": 2.3248572911261033e-05, - "loss": 3.6439, - "step": 80700 - }, - { - "epoch": 20.97, - "learning_rate": 2.3227815256875978e-05, - "loss": 3.6455, - "step": 80800 - }, - { - "epoch": 20.99, - "learning_rate": 2.3207057602490923e-05, - "loss": 3.6419, - "step": 80900 - }, - { - "epoch": 21.02, - "learning_rate": 2.3186299948105864e-05, - "loss": 3.6248, - "step": 81000 - }, - { - "epoch": 21.04, - "learning_rate": 2.316554229372081e-05, - "loss": 3.6155, - "step": 81100 - }, - { - "epoch": 21.07, - "learning_rate": 2.3144784639335754e-05, - "loss": 3.5999, - "step": 81200 - }, - { - "epoch": 21.09, - "learning_rate": 2.31240269849507e-05, - "loss": 3.6312, - "step": 81300 - }, - { - "epoch": 21.12, - "learning_rate": 2.310326933056565e-05, - "loss": 3.6309, - "step": 81400 - }, - { - "epoch": 21.15, - "learning_rate": 2.3082511676180596e-05, - "loss": 3.6292, - "step": 81500 - }, - { - "epoch": 21.17, - "learning_rate": 2.306175402179554e-05, - "loss": 3.6346, - "step": 81600 - }, - { - "epoch": 21.2, - "learning_rate": 2.3040996367410483e-05, - "loss": 3.6325, - "step": 81700 - }, - { - "epoch": 21.22, - "learning_rate": 2.3020238713025427e-05, - "loss": 3.6513, - "step": 81800 - }, - { - "epoch": 21.25, - "learning_rate": 2.2999481058640372e-05, - "loss": 3.618, - "step": 81900 - }, - { - "epoch": 21.28, - "learning_rate": 2.2978723404255324e-05, - "loss": 3.6536, - "step": 82000 - }, - { - "epoch": 21.3, - "learning_rate": 2.295796574987027e-05, - "loss": 3.6265, - "step": 82100 - }, - { - "epoch": 21.33, - "learning_rate": 2.2937208095485214e-05, - "loss": 3.6115, - "step": 82200 - }, - { - "epoch": 21.35, - "learning_rate": 2.291645044110016e-05, - "loss": 3.6246, - "step": 82300 - }, - { - "epoch": 21.38, - "learning_rate": 2.2895692786715104e-05, - "loss": 3.6248, - "step": 82400 - }, - { - "epoch": 21.41, - "learning_rate": 2.2874935132330046e-05, - "loss": 3.6398, - "step": 82500 - }, - { - "epoch": 21.43, - "learning_rate": 2.285417747794499e-05, - "loss": 3.6227, - "step": 82600 - }, - { - "epoch": 21.46, - "learning_rate": 2.2833419823559942e-05, - "loss": 3.6327, - "step": 82700 - }, - { - "epoch": 21.48, - "learning_rate": 2.2812662169174887e-05, - "loss": 3.6141, - "step": 82800 - }, - { - "epoch": 21.51, - "learning_rate": 2.2791904514789832e-05, - "loss": 3.6388, - "step": 82900 - }, - { - "epoch": 21.54, - "learning_rate": 2.2771146860404777e-05, - "loss": 3.5972, - "step": 83000 - }, - { - "epoch": 21.56, - "learning_rate": 2.2750389206019722e-05, - "loss": 3.6391, - "step": 83100 - }, - { - "epoch": 21.59, - "learning_rate": 2.2729631551634664e-05, - "loss": 3.6326, - "step": 83200 - }, - { - "epoch": 21.61, - "learning_rate": 2.2708873897249616e-05, - "loss": 3.6511, - "step": 83300 - }, - { - "epoch": 21.64, - "learning_rate": 2.268811624286456e-05, - "loss": 3.6438, - "step": 83400 - }, - { - "epoch": 21.67, - "learning_rate": 2.2667358588479505e-05, - "loss": 3.6262, - "step": 83500 - }, - { - "epoch": 21.69, - "learning_rate": 2.264660093409445e-05, - "loss": 3.6197, - "step": 83600 - }, - { - "epoch": 21.72, - "learning_rate": 2.2625843279709395e-05, - "loss": 3.6237, - "step": 83700 - }, - { - "epoch": 21.74, - "learning_rate": 2.260508562532434e-05, - "loss": 3.6164, - "step": 83800 - }, - { - "epoch": 21.77, - "learning_rate": 2.2584327970939285e-05, - "loss": 3.6177, - "step": 83900 - }, - { - "epoch": 21.8, - "learning_rate": 2.2563570316554234e-05, - "loss": 3.6335, - "step": 84000 - }, - { - "epoch": 21.82, - "learning_rate": 2.254281266216918e-05, - "loss": 3.6037, - "step": 84100 - }, - { - "epoch": 21.85, - "learning_rate": 2.2522055007784124e-05, - "loss": 3.6256, - "step": 84200 - }, - { - "epoch": 21.87, - "learning_rate": 2.250129735339907e-05, - "loss": 3.6176, - "step": 84300 - }, - { - "epoch": 21.9, - "learning_rate": 2.2480539699014014e-05, - "loss": 3.638, - "step": 84400 - }, - { - "epoch": 21.93, - "learning_rate": 2.245978204462896e-05, - "loss": 3.6471, - "step": 84500 - }, - { - "epoch": 21.95, - "learning_rate": 2.2439024390243907e-05, - "loss": 3.6242, - "step": 84600 - }, - { - "epoch": 21.98, - "learning_rate": 2.2418266735858852e-05, - "loss": 3.6208, - "step": 84700 - }, - { - "epoch": 22.0, - "learning_rate": 2.2397509081473797e-05, - "loss": 3.5985, - "step": 84800 - }, - { - "epoch": 22.03, - "learning_rate": 2.2376751427088742e-05, - "loss": 3.6171, - "step": 84900 - }, - { - "epoch": 22.06, - "learning_rate": 2.2355993772703687e-05, - "loss": 3.6323, - "step": 85000 - }, - { - "epoch": 22.06, - "eval_gen_len": 12.2784, - "eval_loss": 3.6118929386138916, - "eval_rouge1": 38.1155, - "eval_rouge2": 7.0586, - "eval_rougeL": 34.0831, - "eval_rougeLsum": 34.1346, - "eval_runtime": 76.1382, - "eval_samples_per_second": 50.618, - "eval_steps_per_second": 6.331, - "step": 85000 - }, - { - "epoch": 22.08, - "learning_rate": 2.2335236118318632e-05, - "loss": 3.6053, - "step": 85100 - }, - { - "epoch": 22.11, - "learning_rate": 2.2314478463933577e-05, - "loss": 3.6152, - "step": 85200 - }, - { - "epoch": 22.13, - "learning_rate": 2.2293720809548525e-05, - "loss": 3.6379, - "step": 85300 - }, - { - "epoch": 22.16, - "learning_rate": 2.227296315516347e-05, - "loss": 3.6244, - "step": 85400 - }, - { - "epoch": 22.18, - "learning_rate": 2.2252205500778415e-05, - "loss": 3.6135, - "step": 85500 - }, - { - "epoch": 22.21, - "learning_rate": 2.223144784639336e-05, - "loss": 3.5963, - "step": 85600 - }, - { - "epoch": 22.24, - "learning_rate": 2.2210690192008305e-05, - "loss": 3.6271, - "step": 85700 - }, - { - "epoch": 22.26, - "learning_rate": 2.218993253762325e-05, - "loss": 3.6101, - "step": 85800 - }, - { - "epoch": 22.29, - "learning_rate": 2.2169174883238198e-05, - "loss": 3.6205, - "step": 85900 - }, - { - "epoch": 22.31, - "learning_rate": 2.2148417228853143e-05, - "loss": 3.6117, - "step": 86000 - }, - { - "epoch": 22.34, - "learning_rate": 2.2127659574468088e-05, - "loss": 3.6298, - "step": 86100 - }, - { - "epoch": 22.37, - "learning_rate": 2.2106901920083033e-05, - "loss": 3.6262, - "step": 86200 - }, - { - "epoch": 22.39, - "learning_rate": 2.2086144265697978e-05, - "loss": 3.6343, - "step": 86300 - }, - { - "epoch": 22.42, - "learning_rate": 2.2065386611312923e-05, - "loss": 3.6137, - "step": 86400 - }, - { - "epoch": 22.44, - "learning_rate": 2.2044628956927868e-05, - "loss": 3.6192, - "step": 86500 - }, - { - "epoch": 22.47, - "learning_rate": 2.2023871302542816e-05, - "loss": 3.6328, - "step": 86600 - }, - { - "epoch": 22.5, - "learning_rate": 2.200311364815776e-05, - "loss": 3.6345, - "step": 86700 - }, - { - "epoch": 22.52, - "learning_rate": 2.1982355993772706e-05, - "loss": 3.6275, - "step": 86800 - }, - { - "epoch": 22.55, - "learning_rate": 2.196159833938765e-05, - "loss": 3.6173, - "step": 86900 - }, - { - "epoch": 22.57, - "learning_rate": 2.1940840685002596e-05, - "loss": 3.654, - "step": 87000 - }, - { - "epoch": 22.6, - "learning_rate": 2.192008303061754e-05, - "loss": 3.6459, - "step": 87100 - }, - { - "epoch": 22.63, - "learning_rate": 2.1899325376232486e-05, - "loss": 3.6171, - "step": 87200 - }, - { - "epoch": 22.65, - "learning_rate": 2.1878567721847435e-05, - "loss": 3.6092, - "step": 87300 - }, - { - "epoch": 22.68, - "learning_rate": 2.185781006746238e-05, - "loss": 3.6502, - "step": 87400 - }, - { - "epoch": 22.7, - "learning_rate": 2.1837052413077325e-05, - "loss": 3.6308, - "step": 87500 - }, - { - "epoch": 22.73, - "learning_rate": 2.181629475869227e-05, - "loss": 3.6208, - "step": 87600 - }, - { - "epoch": 22.76, - "learning_rate": 2.1795537104307214e-05, - "loss": 3.6293, - "step": 87700 - }, - { - "epoch": 22.78, - "learning_rate": 2.177477944992216e-05, - "loss": 3.6228, - "step": 87800 - }, - { - "epoch": 22.81, - "learning_rate": 2.1754021795537108e-05, - "loss": 3.6328, - "step": 87900 - }, - { - "epoch": 22.83, - "learning_rate": 2.1733264141152053e-05, - "loss": 3.6413, - "step": 88000 - }, - { - "epoch": 22.86, - "learning_rate": 2.1712506486766998e-05, - "loss": 3.623, - "step": 88100 - }, - { - "epoch": 22.89, - "learning_rate": 2.1691748832381943e-05, - "loss": 3.6183, - "step": 88200 - }, - { - "epoch": 22.91, - "learning_rate": 2.1670991177996888e-05, - "loss": 3.6458, - "step": 88300 - }, - { - "epoch": 22.94, - "learning_rate": 2.1650233523611833e-05, - "loss": 3.6135, - "step": 88400 - }, - { - "epoch": 22.96, - "learning_rate": 2.1629475869226778e-05, - "loss": 3.6387, - "step": 88500 - }, - { - "epoch": 22.99, - "learning_rate": 2.1608718214841726e-05, - "loss": 3.6328, - "step": 88600 - }, - { - "epoch": 23.02, - "learning_rate": 2.158796056045667e-05, - "loss": 3.642, - "step": 88700 - }, - { - "epoch": 23.04, - "learning_rate": 2.1567202906071616e-05, - "loss": 3.6133, - "step": 88800 - }, - { - "epoch": 23.07, - "learning_rate": 2.154644525168656e-05, - "loss": 3.6341, - "step": 88900 - }, - { - "epoch": 23.09, - "learning_rate": 2.1525687597301506e-05, - "loss": 3.6534, - "step": 89000 - }, - { - "epoch": 23.12, - "learning_rate": 2.150492994291645e-05, - "loss": 3.6197, - "step": 89100 - }, - { - "epoch": 23.14, - "learning_rate": 2.14841722885314e-05, - "loss": 3.6146, - "step": 89200 - }, - { - "epoch": 23.17, - "learning_rate": 2.1463414634146344e-05, - "loss": 3.619, - "step": 89300 - }, - { - "epoch": 23.2, - "learning_rate": 2.144265697976129e-05, - "loss": 3.6136, - "step": 89400 - }, - { - "epoch": 23.22, - "learning_rate": 2.1421899325376234e-05, - "loss": 3.6261, - "step": 89500 - }, - { - "epoch": 23.25, - "learning_rate": 2.140114167099118e-05, - "loss": 3.6204, - "step": 89600 - }, - { - "epoch": 23.27, - "learning_rate": 2.1380384016606124e-05, - "loss": 3.6304, - "step": 89700 - }, - { - "epoch": 23.3, - "learning_rate": 2.135962636222107e-05, - "loss": 3.6339, - "step": 89800 - }, - { - "epoch": 23.33, - "learning_rate": 2.1338868707836017e-05, - "loss": 3.643, - "step": 89900 - }, - { - "epoch": 23.35, - "learning_rate": 2.1318111053450962e-05, - "loss": 3.6424, - "step": 90000 - }, - { - "epoch": 23.35, - "eval_gen_len": 12.2787, - "eval_loss": 3.610205888748169, - "eval_rouge1": 38.4968, - "eval_rouge2": 7.1642, - "eval_rougeL": 34.4804, - "eval_rougeLsum": 34.4687, - "eval_runtime": 76.2319, - "eval_samples_per_second": 50.556, - "eval_steps_per_second": 6.323, - "step": 90000 - }, - { - "epoch": 23.38, - "learning_rate": 2.1297353399065907e-05, - "loss": 3.6358, - "step": 90100 - }, - { - "epoch": 23.4, - "learning_rate": 2.1276595744680852e-05, - "loss": 3.6221, - "step": 90200 - }, - { - "epoch": 23.43, - "learning_rate": 2.1255838090295797e-05, - "loss": 3.6116, - "step": 90300 - }, - { - "epoch": 23.46, - "learning_rate": 2.1235080435910742e-05, - "loss": 3.6152, - "step": 90400 - }, - { - "epoch": 23.48, - "learning_rate": 2.121432278152569e-05, - "loss": 3.6206, - "step": 90500 - }, - { - "epoch": 23.51, - "learning_rate": 2.1193565127140635e-05, - "loss": 3.6349, - "step": 90600 - }, - { - "epoch": 23.53, - "learning_rate": 2.117280747275558e-05, - "loss": 3.616, - "step": 90700 - }, - { - "epoch": 23.56, - "learning_rate": 2.1152049818370525e-05, - "loss": 3.6248, - "step": 90800 - }, - { - "epoch": 23.59, - "learning_rate": 2.113129216398547e-05, - "loss": 3.6108, - "step": 90900 - }, - { - "epoch": 23.61, - "learning_rate": 2.1110534509600415e-05, - "loss": 3.6109, - "step": 91000 - }, - { - "epoch": 23.64, - "learning_rate": 2.108977685521536e-05, - "loss": 3.6205, - "step": 91100 - }, - { - "epoch": 23.66, - "learning_rate": 2.106901920083031e-05, - "loss": 3.616, - "step": 91200 - }, - { - "epoch": 23.69, - "learning_rate": 2.1048261546445254e-05, - "loss": 3.6276, - "step": 91300 - }, - { - "epoch": 23.72, - "learning_rate": 2.10275038920602e-05, - "loss": 3.6425, - "step": 91400 - }, - { - "epoch": 23.74, - "learning_rate": 2.1006746237675144e-05, - "loss": 3.6308, - "step": 91500 - }, - { - "epoch": 23.77, - "learning_rate": 2.098598858329009e-05, - "loss": 3.6103, - "step": 91600 - }, - { - "epoch": 23.79, - "learning_rate": 2.0965230928905034e-05, - "loss": 3.6167, - "step": 91700 - }, - { - "epoch": 23.82, - "learning_rate": 2.0944473274519982e-05, - "loss": 3.6349, - "step": 91800 - }, - { - "epoch": 23.85, - "learning_rate": 2.0923715620134927e-05, - "loss": 3.6212, - "step": 91900 - }, - { - "epoch": 23.87, - "learning_rate": 2.0902957965749872e-05, - "loss": 3.6271, - "step": 92000 - }, - { - "epoch": 23.9, - "learning_rate": 2.0882200311364817e-05, - "loss": 3.6428, - "step": 92100 - }, - { - "epoch": 23.92, - "learning_rate": 2.0861442656979762e-05, - "loss": 3.6327, - "step": 92200 - }, - { - "epoch": 23.95, - "learning_rate": 2.0840685002594707e-05, - "loss": 3.6309, - "step": 92300 - }, - { - "epoch": 23.98, - "learning_rate": 2.081992734820965e-05, - "loss": 3.6299, - "step": 92400 - }, - { - "epoch": 24.0, - "learning_rate": 2.07991696938246e-05, - "loss": 3.6317, - "step": 92500 - }, - { - "epoch": 24.03, - "learning_rate": 2.0778412039439545e-05, - "loss": 3.631, - "step": 92600 - }, - { - "epoch": 24.05, - "learning_rate": 2.075765438505449e-05, - "loss": 3.6313, - "step": 92700 - }, - { - "epoch": 24.08, - "learning_rate": 2.0736896730669435e-05, - "loss": 3.6339, - "step": 92800 - }, - { - "epoch": 24.1, - "learning_rate": 2.071613907628438e-05, - "loss": 3.627, - "step": 92900 - }, - { - "epoch": 24.13, - "learning_rate": 2.0695381421899325e-05, - "loss": 3.5829, - "step": 93000 - }, - { - "epoch": 24.16, - "learning_rate": 2.0674623767514273e-05, - "loss": 3.6276, - "step": 93100 - }, - { - "epoch": 24.18, - "learning_rate": 2.0653866113129218e-05, - "loss": 3.634, - "step": 93200 - }, - { - "epoch": 24.21, - "learning_rate": 2.0633108458744163e-05, - "loss": 3.6445, - "step": 93300 - }, - { - "epoch": 24.23, - "learning_rate": 2.0612350804359108e-05, - "loss": 3.6152, - "step": 93400 - }, - { - "epoch": 24.26, - "learning_rate": 2.0591593149974053e-05, - "loss": 3.6221, - "step": 93500 - }, - { - "epoch": 24.29, - "learning_rate": 2.0570835495588998e-05, - "loss": 3.6276, - "step": 93600 - }, - { - "epoch": 24.31, - "learning_rate": 2.0550077841203943e-05, - "loss": 3.6177, - "step": 93700 - }, - { - "epoch": 24.34, - "learning_rate": 2.052932018681889e-05, - "loss": 3.6295, - "step": 93800 - }, - { - "epoch": 24.36, - "learning_rate": 2.0508562532433836e-05, - "loss": 3.6136, - "step": 93900 - }, - { - "epoch": 24.39, - "learning_rate": 2.048780487804878e-05, - "loss": 3.6271, - "step": 94000 - }, - { - "epoch": 24.42, - "learning_rate": 2.0467047223663726e-05, - "loss": 3.6286, - "step": 94100 - }, - { - "epoch": 24.44, - "learning_rate": 2.044628956927867e-05, - "loss": 3.6246, - "step": 94200 - }, - { - "epoch": 24.47, - "learning_rate": 2.0425531914893616e-05, - "loss": 3.6044, - "step": 94300 - }, - { - "epoch": 24.49, - "learning_rate": 2.0404774260508568e-05, - "loss": 3.6185, - "step": 94400 - }, - { - "epoch": 24.52, - "learning_rate": 2.038401660612351e-05, - "loss": 3.6121, - "step": 94500 - }, - { - "epoch": 24.55, - "learning_rate": 2.0363258951738455e-05, - "loss": 3.6212, - "step": 94600 - }, - { - "epoch": 24.57, - "learning_rate": 2.03425012973534e-05, - "loss": 3.656, - "step": 94700 - }, - { - "epoch": 24.6, - "learning_rate": 2.0321743642968344e-05, - "loss": 3.592, - "step": 94800 - }, - { - "epoch": 24.62, - "learning_rate": 2.030098598858329e-05, - "loss": 3.6093, - "step": 94900 - }, - { - "epoch": 24.65, - "learning_rate": 2.0280228334198234e-05, - "loss": 3.6256, - "step": 95000 - }, - { - "epoch": 24.65, - "eval_gen_len": 12.2776, - "eval_loss": 3.6076412200927734, - "eval_rouge1": 39.6155, - "eval_rouge2": 7.3309, - "eval_rougeL": 35.6135, - "eval_rougeLsum": 35.6554, - "eval_runtime": 76.2409, - "eval_samples_per_second": 50.55, - "eval_steps_per_second": 6.322, - "step": 95000 - }, - { - "epoch": 24.68, - "learning_rate": 2.0259470679813186e-05, - "loss": 3.6108, - "step": 95100 - }, - { - "epoch": 24.7, - "learning_rate": 2.023871302542813e-05, - "loss": 3.6275, - "step": 95200 - }, - { - "epoch": 24.73, - "learning_rate": 2.0217955371043073e-05, - "loss": 3.607, - "step": 95300 - }, - { - "epoch": 24.75, - "learning_rate": 2.0197197716658018e-05, - "loss": 3.6376, - "step": 95400 - }, - { - "epoch": 24.78, - "learning_rate": 2.0176440062272963e-05, - "loss": 3.6322, - "step": 95500 - }, - { - "epoch": 24.81, - "learning_rate": 2.0155682407887908e-05, - "loss": 3.6188, - "step": 95600 - }, - { - "epoch": 24.83, - "learning_rate": 2.013492475350286e-05, - "loss": 3.6069, - "step": 95700 - }, - { - "epoch": 24.86, - "learning_rate": 2.0114167099117804e-05, - "loss": 3.6363, - "step": 95800 - }, - { - "epoch": 24.88, - "learning_rate": 2.009340944473275e-05, - "loss": 3.636, - "step": 95900 - }, - { - "epoch": 24.91, - "learning_rate": 2.007265179034769e-05, - "loss": 3.6337, - "step": 96000 - }, - { - "epoch": 24.94, - "learning_rate": 2.0051894135962636e-05, - "loss": 3.6192, - "step": 96100 - }, - { - "epoch": 24.96, - "learning_rate": 2.003113648157758e-05, - "loss": 3.6057, - "step": 96200 - }, - { - "epoch": 24.99, - "learning_rate": 2.0010378827192526e-05, - "loss": 3.6318, - "step": 96300 - }, - { - "epoch": 25.01, - "learning_rate": 1.9989621172807474e-05, - "loss": 3.628, - "step": 96400 - }, - { - "epoch": 25.04, - "learning_rate": 1.996886351842242e-05, - "loss": 3.6271, - "step": 96500 - }, - { - "epoch": 25.06, - "learning_rate": 1.9948105864037367e-05, - "loss": 3.5885, - "step": 96600 - }, - { - "epoch": 25.09, - "learning_rate": 1.9927348209652312e-05, - "loss": 3.6315, - "step": 96700 - }, - { - "epoch": 25.12, - "learning_rate": 1.9906590555267254e-05, - "loss": 3.632, - "step": 96800 - }, - { - "epoch": 25.14, - "learning_rate": 1.9885832900882202e-05, - "loss": 3.6198, - "step": 96900 - }, - { - "epoch": 25.17, - "learning_rate": 1.9865075246497147e-05, - "loss": 3.5705, - "step": 97000 - }, - { - "epoch": 25.19, - "learning_rate": 1.9844317592112092e-05, - "loss": 3.624, - "step": 97100 - }, - { - "epoch": 25.22, - "learning_rate": 1.982355993772704e-05, - "loss": 3.6397, - "step": 97200 - }, - { - "epoch": 25.25, - "learning_rate": 1.9802802283341986e-05, - "loss": 3.6237, - "step": 97300 - }, - { - "epoch": 25.27, - "learning_rate": 1.978204462895693e-05, - "loss": 3.6085, - "step": 97400 - }, - { - "epoch": 25.3, - "learning_rate": 1.9761286974571876e-05, - "loss": 3.649, - "step": 97500 - }, - { - "epoch": 25.32, - "learning_rate": 1.974052932018682e-05, - "loss": 3.6188, - "step": 97600 - }, - { - "epoch": 25.35, - "learning_rate": 1.9719771665801765e-05, - "loss": 3.5814, - "step": 97700 - }, - { - "epoch": 25.38, - "learning_rate": 1.969901401141671e-05, - "loss": 3.63, - "step": 97800 - }, - { - "epoch": 25.4, - "learning_rate": 1.967825635703166e-05, - "loss": 3.6045, - "step": 97900 - }, - { - "epoch": 25.43, - "learning_rate": 1.9657498702646604e-05, - "loss": 3.6146, - "step": 98000 - }, - { - "epoch": 25.45, - "learning_rate": 1.963674104826155e-05, - "loss": 3.6045, - "step": 98100 - }, - { - "epoch": 25.48, - "learning_rate": 1.9615983393876494e-05, - "loss": 3.6275, - "step": 98200 - }, - { - "epoch": 25.51, - "learning_rate": 1.959522573949144e-05, - "loss": 3.6167, - "step": 98300 - }, - { - "epoch": 25.53, - "learning_rate": 1.9574468085106384e-05, - "loss": 3.6221, - "step": 98400 - }, - { - "epoch": 25.56, - "learning_rate": 1.9553710430721332e-05, - "loss": 3.6447, - "step": 98500 - }, - { - "epoch": 25.58, - "learning_rate": 1.9532952776336277e-05, - "loss": 3.6243, - "step": 98600 - }, - { - "epoch": 25.61, - "learning_rate": 1.9512195121951222e-05, - "loss": 3.6212, - "step": 98700 - }, - { - "epoch": 25.64, - "learning_rate": 1.9491437467566167e-05, - "loss": 3.6, - "step": 98800 - }, - { - "epoch": 25.66, - "learning_rate": 1.9470679813181112e-05, - "loss": 3.6268, - "step": 98900 - }, - { - "epoch": 25.69, - "learning_rate": 1.9449922158796057e-05, - "loss": 3.6228, - "step": 99000 - }, - { - "epoch": 25.71, - "learning_rate": 1.9429164504411002e-05, - "loss": 3.6269, - "step": 99100 - }, - { - "epoch": 25.74, - "learning_rate": 1.940840685002595e-05, - "loss": 3.6146, - "step": 99200 - }, - { - "epoch": 25.77, - "learning_rate": 1.9387649195640895e-05, - "loss": 3.6174, - "step": 99300 - }, - { - "epoch": 25.79, - "learning_rate": 1.936689154125584e-05, - "loss": 3.6065, - "step": 99400 - }, - { - "epoch": 25.82, - "learning_rate": 1.9346133886870785e-05, - "loss": 3.6325, - "step": 99500 - }, - { - "epoch": 25.84, - "learning_rate": 1.932537623248573e-05, - "loss": 3.5995, - "step": 99600 - }, - { - "epoch": 25.87, - "learning_rate": 1.9304618578100675e-05, - "loss": 3.6314, - "step": 99700 - }, - { - "epoch": 25.9, - "learning_rate": 1.9283860923715623e-05, - "loss": 3.5918, - "step": 99800 - }, - { - "epoch": 25.92, - "learning_rate": 1.926310326933057e-05, - "loss": 3.6165, - "step": 99900 - }, - { - "epoch": 25.95, - "learning_rate": 1.9242345614945513e-05, - "loss": 3.6295, - "step": 100000 - }, - { - "epoch": 25.95, - "eval_gen_len": 12.2774, - "eval_loss": 3.606283664703369, - "eval_rouge1": 39.4383, - "eval_rouge2": 7.6179, - "eval_rougeL": 35.444, - "eval_rougeLsum": 35.4606, - "eval_runtime": 75.8245, - "eval_samples_per_second": 50.828, - "eval_steps_per_second": 6.357, - "step": 100000 - }, - { - "epoch": 25.97, - "learning_rate": 1.9221587960560458e-05, - "loss": 3.6279, - "step": 100100 - }, - { - "epoch": 26.0, - "learning_rate": 1.9200830306175403e-05, - "loss": 3.6296, - "step": 100200 - }, - { - "epoch": 26.02, - "learning_rate": 1.9180072651790348e-05, - "loss": 3.5905, - "step": 100300 - }, - { - "epoch": 26.05, - "learning_rate": 1.9159314997405293e-05, - "loss": 3.5854, - "step": 100400 - }, - { - "epoch": 26.08, - "learning_rate": 1.913855734302024e-05, - "loss": 3.6127, - "step": 100500 - }, - { - "epoch": 26.1, - "learning_rate": 1.9117799688635186e-05, - "loss": 3.6121, - "step": 100600 - }, - { - "epoch": 26.13, - "learning_rate": 1.909704203425013e-05, - "loss": 3.6261, - "step": 100700 - }, - { - "epoch": 26.15, - "learning_rate": 1.9076284379865076e-05, - "loss": 3.6273, - "step": 100800 - }, - { - "epoch": 26.18, - "learning_rate": 1.905552672548002e-05, - "loss": 3.6397, - "step": 100900 - }, - { - "epoch": 26.21, - "learning_rate": 1.9034769071094966e-05, - "loss": 3.6361, - "step": 101000 - }, - { - "epoch": 26.23, - "learning_rate": 1.9014011416709915e-05, - "loss": 3.626, - "step": 101100 - }, - { - "epoch": 26.26, - "learning_rate": 1.899325376232486e-05, - "loss": 3.6197, - "step": 101200 - }, - { - "epoch": 26.28, - "learning_rate": 1.8972496107939805e-05, - "loss": 3.6215, - "step": 101300 - }, - { - "epoch": 26.31, - "learning_rate": 1.895173845355475e-05, - "loss": 3.6079, - "step": 101400 - }, - { - "epoch": 26.34, - "learning_rate": 1.8930980799169695e-05, - "loss": 3.6268, - "step": 101500 - }, - { - "epoch": 26.36, - "learning_rate": 1.891022314478464e-05, - "loss": 3.6208, - "step": 101600 - }, - { - "epoch": 26.39, - "learning_rate": 1.8889465490399585e-05, - "loss": 3.6079, - "step": 101700 - }, - { - "epoch": 26.41, - "learning_rate": 1.8868707836014533e-05, - "loss": 3.6288, - "step": 101800 - }, - { - "epoch": 26.44, - "learning_rate": 1.8847950181629478e-05, - "loss": 3.616, - "step": 101900 - }, - { - "epoch": 26.47, - "learning_rate": 1.8827192527244423e-05, - "loss": 3.6295, - "step": 102000 - }, - { - "epoch": 26.49, - "learning_rate": 1.8806434872859368e-05, - "loss": 3.6294, - "step": 102100 - }, - { - "epoch": 26.52, - "learning_rate": 1.8785677218474313e-05, - "loss": 3.5978, - "step": 102200 - }, - { - "epoch": 26.54, - "learning_rate": 1.8764919564089258e-05, - "loss": 3.6268, - "step": 102300 - }, - { - "epoch": 26.57, - "learning_rate": 1.8744161909704206e-05, - "loss": 3.6098, - "step": 102400 - }, - { - "epoch": 26.6, - "learning_rate": 1.872340425531915e-05, - "loss": 3.6064, - "step": 102500 - }, - { - "epoch": 26.62, - "learning_rate": 1.8702646600934096e-05, - "loss": 3.6046, - "step": 102600 - }, - { - "epoch": 26.65, - "learning_rate": 1.868188894654904e-05, - "loss": 3.6195, - "step": 102700 - }, - { - "epoch": 26.67, - "learning_rate": 1.8661131292163986e-05, - "loss": 3.6087, - "step": 102800 - }, - { - "epoch": 26.7, - "learning_rate": 1.864037363777893e-05, - "loss": 3.6082, - "step": 102900 - }, - { - "epoch": 26.73, - "learning_rate": 1.8619615983393876e-05, - "loss": 3.6098, - "step": 103000 - }, - { - "epoch": 26.75, - "learning_rate": 1.8598858329008824e-05, - "loss": 3.6233, - "step": 103100 - }, - { - "epoch": 26.78, - "learning_rate": 1.857810067462377e-05, - "loss": 3.6173, - "step": 103200 - }, - { - "epoch": 26.8, - "learning_rate": 1.8557343020238714e-05, - "loss": 3.6218, - "step": 103300 - }, - { - "epoch": 26.83, - "learning_rate": 1.8536585365853663e-05, - "loss": 3.6211, - "step": 103400 - }, - { - "epoch": 26.86, - "learning_rate": 1.8515827711468604e-05, - "loss": 3.6139, - "step": 103500 - }, - { - "epoch": 26.88, - "learning_rate": 1.849507005708355e-05, - "loss": 3.6078, - "step": 103600 - }, - { - "epoch": 26.91, - "learning_rate": 1.8474312402698497e-05, - "loss": 3.6302, - "step": 103700 - }, - { - "epoch": 26.93, - "learning_rate": 1.8453554748313442e-05, - "loss": 3.606, - "step": 103800 - }, - { - "epoch": 26.96, - "learning_rate": 1.8432797093928387e-05, - "loss": 3.6303, - "step": 103900 - }, - { - "epoch": 26.98, - "learning_rate": 1.8412039439543332e-05, - "loss": 3.6284, - "step": 104000 - }, - { - "epoch": 27.01, - "learning_rate": 1.839128178515828e-05, - "loss": 3.63, - "step": 104100 - }, - { - "epoch": 27.04, - "learning_rate": 1.8370524130773226e-05, - "loss": 3.5811, - "step": 104200 - }, - { - "epoch": 27.06, - "learning_rate": 1.8349766476388167e-05, - "loss": 3.6216, - "step": 104300 - }, - { - "epoch": 27.09, - "learning_rate": 1.8329008822003116e-05, - "loss": 3.6315, - "step": 104400 - }, - { - "epoch": 27.11, - "learning_rate": 1.830825116761806e-05, - "loss": 3.6269, - "step": 104500 - }, - { - "epoch": 27.14, - "learning_rate": 1.8287493513233006e-05, - "loss": 3.6195, - "step": 104600 - }, - { - "epoch": 27.17, - "learning_rate": 1.8266735858847954e-05, - "loss": 3.6127, - "step": 104700 - }, - { - "epoch": 27.19, - "learning_rate": 1.82459782044629e-05, - "loss": 3.6216, - "step": 104800 - }, - { - "epoch": 27.22, - "learning_rate": 1.8225220550077844e-05, - "loss": 3.6112, - "step": 104900 - }, - { - "epoch": 27.24, - "learning_rate": 1.820446289569279e-05, - "loss": 3.6186, - "step": 105000 - }, - { - "epoch": 27.24, - "eval_gen_len": 12.2769, - "eval_loss": 3.6040432453155518, - "eval_rouge1": 40.2581, - "eval_rouge2": 7.9107, - "eval_rougeL": 36.2296, - "eval_rougeLsum": 36.2565, - "eval_runtime": 75.7656, - "eval_samples_per_second": 50.867, - "eval_steps_per_second": 6.362, - "step": 105000 - }, - { - "epoch": 27.27, - "learning_rate": 1.8183705241307734e-05, - "loss": 3.6093, - "step": 105100 - }, - { - "epoch": 27.3, - "learning_rate": 1.816294758692268e-05, - "loss": 3.6051, - "step": 105200 - }, - { - "epoch": 27.32, - "learning_rate": 1.8142189932537624e-05, - "loss": 3.6303, - "step": 105300 - }, - { - "epoch": 27.35, - "learning_rate": 1.8121432278152572e-05, - "loss": 3.6031, - "step": 105400 - }, - { - "epoch": 27.37, - "learning_rate": 1.8100674623767517e-05, - "loss": 3.6177, - "step": 105500 - }, - { - "epoch": 27.4, - "learning_rate": 1.8079916969382462e-05, - "loss": 3.6258, - "step": 105600 - }, - { - "epoch": 27.43, - "learning_rate": 1.8059159314997407e-05, - "loss": 3.642, - "step": 105700 - }, - { - "epoch": 27.45, - "learning_rate": 1.8038401660612352e-05, - "loss": 3.62, - "step": 105800 - }, - { - "epoch": 27.48, - "learning_rate": 1.8017644006227297e-05, - "loss": 3.6238, - "step": 105900 - }, - { - "epoch": 27.5, - "learning_rate": 1.7996886351842245e-05, - "loss": 3.6382, - "step": 106000 - }, - { - "epoch": 27.53, - "learning_rate": 1.797612869745719e-05, - "loss": 3.5665, - "step": 106100 - }, - { - "epoch": 27.56, - "learning_rate": 1.7955371043072135e-05, - "loss": 3.6275, - "step": 106200 - }, - { - "epoch": 27.58, - "learning_rate": 1.793461338868708e-05, - "loss": 3.6112, - "step": 106300 - }, - { - "epoch": 27.61, - "learning_rate": 1.7913855734302025e-05, - "loss": 3.6087, - "step": 106400 - }, - { - "epoch": 27.63, - "learning_rate": 1.789309807991697e-05, - "loss": 3.5993, - "step": 106500 - }, - { - "epoch": 27.66, - "learning_rate": 1.7872340425531915e-05, - "loss": 3.6198, - "step": 106600 - }, - { - "epoch": 27.69, - "learning_rate": 1.7851582771146863e-05, - "loss": 3.6191, - "step": 106700 - }, - { - "epoch": 27.71, - "learning_rate": 1.783082511676181e-05, - "loss": 3.6086, - "step": 106800 - }, - { - "epoch": 27.74, - "learning_rate": 1.7810067462376753e-05, - "loss": 3.6147, - "step": 106900 - }, - { - "epoch": 27.76, - "learning_rate": 1.77893098079917e-05, - "loss": 3.6102, - "step": 107000 - }, - { - "epoch": 27.79, - "learning_rate": 1.7768552153606643e-05, - "loss": 3.6216, - "step": 107100 - }, - { - "epoch": 27.82, - "learning_rate": 1.7747794499221588e-05, - "loss": 3.6088, - "step": 107200 - }, - { - "epoch": 27.84, - "learning_rate": 1.7727036844836537e-05, - "loss": 3.6231, - "step": 107300 - }, - { - "epoch": 27.87, - "learning_rate": 1.770627919045148e-05, - "loss": 3.6177, - "step": 107400 - }, - { - "epoch": 27.89, - "learning_rate": 1.7685521536066427e-05, - "loss": 3.6116, - "step": 107500 - }, - { - "epoch": 27.92, - "learning_rate": 1.766476388168137e-05, - "loss": 3.6395, - "step": 107600 - }, - { - "epoch": 27.94, - "learning_rate": 1.7644006227296316e-05, - "loss": 3.6175, - "step": 107700 - }, - { - "epoch": 27.97, - "learning_rate": 1.762324857291126e-05, - "loss": 3.6288, - "step": 107800 - }, - { - "epoch": 28.0, - "learning_rate": 1.7602490918526206e-05, - "loss": 3.6269, - "step": 107900 - }, - { - "epoch": 28.02, - "learning_rate": 1.7581733264141155e-05, - "loss": 3.6127, - "step": 108000 - }, - { - "epoch": 28.05, - "learning_rate": 1.75609756097561e-05, - "loss": 3.626, - "step": 108100 - }, - { - "epoch": 28.07, - "learning_rate": 1.7540217955371045e-05, - "loss": 3.5972, - "step": 108200 - }, - { - "epoch": 28.1, - "learning_rate": 1.751946030098599e-05, - "loss": 3.6151, - "step": 108300 - }, - { - "epoch": 28.13, - "learning_rate": 1.7498702646600935e-05, - "loss": 3.6286, - "step": 108400 - }, - { - "epoch": 28.15, - "learning_rate": 1.747794499221588e-05, - "loss": 3.5928, - "step": 108500 - }, - { - "epoch": 28.18, - "learning_rate": 1.7457187337830828e-05, - "loss": 3.6095, - "step": 108600 - }, - { - "epoch": 28.2, - "learning_rate": 1.7436429683445773e-05, - "loss": 3.6135, - "step": 108700 - }, - { - "epoch": 28.23, - "learning_rate": 1.7415672029060718e-05, - "loss": 3.6138, - "step": 108800 - }, - { - "epoch": 28.26, - "learning_rate": 1.7394914374675663e-05, - "loss": 3.6359, - "step": 108900 - }, - { - "epoch": 28.28, - "learning_rate": 1.7374156720290608e-05, - "loss": 3.6128, - "step": 109000 - }, - { - "epoch": 28.31, - "learning_rate": 1.7353399065905553e-05, - "loss": 3.5651, - "step": 109100 - }, - { - "epoch": 28.33, - "learning_rate": 1.7332641411520498e-05, - "loss": 3.6242, - "step": 109200 - }, - { - "epoch": 28.36, - "learning_rate": 1.7311883757135446e-05, - "loss": 3.6071, - "step": 109300 - }, - { - "epoch": 28.39, - "learning_rate": 1.729112610275039e-05, - "loss": 3.6131, - "step": 109400 - }, - { - "epoch": 28.41, - "learning_rate": 1.7270368448365336e-05, - "loss": 3.6289, - "step": 109500 - }, - { - "epoch": 28.44, - "learning_rate": 1.724961079398028e-05, - "loss": 3.6139, - "step": 109600 - }, - { - "epoch": 28.46, - "learning_rate": 1.7228853139595226e-05, - "loss": 3.6368, - "step": 109700 - }, - { - "epoch": 28.49, - "learning_rate": 1.720809548521017e-05, - "loss": 3.6212, - "step": 109800 - }, - { - "epoch": 28.52, - "learning_rate": 1.7187337830825116e-05, - "loss": 3.625, - "step": 109900 - }, - { - "epoch": 28.54, - "learning_rate": 1.7166580176440064e-05, - "loss": 3.6196, - "step": 110000 - }, - { - "epoch": 28.54, - "eval_gen_len": 12.2761, - "eval_loss": 3.60282039642334, - "eval_rouge1": 40.2885, - "eval_rouge2": 7.7568, - "eval_rougeL": 36.2473, - "eval_rougeLsum": 36.286, - "eval_runtime": 75.5525, - "eval_samples_per_second": 51.011, - "eval_steps_per_second": 6.38, - "step": 110000 - }, - { - "epoch": 28.57, - "learning_rate": 1.714582252205501e-05, - "loss": 3.6141, - "step": 110100 - }, - { - "epoch": 28.59, - "learning_rate": 1.7125064867669954e-05, - "loss": 3.624, - "step": 110200 - }, - { - "epoch": 28.62, - "learning_rate": 1.71043072132849e-05, - "loss": 3.6169, - "step": 110300 - }, - { - "epoch": 28.65, - "learning_rate": 1.7083549558899844e-05, - "loss": 3.6252, - "step": 110400 - }, - { - "epoch": 28.67, - "learning_rate": 1.706279190451479e-05, - "loss": 3.6386, - "step": 110500 - }, - { - "epoch": 28.7, - "learning_rate": 1.7042034250129737e-05, - "loss": 3.6126, - "step": 110600 - }, - { - "epoch": 28.72, - "learning_rate": 1.7021276595744682e-05, - "loss": 3.6023, - "step": 110700 - }, - { - "epoch": 28.75, - "learning_rate": 1.7000518941359627e-05, - "loss": 3.5926, - "step": 110800 - }, - { - "epoch": 28.78, - "learning_rate": 1.6979761286974576e-05, - "loss": 3.5743, - "step": 110900 - }, - { - "epoch": 28.8, - "learning_rate": 1.695900363258952e-05, - "loss": 3.5902, - "step": 111000 - }, - { - "epoch": 28.83, - "learning_rate": 1.6938245978204462e-05, - "loss": 3.5791, - "step": 111100 - }, - { - "epoch": 28.85, - "learning_rate": 1.6917488323819407e-05, - "loss": 3.6161, - "step": 111200 - }, - { - "epoch": 28.88, - "learning_rate": 1.6896730669434356e-05, - "loss": 3.6272, - "step": 111300 - }, - { - "epoch": 28.91, - "learning_rate": 1.68759730150493e-05, - "loss": 3.6376, - "step": 111400 - }, - { - "epoch": 28.93, - "learning_rate": 1.6855215360664246e-05, - "loss": 3.6284, - "step": 111500 - }, - { - "epoch": 28.96, - "learning_rate": 1.6834457706279194e-05, - "loss": 3.6144, - "step": 111600 - }, - { - "epoch": 28.98, - "learning_rate": 1.681370005189414e-05, - "loss": 3.6209, - "step": 111700 - }, - { - "epoch": 29.01, - "learning_rate": 1.679294239750908e-05, - "loss": 3.6326, - "step": 111800 - }, - { - "epoch": 29.03, - "learning_rate": 1.677218474312403e-05, - "loss": 3.6165, - "step": 111900 - }, - { - "epoch": 29.06, - "learning_rate": 1.6751427088738974e-05, - "loss": 3.6292, - "step": 112000 - }, - { - "epoch": 29.09, - "learning_rate": 1.673066943435392e-05, - "loss": 3.6099, - "step": 112100 - }, - { - "epoch": 29.11, - "learning_rate": 1.6709911779968867e-05, - "loss": 3.6171, - "step": 112200 - }, - { - "epoch": 29.14, - "learning_rate": 1.6689154125583812e-05, - "loss": 3.6182, - "step": 112300 - }, - { - "epoch": 29.16, - "learning_rate": 1.6668396471198757e-05, - "loss": 3.6187, - "step": 112400 - }, - { - "epoch": 29.19, - "learning_rate": 1.6647638816813702e-05, - "loss": 3.6385, - "step": 112500 - }, - { - "epoch": 29.22, - "learning_rate": 1.6626881162428647e-05, - "loss": 3.6077, - "step": 112600 - }, - { - "epoch": 29.24, - "learning_rate": 1.6606123508043592e-05, - "loss": 3.6391, - "step": 112700 - }, - { - "epoch": 29.27, - "learning_rate": 1.6585365853658537e-05, - "loss": 3.6257, - "step": 112800 - }, - { - "epoch": 29.29, - "learning_rate": 1.6564608199273485e-05, - "loss": 3.5785, - "step": 112900 - }, - { - "epoch": 29.32, - "learning_rate": 1.654385054488843e-05, - "loss": 3.6128, - "step": 113000 - }, - { - "epoch": 29.35, - "learning_rate": 1.6523092890503375e-05, - "loss": 3.6107, - "step": 113100 - }, - { - "epoch": 29.37, - "learning_rate": 1.650233523611832e-05, - "loss": 3.609, - "step": 113200 - }, - { - "epoch": 29.4, - "learning_rate": 1.6481577581733265e-05, - "loss": 3.6079, - "step": 113300 - }, - { - "epoch": 29.42, - "learning_rate": 1.646081992734821e-05, - "loss": 3.6008, - "step": 113400 - }, - { - "epoch": 29.45, - "learning_rate": 1.644006227296316e-05, - "loss": 3.6209, - "step": 113500 - }, - { - "epoch": 29.48, - "learning_rate": 1.6419304618578103e-05, - "loss": 3.6079, - "step": 113600 - }, - { - "epoch": 29.5, - "learning_rate": 1.639854696419305e-05, - "loss": 3.6135, - "step": 113700 - }, - { - "epoch": 29.53, - "learning_rate": 1.6377789309807993e-05, - "loss": 3.6233, - "step": 113800 - }, - { - "epoch": 29.55, - "learning_rate": 1.635703165542294e-05, - "loss": 3.5894, - "step": 113900 - }, - { - "epoch": 29.58, - "learning_rate": 1.6336274001037883e-05, - "loss": 3.6198, - "step": 114000 - }, - { - "epoch": 29.61, - "learning_rate": 1.631551634665283e-05, - "loss": 3.6168, - "step": 114100 - }, - { - "epoch": 29.63, - "learning_rate": 1.6294758692267777e-05, - "loss": 3.6076, - "step": 114200 - }, - { - "epoch": 29.66, - "learning_rate": 1.627400103788272e-05, - "loss": 3.6185, - "step": 114300 - }, - { - "epoch": 29.68, - "learning_rate": 1.6253243383497667e-05, - "loss": 3.6049, - "step": 114400 - }, - { - "epoch": 29.71, - "learning_rate": 1.623248572911261e-05, - "loss": 3.5918, - "step": 114500 - }, - { - "epoch": 29.74, - "learning_rate": 1.6211728074727557e-05, - "loss": 3.6093, - "step": 114600 - }, - { - "epoch": 29.76, - "learning_rate": 1.61909704203425e-05, - "loss": 3.6138, - "step": 114700 - }, - { - "epoch": 29.79, - "learning_rate": 1.6170212765957446e-05, - "loss": 3.6159, - "step": 114800 - }, - { - "epoch": 29.81, - "learning_rate": 1.6149455111572395e-05, - "loss": 3.6266, - "step": 114900 - }, - { - "epoch": 29.84, - "learning_rate": 1.612869745718734e-05, - "loss": 3.6218, - "step": 115000 - }, - { - "epoch": 29.84, - "eval_gen_len": 12.2766, - "eval_loss": 3.6007707118988037, - "eval_rouge1": 40.3343, - "eval_rouge2": 8.055, - "eval_rougeL": 36.3397, - "eval_rougeLsum": 36.3797, - "eval_runtime": 75.7103, - "eval_samples_per_second": 50.905, - "eval_steps_per_second": 6.366, - "step": 115000 - }, - { - "epoch": 29.87, - "learning_rate": 1.6107939802802285e-05, - "loss": 3.6301, - "step": 115100 - }, - { - "epoch": 29.89, - "learning_rate": 1.608718214841723e-05, - "loss": 3.615, - "step": 115200 - }, - { - "epoch": 29.92, - "learning_rate": 1.6066424494032175e-05, - "loss": 3.5828, - "step": 115300 - }, - { - "epoch": 29.94, - "learning_rate": 1.604566683964712e-05, - "loss": 3.6307, - "step": 115400 - }, - { - "epoch": 29.97, - "learning_rate": 1.6024909185262068e-05, - "loss": 3.5944, - "step": 115500 - }, - { - "epoch": 29.99, - "learning_rate": 1.6004151530877013e-05, - "loss": 3.6015, - "step": 115600 - }, - { - "epoch": 30.02, - "learning_rate": 1.5983393876491958e-05, - "loss": 3.6145, - "step": 115700 - }, - { - "epoch": 30.05, - "learning_rate": 1.5962636222106903e-05, - "loss": 3.5917, - "step": 115800 - }, - { - "epoch": 30.07, - "learning_rate": 1.5941878567721848e-05, - "loss": 3.6101, - "step": 115900 - }, - { - "epoch": 30.1, - "learning_rate": 1.5921120913336793e-05, - "loss": 3.6065, - "step": 116000 - }, - { - "epoch": 30.12, - "learning_rate": 1.5900363258951738e-05, - "loss": 3.6199, - "step": 116100 - }, - { - "epoch": 30.15, - "learning_rate": 1.5879605604566686e-05, - "loss": 3.6231, - "step": 116200 - }, - { - "epoch": 30.18, - "learning_rate": 1.585884795018163e-05, - "loss": 3.6277, - "step": 116300 - }, - { - "epoch": 30.2, - "learning_rate": 1.5838090295796576e-05, - "loss": 3.6115, - "step": 116400 - }, - { - "epoch": 30.23, - "learning_rate": 1.581733264141152e-05, - "loss": 3.6093, - "step": 116500 - }, - { - "epoch": 30.25, - "learning_rate": 1.5796574987026466e-05, - "loss": 3.6299, - "step": 116600 - }, - { - "epoch": 30.28, - "learning_rate": 1.577581733264141e-05, - "loss": 3.6238, - "step": 116700 - }, - { - "epoch": 30.31, - "learning_rate": 1.575505967825636e-05, - "loss": 3.6178, - "step": 116800 - }, - { - "epoch": 30.33, - "learning_rate": 1.5734302023871304e-05, - "loss": 3.5967, - "step": 116900 - }, - { - "epoch": 30.36, - "learning_rate": 1.571354436948625e-05, - "loss": 3.5902, - "step": 117000 - }, - { - "epoch": 30.38, - "learning_rate": 1.5692786715101194e-05, - "loss": 3.6247, - "step": 117100 - }, - { - "epoch": 30.41, - "learning_rate": 1.567202906071614e-05, - "loss": 3.5995, - "step": 117200 - }, - { - "epoch": 30.44, - "learning_rate": 1.5651271406331084e-05, - "loss": 3.5888, - "step": 117300 - }, - { - "epoch": 30.46, - "learning_rate": 1.563051375194603e-05, - "loss": 3.6221, - "step": 117400 - }, - { - "epoch": 30.49, - "learning_rate": 1.5609756097560978e-05, - "loss": 3.6046, - "step": 117500 - }, - { - "epoch": 30.51, - "learning_rate": 1.5588998443175923e-05, - "loss": 3.6324, - "step": 117600 - }, - { - "epoch": 30.54, - "learning_rate": 1.5568240788790867e-05, - "loss": 3.6292, - "step": 117700 - }, - { - "epoch": 30.57, - "learning_rate": 1.5547483134405812e-05, - "loss": 3.6127, - "step": 117800 - }, - { - "epoch": 30.59, - "learning_rate": 1.5526725480020757e-05, - "loss": 3.5948, - "step": 117900 - }, - { - "epoch": 30.62, - "learning_rate": 1.5505967825635702e-05, - "loss": 3.6113, - "step": 118000 - }, - { - "epoch": 30.64, - "learning_rate": 1.548521017125065e-05, - "loss": 3.6061, - "step": 118100 - }, - { - "epoch": 30.67, - "learning_rate": 1.5464452516865596e-05, - "loss": 3.6211, - "step": 118200 - }, - { - "epoch": 30.7, - "learning_rate": 1.544369486248054e-05, - "loss": 3.6006, - "step": 118300 - }, - { - "epoch": 30.72, - "learning_rate": 1.542293720809549e-05, - "loss": 3.5968, - "step": 118400 - }, - { - "epoch": 30.75, - "learning_rate": 1.5402179553710434e-05, - "loss": 3.6215, - "step": 118500 - }, - { - "epoch": 30.77, - "learning_rate": 1.5381421899325376e-05, - "loss": 3.6059, - "step": 118600 - }, - { - "epoch": 30.8, - "learning_rate": 1.536066424494032e-05, - "loss": 3.6082, - "step": 118700 - }, - { - "epoch": 30.83, - "learning_rate": 1.533990659055527e-05, - "loss": 3.6177, - "step": 118800 - }, - { - "epoch": 30.85, - "learning_rate": 1.5319148936170214e-05, - "loss": 3.6112, - "step": 118900 - }, - { - "epoch": 30.88, - "learning_rate": 1.529839128178516e-05, - "loss": 3.6266, - "step": 119000 - }, - { - "epoch": 30.9, - "learning_rate": 1.5277633627400107e-05, - "loss": 3.6059, - "step": 119100 - }, - { - "epoch": 30.93, - "learning_rate": 1.525687597301505e-05, - "loss": 3.6226, - "step": 119200 - }, - { - "epoch": 30.95, - "learning_rate": 1.5236118318629995e-05, - "loss": 3.6271, - "step": 119300 - }, - { - "epoch": 30.98, - "learning_rate": 1.5215360664244942e-05, - "loss": 3.6145, - "step": 119400 - }, - { - "epoch": 31.01, - "learning_rate": 1.5194603009859887e-05, - "loss": 3.611, - "step": 119500 - }, - { - "epoch": 31.03, - "learning_rate": 1.5173845355474832e-05, - "loss": 3.5825, - "step": 119600 - }, - { - "epoch": 31.06, - "learning_rate": 1.5153087701089777e-05, - "loss": 3.5962, - "step": 119700 - }, - { - "epoch": 31.08, - "learning_rate": 1.5132330046704724e-05, - "loss": 3.5957, - "step": 119800 - }, - { - "epoch": 31.11, - "learning_rate": 1.5111572392319669e-05, - "loss": 3.6402, - "step": 119900 - }, - { - "epoch": 31.14, - "learning_rate": 1.5090814737934614e-05, - "loss": 3.6194, - "step": 120000 - }, - { - "epoch": 31.14, - "eval_gen_len": 12.2761, - "eval_loss": 3.598875045776367, - "eval_rouge1": 40.1727, - "eval_rouge2": 8.0263, - "eval_rougeL": 36.1356, - "eval_rougeLsum": 36.148, - "eval_runtime": 75.8469, - "eval_samples_per_second": 50.813, - "eval_steps_per_second": 6.355, - "step": 120000 - }, - { - "epoch": 31.16, - "learning_rate": 1.507005708354956e-05, - "loss": 3.6332, - "step": 120100 - }, - { - "epoch": 31.19, - "learning_rate": 1.5049299429164505e-05, - "loss": 3.6044, - "step": 120200 - }, - { - "epoch": 31.21, - "learning_rate": 1.502854177477945e-05, - "loss": 3.6413, - "step": 120300 - }, - { - "epoch": 31.24, - "learning_rate": 1.5007784120394397e-05, - "loss": 3.6182, - "step": 120400 - }, - { - "epoch": 31.27, - "learning_rate": 1.4987026466009342e-05, - "loss": 3.6115, - "step": 120500 - }, - { - "epoch": 31.29, - "learning_rate": 1.4966268811624287e-05, - "loss": 3.6008, - "step": 120600 - }, - { - "epoch": 31.32, - "learning_rate": 1.4945511157239233e-05, - "loss": 3.6245, - "step": 120700 - }, - { - "epoch": 31.34, - "learning_rate": 1.4924753502854178e-05, - "loss": 3.6105, - "step": 120800 - }, - { - "epoch": 31.37, - "learning_rate": 1.4903995848469123e-05, - "loss": 3.6147, - "step": 120900 - }, - { - "epoch": 31.4, - "learning_rate": 1.4883238194084068e-05, - "loss": 3.6158, - "step": 121000 - }, - { - "epoch": 31.42, - "learning_rate": 1.4862480539699015e-05, - "loss": 3.6337, - "step": 121100 - }, - { - "epoch": 31.45, - "learning_rate": 1.484172288531396e-05, - "loss": 3.5977, - "step": 121200 - }, - { - "epoch": 31.47, - "learning_rate": 1.4820965230928905e-05, - "loss": 3.6235, - "step": 121300 - }, - { - "epoch": 31.5, - "learning_rate": 1.4800207576543852e-05, - "loss": 3.5937, - "step": 121400 - }, - { - "epoch": 31.53, - "learning_rate": 1.4779449922158797e-05, - "loss": 3.604, - "step": 121500 - }, - { - "epoch": 31.55, - "learning_rate": 1.4758692267773742e-05, - "loss": 3.6288, - "step": 121600 - }, - { - "epoch": 31.58, - "learning_rate": 1.473793461338869e-05, - "loss": 3.6092, - "step": 121700 - }, - { - "epoch": 31.6, - "learning_rate": 1.4717176959003633e-05, - "loss": 3.6009, - "step": 121800 - }, - { - "epoch": 31.63, - "learning_rate": 1.4696419304618578e-05, - "loss": 3.6203, - "step": 121900 - }, - { - "epoch": 31.66, - "learning_rate": 1.4675661650233527e-05, - "loss": 3.6227, - "step": 122000 - }, - { - "epoch": 31.68, - "learning_rate": 1.4654903995848471e-05, - "loss": 3.6076, - "step": 122100 - }, - { - "epoch": 31.71, - "learning_rate": 1.4634146341463415e-05, - "loss": 3.6155, - "step": 122200 - }, - { - "epoch": 31.73, - "learning_rate": 1.461338868707836e-05, - "loss": 3.6202, - "step": 122300 - }, - { - "epoch": 31.76, - "learning_rate": 1.4592631032693308e-05, - "loss": 3.6015, - "step": 122400 - }, - { - "epoch": 31.79, - "learning_rate": 1.4571873378308253e-05, - "loss": 3.6127, - "step": 122500 - }, - { - "epoch": 31.81, - "learning_rate": 1.4551115723923196e-05, - "loss": 3.6039, - "step": 122600 - }, - { - "epoch": 31.84, - "learning_rate": 1.4530358069538145e-05, - "loss": 3.6027, - "step": 122700 - }, - { - "epoch": 31.86, - "learning_rate": 1.450960041515309e-05, - "loss": 3.6075, - "step": 122800 - }, - { - "epoch": 31.89, - "learning_rate": 1.4488842760768033e-05, - "loss": 3.5994, - "step": 122900 - }, - { - "epoch": 31.91, - "learning_rate": 1.4468085106382981e-05, - "loss": 3.6048, - "step": 123000 - }, - { - "epoch": 31.94, - "learning_rate": 1.4447327451997926e-05, - "loss": 3.6216, - "step": 123100 - }, - { - "epoch": 31.97, - "learning_rate": 1.4426569797612871e-05, - "loss": 3.6034, - "step": 123200 - }, - { - "epoch": 31.99, - "learning_rate": 1.4405812143227818e-05, - "loss": 3.5946, - "step": 123300 - }, - { - "epoch": 32.02, - "learning_rate": 1.4385054488842763e-05, - "loss": 3.6223, - "step": 123400 - }, - { - "epoch": 32.04, - "learning_rate": 1.4364296834457708e-05, - "loss": 3.6146, - "step": 123500 - }, - { - "epoch": 32.07, - "learning_rate": 1.4343539180072653e-05, - "loss": 3.6067, - "step": 123600 - }, - { - "epoch": 32.1, - "learning_rate": 1.43227815256876e-05, - "loss": 3.6073, - "step": 123700 - }, - { - "epoch": 32.12, - "learning_rate": 1.4302023871302544e-05, - "loss": 3.6005, - "step": 123800 - }, - { - "epoch": 32.15, - "learning_rate": 1.428126621691749e-05, - "loss": 3.6215, - "step": 123900 - }, - { - "epoch": 32.17, - "learning_rate": 1.4260508562532436e-05, - "loss": 3.6146, - "step": 124000 - }, - { - "epoch": 32.2, - "learning_rate": 1.4239750908147381e-05, - "loss": 3.6269, - "step": 124100 - }, - { - "epoch": 32.23, - "learning_rate": 1.4218993253762326e-05, - "loss": 3.6096, - "step": 124200 - }, - { - "epoch": 32.25, - "learning_rate": 1.4198235599377273e-05, - "loss": 3.5796, - "step": 124300 - }, - { - "epoch": 32.28, - "learning_rate": 1.4177477944992218e-05, - "loss": 3.6166, - "step": 124400 - }, - { - "epoch": 32.3, - "learning_rate": 1.4156720290607163e-05, - "loss": 3.6222, - "step": 124500 - }, - { - "epoch": 32.33, - "learning_rate": 1.4135962636222108e-05, - "loss": 3.6166, - "step": 124600 - }, - { - "epoch": 32.36, - "learning_rate": 1.4115204981837054e-05, - "loss": 3.626, - "step": 124700 - }, - { - "epoch": 32.38, - "learning_rate": 1.4094447327452e-05, - "loss": 3.6202, - "step": 124800 - }, - { - "epoch": 32.41, - "learning_rate": 1.4073689673066944e-05, - "loss": 3.5897, - "step": 124900 - }, - { - "epoch": 32.43, - "learning_rate": 1.405293201868189e-05, - "loss": 3.619, - "step": 125000 - }, - { - "epoch": 32.43, - "eval_gen_len": 12.2763, - "eval_loss": 3.597046136856079, - "eval_rouge1": 40.9033, - "eval_rouge2": 8.2223, - "eval_rougeL": 36.8949, - "eval_rougeLsum": 36.9021, - "eval_runtime": 75.7787, - "eval_samples_per_second": 50.859, - "eval_steps_per_second": 6.361, - "step": 125000 - }, - { - "epoch": 32.46, - "learning_rate": 1.4032174364296836e-05, - "loss": 3.6003, - "step": 125100 - }, - { - "epoch": 32.49, - "learning_rate": 1.401141670991178e-05, - "loss": 3.6193, - "step": 125200 - }, - { - "epoch": 32.51, - "learning_rate": 1.3990659055526727e-05, - "loss": 3.5954, - "step": 125300 - }, - { - "epoch": 32.54, - "learning_rate": 1.3969901401141672e-05, - "loss": 3.6219, - "step": 125400 - }, - { - "epoch": 32.56, - "learning_rate": 1.3949143746756617e-05, - "loss": 3.6268, - "step": 125500 - }, - { - "epoch": 32.59, - "learning_rate": 1.3928386092371564e-05, - "loss": 3.6083, - "step": 125600 - }, - { - "epoch": 32.62, - "learning_rate": 1.3907628437986509e-05, - "loss": 3.6312, - "step": 125700 - }, - { - "epoch": 32.64, - "learning_rate": 1.3886870783601454e-05, - "loss": 3.5935, - "step": 125800 - }, - { - "epoch": 32.67, - "learning_rate": 1.3866113129216399e-05, - "loss": 3.5712, - "step": 125900 - }, - { - "epoch": 32.69, - "learning_rate": 1.3845355474831346e-05, - "loss": 3.597, - "step": 126000 - }, - { - "epoch": 32.72, - "learning_rate": 1.382459782044629e-05, - "loss": 3.6079, - "step": 126100 - }, - { - "epoch": 32.75, - "learning_rate": 1.3803840166061235e-05, - "loss": 3.5818, - "step": 126200 - }, - { - "epoch": 32.77, - "learning_rate": 1.3783082511676182e-05, - "loss": 3.5973, - "step": 126300 - }, - { - "epoch": 32.8, - "learning_rate": 1.3762324857291127e-05, - "loss": 3.5802, - "step": 126400 - }, - { - "epoch": 32.82, - "learning_rate": 1.3741567202906072e-05, - "loss": 3.6062, - "step": 126500 - }, - { - "epoch": 32.85, - "learning_rate": 1.3720809548521019e-05, - "loss": 3.6021, - "step": 126600 - }, - { - "epoch": 32.87, - "learning_rate": 1.3700051894135964e-05, - "loss": 3.6181, - "step": 126700 - }, - { - "epoch": 32.9, - "learning_rate": 1.3679294239750909e-05, - "loss": 3.623, - "step": 126800 - }, - { - "epoch": 32.93, - "learning_rate": 1.3658536585365855e-05, - "loss": 3.6192, - "step": 126900 - }, - { - "epoch": 32.95, - "learning_rate": 1.36377789309808e-05, - "loss": 3.6072, - "step": 127000 - }, - { - "epoch": 32.98, - "learning_rate": 1.3617021276595745e-05, - "loss": 3.6354, - "step": 127100 - }, - { - "epoch": 33.0, - "learning_rate": 1.359626362221069e-05, - "loss": 3.566, - "step": 127200 - }, - { - "epoch": 33.03, - "learning_rate": 1.3575505967825637e-05, - "loss": 3.6098, - "step": 127300 - }, - { - "epoch": 33.06, - "learning_rate": 1.3554748313440582e-05, - "loss": 3.6211, - "step": 127400 - }, - { - "epoch": 33.08, - "learning_rate": 1.3533990659055527e-05, - "loss": 3.6143, - "step": 127500 - }, - { - "epoch": 33.11, - "learning_rate": 1.3513233004670474e-05, - "loss": 3.6278, - "step": 127600 - }, - { - "epoch": 33.13, - "learning_rate": 1.3492475350285418e-05, - "loss": 3.6094, - "step": 127700 - }, - { - "epoch": 33.16, - "learning_rate": 1.3471717695900363e-05, - "loss": 3.5985, - "step": 127800 - }, - { - "epoch": 33.19, - "learning_rate": 1.345096004151531e-05, - "loss": 3.5683, - "step": 127900 - }, - { - "epoch": 33.21, - "learning_rate": 1.3430202387130255e-05, - "loss": 3.617, - "step": 128000 - }, - { - "epoch": 33.24, - "learning_rate": 1.34094447327452e-05, - "loss": 3.6051, - "step": 128100 - }, - { - "epoch": 33.26, - "learning_rate": 1.3388687078360147e-05, - "loss": 3.6116, - "step": 128200 - }, - { - "epoch": 33.29, - "learning_rate": 1.3367929423975092e-05, - "loss": 3.6178, - "step": 128300 - }, - { - "epoch": 33.32, - "learning_rate": 1.3347171769590037e-05, - "loss": 3.6324, - "step": 128400 - }, - { - "epoch": 33.34, - "learning_rate": 1.3326414115204982e-05, - "loss": 3.5958, - "step": 128500 - }, - { - "epoch": 33.37, - "learning_rate": 1.3305656460819928e-05, - "loss": 3.6233, - "step": 128600 - }, - { - "epoch": 33.39, - "learning_rate": 1.3284898806434873e-05, - "loss": 3.5967, - "step": 128700 - }, - { - "epoch": 33.42, - "learning_rate": 1.3264141152049818e-05, - "loss": 3.6049, - "step": 128800 - }, - { - "epoch": 33.45, - "learning_rate": 1.3243383497664765e-05, - "loss": 3.6029, - "step": 128900 - }, - { - "epoch": 33.47, - "learning_rate": 1.322262584327971e-05, - "loss": 3.5684, - "step": 129000 - }, - { - "epoch": 33.5, - "learning_rate": 1.3201868188894655e-05, - "loss": 3.5898, - "step": 129100 - }, - { - "epoch": 33.52, - "learning_rate": 1.3181110534509603e-05, - "loss": 3.6096, - "step": 129200 - }, - { - "epoch": 33.55, - "learning_rate": 1.3160352880124546e-05, - "loss": 3.6109, - "step": 129300 - }, - { - "epoch": 33.58, - "learning_rate": 1.3139595225739491e-05, - "loss": 3.6187, - "step": 129400 - }, - { - "epoch": 33.6, - "learning_rate": 1.3118837571354436e-05, - "loss": 3.6139, - "step": 129500 - }, - { - "epoch": 33.63, - "learning_rate": 1.3098079916969385e-05, - "loss": 3.6026, - "step": 129600 - }, - { - "epoch": 33.65, - "learning_rate": 1.3077322262584328e-05, - "loss": 3.5973, - "step": 129700 - }, - { - "epoch": 33.68, - "learning_rate": 1.3056564608199273e-05, - "loss": 3.5847, - "step": 129800 - }, - { - "epoch": 33.71, - "learning_rate": 1.3035806953814221e-05, - "loss": 3.6199, - "step": 129900 - }, - { - "epoch": 33.73, - "learning_rate": 1.3015049299429166e-05, - "loss": 3.6116, - "step": 130000 - }, - { - "epoch": 33.73, - "eval_gen_len": 12.2763, - "eval_loss": 3.5960943698883057, - "eval_rouge1": 41.6154, - "eval_rouge2": 8.5729, - "eval_rougeL": 37.598, - "eval_rougeLsum": 37.6264, - "eval_runtime": 75.6128, - "eval_samples_per_second": 50.97, - "eval_steps_per_second": 6.375, - "step": 130000 - }, - { - "epoch": 33.76, - "learning_rate": 1.299429164504411e-05, - "loss": 3.6062, - "step": 130100 - }, - { - "epoch": 33.78, - "learning_rate": 1.2973533990659058e-05, - "loss": 3.6131, - "step": 130200 - }, - { - "epoch": 33.81, - "learning_rate": 1.2952776336274003e-05, - "loss": 3.6163, - "step": 130300 - }, - { - "epoch": 33.83, - "learning_rate": 1.2932018681888948e-05, - "loss": 3.6133, - "step": 130400 - }, - { - "epoch": 33.86, - "learning_rate": 1.2911261027503895e-05, - "loss": 3.6176, - "step": 130500 - }, - { - "epoch": 33.89, - "learning_rate": 1.289050337311884e-05, - "loss": 3.588, - "step": 130600 - }, - { - "epoch": 33.91, - "learning_rate": 1.2869745718733784e-05, - "loss": 3.6007, - "step": 130700 - }, - { - "epoch": 33.94, - "learning_rate": 1.2848988064348728e-05, - "loss": 3.5726, - "step": 130800 - }, - { - "epoch": 33.96, - "learning_rate": 1.2828230409963676e-05, - "loss": 3.5923, - "step": 130900 - }, - { - "epoch": 33.99, - "learning_rate": 1.2807472755578621e-05, - "loss": 3.5719, - "step": 131000 - }, - { - "epoch": 34.02, - "learning_rate": 1.2786715101193566e-05, - "loss": 3.6186, - "step": 131100 - }, - { - "epoch": 34.04, - "learning_rate": 1.2765957446808513e-05, - "loss": 3.6011, - "step": 131200 - }, - { - "epoch": 34.07, - "learning_rate": 1.2745199792423458e-05, - "loss": 3.6053, - "step": 131300 - }, - { - "epoch": 34.09, - "learning_rate": 1.2724442138038403e-05, - "loss": 3.6126, - "step": 131400 - }, - { - "epoch": 34.12, - "learning_rate": 1.270368448365335e-05, - "loss": 3.6255, - "step": 131500 - }, - { - "epoch": 34.15, - "learning_rate": 1.2682926829268294e-05, - "loss": 3.6124, - "step": 131600 - }, - { - "epoch": 34.17, - "learning_rate": 1.266216917488324e-05, - "loss": 3.6038, - "step": 131700 - }, - { - "epoch": 34.2, - "learning_rate": 1.2641411520498186e-05, - "loss": 3.6156, - "step": 131800 - }, - { - "epoch": 34.22, - "learning_rate": 1.2620653866113131e-05, - "loss": 3.609, - "step": 131900 - }, - { - "epoch": 34.25, - "learning_rate": 1.2599896211728076e-05, - "loss": 3.6114, - "step": 132000 - }, - { - "epoch": 34.28, - "learning_rate": 1.257913855734302e-05, - "loss": 3.591, - "step": 132100 - }, - { - "epoch": 34.3, - "learning_rate": 1.2558380902957967e-05, - "loss": 3.6103, - "step": 132200 - }, - { - "epoch": 34.33, - "learning_rate": 1.2537623248572912e-05, - "loss": 3.6059, - "step": 132300 - }, - { - "epoch": 34.35, - "learning_rate": 1.2516865594187857e-05, - "loss": 3.6076, - "step": 132400 - }, - { - "epoch": 34.38, - "learning_rate": 1.2496107939802804e-05, - "loss": 3.5984, - "step": 132500 - }, - { - "epoch": 34.41, - "learning_rate": 1.2475350285417749e-05, - "loss": 3.6029, - "step": 132600 - }, - { - "epoch": 34.43, - "learning_rate": 1.2454592631032694e-05, - "loss": 3.6089, - "step": 132700 - }, - { - "epoch": 34.46, - "learning_rate": 1.243383497664764e-05, - "loss": 3.5939, - "step": 132800 - }, - { - "epoch": 34.48, - "learning_rate": 1.2413077322262586e-05, - "loss": 3.6122, - "step": 132900 - }, - { - "epoch": 34.51, - "learning_rate": 1.239231966787753e-05, - "loss": 3.6043, - "step": 133000 - }, - { - "epoch": 34.54, - "learning_rate": 1.2371562013492477e-05, - "loss": 3.6155, - "step": 133100 - }, - { - "epoch": 34.56, - "learning_rate": 1.2350804359107422e-05, - "loss": 3.5631, - "step": 133200 - }, - { - "epoch": 34.59, - "learning_rate": 1.2330046704722367e-05, - "loss": 3.6253, - "step": 133300 - }, - { - "epoch": 34.61, - "learning_rate": 1.2309289050337312e-05, - "loss": 3.625, - "step": 133400 - }, - { - "epoch": 34.64, - "learning_rate": 1.2288531395952259e-05, - "loss": 3.5985, - "step": 133500 - }, - { - "epoch": 34.67, - "learning_rate": 1.2267773741567204e-05, - "loss": 3.6154, - "step": 133600 - }, - { - "epoch": 34.69, - "learning_rate": 1.2247016087182149e-05, - "loss": 3.6042, - "step": 133700 - }, - { - "epoch": 34.72, - "learning_rate": 1.2226258432797095e-05, - "loss": 3.5926, - "step": 133800 - }, - { - "epoch": 34.74, - "learning_rate": 1.220550077841204e-05, - "loss": 3.5777, - "step": 133900 - }, - { - "epoch": 34.77, - "learning_rate": 1.2184743124026985e-05, - "loss": 3.6164, - "step": 134000 - }, - { - "epoch": 34.8, - "learning_rate": 1.2163985469641932e-05, - "loss": 3.5981, - "step": 134100 - }, - { - "epoch": 34.82, - "learning_rate": 1.2143227815256877e-05, - "loss": 3.619, - "step": 134200 - }, - { - "epoch": 34.85, - "learning_rate": 1.2122470160871822e-05, - "loss": 3.5911, - "step": 134300 - }, - { - "epoch": 34.87, - "learning_rate": 1.2101712506486767e-05, - "loss": 3.6282, - "step": 134400 - }, - { - "epoch": 34.9, - "learning_rate": 1.2080954852101714e-05, - "loss": 3.618, - "step": 134500 - }, - { - "epoch": 34.92, - "learning_rate": 1.2060197197716659e-05, - "loss": 3.612, - "step": 134600 - }, - { - "epoch": 34.95, - "learning_rate": 1.2039439543331604e-05, - "loss": 3.5997, - "step": 134700 - }, - { - "epoch": 34.98, - "learning_rate": 1.201868188894655e-05, - "loss": 3.6071, - "step": 134800 - }, - { - "epoch": 35.0, - "learning_rate": 1.1997924234561495e-05, - "loss": 3.6195, - "step": 134900 - }, - { - "epoch": 35.03, - "learning_rate": 1.197716658017644e-05, - "loss": 3.6, - "step": 135000 - }, - { - "epoch": 35.03, - "eval_gen_len": 12.2766, - "eval_loss": 3.5951695442199707, - "eval_rouge1": 41.4914, - "eval_rouge2": 8.5934, - "eval_rougeL": 37.4486, - "eval_rougeLsum": 37.4712, - "eval_runtime": 75.6467, - "eval_samples_per_second": 50.947, - "eval_steps_per_second": 6.372, - "step": 135000 - }, - { - "epoch": 35.05, - "learning_rate": 1.1956408925791387e-05, - "loss": 3.6107, - "step": 135100 - }, - { - "epoch": 35.08, - "learning_rate": 1.1935651271406332e-05, - "loss": 3.613, - "step": 135200 - }, - { - "epoch": 35.11, - "learning_rate": 1.1914893617021277e-05, - "loss": 3.6227, - "step": 135300 - }, - { - "epoch": 35.13, - "learning_rate": 1.1894135962636223e-05, - "loss": 3.62, - "step": 135400 - }, - { - "epoch": 35.16, - "learning_rate": 1.1873378308251168e-05, - "loss": 3.6185, - "step": 135500 - }, - { - "epoch": 35.18, - "learning_rate": 1.1852620653866113e-05, - "loss": 3.5743, - "step": 135600 - }, - { - "epoch": 35.21, - "learning_rate": 1.1831862999481058e-05, - "loss": 3.5984, - "step": 135700 - }, - { - "epoch": 35.24, - "learning_rate": 1.1811105345096005e-05, - "loss": 3.5912, - "step": 135800 - }, - { - "epoch": 35.26, - "learning_rate": 1.179034769071095e-05, - "loss": 3.6083, - "step": 135900 - }, - { - "epoch": 35.29, - "learning_rate": 1.1769590036325895e-05, - "loss": 3.604, - "step": 136000 - }, - { - "epoch": 35.31, - "learning_rate": 1.1748832381940842e-05, - "loss": 3.6075, - "step": 136100 - }, - { - "epoch": 35.34, - "learning_rate": 1.1728074727555786e-05, - "loss": 3.6021, - "step": 136200 - }, - { - "epoch": 35.37, - "learning_rate": 1.1707317073170731e-05, - "loss": 3.6198, - "step": 136300 - }, - { - "epoch": 35.39, - "learning_rate": 1.168655941878568e-05, - "loss": 3.6024, - "step": 136400 - }, - { - "epoch": 35.42, - "learning_rate": 1.1665801764400623e-05, - "loss": 3.6125, - "step": 136500 - }, - { - "epoch": 35.44, - "learning_rate": 1.1645044110015568e-05, - "loss": 3.5958, - "step": 136600 - }, - { - "epoch": 35.47, - "learning_rate": 1.1624286455630516e-05, - "loss": 3.5855, - "step": 136700 - }, - { - "epoch": 35.5, - "learning_rate": 1.1603528801245461e-05, - "loss": 3.5616, - "step": 136800 - }, - { - "epoch": 35.52, - "learning_rate": 1.1582771146860405e-05, - "loss": 3.6106, - "step": 136900 - }, - { - "epoch": 35.55, - "learning_rate": 1.156201349247535e-05, - "loss": 3.6263, - "step": 137000 - }, - { - "epoch": 35.57, - "learning_rate": 1.1541255838090298e-05, - "loss": 3.6063, - "step": 137100 - }, - { - "epoch": 35.6, - "learning_rate": 1.1520498183705241e-05, - "loss": 3.608, - "step": 137200 - }, - { - "epoch": 35.63, - "learning_rate": 1.1499740529320186e-05, - "loss": 3.6091, - "step": 137300 - }, - { - "epoch": 35.65, - "learning_rate": 1.1478982874935135e-05, - "loss": 3.6027, - "step": 137400 - }, - { - "epoch": 35.68, - "learning_rate": 1.145822522055008e-05, - "loss": 3.6037, - "step": 137500 - }, - { - "epoch": 35.7, - "learning_rate": 1.1437467566165023e-05, - "loss": 3.5811, - "step": 137600 - }, - { - "epoch": 35.73, - "learning_rate": 1.1416709911779971e-05, - "loss": 3.609, - "step": 137700 - }, - { - "epoch": 35.76, - "learning_rate": 1.1395952257394916e-05, - "loss": 3.6111, - "step": 137800 - }, - { - "epoch": 35.78, - "learning_rate": 1.1375194603009861e-05, - "loss": 3.598, - "step": 137900 - }, - { - "epoch": 35.81, - "learning_rate": 1.1354436948624808e-05, - "loss": 3.6237, - "step": 138000 - }, - { - "epoch": 35.83, - "learning_rate": 1.1333679294239753e-05, - "loss": 3.6093, - "step": 138100 - }, - { - "epoch": 35.86, - "learning_rate": 1.1312921639854698e-05, - "loss": 3.6202, - "step": 138200 - }, - { - "epoch": 35.88, - "learning_rate": 1.1292163985469643e-05, - "loss": 3.5869, - "step": 138300 - }, - { - "epoch": 35.91, - "learning_rate": 1.127140633108459e-05, - "loss": 3.6183, - "step": 138400 - }, - { - "epoch": 35.94, - "learning_rate": 1.1250648676699534e-05, - "loss": 3.6089, - "step": 138500 - }, - { - "epoch": 35.96, - "learning_rate": 1.122989102231448e-05, - "loss": 3.6135, - "step": 138600 - }, - { - "epoch": 35.99, - "learning_rate": 1.1209133367929426e-05, - "loss": 3.6, - "step": 138700 - }, - { - "epoch": 36.01, - "learning_rate": 1.1188375713544371e-05, - "loss": 3.5911, - "step": 138800 - }, - { - "epoch": 36.04, - "learning_rate": 1.1167618059159316e-05, - "loss": 3.5952, - "step": 138900 - }, - { - "epoch": 36.07, - "learning_rate": 1.1146860404774263e-05, - "loss": 3.6088, - "step": 139000 - }, - { - "epoch": 36.09, - "learning_rate": 1.1126102750389208e-05, - "loss": 3.6096, - "step": 139100 - }, - { - "epoch": 36.12, - "learning_rate": 1.1105345096004152e-05, - "loss": 3.5832, - "step": 139200 - }, - { - "epoch": 36.14, - "learning_rate": 1.1084587441619099e-05, - "loss": 3.6094, - "step": 139300 - }, - { - "epoch": 36.17, - "learning_rate": 1.1063829787234044e-05, - "loss": 3.62, - "step": 139400 - }, - { - "epoch": 36.2, - "learning_rate": 1.1043072132848989e-05, - "loss": 3.5665, - "step": 139500 - }, - { - "epoch": 36.22, - "learning_rate": 1.1022314478463934e-05, - "loss": 3.5996, - "step": 139600 - }, - { - "epoch": 36.25, - "learning_rate": 1.100155682407888e-05, - "loss": 3.579, - "step": 139700 - }, - { - "epoch": 36.27, - "learning_rate": 1.0980799169693826e-05, - "loss": 3.6229, - "step": 139800 - }, - { - "epoch": 36.3, - "learning_rate": 1.096004151530877e-05, - "loss": 3.6143, - "step": 139900 - }, - { - "epoch": 36.33, - "learning_rate": 1.0939283860923717e-05, - "loss": 3.5896, - "step": 140000 - }, - { - "epoch": 36.33, - "eval_gen_len": 12.2763, - "eval_loss": 3.5935587882995605, - "eval_rouge1": 41.7094, - "eval_rouge2": 8.5817, - "eval_rougeL": 37.677, - "eval_rougeLsum": 37.72, - "eval_runtime": 75.9538, - "eval_samples_per_second": 50.741, - "eval_steps_per_second": 6.346, - "step": 140000 - }, - { - "epoch": 36.35, - "learning_rate": 1.0918526206538662e-05, - "loss": 3.6109, - "step": 140100 - }, - { - "epoch": 36.38, - "learning_rate": 1.0897768552153607e-05, - "loss": 3.5871, - "step": 140200 - }, - { - "epoch": 36.4, - "learning_rate": 1.0877010897768554e-05, - "loss": 3.5966, - "step": 140300 - }, - { - "epoch": 36.43, - "learning_rate": 1.0856253243383499e-05, - "loss": 3.5886, - "step": 140400 - }, - { - "epoch": 36.46, - "learning_rate": 1.0835495588998444e-05, - "loss": 3.6128, - "step": 140500 - }, - { - "epoch": 36.48, - "learning_rate": 1.0814737934613389e-05, - "loss": 3.5788, - "step": 140600 - }, - { - "epoch": 36.51, - "learning_rate": 1.0793980280228335e-05, - "loss": 3.5739, - "step": 140700 - }, - { - "epoch": 36.53, - "learning_rate": 1.077322262584328e-05, - "loss": 3.6301, - "step": 140800 - }, - { - "epoch": 36.56, - "learning_rate": 1.0752464971458225e-05, - "loss": 3.5858, - "step": 140900 - }, - { - "epoch": 36.59, - "learning_rate": 1.0731707317073172e-05, - "loss": 3.5925, - "step": 141000 - }, - { - "epoch": 36.61, - "learning_rate": 1.0710949662688117e-05, - "loss": 3.6174, - "step": 141100 - }, - { - "epoch": 36.64, - "learning_rate": 1.0690192008303062e-05, - "loss": 3.6002, - "step": 141200 - }, - { - "epoch": 36.66, - "learning_rate": 1.0669434353918009e-05, - "loss": 3.6139, - "step": 141300 - }, - { - "epoch": 36.69, - "learning_rate": 1.0648676699532954e-05, - "loss": 3.6358, - "step": 141400 - }, - { - "epoch": 36.72, - "learning_rate": 1.0627919045147899e-05, - "loss": 3.6055, - "step": 141500 - }, - { - "epoch": 36.74, - "learning_rate": 1.0607161390762845e-05, - "loss": 3.6213, - "step": 141600 - }, - { - "epoch": 36.77, - "learning_rate": 1.058640373637779e-05, - "loss": 3.5881, - "step": 141700 - }, - { - "epoch": 36.79, - "learning_rate": 1.0565646081992735e-05, - "loss": 3.6262, - "step": 141800 - }, - { - "epoch": 36.82, - "learning_rate": 1.054488842760768e-05, - "loss": 3.6123, - "step": 141900 - }, - { - "epoch": 36.84, - "learning_rate": 1.0524130773222627e-05, - "loss": 3.6091, - "step": 142000 - }, - { - "epoch": 36.87, - "learning_rate": 1.0503373118837572e-05, - "loss": 3.6199, - "step": 142100 - }, - { - "epoch": 36.9, - "learning_rate": 1.0482615464452517e-05, - "loss": 3.6182, - "step": 142200 - }, - { - "epoch": 36.92, - "learning_rate": 1.0461857810067463e-05, - "loss": 3.5861, - "step": 142300 - }, - { - "epoch": 36.95, - "learning_rate": 1.0441100155682408e-05, - "loss": 3.5922, - "step": 142400 - }, - { - "epoch": 36.97, - "learning_rate": 1.0420342501297353e-05, - "loss": 3.6228, - "step": 142500 - }, - { - "epoch": 37.0, - "learning_rate": 1.03995848469123e-05, - "loss": 3.5662, - "step": 142600 - }, - { - "epoch": 37.03, - "learning_rate": 1.0378827192527245e-05, - "loss": 3.5931, - "step": 142700 - }, - { - "epoch": 37.05, - "learning_rate": 1.035806953814219e-05, - "loss": 3.6024, - "step": 142800 - }, - { - "epoch": 37.08, - "learning_rate": 1.0337311883757137e-05, - "loss": 3.5865, - "step": 142900 - }, - { - "epoch": 37.1, - "learning_rate": 1.0316554229372082e-05, - "loss": 3.6059, - "step": 143000 - }, - { - "epoch": 37.13, - "learning_rate": 1.0295796574987027e-05, - "loss": 3.6064, - "step": 143100 - }, - { - "epoch": 37.16, - "learning_rate": 1.0275038920601972e-05, - "loss": 3.5973, - "step": 143200 - }, - { - "epoch": 37.18, - "learning_rate": 1.0254281266216918e-05, - "loss": 3.6123, - "step": 143300 - }, - { - "epoch": 37.21, - "learning_rate": 1.0233523611831863e-05, - "loss": 3.5877, - "step": 143400 - }, - { - "epoch": 37.23, - "learning_rate": 1.0212765957446808e-05, - "loss": 3.6036, - "step": 143500 - }, - { - "epoch": 37.26, - "learning_rate": 1.0192008303061755e-05, - "loss": 3.6061, - "step": 143600 - }, - { - "epoch": 37.29, - "learning_rate": 1.01712506486767e-05, - "loss": 3.6231, - "step": 143700 - }, - { - "epoch": 37.31, - "learning_rate": 1.0150492994291645e-05, - "loss": 3.6303, - "step": 143800 - }, - { - "epoch": 37.34, - "learning_rate": 1.0129735339906593e-05, - "loss": 3.5989, - "step": 143900 - }, - { - "epoch": 37.36, - "learning_rate": 1.0108977685521536e-05, - "loss": 3.6216, - "step": 144000 - }, - { - "epoch": 37.39, - "learning_rate": 1.0088220031136481e-05, - "loss": 3.5965, - "step": 144100 - }, - { - "epoch": 37.42, - "learning_rate": 1.006746237675143e-05, - "loss": 3.6071, - "step": 144200 - }, - { - "epoch": 37.44, - "learning_rate": 1.0046704722366375e-05, - "loss": 3.6013, - "step": 144300 - }, - { - "epoch": 37.47, - "learning_rate": 1.0025947067981318e-05, - "loss": 3.6158, - "step": 144400 - }, - { - "epoch": 37.49, - "learning_rate": 1.0005189413596263e-05, - "loss": 3.601, - "step": 144500 - }, - { - "epoch": 37.52, - "learning_rate": 9.98443175921121e-06, - "loss": 3.5991, - "step": 144600 - }, - { - "epoch": 37.55, - "learning_rate": 9.963674104826156e-06, - "loss": 3.5952, - "step": 144700 - }, - { - "epoch": 37.57, - "learning_rate": 9.942916450441101e-06, - "loss": 3.6145, - "step": 144800 - }, - { - "epoch": 37.6, - "learning_rate": 9.922158796056046e-06, - "loss": 3.5897, - "step": 144900 - }, - { - "epoch": 37.62, - "learning_rate": 9.901401141670993e-06, - "loss": 3.5978, - "step": 145000 - }, - { - "epoch": 37.62, - "eval_gen_len": 12.2774, - "eval_loss": 3.592960834503174, - "eval_rouge1": 41.3445, - "eval_rouge2": 8.728, - "eval_rougeL": 37.3647, - "eval_rougeLsum": 37.3694, - "eval_runtime": 75.9693, - "eval_samples_per_second": 50.731, - "eval_steps_per_second": 6.345, - "step": 145000 - }, - { - "epoch": 37.65, - "learning_rate": 9.880643487285938e-06, - "loss": 3.6027, - "step": 145100 - }, - { - "epoch": 37.68, - "learning_rate": 9.859885832900883e-06, - "loss": 3.5978, - "step": 145200 - }, - { - "epoch": 37.7, - "learning_rate": 9.83912817851583e-06, - "loss": 3.5987, - "step": 145300 - }, - { - "epoch": 37.73, - "learning_rate": 9.818370524130774e-06, - "loss": 3.6126, - "step": 145400 - }, - { - "epoch": 37.75, - "learning_rate": 9.79761286974572e-06, - "loss": 3.6129, - "step": 145500 - }, - { - "epoch": 37.78, - "learning_rate": 9.776855215360666e-06, - "loss": 3.6129, - "step": 145600 - }, - { - "epoch": 37.8, - "learning_rate": 9.756097560975611e-06, - "loss": 3.5921, - "step": 145700 - }, - { - "epoch": 37.83, - "learning_rate": 9.735339906590556e-06, - "loss": 3.6137, - "step": 145800 - }, - { - "epoch": 37.86, - "learning_rate": 9.714582252205501e-06, - "loss": 3.6208, - "step": 145900 - }, - { - "epoch": 37.88, - "learning_rate": 9.693824597820448e-06, - "loss": 3.6154, - "step": 146000 - }, - { - "epoch": 37.91, - "learning_rate": 9.673066943435393e-06, - "loss": 3.6112, - "step": 146100 - }, - { - "epoch": 37.93, - "learning_rate": 9.652309289050338e-06, - "loss": 3.5969, - "step": 146200 - }, - { - "epoch": 37.96, - "learning_rate": 9.631551634665284e-06, - "loss": 3.6015, - "step": 146300 - }, - { - "epoch": 37.99, - "learning_rate": 9.610793980280229e-06, - "loss": 3.6054, - "step": 146400 - }, - { - "epoch": 38.01, - "learning_rate": 9.590036325895174e-06, - "loss": 3.6147, - "step": 146500 - }, - { - "epoch": 38.04, - "learning_rate": 9.56927867151012e-06, - "loss": 3.5835, - "step": 146600 - }, - { - "epoch": 38.06, - "learning_rate": 9.548521017125066e-06, - "loss": 3.5983, - "step": 146700 - }, - { - "epoch": 38.09, - "learning_rate": 9.52776336274001e-06, - "loss": 3.6203, - "step": 146800 - }, - { - "epoch": 38.12, - "learning_rate": 9.507005708354957e-06, - "loss": 3.6003, - "step": 146900 - }, - { - "epoch": 38.14, - "learning_rate": 9.486248053969902e-06, - "loss": 3.5873, - "step": 147000 - }, - { - "epoch": 38.17, - "learning_rate": 9.465490399584847e-06, - "loss": 3.5896, - "step": 147100 - }, - { - "epoch": 38.19, - "learning_rate": 9.444732745199792e-06, - "loss": 3.6103, - "step": 147200 - }, - { - "epoch": 38.22, - "learning_rate": 9.423975090814739e-06, - "loss": 3.6003, - "step": 147300 - }, - { - "epoch": 38.25, - "learning_rate": 9.403217436429684e-06, - "loss": 3.6127, - "step": 147400 - }, - { - "epoch": 38.27, - "learning_rate": 9.382459782044629e-06, - "loss": 3.6222, - "step": 147500 - }, - { - "epoch": 38.3, - "learning_rate": 9.361702127659576e-06, - "loss": 3.5931, - "step": 147600 - }, - { - "epoch": 38.32, - "learning_rate": 9.34094447327452e-06, - "loss": 3.5879, - "step": 147700 - }, - { - "epoch": 38.35, - "learning_rate": 9.320186818889465e-06, - "loss": 3.6139, - "step": 147800 - }, - { - "epoch": 38.38, - "learning_rate": 9.299429164504412e-06, - "loss": 3.5875, - "step": 147900 - }, - { - "epoch": 38.4, - "learning_rate": 9.278671510119357e-06, - "loss": 3.6209, - "step": 148000 - }, - { - "epoch": 38.43, - "learning_rate": 9.257913855734302e-06, - "loss": 3.6078, - "step": 148100 - }, - { - "epoch": 38.45, - "learning_rate": 9.237156201349249e-06, - "loss": 3.6101, - "step": 148200 - }, - { - "epoch": 38.48, - "learning_rate": 9.216398546964194e-06, - "loss": 3.6003, - "step": 148300 - }, - { - "epoch": 38.51, - "learning_rate": 9.19564089257914e-06, - "loss": 3.5993, - "step": 148400 - }, - { - "epoch": 38.53, - "learning_rate": 9.174883238194084e-06, - "loss": 3.5977, - "step": 148500 - }, - { - "epoch": 38.56, - "learning_rate": 9.15412558380903e-06, - "loss": 3.6018, - "step": 148600 - }, - { - "epoch": 38.58, - "learning_rate": 9.133367929423977e-06, - "loss": 3.5985, - "step": 148700 - }, - { - "epoch": 38.61, - "learning_rate": 9.112610275038922e-06, - "loss": 3.6143, - "step": 148800 - }, - { - "epoch": 38.64, - "learning_rate": 9.091852620653867e-06, - "loss": 3.6048, - "step": 148900 - }, - { - "epoch": 38.66, - "learning_rate": 9.071094966268812e-06, - "loss": 3.6186, - "step": 149000 - }, - { - "epoch": 38.69, - "learning_rate": 9.050337311883759e-06, - "loss": 3.5875, - "step": 149100 - }, - { - "epoch": 38.71, - "learning_rate": 9.029579657498703e-06, - "loss": 3.6165, - "step": 149200 - }, - { - "epoch": 38.74, - "learning_rate": 9.008822003113648e-06, - "loss": 3.595, - "step": 149300 - }, - { - "epoch": 38.76, - "learning_rate": 8.988064348728595e-06, - "loss": 3.6045, - "step": 149400 - }, - { - "epoch": 38.79, - "learning_rate": 8.96730669434354e-06, - "loss": 3.5736, - "step": 149500 - }, - { - "epoch": 38.82, - "learning_rate": 8.946549039958485e-06, - "loss": 3.6124, - "step": 149600 - }, - { - "epoch": 38.84, - "learning_rate": 8.925791385573432e-06, - "loss": 3.5944, - "step": 149700 - }, - { - "epoch": 38.87, - "learning_rate": 8.905033731188377e-06, - "loss": 3.6054, - "step": 149800 - }, - { - "epoch": 38.89, - "learning_rate": 8.884276076803322e-06, - "loss": 3.602, - "step": 149900 - }, - { - "epoch": 38.92, - "learning_rate": 8.863518422418268e-06, - "loss": 3.6121, - "step": 150000 - }, - { - "epoch": 38.92, - "eval_gen_len": 12.2769, - "eval_loss": 3.591606855392456, - "eval_rouge1": 41.9288, - "eval_rouge2": 8.6859, - "eval_rougeL": 37.8925, - "eval_rougeLsum": 37.9125, - "eval_runtime": 75.6977, - "eval_samples_per_second": 50.913, - "eval_steps_per_second": 6.367, - "step": 150000 - }, - { - "epoch": 38.95, - "learning_rate": 8.842760768033213e-06, - "loss": 3.617, - "step": 150100 - }, - { - "epoch": 38.97, - "learning_rate": 8.822003113648158e-06, - "loss": 3.5798, - "step": 150200 - }, - { - "epoch": 39.0, - "learning_rate": 8.801245459263103e-06, - "loss": 3.5933, - "step": 150300 - }, - { - "epoch": 39.02, - "learning_rate": 8.78048780487805e-06, - "loss": 3.594, - "step": 150400 - }, - { - "epoch": 39.05, - "learning_rate": 8.759730150492995e-06, - "loss": 3.6042, - "step": 150500 - }, - { - "epoch": 39.08, - "learning_rate": 8.73897249610794e-06, - "loss": 3.6139, - "step": 150600 - }, - { - "epoch": 39.1, - "learning_rate": 8.718214841722886e-06, - "loss": 3.6119, - "step": 150700 - }, - { - "epoch": 39.13, - "learning_rate": 8.697457187337831e-06, - "loss": 3.5806, - "step": 150800 - }, - { - "epoch": 39.15, - "learning_rate": 8.676699532952776e-06, - "loss": 3.5956, - "step": 150900 - }, - { - "epoch": 39.18, - "learning_rate": 8.655941878567723e-06, - "loss": 3.5559, - "step": 151000 - }, - { - "epoch": 39.21, - "learning_rate": 8.635184224182668e-06, - "loss": 3.6148, - "step": 151100 - }, - { - "epoch": 39.23, - "learning_rate": 8.614426569797613e-06, - "loss": 3.5941, - "step": 151200 - }, - { - "epoch": 39.26, - "learning_rate": 8.593668915412558e-06, - "loss": 3.5734, - "step": 151300 - }, - { - "epoch": 39.28, - "learning_rate": 8.572911261027505e-06, - "loss": 3.5983, - "step": 151400 - }, - { - "epoch": 39.31, - "learning_rate": 8.55215360664245e-06, - "loss": 3.6197, - "step": 151500 - }, - { - "epoch": 39.34, - "learning_rate": 8.531395952257395e-06, - "loss": 3.6068, - "step": 151600 - }, - { - "epoch": 39.36, - "learning_rate": 8.510638297872341e-06, - "loss": 3.5989, - "step": 151700 - }, - { - "epoch": 39.39, - "learning_rate": 8.489880643487288e-06, - "loss": 3.6053, - "step": 151800 - }, - { - "epoch": 39.41, - "learning_rate": 8.469122989102231e-06, - "loss": 3.6161, - "step": 151900 - }, - { - "epoch": 39.44, - "learning_rate": 8.448365334717178e-06, - "loss": 3.6292, - "step": 152000 - }, - { - "epoch": 39.47, - "learning_rate": 8.427607680332123e-06, - "loss": 3.6121, - "step": 152100 - }, - { - "epoch": 39.49, - "learning_rate": 8.40685002594707e-06, - "loss": 3.6004, - "step": 152200 - }, - { - "epoch": 39.52, - "learning_rate": 8.386092371562014e-06, - "loss": 3.6157, - "step": 152300 - }, - { - "epoch": 39.54, - "learning_rate": 8.36533471717696e-06, - "loss": 3.622, - "step": 152400 - }, - { - "epoch": 39.57, - "learning_rate": 8.344577062791906e-06, - "loss": 3.6119, - "step": 152500 - }, - { - "epoch": 39.6, - "learning_rate": 8.323819408406851e-06, - "loss": 3.6092, - "step": 152600 - }, - { - "epoch": 39.62, - "learning_rate": 8.303061754021796e-06, - "loss": 3.5876, - "step": 152700 - }, - { - "epoch": 39.65, - "learning_rate": 8.282304099636743e-06, - "loss": 3.6196, - "step": 152800 - }, - { - "epoch": 39.67, - "learning_rate": 8.261546445251688e-06, - "loss": 3.6121, - "step": 152900 - }, - { - "epoch": 39.7, - "learning_rate": 8.240788790866633e-06, - "loss": 3.6184, - "step": 153000 - }, - { - "epoch": 39.72, - "learning_rate": 8.22003113648158e-06, - "loss": 3.5772, - "step": 153100 - }, - { - "epoch": 39.75, - "learning_rate": 8.199273482096524e-06, - "loss": 3.6062, - "step": 153200 - }, - { - "epoch": 39.78, - "learning_rate": 8.17851582771147e-06, - "loss": 3.6152, - "step": 153300 - }, - { - "epoch": 39.8, - "learning_rate": 8.157758173326414e-06, - "loss": 3.5901, - "step": 153400 - }, - { - "epoch": 39.83, - "learning_rate": 8.13700051894136e-06, - "loss": 3.5957, - "step": 153500 - }, - { - "epoch": 39.85, - "learning_rate": 8.116242864556306e-06, - "loss": 3.6184, - "step": 153600 - }, - { - "epoch": 39.88, - "learning_rate": 8.09548521017125e-06, - "loss": 3.5991, - "step": 153700 - }, - { - "epoch": 39.91, - "learning_rate": 8.074727555786197e-06, - "loss": 3.6069, - "step": 153800 - }, - { - "epoch": 39.93, - "learning_rate": 8.053969901401142e-06, - "loss": 3.5976, - "step": 153900 - }, - { - "epoch": 39.96, - "learning_rate": 8.033212247016087e-06, - "loss": 3.5954, - "step": 154000 - }, - { - "epoch": 39.98, - "learning_rate": 8.012454592631034e-06, - "loss": 3.6192, - "step": 154100 - }, - { - "epoch": 40.01, - "learning_rate": 7.991696938245979e-06, - "loss": 3.5992, - "step": 154200 - }, - { - "epoch": 40.04, - "learning_rate": 7.970939283860924e-06, - "loss": 3.6007, - "step": 154300 - }, - { - "epoch": 40.06, - "learning_rate": 7.950181629475869e-06, - "loss": 3.5923, - "step": 154400 - }, - { - "epoch": 40.09, - "learning_rate": 7.929423975090816e-06, - "loss": 3.6094, - "step": 154500 - }, - { - "epoch": 40.11, - "learning_rate": 7.90866632070576e-06, - "loss": 3.618, - "step": 154600 - }, - { - "epoch": 40.14, - "learning_rate": 7.887908666320706e-06, - "loss": 3.6084, - "step": 154700 - }, - { - "epoch": 40.17, - "learning_rate": 7.867151011935652e-06, - "loss": 3.5987, - "step": 154800 - }, - { - "epoch": 40.19, - "learning_rate": 7.846393357550597e-06, - "loss": 3.5991, - "step": 154900 - }, - { - "epoch": 40.22, - "learning_rate": 7.825635703165542e-06, - "loss": 3.6154, - "step": 155000 - }, - { - "epoch": 40.22, - "eval_gen_len": 12.2763, - "eval_loss": 3.590810775756836, - "eval_rouge1": 41.7854, - "eval_rouge2": 8.6556, - "eval_rougeL": 37.7725, - "eval_rougeLsum": 37.7749, - "eval_runtime": 76.5307, - "eval_samples_per_second": 50.359, - "eval_steps_per_second": 6.298, - "step": 155000 - }, - { - "epoch": 40.24, - "learning_rate": 7.804878048780489e-06, - "loss": 3.626, - "step": 155100 - }, - { - "epoch": 40.27, - "learning_rate": 7.784120394395434e-06, - "loss": 3.6092, - "step": 155200 - }, - { - "epoch": 40.3, - "learning_rate": 7.763362740010379e-06, - "loss": 3.5901, - "step": 155300 - }, - { - "epoch": 40.32, - "learning_rate": 7.742605085625325e-06, - "loss": 3.6067, - "step": 155400 - }, - { - "epoch": 40.35, - "learning_rate": 7.72184743124027e-06, - "loss": 3.6106, - "step": 155500 - }, - { - "epoch": 40.37, - "learning_rate": 7.701089776855217e-06, - "loss": 3.6024, - "step": 155600 - }, - { - "epoch": 40.4, - "learning_rate": 7.68033212247016e-06, - "loss": 3.6023, - "step": 155700 - }, - { - "epoch": 40.43, - "learning_rate": 7.659574468085107e-06, - "loss": 3.5921, - "step": 155800 - }, - { - "epoch": 40.45, - "learning_rate": 7.638816813700054e-06, - "loss": 3.6076, - "step": 155900 - }, - { - "epoch": 40.48, - "learning_rate": 7.618059159314998e-06, - "loss": 3.6178, - "step": 156000 - }, - { - "epoch": 40.5, - "learning_rate": 7.5973015049299435e-06, - "loss": 3.6246, - "step": 156100 - }, - { - "epoch": 40.53, - "learning_rate": 7.5765438505448885e-06, - "loss": 3.591, - "step": 156200 - }, - { - "epoch": 40.56, - "learning_rate": 7.555786196159834e-06, - "loss": 3.5834, - "step": 156300 - }, - { - "epoch": 40.58, - "learning_rate": 7.53502854177478e-06, - "loss": 3.6077, - "step": 156400 - }, - { - "epoch": 40.61, - "learning_rate": 7.514270887389725e-06, - "loss": 3.6109, - "step": 156500 - }, - { - "epoch": 40.63, - "learning_rate": 7.493513233004671e-06, - "loss": 3.6024, - "step": 156600 - }, - { - "epoch": 40.66, - "learning_rate": 7.472755578619617e-06, - "loss": 3.6082, - "step": 156700 - }, - { - "epoch": 40.69, - "learning_rate": 7.451997924234562e-06, - "loss": 3.6214, - "step": 156800 - }, - { - "epoch": 40.71, - "learning_rate": 7.4312402698495075e-06, - "loss": 3.5791, - "step": 156900 - }, - { - "epoch": 40.74, - "learning_rate": 7.4104826154644525e-06, - "loss": 3.6122, - "step": 157000 - }, - { - "epoch": 40.76, - "learning_rate": 7.389724961079398e-06, - "loss": 3.6106, - "step": 157100 - }, - { - "epoch": 40.79, - "learning_rate": 7.368967306694345e-06, - "loss": 3.6073, - "step": 157200 - }, - { - "epoch": 40.81, - "learning_rate": 7.348209652309289e-06, - "loss": 3.6024, - "step": 157300 - }, - { - "epoch": 40.84, - "learning_rate": 7.327451997924236e-06, - "loss": 3.5964, - "step": 157400 - }, - { - "epoch": 40.87, - "learning_rate": 7.30669434353918e-06, - "loss": 3.5994, - "step": 157500 - }, - { - "epoch": 40.89, - "learning_rate": 7.2859366891541265e-06, - "loss": 3.6074, - "step": 157600 - }, - { - "epoch": 40.92, - "learning_rate": 7.265179034769072e-06, - "loss": 3.5946, - "step": 157700 - }, - { - "epoch": 40.94, - "learning_rate": 7.2444213803840165e-06, - "loss": 3.5902, - "step": 157800 - }, - { - "epoch": 40.97, - "learning_rate": 7.223663725998963e-06, - "loss": 3.59, - "step": 157900 - }, - { - "epoch": 41.0, - "learning_rate": 7.202906071613909e-06, - "loss": 3.6009, - "step": 158000 - }, - { - "epoch": 41.02, - "learning_rate": 7.182148417228854e-06, - "loss": 3.5957, - "step": 158100 - }, - { - "epoch": 41.05, - "learning_rate": 7.1613907628438e-06, - "loss": 3.5913, - "step": 158200 - }, - { - "epoch": 41.07, - "learning_rate": 7.140633108458745e-06, - "loss": 3.5876, - "step": 158300 - }, - { - "epoch": 41.1, - "learning_rate": 7.1198754540736905e-06, - "loss": 3.6022, - "step": 158400 - }, - { - "epoch": 41.13, - "learning_rate": 7.099117799688636e-06, - "loss": 3.6123, - "step": 158500 - }, - { - "epoch": 41.15, - "learning_rate": 7.078360145303581e-06, - "loss": 3.5919, - "step": 158600 - }, - { - "epoch": 41.18, - "learning_rate": 7.057602490918527e-06, - "loss": 3.6186, - "step": 158700 - }, - { - "epoch": 41.2, - "learning_rate": 7.036844836533472e-06, - "loss": 3.6035, - "step": 158800 - }, - { - "epoch": 41.23, - "learning_rate": 7.016087182148418e-06, - "loss": 3.5958, - "step": 158900 - }, - { - "epoch": 41.26, - "learning_rate": 6.995329527763364e-06, - "loss": 3.5873, - "step": 159000 - }, - { - "epoch": 41.28, - "learning_rate": 6.974571873378309e-06, - "loss": 3.6196, - "step": 159100 - }, - { - "epoch": 41.31, - "learning_rate": 6.9538142189932545e-06, - "loss": 3.6018, - "step": 159200 - }, - { - "epoch": 41.33, - "learning_rate": 6.9330565646081994e-06, - "loss": 3.5975, - "step": 159300 - }, - { - "epoch": 41.36, - "learning_rate": 6.912298910223145e-06, - "loss": 3.5927, - "step": 159400 - }, - { - "epoch": 41.39, - "learning_rate": 6.891541255838091e-06, - "loss": 3.6108, - "step": 159500 - }, - { - "epoch": 41.41, - "learning_rate": 6.870783601453036e-06, - "loss": 3.617, - "step": 159600 - }, - { - "epoch": 41.44, - "learning_rate": 6.850025947067982e-06, - "loss": 3.6035, - "step": 159700 - }, - { - "epoch": 41.46, - "learning_rate": 6.829268292682928e-06, - "loss": 3.6088, - "step": 159800 - }, - { - "epoch": 41.49, - "learning_rate": 6.808510638297873e-06, - "loss": 3.6014, - "step": 159900 - }, - { - "epoch": 41.52, - "learning_rate": 6.7877529839128185e-06, - "loss": 3.574, - "step": 160000 - }, - { - "epoch": 41.52, - "eval_gen_len": 12.2779, - "eval_loss": 3.5903589725494385, - "eval_rouge1": 41.8234, - "eval_rouge2": 8.7383, - "eval_rougeL": 37.8126, - "eval_rougeLsum": 37.813, - "eval_runtime": 76.3545, - "eval_samples_per_second": 50.475, - "eval_steps_per_second": 6.313, - "step": 160000 - }, - { - "epoch": 41.54, - "learning_rate": 6.7669953295277634e-06, - "loss": 3.6005, - "step": 160100 - }, - { - "epoch": 41.57, - "learning_rate": 6.746237675142709e-06, - "loss": 3.571, - "step": 160200 - }, - { - "epoch": 41.59, - "learning_rate": 6.725480020757655e-06, - "loss": 3.616, - "step": 160300 - }, - { - "epoch": 41.62, - "learning_rate": 6.7047223663726e-06, - "loss": 3.605, - "step": 160400 - }, - { - "epoch": 41.65, - "learning_rate": 6.683964711987546e-06, - "loss": 3.5991, - "step": 160500 - }, - { - "epoch": 41.67, - "learning_rate": 6.663207057602491e-06, - "loss": 3.6063, - "step": 160600 - }, - { - "epoch": 41.7, - "learning_rate": 6.642449403217437e-06, - "loss": 3.5771, - "step": 160700 - }, - { - "epoch": 41.72, - "learning_rate": 6.6216917488323824e-06, - "loss": 3.6011, - "step": 160800 - }, - { - "epoch": 41.75, - "learning_rate": 6.600934094447327e-06, - "loss": 3.6073, - "step": 160900 - }, - { - "epoch": 41.77, - "learning_rate": 6.580176440062273e-06, - "loss": 3.6129, - "step": 161000 - }, - { - "epoch": 41.8, - "learning_rate": 6.559418785677218e-06, - "loss": 3.5917, - "step": 161100 - }, - { - "epoch": 41.83, - "learning_rate": 6.538661131292164e-06, - "loss": 3.6049, - "step": 161200 - }, - { - "epoch": 41.85, - "learning_rate": 6.517903476907111e-06, - "loss": 3.5975, - "step": 161300 - }, - { - "epoch": 41.88, - "learning_rate": 6.497145822522055e-06, - "loss": 3.5925, - "step": 161400 - }, - { - "epoch": 41.9, - "learning_rate": 6.4763881681370014e-06, - "loss": 3.6042, - "step": 161500 - }, - { - "epoch": 41.93, - "learning_rate": 6.455630513751947e-06, - "loss": 3.5978, - "step": 161600 - }, - { - "epoch": 41.96, - "learning_rate": 6.434872859366892e-06, - "loss": 3.5942, - "step": 161700 - }, - { - "epoch": 41.98, - "learning_rate": 6.414115204981838e-06, - "loss": 3.6045, - "step": 161800 - }, - { - "epoch": 42.01, - "learning_rate": 6.393357550596783e-06, - "loss": 3.5946, - "step": 161900 - }, - { - "epoch": 42.03, - "learning_rate": 6.372599896211729e-06, - "loss": 3.6214, - "step": 162000 - }, - { - "epoch": 42.06, - "learning_rate": 6.351842241826675e-06, - "loss": 3.6172, - "step": 162100 - }, - { - "epoch": 42.09, - "learning_rate": 6.33108458744162e-06, - "loss": 3.6064, - "step": 162200 - }, - { - "epoch": 42.11, - "learning_rate": 6.3103269330565654e-06, - "loss": 3.5931, - "step": 162300 - }, - { - "epoch": 42.14, - "learning_rate": 6.28956927867151e-06, - "loss": 3.5975, - "step": 162400 - }, - { - "epoch": 42.16, - "learning_rate": 6.268811624286456e-06, - "loss": 3.5952, - "step": 162500 - }, - { - "epoch": 42.19, - "learning_rate": 6.248053969901402e-06, - "loss": 3.6057, - "step": 162600 - }, - { - "epoch": 42.22, - "learning_rate": 6.227296315516347e-06, - "loss": 3.5965, - "step": 162700 - }, - { - "epoch": 42.24, - "learning_rate": 6.206538661131293e-06, - "loss": 3.6234, - "step": 162800 - }, - { - "epoch": 42.27, - "learning_rate": 6.185781006746239e-06, - "loss": 3.5977, - "step": 162900 - }, - { - "epoch": 42.29, - "learning_rate": 6.165023352361184e-06, - "loss": 3.6099, - "step": 163000 - }, - { - "epoch": 42.32, - "learning_rate": 6.144265697976129e-06, - "loss": 3.5625, - "step": 163100 - }, - { - "epoch": 42.35, - "learning_rate": 6.123508043591074e-06, - "loss": 3.6009, - "step": 163200 - }, - { - "epoch": 42.37, - "learning_rate": 6.10275038920602e-06, - "loss": 3.5834, - "step": 163300 - }, - { - "epoch": 42.4, - "learning_rate": 6.081992734820966e-06, - "loss": 3.564, - "step": 163400 - }, - { - "epoch": 42.42, - "learning_rate": 6.061235080435911e-06, - "loss": 3.6205, - "step": 163500 - }, - { - "epoch": 42.45, - "learning_rate": 6.040477426050857e-06, - "loss": 3.5879, - "step": 163600 - }, - { - "epoch": 42.48, - "learning_rate": 6.019719771665802e-06, - "loss": 3.6, - "step": 163700 - }, - { - "epoch": 42.5, - "learning_rate": 5.9989621172807476e-06, - "loss": 3.5986, - "step": 163800 - }, - { - "epoch": 42.53, - "learning_rate": 5.978204462895693e-06, - "loss": 3.6023, - "step": 163900 - }, - { - "epoch": 42.55, - "learning_rate": 5.957446808510638e-06, - "loss": 3.6012, - "step": 164000 - }, - { - "epoch": 42.58, - "learning_rate": 5.936689154125584e-06, - "loss": 3.5837, - "step": 164100 - }, - { - "epoch": 42.61, - "learning_rate": 5.915931499740529e-06, - "loss": 3.6029, - "step": 164200 - }, - { - "epoch": 42.63, - "learning_rate": 5.895173845355475e-06, - "loss": 3.6119, - "step": 164300 - }, - { - "epoch": 42.66, - "learning_rate": 5.874416190970421e-06, - "loss": 3.5922, - "step": 164400 - }, - { - "epoch": 42.68, - "learning_rate": 5.853658536585366e-06, - "loss": 3.6092, - "step": 164500 - }, - { - "epoch": 42.71, - "learning_rate": 5.8329008822003115e-06, - "loss": 3.5925, - "step": 164600 - }, - { - "epoch": 42.73, - "learning_rate": 5.812143227815258e-06, - "loss": 3.6074, - "step": 164700 - }, - { - "epoch": 42.76, - "learning_rate": 5.791385573430202e-06, - "loss": 3.6085, - "step": 164800 - }, - { - "epoch": 42.79, - "learning_rate": 5.770627919045149e-06, - "loss": 3.6101, - "step": 164900 - }, - { - "epoch": 42.81, - "learning_rate": 5.749870264660093e-06, - "loss": 3.5595, - "step": 165000 - }, - { - "epoch": 42.81, - "eval_gen_len": 12.2761, - "eval_loss": 3.5897645950317383, - "eval_rouge1": 41.7274, - "eval_rouge2": 8.7877, - "eval_rougeL": 37.7333, - "eval_rougeLsum": 37.7388, - "eval_runtime": 76.3561, - "eval_samples_per_second": 50.474, - "eval_steps_per_second": 6.313, - "step": 165000 - }, - { - "epoch": 42.84, - "learning_rate": 5.72911261027504e-06, - "loss": 3.624, - "step": 165100 - }, - { - "epoch": 42.86, - "learning_rate": 5.708354955889986e-06, - "loss": 3.6039, - "step": 165200 - }, - { - "epoch": 42.89, - "learning_rate": 5.6875973015049306e-06, - "loss": 3.6057, - "step": 165300 - }, - { - "epoch": 42.92, - "learning_rate": 5.666839647119876e-06, - "loss": 3.5824, - "step": 165400 - }, - { - "epoch": 42.94, - "learning_rate": 5.646081992734821e-06, - "loss": 3.5717, - "step": 165500 - }, - { - "epoch": 42.97, - "learning_rate": 5.625324338349767e-06, - "loss": 3.5933, - "step": 165600 - }, - { - "epoch": 42.99, - "learning_rate": 5.604566683964713e-06, - "loss": 3.5948, - "step": 165700 - }, - { - "epoch": 43.02, - "learning_rate": 5.583809029579658e-06, - "loss": 3.5862, - "step": 165800 - }, - { - "epoch": 43.05, - "learning_rate": 5.563051375194604e-06, - "loss": 3.5895, - "step": 165900 - }, - { - "epoch": 43.07, - "learning_rate": 5.5422937208095496e-06, - "loss": 3.6074, - "step": 166000 - }, - { - "epoch": 43.1, - "learning_rate": 5.5215360664244945e-06, - "loss": 3.6064, - "step": 166100 - }, - { - "epoch": 43.12, - "learning_rate": 5.50077841203944e-06, - "loss": 3.6145, - "step": 166200 - }, - { - "epoch": 43.15, - "learning_rate": 5.480020757654385e-06, - "loss": 3.5731, - "step": 166300 - }, - { - "epoch": 43.18, - "learning_rate": 5.459263103269331e-06, - "loss": 3.6008, - "step": 166400 - }, - { - "epoch": 43.2, - "learning_rate": 5.438505448884277e-06, - "loss": 3.6109, - "step": 166500 - }, - { - "epoch": 43.23, - "learning_rate": 5.417747794499222e-06, - "loss": 3.6112, - "step": 166600 - }, - { - "epoch": 43.25, - "learning_rate": 5.396990140114168e-06, - "loss": 3.5723, - "step": 166700 - }, - { - "epoch": 43.28, - "learning_rate": 5.376232485729113e-06, - "loss": 3.6063, - "step": 166800 - }, - { - "epoch": 43.31, - "learning_rate": 5.3554748313440585e-06, - "loss": 3.6037, - "step": 166900 - }, - { - "epoch": 43.33, - "learning_rate": 5.334717176959004e-06, - "loss": 3.5815, - "step": 167000 - }, - { - "epoch": 43.36, - "learning_rate": 5.313959522573949e-06, - "loss": 3.6042, - "step": 167100 - }, - { - "epoch": 43.38, - "learning_rate": 5.293201868188895e-06, - "loss": 3.621, - "step": 167200 - }, - { - "epoch": 43.41, - "learning_rate": 5.27244421380384e-06, - "loss": 3.5913, - "step": 167300 - }, - { - "epoch": 43.44, - "learning_rate": 5.251686559418786e-06, - "loss": 3.5908, - "step": 167400 - }, - { - "epoch": 43.46, - "learning_rate": 5.230928905033732e-06, - "loss": 3.5799, - "step": 167500 - }, - { - "epoch": 43.49, - "learning_rate": 5.210171250648677e-06, - "loss": 3.5886, - "step": 167600 - }, - { - "epoch": 43.51, - "learning_rate": 5.1894135962636225e-06, - "loss": 3.6089, - "step": 167700 - }, - { - "epoch": 43.54, - "learning_rate": 5.168655941878568e-06, - "loss": 3.5928, - "step": 167800 - }, - { - "epoch": 43.57, - "learning_rate": 5.147898287493513e-06, - "loss": 3.5681, - "step": 167900 - }, - { - "epoch": 43.59, - "learning_rate": 5.127140633108459e-06, - "loss": 3.611, - "step": 168000 - }, - { - "epoch": 43.62, - "learning_rate": 5.106382978723404e-06, - "loss": 3.612, - "step": 168100 - }, - { - "epoch": 43.64, - "learning_rate": 5.08562532433835e-06, - "loss": 3.5959, - "step": 168200 - }, - { - "epoch": 43.67, - "learning_rate": 5.0648676699532965e-06, - "loss": 3.5898, - "step": 168300 - }, - { - "epoch": 43.69, - "learning_rate": 5.044110015568241e-06, - "loss": 3.5836, - "step": 168400 - }, - { - "epoch": 43.72, - "learning_rate": 5.023352361183187e-06, - "loss": 3.5974, - "step": 168500 - }, - { - "epoch": 43.75, - "learning_rate": 5.0025947067981314e-06, - "loss": 3.6092, - "step": 168600 - }, - { - "epoch": 43.77, - "learning_rate": 4.981837052413078e-06, - "loss": 3.6135, - "step": 168700 - }, - { - "epoch": 43.8, - "learning_rate": 4.961079398028023e-06, - "loss": 3.6055, - "step": 168800 - }, - { - "epoch": 43.82, - "learning_rate": 4.940321743642969e-06, - "loss": 3.5622, - "step": 168900 - }, - { - "epoch": 43.85, - "learning_rate": 4.919564089257915e-06, - "loss": 3.6091, - "step": 169000 - }, - { - "epoch": 43.88, - "learning_rate": 4.89880643487286e-06, - "loss": 3.5868, - "step": 169100 - }, - { - "epoch": 43.9, - "learning_rate": 4.8780487804878055e-06, - "loss": 3.6074, - "step": 169200 - }, - { - "epoch": 43.93, - "learning_rate": 4.8572911261027505e-06, - "loss": 3.6067, - "step": 169300 - }, - { - "epoch": 43.95, - "learning_rate": 4.836533471717696e-06, - "loss": 3.604, - "step": 169400 - }, - { - "epoch": 43.98, - "learning_rate": 4.815775817332642e-06, - "loss": 3.5873, - "step": 169500 - }, - { - "epoch": 44.01, - "learning_rate": 4.795018162947587e-06, - "loss": 3.6056, - "step": 169600 - }, - { - "epoch": 44.03, - "learning_rate": 4.774260508562533e-06, - "loss": 3.6109, - "step": 169700 - }, - { - "epoch": 44.06, - "learning_rate": 4.753502854177479e-06, - "loss": 3.6052, - "step": 169800 - }, - { - "epoch": 44.08, - "learning_rate": 4.732745199792424e-06, - "loss": 3.6063, - "step": 169900 - }, - { - "epoch": 44.11, - "learning_rate": 4.7119875454073695e-06, - "loss": 3.5897, - "step": 170000 - }, - { - "epoch": 44.11, - "eval_gen_len": 12.2763, - "eval_loss": 3.588927984237671, - "eval_rouge1": 41.8372, - "eval_rouge2": 8.7034, - "eval_rougeL": 37.8307, - "eval_rougeLsum": 37.841, - "eval_runtime": 89.0261, - "eval_samples_per_second": 43.291, - "eval_steps_per_second": 5.414, - "step": 170000 - }, - { - "epoch": 44.14, - "learning_rate": 4.6912298910223144e-06, - "loss": 3.5712, - "step": 170100 - }, - { - "epoch": 44.16, - "learning_rate": 4.67047223663726e-06, - "loss": 3.6044, - "step": 170200 - }, - { - "epoch": 44.19, - "learning_rate": 4.649714582252206e-06, - "loss": 3.6017, - "step": 170300 - }, - { - "epoch": 44.21, - "learning_rate": 4.628956927867151e-06, - "loss": 3.5832, - "step": 170400 - }, - { - "epoch": 44.24, - "learning_rate": 4.608199273482097e-06, - "loss": 3.6089, - "step": 170500 - }, - { - "epoch": 44.27, - "learning_rate": 4.587441619097042e-06, - "loss": 3.6157, - "step": 170600 - }, - { - "epoch": 44.29, - "learning_rate": 4.5666839647119885e-06, - "loss": 3.586, - "step": 170700 - }, - { - "epoch": 44.32, - "learning_rate": 4.5459263103269334e-06, - "loss": 3.6016, - "step": 170800 - }, - { - "epoch": 44.34, - "learning_rate": 4.525168655941879e-06, - "loss": 3.6018, - "step": 170900 - }, - { - "epoch": 44.37, - "learning_rate": 4.504411001556824e-06, - "loss": 3.5958, - "step": 171000 - }, - { - "epoch": 44.4, - "learning_rate": 4.48365334717177e-06, - "loss": 3.5539, - "step": 171100 - }, - { - "epoch": 44.42, - "learning_rate": 4.462895692786716e-06, - "loss": 3.5958, - "step": 171200 - }, - { - "epoch": 44.45, - "learning_rate": 4.442138038401661e-06, - "loss": 3.5845, - "step": 171300 - }, - { - "epoch": 44.47, - "learning_rate": 4.421380384016607e-06, - "loss": 3.5792, - "step": 171400 - }, - { - "epoch": 44.5, - "learning_rate": 4.400622729631552e-06, - "loss": 3.6142, - "step": 171500 - }, - { - "epoch": 44.53, - "learning_rate": 4.379865075246497e-06, - "loss": 3.6097, - "step": 171600 - }, - { - "epoch": 44.55, - "learning_rate": 4.359107420861443e-06, - "loss": 3.6089, - "step": 171700 - }, - { - "epoch": 44.58, - "learning_rate": 4.338349766476388e-06, - "loss": 3.6002, - "step": 171800 - }, - { - "epoch": 44.6, - "learning_rate": 4.317592112091334e-06, - "loss": 3.5973, - "step": 171900 - }, - { - "epoch": 44.63, - "learning_rate": 4.296834457706279e-06, - "loss": 3.6005, - "step": 172000 - }, - { - "epoch": 44.65, - "learning_rate": 4.276076803321225e-06, - "loss": 3.6128, - "step": 172100 - }, - { - "epoch": 44.68, - "learning_rate": 4.255319148936171e-06, - "loss": 3.5831, - "step": 172200 - }, - { - "epoch": 44.71, - "learning_rate": 4.234561494551116e-06, - "loss": 3.6096, - "step": 172300 - }, - { - "epoch": 44.73, - "learning_rate": 4.213803840166061e-06, - "loss": 3.5979, - "step": 172400 - }, - { - "epoch": 44.76, - "learning_rate": 4.193046185781007e-06, - "loss": 3.6007, - "step": 172500 - }, - { - "epoch": 44.78, - "learning_rate": 4.172288531395953e-06, - "loss": 3.5962, - "step": 172600 - }, - { - "epoch": 44.81, - "learning_rate": 4.151530877010898e-06, - "loss": 3.6065, - "step": 172700 - }, - { - "epoch": 44.84, - "learning_rate": 4.130773222625844e-06, - "loss": 3.5907, - "step": 172800 - }, - { - "epoch": 44.86, - "learning_rate": 4.11001556824079e-06, - "loss": 3.5942, - "step": 172900 - }, - { - "epoch": 44.89, - "learning_rate": 4.089257913855735e-06, - "loss": 3.6046, - "step": 173000 - }, - { - "epoch": 44.91, - "learning_rate": 4.06850025947068e-06, - "loss": 3.5897, - "step": 173100 - }, - { - "epoch": 44.94, - "learning_rate": 4.047742605085625e-06, - "loss": 3.6068, - "step": 173200 - }, - { - "epoch": 44.97, - "learning_rate": 4.026984950700571e-06, - "loss": 3.5986, - "step": 173300 - }, - { - "epoch": 44.99, - "learning_rate": 4.006227296315517e-06, - "loss": 3.6101, - "step": 173400 - }, - { - "epoch": 45.02, - "learning_rate": 3.985469641930462e-06, - "loss": 3.6068, - "step": 173500 - }, - { - "epoch": 45.04, - "learning_rate": 3.964711987545408e-06, - "loss": 3.6006, - "step": 173600 - }, - { - "epoch": 45.07, - "learning_rate": 3.943954333160353e-06, - "loss": 3.5955, - "step": 173700 - }, - { - "epoch": 45.1, - "learning_rate": 3.9231966787752986e-06, - "loss": 3.6147, - "step": 173800 - }, - { - "epoch": 45.12, - "learning_rate": 3.902439024390244e-06, - "loss": 3.6057, - "step": 173900 - }, - { - "epoch": 45.15, - "learning_rate": 3.881681370005189e-06, - "loss": 3.6, - "step": 174000 - }, - { - "epoch": 45.17, - "learning_rate": 3.860923715620135e-06, - "loss": 3.6016, - "step": 174100 - }, - { - "epoch": 45.2, - "learning_rate": 3.84016606123508e-06, - "loss": 3.6131, - "step": 174200 - }, - { - "epoch": 45.23, - "learning_rate": 3.819408406850027e-06, - "loss": 3.6014, - "step": 174300 - }, - { - "epoch": 45.25, - "learning_rate": 3.7986507524649718e-06, - "loss": 3.6189, - "step": 174400 - }, - { - "epoch": 45.28, - "learning_rate": 3.777893098079917e-06, - "loss": 3.601, - "step": 174500 - }, - { - "epoch": 45.3, - "learning_rate": 3.7571354436948626e-06, - "loss": 3.6047, - "step": 174600 - }, - { - "epoch": 45.33, - "learning_rate": 3.7363777893098084e-06, - "loss": 3.6055, - "step": 174700 - }, - { - "epoch": 45.36, - "learning_rate": 3.7156201349247538e-06, - "loss": 3.5924, - "step": 174800 - }, - { - "epoch": 45.38, - "learning_rate": 3.694862480539699e-06, - "loss": 3.5623, - "step": 174900 - }, - { - "epoch": 45.41, - "learning_rate": 3.6741048261546445e-06, - "loss": 3.603, - "step": 175000 - }, - { - "epoch": 45.41, - "eval_gen_len": 12.2761, - "eval_loss": 3.588848114013672, - "eval_rouge1": 41.7972, - "eval_rouge2": 8.7107, - "eval_rougeL": 37.7776, - "eval_rougeLsum": 37.818, - "eval_runtime": 75.5493, - "eval_samples_per_second": 51.013, - "eval_steps_per_second": 6.38, - "step": 175000 - }, - { - "epoch": 45.43, - "learning_rate": 3.65334717176959e-06, - "loss": 3.5854, - "step": 175100 - }, - { - "epoch": 45.46, - "learning_rate": 3.632589517384536e-06, - "loss": 3.6205, - "step": 175200 - }, - { - "epoch": 45.49, - "learning_rate": 3.6118318629994816e-06, - "loss": 3.5974, - "step": 175300 - }, - { - "epoch": 45.51, - "learning_rate": 3.591074208614427e-06, - "loss": 3.5544, - "step": 175400 - }, - { - "epoch": 45.54, - "learning_rate": 3.5703165542293723e-06, - "loss": 3.6007, - "step": 175500 - }, - { - "epoch": 45.56, - "learning_rate": 3.549558899844318e-06, - "loss": 3.5852, - "step": 175600 - }, - { - "epoch": 45.59, - "learning_rate": 3.5288012454592636e-06, - "loss": 3.5849, - "step": 175700 - }, - { - "epoch": 45.61, - "learning_rate": 3.508043591074209e-06, - "loss": 3.6133, - "step": 175800 - }, - { - "epoch": 45.64, - "learning_rate": 3.4872859366891543e-06, - "loss": 3.6079, - "step": 175900 - }, - { - "epoch": 45.67, - "learning_rate": 3.4665282823040997e-06, - "loss": 3.6056, - "step": 176000 - }, - { - "epoch": 45.69, - "learning_rate": 3.4457706279190455e-06, - "loss": 3.6158, - "step": 176100 - }, - { - "epoch": 45.72, - "learning_rate": 3.425012973533991e-06, - "loss": 3.5834, - "step": 176200 - }, - { - "epoch": 45.74, - "learning_rate": 3.4042553191489363e-06, - "loss": 3.6131, - "step": 176300 - }, - { - "epoch": 45.77, - "learning_rate": 3.3834976647638817e-06, - "loss": 3.599, - "step": 176400 - }, - { - "epoch": 45.8, - "learning_rate": 3.3627400103788275e-06, - "loss": 3.5958, - "step": 176500 - }, - { - "epoch": 45.82, - "learning_rate": 3.341982355993773e-06, - "loss": 3.603, - "step": 176600 - }, - { - "epoch": 45.85, - "learning_rate": 3.3212247016087183e-06, - "loss": 3.6061, - "step": 176700 - }, - { - "epoch": 45.87, - "learning_rate": 3.3004670472236637e-06, - "loss": 3.6092, - "step": 176800 - }, - { - "epoch": 45.9, - "learning_rate": 3.279709392838609e-06, - "loss": 3.5822, - "step": 176900 - }, - { - "epoch": 45.93, - "learning_rate": 3.2589517384535553e-06, - "loss": 3.6168, - "step": 177000 - }, - { - "epoch": 45.95, - "learning_rate": 3.2381940840685007e-06, - "loss": 3.5881, - "step": 177100 - }, - { - "epoch": 45.98, - "learning_rate": 3.217436429683446e-06, - "loss": 3.5902, - "step": 177200 - }, - { - "epoch": 46.0, - "learning_rate": 3.1966787752983915e-06, - "loss": 3.6043, - "step": 177300 - }, - { - "epoch": 46.03, - "learning_rate": 3.1759211209133373e-06, - "loss": 3.622, - "step": 177400 - }, - { - "epoch": 46.06, - "learning_rate": 3.1551634665282827e-06, - "loss": 3.596, - "step": 177500 - }, - { - "epoch": 46.08, - "learning_rate": 3.134405812143228e-06, - "loss": 3.6066, - "step": 177600 - }, - { - "epoch": 46.11, - "learning_rate": 3.1136481577581735e-06, - "loss": 3.5779, - "step": 177700 - }, - { - "epoch": 46.13, - "learning_rate": 3.0928905033731193e-06, - "loss": 3.5979, - "step": 177800 - }, - { - "epoch": 46.16, - "learning_rate": 3.0721328489880647e-06, - "loss": 3.6115, - "step": 177900 - }, - { - "epoch": 46.19, - "learning_rate": 3.05137519460301e-06, - "loss": 3.5964, - "step": 178000 - }, - { - "epoch": 46.21, - "learning_rate": 3.0306175402179555e-06, - "loss": 3.601, - "step": 178100 - }, - { - "epoch": 46.24, - "learning_rate": 3.009859885832901e-06, - "loss": 3.605, - "step": 178200 - }, - { - "epoch": 46.26, - "learning_rate": 2.9891022314478467e-06, - "loss": 3.5961, - "step": 178300 - }, - { - "epoch": 46.29, - "learning_rate": 2.968344577062792e-06, - "loss": 3.5992, - "step": 178400 - }, - { - "epoch": 46.32, - "learning_rate": 2.9475869226777375e-06, - "loss": 3.5911, - "step": 178500 - }, - { - "epoch": 46.34, - "learning_rate": 2.926829268292683e-06, - "loss": 3.6145, - "step": 178600 - }, - { - "epoch": 46.37, - "learning_rate": 2.906071613907629e-06, - "loss": 3.6187, - "step": 178700 - }, - { - "epoch": 46.39, - "learning_rate": 2.8853139595225745e-06, - "loss": 3.5984, - "step": 178800 - }, - { - "epoch": 46.42, - "learning_rate": 2.86455630513752e-06, - "loss": 3.6209, - "step": 178900 - }, - { - "epoch": 46.45, - "learning_rate": 2.8437986507524653e-06, - "loss": 3.6118, - "step": 179000 - }, - { - "epoch": 46.47, - "learning_rate": 2.8230409963674107e-06, - "loss": 3.5919, - "step": 179100 - }, - { - "epoch": 46.5, - "learning_rate": 2.8022833419823565e-06, - "loss": 3.5834, - "step": 179200 - }, - { - "epoch": 46.52, - "learning_rate": 2.781525687597302e-06, - "loss": 3.6115, - "step": 179300 - }, - { - "epoch": 46.55, - "learning_rate": 2.7607680332122473e-06, - "loss": 3.5936, - "step": 179400 - }, - { - "epoch": 46.57, - "learning_rate": 2.7400103788271927e-06, - "loss": 3.6, - "step": 179500 - }, - { - "epoch": 46.6, - "learning_rate": 2.7192527244421385e-06, - "loss": 3.5933, - "step": 179600 - }, - { - "epoch": 46.63, - "learning_rate": 2.698495070057084e-06, - "loss": 3.6119, - "step": 179700 - }, - { - "epoch": 46.65, - "learning_rate": 2.6777374156720293e-06, - "loss": 3.5724, - "step": 179800 - }, - { - "epoch": 46.68, - "learning_rate": 2.6569797612869746e-06, - "loss": 3.6003, - "step": 179900 - }, - { - "epoch": 46.7, - "learning_rate": 2.63622210690192e-06, - "loss": 3.5871, - "step": 180000 - }, - { - "epoch": 46.7, - "eval_gen_len": 12.2763, - "eval_loss": 3.588209390640259, - "eval_rouge1": 42.036, - "eval_rouge2": 8.9032, - "eval_rougeL": 38.038, - "eval_rougeLsum": 38.0396, - "eval_runtime": 75.5618, - "eval_samples_per_second": 51.005, - "eval_steps_per_second": 6.379, - "step": 180000 - }, - { - "epoch": 46.73, - "learning_rate": 2.615464452516866e-06, - "loss": 3.6057, - "step": 180100 - }, - { - "epoch": 46.76, - "learning_rate": 2.5947067981318112e-06, - "loss": 3.6021, - "step": 180200 - }, - { - "epoch": 46.78, - "learning_rate": 2.5739491437467566e-06, - "loss": 3.6029, - "step": 180300 - }, - { - "epoch": 46.81, - "learning_rate": 2.553191489361702e-06, - "loss": 3.5808, - "step": 180400 - }, - { - "epoch": 46.83, - "learning_rate": 2.5324338349766483e-06, - "loss": 3.5926, - "step": 180500 - }, - { - "epoch": 46.86, - "learning_rate": 2.5116761805915937e-06, - "loss": 3.5993, - "step": 180600 - }, - { - "epoch": 46.89, - "learning_rate": 2.490918526206539e-06, - "loss": 3.5895, - "step": 180700 - }, - { - "epoch": 46.91, - "learning_rate": 2.4701608718214844e-06, - "loss": 3.5926, - "step": 180800 - }, - { - "epoch": 46.94, - "learning_rate": 2.44940321743643e-06, - "loss": 3.5833, - "step": 180900 - }, - { - "epoch": 46.96, - "learning_rate": 2.4286455630513752e-06, - "loss": 3.5937, - "step": 181000 - }, - { - "epoch": 46.99, - "learning_rate": 2.407887908666321e-06, - "loss": 3.5956, - "step": 181100 - }, - { - "epoch": 47.02, - "learning_rate": 2.3871302542812664e-06, - "loss": 3.5819, - "step": 181200 - }, - { - "epoch": 47.04, - "learning_rate": 2.366372599896212e-06, - "loss": 3.5829, - "step": 181300 - }, - { - "epoch": 47.07, - "learning_rate": 2.3456149455111572e-06, - "loss": 3.5862, - "step": 181400 - }, - { - "epoch": 47.09, - "learning_rate": 2.324857291126103e-06, - "loss": 3.596, - "step": 181500 - }, - { - "epoch": 47.12, - "learning_rate": 2.3040996367410484e-06, - "loss": 3.5806, - "step": 181600 - }, - { - "epoch": 47.15, - "learning_rate": 2.2833419823559942e-06, - "loss": 3.6108, - "step": 181700 - }, - { - "epoch": 47.17, - "learning_rate": 2.2625843279709396e-06, - "loss": 3.6012, - "step": 181800 - }, - { - "epoch": 47.2, - "learning_rate": 2.241826673585885e-06, - "loss": 3.6083, - "step": 181900 - }, - { - "epoch": 47.22, - "learning_rate": 2.2210690192008304e-06, - "loss": 3.6121, - "step": 182000 - }, - { - "epoch": 47.25, - "learning_rate": 2.200311364815776e-06, - "loss": 3.5922, - "step": 182100 - }, - { - "epoch": 47.28, - "learning_rate": 2.1795537104307216e-06, - "loss": 3.6015, - "step": 182200 - }, - { - "epoch": 47.3, - "learning_rate": 2.158796056045667e-06, - "loss": 3.5939, - "step": 182300 - }, - { - "epoch": 47.33, - "learning_rate": 2.1380384016606124e-06, - "loss": 3.5858, - "step": 182400 - }, - { - "epoch": 47.35, - "learning_rate": 2.117280747275558e-06, - "loss": 3.596, - "step": 182500 - }, - { - "epoch": 47.38, - "learning_rate": 2.0965230928905036e-06, - "loss": 3.5993, - "step": 182600 - }, - { - "epoch": 47.41, - "learning_rate": 2.075765438505449e-06, - "loss": 3.6232, - "step": 182700 - }, - { - "epoch": 47.43, - "learning_rate": 2.055007784120395e-06, - "loss": 3.6277, - "step": 182800 - }, - { - "epoch": 47.46, - "learning_rate": 2.03425012973534e-06, - "loss": 3.6111, - "step": 182900 - }, - { - "epoch": 47.48, - "learning_rate": 2.0134924753502856e-06, - "loss": 3.5782, - "step": 183000 - }, - { - "epoch": 47.51, - "learning_rate": 1.992734820965231e-06, - "loss": 3.6124, - "step": 183100 - }, - { - "epoch": 47.54, - "learning_rate": 1.9719771665801764e-06, - "loss": 3.5887, - "step": 183200 - }, - { - "epoch": 47.56, - "learning_rate": 1.951219512195122e-06, - "loss": 3.5961, - "step": 183300 - }, - { - "epoch": 47.59, - "learning_rate": 1.9304618578100676e-06, - "loss": 3.6037, - "step": 183400 - }, - { - "epoch": 47.61, - "learning_rate": 1.9097042034250134e-06, - "loss": 3.5906, - "step": 183500 - }, - { - "epoch": 47.64, - "learning_rate": 1.8889465490399586e-06, - "loss": 3.6008, - "step": 183600 - }, - { - "epoch": 47.66, - "learning_rate": 1.8681888946549042e-06, - "loss": 3.6003, - "step": 183700 - }, - { - "epoch": 47.69, - "learning_rate": 1.8474312402698496e-06, - "loss": 3.5995, - "step": 183800 - }, - { - "epoch": 47.72, - "learning_rate": 1.826673585884795e-06, - "loss": 3.6065, - "step": 183900 - }, - { - "epoch": 47.74, - "learning_rate": 1.8059159314997408e-06, - "loss": 3.603, - "step": 184000 - }, - { - "epoch": 47.77, - "learning_rate": 1.7851582771146862e-06, - "loss": 3.6115, - "step": 184100 - }, - { - "epoch": 47.79, - "learning_rate": 1.7644006227296318e-06, - "loss": 3.607, - "step": 184200 - }, - { - "epoch": 47.82, - "learning_rate": 1.7436429683445772e-06, - "loss": 3.5868, - "step": 184300 - }, - { - "epoch": 47.85, - "learning_rate": 1.7228853139595228e-06, - "loss": 3.5955, - "step": 184400 - }, - { - "epoch": 47.87, - "learning_rate": 1.7021276595744682e-06, - "loss": 3.5936, - "step": 184500 - }, - { - "epoch": 47.9, - "learning_rate": 1.6813700051894138e-06, - "loss": 3.5856, - "step": 184600 - }, - { - "epoch": 47.92, - "learning_rate": 1.6606123508043592e-06, - "loss": 3.5826, - "step": 184700 - }, - { - "epoch": 47.95, - "learning_rate": 1.6398546964193045e-06, - "loss": 3.596, - "step": 184800 - }, - { - "epoch": 47.98, - "learning_rate": 1.6190970420342504e-06, - "loss": 3.5876, - "step": 184900 - }, - { - "epoch": 48.0, - "learning_rate": 1.5983393876491958e-06, - "loss": 3.5989, - "step": 185000 - }, - { - "epoch": 48.0, - "eval_gen_len": 12.2766, - "eval_loss": 3.588301181793213, - "eval_rouge1": 42.0728, - "eval_rouge2": 8.8115, - "eval_rougeL": 38.0677, - "eval_rougeLsum": 38.0791, - "eval_runtime": 76.2595, - "eval_samples_per_second": 50.538, - "eval_steps_per_second": 6.321, - "step": 185000 - }, - { - "epoch": 48.03, - "learning_rate": 1.5775817332641414e-06, - "loss": 3.5737, - "step": 185100 - }, - { - "epoch": 48.05, - "learning_rate": 1.5568240788790867e-06, - "loss": 3.5853, - "step": 185200 - }, - { - "epoch": 48.08, - "learning_rate": 1.5360664244940324e-06, - "loss": 3.5972, - "step": 185300 - }, - { - "epoch": 48.11, - "learning_rate": 1.5153087701089777e-06, - "loss": 3.5913, - "step": 185400 - }, - { - "epoch": 48.13, - "learning_rate": 1.4945511157239233e-06, - "loss": 3.6013, - "step": 185500 - }, - { - "epoch": 48.16, - "learning_rate": 1.4737934613388687e-06, - "loss": 3.6086, - "step": 185600 - }, - { - "epoch": 48.18, - "learning_rate": 1.4530358069538146e-06, - "loss": 3.6082, - "step": 185700 - }, - { - "epoch": 48.21, - "learning_rate": 1.43227815256876e-06, - "loss": 3.6127, - "step": 185800 - }, - { - "epoch": 48.24, - "learning_rate": 1.4115204981837053e-06, - "loss": 3.5891, - "step": 185900 - }, - { - "epoch": 48.26, - "learning_rate": 1.390762843798651e-06, - "loss": 3.5713, - "step": 186000 - }, - { - "epoch": 48.29, - "learning_rate": 1.3700051894135963e-06, - "loss": 3.6016, - "step": 186100 - }, - { - "epoch": 48.31, - "learning_rate": 1.349247535028542e-06, - "loss": 3.6099, - "step": 186200 - }, - { - "epoch": 48.34, - "learning_rate": 1.3284898806434873e-06, - "loss": 3.6008, - "step": 186300 - }, - { - "epoch": 48.37, - "learning_rate": 1.307732226258433e-06, - "loss": 3.6014, - "step": 186400 - }, - { - "epoch": 48.39, - "learning_rate": 1.2869745718733783e-06, - "loss": 3.5783, - "step": 186500 - }, - { - "epoch": 48.42, - "learning_rate": 1.2662169174883241e-06, - "loss": 3.6134, - "step": 186600 - }, - { - "epoch": 48.44, - "learning_rate": 1.2454592631032695e-06, - "loss": 3.6004, - "step": 186700 - }, - { - "epoch": 48.47, - "learning_rate": 1.224701608718215e-06, - "loss": 3.5965, - "step": 186800 - }, - { - "epoch": 48.5, - "learning_rate": 1.2039439543331605e-06, - "loss": 3.6013, - "step": 186900 - }, - { - "epoch": 48.52, - "learning_rate": 1.183186299948106e-06, - "loss": 3.6047, - "step": 187000 - }, - { - "epoch": 48.55, - "learning_rate": 1.1624286455630515e-06, - "loss": 3.6083, - "step": 187100 - }, - { - "epoch": 48.57, - "learning_rate": 1.1416709911779971e-06, - "loss": 3.5758, - "step": 187200 - }, - { - "epoch": 48.6, - "learning_rate": 1.1209133367929425e-06, - "loss": 3.5949, - "step": 187300 - }, - { - "epoch": 48.62, - "learning_rate": 1.100155682407888e-06, - "loss": 3.6031, - "step": 187400 - }, - { - "epoch": 48.65, - "learning_rate": 1.0793980280228335e-06, - "loss": 3.6055, - "step": 187500 - }, - { - "epoch": 48.68, - "learning_rate": 1.058640373637779e-06, - "loss": 3.5967, - "step": 187600 - }, - { - "epoch": 48.7, - "learning_rate": 1.0378827192527245e-06, - "loss": 3.6363, - "step": 187700 - }, - { - "epoch": 48.73, - "learning_rate": 1.01712506486767e-06, - "loss": 3.5949, - "step": 187800 - }, - { - "epoch": 48.75, - "learning_rate": 9.963674104826155e-07, - "loss": 3.6098, - "step": 187900 - }, - { - "epoch": 48.78, - "learning_rate": 9.75609756097561e-07, - "loss": 3.5849, - "step": 188000 - }, - { - "epoch": 48.81, - "learning_rate": 9.548521017125067e-07, - "loss": 3.5787, - "step": 188100 - }, - { - "epoch": 48.83, - "learning_rate": 9.340944473274521e-07, - "loss": 3.6023, - "step": 188200 - }, - { - "epoch": 48.86, - "learning_rate": 9.133367929423975e-07, - "loss": 3.596, - "step": 188300 - }, - { - "epoch": 48.88, - "learning_rate": 8.925791385573431e-07, - "loss": 3.5904, - "step": 188400 - }, - { - "epoch": 48.91, - "learning_rate": 8.718214841722886e-07, - "loss": 3.5774, - "step": 188500 - }, - { - "epoch": 48.94, - "learning_rate": 8.510638297872341e-07, - "loss": 3.5695, - "step": 188600 - }, - { - "epoch": 48.96, - "learning_rate": 8.303061754021796e-07, - "loss": 3.61, - "step": 188700 - }, - { - "epoch": 48.99, - "learning_rate": 8.095485210171252e-07, - "loss": 3.5872, - "step": 188800 - }, - { - "epoch": 49.01, - "learning_rate": 7.887908666320707e-07, - "loss": 3.6107, - "step": 188900 - }, - { - "epoch": 49.04, - "learning_rate": 7.680332122470162e-07, - "loss": 3.5763, - "step": 189000 - }, - { - "epoch": 49.07, - "learning_rate": 7.472755578619617e-07, - "loss": 3.5971, - "step": 189100 - }, - { - "epoch": 49.09, - "learning_rate": 7.265179034769073e-07, - "loss": 3.6019, - "step": 189200 - }, - { - "epoch": 49.12, - "learning_rate": 7.057602490918527e-07, - "loss": 3.6155, - "step": 189300 - }, - { - "epoch": 49.14, - "learning_rate": 6.850025947067982e-07, - "loss": 3.5935, - "step": 189400 - }, - { - "epoch": 49.17, - "learning_rate": 6.642449403217437e-07, - "loss": 3.6014, - "step": 189500 - }, - { - "epoch": 49.2, - "learning_rate": 6.434872859366892e-07, - "loss": 3.6099, - "step": 189600 - }, - { - "epoch": 49.22, - "learning_rate": 6.227296315516348e-07, - "loss": 3.5569, - "step": 189700 - }, - { - "epoch": 49.25, - "learning_rate": 6.019719771665803e-07, - "loss": 3.6024, - "step": 189800 - }, - { - "epoch": 49.27, - "learning_rate": 5.812143227815258e-07, - "loss": 3.5954, - "step": 189900 - }, - { - "epoch": 49.3, - "learning_rate": 5.604566683964713e-07, - "loss": 3.5932, - "step": 190000 - }, - { - "epoch": 49.3, - "eval_gen_len": 12.2761, - "eval_loss": 3.588038682937622, - "eval_rouge1": 41.988, - "eval_rouge2": 8.8748, - "eval_rougeL": 37.9796, - "eval_rougeLsum": 37.984, - "eval_runtime": 75.4468, - "eval_samples_per_second": 51.082, - "eval_steps_per_second": 6.389, - "step": 190000 - } - ], - "logging_steps": 100, - "max_steps": 192700, - "num_train_epochs": 50, - "save_steps": 5000, - "total_flos": 9.256502644614758e+16, - "trial_name": null, - "trial_params": null -}