End of training

README.md

@@ -4,7 +4,7 @@ base_model: BridgeTower/bridgetower-large-itm-mlm-itc
 tags:
 - generated_from_trainer
 datasets:
-- newyorker_caption_contest
+- jmhessel/newyorker_caption_contest
 model-index:
 - name: bridgetower
   results: []
@@ -15,7 +15,12 @@ should probably proofread and complete it, then remove this comment. -->
 
 # bridgetower
 
-This model is a fine-tuned version of [BridgeTower/bridgetower-large-itm-mlm-itc](https://huggingface.co/BridgeTower/bridgetower-large-itm-mlm-itc) on the newyorker_caption_contest dataset.
+This model is a fine-tuned version of [BridgeTower/bridgetower-large-itm-mlm-itc](https://huggingface.co/BridgeTower/bridgetower-large-itm-mlm-itc) on the jmhessel/newyorker_caption_contest matching dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.1243
+- Memory Allocated (gb): 51.27
+- Max Memory Allocated (gb): 60.52
+- Total Memory Available (gb): 94.62
 
 ## Model description
 

all_results.json

@@ -1,15 +1,15 @@
 {
     "epoch": 5.0,
-    "eval_loss": 0.1326669603586197,
-    "eval_runtime": 7.2012,
-    "eval_samples_per_second": 81.55,
-    "eval_steps_per_second": 5.097,
-    "max_memory_allocated (GB)": 57.18,
+    "eval_loss": 0.12431247532367706,
+    "eval_runtime": 8.9978,
+    "eval_samples_per_second": 65.301,
+    "eval_steps_per_second": 4.081,
+    "max_memory_allocated (GB)": 60.52,
     "memory_allocated (GB)": 51.27,
     "total_flos": 3.0598946525952e+16,
     "total_memory_available (GB)": 94.62,
-    "train_loss": 0.06081954878203723,
-    "train_runtime": 1056.8939,
-    "train_samples_per_second": 53.754,
-    "train_steps_per_second": 1.345
+    "train_loss": 0.06082177159737567,
+    "train_runtime": 1079.1005,
+    "train_samples_per_second": 52.128,
+    "train_steps_per_second": 1.304
 }

runs/Oct16_20-33-19_workload-ai-workshop/events.out.tfevents.1729112043.workload-ai-workshop.16989.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f5e47f19db166d9f51f5eb93aacd1eb69eddbf6ccdd0478eec9ad6f1192e5948
+size 998

test_results.json

@@ -1,10 +1,10 @@
 {
     "epoch": 5.0,
-    "eval_loss": 0.1326669603586197,
-    "eval_runtime": 7.2012,
-    "eval_samples_per_second": 81.55,
-    "eval_steps_per_second": 5.097,
-    "max_memory_allocated (GB)": 57.18,
+    "eval_loss": 0.12431247532367706,
+    "eval_runtime": 8.9978,
+    "eval_samples_per_second": 65.301,
+    "eval_steps_per_second": 4.081,
+    "max_memory_allocated (GB)": 60.52,
     "memory_allocated (GB)": 51.27,
     "total_memory_available (GB)": 94.62
 }

train_results.json

@@ -1,11 +1,11 @@
 {
     "epoch": 5.0,
-    "max_memory_allocated (GB)": 57.18,
+    "max_memory_allocated (GB)": 60.52,
     "memory_allocated (GB)": 50.57,
     "total_flos": 3.0598946525952e+16,
     "total_memory_available (GB)": 94.62,
-    "train_loss": 0.06081954878203723,
-    "train_runtime": 1056.8939,
-    "train_samples_per_second": 53.754,
-    "train_steps_per_second": 1.345
+    "train_loss": 0.06082177159737567,
+    "train_runtime": 1079.1005,
+    "train_samples_per_second": 52.128,
+    "train_steps_per_second": 1.304
 }

trainer_state.json

@@ -10,9 +10,9 @@
   "log_history": [
     {
       "epoch": 0.04081632653061224,
-      "grad_norm": 9.589848518371582,
+      "grad_norm": 16.731555938720703,
       "learning_rate": 9.918367346938776e-06,
-      "loss": 0.2612,
+      "loss": 0.2616,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 10,
@@ -20,9 +20,9 @@
     },
     {
       "epoch": 0.08163265306122448,
-      "grad_norm": 6.701302528381348,
+      "grad_norm": 7.408310890197754,
       "learning_rate": 9.836734693877552e-06,
-      "loss": 0.154,
+      "loss": 0.1553,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 20,
@@ -30,9 +30,9 @@
     },
     {
       "epoch": 0.12244897959183673,
-      "grad_norm": 5.337311267852783,
+      "grad_norm": 5.734114170074463,
       "learning_rate": 9.755102040816327e-06,
-      "loss": 0.1235,
+      "loss": 0.1248,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 30,
@@ -40,9 +40,9 @@
     },
     {
       "epoch": 0.16326530612244897,
-      "grad_norm": 4.59727144241333,
+      "grad_norm": 4.232840061187744,
       "learning_rate": 9.673469387755103e-06,
-      "loss": 0.1094,
+      "loss": 0.1106,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 40,
@@ -50,9 +50,9 @@
     },
     {
       "epoch": 0.20408163265306123,
-      "grad_norm": 4.675155162811279,
+      "grad_norm": 4.877067565917969,
       "learning_rate": 9.591836734693878e-06,
-      "loss": 0.1191,
+      "loss": 0.1233,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 50,
@@ -60,9 +60,9 @@
     },
     {
       "epoch": 0.24489795918367346,
-      "grad_norm": 2.7298097610473633,
+      "grad_norm": 2.0087852478027344,
       "learning_rate": 9.510204081632653e-06,
-      "loss": 0.08,
+      "loss": 0.0799,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 60,
@@ -70,7 +70,7 @@
     },
     {
       "epoch": 0.2857142857142857,
-      "grad_norm": 3.581591844558716,
+      "grad_norm": 4.425936698913574,
       "learning_rate": 9.42857142857143e-06,
       "loss": 0.1015,
       "max_memory_allocated (GB)": 57.18,
@@ -80,9 +80,9 @@
     },
     {
       "epoch": 0.32653061224489793,
-      "grad_norm": 1.5557887554168701,
+      "grad_norm": 1.8348758220672607,
       "learning_rate": 9.346938775510204e-06,
-      "loss": 0.1117,
+      "loss": 0.1094,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 80,
@@ -90,9 +90,9 @@
     },
     {
       "epoch": 0.3673469387755102,
-      "grad_norm": 8.605263710021973,
+      "grad_norm": 5.009132385253906,
       "learning_rate": 9.26530612244898e-06,
-      "loss": 0.0872,
+      "loss": 0.0837,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 90,
@@ -100,9 +100,9 @@
     },
     {
       "epoch": 0.40816326530612246,
-      "grad_norm": 3.075289249420166,
+      "grad_norm": 6.888875484466553,
       "learning_rate": 9.183673469387756e-06,
-      "loss": 0.076,
+      "loss": 0.075,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 100,
@@ -110,9 +110,9 @@
     },
     {
       "epoch": 0.4489795918367347,
-      "grad_norm": 3.8531429767608643,
+      "grad_norm": 3.038123369216919,
       "learning_rate": 9.102040816326532e-06,
-      "loss": 0.0726,
+      "loss": 0.0709,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 110,
@@ -120,9 +120,9 @@
     },
     {
       "epoch": 0.4897959183673469,
-      "grad_norm": 3.202249765396118,
+      "grad_norm": 2.6441116333007812,
       "learning_rate": 9.020408163265307e-06,
-      "loss": 0.0708,
+      "loss": 0.0706,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 120,
@@ -130,9 +130,9 @@
     },
     {
       "epoch": 0.5306122448979592,
-      "grad_norm": 1.818903923034668,
+      "grad_norm": 3.5273971557617188,
       "learning_rate": 8.938775510204082e-06,
-      "loss": 0.059,
+      "loss": 0.0597,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 130,
@@ -140,9 +140,9 @@
     },
     {
       "epoch": 0.5714285714285714,
-      "grad_norm": 1.1054496765136719,
+      "grad_norm": 1.2711573839187622,
       "learning_rate": 8.857142857142858e-06,
-      "loss": 0.0487,
+      "loss": 0.0499,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 140,
@@ -150,9 +150,9 @@
     },
     {
       "epoch": 0.6122448979591837,
-      "grad_norm": 2.9758386611938477,
+      "grad_norm": 2.2461705207824707,
       "learning_rate": 8.775510204081633e-06,
-      "loss": 0.0629,
+      "loss": 0.0636,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 150,
@@ -160,9 +160,9 @@
     },
     {
       "epoch": 0.6530612244897959,
-      "grad_norm": 1.993240237236023,
+      "grad_norm": 2.0061120986938477,
       "learning_rate": 8.69387755102041e-06,
-      "loss": 0.0713,
+      "loss": 0.0708,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 160,
@@ -170,9 +170,9 @@
     },
     {
       "epoch": 0.6938775510204082,
-      "grad_norm": 3.4001684188842773,
+      "grad_norm": 1.7886881828308105,
       "learning_rate": 8.612244897959184e-06,
-      "loss": 0.0748,
+      "loss": 0.0753,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 170,
@@ -180,9 +180,9 @@
     },
     {
       "epoch": 0.7346938775510204,
-      "grad_norm": 1.881607174873352,
+      "grad_norm": 1.8707109689712524,
       "learning_rate": 8.530612244897961e-06,
-      "loss": 0.0603,
+      "loss": 0.0612,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 180,
@@ -190,9 +190,9 @@
     },
     {
       "epoch": 0.7755102040816326,
-      "grad_norm": 3.133998155593872,
+      "grad_norm": 2.3418948650360107,
       "learning_rate": 8.448979591836736e-06,
-      "loss": 0.0592,
+      "loss": 0.0586,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 190,
@@ -200,9 +200,9 @@
     },
     {
       "epoch": 0.8163265306122449,
-      "grad_norm": 0.934773325920105,
+      "grad_norm": 0.7759063839912415,
       "learning_rate": 8.36734693877551e-06,
-      "loss": 0.0626,
+      "loss": 0.063,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 200,
@@ -210,9 +210,9 @@
     },
     {
       "epoch": 0.8571428571428571,
-      "grad_norm": 0.8929204940795898,
+      "grad_norm": 0.9075489044189453,
       "learning_rate": 8.285714285714287e-06,
-      "loss": 0.0648,
+      "loss": 0.063,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 210,
@@ -220,9 +220,9 @@
     },
     {
       "epoch": 0.8979591836734694,
-      "grad_norm": 1.0954375267028809,
+      "grad_norm": 1.0522433519363403,
       "learning_rate": 8.204081632653062e-06,
-      "loss": 0.0596,
+      "loss": 0.0605,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 220,
@@ -230,9 +230,9 @@
     },
     {
       "epoch": 0.9387755102040817,
-      "grad_norm": 0.8736925721168518,
+      "grad_norm": 0.7449356913566589,
       "learning_rate": 8.122448979591837e-06,
-      "loss": 0.0567,
+      "loss": 0.0564,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 230,
@@ -240,9 +240,9 @@
     },
     {
       "epoch": 0.9795918367346939,
-      "grad_norm": 0.6469990611076355,
+      "grad_norm": 0.878478467464447,
       "learning_rate": 8.040816326530613e-06,
-      "loss": 0.0501,
+      "loss": 0.0496,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 240,
@@ -250,9 +250,9 @@
     },
     {
       "epoch": 1.0204081632653061,
-      "grad_norm": 0.7664241790771484,
+      "grad_norm": 1.1043357849121094,
       "learning_rate": 7.959183673469388e-06,
-      "loss": 0.06,
+      "loss": 0.0593,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 250,
@@ -260,9 +260,9 @@
     },
     {
       "epoch": 1.0612244897959184,
-      "grad_norm": 3.3073790073394775,
+      "grad_norm": 1.5091803073883057,
       "learning_rate": 7.877551020408164e-06,
-      "loss": 0.0538,
+      "loss": 0.0542,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 260,
@@ -270,9 +270,9 @@
     },
     {
       "epoch": 1.1020408163265305,
-      "grad_norm": 0.7618402242660522,
+      "grad_norm": 0.7768255472183228,
       "learning_rate": 7.79591836734694e-06,
-      "loss": 0.0621,
+      "loss": 0.0636,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 270,
@@ -280,9 +280,9 @@
     },
     {
       "epoch": 1.1428571428571428,
-      "grad_norm": 2.667276620864868,
+      "grad_norm": 1.0384677648544312,
       "learning_rate": 7.714285714285716e-06,
-      "loss": 0.0578,
+      "loss": 0.0585,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 280,
@@ -290,9 +290,9 @@
     },
     {
       "epoch": 1.183673469387755,
-      "grad_norm": 0.7617210149765015,
+      "grad_norm": 1.9693238735198975,
       "learning_rate": 7.63265306122449e-06,
-      "loss": 0.0505,
+      "loss": 0.0512,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 290,
@@ -300,9 +300,9 @@
     },
     {
       "epoch": 1.2244897959183674,
-      "grad_norm": 1.4268369674682617,
+      "grad_norm": 1.0379549264907837,
       "learning_rate": 7.551020408163265e-06,
-      "loss": 0.0638,
+      "loss": 0.0647,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 300,
@@ -310,7 +310,7 @@
     },
     {
       "epoch": 1.2653061224489797,
-      "grad_norm": 1.0776606798171997,
+      "grad_norm": 1.01348078250885,
       "learning_rate": 7.469387755102041e-06,
       "loss": 0.0668,
       "max_memory_allocated (GB)": 57.18,
@@ -320,9 +320,9 @@
     },
     {
       "epoch": 1.306122448979592,
-      "grad_norm": 1.2839634418487549,
+      "grad_norm": 0.9114493727684021,
       "learning_rate": 7.387755102040817e-06,
-      "loss": 0.049,
+      "loss": 0.0509,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 320,
@@ -330,9 +330,9 @@
     },
     {
       "epoch": 1.346938775510204,
-      "grad_norm": 0.6254063844680786,
+      "grad_norm": 0.5102170705795288,
       "learning_rate": 7.306122448979592e-06,
-      "loss": 0.06,
+      "loss": 0.0612,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 330,
@@ -340,7 +340,7 @@
     },
     {
       "epoch": 1.3877551020408163,
-      "grad_norm": 1.5291829109191895,
+      "grad_norm": 1.097835659980774,
       "learning_rate": 7.224489795918368e-06,
       "loss": 0.055,
       "max_memory_allocated (GB)": 57.18,
@@ -350,9 +350,9 @@
     },
     {
       "epoch": 1.4285714285714286,
-      "grad_norm": 0.6570836305618286,
+      "grad_norm": 0.5733208656311035,
       "learning_rate": 7.1428571428571436e-06,
-      "loss": 0.0488,
+      "loss": 0.05,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 350,
@@ -360,9 +360,9 @@
     },
     {
       "epoch": 1.469387755102041,
-      "grad_norm": 4.310382843017578,
+      "grad_norm": 2.428065538406372,
       "learning_rate": 7.061224489795919e-06,
-      "loss": 0.0464,
+      "loss": 0.0458,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 360,
@@ -370,9 +370,9 @@
     },
     {
       "epoch": 1.510204081632653,
-      "grad_norm": 1.3656598329544067,
+      "grad_norm": 0.9827563166618347,
       "learning_rate": 6.979591836734695e-06,
-      "loss": 0.0585,
+      "loss": 0.0588,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 370,
@@ -380,9 +380,9 @@
     },
     {
       "epoch": 1.5510204081632653,
-      "grad_norm": 0.9290764927864075,
+      "grad_norm": 0.5563956499099731,
       "learning_rate": 6.8979591836734705e-06,
-      "loss": 0.0468,
+      "loss": 0.0473,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 380,
@@ -390,9 +390,9 @@
     },
     {
       "epoch": 1.5918367346938775,
-      "grad_norm": 1.746748447418213,
+      "grad_norm": 0.9855480194091797,
       "learning_rate": 6.816326530612245e-06,
-      "loss": 0.0648,
+      "loss": 0.0653,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 390,
@@ -400,9 +400,9 @@
     },
     {
       "epoch": 1.6326530612244898,
-      "grad_norm": 0.5461153388023376,
+      "grad_norm": 0.7717217803001404,
       "learning_rate": 6.734693877551021e-06,
-      "loss": 0.0613,
+      "loss": 0.0618,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 400,
@@ -410,9 +410,9 @@
     },
     {
       "epoch": 1.6734693877551021,
-      "grad_norm": 0.4624268710613251,
+      "grad_norm": 0.5391837358474731,
       "learning_rate": 6.653061224489797e-06,
-      "loss": 0.0604,
+      "loss": 0.061,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 410,
@@ -420,9 +420,9 @@
     },
     {
       "epoch": 1.7142857142857144,
-      "grad_norm": 4.135178089141846,
+      "grad_norm": 1.2298884391784668,
       "learning_rate": 6.571428571428572e-06,
-      "loss": 0.053,
+      "loss": 0.0534,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 420,
@@ -430,9 +430,9 @@
     },
     {
       "epoch": 1.7551020408163265,
-      "grad_norm": 0.19531233608722687,
+      "grad_norm": 0.22150574624538422,
       "learning_rate": 6.489795918367348e-06,
-      "loss": 0.0675,
+      "loss": 0.0681,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 430,
@@ -440,7 +440,7 @@
     },
     {
       "epoch": 1.7959183673469388,
-      "grad_norm": 1.667291522026062,
+      "grad_norm": 0.8992190957069397,
       "learning_rate": 6.408163265306124e-06,
       "loss": 0.0606,
       "max_memory_allocated (GB)": 57.18,
@@ -450,9 +450,9 @@
     },
     {
       "epoch": 1.836734693877551,
-      "grad_norm": 0.4277726709842682,
+      "grad_norm": 0.9748144149780273,
       "learning_rate": 6.326530612244899e-06,
-      "loss": 0.044,
+      "loss": 0.0436,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 450,
@@ -460,9 +460,9 @@
     },
     {
       "epoch": 1.8775510204081631,
-      "grad_norm": 0.3809329569339752,
+      "grad_norm": 0.3272276520729065,
       "learning_rate": 6.244897959183675e-06,
-      "loss": 0.0537,
+      "loss": 0.0545,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 460,
@@ -470,7 +470,7 @@
     },
     {
       "epoch": 1.9183673469387754,
-      "grad_norm": 1.0573110580444336,
+      "grad_norm": 0.7515161633491516,
       "learning_rate": 6.163265306122449e-06,
       "loss": 0.0427,
       "max_memory_allocated (GB)": 57.18,
@@ -480,9 +480,9 @@
     },
     {
       "epoch": 1.9591836734693877,
-      "grad_norm": 0.20000681281089783,
+      "grad_norm": 0.35405218601226807,
       "learning_rate": 6.0816326530612245e-06,
-      "loss": 0.0502,
+      "loss": 0.0501,
       "max_memory_allocated (GB)": 57.18,
       "memory_allocated (GB)": 50.57,
       "step": 480,
@@ -490,755 +490,755 @@
     },
     {
       "epoch": 2.0,
-      "grad_norm": 0.9971123337745667,
+      "grad_norm": 0.9796536564826965,
       "learning_rate": 6e-06,
-      "loss": 0.0636,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0641,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 490,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.0408163265306123,
-      "grad_norm": 0.673917293548584,
+      "grad_norm": 0.639869213104248,
       "learning_rate": 5.918367346938776e-06,
-      "loss": 0.0591,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0592,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 500,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.0816326530612246,
-      "grad_norm": 0.6252917647361755,
+      "grad_norm": 0.8377614617347717,
       "learning_rate": 5.8367346938775515e-06,
-      "loss": 0.0485,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0497,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 510,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.122448979591837,
-      "grad_norm": 1.9926584959030151,
+      "grad_norm": 1.07124662399292,
       "learning_rate": 5.755102040816327e-06,
-      "loss": 0.0461,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.047,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 520,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.163265306122449,
-      "grad_norm": 1.1171493530273438,
+      "grad_norm": 1.0354053974151611,
       "learning_rate": 5.673469387755103e-06,
-      "loss": 0.063,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0636,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 530,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.204081632653061,
-      "grad_norm": 0.5303067564964294,
+      "grad_norm": 0.5396034717559814,
       "learning_rate": 5.591836734693878e-06,
-      "loss": 0.0594,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0602,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 540,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.2448979591836733,
-      "grad_norm": 1.5552964210510254,
+      "grad_norm": 1.1461529731750488,
       "learning_rate": 5.510204081632653e-06,
-      "loss": 0.0632,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0629,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 550,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.2857142857142856,
-      "grad_norm": 7.580259799957275,
+      "grad_norm": 1.732596516609192,
       "learning_rate": 5.428571428571429e-06,
-      "loss": 0.058,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0575,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 560,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.326530612244898,
-      "grad_norm": 0.8162185549736023,
+      "grad_norm": 0.6468518376350403,
       "learning_rate": 5.3469387755102045e-06,
-      "loss": 0.0585,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0581,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 570,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.36734693877551,
-      "grad_norm": 3.4555113315582275,
+      "grad_norm": 1.7525995969772339,
       "learning_rate": 5.26530612244898e-06,
-      "loss": 0.067,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0678,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 580,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.4081632653061225,
-      "grad_norm": 0.45223644375801086,
+      "grad_norm": 0.5057170391082764,
       "learning_rate": 5.183673469387756e-06,
-      "loss": 0.0537,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0533,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 590,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.4489795918367347,
-      "grad_norm": 0.9515520930290222,
+      "grad_norm": 1.2042871713638306,
       "learning_rate": 5.1020408163265315e-06,
-      "loss": 0.0435,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0444,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 600,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.489795918367347,
-      "grad_norm": 0.5981607437133789,
+      "grad_norm": 0.20784775912761688,
       "learning_rate": 5.020408163265307e-06,
-      "loss": 0.0414,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0413,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 610,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.5306122448979593,
-      "grad_norm": 0.8998698592185974,
+      "grad_norm": 5.924367427825928,
       "learning_rate": 4.938775510204082e-06,
-      "loss": 0.0432,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0424,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 620,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.571428571428571,
-      "grad_norm": 0.5227445960044861,
+      "grad_norm": 0.40434470772743225,
       "learning_rate": 4.857142857142858e-06,
-      "loss": 0.0813,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0815,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 630,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.612244897959184,
-      "grad_norm": 1.117499828338623,
+      "grad_norm": 0.7878668904304504,
       "learning_rate": 4.775510204081633e-06,
-      "loss": 0.0492,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0495,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 640,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.6530612244897958,
-      "grad_norm": 0.6399075984954834,
+      "grad_norm": 0.5937454700469971,
       "learning_rate": 4.693877551020409e-06,
-      "loss": 0.0472,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0471,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 650,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.693877551020408,
-      "grad_norm": 3.8376593589782715,
+      "grad_norm": 0.9869357347488403,
       "learning_rate": 4.612244897959184e-06,
-      "loss": 0.0473,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0472,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 660,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.7346938775510203,
-      "grad_norm": 0.6460541486740112,
+      "grad_norm": 0.7606731653213501,
       "learning_rate": 4.530612244897959e-06,
-      "loss": 0.0872,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0864,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 670,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.7755102040816326,
-      "grad_norm": 1.897635579109192,
+      "grad_norm": 1.5359723567962646,
       "learning_rate": 4.448979591836735e-06,
-      "loss": 0.0487,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0482,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 680,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.816326530612245,
-      "grad_norm": 0.3259105682373047,
+      "grad_norm": 0.370749294757843,
       "learning_rate": 4.367346938775511e-06,
-      "loss": 0.0449,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0447,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 690,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.857142857142857,
-      "grad_norm": 0.3581603467464447,
+      "grad_norm": 0.633791983127594,
       "learning_rate": 4.2857142857142855e-06,
-      "loss": 0.0435,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.044,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 700,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.8979591836734695,
-      "grad_norm": 0.8618249297142029,
+      "grad_norm": 1.1335562467575073,
       "learning_rate": 4.204081632653061e-06,
-      "loss": 0.0549,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0552,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 710,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.938775510204082,
-      "grad_norm": 1.4629887342453003,
+      "grad_norm": 1.5698113441467285,
       "learning_rate": 4.122448979591837e-06,
-      "loss": 0.0584,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0576,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 720,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 2.979591836734694,
-      "grad_norm": 0.21725626289844513,
+      "grad_norm": 0.20579344034194946,
       "learning_rate": 4.040816326530612e-06,
-      "loss": 0.034,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0336,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 730,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.020408163265306,
-      "grad_norm": 1.257843017578125,
+      "grad_norm": 1.3172132968902588,
       "learning_rate": 3.959183673469388e-06,
-      "loss": 0.054,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0534,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 740,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.061224489795918,
-      "grad_norm": 0.7538784742355347,
+      "grad_norm": 1.5193818807601929,
       "learning_rate": 3.877551020408164e-06,
-      "loss": 0.0611,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0605,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 750,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.1020408163265305,
-      "grad_norm": 0.6883149147033691,
+      "grad_norm": 0.7448021769523621,
       "learning_rate": 3.795918367346939e-06,
-      "loss": 0.0523,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0524,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 760,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.142857142857143,
-      "grad_norm": 0.6788195967674255,
+      "grad_norm": 0.6830514073371887,
       "learning_rate": 3.7142857142857146e-06,
-      "loss": 0.0685,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0687,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 770,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.183673469387755,
-      "grad_norm": 0.5011599063873291,
+      "grad_norm": 0.4188525378704071,
       "learning_rate": 3.6326530612244903e-06,
-      "loss": 0.0643,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0646,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 780,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.2244897959183674,
-      "grad_norm": 0.4473969042301178,
+      "grad_norm": 0.562319278717041,
       "learning_rate": 3.5510204081632655e-06,
-      "loss": 0.0525,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0523,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 790,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.2653061224489797,
-      "grad_norm": 0.5276008248329163,
+      "grad_norm": 0.383428156375885,
       "learning_rate": 3.469387755102041e-06,
-      "loss": 0.0422,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0425,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 800,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.306122448979592,
-      "grad_norm": 0.8036576509475708,
+      "grad_norm": 0.6743464469909668,
       "learning_rate": 3.3877551020408168e-06,
-      "loss": 0.0553,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0546,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 810,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.3469387755102042,
-      "grad_norm": 0.7645472884178162,
+      "grad_norm": 0.7104524970054626,
       "learning_rate": 3.3061224489795924e-06,
-      "loss": 0.0457,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0448,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 820,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.387755102040816,
-      "grad_norm": 0.27847281098365784,
+      "grad_norm": 0.5083824396133423,
       "learning_rate": 3.2244897959183672e-06,
-      "loss": 0.0492,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.048,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 830,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.4285714285714284,
-      "grad_norm": 0.3790993094444275,
+      "grad_norm": 0.49159926176071167,
       "learning_rate": 3.142857142857143e-06,
-      "loss": 0.0529,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0521,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 840,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.4693877551020407,
-      "grad_norm": 0.4730982184410095,
+      "grad_norm": 0.635971188545227,
       "learning_rate": 3.0612244897959185e-06,
-      "loss": 0.0828,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.081,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 850,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.510204081632653,
-      "grad_norm": 1.0951735973358154,
+      "grad_norm": 5.446665287017822,
       "learning_rate": 2.979591836734694e-06,
-      "loss": 0.0453,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0449,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 860,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.5510204081632653,
-      "grad_norm": 0.46890848875045776,
+      "grad_norm": 0.3760354816913605,
       "learning_rate": 2.8979591836734694e-06,
-      "loss": 0.0643,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.064,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 870,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.5918367346938775,
-      "grad_norm": 0.9014816880226135,
+      "grad_norm": 1.0110430717468262,
       "learning_rate": 2.816326530612245e-06,
-      "loss": 0.0439,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0433,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 880,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.63265306122449,
-      "grad_norm": 0.1719776690006256,
+      "grad_norm": 0.1704891324043274,
       "learning_rate": 2.7346938775510207e-06,
-      "loss": 0.0288,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0282,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 890,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.673469387755102,
-      "grad_norm": 0.6494420170783997,
+      "grad_norm": 0.8289852142333984,
       "learning_rate": 2.6530612244897964e-06,
-      "loss": 0.0515,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0518,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 900,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.7142857142857144,
-      "grad_norm": 0.4721522629261017,
+      "grad_norm": 0.812251091003418,
       "learning_rate": 2.571428571428571e-06,
-      "loss": 0.0604,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0598,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 910,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.7551020408163263,
-      "grad_norm": 0.13098274171352386,
+      "grad_norm": 0.14789150655269623,
       "learning_rate": 2.489795918367347e-06,
-      "loss": 0.055,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0542,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 920,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.795918367346939,
-      "grad_norm": 0.39990562200546265,
+      "grad_norm": 0.8937461972236633,
       "learning_rate": 2.4081632653061225e-06,
-      "loss": 0.0501,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0499,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 930,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.836734693877551,
-      "grad_norm": 1.020967721939087,
+      "grad_norm": 6.832267761230469,
       "learning_rate": 2.326530612244898e-06,
       "loss": 0.0489,
-      "max_memory_allocated (GB)": 57.18,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 940,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.877551020408163,
-      "grad_norm": 0.6601265072822571,
+      "grad_norm": 0.7444981336593628,
       "learning_rate": 2.244897959183674e-06,
-      "loss": 0.0594,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0591,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 950,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.9183673469387754,
-      "grad_norm": 0.34864309430122375,
+      "grad_norm": 0.38545939326286316,
       "learning_rate": 2.1632653061224495e-06,
-      "loss": 0.0495,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0499,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 960,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 3.9591836734693877,
-      "grad_norm": 1.2881830930709839,
+      "grad_norm": 1.0209099054336548,
       "learning_rate": 2.0816326530612247e-06,
-      "loss": 0.0742,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0736,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 970,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.0,
-      "grad_norm": 0.7139947414398193,
+      "grad_norm": 0.6384561657905579,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.0698,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0705,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 980,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.040816326530612,
-      "grad_norm": 0.7273740172386169,
+      "grad_norm": 0.6284443736076355,
       "learning_rate": 1.9183673469387756e-06,
       "loss": 0.0483,
-      "max_memory_allocated (GB)": 57.18,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 990,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.081632653061225,
-      "grad_norm": 0.811747133731842,
+      "grad_norm": 1.0421397686004639,
       "learning_rate": 1.8367346938775512e-06,
-      "loss": 0.0623,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0627,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1000,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.122448979591836,
-      "grad_norm": 0.595486044883728,
+      "grad_norm": 0.6968585252761841,
       "learning_rate": 1.7551020408163267e-06,
-      "loss": 0.0627,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0632,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1010,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.163265306122449,
-      "grad_norm": 0.40334606170654297,
+      "grad_norm": 0.35346081852912903,
       "learning_rate": 1.6734693877551023e-06,
-      "loss": 0.0404,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0402,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1020,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.204081632653061,
-      "grad_norm": 0.5992821455001831,
+      "grad_norm": 0.8111013174057007,
       "learning_rate": 1.5918367346938775e-06,
-      "loss": 0.0529,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0528,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1030,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.244897959183674,
-      "grad_norm": 0.5139011740684509,
+      "grad_norm": 0.5146018862724304,
       "learning_rate": 1.5102040816326532e-06,
-      "loss": 0.04,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0396,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1040,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.285714285714286,
-      "grad_norm": 0.35282936692237854,
+      "grad_norm": 0.6711630821228027,
       "learning_rate": 1.4285714285714286e-06,
-      "loss": 0.0501,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0493,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1050,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.326530612244898,
-      "grad_norm": 0.32772839069366455,
+      "grad_norm": 0.3221682012081146,
       "learning_rate": 1.3469387755102043e-06,
-      "loss": 0.0456,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0454,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1060,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.36734693877551,
-      "grad_norm": 0.6208099126815796,
+      "grad_norm": 0.9966460466384888,
       "learning_rate": 1.2653061224489795e-06,
-      "loss": 0.0486,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0485,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1070,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.408163265306122,
-      "grad_norm": 0.7009238004684448,
+      "grad_norm": 0.45073866844177246,
       "learning_rate": 1.1836734693877552e-06,
-      "loss": 0.0521,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.052,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1080,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.448979591836735,
-      "grad_norm": 0.3578231930732727,
+      "grad_norm": 0.7129877805709839,
       "learning_rate": 1.1020408163265308e-06,
-      "loss": 0.0509,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0511,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1090,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.489795918367347,
-      "grad_norm": 0.30579662322998047,
+      "grad_norm": 0.333344966173172,
       "learning_rate": 1.0204081632653063e-06,
-      "loss": 0.0365,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0352,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1100,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.530612244897959,
-      "grad_norm": 0.10309538245201111,
+      "grad_norm": 0.11522499471902847,
       "learning_rate": 9.387755102040817e-07,
-      "loss": 0.0462,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0457,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1110,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.571428571428571,
-      "grad_norm": 0.6286255121231079,
+      "grad_norm": 0.3113747239112854,
       "learning_rate": 8.571428571428572e-07,
-      "loss": 0.04,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0402,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1120,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.612244897959184,
-      "grad_norm": 0.6702510714530945,
+      "grad_norm": 0.8939142823219299,
       "learning_rate": 7.755102040816327e-07,
-      "loss": 0.0537,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0536,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1130,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.653061224489796,
-      "grad_norm": 7.299179553985596,
+      "grad_norm": 0.7175144553184509,
       "learning_rate": 6.938775510204082e-07,
-      "loss": 0.0725,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0724,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1140,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.6938775510204085,
-      "grad_norm": 0.8366469144821167,
+      "grad_norm": 0.6586973071098328,
       "learning_rate": 6.122448979591837e-07,
-      "loss": 0.0515,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0514,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1150,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.73469387755102,
-      "grad_norm": 0.7851714491844177,
+      "grad_norm": 1.190521478652954,
       "learning_rate": 5.306122448979592e-07,
-      "loss": 0.061,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0608,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1160,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.775510204081632,
-      "grad_norm": 0.5544641613960266,
+      "grad_norm": 0.47001194953918457,
       "learning_rate": 4.489795918367347e-07,
-      "loss": 0.0552,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0548,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1170,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.816326530612245,
-      "grad_norm": 0.6136904358863831,
+      "grad_norm": 0.9076634049415588,
       "learning_rate": 3.6734693877551025e-07,
-      "loss": 0.044,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0438,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1180,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.857142857142857,
-      "grad_norm": 0.957547128200531,
+      "grad_norm": 1.4096022844314575,
       "learning_rate": 2.8571428571428575e-07,
-      "loss": 0.0545,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0542,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1190,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.8979591836734695,
-      "grad_norm": 0.7660542726516724,
+      "grad_norm": 0.7559374570846558,
       "learning_rate": 2.0408163265306121e-07,
-      "loss": 0.0725,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0715,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1200,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.938775510204081,
-      "grad_norm": 1.0952503681182861,
+      "grad_norm": 0.6551031470298767,
       "learning_rate": 1.2244897959183673e-07,
-      "loss": 0.0648,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0647,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1210,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 4.979591836734694,
-      "grad_norm": 1.1168698072433472,
+      "grad_norm": 0.9617031812667847,
       "learning_rate": 4.0816326530612253e-08,
-      "loss": 0.0415,
-      "max_memory_allocated (GB)": 57.18,
+      "loss": 0.0418,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1220,
       "total_memory_available (GB)": 94.62
     },
     {
       "epoch": 5.0,
-      "max_memory_allocated (GB)": 57.18,
+      "max_memory_allocated (GB)": 60.52,
       "memory_allocated (GB)": 50.57,
       "step": 1225,
       "total_flos": 3.0598946525952e+16,
       "total_memory_available (GB)": 94.62,
-      "train_loss": 0.06081954878203723,
-      "train_runtime": 1056.8939,
-      "train_samples_per_second": 53.754,
-      "train_steps_per_second": 1.345
+      "train_loss": 0.06082177159737567,
+      "train_runtime": 1079.1005,
+      "train_samples_per_second": 52.128,
+      "train_steps_per_second": 1.304
     }
   ],
   "logging_steps": 10,

validation_results.json

@@ -1,10 +1,10 @@
 {
     "epoch": 5.0,
-    "eval_loss": 0.12077265232801437,
-    "eval_runtime": 29.4902,
-    "eval_samples_per_second": 49.245,
-    "eval_steps_per_second": 3.161,
-    "max_memory_allocated (GB)": 57.18,
+    "eval_loss": 0.12328128516674042,
+    "eval_runtime": 24.5719,
+    "eval_samples_per_second": 36.301,
+    "eval_steps_per_second": 2.33,
+    "max_memory_allocated (GB)": 60.52,
     "memory_allocated (GB)": 51.27,
     "total_memory_available (GB)": 94.62
 }