diff --git "a/checkpoint-431935/trainer_state.json" "b/checkpoint-431935/trainer_state.json" deleted file mode 100644--- "a/checkpoint-431935/trainer_state.json" +++ /dev/null @@ -1,5299 +0,0 @@ -{ - "best_metric": 0.21080316603183746, - "best_model_checkpoint": "./models/bart/bart_balanced_subset_without_title_less_than_1024_double_comment/checkpoint-431935", - "epoch": 5.0, - "global_step": 431935, - "is_hyper_param_search": false, - "is_local_process_zero": true, - "is_world_process_zero": true, - "log_history": [ - { - "epoch": 0.01, - "learning_rate": 1.9996141394731464e-05, - "loss": 1.4863, - "step": 500 - }, - { - "epoch": 0.01, - "learning_rate": 1.9992282789462922e-05, - "loss": 0.2641, - "step": 1000 - }, - { - "epoch": 0.02, - "learning_rate": 1.9988424184194384e-05, - "loss": 0.246, - "step": 1500 - }, - { - "epoch": 0.02, - "learning_rate": 1.9984565578925843e-05, - "loss": 0.2419, - "step": 2000 - }, - { - "epoch": 0.03, - "learning_rate": 1.9980706973657305e-05, - "loss": 0.2478, - "step": 2500 - }, - { - "epoch": 0.03, - "learning_rate": 1.9976848368388764e-05, - "loss": 0.2509, - "step": 3000 - }, - { - "epoch": 0.04, - "learning_rate": 1.9972989763120226e-05, - "loss": 0.2526, - "step": 3500 - }, - { - "epoch": 0.05, - "learning_rate": 1.9969131157851684e-05, - "loss": 0.2584, - "step": 4000 - }, - { - "epoch": 0.05, - "learning_rate": 1.9965272552583146e-05, - "loss": 0.2449, - "step": 4500 - }, - { - "epoch": 0.06, - "learning_rate": 1.9961413947314605e-05, - "loss": 0.2346, - "step": 5000 - }, - { - "epoch": 0.06, - "learning_rate": 1.9957555342046067e-05, - "loss": 0.239, - "step": 5500 - }, - { - "epoch": 0.07, - "learning_rate": 1.9953696736777526e-05, - "loss": 0.2449, - "step": 6000 - }, - { - "epoch": 0.08, - "learning_rate": 1.9949838131508988e-05, - "loss": 0.2428, - "step": 6500 - }, - { - "epoch": 0.08, - "learning_rate": 1.9945979526240446e-05, - "loss": 0.2435, - "step": 7000 - }, - { - "epoch": 0.09, - "learning_rate": 1.9942120920971908e-05, - "loss": 0.2322, - "step": 7500 - }, - { - "epoch": 0.09, - "learning_rate": 1.9938262315703367e-05, - "loss": 0.2374, - "step": 8000 - }, - { - "epoch": 0.1, - "learning_rate": 1.993440371043483e-05, - "loss": 0.2335, - "step": 8500 - }, - { - "epoch": 0.1, - "learning_rate": 1.9930545105166287e-05, - "loss": 0.2344, - "step": 9000 - }, - { - "epoch": 0.11, - "learning_rate": 1.992668649989775e-05, - "loss": 0.2241, - "step": 9500 - }, - { - "epoch": 0.12, - "learning_rate": 1.9922827894629208e-05, - "loss": 0.2374, - "step": 10000 - }, - { - "epoch": 0.12, - "learning_rate": 1.991896928936067e-05, - "loss": 0.2461, - "step": 10500 - }, - { - "epoch": 0.13, - "learning_rate": 1.991511068409213e-05, - "loss": 0.243, - "step": 11000 - }, - { - "epoch": 0.13, - "learning_rate": 1.991125207882359e-05, - "loss": 0.2309, - "step": 11500 - }, - { - "epoch": 0.14, - "learning_rate": 1.990739347355505e-05, - "loss": 0.2403, - "step": 12000 - }, - { - "epoch": 0.14, - "learning_rate": 1.990353486828651e-05, - "loss": 0.2425, - "step": 12500 - }, - { - "epoch": 0.15, - "learning_rate": 1.989967626301797e-05, - "loss": 0.2204, - "step": 13000 - }, - { - "epoch": 0.16, - "learning_rate": 1.9895817657749432e-05, - "loss": 0.237, - "step": 13500 - }, - { - "epoch": 0.16, - "learning_rate": 1.989195905248089e-05, - "loss": 0.2314, - "step": 14000 - }, - { - "epoch": 0.17, - "learning_rate": 1.9888100447212353e-05, - "loss": 0.2321, - "step": 14500 - }, - { - "epoch": 0.17, - "learning_rate": 1.988424184194381e-05, - "loss": 0.2366, - "step": 15000 - }, - { - "epoch": 0.18, - "learning_rate": 1.9880383236675273e-05, - "loss": 0.2368, - "step": 15500 - }, - { - "epoch": 0.19, - "learning_rate": 1.9876524631406732e-05, - "loss": 0.2292, - "step": 16000 - }, - { - "epoch": 0.19, - "learning_rate": 1.9872666026138194e-05, - "loss": 0.2334, - "step": 16500 - }, - { - "epoch": 0.2, - "learning_rate": 1.9868807420869653e-05, - "loss": 0.237, - "step": 17000 - }, - { - "epoch": 0.2, - "learning_rate": 1.9864948815601115e-05, - "loss": 0.237, - "step": 17500 - }, - { - "epoch": 0.21, - "learning_rate": 1.9861090210332573e-05, - "loss": 0.2338, - "step": 18000 - }, - { - "epoch": 0.21, - "learning_rate": 1.9857231605064035e-05, - "loss": 0.2305, - "step": 18500 - }, - { - "epoch": 0.22, - "learning_rate": 1.9853372999795494e-05, - "loss": 0.2334, - "step": 19000 - }, - { - "epoch": 0.23, - "learning_rate": 1.9849514394526956e-05, - "loss": 0.2396, - "step": 19500 - }, - { - "epoch": 0.23, - "learning_rate": 1.9845655789258415e-05, - "loss": 0.2432, - "step": 20000 - }, - { - "epoch": 0.24, - "learning_rate": 1.9841797183989877e-05, - "loss": 0.2294, - "step": 20500 - }, - { - "epoch": 0.24, - "learning_rate": 1.9837938578721335e-05, - "loss": 0.2305, - "step": 21000 - }, - { - "epoch": 0.25, - "learning_rate": 1.9834079973452797e-05, - "loss": 0.2415, - "step": 21500 - }, - { - "epoch": 0.25, - "learning_rate": 1.9830221368184256e-05, - "loss": 0.2412, - "step": 22000 - }, - { - "epoch": 0.26, - "learning_rate": 1.9826362762915718e-05, - "loss": 0.2329, - "step": 22500 - }, - { - "epoch": 0.27, - "learning_rate": 1.9822504157647176e-05, - "loss": 0.2354, - "step": 23000 - }, - { - "epoch": 0.27, - "learning_rate": 1.981864555237864e-05, - "loss": 0.2372, - "step": 23500 - }, - { - "epoch": 0.28, - "learning_rate": 1.9814786947110097e-05, - "loss": 0.2401, - "step": 24000 - }, - { - "epoch": 0.28, - "learning_rate": 1.981092834184156e-05, - "loss": 0.2333, - "step": 24500 - }, - { - "epoch": 0.29, - "learning_rate": 1.9807069736573018e-05, - "loss": 0.2301, - "step": 25000 - }, - { - "epoch": 0.3, - "learning_rate": 1.980321113130448e-05, - "loss": 0.2423, - "step": 25500 - }, - { - "epoch": 0.3, - "learning_rate": 1.979935252603594e-05, - "loss": 0.2171, - "step": 26000 - }, - { - "epoch": 0.31, - "learning_rate": 1.97954939207674e-05, - "loss": 0.2347, - "step": 26500 - }, - { - "epoch": 0.31, - "learning_rate": 1.979163531549886e-05, - "loss": 0.2294, - "step": 27000 - }, - { - "epoch": 0.32, - "learning_rate": 1.978777671023032e-05, - "loss": 0.2286, - "step": 27500 - }, - { - "epoch": 0.32, - "learning_rate": 1.978391810496178e-05, - "loss": 0.2355, - "step": 28000 - }, - { - "epoch": 0.33, - "learning_rate": 1.9780059499693242e-05, - "loss": 0.2211, - "step": 28500 - }, - { - "epoch": 0.34, - "learning_rate": 1.97762008944247e-05, - "loss": 0.2222, - "step": 29000 - }, - { - "epoch": 0.34, - "learning_rate": 1.9772342289156162e-05, - "loss": 0.2264, - "step": 29500 - }, - { - "epoch": 0.35, - "learning_rate": 1.976848368388762e-05, - "loss": 0.2388, - "step": 30000 - }, - { - "epoch": 0.35, - "learning_rate": 1.9764625078619083e-05, - "loss": 0.219, - "step": 30500 - }, - { - "epoch": 0.36, - "learning_rate": 1.9760766473350545e-05, - "loss": 0.2357, - "step": 31000 - }, - { - "epoch": 0.36, - "learning_rate": 1.9756907868082004e-05, - "loss": 0.227, - "step": 31500 - }, - { - "epoch": 0.37, - "learning_rate": 1.9753049262813466e-05, - "loss": 0.2228, - "step": 32000 - }, - { - "epoch": 0.38, - "learning_rate": 1.9749190657544924e-05, - "loss": 0.2368, - "step": 32500 - }, - { - "epoch": 0.38, - "learning_rate": 1.9745332052276386e-05, - "loss": 0.2261, - "step": 33000 - }, - { - "epoch": 0.39, - "learning_rate": 1.9741473447007845e-05, - "loss": 0.2299, - "step": 33500 - }, - { - "epoch": 0.39, - "learning_rate": 1.9737614841739307e-05, - "loss": 0.2293, - "step": 34000 - }, - { - "epoch": 0.4, - "learning_rate": 1.9733756236470766e-05, - "loss": 0.2346, - "step": 34500 - }, - { - "epoch": 0.41, - "learning_rate": 1.9729897631202228e-05, - "loss": 0.2277, - "step": 35000 - }, - { - "epoch": 0.41, - "learning_rate": 1.9726039025933686e-05, - "loss": 0.2304, - "step": 35500 - }, - { - "epoch": 0.42, - "learning_rate": 1.9722180420665148e-05, - "loss": 0.2336, - "step": 36000 - }, - { - "epoch": 0.42, - "learning_rate": 1.9718321815396607e-05, - "loss": 0.243, - "step": 36500 - }, - { - "epoch": 0.43, - "learning_rate": 1.971446321012807e-05, - "loss": 0.2327, - "step": 37000 - }, - { - "epoch": 0.43, - "learning_rate": 1.9710604604859528e-05, - "loss": 0.2234, - "step": 37500 - }, - { - "epoch": 0.44, - "learning_rate": 1.970674599959099e-05, - "loss": 0.2265, - "step": 38000 - }, - { - "epoch": 0.45, - "learning_rate": 1.9702887394322448e-05, - "loss": 0.2273, - "step": 38500 - }, - { - "epoch": 0.45, - "learning_rate": 1.969902878905391e-05, - "loss": 0.2335, - "step": 39000 - }, - { - "epoch": 0.46, - "learning_rate": 1.969517018378537e-05, - "loss": 0.2294, - "step": 39500 - }, - { - "epoch": 0.46, - "learning_rate": 1.969131157851683e-05, - "loss": 0.2265, - "step": 40000 - }, - { - "epoch": 0.47, - "learning_rate": 1.968745297324829e-05, - "loss": 0.2246, - "step": 40500 - }, - { - "epoch": 0.47, - "learning_rate": 1.968359436797975e-05, - "loss": 0.2334, - "step": 41000 - }, - { - "epoch": 0.48, - "learning_rate": 1.967973576271121e-05, - "loss": 0.2305, - "step": 41500 - }, - { - "epoch": 0.49, - "learning_rate": 1.9675877157442672e-05, - "loss": 0.2199, - "step": 42000 - }, - { - "epoch": 0.49, - "learning_rate": 1.9672018552174134e-05, - "loss": 0.2193, - "step": 42500 - }, - { - "epoch": 0.5, - "learning_rate": 1.9668159946905593e-05, - "loss": 0.233, - "step": 43000 - }, - { - "epoch": 0.5, - "learning_rate": 1.9664301341637055e-05, - "loss": 0.227, - "step": 43500 - }, - { - "epoch": 0.51, - "learning_rate": 1.9660442736368513e-05, - "loss": 0.2221, - "step": 44000 - }, - { - "epoch": 0.52, - "learning_rate": 1.9656584131099975e-05, - "loss": 0.2312, - "step": 44500 - }, - { - "epoch": 0.52, - "learning_rate": 1.9652725525831434e-05, - "loss": 0.2187, - "step": 45000 - }, - { - "epoch": 0.53, - "learning_rate": 1.9648866920562896e-05, - "loss": 0.2156, - "step": 45500 - }, - { - "epoch": 0.53, - "learning_rate": 1.9645008315294355e-05, - "loss": 0.2227, - "step": 46000 - }, - { - "epoch": 0.54, - "learning_rate": 1.9641149710025817e-05, - "loss": 0.2205, - "step": 46500 - }, - { - "epoch": 0.54, - "learning_rate": 1.9637291104757275e-05, - "loss": 0.2324, - "step": 47000 - }, - { - "epoch": 0.55, - "learning_rate": 1.9633432499488737e-05, - "loss": 0.2159, - "step": 47500 - }, - { - "epoch": 0.56, - "learning_rate": 1.9629573894220196e-05, - "loss": 0.225, - "step": 48000 - }, - { - "epoch": 0.56, - "learning_rate": 1.9625715288951658e-05, - "loss": 0.2124, - "step": 48500 - }, - { - "epoch": 0.57, - "learning_rate": 1.9621856683683117e-05, - "loss": 0.231, - "step": 49000 - }, - { - "epoch": 0.57, - "learning_rate": 1.961799807841458e-05, - "loss": 0.2264, - "step": 49500 - }, - { - "epoch": 0.58, - "learning_rate": 1.9614139473146037e-05, - "loss": 0.2243, - "step": 50000 - }, - { - "epoch": 0.58, - "learning_rate": 1.96102808678775e-05, - "loss": 0.2271, - "step": 50500 - }, - { - "epoch": 0.59, - "learning_rate": 1.9606422262608958e-05, - "loss": 0.2227, - "step": 51000 - }, - { - "epoch": 0.6, - "learning_rate": 1.960256365734042e-05, - "loss": 0.234, - "step": 51500 - }, - { - "epoch": 0.6, - "learning_rate": 1.959870505207188e-05, - "loss": 0.2233, - "step": 52000 - }, - { - "epoch": 0.61, - "learning_rate": 1.959484644680334e-05, - "loss": 0.2253, - "step": 52500 - }, - { - "epoch": 0.61, - "learning_rate": 1.9590987841534803e-05, - "loss": 0.2229, - "step": 53000 - }, - { - "epoch": 0.62, - "learning_rate": 1.958712923626626e-05, - "loss": 0.2359, - "step": 53500 - }, - { - "epoch": 0.63, - "learning_rate": 1.9583270630997723e-05, - "loss": 0.2232, - "step": 54000 - }, - { - "epoch": 0.63, - "learning_rate": 1.9579412025729182e-05, - "loss": 0.2238, - "step": 54500 - }, - { - "epoch": 0.64, - "learning_rate": 1.9575553420460644e-05, - "loss": 0.2239, - "step": 55000 - }, - { - "epoch": 0.64, - "learning_rate": 1.9571694815192103e-05, - "loss": 0.2401, - "step": 55500 - }, - { - "epoch": 0.65, - "learning_rate": 1.9567836209923565e-05, - "loss": 0.2299, - "step": 56000 - }, - { - "epoch": 0.65, - "learning_rate": 1.9563977604655023e-05, - "loss": 0.223, - "step": 56500 - }, - { - "epoch": 0.66, - "learning_rate": 1.9560118999386485e-05, - "loss": 0.2191, - "step": 57000 - }, - { - "epoch": 0.67, - "learning_rate": 1.9556260394117944e-05, - "loss": 0.2291, - "step": 57500 - }, - { - "epoch": 0.67, - "learning_rate": 1.9552401788849406e-05, - "loss": 0.2176, - "step": 58000 - }, - { - "epoch": 0.68, - "learning_rate": 1.9548543183580864e-05, - "loss": 0.2133, - "step": 58500 - }, - { - "epoch": 0.68, - "learning_rate": 1.9544684578312326e-05, - "loss": 0.218, - "step": 59000 - }, - { - "epoch": 0.69, - "learning_rate": 1.9540825973043785e-05, - "loss": 0.2279, - "step": 59500 - }, - { - "epoch": 0.69, - "learning_rate": 1.9536967367775247e-05, - "loss": 0.2174, - "step": 60000 - }, - { - "epoch": 0.7, - "learning_rate": 1.9533108762506706e-05, - "loss": 0.2236, - "step": 60500 - }, - { - "epoch": 0.71, - "learning_rate": 1.9529250157238168e-05, - "loss": 0.2231, - "step": 61000 - }, - { - "epoch": 0.71, - "learning_rate": 1.9525391551969626e-05, - "loss": 0.2193, - "step": 61500 - }, - { - "epoch": 0.72, - "learning_rate": 1.952153294670109e-05, - "loss": 0.213, - "step": 62000 - }, - { - "epoch": 0.72, - "learning_rate": 1.9517674341432547e-05, - "loss": 0.2459, - "step": 62500 - }, - { - "epoch": 0.73, - "learning_rate": 1.951381573616401e-05, - "loss": 0.2233, - "step": 63000 - }, - { - "epoch": 0.74, - "learning_rate": 1.9509957130895468e-05, - "loss": 0.2391, - "step": 63500 - }, - { - "epoch": 0.74, - "learning_rate": 1.950609852562693e-05, - "loss": 0.2248, - "step": 64000 - }, - { - "epoch": 0.75, - "learning_rate": 1.950223992035839e-05, - "loss": 0.225, - "step": 64500 - }, - { - "epoch": 0.75, - "learning_rate": 1.949838131508985e-05, - "loss": 0.2222, - "step": 65000 - }, - { - "epoch": 0.76, - "learning_rate": 1.949452270982131e-05, - "loss": 0.2346, - "step": 65500 - }, - { - "epoch": 0.76, - "learning_rate": 1.949066410455277e-05, - "loss": 0.2231, - "step": 66000 - }, - { - "epoch": 0.77, - "learning_rate": 1.948680549928423e-05, - "loss": 0.2254, - "step": 66500 - }, - { - "epoch": 0.78, - "learning_rate": 1.948294689401569e-05, - "loss": 0.228, - "step": 67000 - }, - { - "epoch": 0.78, - "learning_rate": 1.947908828874715e-05, - "loss": 0.2155, - "step": 67500 - }, - { - "epoch": 0.79, - "learning_rate": 1.9475229683478612e-05, - "loss": 0.228, - "step": 68000 - }, - { - "epoch": 0.79, - "learning_rate": 1.9471371078210074e-05, - "loss": 0.2197, - "step": 68500 - }, - { - "epoch": 0.8, - "learning_rate": 1.9467512472941533e-05, - "loss": 0.2074, - "step": 69000 - }, - { - "epoch": 0.8, - "learning_rate": 1.9463653867672995e-05, - "loss": 0.2183, - "step": 69500 - }, - { - "epoch": 0.81, - "learning_rate": 1.9459795262404454e-05, - "loss": 0.2262, - "step": 70000 - }, - { - "epoch": 0.82, - "learning_rate": 1.9455936657135916e-05, - "loss": 0.2234, - "step": 70500 - }, - { - "epoch": 0.82, - "learning_rate": 1.9452078051867374e-05, - "loss": 0.2309, - "step": 71000 - }, - { - "epoch": 0.83, - "learning_rate": 1.9448219446598836e-05, - "loss": 0.2221, - "step": 71500 - }, - { - "epoch": 0.83, - "learning_rate": 1.9444360841330295e-05, - "loss": 0.2099, - "step": 72000 - }, - { - "epoch": 0.84, - "learning_rate": 1.9440502236061757e-05, - "loss": 0.224, - "step": 72500 - }, - { - "epoch": 0.85, - "learning_rate": 1.9436643630793215e-05, - "loss": 0.2205, - "step": 73000 - }, - { - "epoch": 0.85, - "learning_rate": 1.9432785025524678e-05, - "loss": 0.2203, - "step": 73500 - }, - { - "epoch": 0.86, - "learning_rate": 1.9428926420256136e-05, - "loss": 0.231, - "step": 74000 - }, - { - "epoch": 0.86, - "learning_rate": 1.9425067814987598e-05, - "loss": 0.2247, - "step": 74500 - }, - { - "epoch": 0.87, - "learning_rate": 1.9421209209719057e-05, - "loss": 0.2291, - "step": 75000 - }, - { - "epoch": 0.87, - "learning_rate": 1.941735060445052e-05, - "loss": 0.2284, - "step": 75500 - }, - { - "epoch": 0.88, - "learning_rate": 1.9413491999181977e-05, - "loss": 0.2292, - "step": 76000 - }, - { - "epoch": 0.89, - "learning_rate": 1.940963339391344e-05, - "loss": 0.2257, - "step": 76500 - }, - { - "epoch": 0.89, - "learning_rate": 1.9405774788644898e-05, - "loss": 0.2409, - "step": 77000 - }, - { - "epoch": 0.9, - "learning_rate": 1.940191618337636e-05, - "loss": 0.2293, - "step": 77500 - }, - { - "epoch": 0.9, - "learning_rate": 1.939805757810782e-05, - "loss": 0.2214, - "step": 78000 - }, - { - "epoch": 0.91, - "learning_rate": 1.939419897283928e-05, - "loss": 0.2246, - "step": 78500 - }, - { - "epoch": 0.91, - "learning_rate": 1.939034036757074e-05, - "loss": 0.2216, - "step": 79000 - }, - { - "epoch": 0.92, - "learning_rate": 1.93864817623022e-05, - "loss": 0.2226, - "step": 79500 - }, - { - "epoch": 0.93, - "learning_rate": 1.938262315703366e-05, - "loss": 0.2324, - "step": 80000 - }, - { - "epoch": 0.93, - "learning_rate": 1.9378764551765122e-05, - "loss": 0.2224, - "step": 80500 - }, - { - "epoch": 0.94, - "learning_rate": 1.937490594649658e-05, - "loss": 0.2378, - "step": 81000 - }, - { - "epoch": 0.94, - "learning_rate": 1.9371047341228043e-05, - "loss": 0.2169, - "step": 81500 - }, - { - "epoch": 0.95, - "learning_rate": 1.93671887359595e-05, - "loss": 0.2206, - "step": 82000 - }, - { - "epoch": 0.96, - "learning_rate": 1.9363330130690963e-05, - "loss": 0.2191, - "step": 82500 - }, - { - "epoch": 0.96, - "learning_rate": 1.9359471525422422e-05, - "loss": 0.2363, - "step": 83000 - }, - { - "epoch": 0.97, - "learning_rate": 1.9355612920153884e-05, - "loss": 0.2175, - "step": 83500 - }, - { - "epoch": 0.97, - "learning_rate": 1.9351754314885343e-05, - "loss": 0.21, - "step": 84000 - }, - { - "epoch": 0.98, - "learning_rate": 1.9347895709616805e-05, - "loss": 0.2203, - "step": 84500 - }, - { - "epoch": 0.98, - "learning_rate": 1.9344037104348263e-05, - "loss": 0.2166, - "step": 85000 - }, - { - "epoch": 0.99, - "learning_rate": 1.9340178499079725e-05, - "loss": 0.211, - "step": 85500 - }, - { - "epoch": 1.0, - "learning_rate": 1.9336319893811184e-05, - "loss": 0.2254, - "step": 86000 - }, - { - "epoch": 1.0, - "eval_bart_score": -6.78, - "eval_bertscore_f1": 0.75, - "eval_bertscore_precision": 0.79, - "eval_bertscore_recall": 0.72, - "eval_bleu": 0.0, - "eval_loss": 0.21441325545310974, - "eval_meteor": 0.09, - "eval_rogue_score": { - "rouge1": 0.1571033255826546, - "rouge2": 0.024564791577206757, - "rougeL": 0.12796016101205676, - "rougeLsum": 0.12797893703375637 - }, - "eval_runtime": 5950.6237, - "eval_samples_per_second": 3.629, - "eval_simple_accuracy": 0.74, - "eval_steps_per_second": 1.815, - "step": 86387 - }, - { - "epoch": 1.0, - "learning_rate": 1.9332461288542646e-05, - "loss": 0.2202, - "step": 86500 - }, - { - "epoch": 1.01, - "learning_rate": 1.9328602683274105e-05, - "loss": 0.214, - "step": 87000 - }, - { - "epoch": 1.01, - "learning_rate": 1.9324744078005567e-05, - "loss": 0.2221, - "step": 87500 - }, - { - "epoch": 1.02, - "learning_rate": 1.9320885472737025e-05, - "loss": 0.2164, - "step": 88000 - }, - { - "epoch": 1.02, - "learning_rate": 1.9317026867468487e-05, - "loss": 0.2125, - "step": 88500 - }, - { - "epoch": 1.03, - "learning_rate": 1.9313168262199946e-05, - "loss": 0.221, - "step": 89000 - }, - { - "epoch": 1.04, - "learning_rate": 1.9309309656931408e-05, - "loss": 0.1987, - "step": 89500 - }, - { - "epoch": 1.04, - "learning_rate": 1.9305451051662866e-05, - "loss": 0.213, - "step": 90000 - }, - { - "epoch": 1.05, - "learning_rate": 1.930159244639433e-05, - "loss": 0.2199, - "step": 90500 - }, - { - "epoch": 1.05, - "learning_rate": 1.9297733841125787e-05, - "loss": 0.2154, - "step": 91000 - }, - { - "epoch": 1.06, - "learning_rate": 1.929387523585725e-05, - "loss": 0.2205, - "step": 91500 - }, - { - "epoch": 1.06, - "learning_rate": 1.9290016630588708e-05, - "loss": 0.2095, - "step": 92000 - }, - { - "epoch": 1.07, - "learning_rate": 1.928615802532017e-05, - "loss": 0.2091, - "step": 92500 - }, - { - "epoch": 1.08, - "learning_rate": 1.928229942005163e-05, - "loss": 0.2237, - "step": 93000 - }, - { - "epoch": 1.08, - "learning_rate": 1.927844081478309e-05, - "loss": 0.2231, - "step": 93500 - }, - { - "epoch": 1.09, - "learning_rate": 1.927458220951455e-05, - "loss": 0.2099, - "step": 94000 - }, - { - "epoch": 1.09, - "learning_rate": 1.927072360424601e-05, - "loss": 0.2142, - "step": 94500 - }, - { - "epoch": 1.1, - "learning_rate": 1.926686499897747e-05, - "loss": 0.2092, - "step": 95000 - }, - { - "epoch": 1.11, - "learning_rate": 1.926300639370893e-05, - "loss": 0.2152, - "step": 95500 - }, - { - "epoch": 1.11, - "learning_rate": 1.925914778844039e-05, - "loss": 0.2156, - "step": 96000 - }, - { - "epoch": 1.12, - "learning_rate": 1.9255289183171852e-05, - "loss": 0.2097, - "step": 96500 - }, - { - "epoch": 1.12, - "learning_rate": 1.925143057790331e-05, - "loss": 0.2114, - "step": 97000 - }, - { - "epoch": 1.13, - "learning_rate": 1.9247571972634773e-05, - "loss": 0.2254, - "step": 97500 - }, - { - "epoch": 1.13, - "learning_rate": 1.924371336736623e-05, - "loss": 0.2287, - "step": 98000 - }, - { - "epoch": 1.14, - "learning_rate": 1.9239854762097694e-05, - "loss": 0.2145, - "step": 98500 - }, - { - "epoch": 1.15, - "learning_rate": 1.9235996156829152e-05, - "loss": 0.2137, - "step": 99000 - }, - { - "epoch": 1.15, - "learning_rate": 1.9232137551560614e-05, - "loss": 0.213, - "step": 99500 - }, - { - "epoch": 1.16, - "learning_rate": 1.9228278946292073e-05, - "loss": 0.2161, - "step": 100000 - }, - { - "epoch": 1.16, - "learning_rate": 1.9224420341023535e-05, - "loss": 0.2103, - "step": 100500 - }, - { - "epoch": 1.17, - "learning_rate": 1.9220561735754994e-05, - "loss": 0.2161, - "step": 101000 - }, - { - "epoch": 1.17, - "learning_rate": 1.9216703130486456e-05, - "loss": 0.2086, - "step": 101500 - }, - { - "epoch": 1.18, - "learning_rate": 1.9212844525217914e-05, - "loss": 0.2084, - "step": 102000 - }, - { - "epoch": 1.19, - "learning_rate": 1.9208985919949376e-05, - "loss": 0.2172, - "step": 102500 - }, - { - "epoch": 1.19, - "learning_rate": 1.9205127314680838e-05, - "loss": 0.2099, - "step": 103000 - }, - { - "epoch": 1.2, - "learning_rate": 1.9201268709412297e-05, - "loss": 0.2137, - "step": 103500 - }, - { - "epoch": 1.2, - "learning_rate": 1.919741010414376e-05, - "loss": 0.2189, - "step": 104000 - }, - { - "epoch": 1.21, - "learning_rate": 1.9193551498875217e-05, - "loss": 0.2327, - "step": 104500 - }, - { - "epoch": 1.22, - "learning_rate": 1.918969289360668e-05, - "loss": 0.2216, - "step": 105000 - }, - { - "epoch": 1.22, - "learning_rate": 1.9185834288338138e-05, - "loss": 0.2269, - "step": 105500 - }, - { - "epoch": 1.23, - "learning_rate": 1.91819756830696e-05, - "loss": 0.2005, - "step": 106000 - }, - { - "epoch": 1.23, - "learning_rate": 1.917811707780106e-05, - "loss": 0.2256, - "step": 106500 - }, - { - "epoch": 1.24, - "learning_rate": 1.917425847253252e-05, - "loss": 0.216, - "step": 107000 - }, - { - "epoch": 1.24, - "learning_rate": 1.917039986726398e-05, - "loss": 0.2208, - "step": 107500 - }, - { - "epoch": 1.25, - "learning_rate": 1.916654126199544e-05, - "loss": 0.2083, - "step": 108000 - }, - { - "epoch": 1.26, - "learning_rate": 1.91626826567269e-05, - "loss": 0.2169, - "step": 108500 - }, - { - "epoch": 1.26, - "learning_rate": 1.9158824051458362e-05, - "loss": 0.2078, - "step": 109000 - }, - { - "epoch": 1.27, - "learning_rate": 1.915496544618982e-05, - "loss": 0.2169, - "step": 109500 - }, - { - "epoch": 1.27, - "learning_rate": 1.9151106840921283e-05, - "loss": 0.2129, - "step": 110000 - }, - { - "epoch": 1.28, - "learning_rate": 1.914724823565274e-05, - "loss": 0.2267, - "step": 110500 - }, - { - "epoch": 1.28, - "learning_rate": 1.9143389630384203e-05, - "loss": 0.2084, - "step": 111000 - }, - { - "epoch": 1.29, - "learning_rate": 1.9139531025115662e-05, - "loss": 0.2116, - "step": 111500 - }, - { - "epoch": 1.3, - "learning_rate": 1.9135672419847124e-05, - "loss": 0.2181, - "step": 112000 - }, - { - "epoch": 1.3, - "learning_rate": 1.9131813814578583e-05, - "loss": 0.2233, - "step": 112500 - }, - { - "epoch": 1.31, - "learning_rate": 1.9127955209310045e-05, - "loss": 0.2162, - "step": 113000 - }, - { - "epoch": 1.31, - "learning_rate": 1.9124096604041503e-05, - "loss": 0.2271, - "step": 113500 - }, - { - "epoch": 1.32, - "learning_rate": 1.9120237998772965e-05, - "loss": 0.2173, - "step": 114000 - }, - { - "epoch": 1.33, - "learning_rate": 1.9116379393504424e-05, - "loss": 0.2141, - "step": 114500 - }, - { - "epoch": 1.33, - "learning_rate": 1.9112520788235886e-05, - "loss": 0.2227, - "step": 115000 - }, - { - "epoch": 1.34, - "learning_rate": 1.9108662182967345e-05, - "loss": 0.2234, - "step": 115500 - }, - { - "epoch": 1.34, - "learning_rate": 1.9104803577698807e-05, - "loss": 0.248, - "step": 116000 - }, - { - "epoch": 1.35, - "learning_rate": 1.9100944972430265e-05, - "loss": 0.2096, - "step": 116500 - }, - { - "epoch": 1.35, - "learning_rate": 1.9097086367161727e-05, - "loss": 0.2293, - "step": 117000 - }, - { - "epoch": 1.36, - "learning_rate": 1.9093227761893186e-05, - "loss": 0.2123, - "step": 117500 - }, - { - "epoch": 1.37, - "learning_rate": 1.9089369156624648e-05, - "loss": 0.2301, - "step": 118000 - }, - { - "epoch": 1.37, - "learning_rate": 1.9085510551356107e-05, - "loss": 0.2186, - "step": 118500 - }, - { - "epoch": 1.38, - "learning_rate": 1.908165194608757e-05, - "loss": 0.2147, - "step": 119000 - }, - { - "epoch": 1.38, - "learning_rate": 1.9077793340819027e-05, - "loss": 0.2113, - "step": 119500 - }, - { - "epoch": 1.39, - "learning_rate": 1.907393473555049e-05, - "loss": 0.2166, - "step": 120000 - }, - { - "epoch": 1.39, - "learning_rate": 1.9070076130281948e-05, - "loss": 0.2179, - "step": 120500 - }, - { - "epoch": 1.4, - "learning_rate": 1.906621752501341e-05, - "loss": 0.2195, - "step": 121000 - }, - { - "epoch": 1.41, - "learning_rate": 1.906235891974487e-05, - "loss": 0.2145, - "step": 121500 - }, - { - "epoch": 1.41, - "learning_rate": 1.905850031447633e-05, - "loss": 0.2066, - "step": 122000 - }, - { - "epoch": 1.42, - "learning_rate": 1.905464170920779e-05, - "loss": 0.2126, - "step": 122500 - }, - { - "epoch": 1.42, - "learning_rate": 1.905078310393925e-05, - "loss": 0.213, - "step": 123000 - }, - { - "epoch": 1.43, - "learning_rate": 1.904692449867071e-05, - "loss": 0.2131, - "step": 123500 - }, - { - "epoch": 1.44, - "learning_rate": 1.9043065893402172e-05, - "loss": 0.2212, - "step": 124000 - }, - { - "epoch": 1.44, - "learning_rate": 1.903920728813363e-05, - "loss": 0.2128, - "step": 124500 - }, - { - "epoch": 1.45, - "learning_rate": 1.9035348682865092e-05, - "loss": 0.21, - "step": 125000 - }, - { - "epoch": 1.45, - "learning_rate": 1.9031490077596554e-05, - "loss": 0.2242, - "step": 125500 - }, - { - "epoch": 1.46, - "learning_rate": 1.9027631472328013e-05, - "loss": 0.2108, - "step": 126000 - }, - { - "epoch": 1.46, - "learning_rate": 1.9023772867059475e-05, - "loss": 0.2146, - "step": 126500 - }, - { - "epoch": 1.47, - "learning_rate": 1.9019914261790934e-05, - "loss": 0.2113, - "step": 127000 - }, - { - "epoch": 1.48, - "learning_rate": 1.9016055656522396e-05, - "loss": 0.2182, - "step": 127500 - }, - { - "epoch": 1.48, - "learning_rate": 1.9012197051253854e-05, - "loss": 0.2126, - "step": 128000 - }, - { - "epoch": 1.49, - "learning_rate": 1.9008338445985316e-05, - "loss": 0.2153, - "step": 128500 - }, - { - "epoch": 1.49, - "learning_rate": 1.9004479840716775e-05, - "loss": 0.2256, - "step": 129000 - }, - { - "epoch": 1.5, - "learning_rate": 1.9000621235448237e-05, - "loss": 0.2197, - "step": 129500 - }, - { - "epoch": 1.5, - "learning_rate": 1.8996762630179696e-05, - "loss": 0.2269, - "step": 130000 - }, - { - "epoch": 1.51, - "learning_rate": 1.8992904024911158e-05, - "loss": 0.2277, - "step": 130500 - }, - { - "epoch": 1.52, - "learning_rate": 1.8989045419642616e-05, - "loss": 0.2209, - "step": 131000 - }, - { - "epoch": 1.52, - "learning_rate": 1.8985186814374078e-05, - "loss": 0.2263, - "step": 131500 - }, - { - "epoch": 1.53, - "learning_rate": 1.8981328209105537e-05, - "loss": 0.2171, - "step": 132000 - }, - { - "epoch": 1.53, - "learning_rate": 1.8977469603837e-05, - "loss": 0.2275, - "step": 132500 - }, - { - "epoch": 1.54, - "learning_rate": 1.8973610998568458e-05, - "loss": 0.2185, - "step": 133000 - }, - { - "epoch": 1.55, - "learning_rate": 1.896975239329992e-05, - "loss": 0.2281, - "step": 133500 - }, - { - "epoch": 1.55, - "learning_rate": 1.8965893788031378e-05, - "loss": 0.2125, - "step": 134000 - }, - { - "epoch": 1.56, - "learning_rate": 1.896203518276284e-05, - "loss": 0.2184, - "step": 134500 - }, - { - "epoch": 1.56, - "learning_rate": 1.89581765774943e-05, - "loss": 0.2189, - "step": 135000 - }, - { - "epoch": 1.57, - "learning_rate": 1.895431797222576e-05, - "loss": 0.2203, - "step": 135500 - }, - { - "epoch": 1.57, - "learning_rate": 1.8950459366957223e-05, - "loss": 0.2199, - "step": 136000 - }, - { - "epoch": 1.58, - "learning_rate": 1.894660076168868e-05, - "loss": 0.2251, - "step": 136500 - }, - { - "epoch": 1.59, - "learning_rate": 1.8942742156420144e-05, - "loss": 0.2167, - "step": 137000 - }, - { - "epoch": 1.59, - "learning_rate": 1.8938883551151602e-05, - "loss": 0.2149, - "step": 137500 - }, - { - "epoch": 1.6, - "learning_rate": 1.8935024945883064e-05, - "loss": 0.2071, - "step": 138000 - }, - { - "epoch": 1.6, - "learning_rate": 1.8931166340614523e-05, - "loss": 0.2185, - "step": 138500 - }, - { - "epoch": 1.61, - "learning_rate": 1.8927307735345985e-05, - "loss": 0.2177, - "step": 139000 - }, - { - "epoch": 1.61, - "learning_rate": 1.8923449130077443e-05, - "loss": 0.2144, - "step": 139500 - }, - { - "epoch": 1.62, - "learning_rate": 1.8919590524808905e-05, - "loss": 0.2183, - "step": 140000 - }, - { - "epoch": 1.63, - "learning_rate": 1.8915731919540364e-05, - "loss": 0.2252, - "step": 140500 - }, - { - "epoch": 1.63, - "learning_rate": 1.8911873314271826e-05, - "loss": 0.2193, - "step": 141000 - }, - { - "epoch": 1.64, - "learning_rate": 1.8908014709003285e-05, - "loss": 0.2004, - "step": 141500 - }, - { - "epoch": 1.64, - "learning_rate": 1.8904156103734747e-05, - "loss": 0.2087, - "step": 142000 - }, - { - "epoch": 1.65, - "learning_rate": 1.8900297498466205e-05, - "loss": 0.212, - "step": 142500 - }, - { - "epoch": 1.66, - "learning_rate": 1.8896438893197667e-05, - "loss": 0.2102, - "step": 143000 - }, - { - "epoch": 1.66, - "learning_rate": 1.8892580287929126e-05, - "loss": 0.2225, - "step": 143500 - }, - { - "epoch": 1.67, - "learning_rate": 1.8888721682660588e-05, - "loss": 0.2155, - "step": 144000 - }, - { - "epoch": 1.67, - "learning_rate": 1.8884863077392047e-05, - "loss": 0.2233, - "step": 144500 - }, - { - "epoch": 1.68, - "learning_rate": 1.888100447212351e-05, - "loss": 0.2187, - "step": 145000 - }, - { - "epoch": 1.68, - "learning_rate": 1.8877145866854967e-05, - "loss": 0.2213, - "step": 145500 - }, - { - "epoch": 1.69, - "learning_rate": 1.887328726158643e-05, - "loss": 0.2218, - "step": 146000 - }, - { - "epoch": 1.7, - "learning_rate": 1.886942865631789e-05, - "loss": 0.2169, - "step": 146500 - }, - { - "epoch": 1.7, - "learning_rate": 1.886557005104935e-05, - "loss": 0.2116, - "step": 147000 - }, - { - "epoch": 1.71, - "learning_rate": 1.8861711445780812e-05, - "loss": 0.2143, - "step": 147500 - }, - { - "epoch": 1.71, - "learning_rate": 1.885785284051227e-05, - "loss": 0.2143, - "step": 148000 - }, - { - "epoch": 1.72, - "learning_rate": 1.8853994235243733e-05, - "loss": 0.2101, - "step": 148500 - }, - { - "epoch": 1.72, - "learning_rate": 1.885013562997519e-05, - "loss": 0.2144, - "step": 149000 - }, - { - "epoch": 1.73, - "learning_rate": 1.8846277024706653e-05, - "loss": 0.2074, - "step": 149500 - }, - { - "epoch": 1.74, - "learning_rate": 1.8842418419438112e-05, - "loss": 0.2179, - "step": 150000 - }, - { - "epoch": 1.74, - "learning_rate": 1.8838559814169574e-05, - "loss": 0.2129, - "step": 150500 - }, - { - "epoch": 1.75, - "learning_rate": 1.8834701208901033e-05, - "loss": 0.223, - "step": 151000 - }, - { - "epoch": 1.75, - "learning_rate": 1.8830842603632495e-05, - "loss": 0.2211, - "step": 151500 - }, - { - "epoch": 1.76, - "learning_rate": 1.8826983998363953e-05, - "loss": 0.2135, - "step": 152000 - }, - { - "epoch": 1.77, - "learning_rate": 1.8823125393095415e-05, - "loss": 0.2194, - "step": 152500 - }, - { - "epoch": 1.77, - "learning_rate": 1.8819266787826874e-05, - "loss": 0.227, - "step": 153000 - }, - { - "epoch": 1.78, - "learning_rate": 1.8815408182558336e-05, - "loss": 0.2168, - "step": 153500 - }, - { - "epoch": 1.78, - "learning_rate": 1.8811549577289795e-05, - "loss": 0.2179, - "step": 154000 - }, - { - "epoch": 1.79, - "learning_rate": 1.8807690972021257e-05, - "loss": 0.2077, - "step": 154500 - }, - { - "epoch": 1.79, - "learning_rate": 1.8803832366752715e-05, - "loss": 0.2145, - "step": 155000 - }, - { - "epoch": 1.8, - "learning_rate": 1.8799973761484177e-05, - "loss": 0.215, - "step": 155500 - }, - { - "epoch": 1.81, - "learning_rate": 1.8796115156215636e-05, - "loss": 0.2131, - "step": 156000 - }, - { - "epoch": 1.81, - "learning_rate": 1.8792256550947098e-05, - "loss": 0.2105, - "step": 156500 - }, - { - "epoch": 1.82, - "learning_rate": 1.8788397945678556e-05, - "loss": 0.2007, - "step": 157000 - }, - { - "epoch": 1.82, - "learning_rate": 1.878453934041002e-05, - "loss": 0.2126, - "step": 157500 - }, - { - "epoch": 1.83, - "learning_rate": 1.8780680735141477e-05, - "loss": 0.2001, - "step": 158000 - }, - { - "epoch": 1.83, - "learning_rate": 1.877682212987294e-05, - "loss": 0.216, - "step": 158500 - }, - { - "epoch": 1.84, - "learning_rate": 1.8772963524604398e-05, - "loss": 0.2141, - "step": 159000 - }, - { - "epoch": 1.85, - "learning_rate": 1.876910491933586e-05, - "loss": 0.2238, - "step": 159500 - }, - { - "epoch": 1.85, - "learning_rate": 1.876524631406732e-05, - "loss": 0.2206, - "step": 160000 - }, - { - "epoch": 1.86, - "learning_rate": 1.876138770879878e-05, - "loss": 0.2142, - "step": 160500 - }, - { - "epoch": 1.86, - "learning_rate": 1.875752910353024e-05, - "loss": 0.2031, - "step": 161000 - }, - { - "epoch": 1.87, - "learning_rate": 1.87536704982617e-05, - "loss": 0.2244, - "step": 161500 - }, - { - "epoch": 1.88, - "learning_rate": 1.874981189299316e-05, - "loss": 0.2138, - "step": 162000 - }, - { - "epoch": 1.88, - "learning_rate": 1.874595328772462e-05, - "loss": 0.2192, - "step": 162500 - }, - { - "epoch": 1.89, - "learning_rate": 1.874209468245608e-05, - "loss": 0.2242, - "step": 163000 - }, - { - "epoch": 1.89, - "learning_rate": 1.8738236077187542e-05, - "loss": 0.2166, - "step": 163500 - }, - { - "epoch": 1.9, - "learning_rate": 1.8734377471919e-05, - "loss": 0.2148, - "step": 164000 - }, - { - "epoch": 1.9, - "learning_rate": 1.8730518866650463e-05, - "loss": 0.2141, - "step": 164500 - }, - { - "epoch": 1.91, - "learning_rate": 1.872666026138192e-05, - "loss": 0.2236, - "step": 165000 - }, - { - "epoch": 1.92, - "learning_rate": 1.8722801656113384e-05, - "loss": 0.2218, - "step": 165500 - }, - { - "epoch": 1.92, - "learning_rate": 1.8718943050844842e-05, - "loss": 0.2188, - "step": 166000 - }, - { - "epoch": 1.93, - "learning_rate": 1.8715084445576304e-05, - "loss": 0.2179, - "step": 166500 - }, - { - "epoch": 1.93, - "learning_rate": 1.8711225840307763e-05, - "loss": 0.2305, - "step": 167000 - }, - { - "epoch": 1.94, - "learning_rate": 1.8707367235039225e-05, - "loss": 0.2174, - "step": 167500 - }, - { - "epoch": 1.94, - "learning_rate": 1.8703508629770684e-05, - "loss": 0.2067, - "step": 168000 - }, - { - "epoch": 1.95, - "learning_rate": 1.8699650024502146e-05, - "loss": 0.2163, - "step": 168500 - }, - { - "epoch": 1.96, - "learning_rate": 1.8695791419233604e-05, - "loss": 0.2188, - "step": 169000 - }, - { - "epoch": 1.96, - "learning_rate": 1.8691932813965066e-05, - "loss": 0.2166, - "step": 169500 - }, - { - "epoch": 1.97, - "learning_rate": 1.8688074208696525e-05, - "loss": 0.2215, - "step": 170000 - }, - { - "epoch": 1.97, - "learning_rate": 1.8684215603427987e-05, - "loss": 0.2269, - "step": 170500 - }, - { - "epoch": 1.98, - "learning_rate": 1.868035699815945e-05, - "loss": 0.2182, - "step": 171000 - }, - { - "epoch": 1.99, - "learning_rate": 1.8676498392890907e-05, - "loss": 0.2131, - "step": 171500 - }, - { - "epoch": 1.99, - "learning_rate": 1.867263978762237e-05, - "loss": 0.2125, - "step": 172000 - }, - { - "epoch": 2.0, - "learning_rate": 1.8668781182353828e-05, - "loss": 0.2141, - "step": 172500 - }, - { - "epoch": 2.0, - "eval_bart_score": -6.65, - "eval_bertscore_f1": 0.75, - "eval_bertscore_precision": 0.79, - "eval_bertscore_recall": 0.72, - "eval_bleu": 0.0, - "eval_loss": 0.21217894554138184, - "eval_meteor": 0.09, - "eval_rogue_score": { - "rouge1": 0.15388393304626252, - "rouge2": 0.024424609903622555, - "rougeL": 0.1256245521025564, - "rougeLsum": 0.125613877504259 - }, - "eval_runtime": 4279.9159, - "eval_samples_per_second": 5.046, - "eval_simple_accuracy": 0.72, - "eval_steps_per_second": 2.523, - "step": 172774 - }, - { - "epoch": 2.0, - "learning_rate": 1.866492257708529e-05, - "loss": 0.2107, - "step": 173000 - }, - { - "epoch": 2.01, - "learning_rate": 1.866106397181675e-05, - "loss": 0.2049, - "step": 173500 - }, - { - "epoch": 2.01, - "learning_rate": 1.865720536654821e-05, - "loss": 0.2227, - "step": 174000 - }, - { - "epoch": 2.02, - "learning_rate": 1.865334676127967e-05, - "loss": 0.2124, - "step": 174500 - }, - { - "epoch": 2.03, - "learning_rate": 1.864948815601113e-05, - "loss": 0.209, - "step": 175000 - }, - { - "epoch": 2.03, - "learning_rate": 1.864562955074259e-05, - "loss": 0.2163, - "step": 175500 - }, - { - "epoch": 2.04, - "learning_rate": 1.8641770945474052e-05, - "loss": 0.2165, - "step": 176000 - }, - { - "epoch": 2.04, - "learning_rate": 1.863791234020551e-05, - "loss": 0.2116, - "step": 176500 - }, - { - "epoch": 2.05, - "learning_rate": 1.8634053734936973e-05, - "loss": 0.2007, - "step": 177000 - }, - { - "epoch": 2.05, - "learning_rate": 1.863019512966843e-05, - "loss": 0.202, - "step": 177500 - }, - { - "epoch": 2.06, - "learning_rate": 1.8626336524399893e-05, - "loss": 0.2164, - "step": 178000 - }, - { - "epoch": 2.07, - "learning_rate": 1.8622477919131352e-05, - "loss": 0.2102, - "step": 178500 - }, - { - "epoch": 2.07, - "learning_rate": 1.8618619313862814e-05, - "loss": 0.208, - "step": 179000 - }, - { - "epoch": 2.08, - "learning_rate": 1.8614760708594273e-05, - "loss": 0.2062, - "step": 179500 - }, - { - "epoch": 2.08, - "learning_rate": 1.8610902103325735e-05, - "loss": 0.2099, - "step": 180000 - }, - { - "epoch": 2.09, - "learning_rate": 1.8607043498057193e-05, - "loss": 0.2065, - "step": 180500 - }, - { - "epoch": 2.1, - "learning_rate": 1.8603184892788655e-05, - "loss": 0.211, - "step": 181000 - }, - { - "epoch": 2.1, - "learning_rate": 1.8599326287520114e-05, - "loss": 0.206, - "step": 181500 - }, - { - "epoch": 2.11, - "learning_rate": 1.8595467682251576e-05, - "loss": 0.2096, - "step": 182000 - }, - { - "epoch": 2.11, - "learning_rate": 1.8591609076983035e-05, - "loss": 0.2172, - "step": 182500 - }, - { - "epoch": 2.12, - "learning_rate": 1.8587750471714497e-05, - "loss": 0.2085, - "step": 183000 - }, - { - "epoch": 2.12, - "learning_rate": 1.8583891866445955e-05, - "loss": 0.2056, - "step": 183500 - }, - { - "epoch": 2.13, - "learning_rate": 1.8580033261177417e-05, - "loss": 0.2204, - "step": 184000 - }, - { - "epoch": 2.14, - "learning_rate": 1.8576174655908876e-05, - "loss": 0.2164, - "step": 184500 - }, - { - "epoch": 2.14, - "learning_rate": 1.8572316050640338e-05, - "loss": 0.2, - "step": 185000 - }, - { - "epoch": 2.15, - "learning_rate": 1.8568457445371797e-05, - "loss": 0.2051, - "step": 185500 - }, - { - "epoch": 2.15, - "learning_rate": 1.856459884010326e-05, - "loss": 0.2133, - "step": 186000 - }, - { - "epoch": 2.16, - "learning_rate": 1.8560740234834717e-05, - "loss": 0.2145, - "step": 186500 - }, - { - "epoch": 2.16, - "learning_rate": 1.855688162956618e-05, - "loss": 0.2151, - "step": 187000 - }, - { - "epoch": 2.17, - "learning_rate": 1.8553023024297638e-05, - "loss": 0.2026, - "step": 187500 - }, - { - "epoch": 2.18, - "learning_rate": 1.85491644190291e-05, - "loss": 0.201, - "step": 188000 - }, - { - "epoch": 2.18, - "learning_rate": 1.854530581376056e-05, - "loss": 0.1968, - "step": 188500 - }, - { - "epoch": 2.19, - "learning_rate": 1.854144720849202e-05, - "loss": 0.2079, - "step": 189000 - }, - { - "epoch": 2.19, - "learning_rate": 1.853758860322348e-05, - "loss": 0.2186, - "step": 189500 - }, - { - "epoch": 2.2, - "learning_rate": 1.853372999795494e-05, - "loss": 0.2127, - "step": 190000 - }, - { - "epoch": 2.21, - "learning_rate": 1.85298713926864e-05, - "loss": 0.2043, - "step": 190500 - }, - { - "epoch": 2.21, - "learning_rate": 1.8526012787417862e-05, - "loss": 0.2161, - "step": 191000 - }, - { - "epoch": 2.22, - "learning_rate": 1.852215418214932e-05, - "loss": 0.2066, - "step": 191500 - }, - { - "epoch": 2.22, - "learning_rate": 1.8518295576880782e-05, - "loss": 0.2058, - "step": 192000 - }, - { - "epoch": 2.23, - "learning_rate": 1.851443697161224e-05, - "loss": 0.215, - "step": 192500 - }, - { - "epoch": 2.23, - "learning_rate": 1.8510578366343703e-05, - "loss": 0.2035, - "step": 193000 - }, - { - "epoch": 2.24, - "learning_rate": 1.850671976107516e-05, - "loss": 0.2185, - "step": 193500 - }, - { - "epoch": 2.25, - "learning_rate": 1.8502861155806624e-05, - "loss": 0.2193, - "step": 194000 - }, - { - "epoch": 2.25, - "learning_rate": 1.8499002550538082e-05, - "loss": 0.2254, - "step": 194500 - }, - { - "epoch": 2.26, - "learning_rate": 1.8495143945269544e-05, - "loss": 0.1887, - "step": 195000 - }, - { - "epoch": 2.26, - "learning_rate": 1.8491285340001003e-05, - "loss": 0.215, - "step": 195500 - }, - { - "epoch": 2.27, - "learning_rate": 1.8487426734732465e-05, - "loss": 0.2042, - "step": 196000 - }, - { - "epoch": 2.27, - "learning_rate": 1.8483568129463924e-05, - "loss": 0.2144, - "step": 196500 - }, - { - "epoch": 2.28, - "learning_rate": 1.8479709524195386e-05, - "loss": 0.2181, - "step": 197000 - }, - { - "epoch": 2.29, - "learning_rate": 1.8475850918926844e-05, - "loss": 0.207, - "step": 197500 - }, - { - "epoch": 2.29, - "learning_rate": 1.8471992313658306e-05, - "loss": 0.2116, - "step": 198000 - }, - { - "epoch": 2.3, - "learning_rate": 1.8468133708389765e-05, - "loss": 0.2043, - "step": 198500 - }, - { - "epoch": 2.3, - "learning_rate": 1.8464275103121227e-05, - "loss": 0.2182, - "step": 199000 - }, - { - "epoch": 2.31, - "learning_rate": 1.8460416497852686e-05, - "loss": 0.2135, - "step": 199500 - }, - { - "epoch": 2.32, - "learning_rate": 1.8456557892584148e-05, - "loss": 0.2106, - "step": 200000 - }, - { - "epoch": 2.32, - "learning_rate": 1.8452699287315606e-05, - "loss": 0.2089, - "step": 200500 - }, - { - "epoch": 2.33, - "learning_rate": 1.8448840682047068e-05, - "loss": 0.1977, - "step": 201000 - }, - { - "epoch": 2.33, - "learning_rate": 1.8444982076778527e-05, - "loss": 0.214, - "step": 201500 - }, - { - "epoch": 2.34, - "learning_rate": 1.844112347150999e-05, - "loss": 0.2093, - "step": 202000 - }, - { - "epoch": 2.34, - "learning_rate": 1.8437264866241447e-05, - "loss": 0.2065, - "step": 202500 - }, - { - "epoch": 2.35, - "learning_rate": 1.843340626097291e-05, - "loss": 0.214, - "step": 203000 - }, - { - "epoch": 2.36, - "learning_rate": 1.8429547655704368e-05, - "loss": 0.1995, - "step": 203500 - }, - { - "epoch": 2.36, - "learning_rate": 1.842568905043583e-05, - "loss": 0.2057, - "step": 204000 - }, - { - "epoch": 2.37, - "learning_rate": 1.842183044516729e-05, - "loss": 0.2043, - "step": 204500 - }, - { - "epoch": 2.37, - "learning_rate": 1.841797183989875e-05, - "loss": 0.212, - "step": 205000 - }, - { - "epoch": 2.38, - "learning_rate": 1.8414113234630213e-05, - "loss": 0.2144, - "step": 205500 - }, - { - "epoch": 2.38, - "learning_rate": 1.841025462936167e-05, - "loss": 0.2184, - "step": 206000 - }, - { - "epoch": 2.39, - "learning_rate": 1.8406396024093133e-05, - "loss": 0.2056, - "step": 206500 - }, - { - "epoch": 2.4, - "learning_rate": 1.8402537418824592e-05, - "loss": 0.2134, - "step": 207000 - }, - { - "epoch": 2.4, - "learning_rate": 1.8398678813556054e-05, - "loss": 0.2065, - "step": 207500 - }, - { - "epoch": 2.41, - "learning_rate": 1.8394820208287513e-05, - "loss": 0.2027, - "step": 208000 - }, - { - "epoch": 2.41, - "learning_rate": 1.8390961603018975e-05, - "loss": 0.2082, - "step": 208500 - }, - { - "epoch": 2.42, - "learning_rate": 1.8387102997750433e-05, - "loss": 0.1966, - "step": 209000 - }, - { - "epoch": 2.43, - "learning_rate": 1.8383244392481895e-05, - "loss": 0.1995, - "step": 209500 - }, - { - "epoch": 2.43, - "learning_rate": 1.8379385787213354e-05, - "loss": 0.2144, - "step": 210000 - }, - { - "epoch": 2.44, - "learning_rate": 1.8375527181944816e-05, - "loss": 0.2134, - "step": 210500 - }, - { - "epoch": 2.44, - "learning_rate": 1.8371668576676275e-05, - "loss": 0.2119, - "step": 211000 - }, - { - "epoch": 2.45, - "learning_rate": 1.8367809971407737e-05, - "loss": 0.2095, - "step": 211500 - }, - { - "epoch": 2.45, - "learning_rate": 1.8363951366139195e-05, - "loss": 0.2183, - "step": 212000 - }, - { - "epoch": 2.46, - "learning_rate": 1.8360092760870657e-05, - "loss": 0.2161, - "step": 212500 - }, - { - "epoch": 2.47, - "learning_rate": 1.8356234155602116e-05, - "loss": 0.222, - "step": 213000 - }, - { - "epoch": 2.47, - "learning_rate": 1.8352375550333578e-05, - "loss": 0.217, - "step": 213500 - }, - { - "epoch": 2.48, - "learning_rate": 1.8348516945065037e-05, - "loss": 0.2002, - "step": 214000 - }, - { - "epoch": 2.48, - "learning_rate": 1.83446583397965e-05, - "loss": 0.2039, - "step": 214500 - }, - { - "epoch": 2.49, - "learning_rate": 1.8340799734527957e-05, - "loss": 0.1983, - "step": 215000 - }, - { - "epoch": 2.49, - "learning_rate": 1.833694112925942e-05, - "loss": 0.2086, - "step": 215500 - }, - { - "epoch": 2.5, - "learning_rate": 1.8333082523990878e-05, - "loss": 0.2134, - "step": 216000 - }, - { - "epoch": 2.51, - "learning_rate": 1.832922391872234e-05, - "loss": 0.2086, - "step": 216500 - }, - { - "epoch": 2.51, - "learning_rate": 1.83253653134538e-05, - "loss": 0.2135, - "step": 217000 - }, - { - "epoch": 2.52, - "learning_rate": 1.832150670818526e-05, - "loss": 0.2074, - "step": 217500 - }, - { - "epoch": 2.52, - "learning_rate": 1.831764810291672e-05, - "loss": 0.2122, - "step": 218000 - }, - { - "epoch": 2.53, - "learning_rate": 1.831378949764818e-05, - "loss": 0.2052, - "step": 218500 - }, - { - "epoch": 2.54, - "learning_rate": 1.830993089237964e-05, - "loss": 0.2106, - "step": 219000 - }, - { - "epoch": 2.54, - "learning_rate": 1.8306072287111102e-05, - "loss": 0.2095, - "step": 219500 - }, - { - "epoch": 2.55, - "learning_rate": 1.8302213681842564e-05, - "loss": 0.2093, - "step": 220000 - }, - { - "epoch": 2.55, - "learning_rate": 1.8298355076574022e-05, - "loss": 0.2108, - "step": 220500 - }, - { - "epoch": 2.56, - "learning_rate": 1.8294496471305484e-05, - "loss": 0.2116, - "step": 221000 - }, - { - "epoch": 2.56, - "learning_rate": 1.8290637866036943e-05, - "loss": 0.2107, - "step": 221500 - }, - { - "epoch": 2.57, - "learning_rate": 1.8286779260768405e-05, - "loss": 0.205, - "step": 222000 - }, - { - "epoch": 2.58, - "learning_rate": 1.8282920655499864e-05, - "loss": 0.2249, - "step": 222500 - }, - { - "epoch": 2.58, - "learning_rate": 1.8279062050231326e-05, - "loss": 0.2167, - "step": 223000 - }, - { - "epoch": 2.59, - "learning_rate": 1.8275203444962784e-05, - "loss": 0.2096, - "step": 223500 - }, - { - "epoch": 2.59, - "learning_rate": 1.8271344839694246e-05, - "loss": 0.2097, - "step": 224000 - }, - { - "epoch": 2.6, - "learning_rate": 1.8267486234425705e-05, - "loss": 0.2161, - "step": 224500 - }, - { - "epoch": 2.6, - "learning_rate": 1.8263627629157167e-05, - "loss": 0.2248, - "step": 225000 - }, - { - "epoch": 2.61, - "learning_rate": 1.8259769023888626e-05, - "loss": 0.2203, - "step": 225500 - }, - { - "epoch": 2.62, - "learning_rate": 1.8255910418620088e-05, - "loss": 0.2082, - "step": 226000 - }, - { - "epoch": 2.62, - "learning_rate": 1.8252051813351546e-05, - "loss": 0.214, - "step": 226500 - }, - { - "epoch": 2.63, - "learning_rate": 1.824819320808301e-05, - "loss": 0.2061, - "step": 227000 - }, - { - "epoch": 2.63, - "learning_rate": 1.8244334602814467e-05, - "loss": 0.2137, - "step": 227500 - }, - { - "epoch": 2.64, - "learning_rate": 1.824047599754593e-05, - "loss": 0.2112, - "step": 228000 - }, - { - "epoch": 2.65, - "learning_rate": 1.8236617392277388e-05, - "loss": 0.2074, - "step": 228500 - }, - { - "epoch": 2.65, - "learning_rate": 1.823275878700885e-05, - "loss": 0.2163, - "step": 229000 - }, - { - "epoch": 2.66, - "learning_rate": 1.8228900181740308e-05, - "loss": 0.2022, - "step": 229500 - }, - { - "epoch": 2.66, - "learning_rate": 1.822504157647177e-05, - "loss": 0.2119, - "step": 230000 - }, - { - "epoch": 2.67, - "learning_rate": 1.8221182971203232e-05, - "loss": 0.2286, - "step": 230500 - }, - { - "epoch": 2.67, - "learning_rate": 1.821732436593469e-05, - "loss": 0.2184, - "step": 231000 - }, - { - "epoch": 2.68, - "learning_rate": 1.8213465760666153e-05, - "loss": 0.2022, - "step": 231500 - }, - { - "epoch": 2.69, - "learning_rate": 1.820960715539761e-05, - "loss": 0.2126, - "step": 232000 - }, - { - "epoch": 2.69, - "learning_rate": 1.8205748550129074e-05, - "loss": 0.2115, - "step": 232500 - }, - { - "epoch": 2.7, - "learning_rate": 1.8201889944860532e-05, - "loss": 0.2203, - "step": 233000 - }, - { - "epoch": 2.7, - "learning_rate": 1.8198031339591994e-05, - "loss": 0.2111, - "step": 233500 - }, - { - "epoch": 2.71, - "learning_rate": 1.8194172734323453e-05, - "loss": 0.2151, - "step": 234000 - }, - { - "epoch": 2.71, - "learning_rate": 1.8190314129054915e-05, - "loss": 0.217, - "step": 234500 - }, - { - "epoch": 2.72, - "learning_rate": 1.8186455523786374e-05, - "loss": 0.2146, - "step": 235000 - }, - { - "epoch": 2.73, - "learning_rate": 1.8182596918517836e-05, - "loss": 0.2045, - "step": 235500 - }, - { - "epoch": 2.73, - "learning_rate": 1.8178738313249294e-05, - "loss": 0.2103, - "step": 236000 - }, - { - "epoch": 2.74, - "learning_rate": 1.8174879707980756e-05, - "loss": 0.2011, - "step": 236500 - }, - { - "epoch": 2.74, - "learning_rate": 1.8171021102712215e-05, - "loss": 0.2086, - "step": 237000 - }, - { - "epoch": 2.75, - "learning_rate": 1.8167162497443677e-05, - "loss": 0.215, - "step": 237500 - }, - { - "epoch": 2.76, - "learning_rate": 1.8163303892175135e-05, - "loss": 0.2273, - "step": 238000 - }, - { - "epoch": 2.76, - "learning_rate": 1.8159445286906597e-05, - "loss": 0.2007, - "step": 238500 - }, - { - "epoch": 2.77, - "learning_rate": 1.8155586681638056e-05, - "loss": 0.1992, - "step": 239000 - }, - { - "epoch": 2.77, - "learning_rate": 1.8151728076369518e-05, - "loss": 0.2022, - "step": 239500 - }, - { - "epoch": 2.78, - "learning_rate": 1.8147869471100977e-05, - "loss": 0.2108, - "step": 240000 - }, - { - "epoch": 2.78, - "learning_rate": 1.814401086583244e-05, - "loss": 0.2087, - "step": 240500 - }, - { - "epoch": 2.79, - "learning_rate": 1.81401522605639e-05, - "loss": 0.2147, - "step": 241000 - }, - { - "epoch": 2.8, - "learning_rate": 1.813629365529536e-05, - "loss": 0.2164, - "step": 241500 - }, - { - "epoch": 2.8, - "learning_rate": 1.813243505002682e-05, - "loss": 0.2077, - "step": 242000 - }, - { - "epoch": 2.81, - "learning_rate": 1.812857644475828e-05, - "loss": 0.2195, - "step": 242500 - }, - { - "epoch": 2.81, - "learning_rate": 1.8124717839489742e-05, - "loss": 0.2205, - "step": 243000 - }, - { - "epoch": 2.82, - "learning_rate": 1.81208592342212e-05, - "loss": 0.2114, - "step": 243500 - }, - { - "epoch": 2.82, - "learning_rate": 1.8117000628952663e-05, - "loss": 0.2166, - "step": 244000 - }, - { - "epoch": 2.83, - "learning_rate": 1.811314202368412e-05, - "loss": 0.2018, - "step": 244500 - }, - { - "epoch": 2.84, - "learning_rate": 1.8109283418415583e-05, - "loss": 0.2214, - "step": 245000 - }, - { - "epoch": 2.84, - "learning_rate": 1.8105424813147042e-05, - "loss": 0.2076, - "step": 245500 - }, - { - "epoch": 2.85, - "learning_rate": 1.8101566207878504e-05, - "loss": 0.2237, - "step": 246000 - }, - { - "epoch": 2.85, - "learning_rate": 1.8097707602609963e-05, - "loss": 0.2065, - "step": 246500 - }, - { - "epoch": 2.86, - "learning_rate": 1.8093848997341425e-05, - "loss": 0.2041, - "step": 247000 - }, - { - "epoch": 2.87, - "learning_rate": 1.8089990392072883e-05, - "loss": 0.2188, - "step": 247500 - }, - { - "epoch": 2.87, - "learning_rate": 1.8086131786804345e-05, - "loss": 0.2124, - "step": 248000 - }, - { - "epoch": 2.88, - "learning_rate": 1.8082273181535804e-05, - "loss": 0.2011, - "step": 248500 - }, - { - "epoch": 2.88, - "learning_rate": 1.8078414576267266e-05, - "loss": 0.2155, - "step": 249000 - }, - { - "epoch": 2.89, - "learning_rate": 1.8074555970998725e-05, - "loss": 0.213, - "step": 249500 - }, - { - "epoch": 2.89, - "learning_rate": 1.8070697365730187e-05, - "loss": 0.2052, - "step": 250000 - }, - { - "epoch": 2.9, - "learning_rate": 1.8066838760461645e-05, - "loss": 0.2012, - "step": 250500 - }, - { - "epoch": 2.91, - "learning_rate": 1.8062980155193107e-05, - "loss": 0.2143, - "step": 251000 - }, - { - "epoch": 2.91, - "learning_rate": 1.8059121549924566e-05, - "loss": 0.216, - "step": 251500 - }, - { - "epoch": 2.92, - "learning_rate": 1.8055262944656028e-05, - "loss": 0.2058, - "step": 252000 - }, - { - "epoch": 2.92, - "learning_rate": 1.8051404339387486e-05, - "loss": 0.2036, - "step": 252500 - }, - { - "epoch": 2.93, - "learning_rate": 1.804754573411895e-05, - "loss": 0.207, - "step": 253000 - }, - { - "epoch": 2.93, - "learning_rate": 1.8043687128850407e-05, - "loss": 0.208, - "step": 253500 - }, - { - "epoch": 2.94, - "learning_rate": 1.803982852358187e-05, - "loss": 0.2076, - "step": 254000 - }, - { - "epoch": 2.95, - "learning_rate": 1.8035969918313328e-05, - "loss": 0.2008, - "step": 254500 - }, - { - "epoch": 2.95, - "learning_rate": 1.803211131304479e-05, - "loss": 0.2127, - "step": 255000 - }, - { - "epoch": 2.96, - "learning_rate": 1.802825270777625e-05, - "loss": 0.2069, - "step": 255500 - }, - { - "epoch": 2.96, - "learning_rate": 1.802439410250771e-05, - "loss": 0.2116, - "step": 256000 - }, - { - "epoch": 2.97, - "learning_rate": 1.802053549723917e-05, - "loss": 0.2151, - "step": 256500 - }, - { - "epoch": 2.97, - "learning_rate": 1.801667689197063e-05, - "loss": 0.2058, - "step": 257000 - }, - { - "epoch": 2.98, - "learning_rate": 1.801281828670209e-05, - "loss": 0.2093, - "step": 257500 - }, - { - "epoch": 2.99, - "learning_rate": 1.8008959681433552e-05, - "loss": 0.2013, - "step": 258000 - }, - { - "epoch": 2.99, - "learning_rate": 1.800510107616501e-05, - "loss": 0.2152, - "step": 258500 - }, - { - "epoch": 3.0, - "learning_rate": 1.8001242470896472e-05, - "loss": 0.2219, - "step": 259000 - }, - { - "epoch": 3.0, - "eval_bart_score": -6.71, - "eval_bertscore_f1": 0.75, - "eval_bertscore_precision": 0.78, - "eval_bertscore_recall": 0.71, - "eval_bleu": 0.0, - "eval_loss": 0.21118199825286865, - "eval_meteor": 0.09, - "eval_rogue_score": { - "rouge1": 0.15362210929602754, - "rouge2": 0.02403959138428216, - "rougeL": 0.12578179536151207, - "rougeLsum": 0.1257791367260927 - }, - "eval_runtime": 4356.2183, - "eval_samples_per_second": 4.958, - "eval_simple_accuracy": 0.75, - "eval_steps_per_second": 2.479, - "step": 259161 - }, - { - "epoch": 3.0, - "learning_rate": 1.799738386562793e-05, - "loss": 0.1986, - "step": 259500 - }, - { - "epoch": 3.01, - "learning_rate": 1.7993525260359393e-05, - "loss": 0.2009, - "step": 260000 - }, - { - "epoch": 3.02, - "learning_rate": 1.798966665509085e-05, - "loss": 0.1981, - "step": 260500 - }, - { - "epoch": 3.02, - "learning_rate": 1.7985808049822314e-05, - "loss": 0.2096, - "step": 261000 - }, - { - "epoch": 3.03, - "learning_rate": 1.7981949444553772e-05, - "loss": 0.1995, - "step": 261500 - }, - { - "epoch": 3.03, - "learning_rate": 1.7978090839285234e-05, - "loss": 0.201, - "step": 262000 - }, - { - "epoch": 3.04, - "learning_rate": 1.7974232234016693e-05, - "loss": 0.1993, - "step": 262500 - }, - { - "epoch": 3.04, - "learning_rate": 1.7970373628748155e-05, - "loss": 0.2068, - "step": 263000 - }, - { - "epoch": 3.05, - "learning_rate": 1.7966515023479614e-05, - "loss": 0.2091, - "step": 263500 - }, - { - "epoch": 3.06, - "learning_rate": 1.7962656418211076e-05, - "loss": 0.2002, - "step": 264000 - }, - { - "epoch": 3.06, - "learning_rate": 1.7958797812942534e-05, - "loss": 0.2058, - "step": 264500 - }, - { - "epoch": 3.07, - "learning_rate": 1.7954939207673996e-05, - "loss": 0.1924, - "step": 265000 - }, - { - "epoch": 3.07, - "learning_rate": 1.7951080602405455e-05, - "loss": 0.1967, - "step": 265500 - }, - { - "epoch": 3.08, - "learning_rate": 1.7947221997136917e-05, - "loss": 0.2102, - "step": 266000 - }, - { - "epoch": 3.08, - "learning_rate": 1.7943363391868376e-05, - "loss": 0.1996, - "step": 266500 - }, - { - "epoch": 3.09, - "learning_rate": 1.7939504786599838e-05, - "loss": 0.195, - "step": 267000 - }, - { - "epoch": 3.1, - "learning_rate": 1.7935646181331296e-05, - "loss": 0.1942, - "step": 267500 - }, - { - "epoch": 3.1, - "learning_rate": 1.7931787576062758e-05, - "loss": 0.1986, - "step": 268000 - }, - { - "epoch": 3.11, - "learning_rate": 1.7927928970794217e-05, - "loss": 0.2123, - "step": 268500 - }, - { - "epoch": 3.11, - "learning_rate": 1.792407036552568e-05, - "loss": 0.1924, - "step": 269000 - }, - { - "epoch": 3.12, - "learning_rate": 1.7920211760257137e-05, - "loss": 0.2026, - "step": 269500 - }, - { - "epoch": 3.13, - "learning_rate": 1.79163531549886e-05, - "loss": 0.2038, - "step": 270000 - }, - { - "epoch": 3.13, - "learning_rate": 1.7912494549720058e-05, - "loss": 0.1966, - "step": 270500 - }, - { - "epoch": 3.14, - "learning_rate": 1.790863594445152e-05, - "loss": 0.1984, - "step": 271000 - }, - { - "epoch": 3.14, - "learning_rate": 1.790477733918298e-05, - "loss": 0.1985, - "step": 271500 - }, - { - "epoch": 3.15, - "learning_rate": 1.790091873391444e-05, - "loss": 0.2068, - "step": 272000 - }, - { - "epoch": 3.15, - "learning_rate": 1.78970601286459e-05, - "loss": 0.2091, - "step": 272500 - }, - { - "epoch": 3.16, - "learning_rate": 1.789320152337736e-05, - "loss": 0.2041, - "step": 273000 - }, - { - "epoch": 3.17, - "learning_rate": 1.7889342918108823e-05, - "loss": 0.2076, - "step": 273500 - }, - { - "epoch": 3.17, - "learning_rate": 1.7885484312840282e-05, - "loss": 0.2049, - "step": 274000 - }, - { - "epoch": 3.18, - "learning_rate": 1.7881625707571744e-05, - "loss": 0.2042, - "step": 274500 - }, - { - "epoch": 3.18, - "learning_rate": 1.7877767102303203e-05, - "loss": 0.2046, - "step": 275000 - }, - { - "epoch": 3.19, - "learning_rate": 1.7873908497034665e-05, - "loss": 0.2111, - "step": 275500 - }, - { - "epoch": 3.19, - "learning_rate": 1.7870049891766123e-05, - "loss": 0.2082, - "step": 276000 - }, - { - "epoch": 3.2, - "learning_rate": 1.7866191286497585e-05, - "loss": 0.2052, - "step": 276500 - }, - { - "epoch": 3.21, - "learning_rate": 1.7862332681229044e-05, - "loss": 0.1885, - "step": 277000 - }, - { - "epoch": 3.21, - "learning_rate": 1.7858474075960506e-05, - "loss": 0.2165, - "step": 277500 - }, - { - "epoch": 3.22, - "learning_rate": 1.7854615470691965e-05, - "loss": 0.2122, - "step": 278000 - }, - { - "epoch": 3.22, - "learning_rate": 1.7850756865423427e-05, - "loss": 0.2084, - "step": 278500 - }, - { - "epoch": 3.23, - "learning_rate": 1.7846898260154885e-05, - "loss": 0.2104, - "step": 279000 - }, - { - "epoch": 3.24, - "learning_rate": 1.7843039654886347e-05, - "loss": 0.2143, - "step": 279500 - }, - { - "epoch": 3.24, - "learning_rate": 1.7839181049617806e-05, - "loss": 0.207, - "step": 280000 - }, - { - "epoch": 3.25, - "learning_rate": 1.7835322444349268e-05, - "loss": 0.2075, - "step": 280500 - }, - { - "epoch": 3.25, - "learning_rate": 1.7831463839080727e-05, - "loss": 0.2019, - "step": 281000 - }, - { - "epoch": 3.26, - "learning_rate": 1.782760523381219e-05, - "loss": 0.2103, - "step": 281500 - }, - { - "epoch": 3.26, - "learning_rate": 1.7823746628543647e-05, - "loss": 0.2108, - "step": 282000 - }, - { - "epoch": 3.27, - "learning_rate": 1.781988802327511e-05, - "loss": 0.2062, - "step": 282500 - }, - { - "epoch": 3.28, - "learning_rate": 1.7816029418006568e-05, - "loss": 0.1966, - "step": 283000 - }, - { - "epoch": 3.28, - "learning_rate": 1.781217081273803e-05, - "loss": 0.1954, - "step": 283500 - }, - { - "epoch": 3.29, - "learning_rate": 1.780831220746949e-05, - "loss": 0.1995, - "step": 284000 - }, - { - "epoch": 3.29, - "learning_rate": 1.780445360220095e-05, - "loss": 0.2042, - "step": 284500 - }, - { - "epoch": 3.3, - "learning_rate": 1.780059499693241e-05, - "loss": 0.2065, - "step": 285000 - }, - { - "epoch": 3.3, - "learning_rate": 1.779673639166387e-05, - "loss": 0.2019, - "step": 285500 - }, - { - "epoch": 3.31, - "learning_rate": 1.779287778639533e-05, - "loss": 0.2114, - "step": 286000 - }, - { - "epoch": 3.32, - "learning_rate": 1.7789019181126792e-05, - "loss": 0.2043, - "step": 286500 - }, - { - "epoch": 3.32, - "learning_rate": 1.778516057585825e-05, - "loss": 0.2149, - "step": 287000 - }, - { - "epoch": 3.33, - "learning_rate": 1.7781301970589712e-05, - "loss": 0.198, - "step": 287500 - }, - { - "epoch": 3.33, - "learning_rate": 1.777744336532117e-05, - "loss": 0.2076, - "step": 288000 - }, - { - "epoch": 3.34, - "learning_rate": 1.7773584760052633e-05, - "loss": 0.2019, - "step": 288500 - }, - { - "epoch": 3.35, - "learning_rate": 1.7769726154784092e-05, - "loss": 0.2132, - "step": 289000 - }, - { - "epoch": 3.35, - "learning_rate": 1.7765867549515554e-05, - "loss": 0.2149, - "step": 289500 - }, - { - "epoch": 3.36, - "learning_rate": 1.7762008944247012e-05, - "loss": 0.2053, - "step": 290000 - }, - { - "epoch": 3.36, - "learning_rate": 1.7758150338978474e-05, - "loss": 0.1947, - "step": 290500 - }, - { - "epoch": 3.37, - "learning_rate": 1.7754291733709933e-05, - "loss": 0.2054, - "step": 291000 - }, - { - "epoch": 3.37, - "learning_rate": 1.7750433128441395e-05, - "loss": 0.2098, - "step": 291500 - }, - { - "epoch": 3.38, - "learning_rate": 1.7746574523172854e-05, - "loss": 0.211, - "step": 292000 - }, - { - "epoch": 3.39, - "learning_rate": 1.7742715917904316e-05, - "loss": 0.2129, - "step": 292500 - }, - { - "epoch": 3.39, - "learning_rate": 1.7738857312635774e-05, - "loss": 0.2183, - "step": 293000 - }, - { - "epoch": 3.4, - "learning_rate": 1.7734998707367236e-05, - "loss": 0.2111, - "step": 293500 - }, - { - "epoch": 3.4, - "learning_rate": 1.7731140102098695e-05, - "loss": 0.2039, - "step": 294000 - }, - { - "epoch": 3.41, - "learning_rate": 1.7727281496830157e-05, - "loss": 0.2056, - "step": 294500 - }, - { - "epoch": 3.41, - "learning_rate": 1.7723422891561616e-05, - "loss": 0.2209, - "step": 295000 - }, - { - "epoch": 3.42, - "learning_rate": 1.7719564286293078e-05, - "loss": 0.2174, - "step": 295500 - }, - { - "epoch": 3.43, - "learning_rate": 1.7715705681024536e-05, - "loss": 0.2048, - "step": 296000 - }, - { - "epoch": 3.43, - "learning_rate": 1.7711847075755998e-05, - "loss": 0.203, - "step": 296500 - }, - { - "epoch": 3.44, - "learning_rate": 1.7707988470487457e-05, - "loss": 0.2019, - "step": 297000 - }, - { - "epoch": 3.44, - "learning_rate": 1.770412986521892e-05, - "loss": 0.2108, - "step": 297500 - }, - { - "epoch": 3.45, - "learning_rate": 1.7700271259950378e-05, - "loss": 0.2037, - "step": 298000 - }, - { - "epoch": 3.46, - "learning_rate": 1.769641265468184e-05, - "loss": 0.2025, - "step": 298500 - }, - { - "epoch": 3.46, - "learning_rate": 1.7692554049413298e-05, - "loss": 0.208, - "step": 299000 - }, - { - "epoch": 3.47, - "learning_rate": 1.768869544414476e-05, - "loss": 0.21, - "step": 299500 - }, - { - "epoch": 3.47, - "learning_rate": 1.768483683887622e-05, - "loss": 0.2082, - "step": 300000 - }, - { - "epoch": 3.48, - "learning_rate": 1.768097823360768e-05, - "loss": 0.2073, - "step": 300500 - }, - { - "epoch": 3.48, - "learning_rate": 1.767711962833914e-05, - "loss": 0.2088, - "step": 301000 - }, - { - "epoch": 3.49, - "learning_rate": 1.76732610230706e-05, - "loss": 0.221, - "step": 301500 - }, - { - "epoch": 3.5, - "learning_rate": 1.766940241780206e-05, - "loss": 0.2059, - "step": 302000 - }, - { - "epoch": 3.5, - "learning_rate": 1.7665543812533522e-05, - "loss": 0.2065, - "step": 302500 - }, - { - "epoch": 3.51, - "learning_rate": 1.7661685207264984e-05, - "loss": 0.203, - "step": 303000 - }, - { - "epoch": 3.51, - "learning_rate": 1.7657826601996443e-05, - "loss": 0.1965, - "step": 303500 - }, - { - "epoch": 3.52, - "learning_rate": 1.7653967996727905e-05, - "loss": 0.1983, - "step": 304000 - }, - { - "epoch": 3.52, - "learning_rate": 1.7650109391459363e-05, - "loss": 0.2099, - "step": 304500 - }, - { - "epoch": 3.53, - "learning_rate": 1.7646250786190825e-05, - "loss": 0.2169, - "step": 305000 - }, - { - "epoch": 3.54, - "learning_rate": 1.7642392180922284e-05, - "loss": 0.2161, - "step": 305500 - }, - { - "epoch": 3.54, - "learning_rate": 1.7638533575653746e-05, - "loss": 0.199, - "step": 306000 - }, - { - "epoch": 3.55, - "learning_rate": 1.7634674970385205e-05, - "loss": 0.2072, - "step": 306500 - }, - { - "epoch": 3.55, - "learning_rate": 1.7630816365116667e-05, - "loss": 0.2115, - "step": 307000 - }, - { - "epoch": 3.56, - "learning_rate": 1.7626957759848125e-05, - "loss": 0.2051, - "step": 307500 - }, - { - "epoch": 3.57, - "learning_rate": 1.7623099154579587e-05, - "loss": 0.2063, - "step": 308000 - }, - { - "epoch": 3.57, - "learning_rate": 1.7619240549311046e-05, - "loss": 0.2062, - "step": 308500 - }, - { - "epoch": 3.58, - "learning_rate": 1.7615381944042508e-05, - "loss": 0.2107, - "step": 309000 - }, - { - "epoch": 3.58, - "learning_rate": 1.7611523338773967e-05, - "loss": 0.1978, - "step": 309500 - }, - { - "epoch": 3.59, - "learning_rate": 1.760766473350543e-05, - "loss": 0.2074, - "step": 310000 - }, - { - "epoch": 3.59, - "learning_rate": 1.7603806128236887e-05, - "loss": 0.2014, - "step": 310500 - }, - { - "epoch": 3.6, - "learning_rate": 1.759994752296835e-05, - "loss": 0.2128, - "step": 311000 - }, - { - "epoch": 3.61, - "learning_rate": 1.7596088917699808e-05, - "loss": 0.211, - "step": 311500 - }, - { - "epoch": 3.61, - "learning_rate": 1.759223031243127e-05, - "loss": 0.2056, - "step": 312000 - }, - { - "epoch": 3.62, - "learning_rate": 1.758837170716273e-05, - "loss": 0.2023, - "step": 312500 - }, - { - "epoch": 3.62, - "learning_rate": 1.758451310189419e-05, - "loss": 0.212, - "step": 313000 - }, - { - "epoch": 3.63, - "learning_rate": 1.7580654496625653e-05, - "loss": 0.2084, - "step": 313500 - }, - { - "epoch": 3.63, - "learning_rate": 1.757679589135711e-05, - "loss": 0.1958, - "step": 314000 - }, - { - "epoch": 3.64, - "learning_rate": 1.7572937286088573e-05, - "loss": 0.1997, - "step": 314500 - }, - { - "epoch": 3.65, - "learning_rate": 1.7569078680820032e-05, - "loss": 0.206, - "step": 315000 - }, - { - "epoch": 3.65, - "learning_rate": 1.7565220075551494e-05, - "loss": 0.2068, - "step": 315500 - }, - { - "epoch": 3.66, - "learning_rate": 1.7561361470282953e-05, - "loss": 0.2209, - "step": 316000 - }, - { - "epoch": 3.66, - "learning_rate": 1.7557502865014415e-05, - "loss": 0.2086, - "step": 316500 - }, - { - "epoch": 3.67, - "learning_rate": 1.7553644259745873e-05, - "loss": 0.1992, - "step": 317000 - }, - { - "epoch": 3.68, - "learning_rate": 1.7549785654477335e-05, - "loss": 0.1929, - "step": 317500 - }, - { - "epoch": 3.68, - "learning_rate": 1.7545927049208794e-05, - "loss": 0.2042, - "step": 318000 - }, - { - "epoch": 3.69, - "learning_rate": 1.7542068443940256e-05, - "loss": 0.2031, - "step": 318500 - }, - { - "epoch": 3.69, - "learning_rate": 1.7538209838671714e-05, - "loss": 0.2076, - "step": 319000 - }, - { - "epoch": 3.7, - "learning_rate": 1.7534351233403176e-05, - "loss": 0.2038, - "step": 319500 - }, - { - "epoch": 3.7, - "learning_rate": 1.7530492628134635e-05, - "loss": 0.2113, - "step": 320000 - }, - { - "epoch": 3.71, - "learning_rate": 1.7526634022866097e-05, - "loss": 0.207, - "step": 320500 - }, - { - "epoch": 3.72, - "learning_rate": 1.7522775417597556e-05, - "loss": 0.2133, - "step": 321000 - }, - { - "epoch": 3.72, - "learning_rate": 1.7518916812329018e-05, - "loss": 0.2027, - "step": 321500 - }, - { - "epoch": 3.73, - "learning_rate": 1.7515058207060476e-05, - "loss": 0.2002, - "step": 322000 - }, - { - "epoch": 3.73, - "learning_rate": 1.751119960179194e-05, - "loss": 0.2173, - "step": 322500 - }, - { - "epoch": 3.74, - "learning_rate": 1.7507340996523397e-05, - "loss": 0.2068, - "step": 323000 - }, - { - "epoch": 3.74, - "learning_rate": 1.750348239125486e-05, - "loss": 0.1951, - "step": 323500 - }, - { - "epoch": 3.75, - "learning_rate": 1.749962378598632e-05, - "loss": 0.2117, - "step": 324000 - }, - { - "epoch": 3.76, - "learning_rate": 1.749576518071778e-05, - "loss": 0.199, - "step": 324500 - }, - { - "epoch": 3.76, - "learning_rate": 1.7491906575449242e-05, - "loss": 0.2016, - "step": 325000 - }, - { - "epoch": 3.77, - "learning_rate": 1.74880479701807e-05, - "loss": 0.203, - "step": 325500 - }, - { - "epoch": 3.77, - "learning_rate": 1.7484189364912162e-05, - "loss": 0.204, - "step": 326000 - }, - { - "epoch": 3.78, - "learning_rate": 1.748033075964362e-05, - "loss": 0.2135, - "step": 326500 - }, - { - "epoch": 3.79, - "learning_rate": 1.7476472154375083e-05, - "loss": 0.2115, - "step": 327000 - }, - { - "epoch": 3.79, - "learning_rate": 1.747261354910654e-05, - "loss": 0.2088, - "step": 327500 - }, - { - "epoch": 3.8, - "learning_rate": 1.7468754943838004e-05, - "loss": 0.2045, - "step": 328000 - }, - { - "epoch": 3.8, - "learning_rate": 1.7464896338569462e-05, - "loss": 0.2076, - "step": 328500 - }, - { - "epoch": 3.81, - "learning_rate": 1.7461037733300924e-05, - "loss": 0.2094, - "step": 329000 - }, - { - "epoch": 3.81, - "learning_rate": 1.7457179128032383e-05, - "loss": 0.2053, - "step": 329500 - }, - { - "epoch": 3.82, - "learning_rate": 1.7453320522763845e-05, - "loss": 0.222, - "step": 330000 - }, - { - "epoch": 3.83, - "learning_rate": 1.7449461917495304e-05, - "loss": 0.2011, - "step": 330500 - }, - { - "epoch": 3.83, - "learning_rate": 1.7445603312226766e-05, - "loss": 0.1948, - "step": 331000 - }, - { - "epoch": 3.84, - "learning_rate": 1.7441744706958224e-05, - "loss": 0.2022, - "step": 331500 - }, - { - "epoch": 3.84, - "learning_rate": 1.7437886101689686e-05, - "loss": 0.2084, - "step": 332000 - }, - { - "epoch": 3.85, - "learning_rate": 1.7434027496421145e-05, - "loss": 0.2012, - "step": 332500 - }, - { - "epoch": 3.85, - "learning_rate": 1.7430168891152607e-05, - "loss": 0.2099, - "step": 333000 - }, - { - "epoch": 3.86, - "learning_rate": 1.7426310285884066e-05, - "loss": 0.2229, - "step": 333500 - }, - { - "epoch": 3.87, - "learning_rate": 1.7422451680615528e-05, - "loss": 0.2107, - "step": 334000 - }, - { - "epoch": 3.87, - "learning_rate": 1.7418593075346986e-05, - "loss": 0.2112, - "step": 334500 - }, - { - "epoch": 3.88, - "learning_rate": 1.7414734470078448e-05, - "loss": 0.2041, - "step": 335000 - }, - { - "epoch": 3.88, - "learning_rate": 1.7410875864809907e-05, - "loss": 0.198, - "step": 335500 - }, - { - "epoch": 3.89, - "learning_rate": 1.740701725954137e-05, - "loss": 0.2063, - "step": 336000 - }, - { - "epoch": 3.9, - "learning_rate": 1.7403158654272827e-05, - "loss": 0.2216, - "step": 336500 - }, - { - "epoch": 3.9, - "learning_rate": 1.739930004900429e-05, - "loss": 0.2078, - "step": 337000 - }, - { - "epoch": 3.91, - "learning_rate": 1.7395441443735748e-05, - "loss": 0.2013, - "step": 337500 - }, - { - "epoch": 3.91, - "learning_rate": 1.739158283846721e-05, - "loss": 0.2011, - "step": 338000 - }, - { - "epoch": 3.92, - "learning_rate": 1.738772423319867e-05, - "loss": 0.2074, - "step": 338500 - }, - { - "epoch": 3.92, - "learning_rate": 1.738386562793013e-05, - "loss": 0.2103, - "step": 339000 - }, - { - "epoch": 3.93, - "learning_rate": 1.738000702266159e-05, - "loss": 0.2089, - "step": 339500 - }, - { - "epoch": 3.94, - "learning_rate": 1.737614841739305e-05, - "loss": 0.2029, - "step": 340000 - }, - { - "epoch": 3.94, - "learning_rate": 1.737228981212451e-05, - "loss": 0.2057, - "step": 340500 - }, - { - "epoch": 3.95, - "learning_rate": 1.7368431206855972e-05, - "loss": 0.1963, - "step": 341000 - }, - { - "epoch": 3.95, - "learning_rate": 1.736457260158743e-05, - "loss": 0.2057, - "step": 341500 - }, - { - "epoch": 3.96, - "learning_rate": 1.7360713996318893e-05, - "loss": 0.1949, - "step": 342000 - }, - { - "epoch": 3.96, - "learning_rate": 1.7356855391050355e-05, - "loss": 0.1999, - "step": 342500 - }, - { - "epoch": 3.97, - "learning_rate": 1.7352996785781813e-05, - "loss": 0.2147, - "step": 343000 - }, - { - "epoch": 3.98, - "learning_rate": 1.7349138180513275e-05, - "loss": 0.2092, - "step": 343500 - }, - { - "epoch": 3.98, - "learning_rate": 1.7345279575244734e-05, - "loss": 0.2003, - "step": 344000 - }, - { - "epoch": 3.99, - "learning_rate": 1.7341420969976196e-05, - "loss": 0.2057, - "step": 344500 - }, - { - "epoch": 3.99, - "learning_rate": 1.7337562364707655e-05, - "loss": 0.1958, - "step": 345000 - }, - { - "epoch": 4.0, - "learning_rate": 1.7333703759439117e-05, - "loss": 0.201, - "step": 345500 - }, - { - "epoch": 4.0, - "eval_bart_score": -6.65, - "eval_bertscore_f1": 0.75, - "eval_bertscore_precision": 0.79, - "eval_bertscore_recall": 0.72, - "eval_bleu": 0.0, - "eval_loss": 0.2109740972518921, - "eval_meteor": 0.09, - "eval_rogue_score": { - "rouge1": 0.16131107323752153, - "rouge2": 0.025998526352501692, - "rougeL": 0.1308022660230437, - "rougeLsum": 0.13079982676404756 - }, - "eval_runtime": 4284.4815, - "eval_samples_per_second": 5.041, - "eval_simple_accuracy": 0.78, - "eval_steps_per_second": 2.52, - "step": 345548 - }, - { - "epoch": 4.01, - "learning_rate": 1.7329845154170575e-05, - "loss": 0.192, - "step": 346000 - }, - { - "epoch": 4.01, - "learning_rate": 1.7325986548902037e-05, - "loss": 0.2071, - "step": 346500 - }, - { - "epoch": 4.02, - "learning_rate": 1.7322127943633496e-05, - "loss": 0.2114, - "step": 347000 - }, - { - "epoch": 4.02, - "learning_rate": 1.7318269338364958e-05, - "loss": 0.1951, - "step": 347500 - }, - { - "epoch": 4.03, - "learning_rate": 1.7314410733096417e-05, - "loss": 0.2019, - "step": 348000 - }, - { - "epoch": 4.03, - "learning_rate": 1.731055212782788e-05, - "loss": 0.1942, - "step": 348500 - }, - { - "epoch": 4.04, - "learning_rate": 1.7306693522559337e-05, - "loss": 0.2016, - "step": 349000 - }, - { - "epoch": 4.05, - "learning_rate": 1.73028349172908e-05, - "loss": 0.2076, - "step": 349500 - }, - { - "epoch": 4.05, - "learning_rate": 1.7298976312022258e-05, - "loss": 0.206, - "step": 350000 - }, - { - "epoch": 4.06, - "learning_rate": 1.729511770675372e-05, - "loss": 0.1963, - "step": 350500 - }, - { - "epoch": 4.06, - "learning_rate": 1.729125910148518e-05, - "loss": 0.1892, - "step": 351000 - }, - { - "epoch": 4.07, - "learning_rate": 1.728740049621664e-05, - "loss": 0.2025, - "step": 351500 - }, - { - "epoch": 4.07, - "learning_rate": 1.72835418909481e-05, - "loss": 0.201, - "step": 352000 - }, - { - "epoch": 4.08, - "learning_rate": 1.727968328567956e-05, - "loss": 0.2041, - "step": 352500 - }, - { - "epoch": 4.09, - "learning_rate": 1.727582468041102e-05, - "loss": 0.2077, - "step": 353000 - }, - { - "epoch": 4.09, - "learning_rate": 1.7271966075142482e-05, - "loss": 0.1999, - "step": 353500 - }, - { - "epoch": 4.1, - "learning_rate": 1.726810746987394e-05, - "loss": 0.1976, - "step": 354000 - }, - { - "epoch": 4.1, - "learning_rate": 1.7264248864605402e-05, - "loss": 0.1915, - "step": 354500 - }, - { - "epoch": 4.11, - "learning_rate": 1.726039025933686e-05, - "loss": 0.1992, - "step": 355000 - }, - { - "epoch": 4.12, - "learning_rate": 1.7256531654068323e-05, - "loss": 0.1924, - "step": 355500 - }, - { - "epoch": 4.12, - "learning_rate": 1.7252673048799782e-05, - "loss": 0.1934, - "step": 356000 - }, - { - "epoch": 4.13, - "learning_rate": 1.7248814443531244e-05, - "loss": 0.2034, - "step": 356500 - }, - { - "epoch": 4.13, - "learning_rate": 1.7244955838262702e-05, - "loss": 0.1922, - "step": 357000 - }, - { - "epoch": 4.14, - "learning_rate": 1.7241097232994164e-05, - "loss": 0.191, - "step": 357500 - }, - { - "epoch": 4.14, - "learning_rate": 1.7237238627725623e-05, - "loss": 0.2009, - "step": 358000 - }, - { - "epoch": 4.15, - "learning_rate": 1.7233380022457085e-05, - "loss": 0.1917, - "step": 358500 - }, - { - "epoch": 4.16, - "learning_rate": 1.7229521417188544e-05, - "loss": 0.1917, - "step": 359000 - }, - { - "epoch": 4.16, - "learning_rate": 1.7225662811920006e-05, - "loss": 0.1958, - "step": 359500 - }, - { - "epoch": 4.17, - "learning_rate": 1.7221804206651464e-05, - "loss": 0.2036, - "step": 360000 - }, - { - "epoch": 4.17, - "learning_rate": 1.7217945601382926e-05, - "loss": 0.2019, - "step": 360500 - }, - { - "epoch": 4.18, - "learning_rate": 1.7214086996114385e-05, - "loss": 0.2069, - "step": 361000 - }, - { - "epoch": 4.18, - "learning_rate": 1.7210228390845847e-05, - "loss": 0.2, - "step": 361500 - }, - { - "epoch": 4.19, - "learning_rate": 1.7206369785577306e-05, - "loss": 0.192, - "step": 362000 - }, - { - "epoch": 4.2, - "learning_rate": 1.7202511180308768e-05, - "loss": 0.2127, - "step": 362500 - }, - { - "epoch": 4.2, - "learning_rate": 1.7198652575040226e-05, - "loss": 0.2039, - "step": 363000 - }, - { - "epoch": 4.21, - "learning_rate": 1.7194793969771688e-05, - "loss": 0.2094, - "step": 363500 - }, - { - "epoch": 4.21, - "learning_rate": 1.7190935364503147e-05, - "loss": 0.2071, - "step": 364000 - }, - { - "epoch": 4.22, - "learning_rate": 1.718707675923461e-05, - "loss": 0.189, - "step": 364500 - }, - { - "epoch": 4.23, - "learning_rate": 1.7183218153966068e-05, - "loss": 0.2053, - "step": 365000 - }, - { - "epoch": 4.23, - "learning_rate": 1.717935954869753e-05, - "loss": 0.2061, - "step": 365500 - }, - { - "epoch": 4.24, - "learning_rate": 1.7175500943428988e-05, - "loss": 0.2076, - "step": 366000 - }, - { - "epoch": 4.24, - "learning_rate": 1.717164233816045e-05, - "loss": 0.2189, - "step": 366500 - }, - { - "epoch": 4.25, - "learning_rate": 1.716778373289191e-05, - "loss": 0.2027, - "step": 367000 - }, - { - "epoch": 4.25, - "learning_rate": 1.716392512762337e-05, - "loss": 0.2017, - "step": 367500 - }, - { - "epoch": 4.26, - "learning_rate": 1.716006652235483e-05, - "loss": 0.1943, - "step": 368000 - }, - { - "epoch": 4.27, - "learning_rate": 1.715620791708629e-05, - "loss": 0.1958, - "step": 368500 - }, - { - "epoch": 4.27, - "learning_rate": 1.715234931181775e-05, - "loss": 0.2038, - "step": 369000 - }, - { - "epoch": 4.28, - "learning_rate": 1.7148490706549212e-05, - "loss": 0.2037, - "step": 369500 - }, - { - "epoch": 4.28, - "learning_rate": 1.714463210128067e-05, - "loss": 0.1953, - "step": 370000 - }, - { - "epoch": 4.29, - "learning_rate": 1.7140773496012133e-05, - "loss": 0.1976, - "step": 370500 - }, - { - "epoch": 4.29, - "learning_rate": 1.713691489074359e-05, - "loss": 0.1929, - "step": 371000 - }, - { - "epoch": 4.3, - "learning_rate": 1.7133056285475053e-05, - "loss": 0.205, - "step": 371500 - }, - { - "epoch": 4.31, - "learning_rate": 1.7129197680206512e-05, - "loss": 0.2011, - "step": 372000 - }, - { - "epoch": 4.31, - "learning_rate": 1.7125339074937974e-05, - "loss": 0.2038, - "step": 372500 - }, - { - "epoch": 4.32, - "learning_rate": 1.7121480469669433e-05, - "loss": 0.2029, - "step": 373000 - }, - { - "epoch": 4.32, - "learning_rate": 1.7117621864400895e-05, - "loss": 0.2002, - "step": 373500 - }, - { - "epoch": 4.33, - "learning_rate": 1.7113763259132353e-05, - "loss": 0.1967, - "step": 374000 - }, - { - "epoch": 4.34, - "learning_rate": 1.7109904653863815e-05, - "loss": 0.2038, - "step": 374500 - }, - { - "epoch": 4.34, - "learning_rate": 1.7106046048595274e-05, - "loss": 0.206, - "step": 375000 - }, - { - "epoch": 4.35, - "learning_rate": 1.7102187443326736e-05, - "loss": 0.207, - "step": 375500 - }, - { - "epoch": 4.35, - "learning_rate": 1.7098328838058198e-05, - "loss": 0.2014, - "step": 376000 - }, - { - "epoch": 4.36, - "learning_rate": 1.7094470232789657e-05, - "loss": 0.1977, - "step": 376500 - }, - { - "epoch": 4.36, - "learning_rate": 1.709061162752112e-05, - "loss": 0.1958, - "step": 377000 - }, - { - "epoch": 4.37, - "learning_rate": 1.7086753022252577e-05, - "loss": 0.1975, - "step": 377500 - }, - { - "epoch": 4.38, - "learning_rate": 1.708289441698404e-05, - "loss": 0.2001, - "step": 378000 - }, - { - "epoch": 4.38, - "learning_rate": 1.7079035811715498e-05, - "loss": 0.201, - "step": 378500 - }, - { - "epoch": 4.39, - "learning_rate": 1.707517720644696e-05, - "loss": 0.2023, - "step": 379000 - }, - { - "epoch": 4.39, - "learning_rate": 1.707131860117842e-05, - "loss": 0.2018, - "step": 379500 - }, - { - "epoch": 4.4, - "learning_rate": 1.706745999590988e-05, - "loss": 0.2061, - "step": 380000 - }, - { - "epoch": 4.4, - "learning_rate": 1.706360139064134e-05, - "loss": 0.2104, - "step": 380500 - }, - { - "epoch": 4.41, - "learning_rate": 1.70597427853728e-05, - "loss": 0.2083, - "step": 381000 - }, - { - "epoch": 4.42, - "learning_rate": 1.705588418010426e-05, - "loss": 0.1992, - "step": 381500 - }, - { - "epoch": 4.42, - "learning_rate": 1.7052025574835722e-05, - "loss": 0.204, - "step": 382000 - }, - { - "epoch": 4.43, - "learning_rate": 1.704816696956718e-05, - "loss": 0.2064, - "step": 382500 - }, - { - "epoch": 4.43, - "learning_rate": 1.7044308364298643e-05, - "loss": 0.1923, - "step": 383000 - }, - { - "epoch": 4.44, - "learning_rate": 1.70404497590301e-05, - "loss": 0.1915, - "step": 383500 - }, - { - "epoch": 4.45, - "learning_rate": 1.7036591153761563e-05, - "loss": 0.2041, - "step": 384000 - }, - { - "epoch": 4.45, - "learning_rate": 1.7032732548493022e-05, - "loss": 0.2057, - "step": 384500 - }, - { - "epoch": 4.46, - "learning_rate": 1.7028873943224484e-05, - "loss": 0.1899, - "step": 385000 - }, - { - "epoch": 4.46, - "learning_rate": 1.7025015337955942e-05, - "loss": 0.1987, - "step": 385500 - }, - { - "epoch": 4.47, - "learning_rate": 1.7021156732687404e-05, - "loss": 0.211, - "step": 386000 - }, - { - "epoch": 4.47, - "learning_rate": 1.7017298127418863e-05, - "loss": 0.2023, - "step": 386500 - }, - { - "epoch": 4.48, - "learning_rate": 1.7013439522150325e-05, - "loss": 0.2048, - "step": 387000 - }, - { - "epoch": 4.49, - "learning_rate": 1.7009580916881784e-05, - "loss": 0.2035, - "step": 387500 - }, - { - "epoch": 4.49, - "learning_rate": 1.7005722311613246e-05, - "loss": 0.2011, - "step": 388000 - }, - { - "epoch": 4.5, - "learning_rate": 1.7001863706344704e-05, - "loss": 0.207, - "step": 388500 - }, - { - "epoch": 4.5, - "learning_rate": 1.6998005101076166e-05, - "loss": 0.2022, - "step": 389000 - }, - { - "epoch": 4.51, - "learning_rate": 1.6994146495807625e-05, - "loss": 0.1954, - "step": 389500 - }, - { - "epoch": 4.51, - "learning_rate": 1.6990287890539087e-05, - "loss": 0.2071, - "step": 390000 - }, - { - "epoch": 4.52, - "learning_rate": 1.6986429285270546e-05, - "loss": 0.1948, - "step": 390500 - }, - { - "epoch": 4.53, - "learning_rate": 1.6982570680002008e-05, - "loss": 0.197, - "step": 391000 - }, - { - "epoch": 4.53, - "learning_rate": 1.6978712074733466e-05, - "loss": 0.1984, - "step": 391500 - }, - { - "epoch": 4.54, - "learning_rate": 1.697485346946493e-05, - "loss": 0.2043, - "step": 392000 - }, - { - "epoch": 4.54, - "learning_rate": 1.6970994864196387e-05, - "loss": 0.1983, - "step": 392500 - }, - { - "epoch": 4.55, - "learning_rate": 1.696713625892785e-05, - "loss": 0.1983, - "step": 393000 - }, - { - "epoch": 4.56, - "learning_rate": 1.6963277653659308e-05, - "loss": 0.2012, - "step": 393500 - }, - { - "epoch": 4.56, - "learning_rate": 1.695941904839077e-05, - "loss": 0.2066, - "step": 394000 - }, - { - "epoch": 4.57, - "learning_rate": 1.6955560443122228e-05, - "loss": 0.2046, - "step": 394500 - }, - { - "epoch": 4.57, - "learning_rate": 1.695170183785369e-05, - "loss": 0.1981, - "step": 395000 - }, - { - "epoch": 4.58, - "learning_rate": 1.694784323258515e-05, - "loss": 0.2034, - "step": 395500 - }, - { - "epoch": 4.58, - "learning_rate": 1.694398462731661e-05, - "loss": 0.2048, - "step": 396000 - }, - { - "epoch": 4.59, - "learning_rate": 1.6940126022048073e-05, - "loss": 0.2106, - "step": 396500 - }, - { - "epoch": 4.6, - "learning_rate": 1.693626741677953e-05, - "loss": 0.1976, - "step": 397000 - }, - { - "epoch": 4.6, - "learning_rate": 1.6932408811510994e-05, - "loss": 0.2032, - "step": 397500 - }, - { - "epoch": 4.61, - "learning_rate": 1.6928550206242452e-05, - "loss": 0.2059, - "step": 398000 - }, - { - "epoch": 4.61, - "learning_rate": 1.6924691600973914e-05, - "loss": 0.2039, - "step": 398500 - }, - { - "epoch": 4.62, - "learning_rate": 1.6920832995705373e-05, - "loss": 0.1946, - "step": 399000 - }, - { - "epoch": 4.62, - "learning_rate": 1.6916974390436835e-05, - "loss": 0.2094, - "step": 399500 - }, - { - "epoch": 4.63, - "learning_rate": 1.6913115785168293e-05, - "loss": 0.2111, - "step": 400000 - }, - { - "epoch": 4.64, - "learning_rate": 1.6909257179899755e-05, - "loss": 0.2026, - "step": 400500 - }, - { - "epoch": 4.64, - "learning_rate": 1.6905398574631214e-05, - "loss": 0.2056, - "step": 401000 - }, - { - "epoch": 4.65, - "learning_rate": 1.6901539969362676e-05, - "loss": 0.2042, - "step": 401500 - }, - { - "epoch": 4.65, - "learning_rate": 1.6897681364094135e-05, - "loss": 0.1958, - "step": 402000 - }, - { - "epoch": 4.66, - "learning_rate": 1.6893822758825597e-05, - "loss": 0.1907, - "step": 402500 - }, - { - "epoch": 4.67, - "learning_rate": 1.6889964153557055e-05, - "loss": 0.2064, - "step": 403000 - }, - { - "epoch": 4.67, - "learning_rate": 1.6886105548288517e-05, - "loss": 0.2025, - "step": 403500 - }, - { - "epoch": 4.68, - "learning_rate": 1.6882246943019976e-05, - "loss": 0.2003, - "step": 404000 - }, - { - "epoch": 4.68, - "learning_rate": 1.6878388337751438e-05, - "loss": 0.1886, - "step": 404500 - }, - { - "epoch": 4.69, - "learning_rate": 1.6874529732482897e-05, - "loss": 0.1954, - "step": 405000 - }, - { - "epoch": 4.69, - "learning_rate": 1.687067112721436e-05, - "loss": 0.2081, - "step": 405500 - }, - { - "epoch": 4.7, - "learning_rate": 1.6866812521945817e-05, - "loss": 0.2059, - "step": 406000 - }, - { - "epoch": 4.71, - "learning_rate": 1.686295391667728e-05, - "loss": 0.1964, - "step": 406500 - }, - { - "epoch": 4.71, - "learning_rate": 1.6859095311408738e-05, - "loss": 0.2028, - "step": 407000 - }, - { - "epoch": 4.72, - "learning_rate": 1.68552367061402e-05, - "loss": 0.2032, - "step": 407500 - }, - { - "epoch": 4.72, - "learning_rate": 1.6851378100871662e-05, - "loss": 0.2033, - "step": 408000 - }, - { - "epoch": 4.73, - "learning_rate": 1.684751949560312e-05, - "loss": 0.1917, - "step": 408500 - }, - { - "epoch": 4.73, - "learning_rate": 1.6843660890334583e-05, - "loss": 0.2036, - "step": 409000 - }, - { - "epoch": 4.74, - "learning_rate": 1.683980228506604e-05, - "loss": 0.2028, - "step": 409500 - }, - { - "epoch": 4.75, - "learning_rate": 1.6835943679797503e-05, - "loss": 0.2007, - "step": 410000 - }, - { - "epoch": 4.75, - "learning_rate": 1.6832085074528962e-05, - "loss": 0.2072, - "step": 410500 - }, - { - "epoch": 4.76, - "learning_rate": 1.6828226469260424e-05, - "loss": 0.2059, - "step": 411000 - }, - { - "epoch": 4.76, - "learning_rate": 1.6824367863991883e-05, - "loss": 0.1921, - "step": 411500 - }, - { - "epoch": 4.77, - "learning_rate": 1.6820509258723345e-05, - "loss": 0.2001, - "step": 412000 - }, - { - "epoch": 4.78, - "learning_rate": 1.6816650653454803e-05, - "loss": 0.2051, - "step": 412500 - }, - { - "epoch": 4.78, - "learning_rate": 1.6812792048186265e-05, - "loss": 0.2081, - "step": 413000 - }, - { - "epoch": 4.79, - "learning_rate": 1.6808933442917724e-05, - "loss": 0.2087, - "step": 413500 - }, - { - "epoch": 4.79, - "learning_rate": 1.6805074837649186e-05, - "loss": 0.2157, - "step": 414000 - }, - { - "epoch": 4.8, - "learning_rate": 1.6801216232380645e-05, - "loss": 0.2083, - "step": 414500 - }, - { - "epoch": 4.8, - "learning_rate": 1.6797357627112107e-05, - "loss": 0.2008, - "step": 415000 - }, - { - "epoch": 4.81, - "learning_rate": 1.6793499021843565e-05, - "loss": 0.1974, - "step": 415500 - }, - { - "epoch": 4.82, - "learning_rate": 1.6789640416575027e-05, - "loss": 0.2012, - "step": 416000 - }, - { - "epoch": 4.82, - "learning_rate": 1.6785781811306486e-05, - "loss": 0.1853, - "step": 416500 - }, - { - "epoch": 4.83, - "learning_rate": 1.6781923206037948e-05, - "loss": 0.2057, - "step": 417000 - }, - { - "epoch": 4.83, - "learning_rate": 1.6778064600769406e-05, - "loss": 0.2052, - "step": 417500 - }, - { - "epoch": 4.84, - "learning_rate": 1.677420599550087e-05, - "loss": 0.2027, - "step": 418000 - }, - { - "epoch": 4.84, - "learning_rate": 1.677034739023233e-05, - "loss": 0.2042, - "step": 418500 - }, - { - "epoch": 4.85, - "learning_rate": 1.676648878496379e-05, - "loss": 0.1939, - "step": 419000 - }, - { - "epoch": 4.86, - "learning_rate": 1.676263017969525e-05, - "loss": 0.2066, - "step": 419500 - }, - { - "epoch": 4.86, - "learning_rate": 1.675877157442671e-05, - "loss": 0.2091, - "step": 420000 - }, - { - "epoch": 4.87, - "learning_rate": 1.6754912969158172e-05, - "loss": 0.2111, - "step": 420500 - }, - { - "epoch": 4.87, - "learning_rate": 1.675105436388963e-05, - "loss": 0.2057, - "step": 421000 - }, - { - "epoch": 4.88, - "learning_rate": 1.6747195758621092e-05, - "loss": 0.1918, - "step": 421500 - }, - { - "epoch": 4.88, - "learning_rate": 1.674333715335255e-05, - "loss": 0.2003, - "step": 422000 - }, - { - "epoch": 4.89, - "learning_rate": 1.6739478548084013e-05, - "loss": 0.2063, - "step": 422500 - }, - { - "epoch": 4.9, - "learning_rate": 1.673561994281547e-05, - "loss": 0.2117, - "step": 423000 - }, - { - "epoch": 4.9, - "learning_rate": 1.6731761337546934e-05, - "loss": 0.193, - "step": 423500 - }, - { - "epoch": 4.91, - "learning_rate": 1.6727902732278392e-05, - "loss": 0.2022, - "step": 424000 - }, - { - "epoch": 4.91, - "learning_rate": 1.6724044127009854e-05, - "loss": 0.213, - "step": 424500 - }, - { - "epoch": 4.92, - "learning_rate": 1.6720185521741313e-05, - "loss": 0.2084, - "step": 425000 - }, - { - "epoch": 4.93, - "learning_rate": 1.6716326916472775e-05, - "loss": 0.2009, - "step": 425500 - }, - { - "epoch": 4.93, - "learning_rate": 1.6712468311204234e-05, - "loss": 0.198, - "step": 426000 - }, - { - "epoch": 4.94, - "learning_rate": 1.6708609705935696e-05, - "loss": 0.2142, - "step": 426500 - }, - { - "epoch": 4.94, - "learning_rate": 1.6704751100667154e-05, - "loss": 0.1949, - "step": 427000 - }, - { - "epoch": 4.95, - "learning_rate": 1.6700892495398616e-05, - "loss": 0.2135, - "step": 427500 - }, - { - "epoch": 4.95, - "learning_rate": 1.6697033890130075e-05, - "loss": 0.2038, - "step": 428000 - }, - { - "epoch": 4.96, - "learning_rate": 1.6693175284861537e-05, - "loss": 0.1992, - "step": 428500 - }, - { - "epoch": 4.97, - "learning_rate": 1.6689316679592996e-05, - "loss": 0.2016, - "step": 429000 - }, - { - "epoch": 4.97, - "learning_rate": 1.6685458074324458e-05, - "loss": 0.2019, - "step": 429500 - }, - { - "epoch": 4.98, - "learning_rate": 1.6681599469055916e-05, - "loss": 0.1995, - "step": 430000 - }, - { - "epoch": 4.98, - "learning_rate": 1.6677740863787378e-05, - "loss": 0.1968, - "step": 430500 - }, - { - "epoch": 4.99, - "learning_rate": 1.6673882258518837e-05, - "loss": 0.1999, - "step": 431000 - }, - { - "epoch": 4.99, - "learning_rate": 1.66700236532503e-05, - "loss": 0.2088, - "step": 431500 - }, - { - "epoch": 5.0, - "eval_bart_score": -6.65, - "eval_bertscore_f1": 0.75, - "eval_bertscore_precision": 0.79, - "eval_bertscore_recall": 0.72, - "eval_bleu": 0.0, - "eval_loss": 0.21080316603183746, - "eval_meteor": 0.09, - "eval_rogue_score": { - "rouge1": 0.16481557726494733, - "rouge2": 0.02718331876189925, - "rougeL": 0.13357267291161234, - "rougeLsum": 0.1335407688877468 - }, - "eval_runtime": 4313.7823, - "eval_samples_per_second": 5.007, - "eval_simple_accuracy": 0.81, - "eval_steps_per_second": 2.503, - "step": 431935 - } - ], - "max_steps": 2591610, - "num_train_epochs": 30, - "total_flos": 5.267330341797888e+17, - "trial_name": null, - "trial_params": null -}