Training in progress, step 32

adapter_config.json

@@ -15,17 +15,17 @@
   "megatron_core": "megatron.core",
   "modules_to_save": null,
   "peft_type": "LORA",
-  "r": 64,
+  "r": 128,
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "down_proj",
     "q_proj",
     "k_proj",
-    "o_proj",
-    "gate_proj",
     "v_proj",
-    "up_proj"
+    "down_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj"
   ],
   "task_type": "CAUSAL_LM",
   "use_dora": false,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8302097a7c7f9798e5f539519d4c45118daaa85f82558da3129ba7a18dbb233a
-size 156927136
+oid sha256:c02841a6132e93c9d065bbedc5f2c54a0f8a85b0b42a6e1e972d961ba391d91d
+size 313820752

all_results.json

@@ -1,7 +1,7 @@
 {
     "epoch": 3.0,
-    "train_loss": 0.5746448771096766,
-    "train_runtime": 1124.5772,
-    "train_samples_per_second": 7.285,
-    "train_steps_per_second": 0.455
+    "train_loss": 0.5747196031734347,
+    "train_runtime": 1145.6051,
+    "train_samples_per_second": 7.151,
+    "train_steps_per_second": 0.447
 }

train_results.json

@@ -1,7 +1,7 @@
 {
     "epoch": 3.0,
-    "train_loss": 0.5746448771096766,
-    "train_runtime": 1124.5772,
-    "train_samples_per_second": 7.285,
-    "train_steps_per_second": 0.455
+    "train_loss": 0.5747196031734347,
+    "train_runtime": 1145.6051,
+    "train_samples_per_second": 7.151,
+    "train_steps_per_second": 0.447
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.43469348549842834,
+  "best_metric": 0.4347132444381714,
   "best_model_checkpoint": "outputs/checkpoint-512",
   "epoch": 3.002932551319648,
   "eval_steps": 32,
@@ -10,588 +10,588 @@
   "log_history": [
     {
       "epoch": 0.05,
-      "grad_norm": 0.103515625,
+      "grad_norm": 0.0712890625,
       "learning_rate": 2.5e-05,
-      "loss": 1.2622,
+      "loss": 1.2621,
       "step": 8
     },
     {
       "epoch": 0.09,
-      "grad_norm": 0.08984375,
+      "grad_norm": 0.0625,
       "learning_rate": 5e-05,
-      "loss": 1.2695,
+      "loss": 1.2701,
       "step": 16
     },
     {
       "epoch": 0.14,
-      "grad_norm": 0.1181640625,
+      "grad_norm": 0.08203125,
       "learning_rate": 7.500000000000001e-05,
-      "loss": 1.1917,
+      "loss": 1.1933,
       "step": 24
     },
     {
       "epoch": 0.19,
-      "grad_norm": 0.1689453125,
+      "grad_norm": 0.11572265625,
       "learning_rate": 0.0001,
-      "loss": 1.0581,
+      "loss": 1.0597,
       "step": 32
     },
     {
       "epoch": 0.19,
-      "eval_loss": 0.8670921325683594,
-      "eval_runtime": 2.7914,
-      "eval_samples_per_second": 17.195,
-      "eval_steps_per_second": 2.149,
+      "eval_loss": 0.8685370087623596,
+      "eval_runtime": 2.8151,
+      "eval_samples_per_second": 17.051,
+      "eval_steps_per_second": 2.131,
       "step": 32
     },
     {
       "epoch": 0.23,
-      "grad_norm": 0.12158203125,
+      "grad_norm": 0.087890625,
       "learning_rate": 9.833333333333333e-05,
-      "loss": 0.8582,
+      "loss": 0.8593,
       "step": 40
     },
     {
       "epoch": 0.28,
-      "grad_norm": 0.09619140625,
+      "grad_norm": 0.06640625,
       "learning_rate": 9.666666666666667e-05,
       "loss": 0.7709,
       "step": 48
     },
     {
       "epoch": 0.33,
-      "grad_norm": 0.08642578125,
+      "grad_norm": 0.060546875,
       "learning_rate": 9.5e-05,
-      "loss": 0.7284,
+      "loss": 0.7282,
       "step": 56
     },
     {
       "epoch": 0.38,
-      "grad_norm": 0.1318359375,
+      "grad_norm": 0.0947265625,
       "learning_rate": 9.333333333333334e-05,
-      "loss": 0.7043,
+      "loss": 0.7039,
       "step": 64
     },
     {
       "epoch": 0.38,
-      "eval_loss": 0.6009237170219421,
-      "eval_runtime": 2.1819,
-      "eval_samples_per_second": 22.0,
-      "eval_steps_per_second": 2.75,
+      "eval_loss": 0.6007441282272339,
+      "eval_runtime": 2.1964,
+      "eval_samples_per_second": 21.854,
+      "eval_steps_per_second": 2.732,
       "step": 64
     },
     {
       "epoch": 0.42,
-      "grad_norm": 0.0986328125,
+      "grad_norm": 0.0693359375,
       "learning_rate": 9.166666666666667e-05,
-      "loss": 0.6693,
+      "loss": 0.669,
       "step": 72
     },
     {
       "epoch": 0.47,
-      "grad_norm": 0.09814453125,
+      "grad_norm": 0.06884765625,
       "learning_rate": 9e-05,
       "loss": 0.6409,
       "step": 80
     },
     {
       "epoch": 0.52,
-      "grad_norm": 0.1279296875,
+      "grad_norm": 0.09130859375,
       "learning_rate": 8.833333333333333e-05,
       "loss": 0.6038,
       "step": 88
     },
     {
       "epoch": 0.56,
-      "grad_norm": 0.0947265625,
+      "grad_norm": 0.06787109375,
       "learning_rate": 8.666666666666667e-05,
-      "loss": 0.6093,
+      "loss": 0.6092,
       "step": 96
     },
     {
       "epoch": 0.56,
-      "eval_loss": 0.5342517495155334,
-      "eval_runtime": 2.1808,
-      "eval_samples_per_second": 22.01,
-      "eval_steps_per_second": 2.751,
+      "eval_loss": 0.5340104699134827,
+      "eval_runtime": 2.1966,
+      "eval_samples_per_second": 21.852,
+      "eval_steps_per_second": 2.732,
       "step": 96
     },
     {
       "epoch": 0.61,
-      "grad_norm": 0.1005859375,
+      "grad_norm": 0.0703125,
       "learning_rate": 8.5e-05,
       "loss": 0.5974,
       "step": 104
     },
     {
       "epoch": 0.66,
-      "grad_norm": 0.11474609375,
+      "grad_norm": 0.0791015625,
       "learning_rate": 8.333333333333334e-05,
       "loss": 0.5732,
       "step": 112
     },
     {
       "epoch": 0.7,
-      "grad_norm": 0.11669921875,
+      "grad_norm": 0.08154296875,
       "learning_rate": 8.166666666666667e-05,
-      "loss": 0.5907,
+      "loss": 0.5904,
       "step": 120
     },
     {
       "epoch": 0.75,
-      "grad_norm": 0.0966796875,
+      "grad_norm": 0.06884765625,
       "learning_rate": 8e-05,
-      "loss": 0.5548,
+      "loss": 0.5545,
       "step": 128
     },
     {
       "epoch": 0.75,
-      "eval_loss": 0.5006037950515747,
-      "eval_runtime": 2.1797,
-      "eval_samples_per_second": 22.021,
-      "eval_steps_per_second": 2.753,
+      "eval_loss": 0.5003952383995056,
+      "eval_runtime": 2.1968,
+      "eval_samples_per_second": 21.85,
+      "eval_steps_per_second": 2.731,
       "step": 128
     },
     {
       "epoch": 0.8,
-      "grad_norm": 0.095703125,
+      "grad_norm": 0.06787109375,
       "learning_rate": 7.833333333333333e-05,
-      "loss": 0.5563,
+      "loss": 0.5562,
       "step": 136
     },
     {
       "epoch": 0.84,
-      "grad_norm": 0.1142578125,
+      "grad_norm": 0.0810546875,
       "learning_rate": 7.666666666666667e-05,
       "loss": 0.5311,
       "step": 144
     },
     {
       "epoch": 0.89,
-      "grad_norm": 0.126953125,
+      "grad_norm": 0.08740234375,
       "learning_rate": 7.500000000000001e-05,
-      "loss": 0.5284,
+      "loss": 0.5281,
       "step": 152
     },
     {
       "epoch": 0.94,
-      "grad_norm": 0.09814453125,
+      "grad_norm": 0.0693359375,
       "learning_rate": 7.333333333333333e-05,
-      "loss": 0.5316,
+      "loss": 0.5312,
       "step": 160
     },
     {
       "epoch": 0.94,
-      "eval_loss": 0.48313403129577637,
-      "eval_runtime": 2.1847,
-      "eval_samples_per_second": 21.971,
-      "eval_steps_per_second": 2.746,
+      "eval_loss": 0.482632040977478,
+      "eval_runtime": 2.1969,
+      "eval_samples_per_second": 21.849,
+      "eval_steps_per_second": 2.731,
       "step": 160
     },
     {
       "epoch": 0.99,
-      "grad_norm": 0.09912109375,
+      "grad_norm": 0.0703125,
       "learning_rate": 7.166666666666667e-05,
-      "loss": 0.5324,
+      "loss": 0.5323,
       "step": 168
     },
     {
       "epoch": 1.03,
-      "grad_norm": 0.1455078125,
+      "grad_norm": 0.10498046875,
       "learning_rate": 7e-05,
-      "loss": 0.5456,
+      "loss": 0.5454,
       "step": 176
     },
     {
       "epoch": 1.08,
-      "grad_norm": 0.1025390625,
+      "grad_norm": 0.072265625,
       "learning_rate": 6.833333333333333e-05,
-      "loss": 0.5274,
+      "loss": 0.5273,
       "step": 184
     },
     {
       "epoch": 1.13,
-      "grad_norm": 0.11181640625,
+      "grad_norm": 0.08154296875,
       "learning_rate": 6.666666666666667e-05,
-      "loss": 0.5161,
+      "loss": 0.5159,
       "step": 192
     },
     {
       "epoch": 1.13,
-      "eval_loss": 0.4703657329082489,
-      "eval_runtime": 2.1798,
-      "eval_samples_per_second": 22.021,
-      "eval_steps_per_second": 2.753,
+      "eval_loss": 0.47037437558174133,
+      "eval_runtime": 2.1959,
+      "eval_samples_per_second": 21.859,
+      "eval_steps_per_second": 2.732,
       "step": 192
     },
     {
       "epoch": 1.17,
-      "grad_norm": 0.12060546875,
+      "grad_norm": 0.0849609375,
       "learning_rate": 6.500000000000001e-05,
-      "loss": 0.5129,
+      "loss": 0.513,
       "step": 200
     },
     {
       "epoch": 1.22,
-      "grad_norm": 0.11083984375,
+      "grad_norm": 0.07763671875,
       "learning_rate": 6.333333333333333e-05,
-      "loss": 0.5455,
+      "loss": 0.5456,
       "step": 208
     },
     {
       "epoch": 1.27,
-      "grad_norm": 0.1142578125,
+      "grad_norm": 0.080078125,
       "learning_rate": 6.166666666666667e-05,
-      "loss": 0.5113,
+      "loss": 0.5109,
       "step": 216
     },
     {
       "epoch": 1.31,
-      "grad_norm": 0.134765625,
+      "grad_norm": 0.09423828125,
       "learning_rate": 6e-05,
       "loss": 0.4937,
       "step": 224
     },
     {
       "epoch": 1.31,
-      "eval_loss": 0.4608299434185028,
-      "eval_runtime": 2.1784,
-      "eval_samples_per_second": 22.035,
-      "eval_steps_per_second": 2.754,
+      "eval_loss": 0.46075066924095154,
+      "eval_runtime": 2.202,
+      "eval_samples_per_second": 21.798,
+      "eval_steps_per_second": 2.725,
       "step": 224
     },
     {
       "epoch": 1.36,
-      "grad_norm": 0.10791015625,
+      "grad_norm": 0.076171875,
       "learning_rate": 5.833333333333334e-05,
-      "loss": 0.5279,
+      "loss": 0.5277,
       "step": 232
     },
     {
       "epoch": 1.41,
-      "grad_norm": 0.1181640625,
+      "grad_norm": 0.0830078125,
       "learning_rate": 5.666666666666667e-05,
       "loss": 0.5057,
       "step": 240
     },
     {
       "epoch": 1.45,
-      "grad_norm": 0.11328125,
+      "grad_norm": 0.0771484375,
       "learning_rate": 5.500000000000001e-05,
-      "loss": 0.4863,
+      "loss": 0.4862,
       "step": 248
     },
     {
       "epoch": 1.5,
-      "grad_norm": 0.1376953125,
+      "grad_norm": 0.09619140625,
       "learning_rate": 5.333333333333333e-05,
-      "loss": 0.5236,
+      "loss": 0.5235,
       "step": 256
     },
     {
       "epoch": 1.5,
-      "eval_loss": 0.454235315322876,
-      "eval_runtime": 2.1818,
-      "eval_samples_per_second": 22.0,
-      "eval_steps_per_second": 2.75,
+      "eval_loss": 0.45398303866386414,
+      "eval_runtime": 2.1965,
+      "eval_samples_per_second": 21.853,
+      "eval_steps_per_second": 2.732,
       "step": 256
     },
     {
       "epoch": 1.55,
-      "grad_norm": 0.11181640625,
+      "grad_norm": 0.078125,
       "learning_rate": 5.166666666666667e-05,
-      "loss": 0.5,
+      "loss": 0.4998,
       "step": 264
     },
     {
       "epoch": 1.6,
-      "grad_norm": 0.1259765625,
+      "grad_norm": 0.08837890625,
       "learning_rate": 5e-05,
-      "loss": 0.4989,
+      "loss": 0.4988,
       "step": 272
     },
     {
       "epoch": 1.64,
-      "grad_norm": 0.12451171875,
+      "grad_norm": 0.08837890625,
       "learning_rate": 4.8333333333333334e-05,
-      "loss": 0.4944,
+      "loss": 0.4945,
       "step": 280
     },
     {
       "epoch": 1.69,
-      "grad_norm": 0.11962890625,
+      "grad_norm": 0.08447265625,
       "learning_rate": 4.666666666666667e-05,
-      "loss": 0.5064,
+      "loss": 0.5067,
       "step": 288
     },
     {
       "epoch": 1.69,
-      "eval_loss": 0.4474702775478363,
-      "eval_runtime": 2.1767,
-      "eval_samples_per_second": 22.052,
-      "eval_steps_per_second": 2.756,
+      "eval_loss": 0.44761922955513,
+      "eval_runtime": 2.1988,
+      "eval_samples_per_second": 21.83,
+      "eval_steps_per_second": 2.729,
       "step": 288
     },
     {
       "epoch": 1.74,
-      "grad_norm": 0.103515625,
+      "grad_norm": 0.0732421875,
       "learning_rate": 4.5e-05,
-      "loss": 0.5323,
+      "loss": 0.5321,
       "step": 296
     },
     {
       "epoch": 1.78,
-      "grad_norm": 0.1328125,
+      "grad_norm": 0.09423828125,
       "learning_rate": 4.3333333333333334e-05,
-      "loss": 0.4955,
+      "loss": 0.4956,
       "step": 304
     },
     {
       "epoch": 1.83,
-      "grad_norm": 0.10546875,
+      "grad_norm": 0.07470703125,
       "learning_rate": 4.166666666666667e-05,
-      "loss": 0.5137,
+      "loss": 0.5138,
       "step": 312
     },
     {
       "epoch": 1.88,
-      "grad_norm": 0.1337890625,
+      "grad_norm": 0.095703125,
       "learning_rate": 4e-05,
-      "loss": 0.497,
+      "loss": 0.4974,
       "step": 320
     },
     {
       "epoch": 1.88,
-      "eval_loss": 0.442803293466568,
-      "eval_runtime": 2.1775,
-      "eval_samples_per_second": 22.044,
-      "eval_steps_per_second": 2.755,
+      "eval_loss": 0.44292059540748596,
+      "eval_runtime": 2.1962,
+      "eval_samples_per_second": 21.856,
+      "eval_steps_per_second": 2.732,
       "step": 320
     },
     {
       "epoch": 1.92,
-      "grad_norm": 0.10791015625,
+      "grad_norm": 0.076171875,
       "learning_rate": 3.8333333333333334e-05,
       "loss": 0.5032,
       "step": 328
     },
     {
       "epoch": 1.97,
-      "grad_norm": 0.130859375,
+      "grad_norm": 0.0927734375,
       "learning_rate": 3.6666666666666666e-05,
       "loss": 0.4914,
       "step": 336
     },
     {
       "epoch": 2.02,
-      "grad_norm": 0.1171875,
+      "grad_norm": 0.08251953125,
       "learning_rate": 3.5e-05,
-      "loss": 0.5013,
+      "loss": 0.5014,
       "step": 344
     },
     {
       "epoch": 2.06,
-      "grad_norm": 0.1328125,
+      "grad_norm": 0.095703125,
       "learning_rate": 3.3333333333333335e-05,
-      "loss": 0.4933,
+      "loss": 0.4935,
       "step": 352
     },
     {
       "epoch": 2.06,
-      "eval_loss": 0.4411480128765106,
-      "eval_runtime": 2.1813,
-      "eval_samples_per_second": 22.005,
-      "eval_steps_per_second": 2.751,
+      "eval_loss": 0.4412342607975006,
+      "eval_runtime": 2.1973,
+      "eval_samples_per_second": 21.845,
+      "eval_steps_per_second": 2.731,
       "step": 352
     },
     {
       "epoch": 2.11,
-      "grad_norm": 0.1201171875,
+      "grad_norm": 0.083984375,
       "learning_rate": 3.1666666666666666e-05,
       "loss": 0.4741,
       "step": 360
     },
     {
       "epoch": 2.16,
-      "grad_norm": 0.1181640625,
+      "grad_norm": 0.08349609375,
       "learning_rate": 3e-05,
-      "loss": 0.4802,
+      "loss": 0.4801,
       "step": 368
     },
     {
       "epoch": 2.21,
-      "grad_norm": 0.1181640625,
+      "grad_norm": 0.08203125,
       "learning_rate": 2.8333333333333335e-05,
-      "loss": 0.4911,
+      "loss": 0.4914,
       "step": 376
     },
     {
       "epoch": 2.25,
-      "grad_norm": 0.1455078125,
+      "grad_norm": 0.1044921875,
       "learning_rate": 2.6666666666666667e-05,
-      "loss": 0.517,
+      "loss": 0.5172,
       "step": 384
     },
     {
       "epoch": 2.25,
-      "eval_loss": 0.4384828805923462,
-      "eval_runtime": 2.1808,
-      "eval_samples_per_second": 22.01,
-      "eval_steps_per_second": 2.751,
+      "eval_loss": 0.438584566116333,
+      "eval_runtime": 2.1974,
+      "eval_samples_per_second": 21.844,
+      "eval_steps_per_second": 2.731,
       "step": 384
     },
     {
       "epoch": 2.3,
-      "grad_norm": 0.1318359375,
+      "grad_norm": 0.09521484375,
       "learning_rate": 2.5e-05,
-      "loss": 0.5024,
+      "loss": 0.5027,
       "step": 392
     },
     {
       "epoch": 2.35,
-      "grad_norm": 0.1259765625,
+      "grad_norm": 0.0888671875,
       "learning_rate": 2.3333333333333336e-05,
-      "loss": 0.4802,
+      "loss": 0.4804,
       "step": 400
     },
     {
       "epoch": 2.39,
-      "grad_norm": 0.12109375,
+      "grad_norm": 0.08740234375,
       "learning_rate": 2.1666666666666667e-05,
-      "loss": 0.4857,
+      "loss": 0.4858,
       "step": 408
     },
     {
       "epoch": 2.44,
-      "grad_norm": 0.11962890625,
+      "grad_norm": 0.08447265625,
       "learning_rate": 2e-05,
-      "loss": 0.4949,
+      "loss": 0.4952,
       "step": 416
     },
     {
       "epoch": 2.44,
-      "eval_loss": 0.4360348880290985,
-      "eval_runtime": 2.179,
-      "eval_samples_per_second": 22.029,
-      "eval_steps_per_second": 2.754,
+      "eval_loss": 0.4362548887729645,
+      "eval_runtime": 2.1965,
+      "eval_samples_per_second": 21.853,
+      "eval_steps_per_second": 2.732,
       "step": 416
     },
     {
       "epoch": 2.49,
-      "grad_norm": 0.12890625,
+      "grad_norm": 0.091796875,
       "learning_rate": 1.8333333333333333e-05,
-      "loss": 0.501,
+      "loss": 0.5012,
       "step": 424
     },
     {
       "epoch": 2.53,
-      "grad_norm": 0.1318359375,
+      "grad_norm": 0.091796875,
       "learning_rate": 1.6666666666666667e-05,
-      "loss": 0.4595,
+      "loss": 0.4597,
       "step": 432
     },
     {
       "epoch": 2.58,
-      "grad_norm": 0.11962890625,
+      "grad_norm": 0.08544921875,
       "learning_rate": 1.5e-05,
-      "loss": 0.4715,
+      "loss": 0.4714,
       "step": 440
     },
     {
       "epoch": 2.63,
-      "grad_norm": 0.12255859375,
+      "grad_norm": 0.08642578125,
       "learning_rate": 1.3333333333333333e-05,
-      "loss": 0.4946,
+      "loss": 0.4948,
       "step": 448
     },
     {
       "epoch": 2.63,
-      "eval_loss": 0.4355371296405792,
-      "eval_runtime": 2.1775,
-      "eval_samples_per_second": 22.044,
-      "eval_steps_per_second": 2.755,
+      "eval_loss": 0.43569865822792053,
+      "eval_runtime": 2.1992,
+      "eval_samples_per_second": 21.826,
+      "eval_steps_per_second": 2.728,
       "step": 448
     },
     {
       "epoch": 2.67,
-      "grad_norm": 0.123046875,
+      "grad_norm": 0.0869140625,
       "learning_rate": 1.1666666666666668e-05,
-      "loss": 0.4692,
+      "loss": 0.4693,
       "step": 456
     },
     {
       "epoch": 2.72,
-      "grad_norm": 0.125,
+      "grad_norm": 0.087890625,
       "learning_rate": 1e-05,
-      "loss": 0.4751,
+      "loss": 0.4752,
       "step": 464
     },
     {
       "epoch": 2.77,
-      "grad_norm": 0.12060546875,
+      "grad_norm": 0.08642578125,
       "learning_rate": 8.333333333333334e-06,
-      "loss": 0.5144,
+      "loss": 0.5147,
       "step": 472
     },
     {
       "epoch": 2.82,
-      "grad_norm": 0.119140625,
+      "grad_norm": 0.0830078125,
       "learning_rate": 6.666666666666667e-06,
-      "loss": 0.4589,
+      "loss": 0.4591,
       "step": 480
     },
     {
       "epoch": 2.82,
-      "eval_loss": 0.4347504675388336,
-      "eval_runtime": 2.1766,
-      "eval_samples_per_second": 22.053,
-      "eval_steps_per_second": 2.757,
+      "eval_loss": 0.4348466098308563,
+      "eval_runtime": 2.2046,
+      "eval_samples_per_second": 21.773,
+      "eval_steps_per_second": 2.722,
       "step": 480
     },
     {
       "epoch": 2.86,
-      "grad_norm": 0.1162109375,
+      "grad_norm": 0.08203125,
       "learning_rate": 5e-06,
-      "loss": 0.4768,
+      "loss": 0.4769,
       "step": 488
     },
     {
       "epoch": 2.91,
-      "grad_norm": 0.130859375,
+      "grad_norm": 0.0927734375,
       "learning_rate": 3.3333333333333333e-06,
       "loss": 0.4795,
       "step": 496
     },
     {
       "epoch": 2.96,
-      "grad_norm": 0.1201171875,
+      "grad_norm": 0.08349609375,
       "learning_rate": 1.6666666666666667e-06,
-      "loss": 0.4992,
+      "loss": 0.4994,
       "step": 504
     },
     {
       "epoch": 3.0,
-      "grad_norm": 0.11328125,
+      "grad_norm": 0.07958984375,
       "learning_rate": 0.0,
-      "loss": 0.466,
+      "loss": 0.4659,
       "step": 512
     },
     {
       "epoch": 3.0,
-      "eval_loss": 0.43469348549842834,
-      "eval_runtime": 2.179,
-      "eval_samples_per_second": 22.029,
-      "eval_steps_per_second": 2.754,
+      "eval_loss": 0.4347132444381714,
+      "eval_runtime": 2.1993,
+      "eval_samples_per_second": 21.826,
+      "eval_steps_per_second": 2.728,
       "step": 512
     },
     {
       "epoch": 3.0,
       "step": 512,
-      "total_flos": 9.863515762146509e+16,
-      "train_loss": 0.5746448771096766,
-      "train_runtime": 1124.5772,
-      "train_samples_per_second": 7.285,
-      "train_steps_per_second": 0.455
+      "total_flos": 1.005493534658642e+17,
+      "train_loss": 0.5747196031734347,
+      "train_runtime": 1145.6051,
+      "train_samples_per_second": 7.151,
+      "train_steps_per_second": 0.447
     }
   ],
   "logging_steps": 8,
@@ -599,7 +599,7 @@
   "num_input_tokens_seen": 0,
   "num_train_epochs": 4,
   "save_steps": 32,
-  "total_flos": 9.863515762146509e+16,
+  "total_flos": 1.005493534658642e+17,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a6ec9b9f5f072535b9595e9544338e214bf19768c7fa12b3894f9ba7b868bfbc
+oid sha256:738a99af48764578616ab76c70ebcf572b54f5fe269bac24e0288a200cfecedd
 size 4920