Model save

all_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 1.0,
     "total_flos": 1.997211509414953e+17,
-    "train_loss": 9.52029996245872,
-    "train_runtime": 979.0233,
+    "train_loss": 9.503099798246195,
+    "train_runtime": 980.6026,
     "train_samples": 46801,
-    "train_samples_per_second": 8.551,
+    "train_samples_per_second": 8.538,
     "train_steps_per_second": 0.134
 }

runs/Nov14_20-46-29_main-ah-train-0-0/events.out.tfevents.1731635747.main-ah-train-0-0.517.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c14cc80a78b3ecc5397ddd5fbd231e3dc6074bbdb912ae8b2337ddb501468fab
-size 12244
+oid sha256:aecca69d4f3d4d451bc52749af401ca45d9ba9697bbbf21df24b6948018b48f2
+size 12869

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 1.0,
     "total_flos": 1.997211509414953e+17,
-    "train_loss": 9.52029996245872,
-    "train_runtime": 979.0233,
+    "train_loss": 9.503099798246195,
+    "train_runtime": 980.6026,
     "train_samples": 46801,
-    "train_samples_per_second": 8.551,
+    "train_samples_per_second": 8.538,
     "train_steps_per_second": 0.134
 }

trainer_state.json

@@ -10,208 +10,208 @@
   "log_history": [
     {
       "epoch": 0.007633587786259542,
-      "grad_norm": 177.0453643798828,
+      "grad_norm": 177.04818725585938,
       "learning_rate": 1.4285714285714285e-05,
       "loss": 47.6977,
       "step": 1
     },
     {
       "epoch": 0.03816793893129771,
-      "grad_norm": 85.61766815185547,
+      "grad_norm": 85.4582748413086,
       "learning_rate": 7.142857142857143e-05,
-      "loss": 43.3804,
+      "loss": 43.3706,
       "step": 5
     },
     {
       "epoch": 0.07633587786259542,
-      "grad_norm": 21.188264846801758,
+      "grad_norm": 21.15643882751465,
       "learning_rate": 0.00014285714285714287,
-      "loss": 32.4675,
+      "loss": 32.4713,
       "step": 10
     },
     {
       "epoch": 0.11450381679389313,
-      "grad_norm": 12.669232368469238,
+      "grad_norm": 12.584202766418457,
       "learning_rate": 0.00019996395276708856,
-      "loss": 26.336,
+      "loss": 26.3606,
       "step": 15
     },
     {
       "epoch": 0.15267175572519084,
-      "grad_norm": 6.164061069488525,
+      "grad_norm": 6.0736002922058105,
       "learning_rate": 0.00019870502626379127,
-      "loss": 23.7885,
+      "loss": 23.8068,
       "step": 20
     },
     {
       "epoch": 0.19083969465648856,
-      "grad_norm": 9.347785949707031,
+      "grad_norm": 9.369542121887207,
       "learning_rate": 0.00019566964208274254,
-      "loss": 22.3163,
+      "loss": 22.3161,
       "step": 25
     },
     {
       "epoch": 0.22900763358778625,
-      "grad_norm": 16.819992065429688,
+      "grad_norm": 16.93528175354004,
       "learning_rate": 0.0001909124299802724,
-      "loss": 20.7837,
+      "loss": 20.7882,
       "step": 30
     },
     {
       "epoch": 0.26717557251908397,
-      "grad_norm": 27.521188735961914,
+      "grad_norm": 27.67462730407715,
       "learning_rate": 0.0001845190085543795,
-      "loss": 17.9141,
+      "loss": 17.8876,
       "step": 35
     },
     {
       "epoch": 0.3053435114503817,
-      "grad_norm": 36.15729522705078,
+      "grad_norm": 36.653324127197266,
       "learning_rate": 0.0001766044443118978,
-      "loss": 12.8744,
+      "loss": 12.7851,
       "step": 40
     },
     {
       "epoch": 0.3435114503816794,
-      "grad_norm": 30.39793586730957,
+      "grad_norm": 29.700342178344727,
       "learning_rate": 0.00016731118074275704,
-      "loss": 7.4619,
+      "loss": 7.3703,
       "step": 45
     },
     {
       "epoch": 0.3816793893129771,
-      "grad_norm": 13.962003707885742,
+      "grad_norm": 13.149837493896484,
       "learning_rate": 0.00015680647467311557,
-      "loss": 4.1925,
+      "loss": 4.148,
       "step": 50
     },
     {
       "epoch": 0.4198473282442748,
-      "grad_norm": 9.878511428833008,
+      "grad_norm": 9.42947769165039,
       "learning_rate": 0.00014527938603696376,
-      "loss": 3.5013,
+      "loss": 3.4698,
       "step": 55
     },
     {
       "epoch": 0.4580152671755725,
-      "grad_norm": 4.401936054229736,
+      "grad_norm": 4.0911455154418945,
       "learning_rate": 0.00013293737524320797,
-      "loss": 2.9085,
+      "loss": 2.8863,
       "step": 60
     },
     {
       "epoch": 0.4961832061068702,
-      "grad_norm": 2.499587297439575,
+      "grad_norm": 2.124103546142578,
       "learning_rate": 0.00012000256937760445,
-      "loss": 2.6199,
+      "loss": 2.5943,
       "step": 65
     },
     {
       "epoch": 0.5343511450381679,
-      "grad_norm": 4.096787452697754,
+      "grad_norm": 3.0718891620635986,
       "learning_rate": 0.00010670776443910024,
-      "loss": 2.4025,
+      "loss": 2.385,
       "step": 70
     },
     {
       "epoch": 0.5725190839694656,
-      "grad_norm": 2.0821242332458496,
+      "grad_norm": 3.1116530895233154,
       "learning_rate": 9.329223556089975e-05,
-      "loss": 2.3207,
+      "loss": 2.2977,
       "step": 75
     },
     {
       "epoch": 0.6106870229007634,
-      "grad_norm": 1.016592025756836,
+      "grad_norm": 0.9759073257446289,
       "learning_rate": 7.999743062239557e-05,
-      "loss": 2.2112,
+      "loss": 2.1913,
       "step": 80
     },
     {
       "epoch": 0.648854961832061,
-      "grad_norm": 2.0435054302215576,
+      "grad_norm": 2.0841591358184814,
       "learning_rate": 6.706262475679205e-05,
-      "loss": 2.1628,
+      "loss": 2.1452,
       "step": 85
     },
     {
       "epoch": 0.6870229007633588,
-      "grad_norm": 1.766175627708435,
+      "grad_norm": 1.6919430494308472,
       "learning_rate": 5.472061396303629e-05,
-      "loss": 2.1301,
+      "loss": 2.1161,
       "step": 90
     },
     {
       "epoch": 0.7251908396946565,
-      "grad_norm": 0.9447934031486511,
+      "grad_norm": 0.9739937782287598,
       "learning_rate": 4.3193525326884435e-05,
-      "loss": 2.0873,
+      "loss": 2.0801,
       "step": 95
     },
     {
       "epoch": 0.7633587786259542,
-      "grad_norm": 1.282375454902649,
+      "grad_norm": 1.5150854587554932,
       "learning_rate": 3.268881925724297e-05,
-      "loss": 2.0758,
+      "loss": 2.0685,
       "step": 100
     },
     {
       "epoch": 0.8015267175572519,
-      "grad_norm": 1.1836905479431152,
+      "grad_norm": 1.2287063598632812,
       "learning_rate": 2.339555568810221e-05,
-      "loss": 2.0401,
+      "loss": 2.0339,
       "step": 105
     },
     {
       "epoch": 0.8396946564885496,
-      "grad_norm": 4.228659629821777,
+      "grad_norm": 25.910524368286133,
       "learning_rate": 1.5480991445620542e-05,
-      "loss": 2.0129,
+      "loss": 2.0006,
       "step": 110
     },
     {
       "epoch": 0.8778625954198473,
-      "grad_norm": 0.9176464676856995,
+      "grad_norm": 0.8792179226875305,
       "learning_rate": 9.08757001972762e-06,
-      "loss": 2.0476,
+      "loss": 2.0391,
       "step": 115
     },
     {
       "epoch": 0.916030534351145,
-      "grad_norm": 2.124687433242798,
+      "grad_norm": 2.187098741531372,
       "learning_rate": 4.3303579172574885e-06,
-      "loss": 2.017,
+      "loss": 2.0099,
       "step": 120
     },
     {
       "epoch": 0.9541984732824428,
-      "grad_norm": 1.6751036643981934,
+      "grad_norm": 1.3658674955368042,
       "learning_rate": 1.2949737362087156e-06,
-      "loss": 2.0379,
+      "loss": 2.0285,
       "step": 125
     },
     {
       "epoch": 0.9923664122137404,
-      "grad_norm": 1.4553974866867065,
+      "grad_norm": 1.4813926219940186,
       "learning_rate": 3.60472329114625e-08,
-      "loss": 2.0437,
+      "loss": 2.034,
       "step": 130
     },
     {
       "epoch": 1.0,
-      "eval_loss": 2.034517288208008,
-      "eval_runtime": 27.6,
-      "eval_samples_per_second": 33.37,
-      "eval_steps_per_second": 1.413,
+      "eval_loss": 2.0285837650299072,
+      "eval_runtime": 27.7829,
+      "eval_samples_per_second": 33.15,
+      "eval_steps_per_second": 1.404,
       "step": 131
     },
     {
       "epoch": 1.0,
       "step": 131,
       "total_flos": 1.997211509414953e+17,
-      "train_loss": 9.52029996245872,
-      "train_runtime": 979.0233,
-      "train_samples_per_second": 8.551,
+      "train_loss": 9.503099798246195,
+      "train_runtime": 980.6026,
+      "train_samples_per_second": 8.538,
       "train_steps_per_second": 0.134
     }
   ],