| { | |
| "best_metric": 0.249, | |
| "best_model_checkpoint": "logs/google-t5/t5-small/checkpoint-28155", | |
| "epoch": 18.0, | |
| "eval_steps": 500, | |
| "global_step": 33786, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 1.3638477325439453, | |
| "learning_rate": 1.997336174746937e-05, | |
| "loss": 2.8409, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 1.420181155204773, | |
| "learning_rate": 1.9946723494938732e-05, | |
| "loss": 2.538, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 1.119789719581604, | |
| "learning_rate": 1.9920085242408102e-05, | |
| "loss": 2.4663, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 1.1085336208343506, | |
| "learning_rate": 1.9893446989877465e-05, | |
| "loss": 2.4737, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.8778786659240723, | |
| "learning_rate": 1.9866808737346832e-05, | |
| "loss": 2.4436, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 1.1101582050323486, | |
| "learning_rate": 1.9840170484816195e-05, | |
| "loss": 2.4545, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 1.1823639869689941, | |
| "learning_rate": 1.9813532232285565e-05, | |
| "loss": 2.3472, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 1.0613980293273926, | |
| "learning_rate": 1.978689397975493e-05, | |
| "loss": 2.3945, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 1.1558998823165894, | |
| "learning_rate": 1.9760255727224296e-05, | |
| "loss": 2.4172, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 1.098973274230957, | |
| "learning_rate": 1.9733617474693662e-05, | |
| "loss": 2.3629, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 1.3629950284957886, | |
| "learning_rate": 1.970697922216303e-05, | |
| "loss": 2.3975, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 1.1299411058425903, | |
| "learning_rate": 1.9680340969632396e-05, | |
| "loss": 2.3322, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 0.9830572605133057, | |
| "learning_rate": 1.965370271710176e-05, | |
| "loss": 2.379, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 1.0205196142196655, | |
| "learning_rate": 1.9627064464571126e-05, | |
| "loss": 2.3454, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 1.230190396308899, | |
| "learning_rate": 1.960042621204049e-05, | |
| "loss": 2.3045, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 1.115490436553955, | |
| "learning_rate": 1.957378795950986e-05, | |
| "loss": 2.3676, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 0.9047977924346924, | |
| "learning_rate": 1.9547149706979226e-05, | |
| "loss": 2.3871, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 1.0352205038070679, | |
| "learning_rate": 1.952051145444859e-05, | |
| "loss": 2.3481, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 0.864746630191803, | |
| "learning_rate": 1.9493873201917956e-05, | |
| "loss": 2.3384, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 1.3858081102371216, | |
| "learning_rate": 1.9467234949387323e-05, | |
| "loss": 2.3271, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 0.9969013333320618, | |
| "learning_rate": 1.944059669685669e-05, | |
| "loss": 2.3267, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 1.035396933555603, | |
| "learning_rate": 1.9413958444326053e-05, | |
| "loss": 2.349, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 1.235714316368103, | |
| "learning_rate": 1.938732019179542e-05, | |
| "loss": 2.3442, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 0.9119770526885986, | |
| "learning_rate": 1.9360681939264787e-05, | |
| "loss": 2.3246, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 0.8794578909873962, | |
| "learning_rate": 1.9334043686734153e-05, | |
| "loss": 2.3633, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 0.9355188012123108, | |
| "learning_rate": 1.930740543420352e-05, | |
| "loss": 2.339, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.9303766489028931, | |
| "learning_rate": 1.9280767181672883e-05, | |
| "loss": 2.3174, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 1.3199224472045898, | |
| "learning_rate": 1.925412892914225e-05, | |
| "loss": 2.3321, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 1.0782549381256104, | |
| "learning_rate": 1.9227490676611617e-05, | |
| "loss": 2.3348, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.9403214454650879, | |
| "learning_rate": 1.9200852424080984e-05, | |
| "loss": 2.3133, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 0.9809348583221436, | |
| "learning_rate": 1.9174214171550347e-05, | |
| "loss": 2.3177, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.9974561333656311, | |
| "learning_rate": 1.9147575919019714e-05, | |
| "loss": 2.311, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 0.9475740194320679, | |
| "learning_rate": 1.912093766648908e-05, | |
| "loss": 2.3208, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 1.0113029479980469, | |
| "learning_rate": 1.9094299413958447e-05, | |
| "loss": 2.3031, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 0.920647144317627, | |
| "learning_rate": 1.9067661161427814e-05, | |
| "loss": 2.3268, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.9876402020454407, | |
| "learning_rate": 1.9041022908897177e-05, | |
| "loss": 2.3, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 1.0489838123321533, | |
| "learning_rate": 1.9014384656366544e-05, | |
| "loss": 2.305, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_bert-score-f1": 0.8753073360155793, | |
| "eval_bert-score-precision": 0.8779077029804112, | |
| "eval_bert-score-recall": 0.873077545509742, | |
| "eval_gen_len": 49.607327717446466, | |
| "eval_loss": 2.088743209838867, | |
| "eval_rouge1": 0.3105, | |
| "eval_rouge2": 0.1136, | |
| "eval_rougeL": 0.2275, | |
| "eval_rougeLsum": 0.2276, | |
| "eval_runtime": 1364.7929, | |
| "eval_samples_per_second": 14.439, | |
| "eval_steps_per_second": 0.344, | |
| "step": 1877 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "grad_norm": 0.8712087869644165, | |
| "learning_rate": 1.898774640383591e-05, | |
| "loss": 2.2941, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "grad_norm": 0.8705251812934875, | |
| "learning_rate": 1.8961108151305278e-05, | |
| "loss": 2.2953, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "grad_norm": 1.098132610321045, | |
| "learning_rate": 1.893446989877464e-05, | |
| "loss": 2.2976, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "grad_norm": 1.1778416633605957, | |
| "learning_rate": 1.8907831646244008e-05, | |
| "loss": 2.2866, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "grad_norm": 0.7667921185493469, | |
| "learning_rate": 1.8881193393713374e-05, | |
| "loss": 2.2641, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "grad_norm": 0.9039889574050903, | |
| "learning_rate": 1.885455514118274e-05, | |
| "loss": 2.2788, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "grad_norm": 0.8681074976921082, | |
| "learning_rate": 1.8827916888652108e-05, | |
| "loss": 2.3297, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 1.0154231786727905, | |
| "learning_rate": 1.880127863612147e-05, | |
| "loss": 2.3267, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "grad_norm": 1.5686161518096924, | |
| "learning_rate": 1.8774640383590838e-05, | |
| "loss": 2.2867, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 1.0235270261764526, | |
| "learning_rate": 1.8748002131060205e-05, | |
| "loss": 2.3132, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "grad_norm": 1.1120573282241821, | |
| "learning_rate": 1.872136387852957e-05, | |
| "loss": 2.2774, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "grad_norm": 0.9750345349311829, | |
| "learning_rate": 1.8694725625998935e-05, | |
| "loss": 2.2903, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "grad_norm": 0.8807668089866638, | |
| "learning_rate": 1.86680873734683e-05, | |
| "loss": 2.2923, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "grad_norm": 0.9335638284683228, | |
| "learning_rate": 1.8641449120937668e-05, | |
| "loss": 2.3168, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "grad_norm": 0.8730989694595337, | |
| "learning_rate": 1.8614810868407035e-05, | |
| "loss": 2.3007, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "grad_norm": 1.0019006729125977, | |
| "learning_rate": 1.8588172615876402e-05, | |
| "loss": 2.3087, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "grad_norm": 0.9429858922958374, | |
| "learning_rate": 1.8561534363345765e-05, | |
| "loss": 2.2802, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "grad_norm": 1.0355122089385986, | |
| "learning_rate": 1.8534896110815132e-05, | |
| "loss": 2.2801, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "grad_norm": 1.3223506212234497, | |
| "learning_rate": 1.85082578582845e-05, | |
| "loss": 2.3245, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "grad_norm": 0.9225859642028809, | |
| "learning_rate": 1.8481619605753865e-05, | |
| "loss": 2.2838, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "grad_norm": 0.9716720581054688, | |
| "learning_rate": 1.845498135322323e-05, | |
| "loss": 2.296, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "grad_norm": 0.947675883769989, | |
| "learning_rate": 1.8428343100692595e-05, | |
| "loss": 2.2768, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "grad_norm": 0.9065707921981812, | |
| "learning_rate": 1.8401704848161962e-05, | |
| "loss": 2.2767, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "grad_norm": 1.0325031280517578, | |
| "learning_rate": 1.837506659563133e-05, | |
| "loss": 2.262, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "grad_norm": 0.9252289533615112, | |
| "learning_rate": 1.8348428343100696e-05, | |
| "loss": 2.2824, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "grad_norm": 0.8026869297027588, | |
| "learning_rate": 1.832179009057006e-05, | |
| "loss": 2.2762, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "grad_norm": 1.017001748085022, | |
| "learning_rate": 1.8295151838039426e-05, | |
| "loss": 2.2742, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "grad_norm": 0.8680307269096375, | |
| "learning_rate": 1.8268513585508792e-05, | |
| "loss": 2.252, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "grad_norm": 0.9704865217208862, | |
| "learning_rate": 1.824187533297816e-05, | |
| "loss": 2.2627, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "grad_norm": 1.1407126188278198, | |
| "learning_rate": 1.8215237080447523e-05, | |
| "loss": 2.305, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "grad_norm": 1.0476961135864258, | |
| "learning_rate": 1.818859882791689e-05, | |
| "loss": 2.2659, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "grad_norm": 0.9194826483726501, | |
| "learning_rate": 1.8161960575386256e-05, | |
| "loss": 2.2728, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "grad_norm": 1.061948299407959, | |
| "learning_rate": 1.8135322322855623e-05, | |
| "loss": 2.262, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "grad_norm": 0.8690770864486694, | |
| "learning_rate": 1.810868407032499e-05, | |
| "loss": 2.2611, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "grad_norm": 1.000588297843933, | |
| "learning_rate": 1.8082045817794353e-05, | |
| "loss": 2.2707, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "grad_norm": 1.2689441442489624, | |
| "learning_rate": 1.805540756526372e-05, | |
| "loss": 2.2688, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "grad_norm": 0.9255685210227966, | |
| "learning_rate": 1.8028769312733086e-05, | |
| "loss": 2.2559, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 1.0140782594680786, | |
| "learning_rate": 1.8002131060202453e-05, | |
| "loss": 2.2838, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_bert-score-f1": 0.8773485723212565, | |
| "eval_bert-score-precision": 0.8805842198545674, | |
| "eval_bert-score-recall": 0.8744701069949739, | |
| "eval_gen_len": 49.16461991271694, | |
| "eval_loss": 2.062197685241699, | |
| "eval_rouge1": 0.3196, | |
| "eval_rouge2": 0.1183, | |
| "eval_rougeL": 0.2349, | |
| "eval_rougeLsum": 0.235, | |
| "eval_runtime": 1347.8022, | |
| "eval_samples_per_second": 14.621, | |
| "eval_steps_per_second": 0.349, | |
| "step": 3754 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "grad_norm": 0.8860589861869812, | |
| "learning_rate": 1.7975492807671817e-05, | |
| "loss": 2.2635, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "grad_norm": 0.9218833446502686, | |
| "learning_rate": 1.7948854555141183e-05, | |
| "loss": 2.2621, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "grad_norm": 0.9549726247787476, | |
| "learning_rate": 1.792221630261055e-05, | |
| "loss": 2.2513, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "grad_norm": 1.135712742805481, | |
| "learning_rate": 1.7895578050079917e-05, | |
| "loss": 2.2721, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "grad_norm": 1.056344985961914, | |
| "learning_rate": 1.7868939797549283e-05, | |
| "loss": 2.2601, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "grad_norm": 0.8976427316665649, | |
| "learning_rate": 1.7842301545018647e-05, | |
| "loss": 2.2441, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "grad_norm": 1.0189875364303589, | |
| "learning_rate": 1.7815663292488014e-05, | |
| "loss": 2.2441, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "grad_norm": 1.1941207647323608, | |
| "learning_rate": 1.778902503995738e-05, | |
| "loss": 2.2518, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "grad_norm": 0.9858642816543579, | |
| "learning_rate": 1.7762386787426747e-05, | |
| "loss": 2.2417, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "grad_norm": 0.8909502625465393, | |
| "learning_rate": 1.7735748534896114e-05, | |
| "loss": 2.2741, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "grad_norm": 1.1010240316390991, | |
| "learning_rate": 1.7709110282365477e-05, | |
| "loss": 2.2538, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "grad_norm": 1.1926771402359009, | |
| "learning_rate": 1.7682472029834844e-05, | |
| "loss": 2.2844, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "grad_norm": 1.048973798751831, | |
| "learning_rate": 1.765583377730421e-05, | |
| "loss": 2.2677, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "grad_norm": 1.0072720050811768, | |
| "learning_rate": 1.7629195524773577e-05, | |
| "loss": 2.2616, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "grad_norm": 0.854369044303894, | |
| "learning_rate": 1.760255727224294e-05, | |
| "loss": 2.2498, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "grad_norm": 0.9605410099029541, | |
| "learning_rate": 1.7575919019712307e-05, | |
| "loss": 2.2541, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "grad_norm": 0.9667823314666748, | |
| "learning_rate": 1.7549280767181674e-05, | |
| "loss": 2.26, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "grad_norm": 2.42110013961792, | |
| "learning_rate": 1.752264251465104e-05, | |
| "loss": 2.2585, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 1.0035040378570557, | |
| "learning_rate": 1.7496004262120408e-05, | |
| "loss": 2.2538, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "grad_norm": 1.1149507761001587, | |
| "learning_rate": 1.746936600958977e-05, | |
| "loss": 2.2643, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 1.4942309856414795, | |
| "learning_rate": 1.7442727757059138e-05, | |
| "loss": 2.3053, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "grad_norm": 1.016640305519104, | |
| "learning_rate": 1.7416089504528505e-05, | |
| "loss": 2.2325, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "grad_norm": 0.8914662003517151, | |
| "learning_rate": 1.738945125199787e-05, | |
| "loss": 2.26, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "grad_norm": 1.0502891540527344, | |
| "learning_rate": 1.7362812999467235e-05, | |
| "loss": 2.2357, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "grad_norm": 0.9784315824508667, | |
| "learning_rate": 1.73361747469366e-05, | |
| "loss": 2.2371, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "grad_norm": 0.9308114647865295, | |
| "learning_rate": 1.7309536494405968e-05, | |
| "loss": 2.2452, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "grad_norm": 0.9364565014839172, | |
| "learning_rate": 1.7282898241875335e-05, | |
| "loss": 2.235, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "grad_norm": 0.8989120721817017, | |
| "learning_rate": 1.72562599893447e-05, | |
| "loss": 2.272, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "grad_norm": 1.2291622161865234, | |
| "learning_rate": 1.7229621736814065e-05, | |
| "loss": 2.278, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "grad_norm": 0.9619302749633789, | |
| "learning_rate": 1.7202983484283432e-05, | |
| "loss": 2.2554, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "grad_norm": 1.0430196523666382, | |
| "learning_rate": 1.71763452317528e-05, | |
| "loss": 2.2546, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "grad_norm": 1.0834633111953735, | |
| "learning_rate": 1.7149706979222165e-05, | |
| "loss": 2.2379, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "grad_norm": 1.0730029344558716, | |
| "learning_rate": 1.712306872669153e-05, | |
| "loss": 2.2789, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "grad_norm": 1.124229907989502, | |
| "learning_rate": 1.7096430474160895e-05, | |
| "loss": 2.2293, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "grad_norm": 0.9467495679855347, | |
| "learning_rate": 1.7069792221630262e-05, | |
| "loss": 2.2606, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "grad_norm": 0.8856106400489807, | |
| "learning_rate": 1.704315396909963e-05, | |
| "loss": 2.2713, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "grad_norm": 0.7882632613182068, | |
| "learning_rate": 1.7016515716568996e-05, | |
| "loss": 2.2508, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_bert-score-f1": 0.8785184490539139, | |
| "eval_bert-score-precision": 0.8821266702464495, | |
| "eval_bert-score-recall": 0.8752563403945616, | |
| "eval_gen_len": 48.600781487871714, | |
| "eval_loss": 2.047697067260742, | |
| "eval_rouge1": 0.324, | |
| "eval_rouge2": 0.1204, | |
| "eval_rougeL": 0.2387, | |
| "eval_rougeLsum": 0.2389, | |
| "eval_runtime": 1313.571, | |
| "eval_samples_per_second": 15.002, | |
| "eval_steps_per_second": 0.358, | |
| "step": 5631 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "grad_norm": 0.9630438685417175, | |
| "learning_rate": 1.698987746403836e-05, | |
| "loss": 2.246, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "grad_norm": 0.8204315900802612, | |
| "learning_rate": 1.6963239211507726e-05, | |
| "loss": 2.225, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "grad_norm": 1.1320478916168213, | |
| "learning_rate": 1.6936600958977092e-05, | |
| "loss": 2.2206, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "grad_norm": 0.9559237360954285, | |
| "learning_rate": 1.690996270644646e-05, | |
| "loss": 2.2459, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "grad_norm": 0.9689226150512695, | |
| "learning_rate": 1.6883324453915822e-05, | |
| "loss": 2.2517, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "grad_norm": 0.9438573718070984, | |
| "learning_rate": 1.685668620138519e-05, | |
| "loss": 2.2579, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "grad_norm": 1.1130074262619019, | |
| "learning_rate": 1.6830047948854556e-05, | |
| "loss": 2.2387, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "grad_norm": 1.1083691120147705, | |
| "learning_rate": 1.6803409696323923e-05, | |
| "loss": 2.2558, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "grad_norm": 0.96318119764328, | |
| "learning_rate": 1.677677144379329e-05, | |
| "loss": 2.257, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "grad_norm": 0.8898953795433044, | |
| "learning_rate": 1.6750133191262653e-05, | |
| "loss": 2.2524, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "grad_norm": 1.2933117151260376, | |
| "learning_rate": 1.672349493873202e-05, | |
| "loss": 2.2551, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "grad_norm": 1.1499016284942627, | |
| "learning_rate": 1.6696856686201386e-05, | |
| "loss": 2.244, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "grad_norm": 0.9967105388641357, | |
| "learning_rate": 1.6670218433670753e-05, | |
| "loss": 2.2514, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "grad_norm": 1.0175275802612305, | |
| "learning_rate": 1.6643580181140116e-05, | |
| "loss": 2.2298, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "grad_norm": 0.8878999352455139, | |
| "learning_rate": 1.6616941928609483e-05, | |
| "loss": 2.2394, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "grad_norm": 1.1751534938812256, | |
| "learning_rate": 1.6590303676078853e-05, | |
| "loss": 2.2321, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "grad_norm": 1.0338590145111084, | |
| "learning_rate": 1.6563665423548217e-05, | |
| "loss": 2.2562, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "grad_norm": 0.8953673243522644, | |
| "learning_rate": 1.6537027171017583e-05, | |
| "loss": 2.2199, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "grad_norm": 1.24599027633667, | |
| "learning_rate": 1.6510388918486947e-05, | |
| "loss": 2.2346, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "grad_norm": 0.953091561794281, | |
| "learning_rate": 1.6483750665956313e-05, | |
| "loss": 2.246, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "grad_norm": 0.919922947883606, | |
| "learning_rate": 1.645711241342568e-05, | |
| "loss": 2.2288, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "grad_norm": 1.1812031269073486, | |
| "learning_rate": 1.6430474160895047e-05, | |
| "loss": 2.2419, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "grad_norm": 0.9377938508987427, | |
| "learning_rate": 1.640383590836441e-05, | |
| "loss": 2.2478, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "grad_norm": 1.247550368309021, | |
| "learning_rate": 1.6377197655833777e-05, | |
| "loss": 2.2495, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "grad_norm": 1.1216537952423096, | |
| "learning_rate": 1.6350559403303147e-05, | |
| "loss": 2.2241, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "grad_norm": 1.038252353668213, | |
| "learning_rate": 1.632392115077251e-05, | |
| "loss": 2.2346, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "grad_norm": 0.8921188116073608, | |
| "learning_rate": 1.6297282898241877e-05, | |
| "loss": 2.2488, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "grad_norm": 1.0422166585922241, | |
| "learning_rate": 1.627064464571124e-05, | |
| "loss": 2.2336, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "grad_norm": 1.0294251441955566, | |
| "learning_rate": 1.624400639318061e-05, | |
| "loss": 2.234, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "grad_norm": 0.8972188830375671, | |
| "learning_rate": 1.6217368140649974e-05, | |
| "loss": 2.2147, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "grad_norm": 1.1645957231521606, | |
| "learning_rate": 1.619072988811934e-05, | |
| "loss": 2.2336, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "grad_norm": 1.2658772468566895, | |
| "learning_rate": 1.6164091635588704e-05, | |
| "loss": 2.2322, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "grad_norm": 0.923834502696991, | |
| "learning_rate": 1.613745338305807e-05, | |
| "loss": 2.2348, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "grad_norm": 0.8885565996170044, | |
| "learning_rate": 1.611081513052744e-05, | |
| "loss": 2.2251, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "grad_norm": 1.1370388269424438, | |
| "learning_rate": 1.6084176877996804e-05, | |
| "loss": 2.2266, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "grad_norm": 0.9725862145423889, | |
| "learning_rate": 1.605753862546617e-05, | |
| "loss": 2.2243, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "grad_norm": 0.8951359987258911, | |
| "learning_rate": 1.6030900372935535e-05, | |
| "loss": 2.2278, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 0.9756078720092773, | |
| "learning_rate": 1.6004262120404905e-05, | |
| "loss": 2.2591, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_bert-score-f1": 0.8792554777484974, | |
| "eval_bert-score-precision": 0.8833729287797178, | |
| "eval_bert-score-recall": 0.8754834527081552, | |
| "eval_gen_len": 47.54663554247438, | |
| "eval_loss": 2.039776086807251, | |
| "eval_rouge1": 0.3271, | |
| "eval_rouge2": 0.1225, | |
| "eval_rougeL": 0.2416, | |
| "eval_rougeLsum": 0.2417, | |
| "eval_runtime": 1382.3306, | |
| "eval_samples_per_second": 14.256, | |
| "eval_steps_per_second": 0.34, | |
| "step": 7508 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "grad_norm": 0.9563359618186951, | |
| "learning_rate": 1.5977623867874268e-05, | |
| "loss": 2.2286, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "grad_norm": 0.9959126114845276, | |
| "learning_rate": 1.5950985615343635e-05, | |
| "loss": 2.2287, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "grad_norm": 0.9405544996261597, | |
| "learning_rate": 1.5924347362813e-05, | |
| "loss": 2.244, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "grad_norm": 1.0261281728744507, | |
| "learning_rate": 1.5897709110282368e-05, | |
| "loss": 2.2141, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "grad_norm": 0.8446668386459351, | |
| "learning_rate": 1.5871070857751735e-05, | |
| "loss": 2.2217, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "grad_norm": 0.981716513633728, | |
| "learning_rate": 1.58444326052211e-05, | |
| "loss": 2.2535, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "grad_norm": 0.9390552043914795, | |
| "learning_rate": 1.5817794352690465e-05, | |
| "loss": 2.1902, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "grad_norm": 0.9392485618591309, | |
| "learning_rate": 1.5791156100159832e-05, | |
| "loss": 2.2325, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "grad_norm": 0.8909654021263123, | |
| "learning_rate": 1.57645178476292e-05, | |
| "loss": 2.2106, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "grad_norm": 0.9464941620826721, | |
| "learning_rate": 1.5737879595098562e-05, | |
| "loss": 2.1985, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "grad_norm": 0.9561291337013245, | |
| "learning_rate": 1.571124134256793e-05, | |
| "loss": 2.2595, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "grad_norm": 1.0075396299362183, | |
| "learning_rate": 1.5684603090037295e-05, | |
| "loss": 2.2328, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "grad_norm": 1.0439025163650513, | |
| "learning_rate": 1.5657964837506662e-05, | |
| "loss": 2.2036, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "grad_norm": 0.9002524614334106, | |
| "learning_rate": 1.563132658497603e-05, | |
| "loss": 2.2157, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "grad_norm": 0.9860824346542358, | |
| "learning_rate": 1.5604688332445392e-05, | |
| "loss": 2.2526, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "grad_norm": 0.8125404715538025, | |
| "learning_rate": 1.557805007991476e-05, | |
| "loss": 2.2458, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "grad_norm": 0.9793021082878113, | |
| "learning_rate": 1.5551411827384126e-05, | |
| "loss": 2.2111, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "grad_norm": 0.9543974995613098, | |
| "learning_rate": 1.5524773574853492e-05, | |
| "loss": 2.2372, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "grad_norm": 1.0786656141281128, | |
| "learning_rate": 1.5498135322322856e-05, | |
| "loss": 2.2256, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "grad_norm": 1.1105990409851074, | |
| "learning_rate": 1.5471497069792223e-05, | |
| "loss": 2.2282, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "grad_norm": 1.4228782653808594, | |
| "learning_rate": 1.544485881726159e-05, | |
| "loss": 2.2482, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "grad_norm": 1.0002621412277222, | |
| "learning_rate": 1.5418220564730956e-05, | |
| "loss": 2.2104, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "grad_norm": 0.8620786666870117, | |
| "learning_rate": 1.5391582312200323e-05, | |
| "loss": 2.2201, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "grad_norm": 0.9315156936645508, | |
| "learning_rate": 1.5364944059669686e-05, | |
| "loss": 2.2519, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "grad_norm": 0.9382575750350952, | |
| "learning_rate": 1.5338305807139053e-05, | |
| "loss": 2.2439, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "grad_norm": 0.9228121042251587, | |
| "learning_rate": 1.531166755460842e-05, | |
| "loss": 2.2105, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "grad_norm": 0.9347973465919495, | |
| "learning_rate": 1.5285029302077786e-05, | |
| "loss": 2.2201, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "grad_norm": 1.0652745962142944, | |
| "learning_rate": 1.525839104954715e-05, | |
| "loss": 2.2117, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "grad_norm": 0.9884467720985413, | |
| "learning_rate": 1.5231752797016516e-05, | |
| "loss": 2.2058, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "grad_norm": 1.2926949262619019, | |
| "learning_rate": 1.5205114544485883e-05, | |
| "loss": 2.2062, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "grad_norm": 0.9000586867332458, | |
| "learning_rate": 1.5178476291955248e-05, | |
| "loss": 2.2578, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "grad_norm": 1.1215986013412476, | |
| "learning_rate": 1.5151838039424615e-05, | |
| "loss": 2.2035, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "grad_norm": 0.9411687254905701, | |
| "learning_rate": 1.512519978689398e-05, | |
| "loss": 2.2392, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "grad_norm": 0.9772723913192749, | |
| "learning_rate": 1.5098561534363347e-05, | |
| "loss": 2.2341, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "grad_norm": 0.9062642455101013, | |
| "learning_rate": 1.5071923281832712e-05, | |
| "loss": 2.2132, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "grad_norm": 0.9838491082191467, | |
| "learning_rate": 1.5045285029302079e-05, | |
| "loss": 2.2316, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "grad_norm": 0.7986493110656738, | |
| "learning_rate": 1.5018646776771445e-05, | |
| "loss": 2.2406, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_bert-score-f1": 0.8796141604796214, | |
| "eval_bert-score-precision": 0.883697147159108, | |
| "eval_bert-score-recall": 0.8758704174409406, | |
| "eval_gen_len": 47.69806150411042, | |
| "eval_loss": 2.032108783721924, | |
| "eval_rouge1": 0.3291, | |
| "eval_rouge2": 0.1235, | |
| "eval_rougeL": 0.2433, | |
| "eval_rougeLsum": 0.2434, | |
| "eval_runtime": 1320.3538, | |
| "eval_samples_per_second": 14.925, | |
| "eval_steps_per_second": 0.356, | |
| "step": 9385 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "grad_norm": 1.181012511253357, | |
| "learning_rate": 1.499200852424081e-05, | |
| "loss": 2.205, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "grad_norm": 0.8982387185096741, | |
| "learning_rate": 1.4965370271710177e-05, | |
| "loss": 2.2044, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "grad_norm": 0.9843763709068298, | |
| "learning_rate": 1.4938732019179542e-05, | |
| "loss": 2.2087, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "grad_norm": 0.9226497411727905, | |
| "learning_rate": 1.4912093766648909e-05, | |
| "loss": 2.2019, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "grad_norm": 1.0900003910064697, | |
| "learning_rate": 1.4885455514118274e-05, | |
| "loss": 2.2219, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "grad_norm": 1.0898627042770386, | |
| "learning_rate": 1.485881726158764e-05, | |
| "loss": 2.2281, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "grad_norm": 1.1963268518447876, | |
| "learning_rate": 1.4832179009057006e-05, | |
| "loss": 2.2289, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "grad_norm": 0.9371781349182129, | |
| "learning_rate": 1.4805540756526373e-05, | |
| "loss": 2.2287, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "grad_norm": 0.9393157958984375, | |
| "learning_rate": 1.4778902503995741e-05, | |
| "loss": 2.1454, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 5.25, | |
| "grad_norm": 1.1463308334350586, | |
| "learning_rate": 1.4752264251465104e-05, | |
| "loss": 2.2116, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "grad_norm": 1.232642412185669, | |
| "learning_rate": 1.4725625998934471e-05, | |
| "loss": 2.2056, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "grad_norm": 0.8566424250602722, | |
| "learning_rate": 1.4698987746403836e-05, | |
| "loss": 2.2077, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "grad_norm": 1.073114275932312, | |
| "learning_rate": 1.4672349493873203e-05, | |
| "loss": 2.2043, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "grad_norm": 1.100190281867981, | |
| "learning_rate": 1.4645711241342568e-05, | |
| "loss": 2.2105, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 5.38, | |
| "grad_norm": 0.8915311694145203, | |
| "learning_rate": 1.4619072988811935e-05, | |
| "loss": 2.2263, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "grad_norm": 1.0838483572006226, | |
| "learning_rate": 1.45924347362813e-05, | |
| "loss": 2.2386, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "grad_norm": 1.0265840291976929, | |
| "learning_rate": 1.4565796483750666e-05, | |
| "loss": 2.2046, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "grad_norm": 1.012404441833496, | |
| "learning_rate": 1.4539158231220035e-05, | |
| "loss": 2.1914, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "grad_norm": 0.8332634568214417, | |
| "learning_rate": 1.4512519978689398e-05, | |
| "loss": 2.2424, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "grad_norm": 0.9419781565666199, | |
| "learning_rate": 1.4485881726158767e-05, | |
| "loss": 2.2081, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 5.54, | |
| "grad_norm": 0.9537687301635742, | |
| "learning_rate": 1.445924347362813e-05, | |
| "loss": 2.2382, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 5.57, | |
| "grad_norm": 0.8523421883583069, | |
| "learning_rate": 1.4432605221097498e-05, | |
| "loss": 2.2479, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "grad_norm": 0.9717277884483337, | |
| "learning_rate": 1.4405966968566862e-05, | |
| "loss": 2.2335, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "grad_norm": 1.2556852102279663, | |
| "learning_rate": 1.4379328716036229e-05, | |
| "loss": 2.207, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 5.65, | |
| "grad_norm": 1.0949528217315674, | |
| "learning_rate": 1.4352690463505594e-05, | |
| "loss": 2.2454, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "grad_norm": 0.8859919309616089, | |
| "learning_rate": 1.432605221097496e-05, | |
| "loss": 2.1991, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "grad_norm": 1.0127480030059814, | |
| "learning_rate": 1.4299413958444329e-05, | |
| "loss": 2.2031, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 5.73, | |
| "grad_norm": 1.0773361921310425, | |
| "learning_rate": 1.4272775705913692e-05, | |
| "loss": 2.2183, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "grad_norm": 1.2354990243911743, | |
| "learning_rate": 1.424613745338306e-05, | |
| "loss": 2.1787, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "grad_norm": 0.8633403182029724, | |
| "learning_rate": 1.4219499200852424e-05, | |
| "loss": 2.2058, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "grad_norm": 0.8349035382270813, | |
| "learning_rate": 1.4192860948321792e-05, | |
| "loss": 2.2102, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "grad_norm": 0.954465389251709, | |
| "learning_rate": 1.4166222695791156e-05, | |
| "loss": 2.2074, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "grad_norm": 1.0947058200836182, | |
| "learning_rate": 1.4139584443260524e-05, | |
| "loss": 2.2063, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "grad_norm": 1.0066620111465454, | |
| "learning_rate": 1.4112946190729891e-05, | |
| "loss": 2.2391, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "grad_norm": 1.3262407779693604, | |
| "learning_rate": 1.4086307938199256e-05, | |
| "loss": 2.1883, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "grad_norm": 0.898461639881134, | |
| "learning_rate": 1.4059669685668623e-05, | |
| "loss": 2.2254, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "grad_norm": 1.2728203535079956, | |
| "learning_rate": 1.4033031433137986e-05, | |
| "loss": 2.2002, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "grad_norm": 0.8816812038421631, | |
| "learning_rate": 1.4006393180607354e-05, | |
| "loss": 2.222, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_bert-score-f1": 0.8800827935451446, | |
| "eval_bert-score-precision": 0.884203289352576, | |
| "eval_bert-score-recall": 0.8763012464314104, | |
| "eval_gen_len": 47.73409113975439, | |
| "eval_loss": 2.0270581245422363, | |
| "eval_rouge1": 0.3308, | |
| "eval_rouge2": 0.125, | |
| "eval_rougeL": 0.2449, | |
| "eval_rougeLsum": 0.245, | |
| "eval_runtime": 1367.5469, | |
| "eval_samples_per_second": 14.41, | |
| "eval_steps_per_second": 0.344, | |
| "step": 11262 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "grad_norm": 0.9405691623687744, | |
| "learning_rate": 1.3979754928076718e-05, | |
| "loss": 2.1931, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "grad_norm": 0.8556106686592102, | |
| "learning_rate": 1.3953116675546086e-05, | |
| "loss": 2.2086, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "grad_norm": 1.0254322290420532, | |
| "learning_rate": 1.392647842301545e-05, | |
| "loss": 2.2235, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "grad_norm": 0.9665789604187012, | |
| "learning_rate": 1.3899840170484818e-05, | |
| "loss": 2.1956, | |
| "step": 11450 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "grad_norm": 1.1138479709625244, | |
| "learning_rate": 1.3873201917954185e-05, | |
| "loss": 2.2233, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "grad_norm": 0.9537137150764465, | |
| "learning_rate": 1.384656366542355e-05, | |
| "loss": 2.2047, | |
| "step": 11550 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "grad_norm": 0.9105952978134155, | |
| "learning_rate": 1.3819925412892917e-05, | |
| "loss": 2.1739, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 6.21, | |
| "grad_norm": 0.9230592250823975, | |
| "learning_rate": 1.3793287160362282e-05, | |
| "loss": 2.1911, | |
| "step": 11650 | |
| }, | |
| { | |
| "epoch": 6.23, | |
| "grad_norm": 0.944442868232727, | |
| "learning_rate": 1.3766648907831648e-05, | |
| "loss": 2.2074, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 6.26, | |
| "grad_norm": 0.9925301671028137, | |
| "learning_rate": 1.3740010655301013e-05, | |
| "loss": 2.1973, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "grad_norm": 1.0720447301864624, | |
| "learning_rate": 1.371337240277038e-05, | |
| "loss": 2.181, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 6.31, | |
| "grad_norm": 0.9020980000495911, | |
| "learning_rate": 1.3686734150239745e-05, | |
| "loss": 2.193, | |
| "step": 11850 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "grad_norm": 0.9980204701423645, | |
| "learning_rate": 1.3660095897709112e-05, | |
| "loss": 2.2342, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 6.37, | |
| "grad_norm": 0.9240878820419312, | |
| "learning_rate": 1.3633457645178479e-05, | |
| "loss": 2.1855, | |
| "step": 11950 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "grad_norm": 1.1149895191192627, | |
| "learning_rate": 1.3606819392647844e-05, | |
| "loss": 2.1918, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "grad_norm": 0.8434773683547974, | |
| "learning_rate": 1.358018114011721e-05, | |
| "loss": 2.2347, | |
| "step": 12050 | |
| }, | |
| { | |
| "epoch": 6.45, | |
| "grad_norm": 0.8160800337791443, | |
| "learning_rate": 1.3553542887586576e-05, | |
| "loss": 2.2275, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "grad_norm": 1.0393248796463013, | |
| "learning_rate": 1.3526904635055942e-05, | |
| "loss": 2.2449, | |
| "step": 12150 | |
| }, | |
| { | |
| "epoch": 6.5, | |
| "grad_norm": 0.9805082082748413, | |
| "learning_rate": 1.3500266382525307e-05, | |
| "loss": 2.2118, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "grad_norm": 0.8884342908859253, | |
| "learning_rate": 1.3473628129994674e-05, | |
| "loss": 2.2184, | |
| "step": 12250 | |
| }, | |
| { | |
| "epoch": 6.55, | |
| "grad_norm": 1.1096868515014648, | |
| "learning_rate": 1.3446989877464039e-05, | |
| "loss": 2.231, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 6.58, | |
| "grad_norm": 0.937127947807312, | |
| "learning_rate": 1.3420351624933406e-05, | |
| "loss": 2.2032, | |
| "step": 12350 | |
| }, | |
| { | |
| "epoch": 6.61, | |
| "grad_norm": 1.0019285678863525, | |
| "learning_rate": 1.3393713372402773e-05, | |
| "loss": 2.2028, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 6.63, | |
| "grad_norm": 1.070123553276062, | |
| "learning_rate": 1.3367075119872138e-05, | |
| "loss": 2.2673, | |
| "step": 12450 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "grad_norm": 0.8974484801292419, | |
| "learning_rate": 1.3340436867341504e-05, | |
| "loss": 2.1909, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 6.69, | |
| "grad_norm": 0.9534024596214294, | |
| "learning_rate": 1.331379861481087e-05, | |
| "loss": 2.2061, | |
| "step": 12550 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "grad_norm": 0.8722013831138611, | |
| "learning_rate": 1.3287160362280236e-05, | |
| "loss": 2.2054, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 6.74, | |
| "grad_norm": 1.222458004951477, | |
| "learning_rate": 1.3260522109749601e-05, | |
| "loss": 2.167, | |
| "step": 12650 | |
| }, | |
| { | |
| "epoch": 6.77, | |
| "grad_norm": 0.8800060153007507, | |
| "learning_rate": 1.3233883857218968e-05, | |
| "loss": 2.1955, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 6.79, | |
| "grad_norm": 1.005924105644226, | |
| "learning_rate": 1.3207245604688335e-05, | |
| "loss": 2.1644, | |
| "step": 12750 | |
| }, | |
| { | |
| "epoch": 6.82, | |
| "grad_norm": 1.0622010231018066, | |
| "learning_rate": 1.31806073521577e-05, | |
| "loss": 2.226, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 6.85, | |
| "grad_norm": 0.9073338508605957, | |
| "learning_rate": 1.3153969099627067e-05, | |
| "loss": 2.1781, | |
| "step": 12850 | |
| }, | |
| { | |
| "epoch": 6.87, | |
| "grad_norm": 1.0695091485977173, | |
| "learning_rate": 1.3127330847096432e-05, | |
| "loss": 2.1965, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 6.9, | |
| "grad_norm": 1.0539031028747559, | |
| "learning_rate": 1.3100692594565798e-05, | |
| "loss": 2.1864, | |
| "step": 12950 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "grad_norm": 1.0355671644210815, | |
| "learning_rate": 1.3074054342035163e-05, | |
| "loss": 2.21, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 6.95, | |
| "grad_norm": 0.9524690508842468, | |
| "learning_rate": 1.304741608950453e-05, | |
| "loss": 2.2146, | |
| "step": 13050 | |
| }, | |
| { | |
| "epoch": 6.98, | |
| "grad_norm": 0.8750210404396057, | |
| "learning_rate": 1.3020777836973895e-05, | |
| "loss": 2.2185, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_bert-score-f1": 0.8803590643584976, | |
| "eval_bert-score-precision": 0.8845637915861083, | |
| "eval_bert-score-recall": 0.8764868623034598, | |
| "eval_gen_len": 47.655891606617274, | |
| "eval_loss": 2.022202253341675, | |
| "eval_rouge1": 0.3314, | |
| "eval_rouge2": 0.1248, | |
| "eval_rougeL": 0.2451, | |
| "eval_rougeLsum": 0.2452, | |
| "eval_runtime": 1324.8965, | |
| "eval_samples_per_second": 14.874, | |
| "eval_steps_per_second": 0.355, | |
| "step": 13139 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "grad_norm": 1.8067052364349365, | |
| "learning_rate": 1.2994139584443262e-05, | |
| "loss": 2.2044, | |
| "step": 13150 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "grad_norm": 1.0369492769241333, | |
| "learning_rate": 1.2967501331912629e-05, | |
| "loss": 2.2093, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "grad_norm": 0.7998984456062317, | |
| "learning_rate": 1.2940863079381994e-05, | |
| "loss": 2.1803, | |
| "step": 13250 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "grad_norm": 0.8763892650604248, | |
| "learning_rate": 1.291422482685136e-05, | |
| "loss": 2.2028, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "grad_norm": 1.031553864479065, | |
| "learning_rate": 1.2887586574320726e-05, | |
| "loss": 2.1598, | |
| "step": 13350 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "grad_norm": 1.2158304452896118, | |
| "learning_rate": 1.2860948321790092e-05, | |
| "loss": 2.2008, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "grad_norm": 1.0015206336975098, | |
| "learning_rate": 1.2834310069259457e-05, | |
| "loss": 2.2118, | |
| "step": 13450 | |
| }, | |
| { | |
| "epoch": 7.19, | |
| "grad_norm": 1.3029738664627075, | |
| "learning_rate": 1.2807671816728824e-05, | |
| "loss": 2.163, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 7.22, | |
| "grad_norm": 1.1144462823867798, | |
| "learning_rate": 1.2781033564198189e-05, | |
| "loss": 2.1821, | |
| "step": 13550 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "grad_norm": 0.9808390736579895, | |
| "learning_rate": 1.2754395311667556e-05, | |
| "loss": 2.1844, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 7.27, | |
| "grad_norm": 0.8838719129562378, | |
| "learning_rate": 1.2727757059136923e-05, | |
| "loss": 2.1936, | |
| "step": 13650 | |
| }, | |
| { | |
| "epoch": 7.3, | |
| "grad_norm": 1.0219964981079102, | |
| "learning_rate": 1.2701118806606288e-05, | |
| "loss": 2.1864, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 7.33, | |
| "grad_norm": 0.9708404541015625, | |
| "learning_rate": 1.2674480554075654e-05, | |
| "loss": 2.1734, | |
| "step": 13750 | |
| }, | |
| { | |
| "epoch": 7.35, | |
| "grad_norm": 0.9595683217048645, | |
| "learning_rate": 1.264784230154502e-05, | |
| "loss": 2.2069, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 7.38, | |
| "grad_norm": 1.2408117055892944, | |
| "learning_rate": 1.2621204049014386e-05, | |
| "loss": 2.1918, | |
| "step": 13850 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "grad_norm": 0.9526282548904419, | |
| "learning_rate": 1.2594565796483751e-05, | |
| "loss": 2.2069, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 7.43, | |
| "grad_norm": 0.9085534811019897, | |
| "learning_rate": 1.2567927543953118e-05, | |
| "loss": 2.1929, | |
| "step": 13950 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "grad_norm": 0.9988005757331848, | |
| "learning_rate": 1.2541289291422483e-05, | |
| "loss": 2.212, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 7.49, | |
| "grad_norm": 0.9731308221817017, | |
| "learning_rate": 1.251465103889185e-05, | |
| "loss": 2.2168, | |
| "step": 14050 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "grad_norm": 1.2218581438064575, | |
| "learning_rate": 1.2488012786361216e-05, | |
| "loss": 2.2136, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 7.54, | |
| "grad_norm": 1.3860177993774414, | |
| "learning_rate": 1.2461374533830582e-05, | |
| "loss": 2.1844, | |
| "step": 14150 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "grad_norm": 1.3227049112319946, | |
| "learning_rate": 1.2434736281299948e-05, | |
| "loss": 2.2017, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 7.59, | |
| "grad_norm": 0.9687981009483337, | |
| "learning_rate": 1.2408098028769313e-05, | |
| "loss": 2.2242, | |
| "step": 14250 | |
| }, | |
| { | |
| "epoch": 7.62, | |
| "grad_norm": 1.0260825157165527, | |
| "learning_rate": 1.238145977623868e-05, | |
| "loss": 2.2096, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 7.65, | |
| "grad_norm": 0.9353439211845398, | |
| "learning_rate": 1.2354821523708045e-05, | |
| "loss": 2.2122, | |
| "step": 14350 | |
| }, | |
| { | |
| "epoch": 7.67, | |
| "grad_norm": 1.0099036693572998, | |
| "learning_rate": 1.2328183271177412e-05, | |
| "loss": 2.2187, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 7.7, | |
| "grad_norm": 0.8156920075416565, | |
| "learning_rate": 1.2301545018646779e-05, | |
| "loss": 2.1853, | |
| "step": 14450 | |
| }, | |
| { | |
| "epoch": 7.73, | |
| "grad_norm": 14.783089637756348, | |
| "learning_rate": 1.2274906766116144e-05, | |
| "loss": 2.1917, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 7.75, | |
| "grad_norm": 0.9252942204475403, | |
| "learning_rate": 1.224826851358551e-05, | |
| "loss": 2.2074, | |
| "step": 14550 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "grad_norm": 0.9822306632995605, | |
| "learning_rate": 1.2221630261054875e-05, | |
| "loss": 2.193, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "grad_norm": 1.1552455425262451, | |
| "learning_rate": 1.2194992008524242e-05, | |
| "loss": 2.1851, | |
| "step": 14650 | |
| }, | |
| { | |
| "epoch": 7.83, | |
| "grad_norm": 1.0293680429458618, | |
| "learning_rate": 1.2168353755993607e-05, | |
| "loss": 2.2109, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "grad_norm": 1.0043119192123413, | |
| "learning_rate": 1.2141715503462974e-05, | |
| "loss": 2.2112, | |
| "step": 14750 | |
| }, | |
| { | |
| "epoch": 7.88, | |
| "grad_norm": 0.9331013560295105, | |
| "learning_rate": 1.2115077250932339e-05, | |
| "loss": 2.1837, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 7.91, | |
| "grad_norm": 0.9587385058403015, | |
| "learning_rate": 1.2088438998401706e-05, | |
| "loss": 2.1773, | |
| "step": 14850 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "grad_norm": 1.1071295738220215, | |
| "learning_rate": 1.2061800745871073e-05, | |
| "loss": 2.2065, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 7.96, | |
| "grad_norm": 1.34778892993927, | |
| "learning_rate": 1.2035162493340438e-05, | |
| "loss": 2.1854, | |
| "step": 14950 | |
| }, | |
| { | |
| "epoch": 7.99, | |
| "grad_norm": 3.2136006355285645, | |
| "learning_rate": 1.2008524240809804e-05, | |
| "loss": 2.1886, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_bert-score-f1": 0.8805449074880007, | |
| "eval_bert-score-precision": 0.8848301609899627, | |
| "eval_bert-score-recall": 0.8765954361686147, | |
| "eval_gen_len": 47.339947224195676, | |
| "eval_loss": 2.0204520225524902, | |
| "eval_rouge1": 0.3323, | |
| "eval_rouge2": 0.1257, | |
| "eval_rougeL": 0.2461, | |
| "eval_rougeLsum": 0.2462, | |
| "eval_runtime": 1323.5674, | |
| "eval_samples_per_second": 14.889, | |
| "eval_steps_per_second": 0.355, | |
| "step": 15016 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "grad_norm": 0.8995711803436279, | |
| "learning_rate": 1.198188598827917e-05, | |
| "loss": 2.1601, | |
| "step": 15050 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "grad_norm": 1.0043758153915405, | |
| "learning_rate": 1.1955247735748536e-05, | |
| "loss": 2.2012, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 8.07, | |
| "grad_norm": 0.9039593935012817, | |
| "learning_rate": 1.1928609483217901e-05, | |
| "loss": 2.2207, | |
| "step": 15150 | |
| }, | |
| { | |
| "epoch": 8.1, | |
| "grad_norm": 0.9619396328926086, | |
| "learning_rate": 1.1901971230687268e-05, | |
| "loss": 2.1609, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 8.12, | |
| "grad_norm": 1.0527337789535522, | |
| "learning_rate": 1.1875332978156633e-05, | |
| "loss": 2.221, | |
| "step": 15250 | |
| }, | |
| { | |
| "epoch": 8.15, | |
| "grad_norm": 0.9329215288162231, | |
| "learning_rate": 1.1848694725626e-05, | |
| "loss": 2.1669, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "grad_norm": 1.0186532735824585, | |
| "learning_rate": 1.1822056473095366e-05, | |
| "loss": 2.1857, | |
| "step": 15350 | |
| }, | |
| { | |
| "epoch": 8.2, | |
| "grad_norm": 1.1034983396530151, | |
| "learning_rate": 1.1795418220564731e-05, | |
| "loss": 2.1956, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 8.23, | |
| "grad_norm": 1.0719212293624878, | |
| "learning_rate": 1.1768779968034098e-05, | |
| "loss": 2.2007, | |
| "step": 15450 | |
| }, | |
| { | |
| "epoch": 8.26, | |
| "grad_norm": 1.5037603378295898, | |
| "learning_rate": 1.1742141715503463e-05, | |
| "loss": 2.1967, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 8.28, | |
| "grad_norm": 0.9770453572273254, | |
| "learning_rate": 1.171550346297283e-05, | |
| "loss": 2.218, | |
| "step": 15550 | |
| }, | |
| { | |
| "epoch": 8.31, | |
| "grad_norm": 1.0334933996200562, | |
| "learning_rate": 1.1688865210442195e-05, | |
| "loss": 2.1892, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 8.34, | |
| "grad_norm": 0.9753350615501404, | |
| "learning_rate": 1.1662226957911562e-05, | |
| "loss": 2.2074, | |
| "step": 15650 | |
| }, | |
| { | |
| "epoch": 8.36, | |
| "grad_norm": 0.9147941470146179, | |
| "learning_rate": 1.1635588705380927e-05, | |
| "loss": 2.2111, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 8.39, | |
| "grad_norm": 1.0154210329055786, | |
| "learning_rate": 1.1608950452850294e-05, | |
| "loss": 2.1971, | |
| "step": 15750 | |
| }, | |
| { | |
| "epoch": 8.42, | |
| "grad_norm": 1.0365736484527588, | |
| "learning_rate": 1.158231220031966e-05, | |
| "loss": 2.1874, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 8.44, | |
| "grad_norm": 0.894719660282135, | |
| "learning_rate": 1.1555673947789025e-05, | |
| "loss": 2.2059, | |
| "step": 15850 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "grad_norm": 0.9420655369758606, | |
| "learning_rate": 1.1529035695258392e-05, | |
| "loss": 2.1473, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 8.5, | |
| "grad_norm": 1.0870007276535034, | |
| "learning_rate": 1.1502397442727757e-05, | |
| "loss": 2.1944, | |
| "step": 15950 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "grad_norm": 1.072520136833191, | |
| "learning_rate": 1.1475759190197124e-05, | |
| "loss": 2.1914, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "grad_norm": 0.9754800200462341, | |
| "learning_rate": 1.1449120937666489e-05, | |
| "loss": 2.1726, | |
| "step": 16050 | |
| }, | |
| { | |
| "epoch": 8.58, | |
| "grad_norm": 0.983051598072052, | |
| "learning_rate": 1.1422482685135856e-05, | |
| "loss": 2.1591, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 8.6, | |
| "grad_norm": 0.9521353244781494, | |
| "learning_rate": 1.1395844432605222e-05, | |
| "loss": 2.1675, | |
| "step": 16150 | |
| }, | |
| { | |
| "epoch": 8.63, | |
| "grad_norm": 1.016432762145996, | |
| "learning_rate": 1.1369206180074587e-05, | |
| "loss": 2.1821, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 8.66, | |
| "grad_norm": 0.9886535406112671, | |
| "learning_rate": 1.1342567927543954e-05, | |
| "loss": 2.1711, | |
| "step": 16250 | |
| }, | |
| { | |
| "epoch": 8.68, | |
| "grad_norm": 0.9348542094230652, | |
| "learning_rate": 1.131592967501332e-05, | |
| "loss": 2.2016, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "grad_norm": 0.8952618837356567, | |
| "learning_rate": 1.1289291422482686e-05, | |
| "loss": 2.1632, | |
| "step": 16350 | |
| }, | |
| { | |
| "epoch": 8.74, | |
| "grad_norm": 0.9894302487373352, | |
| "learning_rate": 1.1262653169952051e-05, | |
| "loss": 2.1792, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 8.76, | |
| "grad_norm": 0.899584949016571, | |
| "learning_rate": 1.1236014917421418e-05, | |
| "loss": 2.1698, | |
| "step": 16450 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "grad_norm": 0.9734613299369812, | |
| "learning_rate": 1.1209376664890783e-05, | |
| "loss": 2.2253, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 8.82, | |
| "grad_norm": 1.0697177648544312, | |
| "learning_rate": 1.118273841236015e-05, | |
| "loss": 2.2307, | |
| "step": 16550 | |
| }, | |
| { | |
| "epoch": 8.84, | |
| "grad_norm": 0.827741801738739, | |
| "learning_rate": 1.1156100159829516e-05, | |
| "loss": 2.179, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 8.87, | |
| "grad_norm": 0.8641231060028076, | |
| "learning_rate": 1.1129461907298881e-05, | |
| "loss": 2.1768, | |
| "step": 16650 | |
| }, | |
| { | |
| "epoch": 8.9, | |
| "grad_norm": 0.9500383734703064, | |
| "learning_rate": 1.1102823654768248e-05, | |
| "loss": 2.185, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 8.92, | |
| "grad_norm": 0.8771828413009644, | |
| "learning_rate": 1.1076185402237613e-05, | |
| "loss": 2.2018, | |
| "step": 16750 | |
| }, | |
| { | |
| "epoch": 8.95, | |
| "grad_norm": 0.7786308526992798, | |
| "learning_rate": 1.104954714970698e-05, | |
| "loss": 2.1856, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 8.98, | |
| "grad_norm": 0.9104407429695129, | |
| "learning_rate": 1.1022908897176345e-05, | |
| "loss": 2.2215, | |
| "step": 16850 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_bert-score-f1": 0.8808141563956924, | |
| "eval_bert-score-precision": 0.8852914021691005, | |
| "eval_bert-score-recall": 0.8766740019322894, | |
| "eval_gen_len": 47.14802598193444, | |
| "eval_loss": 2.0156819820404053, | |
| "eval_rouge1": 0.333, | |
| "eval_rouge2": 0.1262, | |
| "eval_rougeL": 0.2467, | |
| "eval_rougeLsum": 0.2467, | |
| "eval_runtime": 1329.5202, | |
| "eval_samples_per_second": 14.822, | |
| "eval_steps_per_second": 0.354, | |
| "step": 16893 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "grad_norm": 0.9584008455276489, | |
| "learning_rate": 1.0996270644645712e-05, | |
| "loss": 2.1627, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "grad_norm": 1.0817357301712036, | |
| "learning_rate": 1.0969632392115077e-05, | |
| "loss": 2.174, | |
| "step": 16950 | |
| }, | |
| { | |
| "epoch": 9.06, | |
| "grad_norm": 1.1630058288574219, | |
| "learning_rate": 1.0942994139584444e-05, | |
| "loss": 2.1988, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 9.08, | |
| "grad_norm": 1.065058946609497, | |
| "learning_rate": 1.091635588705381e-05, | |
| "loss": 2.1628, | |
| "step": 17050 | |
| }, | |
| { | |
| "epoch": 9.11, | |
| "grad_norm": 1.3880870342254639, | |
| "learning_rate": 1.0889717634523175e-05, | |
| "loss": 2.177, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 9.14, | |
| "grad_norm": 1.0271745920181274, | |
| "learning_rate": 1.0863079381992542e-05, | |
| "loss": 2.212, | |
| "step": 17150 | |
| }, | |
| { | |
| "epoch": 9.16, | |
| "grad_norm": 0.9514613747596741, | |
| "learning_rate": 1.0836441129461907e-05, | |
| "loss": 2.1729, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 9.19, | |
| "grad_norm": 0.9049180746078491, | |
| "learning_rate": 1.0809802876931274e-05, | |
| "loss": 2.1928, | |
| "step": 17250 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "grad_norm": 1.0372991561889648, | |
| "learning_rate": 1.0783164624400639e-05, | |
| "loss": 2.207, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 9.24, | |
| "grad_norm": 0.9088106751441956, | |
| "learning_rate": 1.0756526371870006e-05, | |
| "loss": 2.1818, | |
| "step": 17350 | |
| }, | |
| { | |
| "epoch": 9.27, | |
| "grad_norm": 1.0153067111968994, | |
| "learning_rate": 1.072988811933937e-05, | |
| "loss": 2.1996, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 9.3, | |
| "grad_norm": 1.0672743320465088, | |
| "learning_rate": 1.0703249866808737e-05, | |
| "loss": 2.2183, | |
| "step": 17450 | |
| }, | |
| { | |
| "epoch": 9.32, | |
| "grad_norm": 1.0173332691192627, | |
| "learning_rate": 1.0676611614278106e-05, | |
| "loss": 2.2153, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 9.35, | |
| "grad_norm": 1.1739368438720703, | |
| "learning_rate": 1.064997336174747e-05, | |
| "loss": 2.1694, | |
| "step": 17550 | |
| }, | |
| { | |
| "epoch": 9.38, | |
| "grad_norm": 0.8909546136856079, | |
| "learning_rate": 1.0623335109216838e-05, | |
| "loss": 2.1908, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "grad_norm": 0.9530623555183411, | |
| "learning_rate": 1.0596696856686201e-05, | |
| "loss": 2.2027, | |
| "step": 17650 | |
| }, | |
| { | |
| "epoch": 9.43, | |
| "grad_norm": 0.8588944673538208, | |
| "learning_rate": 1.057005860415557e-05, | |
| "loss": 2.143, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 9.46, | |
| "grad_norm": 1.096243977546692, | |
| "learning_rate": 1.0543420351624933e-05, | |
| "loss": 2.1315, | |
| "step": 17750 | |
| }, | |
| { | |
| "epoch": 9.48, | |
| "grad_norm": 0.888977587223053, | |
| "learning_rate": 1.05167820990943e-05, | |
| "loss": 2.1478, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 9.51, | |
| "grad_norm": 1.374234676361084, | |
| "learning_rate": 1.0490143846563668e-05, | |
| "loss": 2.1564, | |
| "step": 17850 | |
| }, | |
| { | |
| "epoch": 9.54, | |
| "grad_norm": 1.0646979808807373, | |
| "learning_rate": 1.0463505594033031e-05, | |
| "loss": 2.1963, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "grad_norm": 0.8910280466079712, | |
| "learning_rate": 1.04368673415024e-05, | |
| "loss": 2.1892, | |
| "step": 17950 | |
| }, | |
| { | |
| "epoch": 9.59, | |
| "grad_norm": 1.0399140119552612, | |
| "learning_rate": 1.0410229088971763e-05, | |
| "loss": 2.2132, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 9.62, | |
| "grad_norm": 1.1613365411758423, | |
| "learning_rate": 1.0383590836441132e-05, | |
| "loss": 2.1946, | |
| "step": 18050 | |
| }, | |
| { | |
| "epoch": 9.64, | |
| "grad_norm": 0.9431652426719666, | |
| "learning_rate": 1.0356952583910495e-05, | |
| "loss": 2.1825, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 9.67, | |
| "grad_norm": 1.1334350109100342, | |
| "learning_rate": 1.0330314331379863e-05, | |
| "loss": 2.1937, | |
| "step": 18150 | |
| }, | |
| { | |
| "epoch": 9.7, | |
| "grad_norm": 1.1028201580047607, | |
| "learning_rate": 1.0303676078849227e-05, | |
| "loss": 2.2181, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 9.72, | |
| "grad_norm": 0.9916718602180481, | |
| "learning_rate": 1.0277037826318595e-05, | |
| "loss": 2.174, | |
| "step": 18250 | |
| }, | |
| { | |
| "epoch": 9.75, | |
| "grad_norm": 1.0355281829833984, | |
| "learning_rate": 1.0250399573787962e-05, | |
| "loss": 2.1565, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 9.78, | |
| "grad_norm": 0.901858925819397, | |
| "learning_rate": 1.0223761321257327e-05, | |
| "loss": 2.1732, | |
| "step": 18350 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "grad_norm": 1.1691358089447021, | |
| "learning_rate": 1.0197123068726694e-05, | |
| "loss": 2.1854, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 9.83, | |
| "grad_norm": 1.1045140027999878, | |
| "learning_rate": 1.0170484816196057e-05, | |
| "loss": 2.1906, | |
| "step": 18450 | |
| }, | |
| { | |
| "epoch": 9.86, | |
| "grad_norm": 1.4803402423858643, | |
| "learning_rate": 1.0143846563665425e-05, | |
| "loss": 2.1744, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 9.88, | |
| "grad_norm": 1.0411149263381958, | |
| "learning_rate": 1.0117208311134789e-05, | |
| "loss": 2.2064, | |
| "step": 18550 | |
| }, | |
| { | |
| "epoch": 9.91, | |
| "grad_norm": 0.9394697546958923, | |
| "learning_rate": 1.0090570058604157e-05, | |
| "loss": 2.1646, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 9.94, | |
| "grad_norm": 1.0079654455184937, | |
| "learning_rate": 1.006393180607352e-05, | |
| "loss": 2.1638, | |
| "step": 18650 | |
| }, | |
| { | |
| "epoch": 9.96, | |
| "grad_norm": 0.9374598264694214, | |
| "learning_rate": 1.0037293553542889e-05, | |
| "loss": 2.2037, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 9.99, | |
| "grad_norm": 0.9795972108840942, | |
| "learning_rate": 1.0010655301012256e-05, | |
| "loss": 2.2119, | |
| "step": 18750 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_bert-score-f1": 0.8809602989904121, | |
| "eval_bert-score-precision": 0.8852206096635378, | |
| "eval_bert-score-recall": 0.8770327499302876, | |
| "eval_gen_len": 47.48157921445245, | |
| "eval_loss": 2.0146677494049072, | |
| "eval_rouge1": 0.3344, | |
| "eval_rouge2": 0.1266, | |
| "eval_rougeL": 0.2474, | |
| "eval_rougeLsum": 0.2475, | |
| "eval_runtime": 1332.7051, | |
| "eval_samples_per_second": 14.786, | |
| "eval_steps_per_second": 0.353, | |
| "step": 18770 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "grad_norm": 1.0391963720321655, | |
| "learning_rate": 9.984017048481621e-06, | |
| "loss": 2.1618, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "grad_norm": 0.9638227820396423, | |
| "learning_rate": 9.957378795950986e-06, | |
| "loss": 2.1847, | |
| "step": 18850 | |
| }, | |
| { | |
| "epoch": 10.07, | |
| "grad_norm": 0.9130365252494812, | |
| "learning_rate": 9.930740543420353e-06, | |
| "loss": 2.1409, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 10.1, | |
| "grad_norm": 0.97170490026474, | |
| "learning_rate": 9.90410229088972e-06, | |
| "loss": 2.2007, | |
| "step": 18950 | |
| }, | |
| { | |
| "epoch": 10.12, | |
| "grad_norm": 0.9903939962387085, | |
| "learning_rate": 9.877464038359084e-06, | |
| "loss": 2.1873, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 10.15, | |
| "grad_norm": 1.0155619382858276, | |
| "learning_rate": 9.850825785828451e-06, | |
| "loss": 2.2185, | |
| "step": 19050 | |
| }, | |
| { | |
| "epoch": 10.18, | |
| "grad_norm": 1.0440953969955444, | |
| "learning_rate": 9.824187533297816e-06, | |
| "loss": 2.183, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 10.2, | |
| "grad_norm": 1.002216100692749, | |
| "learning_rate": 9.797549280767183e-06, | |
| "loss": 2.1827, | |
| "step": 19150 | |
| }, | |
| { | |
| "epoch": 10.23, | |
| "grad_norm": 0.9036744236946106, | |
| "learning_rate": 9.770911028236548e-06, | |
| "loss": 2.157, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 10.26, | |
| "grad_norm": 1.0186132192611694, | |
| "learning_rate": 9.744272775705915e-06, | |
| "loss": 2.1458, | |
| "step": 19250 | |
| }, | |
| { | |
| "epoch": 10.28, | |
| "grad_norm": 1.157223105430603, | |
| "learning_rate": 9.71763452317528e-06, | |
| "loss": 2.1578, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 10.31, | |
| "grad_norm": 1.0317802429199219, | |
| "learning_rate": 9.690996270644647e-06, | |
| "loss": 2.1852, | |
| "step": 19350 | |
| }, | |
| { | |
| "epoch": 10.34, | |
| "grad_norm": 0.9137316942214966, | |
| "learning_rate": 9.664358018114013e-06, | |
| "loss": 2.228, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 10.36, | |
| "grad_norm": 0.8266160488128662, | |
| "learning_rate": 9.637719765583378e-06, | |
| "loss": 2.2115, | |
| "step": 19450 | |
| }, | |
| { | |
| "epoch": 10.39, | |
| "grad_norm": 0.9575254917144775, | |
| "learning_rate": 9.611081513052745e-06, | |
| "loss": 2.1671, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 10.42, | |
| "grad_norm": 1.1885944604873657, | |
| "learning_rate": 9.58444326052211e-06, | |
| "loss": 2.1953, | |
| "step": 19550 | |
| }, | |
| { | |
| "epoch": 10.44, | |
| "grad_norm": 0.8843189477920532, | |
| "learning_rate": 9.557805007991477e-06, | |
| "loss": 2.1859, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 10.47, | |
| "grad_norm": 1.5487014055252075, | |
| "learning_rate": 9.531166755460842e-06, | |
| "loss": 2.1678, | |
| "step": 19650 | |
| }, | |
| { | |
| "epoch": 10.5, | |
| "grad_norm": 1.1339489221572876, | |
| "learning_rate": 9.504528502930209e-06, | |
| "loss": 2.1917, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 10.52, | |
| "grad_norm": 0.973892331123352, | |
| "learning_rate": 9.477890250399575e-06, | |
| "loss": 2.1741, | |
| "step": 19750 | |
| }, | |
| { | |
| "epoch": 10.55, | |
| "grad_norm": 1.0158993005752563, | |
| "learning_rate": 9.45125199786894e-06, | |
| "loss": 2.1949, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 10.58, | |
| "grad_norm": 0.8538472056388855, | |
| "learning_rate": 9.424613745338307e-06, | |
| "loss": 2.1792, | |
| "step": 19850 | |
| }, | |
| { | |
| "epoch": 10.6, | |
| "grad_norm": 1.1701879501342773, | |
| "learning_rate": 9.397975492807672e-06, | |
| "loss": 2.1807, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 10.63, | |
| "grad_norm": 1.2362642288208008, | |
| "learning_rate": 9.371337240277039e-06, | |
| "loss": 2.1767, | |
| "step": 19950 | |
| }, | |
| { | |
| "epoch": 10.66, | |
| "grad_norm": 1.123152494430542, | |
| "learning_rate": 9.344698987746404e-06, | |
| "loss": 2.1598, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 10.68, | |
| "grad_norm": 0.953671932220459, | |
| "learning_rate": 9.31806073521577e-06, | |
| "loss": 2.1543, | |
| "step": 20050 | |
| }, | |
| { | |
| "epoch": 10.71, | |
| "grad_norm": 1.0523099899291992, | |
| "learning_rate": 9.291422482685136e-06, | |
| "loss": 2.2153, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 10.74, | |
| "grad_norm": 1.3634711503982544, | |
| "learning_rate": 9.264784230154503e-06, | |
| "loss": 2.1659, | |
| "step": 20150 | |
| }, | |
| { | |
| "epoch": 10.76, | |
| "grad_norm": 0.9536843299865723, | |
| "learning_rate": 9.23814597762387e-06, | |
| "loss": 2.1771, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 10.79, | |
| "grad_norm": 0.9196369647979736, | |
| "learning_rate": 9.211507725093234e-06, | |
| "loss": 2.1675, | |
| "step": 20250 | |
| }, | |
| { | |
| "epoch": 10.82, | |
| "grad_norm": 0.846582293510437, | |
| "learning_rate": 9.184869472562601e-06, | |
| "loss": 2.1461, | |
| "step": 20300 | |
| }, | |
| { | |
| "epoch": 10.84, | |
| "grad_norm": 0.9845914840698242, | |
| "learning_rate": 9.158231220031966e-06, | |
| "loss": 2.2221, | |
| "step": 20350 | |
| }, | |
| { | |
| "epoch": 10.87, | |
| "grad_norm": 1.0751748085021973, | |
| "learning_rate": 9.131592967501333e-06, | |
| "loss": 2.1662, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 10.9, | |
| "grad_norm": 0.9254376292228699, | |
| "learning_rate": 9.104954714970698e-06, | |
| "loss": 2.1693, | |
| "step": 20450 | |
| }, | |
| { | |
| "epoch": 10.92, | |
| "grad_norm": 0.9914495348930359, | |
| "learning_rate": 9.078316462440065e-06, | |
| "loss": 2.1774, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 10.95, | |
| "grad_norm": 1.060456395149231, | |
| "learning_rate": 9.05167820990943e-06, | |
| "loss": 2.1928, | |
| "step": 20550 | |
| }, | |
| { | |
| "epoch": 10.97, | |
| "grad_norm": 0.9841185212135315, | |
| "learning_rate": 9.025039957378797e-06, | |
| "loss": 2.1774, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_bert-score-f1": 0.8810000661683061, | |
| "eval_bert-score-precision": 0.8851854187195088, | |
| "eval_bert-score-recall": 0.8771485046185927, | |
| "eval_gen_len": 47.51730437430224, | |
| "eval_loss": 2.013378381729126, | |
| "eval_rouge1": 0.3348, | |
| "eval_rouge2": 0.1268, | |
| "eval_rougeL": 0.2476, | |
| "eval_rougeLsum": 0.2477, | |
| "eval_runtime": 1329.4509, | |
| "eval_samples_per_second": 14.823, | |
| "eval_steps_per_second": 0.354, | |
| "step": 20647 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "grad_norm": 0.9042637944221497, | |
| "learning_rate": 8.998401704848163e-06, | |
| "loss": 2.1635, | |
| "step": 20650 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "grad_norm": 0.9590170979499817, | |
| "learning_rate": 8.971763452317528e-06, | |
| "loss": 2.165, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 11.05, | |
| "grad_norm": 0.8360864520072937, | |
| "learning_rate": 8.945125199786895e-06, | |
| "loss": 2.1263, | |
| "step": 20750 | |
| }, | |
| { | |
| "epoch": 11.08, | |
| "grad_norm": 1.0940536260604858, | |
| "learning_rate": 8.91848694725626e-06, | |
| "loss": 2.1797, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 11.11, | |
| "grad_norm": 0.9590544104576111, | |
| "learning_rate": 8.891848694725627e-06, | |
| "loss": 2.1848, | |
| "step": 20850 | |
| }, | |
| { | |
| "epoch": 11.13, | |
| "grad_norm": 1.009817361831665, | |
| "learning_rate": 8.865210442194992e-06, | |
| "loss": 2.1582, | |
| "step": 20900 | |
| }, | |
| { | |
| "epoch": 11.16, | |
| "grad_norm": 0.9471531510353088, | |
| "learning_rate": 8.838572189664359e-06, | |
| "loss": 2.193, | |
| "step": 20950 | |
| }, | |
| { | |
| "epoch": 11.19, | |
| "grad_norm": 1.0452817678451538, | |
| "learning_rate": 8.811933937133724e-06, | |
| "loss": 2.1514, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 11.21, | |
| "grad_norm": 1.0049973726272583, | |
| "learning_rate": 8.78529568460309e-06, | |
| "loss": 2.1969, | |
| "step": 21050 | |
| }, | |
| { | |
| "epoch": 11.24, | |
| "grad_norm": 1.1841228008270264, | |
| "learning_rate": 8.758657432072457e-06, | |
| "loss": 2.1479, | |
| "step": 21100 | |
| }, | |
| { | |
| "epoch": 11.27, | |
| "grad_norm": 0.9284167289733887, | |
| "learning_rate": 8.732019179541822e-06, | |
| "loss": 2.1597, | |
| "step": 21150 | |
| }, | |
| { | |
| "epoch": 11.29, | |
| "grad_norm": 1.0026378631591797, | |
| "learning_rate": 8.705380927011189e-06, | |
| "loss": 2.1856, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 11.32, | |
| "grad_norm": 1.1704638004302979, | |
| "learning_rate": 8.678742674480554e-06, | |
| "loss": 2.1991, | |
| "step": 21250 | |
| }, | |
| { | |
| "epoch": 11.35, | |
| "grad_norm": 0.9646838307380676, | |
| "learning_rate": 8.65210442194992e-06, | |
| "loss": 2.1673, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 11.37, | |
| "grad_norm": 0.9169173240661621, | |
| "learning_rate": 8.625466169419286e-06, | |
| "loss": 2.1879, | |
| "step": 21350 | |
| }, | |
| { | |
| "epoch": 11.4, | |
| "grad_norm": 1.2698827981948853, | |
| "learning_rate": 8.598827916888653e-06, | |
| "loss": 2.1711, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 11.43, | |
| "grad_norm": 1.0014339685440063, | |
| "learning_rate": 8.57218966435802e-06, | |
| "loss": 2.1574, | |
| "step": 21450 | |
| }, | |
| { | |
| "epoch": 11.45, | |
| "grad_norm": 1.141926884651184, | |
| "learning_rate": 8.545551411827386e-06, | |
| "loss": 2.181, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 11.48, | |
| "grad_norm": 0.9993128180503845, | |
| "learning_rate": 8.518913159296751e-06, | |
| "loss": 2.1954, | |
| "step": 21550 | |
| }, | |
| { | |
| "epoch": 11.51, | |
| "grad_norm": 0.8284913301467896, | |
| "learning_rate": 8.492274906766118e-06, | |
| "loss": 2.1507, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 11.53, | |
| "grad_norm": 0.9283605217933655, | |
| "learning_rate": 8.465636654235483e-06, | |
| "loss": 2.164, | |
| "step": 21650 | |
| }, | |
| { | |
| "epoch": 11.56, | |
| "grad_norm": 1.0683674812316895, | |
| "learning_rate": 8.438998401704848e-06, | |
| "loss": 2.1689, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 11.59, | |
| "grad_norm": 0.9687473177909851, | |
| "learning_rate": 8.412360149174215e-06, | |
| "loss": 2.1447, | |
| "step": 21750 | |
| }, | |
| { | |
| "epoch": 11.61, | |
| "grad_norm": 1.2580022811889648, | |
| "learning_rate": 8.38572189664358e-06, | |
| "loss": 2.1895, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 11.64, | |
| "grad_norm": 0.9309559464454651, | |
| "learning_rate": 8.359083644112946e-06, | |
| "loss": 2.178, | |
| "step": 21850 | |
| }, | |
| { | |
| "epoch": 11.67, | |
| "grad_norm": 0.9081181883811951, | |
| "learning_rate": 8.332445391582313e-06, | |
| "loss": 2.1827, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 11.69, | |
| "grad_norm": 0.9058334231376648, | |
| "learning_rate": 8.30580713905168e-06, | |
| "loss": 2.1636, | |
| "step": 21950 | |
| }, | |
| { | |
| "epoch": 11.72, | |
| "grad_norm": 0.9327389597892761, | |
| "learning_rate": 8.279168886521045e-06, | |
| "loss": 2.175, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 11.75, | |
| "grad_norm": 1.3059749603271484, | |
| "learning_rate": 8.252530633990412e-06, | |
| "loss": 2.1655, | |
| "step": 22050 | |
| }, | |
| { | |
| "epoch": 11.77, | |
| "grad_norm": 0.9360121488571167, | |
| "learning_rate": 8.225892381459777e-06, | |
| "loss": 2.1759, | |
| "step": 22100 | |
| }, | |
| { | |
| "epoch": 11.8, | |
| "grad_norm": 2.760277032852173, | |
| "learning_rate": 8.199254128929144e-06, | |
| "loss": 2.1784, | |
| "step": 22150 | |
| }, | |
| { | |
| "epoch": 11.83, | |
| "grad_norm": 0.9712272882461548, | |
| "learning_rate": 8.172615876398509e-06, | |
| "loss": 2.1736, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 11.85, | |
| "grad_norm": 0.8907904028892517, | |
| "learning_rate": 8.145977623867875e-06, | |
| "loss": 2.1918, | |
| "step": 22250 | |
| }, | |
| { | |
| "epoch": 11.88, | |
| "grad_norm": 1.1707508563995361, | |
| "learning_rate": 8.119339371337242e-06, | |
| "loss": 2.1679, | |
| "step": 22300 | |
| }, | |
| { | |
| "epoch": 11.91, | |
| "grad_norm": 0.8738240003585815, | |
| "learning_rate": 8.092701118806607e-06, | |
| "loss": 2.1831, | |
| "step": 22350 | |
| }, | |
| { | |
| "epoch": 11.93, | |
| "grad_norm": 1.0194109678268433, | |
| "learning_rate": 8.066062866275974e-06, | |
| "loss": 2.1654, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 11.96, | |
| "grad_norm": 2.5614118576049805, | |
| "learning_rate": 8.039424613745339e-06, | |
| "loss": 2.1985, | |
| "step": 22450 | |
| }, | |
| { | |
| "epoch": 11.99, | |
| "grad_norm": 1.0855169296264648, | |
| "learning_rate": 8.012786361214706e-06, | |
| "loss": 2.2065, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_bert-score-f1": 0.8812093867313604, | |
| "eval_bert-score-precision": 0.8855559532955481, | |
| "eval_bert-score-recall": 0.8771963626476008, | |
| "eval_gen_len": 47.54465644981224, | |
| "eval_loss": 2.0089080333709717, | |
| "eval_rouge1": 0.3353, | |
| "eval_rouge2": 0.1269, | |
| "eval_rougeL": 0.2482, | |
| "eval_rougeLsum": 0.2483, | |
| "eval_runtime": 1322.3144, | |
| "eval_samples_per_second": 14.903, | |
| "eval_steps_per_second": 0.355, | |
| "step": 22524 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "grad_norm": 0.9317869544029236, | |
| "learning_rate": 7.98614810868407e-06, | |
| "loss": 2.1648, | |
| "step": 22550 | |
| }, | |
| { | |
| "epoch": 12.04, | |
| "grad_norm": 0.9064331650733948, | |
| "learning_rate": 7.959509856153437e-06, | |
| "loss": 2.1624, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 12.07, | |
| "grad_norm": 0.867645263671875, | |
| "learning_rate": 7.932871603622802e-06, | |
| "loss": 2.135, | |
| "step": 22650 | |
| }, | |
| { | |
| "epoch": 12.09, | |
| "grad_norm": 0.9419721364974976, | |
| "learning_rate": 7.90623335109217e-06, | |
| "loss": 2.1511, | |
| "step": 22700 | |
| }, | |
| { | |
| "epoch": 12.12, | |
| "grad_norm": 1.0487383604049683, | |
| "learning_rate": 7.879595098561536e-06, | |
| "loss": 2.2027, | |
| "step": 22750 | |
| }, | |
| { | |
| "epoch": 12.15, | |
| "grad_norm": 0.9865580797195435, | |
| "learning_rate": 7.852956846030901e-06, | |
| "loss": 2.1988, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 12.17, | |
| "grad_norm": 0.8484292030334473, | |
| "learning_rate": 7.826318593500268e-06, | |
| "loss": 2.1662, | |
| "step": 22850 | |
| }, | |
| { | |
| "epoch": 12.2, | |
| "grad_norm": 1.0033077001571655, | |
| "learning_rate": 7.799680340969633e-06, | |
| "loss": 2.1494, | |
| "step": 22900 | |
| }, | |
| { | |
| "epoch": 12.23, | |
| "grad_norm": 0.967682957649231, | |
| "learning_rate": 7.773042088439e-06, | |
| "loss": 2.149, | |
| "step": 22950 | |
| }, | |
| { | |
| "epoch": 12.25, | |
| "grad_norm": 1.0493502616882324, | |
| "learning_rate": 7.746403835908365e-06, | |
| "loss": 2.1783, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 12.28, | |
| "grad_norm": 1.7216830253601074, | |
| "learning_rate": 7.719765583377731e-06, | |
| "loss": 2.2004, | |
| "step": 23050 | |
| }, | |
| { | |
| "epoch": 12.31, | |
| "grad_norm": 0.8698610067367554, | |
| "learning_rate": 7.693127330847096e-06, | |
| "loss": 2.1873, | |
| "step": 23100 | |
| }, | |
| { | |
| "epoch": 12.33, | |
| "grad_norm": 0.9329569935798645, | |
| "learning_rate": 7.666489078316463e-06, | |
| "loss": 2.1386, | |
| "step": 23150 | |
| }, | |
| { | |
| "epoch": 12.36, | |
| "grad_norm": 1.0915803909301758, | |
| "learning_rate": 7.63985082578583e-06, | |
| "loss": 2.1847, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 12.39, | |
| "grad_norm": 0.945863664150238, | |
| "learning_rate": 7.613212573255196e-06, | |
| "loss": 2.1515, | |
| "step": 23250 | |
| }, | |
| { | |
| "epoch": 12.41, | |
| "grad_norm": 1.3315609693527222, | |
| "learning_rate": 7.586574320724561e-06, | |
| "loss": 2.1758, | |
| "step": 23300 | |
| }, | |
| { | |
| "epoch": 12.44, | |
| "grad_norm": 0.9587283134460449, | |
| "learning_rate": 7.559936068193927e-06, | |
| "loss": 2.1817, | |
| "step": 23350 | |
| }, | |
| { | |
| "epoch": 12.47, | |
| "grad_norm": 1.0035264492034912, | |
| "learning_rate": 7.533297815663293e-06, | |
| "loss": 2.2093, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 12.49, | |
| "grad_norm": 0.9835115671157837, | |
| "learning_rate": 7.5066595631326585e-06, | |
| "loss": 2.1545, | |
| "step": 23450 | |
| }, | |
| { | |
| "epoch": 12.52, | |
| "grad_norm": 1.103174090385437, | |
| "learning_rate": 7.480021310602024e-06, | |
| "loss": 2.1901, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 12.55, | |
| "grad_norm": 0.9141381978988647, | |
| "learning_rate": 7.45338305807139e-06, | |
| "loss": 2.1721, | |
| "step": 23550 | |
| }, | |
| { | |
| "epoch": 12.57, | |
| "grad_norm": 0.9373638033866882, | |
| "learning_rate": 7.426744805540758e-06, | |
| "loss": 2.1484, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 12.6, | |
| "grad_norm": 0.8388597369194031, | |
| "learning_rate": 7.400106553010124e-06, | |
| "loss": 2.1476, | |
| "step": 23650 | |
| }, | |
| { | |
| "epoch": 12.63, | |
| "grad_norm": 1.1504141092300415, | |
| "learning_rate": 7.37346830047949e-06, | |
| "loss": 2.1471, | |
| "step": 23700 | |
| }, | |
| { | |
| "epoch": 12.65, | |
| "grad_norm": 1.0790530443191528, | |
| "learning_rate": 7.3468300479488556e-06, | |
| "loss": 2.1905, | |
| "step": 23750 | |
| }, | |
| { | |
| "epoch": 12.68, | |
| "grad_norm": 1.2390844821929932, | |
| "learning_rate": 7.3201917954182215e-06, | |
| "loss": 2.1447, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 12.71, | |
| "grad_norm": 1.1462445259094238, | |
| "learning_rate": 7.293553542887587e-06, | |
| "loss": 2.176, | |
| "step": 23850 | |
| }, | |
| { | |
| "epoch": 12.73, | |
| "grad_norm": 1.1283916234970093, | |
| "learning_rate": 7.266915290356953e-06, | |
| "loss": 2.2012, | |
| "step": 23900 | |
| }, | |
| { | |
| "epoch": 12.76, | |
| "grad_norm": 0.9534170031547546, | |
| "learning_rate": 7.240277037826319e-06, | |
| "loss": 2.1723, | |
| "step": 23950 | |
| }, | |
| { | |
| "epoch": 12.79, | |
| "grad_norm": 1.0652652978897095, | |
| "learning_rate": 7.213638785295686e-06, | |
| "loss": 2.1637, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 12.81, | |
| "grad_norm": 0.9238381385803223, | |
| "learning_rate": 7.187000532765052e-06, | |
| "loss": 2.1684, | |
| "step": 24050 | |
| }, | |
| { | |
| "epoch": 12.84, | |
| "grad_norm": 1.0384712219238281, | |
| "learning_rate": 7.160362280234418e-06, | |
| "loss": 2.1827, | |
| "step": 24100 | |
| }, | |
| { | |
| "epoch": 12.87, | |
| "grad_norm": 1.0068541765213013, | |
| "learning_rate": 7.133724027703784e-06, | |
| "loss": 2.1631, | |
| "step": 24150 | |
| }, | |
| { | |
| "epoch": 12.89, | |
| "grad_norm": 1.0497870445251465, | |
| "learning_rate": 7.1070857751731495e-06, | |
| "loss": 2.1397, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 12.92, | |
| "grad_norm": 0.9750410318374634, | |
| "learning_rate": 7.080447522642515e-06, | |
| "loss": 2.1806, | |
| "step": 24250 | |
| }, | |
| { | |
| "epoch": 12.95, | |
| "grad_norm": 1.383784532546997, | |
| "learning_rate": 7.053809270111881e-06, | |
| "loss": 2.1835, | |
| "step": 24300 | |
| }, | |
| { | |
| "epoch": 12.97, | |
| "grad_norm": 1.006396770477295, | |
| "learning_rate": 7.027171017581247e-06, | |
| "loss": 2.1553, | |
| "step": 24350 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "grad_norm": 1.2312555313110352, | |
| "learning_rate": 7.000532765050613e-06, | |
| "loss": 2.1748, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_bert-score-f1": 0.8812570747936678, | |
| "eval_bert-score-precision": 0.8856106671139539, | |
| "eval_bert-score-recall": 0.8772335333635659, | |
| "eval_gen_len": 47.40327818938395, | |
| "eval_loss": 2.0081796646118164, | |
| "eval_rouge1": 0.3355, | |
| "eval_rouge2": 0.1268, | |
| "eval_rougeL": 0.2484, | |
| "eval_rougeLsum": 0.2485, | |
| "eval_runtime": 1330.0155, | |
| "eval_samples_per_second": 14.816, | |
| "eval_steps_per_second": 0.353, | |
| "step": 24401 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "grad_norm": 0.9376835227012634, | |
| "learning_rate": 6.97389451251998e-06, | |
| "loss": 2.177, | |
| "step": 24450 | |
| }, | |
| { | |
| "epoch": 13.05, | |
| "grad_norm": 0.8431729078292847, | |
| "learning_rate": 6.947256259989346e-06, | |
| "loss": 2.151, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 13.08, | |
| "grad_norm": 0.9604516625404358, | |
| "learning_rate": 6.920618007458712e-06, | |
| "loss": 2.1812, | |
| "step": 24550 | |
| }, | |
| { | |
| "epoch": 13.11, | |
| "grad_norm": 1.005323886871338, | |
| "learning_rate": 6.8939797549280775e-06, | |
| "loss": 2.1375, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 13.13, | |
| "grad_norm": 0.8559876680374146, | |
| "learning_rate": 6.867341502397443e-06, | |
| "loss": 2.1757, | |
| "step": 24650 | |
| }, | |
| { | |
| "epoch": 13.16, | |
| "grad_norm": 0.9345346689224243, | |
| "learning_rate": 6.840703249866809e-06, | |
| "loss": 2.1612, | |
| "step": 24700 | |
| }, | |
| { | |
| "epoch": 13.19, | |
| "grad_norm": 0.8913391828536987, | |
| "learning_rate": 6.814064997336175e-06, | |
| "loss": 2.1644, | |
| "step": 24750 | |
| }, | |
| { | |
| "epoch": 13.21, | |
| "grad_norm": 1.023695468902588, | |
| "learning_rate": 6.787426744805541e-06, | |
| "loss": 2.1737, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 13.24, | |
| "grad_norm": 0.9663516879081726, | |
| "learning_rate": 6.760788492274908e-06, | |
| "loss": 2.194, | |
| "step": 24850 | |
| }, | |
| { | |
| "epoch": 13.27, | |
| "grad_norm": 2.1459882259368896, | |
| "learning_rate": 6.734150239744274e-06, | |
| "loss": 2.1383, | |
| "step": 24900 | |
| }, | |
| { | |
| "epoch": 13.29, | |
| "grad_norm": 1.0092273950576782, | |
| "learning_rate": 6.70751198721364e-06, | |
| "loss": 2.1482, | |
| "step": 24950 | |
| }, | |
| { | |
| "epoch": 13.32, | |
| "grad_norm": 0.9323935508728027, | |
| "learning_rate": 6.6808737346830055e-06, | |
| "loss": 2.1585, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 13.35, | |
| "grad_norm": 0.9155877828598022, | |
| "learning_rate": 6.654235482152371e-06, | |
| "loss": 2.1752, | |
| "step": 25050 | |
| }, | |
| { | |
| "epoch": 13.37, | |
| "grad_norm": 1.0836352109909058, | |
| "learning_rate": 6.627597229621737e-06, | |
| "loss": 2.1702, | |
| "step": 25100 | |
| }, | |
| { | |
| "epoch": 13.4, | |
| "grad_norm": 1.0014092922210693, | |
| "learning_rate": 6.600958977091103e-06, | |
| "loss": 2.1667, | |
| "step": 25150 | |
| }, | |
| { | |
| "epoch": 13.43, | |
| "grad_norm": 0.917218804359436, | |
| "learning_rate": 6.574320724560469e-06, | |
| "loss": 2.174, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 13.45, | |
| "grad_norm": 0.9018741846084595, | |
| "learning_rate": 6.547682472029835e-06, | |
| "loss": 2.1612, | |
| "step": 25250 | |
| }, | |
| { | |
| "epoch": 13.48, | |
| "grad_norm": 1.0210672616958618, | |
| "learning_rate": 6.521044219499202e-06, | |
| "loss": 2.1516, | |
| "step": 25300 | |
| }, | |
| { | |
| "epoch": 13.51, | |
| "grad_norm": 0.9759948253631592, | |
| "learning_rate": 6.494405966968568e-06, | |
| "loss": 2.1683, | |
| "step": 25350 | |
| }, | |
| { | |
| "epoch": 13.53, | |
| "grad_norm": 0.8901129961013794, | |
| "learning_rate": 6.4677677144379335e-06, | |
| "loss": 2.1777, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 13.56, | |
| "grad_norm": 0.9080635905265808, | |
| "learning_rate": 6.4411294619072994e-06, | |
| "loss": 2.1821, | |
| "step": 25450 | |
| }, | |
| { | |
| "epoch": 13.59, | |
| "grad_norm": 0.9286373257637024, | |
| "learning_rate": 6.414491209376665e-06, | |
| "loss": 2.1779, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 13.61, | |
| "grad_norm": 1.1481138467788696, | |
| "learning_rate": 6.387852956846031e-06, | |
| "loss": 2.1692, | |
| "step": 25550 | |
| }, | |
| { | |
| "epoch": 13.64, | |
| "grad_norm": 0.9707843661308289, | |
| "learning_rate": 6.361214704315397e-06, | |
| "loss": 2.1682, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 13.67, | |
| "grad_norm": 1.042262077331543, | |
| "learning_rate": 6.334576451784763e-06, | |
| "loss": 2.1584, | |
| "step": 25650 | |
| }, | |
| { | |
| "epoch": 13.69, | |
| "grad_norm": 0.9933112859725952, | |
| "learning_rate": 6.30793819925413e-06, | |
| "loss": 2.1558, | |
| "step": 25700 | |
| }, | |
| { | |
| "epoch": 13.72, | |
| "grad_norm": 0.8889453411102295, | |
| "learning_rate": 6.281299946723496e-06, | |
| "loss": 2.1592, | |
| "step": 25750 | |
| }, | |
| { | |
| "epoch": 13.75, | |
| "grad_norm": 0.8605052828788757, | |
| "learning_rate": 6.2546616941928615e-06, | |
| "loss": 2.1993, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 13.77, | |
| "grad_norm": 0.946685791015625, | |
| "learning_rate": 6.2280234416622274e-06, | |
| "loss": 2.1609, | |
| "step": 25850 | |
| }, | |
| { | |
| "epoch": 13.8, | |
| "grad_norm": 0.9031264185905457, | |
| "learning_rate": 6.201385189131593e-06, | |
| "loss": 2.1578, | |
| "step": 25900 | |
| }, | |
| { | |
| "epoch": 13.83, | |
| "grad_norm": 1.1098228693008423, | |
| "learning_rate": 6.174746936600959e-06, | |
| "loss": 2.1487, | |
| "step": 25950 | |
| }, | |
| { | |
| "epoch": 13.85, | |
| "grad_norm": 0.9173837900161743, | |
| "learning_rate": 6.148108684070325e-06, | |
| "loss": 2.1639, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 13.88, | |
| "grad_norm": 1.1397854089736938, | |
| "learning_rate": 6.121470431539691e-06, | |
| "loss": 2.165, | |
| "step": 26050 | |
| }, | |
| { | |
| "epoch": 13.91, | |
| "grad_norm": 1.0020257234573364, | |
| "learning_rate": 6.094832179009057e-06, | |
| "loss": 2.1713, | |
| "step": 26100 | |
| }, | |
| { | |
| "epoch": 13.93, | |
| "grad_norm": 0.9584590792655945, | |
| "learning_rate": 6.068193926478424e-06, | |
| "loss": 2.2049, | |
| "step": 26150 | |
| }, | |
| { | |
| "epoch": 13.96, | |
| "grad_norm": 1.1951069831848145, | |
| "learning_rate": 6.0415556739477896e-06, | |
| "loss": 2.1632, | |
| "step": 26200 | |
| }, | |
| { | |
| "epoch": 13.99, | |
| "grad_norm": 1.0568323135375977, | |
| "learning_rate": 6.0149174214171555e-06, | |
| "loss": 2.1792, | |
| "step": 26250 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_bert-score-f1": 0.8812965099779855, | |
| "eval_bert-score-precision": 0.8854620127155008, | |
| "eval_bert-score-recall": 0.877461512582043, | |
| "eval_gen_len": 47.727494164213944, | |
| "eval_loss": 2.0084121227264404, | |
| "eval_rouge1": 0.3362, | |
| "eval_rouge2": 0.1273, | |
| "eval_rougeL": 0.2487, | |
| "eval_rougeLsum": 0.2487, | |
| "eval_runtime": 1350.2387, | |
| "eval_samples_per_second": 14.594, | |
| "eval_steps_per_second": 0.348, | |
| "step": 26278 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "grad_norm": 0.9785760045051575, | |
| "learning_rate": 5.988279168886521e-06, | |
| "loss": 2.1468, | |
| "step": 26300 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "grad_norm": 1.0162745714187622, | |
| "learning_rate": 5.961640916355887e-06, | |
| "loss": 2.1717, | |
| "step": 26350 | |
| }, | |
| { | |
| "epoch": 14.06, | |
| "grad_norm": 1.0883443355560303, | |
| "learning_rate": 5.935002663825253e-06, | |
| "loss": 2.1565, | |
| "step": 26400 | |
| }, | |
| { | |
| "epoch": 14.09, | |
| "grad_norm": 1.0060546398162842, | |
| "learning_rate": 5.908364411294619e-06, | |
| "loss": 2.1691, | |
| "step": 26450 | |
| }, | |
| { | |
| "epoch": 14.12, | |
| "grad_norm": 0.9702796936035156, | |
| "learning_rate": 5.881726158763985e-06, | |
| "loss": 2.1747, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 14.14, | |
| "grad_norm": 0.9459964632987976, | |
| "learning_rate": 5.8550879062333525e-06, | |
| "loss": 2.1781, | |
| "step": 26550 | |
| }, | |
| { | |
| "epoch": 14.17, | |
| "grad_norm": 0.8751854300498962, | |
| "learning_rate": 5.8284496537027176e-06, | |
| "loss": 2.1531, | |
| "step": 26600 | |
| }, | |
| { | |
| "epoch": 14.2, | |
| "grad_norm": 0.9021329879760742, | |
| "learning_rate": 5.8018114011720835e-06, | |
| "loss": 2.1741, | |
| "step": 26650 | |
| }, | |
| { | |
| "epoch": 14.22, | |
| "grad_norm": 0.9062153100967407, | |
| "learning_rate": 5.775173148641449e-06, | |
| "loss": 2.1424, | |
| "step": 26700 | |
| }, | |
| { | |
| "epoch": 14.25, | |
| "grad_norm": 0.8709902167320251, | |
| "learning_rate": 5.748534896110815e-06, | |
| "loss": 2.1458, | |
| "step": 26750 | |
| }, | |
| { | |
| "epoch": 14.28, | |
| "grad_norm": 0.8705712556838989, | |
| "learning_rate": 5.721896643580181e-06, | |
| "loss": 2.113, | |
| "step": 26800 | |
| }, | |
| { | |
| "epoch": 14.3, | |
| "grad_norm": 0.880886435508728, | |
| "learning_rate": 5.695258391049547e-06, | |
| "loss": 2.2009, | |
| "step": 26850 | |
| }, | |
| { | |
| "epoch": 14.33, | |
| "grad_norm": 1.1115506887435913, | |
| "learning_rate": 5.668620138518913e-06, | |
| "loss": 2.1809, | |
| "step": 26900 | |
| }, | |
| { | |
| "epoch": 14.36, | |
| "grad_norm": 0.9538648128509521, | |
| "learning_rate": 5.641981885988279e-06, | |
| "loss": 2.1776, | |
| "step": 26950 | |
| }, | |
| { | |
| "epoch": 14.38, | |
| "grad_norm": 0.8338669538497925, | |
| "learning_rate": 5.6153436334576464e-06, | |
| "loss": 2.1357, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 14.41, | |
| "grad_norm": 0.9421921372413635, | |
| "learning_rate": 5.588705380927012e-06, | |
| "loss": 2.1514, | |
| "step": 27050 | |
| }, | |
| { | |
| "epoch": 14.44, | |
| "grad_norm": 0.8788719177246094, | |
| "learning_rate": 5.562067128396378e-06, | |
| "loss": 2.1643, | |
| "step": 27100 | |
| }, | |
| { | |
| "epoch": 14.46, | |
| "grad_norm": 0.9146755337715149, | |
| "learning_rate": 5.535428875865744e-06, | |
| "loss": 2.1334, | |
| "step": 27150 | |
| }, | |
| { | |
| "epoch": 14.49, | |
| "grad_norm": 1.042197346687317, | |
| "learning_rate": 5.50879062333511e-06, | |
| "loss": 2.1927, | |
| "step": 27200 | |
| }, | |
| { | |
| "epoch": 14.52, | |
| "grad_norm": 0.9642235636711121, | |
| "learning_rate": 5.482152370804475e-06, | |
| "loss": 2.1628, | |
| "step": 27250 | |
| }, | |
| { | |
| "epoch": 14.54, | |
| "grad_norm": 1.0911927223205566, | |
| "learning_rate": 5.455514118273841e-06, | |
| "loss": 2.1643, | |
| "step": 27300 | |
| }, | |
| { | |
| "epoch": 14.57, | |
| "grad_norm": 1.1119507551193237, | |
| "learning_rate": 5.428875865743207e-06, | |
| "loss": 2.1757, | |
| "step": 27350 | |
| }, | |
| { | |
| "epoch": 14.6, | |
| "grad_norm": 0.9982576370239258, | |
| "learning_rate": 5.4022376132125744e-06, | |
| "loss": 2.1912, | |
| "step": 27400 | |
| }, | |
| { | |
| "epoch": 14.62, | |
| "grad_norm": 0.9314507842063904, | |
| "learning_rate": 5.37559936068194e-06, | |
| "loss": 2.171, | |
| "step": 27450 | |
| }, | |
| { | |
| "epoch": 14.65, | |
| "grad_norm": 1.2115583419799805, | |
| "learning_rate": 5.348961108151306e-06, | |
| "loss": 2.16, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 14.68, | |
| "grad_norm": 0.9271109700202942, | |
| "learning_rate": 5.322322855620672e-06, | |
| "loss": 2.1908, | |
| "step": 27550 | |
| }, | |
| { | |
| "epoch": 14.7, | |
| "grad_norm": 0.8988925218582153, | |
| "learning_rate": 5.295684603090038e-06, | |
| "loss": 2.1852, | |
| "step": 27600 | |
| }, | |
| { | |
| "epoch": 14.73, | |
| "grad_norm": 0.9780814051628113, | |
| "learning_rate": 5.269046350559404e-06, | |
| "loss": 2.1417, | |
| "step": 27650 | |
| }, | |
| { | |
| "epoch": 14.76, | |
| "grad_norm": 1.1456106901168823, | |
| "learning_rate": 5.24240809802877e-06, | |
| "loss": 2.1775, | |
| "step": 27700 | |
| }, | |
| { | |
| "epoch": 14.78, | |
| "grad_norm": 0.813176155090332, | |
| "learning_rate": 5.215769845498136e-06, | |
| "loss": 2.1916, | |
| "step": 27750 | |
| }, | |
| { | |
| "epoch": 14.81, | |
| "grad_norm": 0.9489790797233582, | |
| "learning_rate": 5.189131592967502e-06, | |
| "loss": 2.1561, | |
| "step": 27800 | |
| }, | |
| { | |
| "epoch": 14.84, | |
| "grad_norm": 0.9345203042030334, | |
| "learning_rate": 5.162493340436868e-06, | |
| "loss": 2.1511, | |
| "step": 27850 | |
| }, | |
| { | |
| "epoch": 14.86, | |
| "grad_norm": 0.7995460033416748, | |
| "learning_rate": 5.135855087906234e-06, | |
| "loss": 2.1918, | |
| "step": 27900 | |
| }, | |
| { | |
| "epoch": 14.89, | |
| "grad_norm": 0.8729720711708069, | |
| "learning_rate": 5.1092168353756e-06, | |
| "loss": 2.1598, | |
| "step": 27950 | |
| }, | |
| { | |
| "epoch": 14.92, | |
| "grad_norm": 0.928892970085144, | |
| "learning_rate": 5.082578582844966e-06, | |
| "loss": 2.1602, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 14.94, | |
| "grad_norm": 0.9653081297874451, | |
| "learning_rate": 5.055940330314332e-06, | |
| "loss": 2.1591, | |
| "step": 28050 | |
| }, | |
| { | |
| "epoch": 14.97, | |
| "grad_norm": 1.006082534790039, | |
| "learning_rate": 5.029302077783698e-06, | |
| "loss": 2.1554, | |
| "step": 28100 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "grad_norm": 0.9210333228111267, | |
| "learning_rate": 5.002663825253064e-06, | |
| "loss": 2.1609, | |
| "step": 28150 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_bert-score-f1": 0.8813901242448342, | |
| "eval_bert-score-precision": 0.8855356388553972, | |
| "eval_bert-score-recall": 0.8775755109870953, | |
| "eval_gen_len": 47.89211407693088, | |
| "eval_loss": 2.008052349090576, | |
| "eval_rouge1": 0.3364, | |
| "eval_rouge2": 0.1275, | |
| "eval_rougeL": 0.249, | |
| "eval_rougeLsum": 0.2491, | |
| "eval_runtime": 1342.1803, | |
| "eval_samples_per_second": 14.682, | |
| "eval_steps_per_second": 0.35, | |
| "step": 28155 | |
| }, | |
| { | |
| "epoch": 15.02, | |
| "grad_norm": 0.9067476987838745, | |
| "learning_rate": 4.97602557272243e-06, | |
| "loss": 2.1855, | |
| "step": 28200 | |
| }, | |
| { | |
| "epoch": 15.05, | |
| "grad_norm": 0.9924964904785156, | |
| "learning_rate": 4.9493873201917955e-06, | |
| "loss": 2.1434, | |
| "step": 28250 | |
| }, | |
| { | |
| "epoch": 15.08, | |
| "grad_norm": 0.9526183009147644, | |
| "learning_rate": 4.922749067661161e-06, | |
| "loss": 2.1306, | |
| "step": 28300 | |
| }, | |
| { | |
| "epoch": 15.1, | |
| "grad_norm": 0.8999783396720886, | |
| "learning_rate": 4.896110815130528e-06, | |
| "loss": 2.1683, | |
| "step": 28350 | |
| }, | |
| { | |
| "epoch": 15.13, | |
| "grad_norm": 1.3067386150360107, | |
| "learning_rate": 4.869472562599894e-06, | |
| "loss": 2.175, | |
| "step": 28400 | |
| }, | |
| { | |
| "epoch": 15.16, | |
| "grad_norm": 0.930998682975769, | |
| "learning_rate": 4.84283431006926e-06, | |
| "loss": 2.1571, | |
| "step": 28450 | |
| }, | |
| { | |
| "epoch": 15.18, | |
| "grad_norm": 1.0874801874160767, | |
| "learning_rate": 4.816196057538626e-06, | |
| "loss": 2.1648, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 15.21, | |
| "grad_norm": 0.8937533497810364, | |
| "learning_rate": 4.789557805007992e-06, | |
| "loss": 2.1593, | |
| "step": 28550 | |
| }, | |
| { | |
| "epoch": 15.24, | |
| "grad_norm": 1.1147658824920654, | |
| "learning_rate": 4.762919552477358e-06, | |
| "loss": 2.1554, | |
| "step": 28600 | |
| }, | |
| { | |
| "epoch": 15.26, | |
| "grad_norm": 1.2026888132095337, | |
| "learning_rate": 4.7362812999467235e-06, | |
| "loss": 2.1702, | |
| "step": 28650 | |
| }, | |
| { | |
| "epoch": 15.29, | |
| "grad_norm": 1.02727210521698, | |
| "learning_rate": 4.7096430474160894e-06, | |
| "loss": 2.1316, | |
| "step": 28700 | |
| }, | |
| { | |
| "epoch": 15.32, | |
| "grad_norm": 0.9779881834983826, | |
| "learning_rate": 4.683004794885455e-06, | |
| "loss": 2.1673, | |
| "step": 28750 | |
| }, | |
| { | |
| "epoch": 15.34, | |
| "grad_norm": 1.0025492906570435, | |
| "learning_rate": 4.656366542354822e-06, | |
| "loss": 2.1949, | |
| "step": 28800 | |
| }, | |
| { | |
| "epoch": 15.37, | |
| "grad_norm": 1.1579878330230713, | |
| "learning_rate": 4.629728289824188e-06, | |
| "loss": 2.161, | |
| "step": 28850 | |
| }, | |
| { | |
| "epoch": 15.4, | |
| "grad_norm": 0.9376475811004639, | |
| "learning_rate": 4.603090037293554e-06, | |
| "loss": 2.1849, | |
| "step": 28900 | |
| }, | |
| { | |
| "epoch": 15.42, | |
| "grad_norm": 1.239538550376892, | |
| "learning_rate": 4.57645178476292e-06, | |
| "loss": 2.148, | |
| "step": 28950 | |
| }, | |
| { | |
| "epoch": 15.45, | |
| "grad_norm": 0.9662672281265259, | |
| "learning_rate": 4.5498135322322865e-06, | |
| "loss": 2.1917, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 15.48, | |
| "grad_norm": 0.999487578868866, | |
| "learning_rate": 4.523175279701652e-06, | |
| "loss": 2.1746, | |
| "step": 29050 | |
| }, | |
| { | |
| "epoch": 15.5, | |
| "grad_norm": 1.1087113618850708, | |
| "learning_rate": 4.496537027171018e-06, | |
| "loss": 2.1756, | |
| "step": 29100 | |
| }, | |
| { | |
| "epoch": 15.53, | |
| "grad_norm": 0.9873372316360474, | |
| "learning_rate": 4.469898774640384e-06, | |
| "loss": 2.1333, | |
| "step": 29150 | |
| }, | |
| { | |
| "epoch": 15.56, | |
| "grad_norm": 0.9368143677711487, | |
| "learning_rate": 4.44326052210975e-06, | |
| "loss": 2.1784, | |
| "step": 29200 | |
| }, | |
| { | |
| "epoch": 15.58, | |
| "grad_norm": 1.0396299362182617, | |
| "learning_rate": 4.416622269579116e-06, | |
| "loss": 2.1786, | |
| "step": 29250 | |
| }, | |
| { | |
| "epoch": 15.61, | |
| "grad_norm": 0.9846010804176331, | |
| "learning_rate": 4.389984017048482e-06, | |
| "loss": 2.1886, | |
| "step": 29300 | |
| }, | |
| { | |
| "epoch": 15.64, | |
| "grad_norm": 0.7157499194145203, | |
| "learning_rate": 4.363345764517848e-06, | |
| "loss": 2.1454, | |
| "step": 29350 | |
| }, | |
| { | |
| "epoch": 15.66, | |
| "grad_norm": 0.8496772050857544, | |
| "learning_rate": 4.336707511987214e-06, | |
| "loss": 2.1456, | |
| "step": 29400 | |
| }, | |
| { | |
| "epoch": 15.69, | |
| "grad_norm": 1.0150104761123657, | |
| "learning_rate": 4.31006925945658e-06, | |
| "loss": 2.1725, | |
| "step": 29450 | |
| }, | |
| { | |
| "epoch": 15.72, | |
| "grad_norm": 0.9437685012817383, | |
| "learning_rate": 4.283431006925946e-06, | |
| "loss": 2.1655, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 15.74, | |
| "grad_norm": 0.978090226650238, | |
| "learning_rate": 4.256792754395312e-06, | |
| "loss": 2.1476, | |
| "step": 29550 | |
| }, | |
| { | |
| "epoch": 15.77, | |
| "grad_norm": 0.8782944679260254, | |
| "learning_rate": 4.230154501864678e-06, | |
| "loss": 2.1482, | |
| "step": 29600 | |
| }, | |
| { | |
| "epoch": 15.8, | |
| "grad_norm": 0.830117404460907, | |
| "learning_rate": 4.203516249334044e-06, | |
| "loss": 2.118, | |
| "step": 29650 | |
| }, | |
| { | |
| "epoch": 15.82, | |
| "grad_norm": 0.9760297536849976, | |
| "learning_rate": 4.17687799680341e-06, | |
| "loss": 2.2007, | |
| "step": 29700 | |
| }, | |
| { | |
| "epoch": 15.85, | |
| "grad_norm": 0.8977949619293213, | |
| "learning_rate": 4.150239744272776e-06, | |
| "loss": 2.1008, | |
| "step": 29750 | |
| }, | |
| { | |
| "epoch": 15.88, | |
| "grad_norm": 1.0869261026382446, | |
| "learning_rate": 4.123601491742142e-06, | |
| "loss": 2.1915, | |
| "step": 29800 | |
| }, | |
| { | |
| "epoch": 15.9, | |
| "grad_norm": 1.1186174154281616, | |
| "learning_rate": 4.0969632392115084e-06, | |
| "loss": 2.1653, | |
| "step": 29850 | |
| }, | |
| { | |
| "epoch": 15.93, | |
| "grad_norm": 0.9164936542510986, | |
| "learning_rate": 4.070324986680874e-06, | |
| "loss": 2.1479, | |
| "step": 29900 | |
| }, | |
| { | |
| "epoch": 15.96, | |
| "grad_norm": 0.9677796959877014, | |
| "learning_rate": 4.04368673415024e-06, | |
| "loss": 2.1207, | |
| "step": 29950 | |
| }, | |
| { | |
| "epoch": 15.98, | |
| "grad_norm": 0.8449875116348267, | |
| "learning_rate": 4.017048481619606e-06, | |
| "loss": 2.189, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_bert-score-f1": 0.8815181989403869, | |
| "eval_bert-score-precision": 0.8858505654647171, | |
| "eval_bert-score-recall": 0.8775155795209725, | |
| "eval_gen_len": 47.55181163097534, | |
| "eval_loss": 2.0050902366638184, | |
| "eval_rouge1": 0.3365, | |
| "eval_rouge2": 0.1273, | |
| "eval_rougeL": 0.249, | |
| "eval_rougeLsum": 0.2491, | |
| "eval_runtime": 1330.728, | |
| "eval_samples_per_second": 14.808, | |
| "eval_steps_per_second": 0.353, | |
| "step": 30032 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "grad_norm": 0.9980852603912354, | |
| "learning_rate": 3.990410229088972e-06, | |
| "loss": 2.1959, | |
| "step": 30050 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "grad_norm": 1.3163436651229858, | |
| "learning_rate": 3.963771976558338e-06, | |
| "loss": 2.173, | |
| "step": 30100 | |
| }, | |
| { | |
| "epoch": 16.06, | |
| "grad_norm": 0.9099076986312866, | |
| "learning_rate": 3.937133724027704e-06, | |
| "loss": 2.1724, | |
| "step": 30150 | |
| }, | |
| { | |
| "epoch": 16.09, | |
| "grad_norm": 1.0009883642196655, | |
| "learning_rate": 3.91049547149707e-06, | |
| "loss": 2.1544, | |
| "step": 30200 | |
| }, | |
| { | |
| "epoch": 16.12, | |
| "grad_norm": 0.9975098967552185, | |
| "learning_rate": 3.883857218966436e-06, | |
| "loss": 2.1458, | |
| "step": 30250 | |
| }, | |
| { | |
| "epoch": 16.14, | |
| "grad_norm": 0.8518524169921875, | |
| "learning_rate": 3.857218966435802e-06, | |
| "loss": 2.1569, | |
| "step": 30300 | |
| }, | |
| { | |
| "epoch": 16.17, | |
| "grad_norm": 0.9639623761177063, | |
| "learning_rate": 3.830580713905168e-06, | |
| "loss": 2.15, | |
| "step": 30350 | |
| }, | |
| { | |
| "epoch": 16.2, | |
| "grad_norm": 0.8993538022041321, | |
| "learning_rate": 3.803942461374534e-06, | |
| "loss": 2.1448, | |
| "step": 30400 | |
| }, | |
| { | |
| "epoch": 16.22, | |
| "grad_norm": 1.0397539138793945, | |
| "learning_rate": 3.7773042088439e-06, | |
| "loss": 2.1305, | |
| "step": 30450 | |
| }, | |
| { | |
| "epoch": 16.25, | |
| "grad_norm": 1.0676199197769165, | |
| "learning_rate": 3.7506659563132663e-06, | |
| "loss": 2.141, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 16.28, | |
| "grad_norm": 1.0054043531417847, | |
| "learning_rate": 3.7240277037826322e-06, | |
| "loss": 2.1595, | |
| "step": 30550 | |
| }, | |
| { | |
| "epoch": 16.3, | |
| "grad_norm": 0.949520468711853, | |
| "learning_rate": 3.697389451251998e-06, | |
| "loss": 2.2005, | |
| "step": 30600 | |
| }, | |
| { | |
| "epoch": 16.33, | |
| "grad_norm": 0.9753077030181885, | |
| "learning_rate": 3.670751198721364e-06, | |
| "loss": 2.1924, | |
| "step": 30650 | |
| }, | |
| { | |
| "epoch": 16.36, | |
| "grad_norm": 0.8877549767494202, | |
| "learning_rate": 3.6441129461907304e-06, | |
| "loss": 2.1553, | |
| "step": 30700 | |
| }, | |
| { | |
| "epoch": 16.38, | |
| "grad_norm": 1.0130952596664429, | |
| "learning_rate": 3.6174746936600963e-06, | |
| "loss": 2.1451, | |
| "step": 30750 | |
| }, | |
| { | |
| "epoch": 16.41, | |
| "grad_norm": 1.2479366064071655, | |
| "learning_rate": 3.590836441129462e-06, | |
| "loss": 2.168, | |
| "step": 30800 | |
| }, | |
| { | |
| "epoch": 16.44, | |
| "grad_norm": 1.0229461193084717, | |
| "learning_rate": 3.564198188598828e-06, | |
| "loss": 2.1642, | |
| "step": 30850 | |
| }, | |
| { | |
| "epoch": 16.46, | |
| "grad_norm": 0.9332823157310486, | |
| "learning_rate": 3.5375599360681944e-06, | |
| "loss": 2.1587, | |
| "step": 30900 | |
| }, | |
| { | |
| "epoch": 16.49, | |
| "grad_norm": 1.008484125137329, | |
| "learning_rate": 3.5109216835375603e-06, | |
| "loss": 2.1725, | |
| "step": 30950 | |
| }, | |
| { | |
| "epoch": 16.52, | |
| "grad_norm": 0.9926919341087341, | |
| "learning_rate": 3.484283431006926e-06, | |
| "loss": 2.1694, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 16.54, | |
| "grad_norm": 0.9659560322761536, | |
| "learning_rate": 3.457645178476292e-06, | |
| "loss": 2.1787, | |
| "step": 31050 | |
| }, | |
| { | |
| "epoch": 16.57, | |
| "grad_norm": 0.9388399720191956, | |
| "learning_rate": 3.431006925945658e-06, | |
| "loss": 2.1726, | |
| "step": 31100 | |
| }, | |
| { | |
| "epoch": 16.6, | |
| "grad_norm": 0.8616447448730469, | |
| "learning_rate": 3.4043686734150243e-06, | |
| "loss": 2.1566, | |
| "step": 31150 | |
| }, | |
| { | |
| "epoch": 16.62, | |
| "grad_norm": 0.8849464058876038, | |
| "learning_rate": 3.37773042088439e-06, | |
| "loss": 2.1769, | |
| "step": 31200 | |
| }, | |
| { | |
| "epoch": 16.65, | |
| "grad_norm": 0.9730740785598755, | |
| "learning_rate": 3.351092168353756e-06, | |
| "loss": 2.1858, | |
| "step": 31250 | |
| }, | |
| { | |
| "epoch": 16.68, | |
| "grad_norm": 1.1059538125991821, | |
| "learning_rate": 3.324453915823122e-06, | |
| "loss": 2.1722, | |
| "step": 31300 | |
| }, | |
| { | |
| "epoch": 16.7, | |
| "grad_norm": 1.2941957712173462, | |
| "learning_rate": 3.2978156632924883e-06, | |
| "loss": 2.1652, | |
| "step": 31350 | |
| }, | |
| { | |
| "epoch": 16.73, | |
| "grad_norm": 0.9166463017463684, | |
| "learning_rate": 3.271177410761854e-06, | |
| "loss": 2.158, | |
| "step": 31400 | |
| }, | |
| { | |
| "epoch": 16.76, | |
| "grad_norm": 1.1810513734817505, | |
| "learning_rate": 3.24453915823122e-06, | |
| "loss": 2.1334, | |
| "step": 31450 | |
| }, | |
| { | |
| "epoch": 16.78, | |
| "grad_norm": 0.9561477899551392, | |
| "learning_rate": 3.217900905700586e-06, | |
| "loss": 2.1354, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 16.81, | |
| "grad_norm": 0.8539523482322693, | |
| "learning_rate": 3.1912626531699527e-06, | |
| "loss": 2.1458, | |
| "step": 31550 | |
| }, | |
| { | |
| "epoch": 16.84, | |
| "grad_norm": 1.099672794342041, | |
| "learning_rate": 3.1646244006393186e-06, | |
| "loss": 2.1784, | |
| "step": 31600 | |
| }, | |
| { | |
| "epoch": 16.86, | |
| "grad_norm": 0.8352581858634949, | |
| "learning_rate": 3.1379861481086845e-06, | |
| "loss": 2.1348, | |
| "step": 31650 | |
| }, | |
| { | |
| "epoch": 16.89, | |
| "grad_norm": 1.0823798179626465, | |
| "learning_rate": 3.11134789557805e-06, | |
| "loss": 2.1657, | |
| "step": 31700 | |
| }, | |
| { | |
| "epoch": 16.92, | |
| "grad_norm": 0.9096495509147644, | |
| "learning_rate": 3.0847096430474167e-06, | |
| "loss": 2.1526, | |
| "step": 31750 | |
| }, | |
| { | |
| "epoch": 16.94, | |
| "grad_norm": 0.9881473779678345, | |
| "learning_rate": 3.0580713905167826e-06, | |
| "loss": 2.1759, | |
| "step": 31800 | |
| }, | |
| { | |
| "epoch": 16.97, | |
| "grad_norm": 1.2694969177246094, | |
| "learning_rate": 3.0314331379861485e-06, | |
| "loss": 2.1896, | |
| "step": 31850 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "grad_norm": 1.1195552349090576, | |
| "learning_rate": 3.0047948854555144e-06, | |
| "loss": 2.1417, | |
| "step": 31900 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_bert-score-f1": 0.8814209539291642, | |
| "eval_bert-score-precision": 0.8856624760483045, | |
| "eval_bert-score-recall": 0.8775075128371735, | |
| "eval_gen_len": 47.645996143306604, | |
| "eval_loss": 2.005204439163208, | |
| "eval_rouge1": 0.3363, | |
| "eval_rouge2": 0.1273, | |
| "eval_rougeL": 0.2488, | |
| "eval_rougeLsum": 0.2489, | |
| "eval_runtime": 1335.7013, | |
| "eval_samples_per_second": 14.753, | |
| "eval_steps_per_second": 0.352, | |
| "step": 31909 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "grad_norm": 0.8519843816757202, | |
| "learning_rate": 2.9781566329248803e-06, | |
| "loss": 2.139, | |
| "step": 31950 | |
| }, | |
| { | |
| "epoch": 17.05, | |
| "grad_norm": 0.9282755255699158, | |
| "learning_rate": 2.9515183803942466e-06, | |
| "loss": 2.1555, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 17.08, | |
| "grad_norm": 1.0514159202575684, | |
| "learning_rate": 2.9248801278636125e-06, | |
| "loss": 2.1316, | |
| "step": 32050 | |
| }, | |
| { | |
| "epoch": 17.1, | |
| "grad_norm": 1.034957766532898, | |
| "learning_rate": 2.8982418753329784e-06, | |
| "loss": 2.2094, | |
| "step": 32100 | |
| }, | |
| { | |
| "epoch": 17.13, | |
| "grad_norm": 1.0574824810028076, | |
| "learning_rate": 2.8716036228023443e-06, | |
| "loss": 2.1641, | |
| "step": 32150 | |
| }, | |
| { | |
| "epoch": 17.16, | |
| "grad_norm": 0.9753189086914062, | |
| "learning_rate": 2.8449653702717106e-06, | |
| "loss": 2.1442, | |
| "step": 32200 | |
| }, | |
| { | |
| "epoch": 17.18, | |
| "grad_norm": 1.2037853002548218, | |
| "learning_rate": 2.8183271177410765e-06, | |
| "loss": 2.147, | |
| "step": 32250 | |
| }, | |
| { | |
| "epoch": 17.21, | |
| "grad_norm": 0.8287584185600281, | |
| "learning_rate": 2.7916888652104424e-06, | |
| "loss": 2.1765, | |
| "step": 32300 | |
| }, | |
| { | |
| "epoch": 17.23, | |
| "grad_norm": 1.0360517501831055, | |
| "learning_rate": 2.7650506126798083e-06, | |
| "loss": 2.1864, | |
| "step": 32350 | |
| }, | |
| { | |
| "epoch": 17.26, | |
| "grad_norm": 1.1471740007400513, | |
| "learning_rate": 2.7384123601491746e-06, | |
| "loss": 2.1629, | |
| "step": 32400 | |
| }, | |
| { | |
| "epoch": 17.29, | |
| "grad_norm": 0.9185000658035278, | |
| "learning_rate": 2.7117741076185405e-06, | |
| "loss": 2.1639, | |
| "step": 32450 | |
| }, | |
| { | |
| "epoch": 17.31, | |
| "grad_norm": 1.1663320064544678, | |
| "learning_rate": 2.6851358550879064e-06, | |
| "loss": 2.1174, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 17.34, | |
| "grad_norm": 0.980441689491272, | |
| "learning_rate": 2.6584976025572723e-06, | |
| "loss": 2.1397, | |
| "step": 32550 | |
| }, | |
| { | |
| "epoch": 17.37, | |
| "grad_norm": 1.1273858547210693, | |
| "learning_rate": 2.6318593500266386e-06, | |
| "loss": 2.1462, | |
| "step": 32600 | |
| }, | |
| { | |
| "epoch": 17.39, | |
| "grad_norm": 1.146296501159668, | |
| "learning_rate": 2.6052210974960045e-06, | |
| "loss": 2.1797, | |
| "step": 32650 | |
| }, | |
| { | |
| "epoch": 17.42, | |
| "grad_norm": 0.9205301403999329, | |
| "learning_rate": 2.5785828449653704e-06, | |
| "loss": 2.1909, | |
| "step": 32700 | |
| }, | |
| { | |
| "epoch": 17.45, | |
| "grad_norm": 1.103619933128357, | |
| "learning_rate": 2.5519445924347363e-06, | |
| "loss": 2.1637, | |
| "step": 32750 | |
| }, | |
| { | |
| "epoch": 17.47, | |
| "grad_norm": 0.8971651196479797, | |
| "learning_rate": 2.5253063399041022e-06, | |
| "loss": 2.1693, | |
| "step": 32800 | |
| }, | |
| { | |
| "epoch": 17.5, | |
| "grad_norm": 1.088956356048584, | |
| "learning_rate": 2.4986680873734685e-06, | |
| "loss": 2.151, | |
| "step": 32850 | |
| }, | |
| { | |
| "epoch": 17.53, | |
| "grad_norm": 0.9202156066894531, | |
| "learning_rate": 2.4720298348428344e-06, | |
| "loss": 2.1679, | |
| "step": 32900 | |
| }, | |
| { | |
| "epoch": 17.55, | |
| "grad_norm": 0.9409503936767578, | |
| "learning_rate": 2.4453915823122003e-06, | |
| "loss": 2.157, | |
| "step": 32950 | |
| }, | |
| { | |
| "epoch": 17.58, | |
| "grad_norm": 1.0779467821121216, | |
| "learning_rate": 2.4187533297815667e-06, | |
| "loss": 2.1291, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 17.61, | |
| "grad_norm": 0.918696403503418, | |
| "learning_rate": 2.3921150772509325e-06, | |
| "loss": 2.1844, | |
| "step": 33050 | |
| }, | |
| { | |
| "epoch": 17.63, | |
| "grad_norm": 1.5606318712234497, | |
| "learning_rate": 2.3654768247202984e-06, | |
| "loss": 2.1489, | |
| "step": 33100 | |
| }, | |
| { | |
| "epoch": 17.66, | |
| "grad_norm": 1.205295443534851, | |
| "learning_rate": 2.3388385721896643e-06, | |
| "loss": 2.1667, | |
| "step": 33150 | |
| }, | |
| { | |
| "epoch": 17.69, | |
| "grad_norm": 1.052687644958496, | |
| "learning_rate": 2.3122003196590307e-06, | |
| "loss": 2.1258, | |
| "step": 33200 | |
| }, | |
| { | |
| "epoch": 17.71, | |
| "grad_norm": 0.9863780736923218, | |
| "learning_rate": 2.2855620671283966e-06, | |
| "loss": 2.1722, | |
| "step": 33250 | |
| }, | |
| { | |
| "epoch": 17.74, | |
| "grad_norm": 0.9677025079727173, | |
| "learning_rate": 2.258923814597763e-06, | |
| "loss": 2.1813, | |
| "step": 33300 | |
| }, | |
| { | |
| "epoch": 17.77, | |
| "grad_norm": 1.0325006246566772, | |
| "learning_rate": 2.2322855620671283e-06, | |
| "loss": 2.1258, | |
| "step": 33350 | |
| }, | |
| { | |
| "epoch": 17.79, | |
| "grad_norm": 1.2607378959655762, | |
| "learning_rate": 2.2056473095364947e-06, | |
| "loss": 2.1531, | |
| "step": 33400 | |
| }, | |
| { | |
| "epoch": 17.82, | |
| "grad_norm": 0.8155959844589233, | |
| "learning_rate": 2.1790090570058606e-06, | |
| "loss": 2.1356, | |
| "step": 33450 | |
| }, | |
| { | |
| "epoch": 17.85, | |
| "grad_norm": 1.000097632408142, | |
| "learning_rate": 2.152370804475227e-06, | |
| "loss": 2.1564, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 17.87, | |
| "grad_norm": 1.0435174703598022, | |
| "learning_rate": 2.1257325519445928e-06, | |
| "loss": 2.1442, | |
| "step": 33550 | |
| }, | |
| { | |
| "epoch": 17.9, | |
| "grad_norm": 0.8841784000396729, | |
| "learning_rate": 2.0990942994139587e-06, | |
| "loss": 2.1775, | |
| "step": 33600 | |
| }, | |
| { | |
| "epoch": 17.93, | |
| "grad_norm": 0.971834123134613, | |
| "learning_rate": 2.0724560468833246e-06, | |
| "loss": 2.1737, | |
| "step": 33650 | |
| }, | |
| { | |
| "epoch": 17.95, | |
| "grad_norm": 1.0215661525726318, | |
| "learning_rate": 2.0458177943526905e-06, | |
| "loss": 2.1105, | |
| "step": 33700 | |
| }, | |
| { | |
| "epoch": 17.98, | |
| "grad_norm": 0.964297890663147, | |
| "learning_rate": 2.0191795418220568e-06, | |
| "loss": 2.1731, | |
| "step": 33750 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_bert-score-f1": 0.8815276650811141, | |
| "eval_bert-score-precision": 0.8858032303631735, | |
| "eval_bert-score-recall": 0.8775806915883447, | |
| "eval_gen_len": 47.6104739673196, | |
| "eval_loss": 2.004894495010376, | |
| "eval_rouge1": 0.3366, | |
| "eval_rouge2": 0.1273, | |
| "eval_rougeL": 0.249, | |
| "eval_rougeLsum": 0.2491, | |
| "eval_runtime": 1338.1647, | |
| "eval_samples_per_second": 14.726, | |
| "eval_steps_per_second": 0.351, | |
| "step": 33786 | |
| } | |
| ], | |
| "logging_steps": 50, | |
| "max_steps": 37540, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 20, | |
| "save_steps": 500, | |
| "total_flos": 3.840531477455831e+17, | |
| "train_batch_size": 42, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |