End of training

all_results.json

@@ -1,17 +1,17 @@
 {
     "epoch": 300.0,
-    "eval_bleu": 95.481,
-    "eval_em": 0.6019,
-    "eval_gen_len": 80.4676,
-    "eval_loss": 0.42530182003974915,
-    "eval_rm": 0.6547,
-    "eval_runtime": 118.0533,
+    "eval_bleu": 95.2745,
+    "eval_em": 0.1415,
+    "eval_gen_len": 58.7746,
+    "eval_loss": 0.35256680846214294,
+    "eval_rm": 0.3022,
+    "eval_runtime": 119.1784,
     "eval_samples": 417,
-    "eval_samples_per_second": 3.532,
-    "eval_steps_per_second": 0.229,
-    "train_loss": 0.49965806714693706,
-    "train_runtime": 5702.8364,
+    "eval_samples_per_second": 3.499,
+    "eval_steps_per_second": 0.445,
+    "train_loss": 0.006238417584671934,
+    "train_runtime": 1626.7385,
     "train_samples": 1226,
-    "train_samples_per_second": 64.494,
-    "train_steps_per_second": 1.052
+    "train_samples_per_second": 226.097,
+    "train_steps_per_second": 7.192
 }

eval_results.json

@@ -1,12 +1,12 @@
 {
     "epoch": 300.0,
-    "eval_bleu": 95.481,
-    "eval_em": 0.6019,
-    "eval_gen_len": 80.4676,
-    "eval_loss": 0.42530182003974915,
-    "eval_rm": 0.6547,
-    "eval_runtime": 118.0533,
+    "eval_bleu": 95.2745,
+    "eval_em": 0.1415,
+    "eval_gen_len": 58.7746,
+    "eval_loss": 0.35256680846214294,
+    "eval_rm": 0.3022,
+    "eval_runtime": 119.1784,
     "eval_samples": 417,
-    "eval_samples_per_second": 3.532,
-    "eval_steps_per_second": 0.229
+    "eval_samples_per_second": 3.499,
+    "eval_steps_per_second": 0.445
 }

runs/Jun02_07-11-36_8c8a3ec9844a/events.out.tfevents.1685692840.8c8a3ec9844a.3155.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:822dbd17077c59699f33b113589243b607373bef487b34a77459b5e2b7c24cb2
+size 550

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 300.0,
-    "train_loss": 0.49965806714693706,
-    "train_runtime": 5702.8364,
+    "train_loss": 0.006238417584671934,
+    "train_runtime": 1626.7385,
     "train_samples": 1226,
-    "train_samples_per_second": 64.494,
-    "train_steps_per_second": 1.052
+    "train_samples_per_second": 226.097,
+    "train_steps_per_second": 7.192
 }

trainer_state.json

@@ -2,132 +2,438 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 300.0,
-  "global_step": 6000,
+  "global_step": 11700,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 50.0,
-      "learning_rate": 2.4875e-05,
-      "loss": 2.9129,
+      "epoch": 12.82,
+      "learning_rate": 1.24e-05,
+      "loss": 4.0279,
+      "step": 500
+    },
+    {
+      "epoch": 12.82,
+      "eval_bleu": 49.841,
+      "eval_em": 0.0,
+      "eval_gen_len": 51.6403,
+      "eval_loss": 2.4031472206115723,
+      "eval_rm": 0.0,
+      "eval_runtime": 118.4236,
+      "eval_samples_per_second": 3.521,
+      "eval_steps_per_second": 0.448,
+      "step": 500
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 2.4900000000000002e-05,
+      "loss": 1.3442,
       "step": 1000
     },
     {
-      "epoch": 50.0,
-      "eval_bleu": 92.7571,
-      "eval_em": 0.2302,
-      "eval_gen_len": 79.2806,
-      "eval_loss": 0.3861088156700134,
-      "eval_rm": 0.2422,
-      "eval_runtime": 119.9581,
-      "eval_samples_per_second": 3.476,
-      "eval_steps_per_second": 0.225,
+      "epoch": 25.64,
+      "eval_bleu": 85.0177,
+      "eval_em": 0.0,
+      "eval_gen_len": 57.9784,
+      "eval_loss": 0.501366138458252,
+      "eval_rm": 0.0,
+      "eval_runtime": 163.8536,
+      "eval_samples_per_second": 2.545,
+      "eval_steps_per_second": 0.323,
       "step": 1000
     },
     {
-      "epoch": 100.0,
-      "learning_rate": 4.9875000000000006e-05,
-      "loss": 0.052,
+      "epoch": 38.46,
+      "learning_rate": 3.74e-05,
+      "loss": 0.2522,
+      "step": 1500
+    },
+    {
+      "epoch": 38.46,
+      "eval_bleu": 94.0714,
+      "eval_em": 0.0168,
+      "eval_gen_len": 57.9137,
+      "eval_loss": 0.3293180763721466,
+      "eval_rm": 0.0216,
+      "eval_runtime": 113.0226,
+      "eval_samples_per_second": 3.69,
+      "eval_steps_per_second": 0.469,
+      "step": 1500
+    },
+    {
+      "epoch": 51.28,
+      "learning_rate": 4.99e-05,
+      "loss": 0.1534,
       "step": 2000
     },
     {
-      "epoch": 100.0,
-      "eval_bleu": 95.3257,
-      "eval_em": 0.5156,
-      "eval_gen_len": 80.6859,
-      "eval_loss": 0.3531915545463562,
-      "eval_rm": 0.5635,
-      "eval_runtime": 114.8636,
-      "eval_samples_per_second": 3.63,
-      "eval_steps_per_second": 0.235,
+      "epoch": 51.28,
+      "eval_bleu": 94.4328,
+      "eval_em": 0.0024,
+      "eval_gen_len": 58.9448,
+      "eval_loss": 0.320736825466156,
+      "eval_rm": 0.0072,
+      "eval_runtime": 116.3746,
+      "eval_samples_per_second": 3.583,
+      "eval_steps_per_second": 0.455,
       "step": 2000
     },
     {
-      "epoch": 150.0,
-      "learning_rate": 3.75625e-05,
-      "loss": 0.0155,
+      "epoch": 64.1,
+      "learning_rate": 4.744329896907217e-05,
+      "loss": 0.1305,
+      "step": 2500
+    },
+    {
+      "epoch": 64.1,
+      "eval_bleu": 94.0708,
+      "eval_em": 0.0,
+      "eval_gen_len": 59.6115,
+      "eval_loss": 0.3247060477733612,
+      "eval_rm": 0.0,
+      "eval_runtime": 117.0173,
+      "eval_samples_per_second": 3.564,
+      "eval_steps_per_second": 0.453,
+      "step": 2500
+    },
+    {
+      "epoch": 76.92,
+      "learning_rate": 4.48659793814433e-05,
+      "loss": 0.1226,
       "step": 3000
     },
     {
-      "epoch": 150.0,
-      "eval_bleu": 94.8017,
-      "eval_em": 0.5659,
-      "eval_gen_len": 79.3165,
-      "eval_loss": 0.4279763698577881,
-      "eval_rm": 0.6355,
-      "eval_runtime": 113.6979,
-      "eval_samples_per_second": 3.668,
-      "eval_steps_per_second": 0.237,
+      "epoch": 76.92,
+      "eval_bleu": 94.3143,
+      "eval_em": 0.0024,
+      "eval_gen_len": 58.235,
+      "eval_loss": 0.33251264691352844,
+      "eval_rm": 0.0024,
+      "eval_runtime": 119.1624,
+      "eval_samples_per_second": 3.499,
+      "eval_steps_per_second": 0.445,
       "step": 3000
     },
     {
-      "epoch": 200.0,
-      "learning_rate": 2.50625e-05,
-      "loss": 0.0086,
+      "epoch": 89.74,
+      "learning_rate": 4.228865979381443e-05,
+      "loss": 0.1131,
+      "step": 3500
+    },
+    {
+      "epoch": 89.74,
+      "eval_bleu": 94.5678,
+      "eval_em": 0.0048,
+      "eval_gen_len": 59.6811,
+      "eval_loss": 0.3400600850582123,
+      "eval_rm": 0.0144,
+      "eval_runtime": 116.7251,
+      "eval_samples_per_second": 3.572,
+      "eval_steps_per_second": 0.454,
+      "step": 3500
+    },
+    {
+      "epoch": 102.56,
+      "learning_rate": 3.971134020618557e-05,
+      "loss": 0.1053,
       "step": 4000
     },
     {
-      "epoch": 200.0,
-      "eval_bleu": 95.5885,
-      "eval_em": 0.5803,
-      "eval_gen_len": 80.6978,
-      "eval_loss": 0.40772923827171326,
-      "eval_rm": 0.6283,
-      "eval_runtime": 118.0147,
-      "eval_samples_per_second": 3.533,
-      "eval_steps_per_second": 0.229,
+      "epoch": 102.56,
+      "eval_bleu": 94.4738,
+      "eval_em": 0.0168,
+      "eval_gen_len": 59.0288,
+      "eval_loss": 0.3373829424381256,
+      "eval_rm": 0.0552,
+      "eval_runtime": 118.4954,
+      "eval_samples_per_second": 3.519,
+      "eval_steps_per_second": 0.447,
       "step": 4000
     },
     {
-      "epoch": 250.0,
-      "learning_rate": 1.2562499999999999e-05,
-      "loss": 0.0051,
+      "epoch": 115.38,
+      "learning_rate": 3.71340206185567e-05,
+      "loss": 0.0999,
+      "step": 4500
+    },
+    {
+      "epoch": 115.38,
+      "eval_bleu": 94.6291,
+      "eval_em": 0.0336,
+      "eval_gen_len": 58.6283,
+      "eval_loss": 0.3437003791332245,
+      "eval_rm": 0.0624,
+      "eval_runtime": 119.5949,
+      "eval_samples_per_second": 3.487,
+      "eval_steps_per_second": 0.443,
+      "step": 4500
+    },
+    {
+      "epoch": 128.21,
+      "learning_rate": 3.455670103092783e-05,
+      "loss": 0.0941,
       "step": 5000
     },
     {
-      "epoch": 250.0,
-      "eval_bleu": 95.5923,
-      "eval_em": 0.6019,
-      "eval_gen_len": 80.6523,
-      "eval_loss": 0.41692906618118286,
-      "eval_rm": 0.6523,
-      "eval_runtime": 120.2437,
-      "eval_samples_per_second": 3.468,
-      "eval_steps_per_second": 0.225,
+      "epoch": 128.21,
+      "eval_bleu": 94.7896,
+      "eval_em": 0.0695,
+      "eval_gen_len": 58.4149,
+      "eval_loss": 0.351246178150177,
+      "eval_rm": 0.1271,
+      "eval_runtime": 121.3634,
+      "eval_samples_per_second": 3.436,
+      "eval_steps_per_second": 0.437,
       "step": 5000
     },
     {
-      "epoch": 300.0,
-      "learning_rate": 6.250000000000001e-08,
-      "loss": 0.0038,
+      "epoch": 141.03,
+      "learning_rate": 3.197938144329897e-05,
+      "loss": 0.0904,
+      "step": 5500
+    },
+    {
+      "epoch": 141.03,
+      "eval_bleu": 94.4101,
+      "eval_em": 0.0719,
+      "eval_gen_len": 58.2518,
+      "eval_loss": 0.34235823154449463,
+      "eval_rm": 0.1439,
+      "eval_runtime": 118.818,
+      "eval_samples_per_second": 3.51,
+      "eval_steps_per_second": 0.446,
+      "step": 5500
+    },
+    {
+      "epoch": 153.85,
+      "learning_rate": 2.9402061855670106e-05,
+      "loss": 0.0833,
       "step": 6000
     },
     {
-      "epoch": 300.0,
-      "eval_bleu": 95.481,
-      "eval_em": 0.6019,
-      "eval_gen_len": 80.4676,
-      "eval_loss": 0.42530182003974915,
-      "eval_rm": 0.6547,
-      "eval_runtime": 133.449,
-      "eval_samples_per_second": 3.125,
-      "eval_steps_per_second": 0.202,
+      "epoch": 153.85,
+      "eval_bleu": 94.7141,
+      "eval_em": 0.0887,
+      "eval_gen_len": 59.0312,
+      "eval_loss": 0.3461511433124542,
+      "eval_rm": 0.1775,
+      "eval_runtime": 116.2495,
+      "eval_samples_per_second": 3.587,
+      "eval_steps_per_second": 0.456,
       "step": 6000
     },
+    {
+      "epoch": 166.67,
+      "learning_rate": 2.6824742268041237e-05,
+      "loss": 0.0772,
+      "step": 6500
+    },
+    {
+      "epoch": 166.67,
+      "eval_bleu": 94.6758,
+      "eval_em": 0.0911,
+      "eval_gen_len": 59.0767,
+      "eval_loss": 0.34671926498413086,
+      "eval_rm": 0.2062,
+      "eval_runtime": 116.1647,
+      "eval_samples_per_second": 3.59,
+      "eval_steps_per_second": 0.456,
+      "step": 6500
+    },
+    {
+      "epoch": 179.49,
+      "learning_rate": 2.4247422680412372e-05,
+      "loss": 0.0722,
+      "step": 7000
+    },
+    {
+      "epoch": 179.49,
+      "eval_bleu": 94.5698,
+      "eval_em": 0.1055,
+      "eval_gen_len": 58.1415,
+      "eval_loss": 0.3461613953113556,
+      "eval_rm": 0.2398,
+      "eval_runtime": 119.2771,
+      "eval_samples_per_second": 3.496,
+      "eval_steps_per_second": 0.444,
+      "step": 7000
+    },
+    {
+      "epoch": 192.31,
+      "learning_rate": 2.1670103092783507e-05,
+      "loss": 0.0669,
+      "step": 7500
+    },
+    {
+      "epoch": 192.31,
+      "eval_bleu": 95.0365,
+      "eval_em": 0.1223,
+      "eval_gen_len": 58.7794,
+      "eval_loss": 0.35367459058761597,
+      "eval_rm": 0.2782,
+      "eval_runtime": 115.018,
+      "eval_samples_per_second": 3.626,
+      "eval_steps_per_second": 0.461,
+      "step": 7500
+    },
+    {
+      "epoch": 205.13,
+      "learning_rate": 1.9092783505154642e-05,
+      "loss": 0.062,
+      "step": 8000
+    },
+    {
+      "epoch": 205.13,
+      "eval_bleu": 94.8694,
+      "eval_em": 0.1247,
+      "eval_gen_len": 58.211,
+      "eval_loss": 0.35051023960113525,
+      "eval_rm": 0.2686,
+      "eval_runtime": 113.7476,
+      "eval_samples_per_second": 3.666,
+      "eval_steps_per_second": 0.466,
+      "step": 8000
+    },
+    {
+      "epoch": 217.95,
+      "learning_rate": 1.6515463917525774e-05,
+      "loss": 0.0576,
+      "step": 8500
+    },
+    {
+      "epoch": 217.95,
+      "eval_bleu": 94.8168,
+      "eval_em": 0.1271,
+      "eval_gen_len": 59.0791,
+      "eval_loss": 0.3510896563529968,
+      "eval_rm": 0.2926,
+      "eval_runtime": 117.1223,
+      "eval_samples_per_second": 3.56,
+      "eval_steps_per_second": 0.453,
+      "step": 8500
+    },
+    {
+      "epoch": 230.77,
+      "learning_rate": 1.3938144329896907e-05,
+      "loss": 0.0539,
+      "step": 9000
+    },
+    {
+      "epoch": 230.77,
+      "eval_bleu": 95.1935,
+      "eval_em": 0.1367,
+      "eval_gen_len": 58.6787,
+      "eval_loss": 0.34899094700813293,
+      "eval_rm": 0.3046,
+      "eval_runtime": 117.1796,
+      "eval_samples_per_second": 3.559,
+      "eval_steps_per_second": 0.452,
+      "step": 9000
+    },
+    {
+      "epoch": 243.59,
+      "learning_rate": 1.1360824742268042e-05,
+      "loss": 0.0502,
+      "step": 9500
+    },
+    {
+      "epoch": 243.59,
+      "eval_bleu": 95.1882,
+      "eval_em": 0.1319,
+      "eval_gen_len": 58.5228,
+      "eval_loss": 0.3490062654018402,
+      "eval_rm": 0.3141,
+      "eval_runtime": 118.559,
+      "eval_samples_per_second": 3.517,
+      "eval_steps_per_second": 0.447,
+      "step": 9500
+    },
+    {
+      "epoch": 256.41,
+      "learning_rate": 8.783505154639175e-06,
+      "loss": 0.0473,
+      "step": 10000
+    },
+    {
+      "epoch": 256.41,
+      "eval_bleu": 95.1198,
+      "eval_em": 0.1319,
+      "eval_gen_len": 58.4245,
+      "eval_loss": 0.3504057824611664,
+      "eval_rm": 0.307,
+      "eval_runtime": 118.462,
+      "eval_samples_per_second": 3.52,
+      "eval_steps_per_second": 0.447,
+      "step": 10000
+    },
+    {
+      "epoch": 269.23,
+      "learning_rate": 6.206185567010309e-06,
+      "loss": 0.045,
+      "step": 10500
+    },
+    {
+      "epoch": 269.23,
+      "eval_bleu": 95.047,
+      "eval_em": 0.1343,
+      "eval_gen_len": 58.3213,
+      "eval_loss": 0.35046613216400146,
+      "eval_rm": 0.307,
+      "eval_runtime": 118.1147,
+      "eval_samples_per_second": 3.53,
+      "eval_steps_per_second": 0.449,
+      "step": 10500
+    },
+    {
+      "epoch": 282.05,
+      "learning_rate": 3.6288659793814435e-06,
+      "loss": 0.0429,
+      "step": 11000
+    },
+    {
+      "epoch": 282.05,
+      "eval_bleu": 95.2397,
+      "eval_em": 0.1391,
+      "eval_gen_len": 58.7242,
+      "eval_loss": 0.3522409200668335,
+      "eval_rm": 0.3046,
+      "eval_runtime": 119.4326,
+      "eval_samples_per_second": 3.492,
+      "eval_steps_per_second": 0.444,
+      "step": 11000
+    },
+    {
+      "epoch": 294.87,
+      "learning_rate": 1.0515463917525774e-06,
+      "loss": 0.0416,
+      "step": 11500
+    },
+    {
+      "epoch": 294.87,
+      "eval_bleu": 95.2821,
+      "eval_em": 0.1415,
+      "eval_gen_len": 58.7746,
+      "eval_loss": 0.3522770404815674,
+      "eval_rm": 0.3046,
+      "eval_runtime": 119.2922,
+      "eval_samples_per_second": 3.496,
+      "eval_steps_per_second": 0.444,
+      "step": 11500
+    },
     {
       "epoch": 300.0,
-      "step": 6000,
-      "total_flos": 1.014399554069376e+16,
-      "train_loss": 0.49965806714693706,
-      "train_runtime": 5702.8364,
-      "train_samples_per_second": 64.494,
-      "train_steps_per_second": 1.052
+      "step": 11700,
+      "total_flos": 9455798374608960.0,
+      "train_loss": 0.006238417584671934,
+      "train_runtime": 1626.7385,
+      "train_samples_per_second": 226.097,
+      "train_steps_per_second": 7.192
     }
   ],
-  "max_steps": 6000,
+  "max_steps": 11700,
   "num_train_epochs": 300,
-  "total_flos": 1.014399554069376e+16,
+  "total_flos": 9455798374608960.0,
   "trial_name": null,
   "trial_params": null
 }