Model save

README.md

@@ -15,15 +15,15 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model was trained from scratch on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 1.9828
-- Rewards/chosen: -3.4223
-- Rewards/rejected: -2.1126
-- Rewards/accuracies: 0.3555
-- Rewards/margins: -1.3097
-- Logps/rejected: -521.2875
-- Logps/chosen: -405.8879
-- Logits/rejected: -4.9364
-- Logits/chosen: -4.7068
+- Loss: 0.6931
+- Rewards/chosen: -2.1301
+- Rewards/rejected: -2.1301
+- Rewards/accuracies: 0.0
+- Rewards/margins: 0.0
+- Logps/rejected: -159.8372
+- Logps/chosen: -159.8372
+- Logits/rejected: -3.1995
+- Logits/chosen: -3.1995
 
 ## Model description
 
@@ -42,7 +42,7 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 5e-08
+- learning_rate: 2e-07
 - train_batch_size: 8
 - eval_batch_size: 8
 - seed: 42
@@ -60,9 +60,9 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
 |:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
-| 0.3652        | 0.29  | 100  | 1.7488          | -2.2159        | -1.0097          | 0.3516             | -1.2062         | -519.0817      | -403.4752    | -4.9249         | -4.6962       |
-| 0.312         | 0.57  | 200  | 1.9596          | -3.1949        | -1.8164          | 0.3398             | -1.3786         | -520.6950      | -405.4332    | -4.9391         | -4.7096       |
-| 0.2993        | 0.86  | 300  | 1.9828          | -3.4223        | -2.1126          | 0.3555             | -1.3097         | -521.2875      | -405.8879    | -4.9364         | -4.7068       |
+| 0.3114        | 0.29  | 100  | 0.6931          | -2.8588        | -2.8588          | 0.0                | 0.0             | -161.2947      | -161.2947    | -3.2262         | -3.2262       |
+| 0.2741        | 0.57  | 200  | 0.6931          | -2.6760        | -2.6760          | 0.0                | 0.0             | -160.9289      | -160.9289    | -3.2033         | -3.2033       |
+| 0.2695        | 0.86  | 300  | 0.6931          | -2.1301        | -2.1301          | 0.0                | 0.0             | -159.8372      | -159.8372    | -3.1995         | -3.1995       |
 
 
 ### Framework versions

all_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 1.0,
-    "train_loss": 0.3696309270011661,
-    "train_runtime": 5591.2149,
+    "train_loss": 0.30806411576476,
+    "train_runtime": 5299.0292,
     "train_samples": 44682,
-    "train_samples_per_second": 7.991,
-    "train_steps_per_second": 0.062
+    "train_samples_per_second": 8.432,
+    "train_steps_per_second": 0.066
 }

config.json

@@ -20,7 +20,7 @@
   "sliding_window": 4096,
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
-  "transformers_version": "4.41.1",
+  "transformers_version": "4.38.2",
   "use_cache": false,
   "vocab_size": 32002
 }

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b289b436dc9787cb73fff5a2a0c5fb69ad78dc06d960a4946cb9d6b35bf920b3
+oid sha256:ff3d0ac00c3304811f4f56db0c3a49ea7bf825902fea45c8a9791a2e4f7431ab
 size 4943178720

model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:556d3f80045da81e8b580005d0676f922550a32c41b802ac16391cdb8a56eac7
+oid sha256:1122a41f096dacbe12e70bff78c3951321a0a6f46db2c7995a726fd8f909bf84
 size 4999819336

model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:06047de541b6531a598ee5f5cc63aaa960471d4a4c6da134647cc69aa23f71b0
+oid sha256:4c18a2acaf553956d0cddde18a5d52d992dfb5723fc496a27bcde59c9bb62529
 size 4540532728

runs/May30_11-21-27_n136-129-074/events.out.tfevents.1717039322.n136-129-074.2425775.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:054622cc33ba7e3d84169c49b4b7eed8365d0c64fcf5eb34d2a9a1e5f60e6a4a
-size 28328
+oid sha256:f8fc27d90b3ccda7d2e755a090af78f6ce2111c0b55bbb4c06e4b2693fca26d7
+size 31434

tokenizer.json

@@ -152,7 +152,6 @@
     "end_of_word_suffix": null,
     "fuse_unk": true,
     "byte_fallback": true,
-    "ignore_merges": false,
     "vocab": {
       "<unk>": 0,
       "<s>": 1,

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 1.0,
-    "train_loss": 0.3696309270011661,
-    "train_runtime": 5591.2149,
+    "train_loss": 0.30806411576476,
+    "train_runtime": 5299.0292,
     "train_samples": 44682,
-    "train_samples_per_second": 7.991,
-    "train_steps_per_second": 0.062
+    "train_samples_per_second": 8.432,
+    "train_steps_per_second": 0.066
 }

trainer_state.json

@@ -10,8 +10,8 @@
   "log_history": [
     {
       "epoch": 0.0,
-      "grad_norm": 1482.8898632632786,
-      "learning_rate": 1.4285714285714284e-09,
+      "grad_norm": 1483.079147156977,
+      "learning_rate": 5.7142857142857136e-09,
       "logits/chosen": -4.490396976470947,
       "logits/rejected": -4.787891387939453,
       "logps/chosen": -300.56573486328125,
@@ -25,570 +25,570 @@
     },
     {
       "epoch": 0.03,
-      "grad_norm": 1626.423546399318,
-      "learning_rate": 1.4285714285714284e-08,
-      "logits/chosen": -4.29224967956543,
-      "logits/rejected": -4.469963550567627,
-      "logps/chosen": -285.7841796875,
-      "logps/rejected": -241.37472534179688,
-      "loss": 0.7238,
-      "rewards/accuracies": 0.4097222089767456,
-      "rewards/chosen": 0.00654969597235322,
-      "rewards/margins": -0.022695984691381454,
-      "rewards/rejected": 0.029245682060718536,
+      "grad_norm": 1678.9009350663757,
+      "learning_rate": 5.714285714285714e-08,
+      "logits/chosen": -4.292892932891846,
+      "logits/rejected": -4.4704790115356445,
+      "logps/chosen": -285.7367858886719,
+      "logps/rejected": -241.40879821777344,
+      "loss": 0.7232,
+      "rewards/accuracies": 0.4375,
+      "rewards/chosen": 0.03021715022623539,
+      "rewards/margins": 0.018014244735240936,
+      "rewards/rejected": 0.012202905490994453,
       "step": 10
     },
     {
       "epoch": 0.06,
-      "grad_norm": 1614.6738239528886,
-      "learning_rate": 2.857142857142857e-08,
-      "logits/chosen": -4.210062026977539,
-      "logits/rejected": -4.502069473266602,
-      "logps/chosen": -312.73968505859375,
-      "logps/rejected": -247.76315307617188,
-      "loss": 0.7059,
-      "rewards/accuracies": 0.4625000059604645,
-      "rewards/chosen": 0.01807965151965618,
-      "rewards/margins": 0.015493685379624367,
-      "rewards/rejected": 0.0025859654415398836,
+      "grad_norm": 1307.5912774162387,
+      "learning_rate": 1.1428571428571427e-07,
+      "logits/chosen": -4.212913990020752,
+      "logits/rejected": -4.504936695098877,
+      "logps/chosen": -312.25714111328125,
+      "logps/rejected": -247.5605926513672,
+      "loss": 0.6348,
+      "rewards/accuracies": 0.6187499761581421,
+      "rewards/chosen": 0.25936827063560486,
+      "rewards/margins": 0.15551437437534332,
+      "rewards/rejected": 0.10385391861200333,
       "step": 20
     },
     {
       "epoch": 0.09,
-      "grad_norm": 1444.832849324016,
-      "learning_rate": 4.285714285714285e-08,
-      "logits/chosen": -4.140606880187988,
-      "logits/rejected": -4.359016418457031,
-      "logps/chosen": -323.00701904296875,
-      "logps/rejected": -274.5583801269531,
-      "loss": 0.6754,
-      "rewards/accuracies": 0.6812499761581421,
-      "rewards/chosen": 0.26756447553634644,
-      "rewards/margins": 0.19910377264022827,
-      "rewards/rejected": 0.06846068799495697,
+      "grad_norm": 877.5830541670848,
+      "learning_rate": 1.714285714285714e-07,
+      "logits/chosen": -4.158999919891357,
+      "logits/rejected": -4.378296852111816,
+      "logps/chosen": -320.1895446777344,
+      "logps/rejected": -273.6168518066406,
+      "loss": 0.4394,
+      "rewards/accuracies": 0.793749988079071,
+      "rewards/chosen": 1.6762861013412476,
+      "rewards/margins": 1.1370527744293213,
+      "rewards/rejected": 0.5392333269119263,
       "step": 30
     },
     {
       "epoch": 0.11,
-      "grad_norm": 1322.1199574632033,
-      "learning_rate": 4.9968724814144246e-08,
-      "logits/chosen": -4.296597480773926,
-      "logits/rejected": -4.558178901672363,
-      "logps/chosen": -277.67041015625,
-      "logps/rejected": -227.9638214111328,
-      "loss": 0.5887,
-      "rewards/accuracies": 0.731249988079071,
-      "rewards/chosen": 0.46597638726234436,
-      "rewards/margins": 0.30959779024124146,
-      "rewards/rejected": 0.1563786268234253,
+      "grad_norm": 1094.7268353859254,
+      "learning_rate": 1.9987489925657699e-07,
+      "logits/chosen": -4.327412128448486,
+      "logits/rejected": -4.590438365936279,
+      "logps/chosen": -272.52813720703125,
+      "logps/rejected": -226.4479522705078,
+      "loss": 0.3519,
+      "rewards/accuracies": 0.8374999761581421,
+      "rewards/chosen": 3.0370898246765137,
+      "rewards/margins": 2.1227898597717285,
+      "rewards/rejected": 0.9143003225326538,
       "step": 40
     },
     {
       "epoch": 0.14,
-      "grad_norm": 1226.7969075975607,
-      "learning_rate": 4.971899263739325e-08,
-      "logits/chosen": -4.339926719665527,
-      "logits/rejected": -4.592724323272705,
-      "logps/chosen": -270.9559020996094,
-      "logps/rejected": -228.2990264892578,
-      "loss": 0.5099,
-      "rewards/accuracies": 0.7437499761581421,
-      "rewards/chosen": 0.9494625329971313,
-      "rewards/margins": 0.5203765034675598,
-      "rewards/rejected": 0.4290861189365387,
+      "grad_norm": 1142.545986046605,
+      "learning_rate": 1.98875970549573e-07,
+      "logits/chosen": -4.374251365661621,
+      "logits/rejected": -4.627996921539307,
+      "logps/chosen": -264.49615478515625,
+      "logps/rejected": -226.27197265625,
+      "loss": 0.3689,
+      "rewards/accuracies": 0.831250011920929,
+      "rewards/chosen": 4.1793413162231445,
+      "rewards/margins": 2.736725091934204,
+      "rewards/rejected": 1.4426158666610718,
       "step": 50
     },
     {
       "epoch": 0.17,
-      "grad_norm": 929.2104534934765,
-      "learning_rate": 4.9222026055025726e-08,
-      "logits/chosen": -4.333067893981934,
-      "logits/rejected": -4.564365386962891,
-      "logps/chosen": -303.43310546875,
-      "logps/rejected": -252.1361846923828,
-      "loss": 0.4317,
-      "rewards/accuracies": 0.762499988079071,
-      "rewards/chosen": 1.6878736019134521,
-      "rewards/margins": 1.0692347288131714,
-      "rewards/rejected": 0.6186389923095703,
+      "grad_norm": 635.1242404757879,
+      "learning_rate": 1.968881042201029e-07,
+      "logits/chosen": -4.35978889465332,
+      "logits/rejected": -4.594507694244385,
+      "logps/chosen": -296.86907958984375,
+      "logps/rejected": -251.1016845703125,
+      "loss": 0.3246,
+      "rewards/accuracies": 0.8062499761581421,
+      "rewards/chosen": 4.9698944091796875,
+      "rewards/margins": 3.8340022563934326,
+      "rewards/rejected": 1.1358922719955444,
       "step": 60
     },
     {
       "epoch": 0.2,
-      "grad_norm": 1085.9024365118285,
-      "learning_rate": 4.8482795627104736e-08,
-      "logits/chosen": -4.332296371459961,
-      "logits/rejected": -4.567011833190918,
-      "logps/chosen": -289.8699645996094,
-      "logps/rejected": -241.077880859375,
-      "loss": 0.4089,
-      "rewards/accuracies": 0.862500011920929,
-      "rewards/chosen": 1.8221668004989624,
-      "rewards/margins": 1.204461693763733,
-      "rewards/rejected": 0.6177049875259399,
+      "grad_norm": 871.5895225513464,
+      "learning_rate": 1.9393118250841894e-07,
+      "logits/chosen": -4.349625587463379,
+      "logits/rejected": -4.586968421936035,
+      "logps/chosen": -285.0860595703125,
+      "logps/rejected": -241.4550018310547,
+      "loss": 0.3259,
+      "rewards/accuracies": 0.8812500238418579,
+      "rewards/chosen": 4.21413516998291,
+      "rewards/margins": 3.7849984169006348,
+      "rewards/rejected": 0.42913690209388733,
       "step": 70
     },
     {
       "epoch": 0.23,
-      "grad_norm": 978.9007166471121,
-      "learning_rate": 4.750869498807735e-08,
-      "logits/chosen": -4.28688383102417,
-      "logits/rejected": -4.543013572692871,
-      "logps/chosen": -322.56048583984375,
-      "logps/rejected": -271.0233154296875,
-      "loss": 0.3625,
-      "rewards/accuracies": 0.875,
-      "rewards/chosen": 2.1792995929718018,
-      "rewards/margins": 1.4446518421173096,
-      "rewards/rejected": 0.7346473932266235,
+      "grad_norm": 869.2800598030351,
+      "learning_rate": 1.900347799523094e-07,
+      "logits/chosen": -4.277868747711182,
+      "logits/rejected": -4.535857677459717,
+      "logps/chosen": -318.2683410644531,
+      "logps/rejected": -272.18206787109375,
+      "loss": 0.2501,
+      "rewards/accuracies": 0.9375,
+      "rewards/chosen": 4.3253655433654785,
+      "rewards/margins": 4.170114040374756,
+      "rewards/rejected": 0.15525080263614655,
       "step": 80
     },
     {
       "epoch": 0.26,
-      "grad_norm": 755.2675019751182,
-      "learning_rate": 4.630946689712609e-08,
-      "logits/chosen": -4.393240928649902,
-      "logits/rejected": -4.636073112487793,
-      "logps/chosen": -283.79290771484375,
-      "logps/rejected": -234.5099334716797,
-      "loss": 0.3436,
-      "rewards/accuracies": 0.862500011920929,
-      "rewards/chosen": 2.397665500640869,
-      "rewards/margins": 1.638594388961792,
-      "rewards/rejected": 0.7590711116790771,
+      "grad_norm": 616.9171514391553,
+      "learning_rate": 1.8523786758850436e-07,
+      "logits/chosen": -4.401219844818115,
+      "logits/rejected": -4.6477251052856445,
+      "logps/chosen": -280.7421569824219,
+      "logps/rejected": -236.2926025390625,
+      "loss": 0.2859,
+      "rewards/accuracies": 0.856249988079071,
+      "rewards/chosen": 3.923023223876953,
+      "rewards/margins": 4.055254936218262,
+      "rewards/rejected": -0.13223108649253845,
       "step": 90
     },
     {
       "epoch": 0.29,
-      "grad_norm": 893.2972482207422,
-      "learning_rate": 4.4897105793046024e-08,
-      "logits/chosen": -4.302220344543457,
-      "logits/rejected": -4.582569122314453,
-      "logps/chosen": -270.37457275390625,
-      "logps/rejected": -229.5946502685547,
-      "loss": 0.3652,
-      "rewards/accuracies": 0.8374999761581421,
-      "rewards/chosen": 2.673680067062378,
-      "rewards/margins": 2.020587921142578,
-      "rewards/rejected": 0.6530919075012207,
+      "grad_norm": 735.4670744748103,
+      "learning_rate": 1.795884231721841e-07,
+      "logits/chosen": -4.318561553955078,
+      "logits/rejected": -4.602439880371094,
+      "logps/chosen": -267.45794677734375,
+      "logps/rejected": -232.0417938232422,
+      "loss": 0.3114,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": 4.1319708824157715,
+      "rewards/margins": 4.702456951141357,
+      "rewards/rejected": -0.5704857110977173,
       "step": 100
     },
     {
       "epoch": 0.29,
-      "eval_logits/chosen": -4.696198463439941,
-      "eval_logits/rejected": -4.924900054931641,
-      "eval_logps/chosen": -403.4751892089844,
-      "eval_logps/rejected": -519.0816650390625,
-      "eval_loss": 1.7488452196121216,
-      "eval_rewards/accuracies": 0.3515625,
-      "eval_rewards/chosen": -2.2159295082092285,
-      "eval_rewards/margins": -1.206247329711914,
-      "eval_rewards/rejected": -1.009682059288025,
-      "eval_runtime": 97.4797,
-      "eval_samples_per_second": 20.517,
-      "eval_steps_per_second": 0.328,
+      "eval_logits/chosen": -3.226163148880005,
+      "eval_logits/rejected": -3.226163148880005,
+      "eval_logps/chosen": -161.294677734375,
+      "eval_logps/rejected": -161.294677734375,
+      "eval_loss": 0.6931471824645996,
+      "eval_rewards/accuracies": 0.0,
+      "eval_rewards/chosen": -2.8588294982910156,
+      "eval_rewards/margins": 0.0,
+      "eval_rewards/rejected": -2.8588294982910156,
+      "eval_runtime": 1.5114,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.662,
       "step": 100
     },
     {
       "epoch": 0.31,
-      "grad_norm": 872.5882367590035,
-      "learning_rate": 4.328573782827409e-08,
-      "logits/chosen": -4.3329758644104,
-      "logits/rejected": -4.617272853851318,
-      "logps/chosen": -296.02618408203125,
-      "logps/rejected": -242.73428344726562,
-      "loss": 0.349,
-      "rewards/accuracies": 0.875,
-      "rewards/chosen": 2.7564496994018555,
-      "rewards/margins": 2.076282024383545,
-      "rewards/rejected": 0.6801677942276001,
+      "grad_norm": 817.8641381018862,
+      "learning_rate": 1.7314295131309637e-07,
+      "logits/chosen": -4.324714660644531,
+      "logits/rejected": -4.614729404449463,
+      "logps/chosen": -293.42364501953125,
+      "logps/rejected": -245.01620483398438,
+      "loss": 0.3079,
+      "rewards/accuracies": 0.8812500238418579,
+      "rewards/chosen": 4.0577192306518555,
+      "rewards/margins": 4.518521308898926,
+      "rewards/rejected": -0.4608024060726166,
       "step": 110
     },
     {
       "epoch": 0.34,
-      "grad_norm": 1043.3393138533181,
-      "learning_rate": 4.1491479581946166e-08,
-      "logits/chosen": -4.3465352058410645,
-      "logits/rejected": -4.575117111206055,
-      "logps/chosen": -300.99859619140625,
-      "logps/rejected": -265.1387023925781,
-      "loss": 0.3567,
-      "rewards/accuracies": 0.8500000238418579,
-      "rewards/chosen": 2.6209876537323,
-      "rewards/margins": 1.951716423034668,
-      "rewards/rejected": 0.6692714095115662,
+      "grad_norm": 926.642960527248,
+      "learning_rate": 1.6596591832778466e-07,
+      "logits/chosen": -4.334780216217041,
+      "logits/rejected": -4.567566394805908,
+      "logps/chosen": -298.56976318359375,
+      "logps/rejected": -266.8869323730469,
+      "loss": 0.2848,
+      "rewards/accuracies": 0.84375,
+      "rewards/chosen": 3.83539080619812,
+      "rewards/margins": 4.040234088897705,
+      "rewards/rejected": -0.20484237372875214,
       "step": 120
     },
     {
       "epoch": 0.37,
-      "grad_norm": 837.2392359904699,
-      "learning_rate": 3.953227686510564e-08,
-      "logits/chosen": -4.319238185882568,
-      "logits/rejected": -4.565236568450928,
-      "logps/chosen": -288.36688232421875,
-      "logps/rejected": -240.29177856445312,
-      "loss": 0.3089,
-      "rewards/accuracies": 0.887499988079071,
-      "rewards/chosen": 2.86527419090271,
-      "rewards/margins": 2.168524980545044,
-      "rewards/rejected": 0.6967490911483765,
+      "grad_norm": 675.9811999128359,
+      "learning_rate": 1.5812910746042256e-07,
+      "logits/chosen": -4.3064866065979,
+      "logits/rejected": -4.553826332092285,
+      "logps/chosen": -286.5328063964844,
+      "logps/rejected": -242.8630828857422,
+      "loss": 0.2782,
+      "rewards/accuracies": 0.893750011920929,
+      "rewards/chosen": 3.7822985649108887,
+      "rewards/margins": 4.371214389801025,
+      "rewards/rejected": -0.5889158248901367,
       "step": 130
     },
     {
       "epoch": 0.4,
-      "grad_norm": 1077.8316620275032,
-      "learning_rate": 3.7427725230301354e-08,
-      "logits/chosen": -4.310965538024902,
-      "logits/rejected": -4.550887584686279,
-      "logps/chosen": -295.83868408203125,
-      "logps/rejected": -240.91311645507812,
-      "loss": 0.3113,
-      "rewards/accuracies": 0.8812500238418579,
-      "rewards/chosen": 3.000100612640381,
-      "rewards/margins": 2.261690855026245,
-      "rewards/rejected": 0.7384099960327148,
+      "grad_norm": 1048.4484509272795,
+      "learning_rate": 1.4971090092120542e-07,
+      "logits/chosen": -4.304565906524658,
+      "logits/rejected": -4.545400142669678,
+      "logps/chosen": -294.17095947265625,
+      "logps/rejected": -243.83935546875,
+      "loss": 0.269,
+      "rewards/accuracies": 0.9125000238418579,
+      "rewards/chosen": 3.8339390754699707,
+      "rewards/margins": 4.558647155761719,
+      "rewards/rejected": -0.7247086763381958,
       "step": 140
     },
     {
       "epoch": 0.43,
-      "grad_norm": 805.166891537404,
-      "learning_rate": 3.5198873980801955e-08,
-      "logits/chosen": -4.3193488121032715,
-      "logits/rejected": -4.625092506408691,
-      "logps/chosen": -294.2840881347656,
-      "logps/rejected": -245.18655395507812,
-      "loss": 0.318,
-      "rewards/accuracies": 0.8374999761581421,
-      "rewards/chosen": 3.1946704387664795,
-      "rewards/margins": 2.1761879920959473,
-      "rewards/rejected": 1.0184824466705322,
+      "grad_norm": 844.0558457976633,
+      "learning_rate": 1.4079549592320782e-07,
+      "logits/chosen": -4.313709735870361,
+      "logits/rejected": -4.621119499206543,
+      "logps/chosen": -292.96661376953125,
+      "logps/rejected": -248.0418701171875,
+      "loss": 0.2727,
+      "rewards/accuracies": 0.856249988079071,
+      "rewards/chosen": 3.8533847332000732,
+      "rewards/margins": 4.262558937072754,
+      "rewards/rejected": -0.4091736674308777,
       "step": 150
     },
     {
       "epoch": 0.46,
-      "grad_norm": 711.9582620759091,
-      "learning_rate": 3.2868015639687205e-08,
-      "logits/chosen": -4.259413242340088,
-      "logits/rejected": -4.450949668884277,
-      "logps/chosen": -311.91326904296875,
-      "logps/rejected": -270.4508972167969,
-      "loss": 0.3344,
-      "rewards/accuracies": 0.8812500238418579,
-      "rewards/chosen": 3.2649455070495605,
-      "rewards/margins": 2.298511266708374,
-      "rewards/rejected": 0.966434121131897,
+      "grad_norm": 684.2161612155076,
+      "learning_rate": 1.3147206255874882e-07,
+      "logits/chosen": -4.248518943786621,
+      "logits/rejected": -4.44361686706543,
+      "logps/chosen": -309.7388610839844,
+      "logps/rejected": -272.90252685546875,
+      "loss": 0.2961,
+      "rewards/accuracies": 0.90625,
+      "rewards/chosen": 4.352166652679443,
+      "rewards/margins": 4.611546516418457,
+      "rewards/rejected": -0.2593800127506256,
       "step": 160
     },
     {
       "epoch": 0.49,
-      "grad_norm": 894.588925085696,
-      "learning_rate": 3.0458462984504134e-08,
-      "logits/chosen": -4.303664207458496,
-      "logits/rejected": -4.500936031341553,
-      "logps/chosen": -274.3706359863281,
-      "logps/rejected": -235.48678588867188,
-      "loss": 0.2963,
-      "rewards/accuracies": 0.856249988079071,
-      "rewards/chosen": 3.1083619594573975,
-      "rewards/margins": 2.440159559249878,
-      "rewards/rejected": 0.6682023406028748,
+      "grad_norm": 795.319436926025,
+      "learning_rate": 1.2183385193801653e-07,
+      "logits/chosen": -4.289888381958008,
+      "logits/rejected": -4.489100933074951,
+      "logps/chosen": -272.3153991699219,
+      "logps/rejected": -237.64114379882812,
+      "loss": 0.2547,
+      "rewards/accuracies": 0.918749988079071,
+      "rewards/chosen": 4.135977745056152,
+      "rewards/margins": 4.544943809509277,
+      "rewards/rejected": -0.40896692872047424,
       "step": 170
     },
     {
       "epoch": 0.52,
-      "grad_norm": 787.4594621489149,
-      "learning_rate": 2.7994315877542628e-08,
-      "logits/chosen": -4.386083602905273,
-      "logits/rejected": -4.545100212097168,
-      "logps/chosen": -263.3323059082031,
-      "logps/rejected": -228.91329956054688,
-      "loss": 0.329,
-      "rewards/accuracies": 0.8500000238418579,
-      "rewards/chosen": 2.6448895931243896,
-      "rewards/margins": 2.054766893386841,
-      "rewards/rejected": 0.5901231169700623,
+      "grad_norm": 917.2913745554154,
+      "learning_rate": 1.1197726351017051e-07,
+      "logits/chosen": -4.37355375289917,
+      "logits/rejected": -4.5346221923828125,
+      "logps/chosen": -261.6694641113281,
+      "logps/rejected": -230.89599609375,
+      "loss": 0.3104,
+      "rewards/accuracies": 0.862500011920929,
+      "rewards/chosen": 3.4763107299804688,
+      "rewards/margins": 3.877528429031372,
+      "rewards/rejected": -0.4012181758880615,
       "step": 180
     },
     {
       "epoch": 0.54,
-      "grad_norm": 1024.0437070924656,
-      "learning_rate": 2.5500220223847356e-08,
-      "logits/chosen": -4.346238613128662,
-      "logits/rejected": -4.559948444366455,
-      "logps/chosen": -290.6450500488281,
-      "logps/rejected": -252.0496063232422,
-      "loss": 0.3385,
-      "rewards/accuracies": 0.8374999761581421,
-      "rewards/chosen": 3.277946949005127,
-      "rewards/margins": 2.545849323272705,
-      "rewards/rejected": 0.7320975661277771,
+      "grad_norm": 880.6071936953809,
+      "learning_rate": 1.0200088089538943e-07,
+      "logits/chosen": -4.33815860748291,
+      "logits/rejected": -4.5543646812438965,
+      "logps/chosen": -288.26873779296875,
+      "logps/rejected": -253.7147216796875,
+      "loss": 0.3129,
+      "rewards/accuracies": 0.856249988079071,
+      "rewards/chosen": 4.4660868644714355,
+      "rewards/margins": 4.566544055938721,
+      "rewards/rejected": -0.10045762360095978,
       "step": 190
     },
     {
       "epoch": 0.57,
-      "grad_norm": 937.2977444102543,
-      "learning_rate": 2.3001121467819626e-08,
-      "logits/chosen": -4.290076732635498,
-      "logits/rejected": -4.555208683013916,
-      "logps/chosen": -317.6783447265625,
-      "logps/rejected": -267.9818115234375,
-      "loss": 0.312,
-      "rewards/accuracies": 0.824999988079071,
-      "rewards/chosen": 3.4544365406036377,
-      "rewards/margins": 2.3504996299743652,
-      "rewards/rejected": 1.1039369106292725,
+      "grad_norm": 781.3066688232569,
+      "learning_rate": 9.20044858712785e-08,
+      "logits/chosen": -4.292983055114746,
+      "logits/rejected": -4.566291332244873,
+      "logps/chosen": -315.58251953125,
+      "logps/rejected": -270.0286560058594,
+      "loss": 0.2741,
+      "rewards/accuracies": 0.862500011920929,
+      "rewards/chosen": 4.502341270446777,
+      "rewards/margins": 4.4218339920043945,
+      "rewards/rejected": 0.08050797879695892,
       "step": 200
     },
     {
       "epoch": 0.57,
-      "eval_logits/chosen": -4.7096052169799805,
-      "eval_logits/rejected": -4.93910026550293,
-      "eval_logps/chosen": -405.4331970214844,
-      "eval_logps/rejected": -520.6950073242188,
-      "eval_loss": 1.959585189819336,
-      "eval_rewards/accuracies": 0.33984375,
-      "eval_rewards/chosen": -3.194929599761963,
-      "eval_rewards/margins": -1.3785794973373413,
-      "eval_rewards/rejected": -1.8163501024246216,
-      "eval_runtime": 97.467,
-      "eval_samples_per_second": 20.52,
-      "eval_steps_per_second": 0.328,
+      "eval_logits/chosen": -3.2033309936523438,
+      "eval_logits/rejected": -3.2033309936523438,
+      "eval_logps/chosen": -160.92892456054688,
+      "eval_logps/rejected": -160.92892456054688,
+      "eval_loss": 0.6931471824645996,
+      "eval_rewards/accuracies": 0.0,
+      "eval_rewards/chosen": -2.675952911376953,
+      "eval_rewards/margins": 0.0,
+      "eval_rewards/rejected": -2.675952911376953,
+      "eval_runtime": 1.48,
+      "eval_samples_per_second": 0.676,
+      "eval_steps_per_second": 0.676,
       "step": 200
     },
     {
       "epoch": 0.6,
-      "grad_norm": 839.0813036142636,
-      "learning_rate": 2.0522015093886612e-08,
-      "logits/chosen": -4.317226409912109,
-      "logits/rejected": -4.618459224700928,
-      "logps/chosen": -294.03253173828125,
-      "logps/rejected": -234.196533203125,
-      "loss": 0.2938,
-      "rewards/accuracies": 0.8500000238418579,
-      "rewards/chosen": 3.499882459640503,
-      "rewards/margins": 2.7310853004455566,
-      "rewards/rejected": 0.7687975168228149,
+      "grad_norm": 1021.5696401893487,
+      "learning_rate": 8.208806037554645e-08,
+      "logits/chosen": -4.327147483825684,
+      "logits/rejected": -4.6312479972839355,
+      "logps/chosen": -292.85919189453125,
+      "logps/rejected": -236.7397918701172,
+      "loss": 0.2345,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": 4.086556434631348,
+      "rewards/margins": 4.589372634887695,
+      "rewards/rejected": -0.5028160810470581,
       "step": 210
     },
     {
       "epoch": 0.63,
-      "grad_norm": 996.0408528817461,
-      "learning_rate": 1.808769662668035e-08,
-      "logits/chosen": -4.358494758605957,
-      "logits/rejected": -4.704705238342285,
-      "logps/chosen": -299.7410888671875,
-      "logps/rejected": -235.092041015625,
-      "loss": 0.2943,
-      "rewards/accuracies": 0.8999999761581421,
-      "rewards/chosen": 3.0801563262939453,
-      "rewards/margins": 2.7920236587524414,
-      "rewards/rejected": 0.28813308477401733,
+      "grad_norm": 720.6830161967047,
+      "learning_rate": 7.23507865067214e-08,
+      "logits/chosen": -4.357525825500488,
+      "logits/rejected": -4.7082133293151855,
+      "logps/chosen": -298.2110900878906,
+      "logps/rejected": -237.28903198242188,
+      "loss": 0.241,
+      "rewards/accuracies": 0.9125000238418579,
+      "rewards/chosen": 3.84515118598938,
+      "rewards/margins": 4.655531406402588,
+      "rewards/rejected": -0.8103805780410767,
       "step": 220
     },
     {
       "epoch": 0.66,
-      "grad_norm": 781.9269603792492,
-      "learning_rate": 1.5722513631174444e-08,
-      "logits/chosen": -4.392434597015381,
-      "logits/rejected": -4.748046398162842,
-      "logps/chosen": -289.19927978515625,
-      "logps/rejected": -234.593017578125,
-      "loss": 0.3084,
-      "rewards/accuracies": 0.856249988079071,
-      "rewards/chosen": 2.994288921356201,
-      "rewards/margins": 2.508836507797241,
-      "rewards/rejected": 0.4854525625705719,
+      "grad_norm": 784.9312094778309,
+      "learning_rate": 6.289005452469778e-08,
+      "logits/chosen": -4.388636112213135,
+      "logits/rejected": -4.748753070831299,
+      "logps/chosen": -288.0061340332031,
+      "logps/rejected": -236.6077880859375,
+      "loss": 0.268,
+      "rewards/accuracies": 0.9125000238418579,
+      "rewards/chosen": 3.5908846855163574,
+      "rewards/margins": 4.1128339767456055,
+      "rewards/rejected": -0.5219489932060242,
       "step": 230
     },
     {
       "epoch": 0.69,
-      "grad_norm": 795.1407214961006,
-      "learning_rate": 1.345012219322345e-08,
-      "logits/chosen": -4.238420486450195,
-      "logits/rejected": -4.548759460449219,
-      "logps/chosen": -285.19232177734375,
-      "logps/rejected": -240.72509765625,
-      "loss": 0.3269,
-      "rewards/accuracies": 0.8687499761581421,
-      "rewards/chosen": 2.8941564559936523,
-      "rewards/margins": 2.3365750312805176,
-      "rewards/rejected": 0.5575811266899109,
+      "grad_norm": 765.61870702253,
+      "learning_rate": 5.38004887728938e-08,
+      "logits/chosen": -4.234927177429199,
+      "logits/rejected": -4.550938606262207,
+      "logps/chosen": -283.8558654785156,
+      "logps/rejected": -242.86538696289062,
+      "loss": 0.2908,
+      "rewards/accuracies": 0.887499988079071,
+      "rewards/chosen": 3.562377452850342,
+      "rewards/margins": 4.074954032897949,
+      "rewards/rejected": -0.5125768780708313,
       "step": 240
     },
     {
       "epoch": 0.72,
-      "grad_norm": 879.951904317175,
-      "learning_rate": 1.1293250316137664e-08,
-      "logits/chosen": -4.293431282043457,
-      "logits/rejected": -4.5221476554870605,
-      "logps/chosen": -294.2128601074219,
-      "logps/rejected": -254.7281494140625,
-      "loss": 0.3131,
-      "rewards/accuracies": 0.856249988079071,
-      "rewards/chosen": 3.089404821395874,
-      "rewards/margins": 2.4765713214874268,
-      "rewards/rejected": 0.6128337383270264,
+      "grad_norm": 787.5249084258048,
+      "learning_rate": 4.517300126455066e-08,
+      "logits/chosen": -4.290322303771973,
+      "logits/rejected": -4.522026538848877,
+      "logps/chosen": -292.47442626953125,
+      "logps/rejected": -257.0675354003906,
+      "loss": 0.2635,
+      "rewards/accuracies": 0.8812500238418579,
+      "rewards/chosen": 3.95862078666687,
+      "rewards/margins": 4.515480995178223,
+      "rewards/rejected": -0.5568601489067078,
       "step": 250
     },
     {
       "epoch": 0.74,
-      "grad_norm": 1094.0422336259135,
-      "learning_rate": 9.273470599753375e-09,
-      "logits/chosen": -4.372658729553223,
-      "logits/rejected": -4.525150775909424,
-      "logps/chosen": -294.89874267578125,
-      "logps/rejected": -263.60577392578125,
-      "loss": 0.3409,
-      "rewards/accuracies": 0.831250011920929,
-      "rewards/chosen": 2.7965197563171387,
-      "rewards/margins": 2.07285737991333,
-      "rewards/rejected": 0.7236624956130981,
+      "grad_norm": 894.7487355536125,
+      "learning_rate": 3.70938823990135e-08,
+      "logits/chosen": -4.367636680603027,
+      "logits/rejected": -4.524864196777344,
+      "logps/chosen": -293.8681640625,
+      "logps/rejected": -266.19622802734375,
+      "loss": 0.2999,
+      "rewards/accuracies": 0.887499988079071,
+      "rewards/chosen": 3.311814069747925,
+      "rewards/margins": 3.883411407470703,
+      "rewards/rejected": -0.5715969800949097,
       "step": 260
     },
     {
       "epoch": 0.77,
-      "grad_norm": 775.0184329679462,
-      "learning_rate": 7.410984475616819e-09,
-      "logits/chosen": -4.244694709777832,
-      "logits/rejected": -4.5279622077941895,
-      "logps/chosen": -282.0103454589844,
-      "logps/rejected": -239.31399536132812,
-      "loss": 0.2834,
-      "rewards/accuracies": 0.856249988079071,
-      "rewards/chosen": 3.1753017902374268,
-      "rewards/margins": 2.7872982025146484,
-      "rewards/rejected": 0.38800328969955444,
+      "grad_norm": 582.5138549963486,
+      "learning_rate": 2.9643937902467277e-08,
+      "logits/chosen": -4.23429012298584,
+      "logits/rejected": -4.521827697753906,
+      "logps/chosen": -280.5379333496094,
+      "logps/rejected": -241.7247314453125,
+      "loss": 0.2388,
+      "rewards/accuracies": 0.862500011920929,
+      "rewards/chosen": 3.911522626876831,
+      "rewards/margins": 4.728894233703613,
+      "rewards/rejected": -0.8173705339431763,
       "step": 270
     },
     {
       "epoch": 0.8,
-      "grad_norm": 796.3667611732337,
-      "learning_rate": 5.724420156318405e-09,
-      "logits/chosen": -4.291975498199463,
-      "logits/rejected": -4.621099472045898,
-      "logps/chosen": -285.0769958496094,
-      "logps/rejected": -232.0653839111328,
-      "loss": 0.3144,
-      "rewards/accuracies": 0.8687499761581421,
-      "rewards/chosen": 2.847846508026123,
-      "rewards/margins": 2.3038220405578613,
-      "rewards/rejected": 0.5440241098403931,
+      "grad_norm": 810.3954826895692,
+      "learning_rate": 2.289768062527362e-08,
+      "logits/chosen": -4.273584842681885,
+      "logits/rejected": -4.6050519943237305,
+      "logps/chosen": -283.619384765625,
+      "logps/rejected": -233.9824981689453,
+      "loss": 0.2895,
+      "rewards/accuracies": 0.8812500238418579,
+      "rewards/chosen": 3.576643466949463,
+      "rewards/margins": 3.991179943084717,
+      "rewards/rejected": -0.41453617811203003,
       "step": 280
     },
     {
       "epoch": 0.83,
-      "grad_norm": 1101.8200803185337,
-      "learning_rate": 4.230646319847259e-09,
-      "logits/chosen": -4.424475193023682,
-      "logits/rejected": -4.705140113830566,
-      "logps/chosen": -295.53857421875,
-      "logps/rejected": -232.05245971679688,
-      "loss": 0.3121,
-      "rewards/accuracies": 0.893750011920929,
-      "rewards/chosen": 2.7051048278808594,
-      "rewards/margins": 2.4193003177642822,
-      "rewards/rejected": 0.2858046591281891,
+      "grad_norm": 730.6997584190797,
+      "learning_rate": 1.6922585279389035e-08,
+      "logits/chosen": -4.406073093414307,
+      "logits/rejected": -4.6901140213012695,
+      "logps/chosen": -294.35699462890625,
+      "logps/rejected": -233.74594116210938,
+      "loss": 0.2669,
+      "rewards/accuracies": 0.90625,
+      "rewards/chosen": 3.295900344848633,
+      "rewards/margins": 3.856853485107422,
+      "rewards/rejected": -0.5609533786773682,
       "step": 290
     },
     {
       "epoch": 0.86,
-      "grad_norm": 1165.1379404537536,
-      "learning_rate": 2.944603392457931e-09,
-      "logits/chosen": -4.335474491119385,
-      "logits/rejected": -4.49271297454834,
-      "logps/chosen": -292.6704406738281,
-      "logps/rejected": -252.3407440185547,
-      "loss": 0.2993,
-      "rewards/accuracies": 0.831250011920929,
-      "rewards/chosen": 3.1620662212371826,
-      "rewards/margins": 2.2422008514404297,
-      "rewards/rejected": 0.9198653101921082,
+      "grad_norm": 1254.3257141110294,
+      "learning_rate": 1.1778413569831725e-08,
+      "logits/chosen": -4.323153495788574,
+      "logits/rejected": -4.481599807739258,
+      "logps/chosen": -291.29742431640625,
+      "logps/rejected": -254.1814422607422,
+      "loss": 0.2695,
+      "rewards/accuracies": 0.856249988079071,
+      "rewards/chosen": 3.8485710620880127,
+      "rewards/margins": 3.8490569591522217,
+      "rewards/rejected": -0.0004864335060119629,
       "step": 300
     },
     {
       "epoch": 0.86,
-      "eval_logits/chosen": -4.706819534301758,
-      "eval_logits/rejected": -4.936363697052002,
-      "eval_logps/chosen": -405.887939453125,
-      "eval_logps/rejected": -521.2874755859375,
-      "eval_loss": 1.9827619791030884,
-      "eval_rewards/accuracies": 0.35546875,
-      "eval_rewards/chosen": -3.4222922325134277,
-      "eval_rewards/margins": -1.3097174167633057,
-      "eval_rewards/rejected": -2.112574815750122,
-      "eval_runtime": 97.4214,
-      "eval_samples_per_second": 20.529,
-      "eval_steps_per_second": 0.328,
+      "eval_logits/chosen": -3.199521064758301,
+      "eval_logits/rejected": -3.199521064758301,
+      "eval_logps/chosen": -159.83721923828125,
+      "eval_logps/rejected": -159.83721923828125,
+      "eval_loss": 0.6931471824645996,
+      "eval_rewards/accuracies": 0.0,
+      "eval_rewards/chosen": -2.1301021575927734,
+      "eval_rewards/margins": 0.0,
+      "eval_rewards/rejected": -2.1301021575927734,
+      "eval_runtime": 1.4681,
+      "eval_samples_per_second": 0.681,
+      "eval_steps_per_second": 0.681,
       "step": 300
     },
     {
       "epoch": 0.89,
-      "grad_norm": 1000.5341122694366,
-      "learning_rate": 1.8791541175240787e-09,
-      "logits/chosen": -4.3908562660217285,
-      "logits/rejected": -4.607417583465576,
-      "logps/chosen": -301.6288146972656,
-      "logps/rejected": -246.2473907470703,
-      "loss": 0.2942,
-      "rewards/accuracies": 0.862500011920929,
-      "rewards/chosen": 3.3309147357940674,
-      "rewards/margins": 2.577249765396118,
-      "rewards/rejected": 0.753665030002594,
+      "grad_norm": 1302.095010227008,
+      "learning_rate": 7.516616470096315e-09,
+      "logits/chosen": -4.3807525634765625,
+      "logits/rejected": -4.600883483886719,
+      "logps/chosen": -299.8600158691406,
+      "logps/rejected": -248.4993438720703,
+      "loss": 0.2545,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": 4.215323448181152,
+      "rewards/margins": 4.587640762329102,
+      "rewards/rejected": -0.3723169267177582,
       "step": 310
     },
     {
       "epoch": 0.92,
-      "grad_norm": 942.4369602130394,
-      "learning_rate": 1.0449549049596136e-09,
-      "logits/chosen": -4.230744361877441,
-      "logits/rejected": -4.579751968383789,
-      "logps/chosen": -287.4718322753906,
-      "logps/rejected": -232.3734893798828,
-      "loss": 0.3078,
-      "rewards/accuracies": 0.862500011920929,
-      "rewards/chosen": 3.2846877574920654,
-      "rewards/margins": 2.4689810276031494,
-      "rewards/rejected": 0.8157066106796265,
+      "grad_norm": 875.4939401121867,
+      "learning_rate": 4.179819619838454e-09,
+      "logits/chosen": -4.2180280685424805,
+      "logits/rejected": -4.570239067077637,
+      "logps/chosen": -286.0614013671875,
+      "logps/rejected": -234.3032684326172,
+      "loss": 0.269,
+      "rewards/accuracies": 0.8374999761581421,
+      "rewards/chosen": 3.9898929595947266,
+      "rewards/margins": 4.13907527923584,
+      "rewards/rejected": -0.1491830050945282,
       "step": 320
     },
     {
       "epoch": 0.94,
-      "grad_norm": 1041.373917696444,
-      "learning_rate": 4.5034924794443707e-10,
-      "logits/chosen": -4.3931074142456055,
-      "logits/rejected": -4.620477199554443,
-      "logps/chosen": -280.0606994628906,
-      "logps/rejected": -233.14346313476562,
-      "loss": 0.2917,
-      "rewards/accuracies": 0.856249988079071,
-      "rewards/chosen": 3.2919604778289795,
-      "rewards/margins": 2.5820116996765137,
-      "rewards/rejected": 0.7099487781524658,
+      "grad_norm": 1010.1848642033616,
+      "learning_rate": 1.8013969917777483e-09,
+      "logits/chosen": -4.380679130554199,
+      "logits/rejected": -4.608093738555908,
+      "logps/chosen": -278.8153381347656,
+      "logps/rejected": -235.4047088623047,
+      "loss": 0.247,
+      "rewards/accuracies": 0.887499988079071,
+      "rewards/chosen": 3.914651870727539,
+      "rewards/margins": 4.33533239364624,
+      "rewards/rejected": -0.42068085074424744,
       "step": 330
     },
     {
       "epoch": 0.97,
-      "grad_norm": 859.9709811153568,
-      "learning_rate": 1.0128427297940723e-10,
-      "logits/chosen": -4.328027248382568,
-      "logits/rejected": -4.5136332511901855,
-      "logps/chosen": -284.2748718261719,
-      "logps/rejected": -244.03817749023438,
-      "loss": 0.2952,
-      "rewards/accuracies": 0.862500011920929,
-      "rewards/chosen": 3.204352855682373,
-      "rewards/margins": 2.4567666053771973,
-      "rewards/rejected": 0.7475861310958862,
+      "grad_norm": 702.698454631269,
+      "learning_rate": 4.051370919176289e-10,
+      "logits/chosen": -4.314477920532227,
+      "logits/rejected": -4.502932548522949,
+      "logps/chosen": -282.89666748046875,
+      "logps/rejected": -246.3631134033203,
+      "loss": 0.243,
+      "rewards/accuracies": 0.90625,
+      "rewards/chosen": 3.8934428691864014,
+      "rewards/margins": 4.308337688446045,
+      "rewards/rejected": -0.41489481925964355,
       "step": 340
     },
     {
       "epoch": 1.0,
       "step": 349,
       "total_flos": 0.0,
-      "train_loss": 0.3696309270011661,
-      "train_runtime": 5591.2149,
-      "train_samples_per_second": 7.991,
-      "train_steps_per_second": 0.062
+      "train_loss": 0.30806411576476,
+      "train_runtime": 5299.0292,
+      "train_samples_per_second": 8.432,
+      "train_steps_per_second": 0.066
     }
   ],
   "logging_steps": 10,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3101c501cb9463cb757d57fc1ba9afd767800873490708bfba47496b8c2a73c9
-size 6520
+oid sha256:b3f943ec10c03c0b36e6f7cf63a55e819cb580b98a2a33846c2e1059e1f696c2
+size 6328