🍻 cheers

README.md

@@ -2,6 +2,7 @@
 license: apache-2.0
 base_model: google/vit-base-patch16-224-in21k
 tags:
+- image-classification
 - generated_from_trainer
 model-index:
 - name: ryan04152024_ALLDATA
@@ -13,7 +14,12 @@ should probably proofread and complete it, then remove this comment. -->
 
 # ryan04152024_ALLDATA
 
-This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on the None dataset.
+This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on the properties dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.1193
+- Ordinal Mae: 0.3505
+- Ordinal Accuracy: 0.7757
+- Na Accuracy: 0.9411
 
 ## Model description
 

all_results.json

@@ -1,7 +1,14 @@
 {
-    "epoch": 1.01,
-    "train_loss": 0.21891093194946762,
-    "train_runtime": 6427.7582,
-    "train_samples_per_second": 8.027,
-    "train_steps_per_second": 0.502
+    "epoch": 1.02,
+    "eval_loss": 0.1193477213382721,
+    "eval_na_accuracy": 0.9410737752914429,
+    "eval_ordinal_accuracy": 0.7756819128990173,
+    "eval_ordinal_mae": 0.350502610206604,
+    "eval_runtime": 1992.5951,
+    "eval_samples_per_second": 25.638,
+    "eval_steps_per_second": 3.205,
+    "train_loss": 0.2225519281828883,
+    "train_runtime": 2924.985,
+    "train_samples_per_second": 17.815,
+    "train_steps_per_second": 1.114
 }

eval_results.json

@@ -0,0 +1,10 @@
+{
+    "epoch": 1.02,
+    "eval_loss": 0.1193477213382721,
+    "eval_na_accuracy": 0.9410737752914429,
+    "eval_ordinal_accuracy": 0.7756819128990173,
+    "eval_ordinal_mae": 0.350502610206604,
+    "eval_runtime": 1992.5951,
+    "eval_samples_per_second": 25.638,
+    "eval_steps_per_second": 3.205
+}

runs/Apr15_16-57-14_ryanserver/events.out.tfevents.1713219645.ryanserver.25404.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:921604af27040465e89282dcd46e1b4b4a3be3d6f82fe704e4b642150afd563a
+size 529

train_results.json

@@ -1,7 +1,7 @@
 {
-    "epoch": 1.01,
-    "train_loss": 0.21891093194946762,
-    "train_runtime": 6427.7582,
-    "train_samples_per_second": 8.027,
-    "train_steps_per_second": 0.502
+    "epoch": 1.02,
+    "train_loss": 0.2225519281828883,
+    "train_runtime": 2924.985,
+    "train_samples_per_second": 17.815,
+    "train_steps_per_second": 1.114
 }

trainer_state.json

@@ -1,932 +1,939 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.0100219229564673,
+  "epoch": 1.0200438459129346,
   "eval_steps": 100,
-  "global_step": 3225,
+  "global_step": 3257,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.01,
-      "grad_norm": 0.6999576091766357,
-      "learning_rate": 4.96124031007752e-05,
-      "loss": 0.533,
+      "grad_norm": 0.6797587275505066,
+      "learning_rate": 4.961621123733498e-05,
+      "loss": 0.5293,
       "step": 25
     },
     {
       "epoch": 0.02,
-      "grad_norm": 0.682187020778656,
-      "learning_rate": 4.922480620155039e-05,
-      "loss": 0.3801,
+      "grad_norm": 0.559562087059021,
+      "learning_rate": 4.9232422474669945e-05,
+      "loss": 0.3908,
       "step": 50
     },
     {
       "epoch": 0.02,
-      "grad_norm": 0.6211907267570496,
-      "learning_rate": 4.883720930232558e-05,
-      "loss": 0.3996,
+      "grad_norm": 0.6713371872901917,
+      "learning_rate": 4.884863371200491e-05,
+      "loss": 0.4101,
       "step": 75
     },
     {
       "epoch": 0.03,
-      "grad_norm": 2.131941080093384,
-      "learning_rate": 4.8449612403100775e-05,
-      "loss": 0.3272,
+      "grad_norm": 2.220188617706299,
+      "learning_rate": 4.846484494933989e-05,
+      "loss": 0.3246,
       "step": 100
     },
     {
       "epoch": 0.04,
-      "grad_norm": 0.6068922877311707,
-      "learning_rate": 4.8062015503875976e-05,
-      "loss": 0.3431,
+      "grad_norm": 0.7794200778007507,
+      "learning_rate": 4.8081056186674856e-05,
+      "loss": 0.3521,
       "step": 125
     },
     {
       "epoch": 0.05,
-      "grad_norm": 2.6034927368164062,
-      "learning_rate": 4.7674418604651164e-05,
-      "loss": 0.3298,
+      "grad_norm": 2.3898696899414062,
+      "learning_rate": 4.769726742400983e-05,
+      "loss": 0.338,
       "step": 150
     },
     {
       "epoch": 0.05,
-      "grad_norm": 2.252187490463257,
-      "learning_rate": 4.728682170542636e-05,
-      "loss": 0.3048,
+      "grad_norm": 1.896946668624878,
+      "learning_rate": 4.731347866134479e-05,
+      "loss": 0.3142,
       "step": 175
     },
     {
       "epoch": 0.06,
-      "grad_norm": 0.42226940393447876,
-      "learning_rate": 4.6899224806201553e-05,
-      "loss": 0.2993,
+      "grad_norm": 0.35194501280784607,
+      "learning_rate": 4.692968989867977e-05,
+      "loss": 0.3061,
       "step": 200
     },
     {
       "epoch": 0.07,
-      "grad_norm": 0.5101619958877563,
-      "learning_rate": 4.651162790697675e-05,
+      "grad_norm": 0.6125898957252502,
+      "learning_rate": 4.654590113601474e-05,
       "loss": 0.3291,
       "step": 225
     },
     {
       "epoch": 0.08,
-      "grad_norm": 0.44378426671028137,
-      "learning_rate": 4.6124031007751936e-05,
-      "loss": 0.3254,
+      "grad_norm": 0.36999431252479553,
+      "learning_rate": 4.616211237334971e-05,
+      "loss": 0.3392,
       "step": 250
     },
     {
       "epoch": 0.09,
-      "grad_norm": 0.39292019605636597,
-      "learning_rate": 4.573643410852713e-05,
-      "loss": 0.2908,
+      "grad_norm": 0.5263044834136963,
+      "learning_rate": 4.5793675161191285e-05,
+      "loss": 0.3102,
       "step": 275
     },
     {
       "epoch": 0.09,
-      "grad_norm": 0.3389347791671753,
-      "learning_rate": 4.5348837209302326e-05,
-      "loss": 0.3058,
+      "grad_norm": 0.3138100802898407,
+      "learning_rate": 4.542523794903286e-05,
+      "loss": 0.3065,
       "step": 300
     },
     {
       "epoch": 0.1,
-      "grad_norm": 0.3901497423648834,
-      "learning_rate": 4.496124031007753e-05,
-      "loss": 0.343,
+      "grad_norm": 0.38833898305892944,
+      "learning_rate": 4.504144918636782e-05,
+      "loss": 0.3586,
       "step": 325
     },
     {
       "epoch": 0.11,
-      "grad_norm": 2.7542033195495605,
-      "learning_rate": 4.4573643410852715e-05,
-      "loss": 0.2969,
+      "grad_norm": 2.349430799484253,
+      "learning_rate": 4.4657660423702795e-05,
+      "loss": 0.3072,
       "step": 350
     },
     {
       "epoch": 0.12,
-      "grad_norm": 0.3395516872406006,
-      "learning_rate": 4.418604651162791e-05,
-      "loss": 0.2907,
+      "grad_norm": 0.3345746397972107,
+      "learning_rate": 4.427387166103777e-05,
+      "loss": 0.2855,
       "step": 375
     },
     {
       "epoch": 0.13,
-      "grad_norm": 0.3581525385379791,
-      "learning_rate": 4.3813953488372095e-05,
-      "loss": 0.2783,
+      "grad_norm": 0.46446025371551514,
+      "learning_rate": 4.390543444887934e-05,
+      "loss": 0.2852,
       "step": 400
     },
     {
       "epoch": 0.13,
-      "grad_norm": 0.6102267503738403,
-      "learning_rate": 4.342635658914729e-05,
-      "loss": 0.266,
+      "grad_norm": 0.6488687992095947,
+      "learning_rate": 4.3521645686214306e-05,
+      "loss": 0.2731,
       "step": 425
     },
     {
       "epoch": 0.14,
-      "grad_norm": 1.409443974494934,
-      "learning_rate": 4.3038759689922484e-05,
-      "loss": 0.2898,
+      "grad_norm": 0.7198542356491089,
+      "learning_rate": 4.313785692354928e-05,
+      "loss": 0.2986,
       "step": 450
     },
     {
       "epoch": 0.15,
-      "grad_norm": 2.3266422748565674,
-      "learning_rate": 4.265116279069767e-05,
-      "loss": 0.2967,
+      "grad_norm": 1.6638888120651245,
+      "learning_rate": 4.2754068160884256e-05,
+      "loss": 0.2985,
       "step": 475
     },
     {
       "epoch": 0.16,
-      "grad_norm": 1.4810105562210083,
-      "learning_rate": 4.2263565891472874e-05,
-      "loss": 0.2621,
+      "grad_norm": 1.4277877807617188,
+      "learning_rate": 4.237027939821922e-05,
+      "loss": 0.2559,
       "step": 500
     },
     {
       "epoch": 0.16,
-      "grad_norm": 3.7368390560150146,
-      "learning_rate": 4.187596899224807e-05,
-      "loss": 0.2766,
+      "grad_norm": 3.699174642562866,
+      "learning_rate": 4.198649063555419e-05,
+      "loss": 0.2731,
       "step": 525
     },
     {
       "epoch": 0.17,
-      "grad_norm": 1.7454067468643188,
-      "learning_rate": 4.1488372093023256e-05,
-      "loss": 0.2435,
+      "grad_norm": 1.9921191930770874,
+      "learning_rate": 4.160270187288917e-05,
+      "loss": 0.242,
       "step": 550
     },
     {
       "epoch": 0.18,
-      "grad_norm": 0.34005963802337646,
-      "learning_rate": 4.110077519379845e-05,
-      "loss": 0.2407,
+      "grad_norm": 0.3546009361743927,
+      "learning_rate": 4.1218913110224135e-05,
+      "loss": 0.253,
       "step": 575
     },
     {
       "epoch": 0.19,
-      "grad_norm": 1.942692518234253,
-      "learning_rate": 4.0713178294573646e-05,
-      "loss": 0.2556,
+      "grad_norm": 2.4208102226257324,
+      "learning_rate": 4.08351243475591e-05,
+      "loss": 0.2734,
       "step": 600
     },
     {
       "epoch": 0.2,
-      "grad_norm": 1.3751680850982666,
-      "learning_rate": 4.032558139534884e-05,
-      "loss": 0.2352,
+      "grad_norm": 1.4212377071380615,
+      "learning_rate": 4.045133558489408e-05,
+      "loss": 0.2426,
       "step": 625
     },
     {
       "epoch": 0.2,
-      "grad_norm": 1.911611795425415,
-      "learning_rate": 3.993798449612403e-05,
-      "loss": 0.2635,
+      "grad_norm": 3.6487390995025635,
+      "learning_rate": 4.0067546822229046e-05,
+      "loss": 0.2798,
       "step": 650
     },
     {
       "epoch": 0.21,
-      "grad_norm": 2.1268646717071533,
-      "learning_rate": 3.955038759689922e-05,
-      "loss": 0.2473,
+      "grad_norm": 2.001326322555542,
+      "learning_rate": 3.968375805956402e-05,
+      "loss": 0.2599,
       "step": 675
     },
     {
       "epoch": 0.22,
-      "grad_norm": 1.866083025932312,
-      "learning_rate": 3.9162790697674425e-05,
-      "loss": 0.238,
+      "grad_norm": 1.6598902940750122,
+      "learning_rate": 3.929996929689899e-05,
+      "loss": 0.2578,
       "step": 700
     },
     {
       "epoch": 0.23,
-      "grad_norm": 0.8285239338874817,
-      "learning_rate": 3.877519379844962e-05,
-      "loss": 0.2666,
+      "grad_norm": 1.3611884117126465,
+      "learning_rate": 3.8916180534233957e-05,
+      "loss": 0.2766,
       "step": 725
     },
     {
       "epoch": 0.23,
-      "grad_norm": 0.8000108003616333,
-      "learning_rate": 3.838759689922481e-05,
-      "loss": 0.2343,
+      "grad_norm": 0.8574317097663879,
+      "learning_rate": 3.853239177156893e-05,
+      "loss": 0.2417,
       "step": 750
     },
     {
       "epoch": 0.24,
-      "grad_norm": 0.6987760066986084,
-      "learning_rate": 3.8e-05,
-      "loss": 0.207,
+      "grad_norm": 0.8539307713508606,
+      "learning_rate": 3.81486030089039e-05,
+      "loss": 0.223,
       "step": 775
     },
     {
       "epoch": 0.25,
-      "grad_norm": 1.039829134941101,
-      "learning_rate": 3.76124031007752e-05,
-      "loss": 0.2469,
+      "grad_norm": 1.1913343667984009,
+      "learning_rate": 3.7764814246238874e-05,
+      "loss": 0.2602,
       "step": 800
     },
     {
       "epoch": 0.26,
-      "grad_norm": 1.2499618530273438,
-      "learning_rate": 3.722480620155039e-05,
-      "loss": 0.2589,
+      "grad_norm": 0.5884910821914673,
+      "learning_rate": 3.738102548357384e-05,
+      "loss": 0.2747,
       "step": 825
     },
     {
       "epoch": 0.27,
-      "grad_norm": 2.926412582397461,
-      "learning_rate": 3.683720930232558e-05,
-      "loss": 0.1872,
+      "grad_norm": 0.9413720965385437,
+      "learning_rate": 3.699723672090881e-05,
+      "loss": 0.1835,
       "step": 850
     },
     {
       "epoch": 0.27,
-      "grad_norm": 5.37480354309082,
-      "learning_rate": 3.6449612403100774e-05,
-      "loss": 0.2797,
+      "grad_norm": 2.626718521118164,
+      "learning_rate": 3.6613447958243785e-05,
+      "loss": 0.2883,
       "step": 875
     },
     {
       "epoch": 0.28,
-      "grad_norm": 1.7981369495391846,
-      "learning_rate": 3.6077519379844966e-05,
-      "loss": 0.2423,
+      "grad_norm": 1.573789119720459,
+      "learning_rate": 3.622965919557876e-05,
+      "loss": 0.252,
       "step": 900
     },
     {
       "epoch": 0.29,
-      "grad_norm": 1.1063029766082764,
-      "learning_rate": 3.568992248062016e-05,
-      "loss": 0.2528,
+      "grad_norm": 0.6372594833374023,
+      "learning_rate": 3.584587043291373e-05,
+      "loss": 0.2624,
       "step": 925
     },
     {
       "epoch": 0.3,
-      "grad_norm": 0.5154077410697937,
-      "learning_rate": 3.530232558139535e-05,
-      "loss": 0.1966,
+      "grad_norm": 0.4254266619682312,
+      "learning_rate": 3.5462081670248696e-05,
+      "loss": 0.2012,
       "step": 950
     },
     {
       "epoch": 0.31,
-      "grad_norm": 1.958801507949829,
-      "learning_rate": 3.4914728682170543e-05,
-      "loss": 0.2569,
+      "grad_norm": 1.2674840688705444,
+      "learning_rate": 3.5078292907583664e-05,
+      "loss": 0.2576,
       "step": 975
     },
     {
       "epoch": 0.31,
-      "grad_norm": 0.43896540999412537,
-      "learning_rate": 3.452713178294574e-05,
-      "loss": 0.2294,
+      "grad_norm": 0.6976171731948853,
+      "learning_rate": 3.469450414491864e-05,
+      "loss": 0.2374,
       "step": 1000
     },
     {
       "epoch": 0.32,
-      "grad_norm": 4.940056800842285,
-      "learning_rate": 3.413953488372093e-05,
-      "loss": 0.2138,
+      "grad_norm": 2.898336410522461,
+      "learning_rate": 3.4310715382253614e-05,
+      "loss": 0.2168,
       "step": 1025
     },
     {
       "epoch": 0.33,
-      "grad_norm": 2.0871469974517822,
-      "learning_rate": 3.375193798449612e-05,
-      "loss": 0.2486,
+      "grad_norm": 2.6241133213043213,
+      "learning_rate": 3.3926926619588575e-05,
+      "loss": 0.2721,
       "step": 1050
     },
     {
       "epoch": 0.34,
-      "grad_norm": 0.4600803554058075,
-      "learning_rate": 3.3364341085271316e-05,
-      "loss": 0.1784,
+      "grad_norm": 0.452421635389328,
+      "learning_rate": 3.354313785692355e-05,
+      "loss": 0.1887,
       "step": 1075
     },
     {
       "epoch": 0.34,
-      "grad_norm": 1.7933701276779175,
-      "learning_rate": 3.297674418604652e-05,
-      "loss": 0.2486,
+      "grad_norm": 4.101943492889404,
+      "learning_rate": 3.3159349094258525e-05,
+      "loss": 0.2597,
       "step": 1100
     },
     {
       "epoch": 0.35,
-      "grad_norm": 0.8704222440719604,
-      "learning_rate": 3.258914728682171e-05,
-      "loss": 0.1977,
+      "grad_norm": 0.8068587779998779,
+      "learning_rate": 3.277556033159349e-05,
+      "loss": 0.1803,
       "step": 1125
     },
     {
       "epoch": 0.36,
-      "grad_norm": 2.517425775527954,
-      "learning_rate": 3.22015503875969e-05,
-      "loss": 0.2128,
+      "grad_norm": 3.24800705909729,
+      "learning_rate": 3.239177156892847e-05,
+      "loss": 0.2175,
       "step": 1150
     },
     {
       "epoch": 0.37,
-      "grad_norm": 1.5435869693756104,
-      "learning_rate": 3.1813953488372094e-05,
-      "loss": 0.1889,
+      "grad_norm": 3.0881152153015137,
+      "learning_rate": 3.2007982806263435e-05,
+      "loss": 0.1883,
       "step": 1175
     },
     {
       "epoch": 0.38,
-      "grad_norm": 0.8101773262023926,
-      "learning_rate": 3.142635658914729e-05,
-      "loss": 0.2274,
+      "grad_norm": 0.5309311151504517,
+      "learning_rate": 3.1624194043598403e-05,
+      "loss": 0.243,
       "step": 1200
     },
     {
       "epoch": 0.38,
-      "grad_norm": 0.910406231880188,
-      "learning_rate": 3.1038759689922484e-05,
-      "loss": 0.1939,
+      "grad_norm": 0.8343615531921387,
+      "learning_rate": 3.124040528093338e-05,
+      "loss": 0.198,
       "step": 1225
     },
     {
       "epoch": 0.39,
-      "grad_norm": 1.1105294227600098,
-      "learning_rate": 3.065116279069767e-05,
-      "loss": 0.2712,
+      "grad_norm": 1.199294924736023,
+      "learning_rate": 3.0856616518268346e-05,
+      "loss": 0.2609,
       "step": 1250
     },
     {
       "epoch": 0.4,
-      "grad_norm": 1.6309148073196411,
-      "learning_rate": 3.0263565891472867e-05,
-      "loss": 0.2098,
+      "grad_norm": 1.7239773273468018,
+      "learning_rate": 3.0472827755603318e-05,
+      "loss": 0.2184,
       "step": 1275
     },
     {
       "epoch": 0.41,
-      "grad_norm": 0.6428541541099548,
-      "learning_rate": 2.9875968992248065e-05,
-      "loss": 0.1944,
+      "grad_norm": 0.6922329664230347,
+      "learning_rate": 3.008903899293829e-05,
+      "loss": 0.2026,
       "step": 1300
     },
     {
       "epoch": 0.41,
-      "grad_norm": 1.2899640798568726,
-      "learning_rate": 2.948837209302326e-05,
-      "loss": 0.232,
+      "grad_norm": 1.3839455842971802,
+      "learning_rate": 2.9705250230273257e-05,
+      "loss": 0.2365,
       "step": 1325
     },
     {
       "epoch": 0.42,
-      "grad_norm": 0.35119500756263733,
-      "learning_rate": 2.910077519379845e-05,
-      "loss": 0.2279,
+      "grad_norm": 0.3912523686885834,
+      "learning_rate": 2.932146146760823e-05,
+      "loss": 0.2228,
       "step": 1350
     },
     {
       "epoch": 0.43,
-      "grad_norm": 0.7236185669898987,
-      "learning_rate": 2.8713178294573645e-05,
-      "loss": 0.1759,
+      "grad_norm": 0.7774330973625183,
+      "learning_rate": 2.8937672704943203e-05,
+      "loss": 0.1836,
       "step": 1375
     },
     {
       "epoch": 0.44,
-      "grad_norm": 0.7780153155326843,
-      "learning_rate": 2.8325581395348837e-05,
-      "loss": 0.2006,
+      "grad_norm": 0.7025958895683289,
+      "learning_rate": 2.855388394227817e-05,
+      "loss": 0.2092,
       "step": 1400
     },
     {
       "epoch": 0.45,
-      "grad_norm": 2.4104530811309814,
-      "learning_rate": 2.793798449612403e-05,
-      "loss": 0.2128,
+      "grad_norm": 3.789003849029541,
+      "learning_rate": 2.8170095179613143e-05,
+      "loss": 0.217,
       "step": 1425
     },
     {
       "epoch": 0.45,
-      "grad_norm": 0.5463195443153381,
-      "learning_rate": 2.7550387596899223e-05,
-      "loss": 0.1777,
+      "grad_norm": 0.5001412034034729,
+      "learning_rate": 2.7786306416948114e-05,
+      "loss": 0.1771,
       "step": 1450
     },
     {
       "epoch": 0.46,
-      "grad_norm": 0.48027583956718445,
-      "learning_rate": 2.7162790697674418e-05,
-      "loss": 0.1887,
+      "grad_norm": 0.585847795009613,
+      "learning_rate": 2.7402517654283082e-05,
+      "loss": 0.1899,
       "step": 1475
     },
     {
       "epoch": 0.47,
-      "grad_norm": 1.2007440328598022,
-      "learning_rate": 2.6775193798449616e-05,
-      "loss": 0.2016,
+      "grad_norm": 1.6145225763320923,
+      "learning_rate": 2.7018728891618057e-05,
+      "loss": 0.214,
       "step": 1500
     },
     {
       "epoch": 0.48,
-      "grad_norm": 1.51267409324646,
-      "learning_rate": 2.638759689922481e-05,
-      "loss": 0.174,
+      "grad_norm": 2.724287748336792,
+      "learning_rate": 2.6634940128953022e-05,
+      "loss": 0.1857,
       "step": 1525
     },
     {
       "epoch": 0.49,
-      "grad_norm": 2.8868677616119385,
-      "learning_rate": 2.6000000000000002e-05,
+      "grad_norm": 1.1571438312530518,
+      "learning_rate": 2.6251151366287997e-05,
       "loss": 0.2046,
       "step": 1550
     },
     {
       "epoch": 0.49,
-      "grad_norm": 2.025702476501465,
-      "learning_rate": 2.5612403100775197e-05,
-      "loss": 0.1859,
+      "grad_norm": 3.0686943531036377,
+      "learning_rate": 2.5867362603622968e-05,
+      "loss": 0.1841,
       "step": 1575
     },
     {
       "epoch": 0.5,
-      "grad_norm": 1.341884732246399,
-      "learning_rate": 2.5224806201550388e-05,
-      "loss": 0.2007,
+      "grad_norm": 1.6015386581420898,
+      "learning_rate": 2.5483573840957936e-05,
+      "loss": 0.2095,
       "step": 1600
     },
     {
       "epoch": 0.51,
-      "grad_norm": 0.914476752281189,
-      "learning_rate": 2.4837209302325583e-05,
-      "loss": 0.1863,
+      "grad_norm": 1.5645371675491333,
+      "learning_rate": 2.5099785078292907e-05,
+      "loss": 0.193,
       "step": 1625
     },
     {
       "epoch": 0.52,
-      "grad_norm": 0.8603270649909973,
-      "learning_rate": 2.4465116279069768e-05,
-      "loss": 0.2002,
+      "grad_norm": 1.0582934617996216,
+      "learning_rate": 2.471599631562788e-05,
+      "loss": 0.2065,
       "step": 1650
     },
     {
       "epoch": 0.52,
-      "grad_norm": 3.7743961811065674,
-      "learning_rate": 2.4077519379844962e-05,
-      "loss": 0.1991,
+      "grad_norm": 4.154886722564697,
+      "learning_rate": 2.433220755296285e-05,
+      "loss": 0.1955,
       "step": 1675
     },
     {
       "epoch": 0.53,
-      "grad_norm": 2.3465445041656494,
-      "learning_rate": 2.3689922480620154e-05,
-      "loss": 0.1901,
+      "grad_norm": 2.047689437866211,
+      "learning_rate": 2.3948418790297822e-05,
+      "loss": 0.1999,
       "step": 1700
     },
     {
       "epoch": 0.54,
-      "grad_norm": 1.6707566976547241,
-      "learning_rate": 2.3302325581395352e-05,
-      "loss": 0.1805,
+      "grad_norm": 1.7541192770004272,
+      "learning_rate": 2.3564630027632793e-05,
+      "loss": 0.1906,
       "step": 1725
     },
     {
       "epoch": 0.55,
-      "grad_norm": 0.9617774486541748,
-      "learning_rate": 2.2914728682170543e-05,
-      "loss": 0.2044,
+      "grad_norm": 0.8929910063743591,
+      "learning_rate": 2.318084126496776e-05,
+      "loss": 0.2156,
       "step": 1750
     },
     {
       "epoch": 0.56,
-      "grad_norm": 1.740809440612793,
-      "learning_rate": 2.2527131782945738e-05,
-      "loss": 0.2502,
+      "grad_norm": 2.3634512424468994,
+      "learning_rate": 2.2797052502302736e-05,
+      "loss": 0.251,
       "step": 1775
     },
     {
       "epoch": 0.56,
-      "grad_norm": 1.6407257318496704,
-      "learning_rate": 2.213953488372093e-05,
-      "loss": 0.2071,
+      "grad_norm": 1.0741249322891235,
+      "learning_rate": 2.2413263739637704e-05,
+      "loss": 0.2221,
       "step": 1800
     },
     {
       "epoch": 0.57,
-      "grad_norm": 0.5815154314041138,
-      "learning_rate": 2.1751937984496127e-05,
-      "loss": 0.1962,
+      "grad_norm": 0.6083930134773254,
+      "learning_rate": 2.2029474976972675e-05,
+      "loss": 0.1922,
       "step": 1825
     },
     {
       "epoch": 0.58,
-      "grad_norm": 2.898057699203491,
-      "learning_rate": 2.136434108527132e-05,
-      "loss": 0.168,
+      "grad_norm": 3.046408176422119,
+      "learning_rate": 2.1645686214307644e-05,
+      "loss": 0.1634,
       "step": 1850
     },
     {
       "epoch": 0.59,
-      "grad_norm": 0.7204940319061279,
-      "learning_rate": 2.0976744186046513e-05,
-      "loss": 0.1808,
+      "grad_norm": 1.0440948009490967,
+      "learning_rate": 2.126189745164262e-05,
+      "loss": 0.1798,
       "step": 1875
     },
     {
       "epoch": 0.6,
-      "grad_norm": 1.5870317220687866,
-      "learning_rate": 2.0589147286821705e-05,
-      "loss": 0.1811,
+      "grad_norm": 2.302553176879883,
+      "learning_rate": 2.0878108688977586e-05,
+      "loss": 0.179,
       "step": 1900
     },
     {
       "epoch": 0.6,
-      "grad_norm": 0.8534879088401794,
-      "learning_rate": 2.0201550387596903e-05,
-      "loss": 0.2235,
+      "grad_norm": 0.5894582867622375,
+      "learning_rate": 2.0494319926312558e-05,
+      "loss": 0.2181,
       "step": 1925
     },
     {
       "epoch": 0.61,
-      "grad_norm": 1.5544407367706299,
-      "learning_rate": 1.9813953488372094e-05,
-      "loss": 0.192,
+      "grad_norm": 1.7287715673446655,
+      "learning_rate": 2.011053116364753e-05,
+      "loss": 0.1996,
       "step": 1950
     },
     {
       "epoch": 0.62,
-      "grad_norm": 2.284710645675659,
-      "learning_rate": 1.942635658914729e-05,
-      "loss": 0.2307,
+      "grad_norm": 2.5236544609069824,
+      "learning_rate": 1.97267424009825e-05,
+      "loss": 0.2366,
       "step": 1975
     },
     {
       "epoch": 0.63,
-      "grad_norm": 2.3412628173828125,
-      "learning_rate": 1.903875968992248e-05,
-      "loss": 0.1858,
+      "grad_norm": 1.8819808959960938,
+      "learning_rate": 1.9342953638317472e-05,
+      "loss": 0.1906,
       "step": 2000
     },
     {
       "epoch": 0.63,
-      "grad_norm": 0.5765611529350281,
-      "learning_rate": 1.8651162790697675e-05,
-      "loss": 0.1833,
+      "grad_norm": 0.5952481031417847,
+      "learning_rate": 1.895916487565244e-05,
+      "loss": 0.1749,
       "step": 2025
     },
     {
       "epoch": 0.64,
-      "grad_norm": 1.0757735967636108,
-      "learning_rate": 1.826356589147287e-05,
-      "loss": 0.209,
+      "grad_norm": 0.9825776815414429,
+      "learning_rate": 1.8575376112987415e-05,
+      "loss": 0.2179,
       "step": 2050
     },
     {
       "epoch": 0.65,
-      "grad_norm": 1.0615739822387695,
-      "learning_rate": 1.7875968992248064e-05,
-      "loss": 0.1591,
+      "grad_norm": 1.5762274265289307,
+      "learning_rate": 1.8191587350322383e-05,
+      "loss": 0.169,
       "step": 2075
     },
     {
       "epoch": 0.66,
-      "grad_norm": 0.5009509921073914,
-      "learning_rate": 1.7488372093023256e-05,
-      "loss": 0.1512,
+      "grad_norm": 1.433192253112793,
+      "learning_rate": 1.7807798587657354e-05,
+      "loss": 0.1677,
       "step": 2100
     },
     {
       "epoch": 0.67,
-      "grad_norm": 2.7069976329803467,
-      "learning_rate": 1.710077519379845e-05,
-      "loss": 0.2202,
+      "grad_norm": 2.5697052478790283,
+      "learning_rate": 1.7424009824992322e-05,
+      "loss": 0.2337,
       "step": 2125
     },
     {
       "epoch": 0.67,
-      "grad_norm": 1.9919098615646362,
-      "learning_rate": 1.6713178294573645e-05,
-      "loss": 0.2132,
+      "grad_norm": 1.336775541305542,
+      "learning_rate": 1.7040221062327297e-05,
+      "loss": 0.2168,
       "step": 2150
     },
     {
       "epoch": 0.68,
-      "grad_norm": 2.6684300899505615,
-      "learning_rate": 1.6325581395348837e-05,
+      "grad_norm": 2.746216058731079,
+      "learning_rate": 1.6656432299662265e-05,
       "loss": 0.1887,
       "step": 2175
     },
     {
       "epoch": 0.69,
-      "grad_norm": 2.843564987182617,
-      "learning_rate": 1.593798449612403e-05,
-      "loss": 0.2035,
+      "grad_norm": 2.6653735637664795,
+      "learning_rate": 1.6272643536997237e-05,
+      "loss": 0.2085,
       "step": 2200
     },
     {
       "epoch": 0.7,
-      "grad_norm": 2.938154697418213,
-      "learning_rate": 1.5550387596899226e-05,
-      "loss": 0.1946,
+      "grad_norm": 2.748147487640381,
+      "learning_rate": 1.5888854774332208e-05,
+      "loss": 0.1927,
       "step": 2225
     },
     {
       "epoch": 0.7,
-      "grad_norm": 2.365872859954834,
-      "learning_rate": 1.516279069767442e-05,
-      "loss": 0.159,
+      "grad_norm": 1.3852734565734863,
+      "learning_rate": 1.550506601166718e-05,
+      "loss": 0.1563,
       "step": 2250
     },
     {
       "epoch": 0.71,
-      "grad_norm": 2.8501169681549072,
-      "learning_rate": 1.4775193798449614e-05,
-      "loss": 0.1569,
+      "grad_norm": 3.5002951622009277,
+      "learning_rate": 1.512127724900215e-05,
+      "loss": 0.1662,
       "step": 2275
     },
     {
       "epoch": 0.72,
-      "grad_norm": 0.6475769877433777,
-      "learning_rate": 1.4387596899224807e-05,
-      "loss": 0.1626,
+      "grad_norm": 0.5453472137451172,
+      "learning_rate": 1.473748848633712e-05,
+      "loss": 0.1789,
       "step": 2300
     },
     {
       "epoch": 0.73,
-      "grad_norm": 3.125209093093872,
-      "learning_rate": 1.4000000000000001e-05,
-      "loss": 0.2249,
+      "grad_norm": 2.8527770042419434,
+      "learning_rate": 1.4353699723672092e-05,
+      "loss": 0.2245,
       "step": 2325
     },
     {
       "epoch": 0.74,
-      "grad_norm": 0.6424142718315125,
-      "learning_rate": 1.3612403100775195e-05,
-      "loss": 0.1583,
+      "grad_norm": 0.5989676713943481,
+      "learning_rate": 1.3969910961007064e-05,
+      "loss": 0.1695,
       "step": 2350
     },
     {
       "epoch": 0.74,
-      "grad_norm": 1.333817481994629,
-      "learning_rate": 1.3224806201550388e-05,
-      "loss": 0.1505,
+      "grad_norm": 0.6295841932296753,
+      "learning_rate": 1.3586122198342033e-05,
+      "loss": 0.1611,
       "step": 2375
     },
     {
       "epoch": 0.75,
-      "grad_norm": 0.30924907326698303,
-      "learning_rate": 1.283720930232558e-05,
-      "loss": 0.211,
+      "grad_norm": 0.4394790232181549,
+      "learning_rate": 1.3202333435677003e-05,
+      "loss": 0.2185,
       "step": 2400
     },
     {
       "epoch": 0.76,
-      "grad_norm": 0.9411177039146423,
-      "learning_rate": 1.2449612403100775e-05,
-      "loss": 0.1582,
+      "grad_norm": 0.9019514322280884,
+      "learning_rate": 1.2818544673011976e-05,
+      "loss": 0.1679,
       "step": 2425
     },
     {
       "epoch": 0.77,
-      "grad_norm": 0.6235800981521606,
-      "learning_rate": 1.206201550387597e-05,
-      "loss": 0.1762,
+      "grad_norm": 0.7013410329818726,
+      "learning_rate": 1.2434755910346946e-05,
+      "loss": 0.1752,
       "step": 2450
     },
     {
       "epoch": 0.78,
-      "grad_norm": 1.076593279838562,
-      "learning_rate": 1.1674418604651163e-05,
-      "loss": 0.2124,
+      "grad_norm": 0.7528170347213745,
+      "learning_rate": 1.2050967147681916e-05,
+      "loss": 0.1999,
       "step": 2475
     },
     {
       "epoch": 0.78,
-      "grad_norm": 1.0768756866455078,
-      "learning_rate": 1.1286821705426358e-05,
-      "loss": 0.2067,
+      "grad_norm": 1.6549068689346313,
+      "learning_rate": 1.1667178385016887e-05,
+      "loss": 0.1954,
       "step": 2500
     },
     {
       "epoch": 0.79,
-      "grad_norm": 1.3747373819351196,
-      "learning_rate": 1.089922480620155e-05,
-      "loss": 0.1959,
+      "grad_norm": 2.296861410140991,
+      "learning_rate": 1.1283389622351858e-05,
+      "loss": 0.1999,
       "step": 2525
     },
     {
       "epoch": 0.8,
-      "grad_norm": 3.307833194732666,
-      "learning_rate": 1.0511627906976746e-05,
-      "loss": 0.2249,
+      "grad_norm": 3.1202523708343506,
+      "learning_rate": 1.0899600859686828e-05,
+      "loss": 0.2417,
       "step": 2550
     },
     {
       "epoch": 0.81,
-      "grad_norm": 1.8046529293060303,
-      "learning_rate": 1.0124031007751939e-05,
-      "loss": 0.1566,
+      "grad_norm": 1.385256052017212,
+      "learning_rate": 1.05158120970218e-05,
+      "loss": 0.1598,
       "step": 2575
     },
     {
       "epoch": 0.81,
-      "grad_norm": 1.890415072441101,
-      "learning_rate": 9.736434108527132e-06,
-      "loss": 0.1621,
+      "grad_norm": 1.5936309099197388,
+      "learning_rate": 1.0132023334356771e-05,
+      "loss": 0.1629,
       "step": 2600
     },
     {
       "epoch": 0.82,
-      "grad_norm": 1.755682110786438,
-      "learning_rate": 9.348837209302326e-06,
-      "loss": 0.202,
+      "grad_norm": 1.5769296884536743,
+      "learning_rate": 9.748234571691742e-06,
+      "loss": 0.1823,
       "step": 2625
     },
     {
       "epoch": 0.83,
-      "grad_norm": 0.6612691879272461,
-      "learning_rate": 8.96124031007752e-06,
-      "loss": 0.1863,
+      "grad_norm": 1.9773831367492676,
+      "learning_rate": 9.364445809026712e-06,
+      "loss": 0.1851,
       "step": 2650
     },
     {
       "epoch": 0.84,
-      "grad_norm": 2.275036334991455,
-      "learning_rate": 8.573643410852714e-06,
-      "loss": 0.173,
+      "grad_norm": 2.044830799102783,
+      "learning_rate": 8.980657046361684e-06,
+      "loss": 0.1704,
       "step": 2675
     },
     {
       "epoch": 0.85,
-      "grad_norm": 0.7513935565948486,
-      "learning_rate": 8.186046511627907e-06,
-      "loss": 0.1884,
+      "grad_norm": 0.4414065480232239,
+      "learning_rate": 8.596868283696653e-06,
+      "loss": 0.1756,
       "step": 2700
     },
     {
       "epoch": 0.85,
-      "grad_norm": 1.9435423612594604,
-      "learning_rate": 7.798449612403102e-06,
-      "loss": 0.1795,
+      "grad_norm": 2.6521832942962646,
+      "learning_rate": 8.213079521031625e-06,
+      "loss": 0.1898,
       "step": 2725
     },
     {
       "epoch": 0.86,
-      "grad_norm": 0.5520192384719849,
-      "learning_rate": 7.410852713178294e-06,
-      "loss": 0.1908,
+      "grad_norm": 0.5525100231170654,
+      "learning_rate": 7.829290758366594e-06,
+      "loss": 0.192,
       "step": 2750
     },
     {
       "epoch": 0.87,
-      "grad_norm": 2.2986812591552734,
-      "learning_rate": 7.023255813953489e-06,
-      "loss": 0.1767,
+      "grad_norm": 2.2816734313964844,
+      "learning_rate": 7.445501995701566e-06,
+      "loss": 0.1852,
       "step": 2775
     },
     {
       "epoch": 0.88,
-      "grad_norm": 2.4153175354003906,
-      "learning_rate": 6.635658914728682e-06,
-      "loss": 0.2202,
+      "grad_norm": 1.4024970531463623,
+      "learning_rate": 7.061713233036537e-06,
+      "loss": 0.2279,
       "step": 2800
     },
     {
       "epoch": 0.88,
-      "grad_norm": 2.7443273067474365,
-      "learning_rate": 6.2480620155038765e-06,
-      "loss": 0.1961,
+      "grad_norm": 2.59187912940979,
+      "learning_rate": 6.677924470371508e-06,
+      "loss": 0.195,
       "step": 2825
     },
     {
       "epoch": 0.89,
-      "grad_norm": 5.525525093078613,
-      "learning_rate": 5.86046511627907e-06,
-      "loss": 0.1827,
+      "grad_norm": 4.647696018218994,
+      "learning_rate": 6.294135707706479e-06,
+      "loss": 0.1785,
       "step": 2850
     },
     {
       "epoch": 0.9,
-      "grad_norm": 2.7762510776519775,
-      "learning_rate": 5.472868217054264e-06,
-      "loss": 0.1816,
+      "grad_norm": 2.7766268253326416,
+      "learning_rate": 5.910346945041449e-06,
+      "loss": 0.188,
       "step": 2875
     },
     {
       "epoch": 0.91,
-      "grad_norm": 2.3923604488372803,
-      "learning_rate": 5.085271317829457e-06,
-      "loss": 0.1482,
+      "grad_norm": 2.672715425491333,
+      "learning_rate": 5.52655818237642e-06,
+      "loss": 0.1467,
       "step": 2900
     },
     {
       "epoch": 0.92,
-      "grad_norm": 0.8370789289474487,
-      "learning_rate": 4.697674418604651e-06,
-      "loss": 0.148,
+      "grad_norm": 0.6805928945541382,
+      "learning_rate": 5.142769419711391e-06,
+      "loss": 0.1561,
       "step": 2925
     },
     {
       "epoch": 0.92,
-      "grad_norm": 2.383211374282837,
-      "learning_rate": 4.310077519379845e-06,
-      "loss": 0.1772,
+      "grad_norm": 2.8108584880828857,
+      "learning_rate": 4.7589806570463625e-06,
+      "loss": 0.189,
       "step": 2950
     },
     {
       "epoch": 0.93,
-      "grad_norm": 0.5339698195457458,
-      "learning_rate": 3.922480620155039e-06,
-      "loss": 0.1409,
+      "grad_norm": 0.4544536769390106,
+      "learning_rate": 4.375191894381333e-06,
+      "loss": 0.1464,
       "step": 2975
     },
     {
       "epoch": 0.94,
-      "grad_norm": 0.5133029222488403,
-      "learning_rate": 3.534883720930233e-06,
-      "loss": 0.1532,
+      "grad_norm": 0.4506062865257263,
+      "learning_rate": 3.991403131716304e-06,
+      "loss": 0.1533,
       "step": 3000
     },
     {
       "epoch": 0.95,
-      "grad_norm": 3.2108466625213623,
-      "learning_rate": 3.1472868217054267e-06,
-      "loss": 0.1999,
+      "grad_norm": 1.9615803956985474,
+      "learning_rate": 3.607614369051274e-06,
+      "loss": 0.184,
       "step": 3025
     },
     {
       "epoch": 0.96,
-      "grad_norm": 3.6793041229248047,
-      "learning_rate": 2.7596899224806206e-06,
-      "loss": 0.1732,
+      "grad_norm": 3.426203489303589,
+      "learning_rate": 3.223825606386245e-06,
+      "loss": 0.1744,
       "step": 3050
     },
     {
       "epoch": 0.96,
-      "grad_norm": 2.0069997310638428,
-      "learning_rate": 2.372093023255814e-06,
+      "grad_norm": 2.0022315979003906,
+      "learning_rate": 2.8400368437212158e-06,
       "loss": 0.1839,
       "step": 3075
     },
     {
       "epoch": 0.97,
-      "grad_norm": 0.663833737373352,
-      "learning_rate": 1.984496124031008e-06,
-      "loss": 0.1928,
+      "grad_norm": 0.8066006302833557,
+      "learning_rate": 2.4562480810561868e-06,
+      "loss": 0.199,
       "step": 3100
     },
     {
       "epoch": 0.98,
-      "grad_norm": 3.452453851699829,
-      "learning_rate": 1.5968992248062018e-06,
-      "loss": 0.1304,
+      "grad_norm": 2.2845537662506104,
+      "learning_rate": 2.0724593183911578e-06,
+      "loss": 0.1361,
       "step": 3125
     },
     {
       "epoch": 0.99,
-      "grad_norm": 0.8049046397209167,
-      "learning_rate": 1.2093023255813954e-06,
-      "loss": 0.1972,
+      "grad_norm": 0.5564154982566833,
+      "learning_rate": 1.6886705557261286e-06,
+      "loss": 0.217,
       "step": 3150
     },
     {
       "epoch": 0.99,
-      "grad_norm": 0.5933910608291626,
-      "learning_rate": 8.217054263565891e-07,
-      "loss": 0.159,
+      "grad_norm": 0.6154067516326904,
+      "learning_rate": 1.3048817930610991e-06,
+      "loss": 0.1631,
       "step": 3175
     },
     {
       "epoch": 1.0,
-      "grad_norm": 0.493872731924057,
-      "learning_rate": 4.3410852713178294e-07,
-      "loss": 0.1212,
+      "grad_norm": 0.6170292496681213,
+      "learning_rate": 9.2109303039607e-07,
+      "loss": 0.1161,
       "step": 3200
     },
     {
       "epoch": 1.01,
-      "grad_norm": 1.8325016498565674,
-      "learning_rate": 4.651162790697675e-08,
-      "loss": 0.1036,
+      "grad_norm": 3.1419057846069336,
+      "learning_rate": 5.373042677310408e-07,
+      "loss": 0.106,
       "step": 3225
     },
     {
-      "epoch": 1.01,
-      "step": 3225,
-      "total_flos": 3.9986525216152166e+18,
-      "train_loss": 0.21891093194946762,
-      "train_runtime": 6427.7582,
-      "train_samples_per_second": 8.027,
-      "train_steps_per_second": 0.502
+      "epoch": 1.02,
+      "grad_norm": 0.4944257438182831,
+      "learning_rate": 1.5351550506601167e-07,
+      "loss": 0.1072,
+      "step": 3250
+    },
+    {
+      "epoch": 1.02,
+      "step": 3257,
+      "total_flos": 4.038329842708009e+18,
+      "train_loss": 0.2225519281828883,
+      "train_runtime": 2924.985,
+      "train_samples_per_second": 17.815,
+      "train_steps_per_second": 1.114
     }
   ],
   "logging_steps": 25,
-  "max_steps": 3225,
+  "max_steps": 3257,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 2,
   "save_steps": 100,
-  "total_flos": 3.9986525216152166e+18,
+  "total_flos": 4.038329842708009e+18,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null