Model save

all_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 1.0,
-    "train_loss": 0.5863300847029632,
-    "train_runtime": 6843.0852,
+    "train_loss": 0.5566485422347293,
+    "train_runtime": 6841.2124,
     "train_samples": 56236,
-    "train_samples_per_second": 8.218,
+    "train_samples_per_second": 8.22,
     "train_steps_per_second": 0.064
 }

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d085da4a857fb7f218891d9be471fb86e732b224ff99d833d19b15d462e4ce9d
+oid sha256:ea0d09d1476d8a49fe285bd946375e1cbdcb6b821865850265935b691a3e4643
 size 4943162336

model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7a5dbc0fe2ece97438d30fd6b0f0e3d1f3e734e6156b8e9d5f3c15a5dcaa7cf0
+oid sha256:3cb1982972159f5dbb8063c84e14c867f92ca23eda59ef58fee5ea9fcd73336c
 size 4999819336

model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:220289bd03f2c16bc923749307f2d3eab7817266cb64a00415b0df5008ec88f8
+oid sha256:d179a041fb683849f3965d5a376a01e933a9c09cf0615f02ac2d59791e92b180
 size 4540516344

runs/Jul02_01-45-45_n136-129-074/events.out.tfevents.1719855968.n136-129-074.2088050.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aaaae682e32eeb1643f77b09ee73d6c7d9891d9c650800b94d91bb64ed3b099d
-size 32981
+oid sha256:acdb8fc06236d0ccea2e1cf7411336f1da73ac2c30cd5dfa5e8c7d2e2049a59c
+size 35399

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 1.0,
-    "train_loss": 0.5863300847029632,
-    "train_runtime": 6843.0852,
+    "train_loss": 0.5566485422347293,
+    "train_runtime": 6841.2124,
     "train_samples": 56236,
-    "train_samples_per_second": 8.218,
+    "train_samples_per_second": 8.22,
     "train_steps_per_second": 0.064
 }

trainer_state.json

@@ -10,7 +10,7 @@
   "log_history": [
     {
       "epoch": 0.0,
-      "grad_norm": 22.78893022336454,
+      "grad_norm": 113.90219393876704,
       "learning_rate": 2.2727272727272727e-09,
       "logits/chosen": -1.6768856048583984,
       "logits/rejected": -1.7259055376052856,
@@ -25,656 +25,656 @@
     },
     {
       "epoch": 0.02,
-      "grad_norm": 25.567365971420195,
+      "grad_norm": 129.29173344209107,
       "learning_rate": 2.2727272727272725e-08,
-      "logits/chosen": -1.7031302452087402,
-      "logits/rejected": -1.6688512563705444,
-      "logps/chosen": -1.213205337524414,
-      "logps/rejected": -1.220165729522705,
-      "loss": 0.693,
-      "rewards/accuracies": 0.4513888955116272,
-      "rewards/chosen": 0.0002006387512665242,
-      "rewards/margins": -0.0009609226835891604,
-      "rewards/rejected": 0.001161561463959515,
+      "logits/chosen": -1.702974796295166,
+      "logits/rejected": -1.6685913801193237,
+      "logps/chosen": -1.213309645652771,
+      "logps/rejected": -1.2203179597854614,
+      "loss": 0.6934,
+      "rewards/accuracies": 0.4375,
+      "rewards/chosen": -4.0181395888794214e-05,
+      "rewards/margins": -0.004325871355831623,
+      "rewards/rejected": 0.0042856899090111256,
       "step": 10
     },
     {
       "epoch": 0.05,
-      "grad_norm": 26.2968319036787,
+      "grad_norm": 129.013817774111,
       "learning_rate": 4.545454545454545e-08,
-      "logits/chosen": -1.7800958156585693,
-      "logits/rejected": -1.7349421977996826,
-      "logps/chosen": -1.1448661088943481,
-      "logps/rejected": -1.185571312904358,
-      "loss": 0.6924,
-      "rewards/accuracies": 0.5062500238418579,
-      "rewards/chosen": -0.0032871514558792114,
-      "rewards/margins": 0.0009361729025840759,
-      "rewards/rejected": -0.004223324358463287,
+      "logits/chosen": -1.7797605991363525,
+      "logits/rejected": -1.7345327138900757,
+      "logps/chosen": -1.144714593887329,
+      "logps/rejected": -1.1853687763214111,
+      "loss": 0.6906,
+      "rewards/accuracies": 0.5687500238418579,
+      "rewards/chosen": -0.014920851215720177,
+      "rewards/margins": 0.0041695646941661835,
+      "rewards/rejected": -0.01909041777253151,
       "step": 20
     },
     {
       "epoch": 0.07,
-      "grad_norm": 28.64228481569523,
+      "grad_norm": 125.39732746183036,
       "learning_rate": 6.818181818181817e-08,
-      "logits/chosen": -1.744109869003296,
-      "logits/rejected": -1.6754045486450195,
-      "logps/chosen": -1.195277452468872,
-      "logps/rejected": -1.2481118440628052,
-      "loss": 0.6883,
-      "rewards/accuracies": 0.6187499761581421,
-      "rewards/chosen": -0.02475227788090706,
-      "rewards/margins": 0.011208651587367058,
-      "rewards/rejected": -0.035960931330919266,
+      "logits/chosen": -1.7443883419036865,
+      "logits/rejected": -1.6757313013076782,
+      "logps/chosen": -1.1939805746078491,
+      "logps/rejected": -1.2470417022705078,
+      "loss": 0.6755,
+      "rewards/accuracies": 0.6812499761581421,
+      "rewards/chosen": -0.11079101264476776,
+      "rewards/margins": 0.058313049376010895,
+      "rewards/rejected": -0.16910405457019806,
       "step": 30
     },
     {
       "epoch": 0.09,
-      "grad_norm": 19.497764912088467,
+      "grad_norm": 99.57108978782348,
       "learning_rate": 9.09090909090909e-08,
-      "logits/chosen": -1.7310603857040405,
-      "logits/rejected": -1.6648972034454346,
-      "logps/chosen": -1.2513717412948608,
-      "logps/rejected": -1.3350750207901,
-      "loss": 0.6798,
-      "rewards/accuracies": 0.6812499761581421,
-      "rewards/chosen": -0.08884630352258682,
-      "rewards/margins": 0.06275957077741623,
-      "rewards/rejected": -0.15160587430000305,
+      "logits/chosen": -1.731239914894104,
+      "logits/rejected": -1.6651241779327393,
+      "logps/chosen": -1.2352855205535889,
+      "logps/rejected": -1.3083709478378296,
+      "loss": 0.6603,
+      "rewards/accuracies": 0.637499988079071,
+      "rewards/chosen": -0.2833693027496338,
+      "rewards/margins": 0.20761942863464355,
+      "rewards/rejected": -0.49098867177963257,
       "step": 40
     },
     {
       "epoch": 0.11,
-      "grad_norm": 24.1777149048992,
+      "grad_norm": 94.95534987876165,
       "learning_rate": 9.994307990108962e-08,
-      "logits/chosen": -1.6923043727874756,
-      "logits/rejected": -1.6273339986801147,
-      "logps/chosen": -1.2986948490142822,
-      "logps/rejected": -1.356567144393921,
-      "loss": 0.6636,
-      "rewards/accuracies": 0.71875,
-      "rewards/chosen": -0.18121571838855743,
-      "rewards/margins": 0.0755590870976448,
-      "rewards/rejected": -0.25677481293678284,
+      "logits/chosen": -1.6967604160308838,
+      "logits/rejected": -1.63302743434906,
+      "logps/chosen": -1.2550745010375977,
+      "logps/rejected": -1.304324984550476,
+      "loss": 0.6261,
+      "rewards/accuracies": 0.668749988079071,
+      "rewards/chosen": -0.4698738157749176,
+      "rewards/margins": 0.29158082604408264,
+      "rewards/rejected": -0.7614546418190002,
       "step": 50
     },
     {
       "epoch": 0.14,
-      "grad_norm": 28.137865307641825,
+      "grad_norm": 134.17306631598268,
       "learning_rate": 9.959570405988094e-08,
-      "logits/chosen": -1.7212276458740234,
-      "logits/rejected": -1.6404285430908203,
-      "logps/chosen": -1.2889435291290283,
-      "logps/rejected": -1.3794549703598022,
-      "loss": 0.6598,
-      "rewards/accuracies": 0.643750011920929,
-      "rewards/chosen": -0.35969024896621704,
-      "rewards/margins": 0.07716657221317291,
-      "rewards/rejected": -0.4368568956851959,
+      "logits/chosen": -1.7350832223892212,
+      "logits/rejected": -1.6556373834609985,
+      "logps/chosen": -1.1744005680084229,
+      "logps/rejected": -1.2519078254699707,
+      "loss": 0.6292,
+      "rewards/accuracies": 0.706250011920929,
+      "rewards/chosen": -0.6530232429504395,
+      "rewards/margins": 0.25578898191452026,
+      "rewards/rejected": -0.9088121652603149,
       "step": 60
     },
     {
       "epoch": 0.16,
-      "grad_norm": 21.253003888204287,
+      "grad_norm": 82.25189789646025,
       "learning_rate": 9.893476820924666e-08,
-      "logits/chosen": -1.8007967472076416,
-      "logits/rejected": -1.7120532989501953,
-      "logps/chosen": -1.476588487625122,
-      "logps/rejected": -1.5963420867919922,
-      "loss": 0.6499,
+      "logits/chosen": -1.8206403255462646,
+      "logits/rejected": -1.7381073236465454,
+      "logps/chosen": -1.2853193283081055,
+      "logps/rejected": -1.3663524389266968,
+      "loss": 0.5893,
       "rewards/accuracies": 0.6937500238418579,
-      "rewards/chosen": -0.5274931192398071,
-      "rewards/margins": 0.14617758989334106,
-      "rewards/rejected": -0.6736707091331482,
+      "rewards/chosen": -0.7247735261917114,
+      "rewards/margins": 0.34368234872817993,
+      "rewards/rejected": -1.0684559345245361,
       "step": 70
     },
     {
       "epoch": 0.18,
-      "grad_norm": 26.777703545743453,
+      "grad_norm": 111.68081612756365,
       "learning_rate": 9.796445099843647e-08,
-      "logits/chosen": -1.7857911586761475,
-      "logits/rejected": -1.6999902725219727,
-      "logps/chosen": -1.5350762605667114,
-      "logps/rejected": -1.6896966695785522,
-      "loss": 0.654,
-      "rewards/accuracies": 0.668749988079071,
-      "rewards/chosen": -0.6562157869338989,
-      "rewards/margins": 0.18818075954914093,
-      "rewards/rejected": -0.8443965911865234,
+      "logits/chosen": -1.8098180294036865,
+      "logits/rejected": -1.7271239757537842,
+      "logps/chosen": -1.2911165952682495,
+      "logps/rejected": -1.3880374431610107,
+      "loss": 0.6284,
+      "rewards/accuracies": 0.6625000238418579,
+      "rewards/chosen": -0.8414822816848755,
+      "rewards/margins": 0.3639078140258789,
+      "rewards/rejected": -1.2053899765014648,
       "step": 80
     },
     {
       "epoch": 0.2,
-      "grad_norm": 25.00309177460346,
+      "grad_norm": 98.75254486198716,
       "learning_rate": 9.669088708527066e-08,
-      "logits/chosen": -1.7340404987335205,
-      "logits/rejected": -1.6651910543441772,
-      "logps/chosen": -1.6042931079864502,
-      "logps/rejected": -1.6743539571762085,
-      "loss": 0.6399,
-      "rewards/accuracies": 0.637499988079071,
-      "rewards/chosen": -0.737191379070282,
-      "rewards/margins": 0.12603236734867096,
-      "rewards/rejected": -0.8632237315177917,
+      "logits/chosen": -1.7565476894378662,
+      "logits/rejected": -1.6922813653945923,
+      "logps/chosen": -1.3374165296554565,
+      "logps/rejected": -1.3871676921844482,
+      "loss": 0.5903,
+      "rewards/accuracies": 0.699999988079071,
+      "rewards/chosen": -1.0171922445297241,
+      "rewards/margins": 0.42706432938575745,
+      "rewards/rejected": -1.4442565441131592,
       "step": 90
     },
     {
       "epoch": 0.23,
-      "grad_norm": 27.969594098965942,
+      "grad_norm": 110.42472659036405,
       "learning_rate": 9.512212835085849e-08,
-      "logits/chosen": -1.772202730178833,
-      "logits/rejected": -1.6827232837677002,
-      "logps/chosen": -1.5563361644744873,
-      "logps/rejected": -1.69924795627594,
-      "loss": 0.6231,
-      "rewards/accuracies": 0.699999988079071,
-      "rewards/chosen": -0.7757617831230164,
-      "rewards/margins": 0.20040392875671387,
-      "rewards/rejected": -0.9761656522750854,
+      "logits/chosen": -1.7919187545776367,
+      "logits/rejected": -1.7078930139541626,
+      "logps/chosen": -1.284099817276001,
+      "logps/rejected": -1.3786065578460693,
+      "loss": 0.5835,
+      "rewards/accuracies": 0.7124999761581421,
+      "rewards/chosen": -1.1564443111419678,
+      "rewards/margins": 0.5179702639579773,
+      "rewards/rejected": -1.6744146347045898,
       "step": 100
     },
     {
       "epoch": 0.25,
-      "grad_norm": 25.83218161515289,
+      "grad_norm": 94.2164442304763,
       "learning_rate": 9.326809299301306e-08,
-      "logits/chosen": -1.771267294883728,
-      "logits/rejected": -1.6669820547103882,
-      "logps/chosen": -1.600318193435669,
-      "logps/rejected": -1.7721306085586548,
-      "loss": 0.6211,
-      "rewards/accuracies": 0.7124999761581421,
-      "rewards/chosen": -0.7993522882461548,
-      "rewards/margins": 0.24041876196861267,
-      "rewards/rejected": -1.0397710800170898,
+      "logits/chosen": -1.7935024499893188,
+      "logits/rejected": -1.6928462982177734,
+      "logps/chosen": -1.3194783926010132,
+      "logps/rejected": -1.4348446130752563,
+      "loss": 0.5774,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": -1.188363790512085,
+      "rewards/margins": 0.63763028383255,
+      "rewards/rejected": -1.8259938955307007,
       "step": 110
     },
     {
       "epoch": 0.27,
-      "grad_norm": 26.667996957961257,
+      "grad_norm": 124.23898564185464,
       "learning_rate": 9.114050282021158e-08,
-      "logits/chosen": -1.767559289932251,
-      "logits/rejected": -1.7058799266815186,
-      "logps/chosen": -1.565065622329712,
-      "logps/rejected": -1.7299690246582031,
-      "loss": 0.6144,
-      "rewards/accuracies": 0.731249988079071,
-      "rewards/chosen": -0.8350059390068054,
-      "rewards/margins": 0.2350219041109085,
-      "rewards/rejected": -1.0700278282165527,
+      "logits/chosen": -1.7868391275405884,
+      "logits/rejected": -1.7275841236114502,
+      "logps/chosen": -1.271259069442749,
+      "logps/rejected": -1.3735918998718262,
+      "loss": 0.563,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": -1.2369645833969116,
+      "rewards/margins": 0.5494047403335571,
+      "rewards/rejected": -1.7863690853118896,
       "step": 120
     },
     {
       "epoch": 0.3,
-      "grad_norm": 25.330536994335382,
+      "grad_norm": 94.58029537839892,
       "learning_rate": 8.875280914254802e-08,
-      "logits/chosen": -1.765298843383789,
-      "logits/rejected": -1.6749998331069946,
-      "logps/chosen": -1.7646329402923584,
-      "logps/rejected": -1.9669040441513062,
-      "loss": 0.5993,
-      "rewards/accuracies": 0.675000011920929,
-      "rewards/chosen": -1.1436058282852173,
-      "rewards/margins": 0.34769219160079956,
-      "rewards/rejected": -1.491297960281372,
+      "logits/chosen": -1.7857173681259155,
+      "logits/rejected": -1.6955511569976807,
+      "logps/chosen": -1.3196834325790405,
+      "logps/rejected": -1.4175035953521729,
+      "loss": 0.5654,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": -1.2685333490371704,
+      "rewards/margins": 0.6939533948898315,
+      "rewards/rejected": -1.9624868631362915,
       "step": 130
     },
     {
       "epoch": 0.32,
-      "grad_norm": 24.002049854639584,
+      "grad_norm": 78.26072920533862,
       "learning_rate": 8.612010772821971e-08,
-      "logits/chosen": -1.7989845275878906,
-      "logits/rejected": -1.7554515600204468,
-      "logps/chosen": -1.8291784524917603,
-      "logps/rejected": -1.9761606454849243,
-      "loss": 0.5991,
-      "rewards/accuracies": 0.731249988079071,
-      "rewards/chosen": -1.2305986881256104,
-      "rewards/margins": 0.3437841534614563,
-      "rewards/rejected": -1.5743829011917114,
+      "logits/chosen": -1.8125206232070923,
+      "logits/rejected": -1.767704725265503,
+      "logps/chosen": -1.3238108158111572,
+      "logps/rejected": -1.362679123878479,
+      "loss": 0.5665,
+      "rewards/accuracies": 0.762499988079071,
+      "rewards/chosen": -1.0993187427520752,
+      "rewards/margins": 0.6377816200256348,
+      "rewards/rejected": -1.73710036277771,
       "step": 140
     },
     {
       "epoch": 0.34,
-      "grad_norm": 28.7685379938573,
+      "grad_norm": 79.54962530262117,
       "learning_rate": 8.325904336322055e-08,
-      "logits/chosen": -1.777856469154358,
-      "logits/rejected": -1.7238337993621826,
-      "logps/chosen": -1.9271425008773804,
-      "logps/rejected": -2.141960859298706,
-      "loss": 0.6082,
-      "rewards/accuracies": 0.6875,
-      "rewards/chosen": -1.5195552110671997,
-      "rewards/margins": 0.3352898359298706,
-      "rewards/rejected": -1.8548450469970703,
+      "logits/chosen": -1.784720778465271,
+      "logits/rejected": -1.728867530822754,
+      "logps/chosen": -1.2866287231445312,
+      "logps/rejected": -1.4028968811035156,
+      "loss": 0.565,
+      "rewards/accuracies": 0.731249988079071,
+      "rewards/chosen": -1.192639708518982,
+      "rewards/margins": 0.6909436583518982,
+      "rewards/rejected": -1.883583426475525,
       "step": 150
     },
     {
       "epoch": 0.36,
-      "grad_norm": 22.36425985091183,
+      "grad_norm": 81.2500983198601,
       "learning_rate": 8.01877046176447e-08,
-      "logits/chosen": -1.72470223903656,
-      "logits/rejected": -1.6554687023162842,
-      "logps/chosen": -1.9798545837402344,
-      "logps/rejected": -2.1781439781188965,
-      "loss": 0.5901,
-      "rewards/accuracies": 0.675000011920929,
-      "rewards/chosen": -1.6031091213226318,
-      "rewards/margins": 0.29380664229393005,
-      "rewards/rejected": -1.8969157934188843,
+      "logits/chosen": -1.7332093715667725,
+      "logits/rejected": -1.6599019765853882,
+      "logps/chosen": -1.3141100406646729,
+      "logps/rejected": -1.4209729433059692,
+      "loss": 0.5662,
+      "rewards/accuracies": 0.706250011920929,
+      "rewards/chosen": -1.358100175857544,
+      "rewards/margins": 0.5547691583633423,
+      "rewards/rejected": -1.9128694534301758,
       "step": 160
     },
     {
       "epoch": 0.39,
-      "grad_norm": 19.908864367961982,
+      "grad_norm": 82.5330575937106,
       "learning_rate": 7.692550948392249e-08,
-      "logits/chosen": -1.7723356485366821,
-      "logits/rejected": -1.7198549509048462,
-      "logps/chosen": -1.9558120965957642,
-      "logps/rejected": -2.159667491912842,
-      "loss": 0.5921,
-      "rewards/accuracies": 0.706250011920929,
-      "rewards/chosen": -1.4911130666732788,
-      "rewards/margins": 0.3868214190006256,
-      "rewards/rejected": -1.877934217453003,
+      "logits/chosen": -1.7886979579925537,
+      "logits/rejected": -1.7225795984268188,
+      "logps/chosen": -1.324988603591919,
+      "logps/rejected": -1.400198221206665,
+      "loss": 0.5593,
+      "rewards/accuracies": 0.731249988079071,
+      "rewards/chosen": -1.1473289728164673,
+      "rewards/margins": 0.647649884223938,
+      "rewards/rejected": -1.7949787378311157,
       "step": 170
     },
     {
       "epoch": 0.41,
-      "grad_norm": 31.864529340507623,
+      "grad_norm": 90.45090476516394,
       "learning_rate": 7.349308261002021e-08,
-      "logits/chosen": -1.7342097759246826,
-      "logits/rejected": -1.6827507019042969,
-      "logps/chosen": -1.9941844940185547,
-      "logps/rejected": -2.22194504737854,
-      "loss": 0.5855,
-      "rewards/accuracies": 0.706250011920929,
-      "rewards/chosen": -1.5941965579986572,
-      "rewards/margins": 0.35622045397758484,
-      "rewards/rejected": -1.950416922569275,
+      "logits/chosen": -1.7442381381988525,
+      "logits/rejected": -1.6850671768188477,
+      "logps/chosen": -1.318302869796753,
+      "logps/rejected": -1.4245436191558838,
+      "loss": 0.5533,
+      "rewards/accuracies": 0.7437499761581421,
+      "rewards/chosen": -1.2121667861938477,
+      "rewards/margins": 0.5659047961235046,
+      "rewards/rejected": -1.778071641921997,
       "step": 180
     },
     {
       "epoch": 0.43,
-      "grad_norm": 24.691918932601595,
+      "grad_norm": 84.71027965029923,
       "learning_rate": 6.991212490377531e-08,
-      "logits/chosen": -1.7905107736587524,
-      "logits/rejected": -1.7474826574325562,
-      "logps/chosen": -2.10019850730896,
-      "logps/rejected": -2.3696181774139404,
-      "loss": 0.5604,
-      "rewards/accuracies": 0.75,
-      "rewards/chosen": -1.740191102027893,
-      "rewards/margins": 0.4957551956176758,
-      "rewards/rejected": -2.2359461784362793,
+      "logits/chosen": -1.7995620965957642,
+      "logits/rejected": -1.745843529701233,
+      "logps/chosen": -1.3425184488296509,
+      "logps/rejected": -1.4334449768066406,
+      "loss": 0.5291,
+      "rewards/accuracies": 0.7562500238418579,
+      "rewards/chosen": -1.1241543292999268,
+      "rewards/margins": 0.6938456296920776,
+      "rewards/rejected": -1.818000078201294,
       "step": 190
     },
     {
       "epoch": 0.46,
-      "grad_norm": 30.11763021258028,
+      "grad_norm": 104.41708218825802,
       "learning_rate": 6.620527633276978e-08,
-      "logits/chosen": -1.7320966720581055,
-      "logits/rejected": -1.6665375232696533,
-      "logps/chosen": -2.161935329437256,
-      "logps/rejected": -2.522731304168701,
-      "loss": 0.5808,
-      "rewards/accuracies": 0.7250000238418579,
-      "rewards/chosen": -1.9349641799926758,
-      "rewards/margins": 0.5480148792266846,
-      "rewards/rejected": -2.4829792976379395,
+      "logits/chosen": -1.7381435632705688,
+      "logits/rejected": -1.662602424621582,
+      "logps/chosen": -1.3073325157165527,
+      "logps/rejected": -1.4739927053451538,
+      "loss": 0.5404,
+      "rewards/accuracies": 0.762499988079071,
+      "rewards/chosen": -1.1287935972213745,
+      "rewards/margins": 0.7987143993377686,
+      "rewards/rejected": -1.927507758140564,
       "step": 200
     },
     {
       "epoch": 0.48,
-      "grad_norm": 27.459036499855436,
+      "grad_norm": 98.12458997212138,
       "learning_rate": 6.239597278716581e-08,
-      "logits/chosen": -1.7859094142913818,
-      "logits/rejected": -1.7306747436523438,
-      "logps/chosen": -2.2954204082489014,
-      "logps/rejected": -2.540391445159912,
-      "loss": 0.5619,
-      "rewards/accuracies": 0.7250000238418579,
-      "rewards/chosen": -2.1001808643341064,
-      "rewards/margins": 0.5312689542770386,
-      "rewards/rejected": -2.6314499378204346,
+      "logits/chosen": -1.808566689491272,
+      "logits/rejected": -1.7453174591064453,
+      "logps/chosen": -1.3598778247833252,
+      "logps/rejected": -1.4206109046936035,
+      "loss": 0.5355,
+      "rewards/accuracies": 0.762499988079071,
+      "rewards/chosen": -1.1454766988754272,
+      "rewards/margins": 0.813967227935791,
+      "rewards/rejected": -1.9594440460205078,
       "step": 210
     },
     {
       "epoch": 0.5,
-      "grad_norm": 26.562029580561404,
+      "grad_norm": 73.95327688888813,
       "learning_rate": 5.8508297910462456e-08,
-      "logits/chosen": -1.7313369512557983,
-      "logits/rejected": -1.6572465896606445,
-      "logps/chosen": -2.212656021118164,
-      "logps/rejected": -2.5422561168670654,
-      "loss": 0.5673,
-      "rewards/accuracies": 0.699999988079071,
-      "rewards/chosen": -2.0982813835144043,
-      "rewards/margins": 0.5436533689498901,
-      "rewards/rejected": -2.641934871673584,
+      "logits/chosen": -1.7581428289413452,
+      "logits/rejected": -1.6758205890655518,
+      "logps/chosen": -1.2785987854003906,
+      "logps/rejected": -1.4212281703948975,
+      "loss": 0.5267,
+      "rewards/accuracies": 0.793749988079071,
+      "rewards/chosen": -1.1508338451385498,
+      "rewards/margins": 0.8485604524612427,
+      "rewards/rejected": -1.999394178390503,
       "step": 220
     },
     {
       "epoch": 0.52,
-      "grad_norm": 26.043734197401424,
+      "grad_norm": 101.92396024664481,
       "learning_rate": 5.456683083494731e-08,
-      "logits/chosen": -1.7218725681304932,
-      "logits/rejected": -1.6862335205078125,
-      "logps/chosen": -2.1797163486480713,
-      "logps/rejected": -2.389193058013916,
-      "loss": 0.5843,
-      "rewards/accuracies": 0.65625,
-      "rewards/chosen": -1.9718148708343506,
-      "rewards/margins": 0.37147068977355957,
-      "rewards/rejected": -2.34328556060791,
+      "logits/chosen": -1.7508561611175537,
+      "logits/rejected": -1.7111085653305054,
+      "logps/chosen": -1.2999995946884155,
+      "logps/rejected": -1.3745485544204712,
+      "loss": 0.554,
+      "rewards/accuracies": 0.6937500238418579,
+      "rewards/chosen": -1.0619090795516968,
+      "rewards/margins": 0.5080693364143372,
+      "rewards/rejected": -1.5699783563613892,
       "step": 230
     },
     {
       "epoch": 0.55,
-      "grad_norm": 33.60349201324705,
+      "grad_norm": 110.90195766775575,
       "learning_rate": 5.059649078450834e-08,
-      "logits/chosen": -1.7049477100372314,
-      "logits/rejected": -1.6663001775741577,
-      "logps/chosen": -2.2287259101867676,
-      "logps/rejected": -2.5061872005462646,
-      "loss": 0.5527,
-      "rewards/accuracies": 0.6625000238418579,
-      "rewards/chosen": -2.120744466781616,
-      "rewards/margins": 0.4308454990386963,
-      "rewards/rejected": -2.5515899658203125,
+      "logits/chosen": -1.745862603187561,
+      "logits/rejected": -1.69782292842865,
+      "logps/chosen": -1.2861733436584473,
+      "logps/rejected": -1.4080326557159424,
+      "loss": 0.5271,
+      "rewards/accuracies": 0.7437499761581421,
+      "rewards/chosen": -1.1781983375549316,
+      "rewards/margins": 0.5982070565223694,
+      "rewards/rejected": -1.7764055728912354,
       "step": 240
     },
     {
       "epoch": 0.57,
-      "grad_norm": 24.721682812448297,
+      "grad_norm": 90.78412484714156,
       "learning_rate": 4.6622379527277186e-08,
-      "logits/chosen": -1.716957688331604,
-      "logits/rejected": -1.6692520380020142,
-      "logps/chosen": -2.323068857192993,
-      "logps/rejected": -2.551687240600586,
-      "loss": 0.5645,
-      "rewards/accuracies": 0.625,
-      "rewards/chosen": -2.3683600425720215,
-      "rewards/margins": 0.3910773694515228,
-      "rewards/rejected": -2.759437322616577,
+      "logits/chosen": -1.7498047351837158,
+      "logits/rejected": -1.6966612339019775,
+      "logps/chosen": -1.263024091720581,
+      "logps/rejected": -1.3622833490371704,
+      "loss": 0.5322,
+      "rewards/accuracies": 0.699999988079071,
+      "rewards/chosen": -1.241353988647461,
+      "rewards/margins": 0.6617964506149292,
+      "rewards/rejected": -1.9031503200531006,
       "step": 250
     },
     {
       "epoch": 0.59,
-      "grad_norm": 28.965518895994943,
+      "grad_norm": 80.47587413810997,
       "learning_rate": 4.26696226741691e-08,
-      "logits/chosen": -1.731431007385254,
-      "logits/rejected": -1.6687753200531006,
-      "logps/chosen": -2.4228413105010986,
-      "logps/rejected": -2.701592206954956,
-      "loss": 0.5647,
-      "rewards/accuracies": 0.6499999761581421,
-      "rewards/chosen": -2.380035877227783,
-      "rewards/margins": 0.5175679922103882,
-      "rewards/rejected": -2.897603750228882,
+      "logits/chosen": -1.766274094581604,
+      "logits/rejected": -1.6926915645599365,
+      "logps/chosen": -1.3750711679458618,
+      "logps/rejected": -1.4678473472595215,
+      "loss": 0.5309,
+      "rewards/accuracies": 0.699999988079071,
+      "rewards/chosen": -1.4224761724472046,
+      "rewards/margins": 0.7280942797660828,
+      "rewards/rejected": -2.1505703926086426,
       "step": 260
     },
     {
       "epoch": 0.61,
-      "grad_norm": 24.276560436452733,
+      "grad_norm": 81.2536856430445,
       "learning_rate": 3.876321082668098e-08,
-      "logits/chosen": -1.7877088785171509,
-      "logits/rejected": -1.7258117198944092,
-      "logps/chosen": -2.3104138374328613,
-      "logps/rejected": -2.597568988800049,
-      "loss": 0.5577,
-      "rewards/accuracies": 0.7250000238418579,
-      "rewards/chosen": -2.143846035003662,
-      "rewards/margins": 0.5556932091712952,
-      "rewards/rejected": -2.6995394229888916,
+      "logits/chosen": -1.8292741775512695,
+      "logits/rejected": -1.7543790340423584,
+      "logps/chosen": -1.3468070030212402,
+      "logps/rejected": -1.4356333017349243,
+      "loss": 0.5228,
+      "rewards/accuracies": 0.78125,
+      "rewards/chosen": -1.0831630229949951,
+      "rewards/margins": 0.7951737642288208,
+      "rewards/rejected": -1.8783369064331055,
       "step": 270
     },
     {
       "epoch": 0.64,
-      "grad_norm": 26.01398581499373,
+      "grad_norm": 87.95396768106268,
       "learning_rate": 3.492784157826244e-08,
-      "logits/chosen": -1.7255363464355469,
-      "logits/rejected": -1.6368858814239502,
-      "logps/chosen": -2.307375431060791,
-      "logps/rejected": -2.6467177867889404,
-      "loss": 0.5569,
-      "rewards/accuracies": 0.71875,
-      "rewards/chosen": -2.1689984798431396,
-      "rewards/margins": 0.6079045534133911,
-      "rewards/rejected": -2.7769031524658203,
+      "logits/chosen": -1.754434585571289,
+      "logits/rejected": -1.654488205909729,
+      "logps/chosen": -1.331969141960144,
+      "logps/rejected": -1.4491215944290161,
+      "loss": 0.5257,
+      "rewards/accuracies": 0.7124999761581421,
+      "rewards/chosen": -1.0909301042556763,
+      "rewards/margins": 0.8176229596138,
+      "rewards/rejected": -1.908552885055542,
       "step": 280
     },
     {
       "epoch": 0.66,
-      "grad_norm": 22.449000475495634,
+      "grad_norm": 75.75936548706386,
       "learning_rate": 3.118776336817812e-08,
-      "logits/chosen": -1.7589473724365234,
-      "logits/rejected": -1.7079540491104126,
-      "logps/chosen": -2.2426087856292725,
-      "logps/rejected": -2.575594425201416,
-      "loss": 0.5461,
-      "rewards/accuracies": 0.731249988079071,
-      "rewards/chosen": -2.0787200927734375,
-      "rewards/margins": 0.6515394449234009,
-      "rewards/rejected": -2.730259656906128,
+      "logits/chosen": -1.800217628479004,
+      "logits/rejected": -1.7326066493988037,
+      "logps/chosen": -1.3206775188446045,
+      "logps/rejected": -1.4107261896133423,
+      "loss": 0.5236,
+      "rewards/accuracies": 0.7562500238418579,
+      "rewards/chosen": -1.1742886304855347,
+      "rewards/margins": 0.8283244967460632,
+      "rewards/rejected": -2.002613067626953,
       "step": 290
     },
     {
       "epoch": 0.68,
-      "grad_norm": 23.916342153040954,
+      "grad_norm": 86.61923660491236,
       "learning_rate": 2.7566622175067443e-08,
-      "logits/chosen": -1.7405074834823608,
-      "logits/rejected": -1.6836473941802979,
-      "logps/chosen": -2.341914176940918,
-      "logps/rejected": -2.724348545074463,
-      "loss": 0.5507,
-      "rewards/accuracies": 0.7562500238418579,
-      "rewards/chosen": -2.2666609287261963,
-      "rewards/margins": 0.6088961958885193,
-      "rewards/rejected": -2.875556707382202,
+      "logits/chosen": -1.781106948852539,
+      "logits/rejected": -1.7118265628814697,
+      "logps/chosen": -1.3446478843688965,
+      "logps/rejected": -1.4952938556671143,
+      "loss": 0.5304,
+      "rewards/accuracies": 0.7250000238418579,
+      "rewards/chosen": -1.3606449365615845,
+      "rewards/margins": 0.7265914082527161,
+      "rewards/rejected": -2.0872364044189453,
       "step": 300
     },
     {
       "epoch": 0.71,
-      "grad_norm": 28.87016818276154,
+      "grad_norm": 104.45230542444779,
       "learning_rate": 2.408731201945432e-08,
-      "logits/chosen": -1.738867998123169,
-      "logits/rejected": -1.6946017742156982,
-      "logps/chosen": -2.2607665061950684,
-      "logps/rejected": -2.508131742477417,
-      "loss": 0.5549,
-      "rewards/accuracies": 0.6812499761581421,
-      "rewards/chosen": -2.133742332458496,
-      "rewards/margins": 0.45442089438438416,
-      "rewards/rejected": -2.5881636142730713,
+      "logits/chosen": -1.7766481637954712,
+      "logits/rejected": -1.7281579971313477,
+      "logps/chosen": -1.3146270513534546,
+      "logps/rejected": -1.4018762111663818,
+      "loss": 0.5176,
+      "rewards/accuracies": 0.768750011920929,
+      "rewards/chosen": -1.2073173522949219,
+      "rewards/margins": 0.670944333076477,
+      "rewards/rejected": -1.8782615661621094,
       "step": 310
     },
     {
       "epoch": 0.73,
-      "grad_norm": 26.474891102018546,
+      "grad_norm": 97.89401056240122,
       "learning_rate": 2.0771830220378112e-08,
-      "logits/chosen": -1.6925548315048218,
-      "logits/rejected": -1.6349446773529053,
-      "logps/chosen": -2.292504072189331,
-      "logps/rejected": -2.5281872749328613,
-      "loss": 0.5545,
-      "rewards/accuracies": 0.625,
-      "rewards/chosen": -2.139239549636841,
-      "rewards/margins": 0.48706698417663574,
-      "rewards/rejected": -2.6263065338134766,
+      "logits/chosen": -1.7322553396224976,
+      "logits/rejected": -1.6741504669189453,
+      "logps/chosen": -1.347646951675415,
+      "logps/rejected": -1.4063564538955688,
+      "loss": 0.5268,
+      "rewards/accuracies": 0.737500011920929,
+      "rewards/chosen": -1.2476258277893066,
+      "rewards/margins": 0.6655985713005066,
+      "rewards/rejected": -1.913224458694458,
       "step": 320
     },
     {
       "epoch": 0.75,
-      "grad_norm": 26.153240463475917,
+      "grad_norm": 79.38110491324964,
       "learning_rate": 1.7641138321260257e-08,
-      "logits/chosen": -1.7273342609405518,
-      "logits/rejected": -1.6638519763946533,
-      "logps/chosen": -2.218357563018799,
-      "logps/rejected": -2.6145200729370117,
-      "loss": 0.5485,
-      "rewards/accuracies": 0.824999988079071,
-      "rewards/chosen": -2.069005250930786,
-      "rewards/margins": 0.7372555136680603,
-      "rewards/rejected": -2.806260585784912,
+      "logits/chosen": -1.7678320407867432,
+      "logits/rejected": -1.692814826965332,
+      "logps/chosen": -1.3020561933517456,
+      "logps/rejected": -1.4196479320526123,
+      "loss": 0.5181,
+      "rewards/accuracies": 0.8187500238418579,
+      "rewards/chosen": -1.1820148229599,
+      "rewards/margins": 0.900565505027771,
+      "rewards/rejected": -2.082580327987671,
       "step": 330
     },
     {
       "epoch": 0.77,
-      "grad_norm": 26.27728864294511,
+      "grad_norm": 79.242696260741,
       "learning_rate": 1.4715029564277793e-08,
-      "logits/chosen": -1.7901878356933594,
-      "logits/rejected": -1.7485193014144897,
-      "logps/chosen": -2.2015347480773926,
-      "logps/rejected": -2.5572023391723633,
-      "loss": 0.5577,
-      "rewards/accuracies": 0.7562500238418579,
-      "rewards/chosen": -1.9937489032745361,
-      "rewards/margins": 0.637208104133606,
-      "rewards/rejected": -2.6309566497802734,
+      "logits/chosen": -1.8388326168060303,
+      "logits/rejected": -1.7862517833709717,
+      "logps/chosen": -1.306144118309021,
+      "logps/rejected": -1.4227768182754517,
+      "loss": 0.5342,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": -1.0148370265960693,
+      "rewards/margins": 0.7956900596618652,
+      "rewards/rejected": -1.8105270862579346,
       "step": 340
     },
     {
       "epoch": 0.8,
-      "grad_norm": 26.035613152091976,
+      "grad_norm": 86.49780334887537,
       "learning_rate": 1.2012003751113343e-08,
-      "logits/chosen": -1.7805286645889282,
-      "logits/rejected": -1.7256838083267212,
-      "logps/chosen": -2.3138480186462402,
-      "logps/rejected": -2.6945948600769043,
-      "loss": 0.536,
-      "rewards/accuracies": 0.71875,
-      "rewards/chosen": -2.3033559322357178,
-      "rewards/margins": 0.6611131429672241,
-      "rewards/rejected": -2.9644687175750732,
+      "logits/chosen": -1.822198510169983,
+      "logits/rejected": -1.756326675415039,
+      "logps/chosen": -1.2951385974884033,
+      "logps/rejected": -1.4268760681152344,
+      "loss": 0.5045,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": -1.329688310623169,
+      "rewards/margins": 0.815467357635498,
+      "rewards/rejected": -2.145155668258667,
       "step": 350
     },
     {
       "epoch": 0.82,
-      "grad_norm": 31.4301067051425,
+      "grad_norm": 82.9744571713248,
       "learning_rate": 9.549150281252633e-09,
-      "logits/chosen": -1.7304834127426147,
-      "logits/rejected": -1.6880794763565063,
-      "logps/chosen": -2.2502734661102295,
-      "logps/rejected": -2.593324661254883,
-      "loss": 0.54,
-      "rewards/accuracies": 0.6937500238418579,
-      "rewards/chosen": -2.1562764644622803,
-      "rewards/margins": 0.5826417207717896,
-      "rewards/rejected": -2.7389183044433594,
+      "logits/chosen": -1.7854712009429932,
+      "logits/rejected": -1.7307994365692139,
+      "logps/chosen": -1.2910486459732056,
+      "logps/rejected": -1.4215586185455322,
+      "loss": 0.508,
+      "rewards/accuracies": 0.737500011920929,
+      "rewards/chosen": -1.1891343593597412,
+      "rewards/margins": 0.7877954244613647,
+      "rewards/rejected": -1.9769299030303955,
       "step": 360
     },
     {
       "epoch": 0.84,
-      "grad_norm": 25.13509986712804,
+      "grad_norm": 77.94995731245406,
       "learning_rate": 7.3420401072985306e-09,
-      "logits/chosen": -1.7723455429077148,
-      "logits/rejected": -1.721980094909668,
-      "logps/chosen": -2.3107597827911377,
-      "logps/rejected": -2.679028272628784,
-      "loss": 0.5374,
-      "rewards/accuracies": 0.6875,
-      "rewards/chosen": -2.1922922134399414,
-      "rewards/margins": 0.6216103434562683,
-      "rewards/rejected": -2.8139023780822754,
+      "logits/chosen": -1.8163235187530518,
+      "logits/rejected": -1.7626298666000366,
+      "logps/chosen": -1.3424322605133057,
+      "logps/rejected": -1.483293890953064,
+      "loss": 0.5133,
+      "rewards/accuracies": 0.737500011920929,
+      "rewards/chosen": -1.2781879901885986,
+      "rewards/margins": 0.8339790105819702,
+      "rewards/rejected": -2.1121671199798584,
       "step": 370
     },
     {
       "epoch": 0.86,
-      "grad_norm": 28.235691242335875,
+      "grad_norm": 99.08582811249858,
       "learning_rate": 5.404627290395369e-09,
-      "logits/chosen": -1.733109712600708,
-      "logits/rejected": -1.6734564304351807,
-      "logps/chosen": -2.241391897201538,
-      "logps/rejected": -2.590752363204956,
-      "loss": 0.5365,
-      "rewards/accuracies": 0.7437499761581421,
-      "rewards/chosen": -2.103571653366089,
-      "rewards/margins": 0.6187294721603394,
-      "rewards/rejected": -2.7223010063171387,
+      "logits/chosen": -1.781881332397461,
+      "logits/rejected": -1.712882399559021,
+      "logps/chosen": -1.3003751039505005,
+      "logps/rejected": -1.4186222553253174,
+      "loss": 0.5136,
+      "rewards/accuracies": 0.737500011920929,
+      "rewards/chosen": -1.1076927185058594,
+      "rewards/margins": 0.7825103998184204,
+      "rewards/rejected": -1.8902031183242798,
       "step": 380
     },
     {
       "epoch": 0.89,
-      "grad_norm": 33.12956908569685,
+      "grad_norm": 142.606766968544,
       "learning_rate": 3.74916077816162e-09,
-      "logits/chosen": -1.7393004894256592,
-      "logits/rejected": -1.6843922138214111,
-      "logps/chosen": -2.267721652984619,
-      "logps/rejected": -2.5625317096710205,
-      "loss": 0.5513,
-      "rewards/accuracies": 0.6875,
-      "rewards/chosen": -2.235853672027588,
-      "rewards/margins": 0.5573619604110718,
-      "rewards/rejected": -2.7932159900665283,
+      "logits/chosen": -1.7914397716522217,
+      "logits/rejected": -1.7269777059555054,
+      "logps/chosen": -1.2829267978668213,
+      "logps/rejected": -1.3663567304611206,
+      "loss": 0.5283,
+      "rewards/accuracies": 0.706250011920929,
+      "rewards/chosen": -1.3313188552856445,
+      "rewards/margins": 0.6730116605758667,
+      "rewards/rejected": -2.0043303966522217,
       "step": 390
     },
     {
       "epoch": 0.91,
-      "grad_norm": 28.33250387056753,
+      "grad_norm": 87.75942228209004,
       "learning_rate": 2.386106962899165e-09,
-      "logits/chosen": -1.666548490524292,
-      "logits/rejected": -1.6048628091812134,
-      "logps/chosen": -2.401040554046631,
-      "logps/rejected": -2.733664035797119,
-      "loss": 0.5422,
-      "rewards/accuracies": 0.706250011920929,
-      "rewards/chosen": -2.349238872528076,
-      "rewards/margins": 0.5525677800178528,
-      "rewards/rejected": -2.901806592941284,
+      "logits/chosen": -1.7152767181396484,
+      "logits/rejected": -1.6405121088027954,
+      "logps/chosen": -1.3587360382080078,
+      "logps/rejected": -1.4870920181274414,
+      "loss": 0.523,
+      "rewards/accuracies": 0.7124999761581421,
+      "rewards/chosen": -1.3231509923934937,
+      "rewards/margins": 0.7201642394065857,
+      "rewards/rejected": -2.0433154106140137,
       "step": 400
     },
     {
       "epoch": 0.93,
-      "grad_norm": 31.32317842504756,
+      "grad_norm": 95.93932206444032,
       "learning_rate": 1.3240835096913706e-09,
-      "logits/chosen": -1.6938998699188232,
-      "logits/rejected": -1.6031659841537476,
-      "logps/chosen": -2.251462459564209,
-      "logps/rejected": -2.6835711002349854,
-      "loss": 0.5597,
+      "logits/chosen": -1.7500959634780884,
+      "logits/rejected": -1.6428474187850952,
+      "logps/chosen": -1.280632734298706,
+      "logps/rejected": -1.440298080444336,
+      "loss": 0.5334,
       "rewards/accuracies": 0.7124999761581421,
-      "rewards/chosen": -2.1598236560821533,
-      "rewards/margins": 0.7253878116607666,
-      "rewards/rejected": -2.885211706161499,
+      "rewards/chosen": -1.0908193588256836,
+      "rewards/margins": 0.9025093913078308,
+      "rewards/rejected": -1.9933288097381592,
       "step": 410
     },
     {
       "epoch": 0.96,
-      "grad_norm": 23.813879321456167,
+      "grad_norm": 185.3215699605587,
       "learning_rate": 5.698048727497462e-10,
-      "logits/chosen": -1.7287553548812866,
-      "logits/rejected": -1.6634016036987305,
-      "logps/chosen": -2.358034610748291,
-      "logps/rejected": -2.7596287727355957,
-      "loss": 0.5431,
-      "rewards/accuracies": 0.731249988079071,
-      "rewards/chosen": -2.3334739208221436,
-      "rewards/margins": 0.669273853302002,
-      "rewards/rejected": -3.0027480125427246,
+      "logits/chosen": -1.7732937335968018,
+      "logits/rejected": -1.6997044086456299,
+      "logps/chosen": -1.3186805248260498,
+      "logps/rejected": -1.473042607307434,
+      "loss": 0.5207,
+      "rewards/accuracies": 0.7875000238418579,
+      "rewards/chosen": -1.2738254070281982,
+      "rewards/margins": 0.8740528225898743,
+      "rewards/rejected": -2.1478781700134277,
       "step": 420
     },
     {
       "epoch": 0.98,
-      "grad_norm": 24.13495374841747,
+      "grad_norm": 85.90784633584286,
       "learning_rate": 1.2803984447259387e-10,
-      "logits/chosen": -1.7348169088363647,
-      "logits/rejected": -1.6923096179962158,
-      "logps/chosen": -2.339822769165039,
-      "logps/rejected": -2.7242085933685303,
-      "loss": 0.5338,
-      "rewards/accuracies": 0.7250000238418579,
-      "rewards/chosen": -2.314966917037964,
-      "rewards/margins": 0.6734664440155029,
-      "rewards/rejected": -2.9884331226348877,
+      "logits/chosen": -1.7828824520111084,
+      "logits/rejected": -1.7222639322280884,
+      "logps/chosen": -1.310931921005249,
+      "logps/rejected": -1.4457513093948364,
+      "loss": 0.5116,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": -1.2859268188476562,
+      "rewards/margins": 0.8716639280319214,
+      "rewards/rejected": -2.157590866088867,
       "step": 430
     },
     {
       "epoch": 1.0,
       "step": 439,
       "total_flos": 0.0,
-      "train_loss": 0.5863300847029632,
-      "train_runtime": 6843.0852,
-      "train_samples_per_second": 8.218,
+      "train_loss": 0.5566485422347293,
+      "train_runtime": 6841.2124,
+      "train_samples_per_second": 8.22,
       "train_steps_per_second": 0.064
     }
   ],