Updated model

all_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 20.0,
-    "train_loss": 1.6082039656264058,
-    "train_runtime": 1452.4648,
+    "train_loss": 1.6175778072871518,
+    "train_runtime": 1440.6716,
     "train_samples": 2846,
-    "train_samples_per_second": 39.189,
-    "train_steps_per_second": 4.902
+    "train_samples_per_second": 39.509,
+    "train_steps_per_second": 4.942
 }

config.json

@@ -26,9 +26,9 @@
     "translation": {
       "early_stopping": true,
       "max_length": 64,
-      "num_beams": 4,
+      "num_beams": 8,
       "prefix": ""
-    }
+    },
   },
   "torch_dtype": "float32",
   "transformers_version": "4.11.0.dev0",

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:40686b7396955e900069fa47f553b838b0f55b5d141bbdfbe324230cf16b80d5
+oid sha256:510fc77ae19b27fb6443b9ecfd842338af324333a035f7f8b1b267aaf9d47f87
 size 2950790023

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 20.0,
-    "train_loss": 1.6082039656264058,
-    "train_runtime": 1452.4648,
+    "train_loss": 1.6175778072871518,
+    "train_runtime": 1440.6716,
     "train_samples": 2846,
-    "train_samples_per_second": 39.189,
-    "train_steps_per_second": 4.902
+    "train_samples_per_second": 39.509,
+    "train_steps_per_second": 4.942
 }

trainer_state.json

@@ -9,443 +9,443 @@
   "log_history": [
     {
       "epoch": 0.28,
-      "learning_rate": 2.5e-06,
-      "loss": 5.4155,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 5.4466,
       "step": 100
     },
     {
       "epoch": 0.56,
-      "learning_rate": 5e-06,
-      "loss": 4.1706,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 4.3117,
       "step": 200
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.500000000000001e-06,
-      "loss": 3.4307,
+      "learning_rate": 6e-06,
+      "loss": 3.6079,
       "step": 300
     },
     {
       "epoch": 1.12,
-      "learning_rate": 1e-05,
-      "loss": 2.7978,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.9597,
       "step": 400
     },
     {
       "epoch": 1.4,
-      "learning_rate": 9.851190476190477e-06,
-      "loss": 2.5436,
+      "learning_rate": 1e-05,
+      "loss": 2.5962,
       "step": 500
     },
     {
       "epoch": 1.69,
-      "learning_rate": 9.702380952380953e-06,
-      "loss": 2.4266,
+      "learning_rate": 9.848942598187312e-06,
+      "loss": 2.4697,
       "step": 600
     },
     {
       "epoch": 1.97,
-      "learning_rate": 9.55357142857143e-06,
-      "loss": 2.3508,
+      "learning_rate": 9.697885196374624e-06,
+      "loss": 2.3802,
       "step": 700
     },
     {
       "epoch": 2.25,
-      "learning_rate": 9.404761904761905e-06,
-      "loss": 2.2192,
+      "learning_rate": 9.546827794561935e-06,
+      "loss": 2.2533,
       "step": 800
     },
     {
       "epoch": 2.53,
-      "learning_rate": 9.255952380952381e-06,
-      "loss": 2.1034,
+      "learning_rate": 9.395770392749246e-06,
+      "loss": 2.1236,
       "step": 900
     },
     {
       "epoch": 2.81,
-      "learning_rate": 9.107142857142858e-06,
-      "loss": 2.1141,
+      "learning_rate": 9.244712990936556e-06,
+      "loss": 2.1289,
       "step": 1000
     },
     {
       "epoch": 3.09,
-      "learning_rate": 8.958333333333334e-06,
-      "loss": 2.0817,
+      "learning_rate": 9.093655589123867e-06,
+      "loss": 2.1009,
       "step": 1100
     },
     {
       "epoch": 3.37,
-      "learning_rate": 8.80952380952381e-06,
-      "loss": 1.9094,
+      "learning_rate": 8.94259818731118e-06,
+      "loss": 1.9022,
       "step": 1200
     },
     {
       "epoch": 3.65,
-      "learning_rate": 8.660714285714286e-06,
-      "loss": 1.9655,
+      "learning_rate": 8.79154078549849e-06,
+      "loss": 1.9634,
       "step": 1300
     },
     {
       "epoch": 3.93,
-      "learning_rate": 8.511904761904762e-06,
-      "loss": 1.8821,
+      "learning_rate": 8.640483383685801e-06,
+      "loss": 1.8827,
       "step": 1400
     },
     {
       "epoch": 4.21,
-      "learning_rate": 8.36309523809524e-06,
-      "loss": 1.8639,
+      "learning_rate": 8.489425981873114e-06,
+      "loss": 1.8761,
       "step": 1500
     },
     {
       "epoch": 4.49,
-      "learning_rate": 8.214285714285714e-06,
-      "loss": 1.7393,
+      "learning_rate": 8.338368580060424e-06,
+      "loss": 1.7582,
       "step": 1600
     },
     {
       "epoch": 4.78,
-      "learning_rate": 8.065476190476192e-06,
-      "loss": 1.7837,
+      "learning_rate": 8.187311178247735e-06,
+      "loss": 1.7879,
       "step": 1700
     },
     {
       "epoch": 5.06,
-      "learning_rate": 7.916666666666667e-06,
-      "loss": 1.8089,
+      "learning_rate": 8.036253776435046e-06,
+      "loss": 1.7946,
       "step": 1800
     },
     {
       "epoch": 5.34,
-      "learning_rate": 7.767857142857144e-06,
-      "loss": 1.652,
+      "learning_rate": 7.885196374622357e-06,
+      "loss": 1.6653,
       "step": 1900
     },
     {
       "epoch": 5.62,
-      "learning_rate": 7.61904761904762e-06,
-      "loss": 1.6584,
+      "learning_rate": 7.734138972809667e-06,
+      "loss": 1.6526,
       "step": 2000
     },
     {
       "epoch": 5.9,
-      "learning_rate": 7.470238095238096e-06,
-      "loss": 1.6747,
+      "learning_rate": 7.58308157099698e-06,
+      "loss": 1.6819,
       "step": 2100
     },
     {
       "epoch": 6.18,
-      "learning_rate": 7.321428571428572e-06,
-      "loss": 1.7032,
+      "learning_rate": 7.432024169184291e-06,
+      "loss": 1.7149,
       "step": 2200
     },
     {
       "epoch": 6.46,
-      "learning_rate": 7.172619047619048e-06,
-      "loss": 1.6471,
+      "learning_rate": 7.280966767371602e-06,
+      "loss": 1.6477,
       "step": 2300
     },
     {
       "epoch": 6.74,
-      "learning_rate": 7.023809523809524e-06,
-      "loss": 1.6124,
+      "learning_rate": 7.129909365558912e-06,
+      "loss": 1.6371,
       "step": 2400
     },
     {
       "epoch": 7.02,
-      "learning_rate": 6.875e-06,
-      "loss": 1.4935,
+      "learning_rate": 6.978851963746224e-06,
+      "loss": 1.4978,
       "step": 2500
     },
     {
       "epoch": 7.3,
-      "learning_rate": 6.726190476190477e-06,
-      "loss": 1.5444,
+      "learning_rate": 6.827794561933535e-06,
+      "loss": 1.5412,
       "step": 2600
     },
     {
       "epoch": 7.58,
-      "learning_rate": 6.5773809523809525e-06,
-      "loss": 1.5369,
+      "learning_rate": 6.676737160120846e-06,
+      "loss": 1.5468,
       "step": 2700
     },
     {
       "epoch": 7.87,
-      "learning_rate": 6.4285714285714295e-06,
-      "loss": 1.4954,
+      "learning_rate": 6.525679758308158e-06,
+      "loss": 1.4743,
       "step": 2800
     },
     {
       "epoch": 8.15,
-      "learning_rate": 6.279761904761906e-06,
-      "loss": 1.4372,
+      "learning_rate": 6.374622356495469e-06,
+      "loss": 1.4338,
       "step": 2900
     },
     {
       "epoch": 8.43,
-      "learning_rate": 6.130952380952382e-06,
-      "loss": 1.4984,
+      "learning_rate": 6.22356495468278e-06,
+      "loss": 1.503,
       "step": 3000
     },
     {
       "epoch": 8.71,
-      "learning_rate": 5.982142857142858e-06,
-      "loss": 1.4569,
+      "learning_rate": 6.072507552870092e-06,
+      "loss": 1.4474,
       "step": 3100
     },
     {
       "epoch": 8.99,
-      "learning_rate": 5.833333333333334e-06,
-      "loss": 1.4227,
+      "learning_rate": 5.921450151057402e-06,
+      "loss": 1.4009,
       "step": 3200
     },
     {
       "epoch": 9.27,
-      "learning_rate": 5.68452380952381e-06,
-      "loss": 1.3471,
+      "learning_rate": 5.7703927492447134e-06,
+      "loss": 1.3461,
       "step": 3300
     },
     {
       "epoch": 9.55,
-      "learning_rate": 5.535714285714286e-06,
-      "loss": 1.3956,
+      "learning_rate": 5.619335347432024e-06,
+      "loss": 1.4044,
       "step": 3400
     },
     {
       "epoch": 9.83,
-      "learning_rate": 5.386904761904762e-06,
-      "loss": 1.4264,
+      "learning_rate": 5.468277945619336e-06,
+      "loss": 1.4333,
       "step": 3500
     },
     {
       "epoch": 10.11,
-      "learning_rate": 5.2380952380952384e-06,
-      "loss": 1.4372,
+      "learning_rate": 5.3172205438066474e-06,
+      "loss": 1.4325,
       "step": 3600
     },
     {
       "epoch": 10.39,
-      "learning_rate": 5.0892857142857146e-06,
-      "loss": 1.3805,
+      "learning_rate": 5.166163141993958e-06,
+      "loss": 1.3621,
       "step": 3700
     },
     {
       "epoch": 10.67,
-      "learning_rate": 4.940476190476191e-06,
-      "loss": 1.3379,
+      "learning_rate": 5.01510574018127e-06,
+      "loss": 1.3343,
       "step": 3800
     },
     {
       "epoch": 10.96,
-      "learning_rate": 4.791666666666668e-06,
-      "loss": 1.2706,
+      "learning_rate": 4.864048338368581e-06,
+      "loss": 1.2809,
       "step": 3900
     },
     {
       "epoch": 11.24,
-      "learning_rate": 4.642857142857144e-06,
-      "loss": 1.2747,
+      "learning_rate": 4.712990936555891e-06,
+      "loss": 1.2812,
       "step": 4000
     },
     {
       "epoch": 11.52,
-      "learning_rate": 4.49404761904762e-06,
-      "loss": 1.3314,
+      "learning_rate": 4.561933534743202e-06,
+      "loss": 1.3322,
       "step": 4100
     },
     {
       "epoch": 11.8,
-      "learning_rate": 4.345238095238096e-06,
-      "loss": 1.2674,
+      "learning_rate": 4.410876132930514e-06,
+      "loss": 1.2709,
       "step": 4200
     },
     {
       "epoch": 12.08,
-      "learning_rate": 4.196428571428572e-06,
-      "loss": 1.2883,
+      "learning_rate": 4.259818731117825e-06,
+      "loss": 1.2743,
       "step": 4300
     },
     {
       "epoch": 12.36,
-      "learning_rate": 4.047619047619048e-06,
-      "loss": 1.2421,
+      "learning_rate": 4.108761329305136e-06,
+      "loss": 1.2477,
       "step": 4400
     },
     {
       "epoch": 12.64,
-      "learning_rate": 3.898809523809524e-06,
-      "loss": 1.2721,
+      "learning_rate": 3.957703927492447e-06,
+      "loss": 1.2639,
       "step": 4500
     },
     {
       "epoch": 12.92,
-      "learning_rate": 3.7500000000000005e-06,
-      "loss": 1.2663,
+      "learning_rate": 3.8066465256797586e-06,
+      "loss": 1.251,
       "step": 4600
     },
     {
       "epoch": 13.2,
-      "learning_rate": 3.6011904761904766e-06,
-      "loss": 1.2647,
+      "learning_rate": 3.6555891238670698e-06,
+      "loss": 1.2467,
       "step": 4700
     },
     {
       "epoch": 13.48,
-      "learning_rate": 3.4523809523809528e-06,
-      "loss": 1.225,
+      "learning_rate": 3.5045317220543814e-06,
+      "loss": 1.2274,
       "step": 4800
     },
     {
       "epoch": 13.76,
-      "learning_rate": 3.303571428571429e-06,
-      "loss": 1.2095,
+      "learning_rate": 3.353474320241692e-06,
+      "loss": 1.2029,
       "step": 4900
     },
     {
       "epoch": 14.04,
-      "learning_rate": 3.154761904761905e-06,
-      "loss": 1.2523,
+      "learning_rate": 3.2024169184290033e-06,
+      "loss": 1.2557,
       "step": 5000
     },
     {
       "epoch": 14.33,
-      "learning_rate": 3.005952380952381e-06,
-      "loss": 1.2451,
+      "learning_rate": 3.0513595166163145e-06,
+      "loss": 1.243,
       "step": 5100
     },
     {
       "epoch": 14.61,
-      "learning_rate": 2.8571428571428573e-06,
-      "loss": 1.2037,
+      "learning_rate": 2.9003021148036257e-06,
+      "loss": 1.1927,
       "step": 5200
     },
     {
       "epoch": 14.89,
-      "learning_rate": 2.7083333333333334e-06,
-      "loss": 1.1763,
+      "learning_rate": 2.7492447129909365e-06,
+      "loss": 1.1605,
       "step": 5300
     },
     {
       "epoch": 15.17,
-      "learning_rate": 2.5595238095238095e-06,
-      "loss": 1.1879,
+      "learning_rate": 2.598187311178248e-06,
+      "loss": 1.1765,
       "step": 5400
     },
     {
       "epoch": 15.45,
-      "learning_rate": 2.410714285714286e-06,
-      "loss": 1.195,
+      "learning_rate": 2.447129909365559e-06,
+      "loss": 1.2024,
       "step": 5500
     },
     {
       "epoch": 15.73,
-      "learning_rate": 2.261904761904762e-06,
-      "loss": 1.1531,
+      "learning_rate": 2.2960725075528705e-06,
+      "loss": 1.1623,
       "step": 5600
     },
     {
       "epoch": 16.01,
-      "learning_rate": 2.1130952380952383e-06,
-      "loss": 1.1864,
+      "learning_rate": 2.1450151057401813e-06,
+      "loss": 1.1947,
       "step": 5700
     },
     {
       "epoch": 16.29,
-      "learning_rate": 1.9642857142857144e-06,
-      "loss": 1.188,
+      "learning_rate": 1.9939577039274925e-06,
+      "loss": 1.2003,
       "step": 5800
     },
     {
       "epoch": 16.57,
-      "learning_rate": 1.8154761904761905e-06,
-      "loss": 1.1434,
+      "learning_rate": 1.8429003021148037e-06,
+      "loss": 1.1331,
       "step": 5900
     },
     {
       "epoch": 16.85,
-      "learning_rate": 1.6666666666666667e-06,
-      "loss": 1.1691,
+      "learning_rate": 1.691842900302115e-06,
+      "loss": 1.1626,
       "step": 6000
     },
     {
       "epoch": 17.13,
-      "learning_rate": 1.5178571428571428e-06,
-      "loss": 1.1299,
+      "learning_rate": 1.540785498489426e-06,
+      "loss": 1.1243,
       "step": 6100
     },
     {
       "epoch": 17.42,
-      "learning_rate": 1.3690476190476193e-06,
-      "loss": 1.1265,
+      "learning_rate": 1.3897280966767373e-06,
+      "loss": 1.1287,
       "step": 6200
     },
     {
       "epoch": 17.7,
-      "learning_rate": 1.2202380952380952e-06,
-      "loss": 1.1849,
+      "learning_rate": 1.2386706948640485e-06,
+      "loss": 1.177,
       "step": 6300
     },
     {
       "epoch": 17.98,
-      "learning_rate": 1.0714285714285714e-06,
-      "loss": 1.1179,
+      "learning_rate": 1.0876132930513596e-06,
+      "loss": 1.1077,
       "step": 6400
     },
     {
       "epoch": 18.26,
-      "learning_rate": 9.226190476190477e-07,
-      "loss": 1.1428,
+      "learning_rate": 9.365558912386707e-07,
+      "loss": 1.1487,
       "step": 6500
     },
     {
       "epoch": 18.54,
-      "learning_rate": 7.738095238095239e-07,
-      "loss": 1.1704,
+      "learning_rate": 7.854984894259819e-07,
+      "loss": 1.1825,
       "step": 6600
     },
     {
       "epoch": 18.82,
-      "learning_rate": 6.25e-07,
-      "loss": 1.1172,
+      "learning_rate": 6.344410876132931e-07,
+      "loss": 1.0969,
       "step": 6700
     },
     {
       "epoch": 19.1,
-      "learning_rate": 4.7619047619047623e-07,
-      "loss": 1.1291,
+      "learning_rate": 4.833836858006042e-07,
+      "loss": 1.1235,
       "step": 6800
     },
     {
       "epoch": 19.38,
-      "learning_rate": 3.273809523809524e-07,
-      "loss": 1.1492,
+      "learning_rate": 3.3232628398791547e-07,
+      "loss": 1.1588,
       "step": 6900
     },
     {
       "epoch": 19.66,
-      "learning_rate": 1.7857142857142858e-07,
-      "loss": 1.1023,
+      "learning_rate": 1.812688821752266e-07,
+      "loss": 1.1082,
       "step": 7000
     },
     {
       "epoch": 19.94,
-      "learning_rate": 2.9761904761904764e-08,
-      "loss": 1.1626,
+      "learning_rate": 3.0211480362537764e-08,
+      "loss": 1.1526,
       "step": 7100
     },
     {
       "epoch": 20.0,
       "step": 7120,
-      "total_flos": 1.3000229462016e+16,
-      "train_loss": 1.6082039656264058,
-      "train_runtime": 1452.4648,
-      "train_samples_per_second": 39.189,
-      "train_steps_per_second": 4.902
+      "total_flos": 1.2829190725632e+16,
+      "train_loss": 1.6175778072871518,
+      "train_runtime": 1440.6716,
+      "train_samples_per_second": 39.509,
+      "train_steps_per_second": 4.942
     }
   ],
   "max_steps": 7120,
   "num_train_epochs": 20,
-  "total_flos": 1.3000229462016e+16,
+  "total_flos": 1.2829190725632e+16,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:ad65936847fc670c8449b7c2ff71d343d79d1c4c1c93260a6b6c8295ffd32c61
-size 2927