Training in progress, step 13200

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d5e446fa4d8a8bacef6da1c91f421e9f87be43e206bc7b9990571f79b45e60d0
+oid sha256:31342273fb602a0e10624e2c9852841a2091a5260f85fb0f0a0089eb3262870b
 size 88891320

last-checkpoint/adapter_config.json

@@ -30,15 +30,15 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "w3",
-    "o_proj",
     "w2",
-    "v_proj",
     "in_proj",
-    "k_proj",
-    "w1",
     "q_proj",
-    "out_proj"
+    "w1",
+    "k_proj",
+    "v_proj",
+    "o_proj",
+    "out_proj",
+    "w3"
   ],
   "target_parameters": null,
   "task_type": "CAUSAL_LM",

last-checkpoint/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d5e446fa4d8a8bacef6da1c91f421e9f87be43e206bc7b9990571f79b45e60d0
+oid sha256:d59a26df247a9b3bd089944c699df5332df994dda9c303c5de94dc6e76616eb9
 size 88891320

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9ebede7fb03cef2a8824b554904f7673360bc6df18d358569dbf0aacc1598b36
+oid sha256:ca1ac60d2a733c3f6db08f04b7d8d4f6547bd586f287c155721433e984edc7db
 size 177891851

last-checkpoint/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dfb9d2e6ca1ab82f88c8161656f06fa5568726a52b847e8b4bd05510e86e2154
+oid sha256:e6b6744a57f42273e871a115e82b156c17e0c93e4839b3bac6dc6f53f8600e91
 size 14917

last-checkpoint/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6c476c226715fe8efc3e1eecf1d728c3e93abbcac806d565885a031a5ecaafaf
+oid sha256:79bfae6c501e0b8cd471ca11801529a5af1b0faa0b363fbd03730ca881075ee9
 size 14917

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:85fca9af1c746cc4034e53686fe77cfc7e3022d0dfdd46017391d6be803268a9
+oid sha256:cb1f23b3450b5ea7f3c2c09c828367f5357644ec38f3b3af648f0d2bda902fef
 size 1465

last-checkpoint/trainer_state.json

@@ -1,10 +1,10 @@
 {
-  "best_global_step": 13000,
-  "best_metric": 0.2847670912742615,
-  "best_model_checkpoint": "./lfm25-kodcode-lora/checkpoint-13000",
-  "epoch": 0.9848671376351825,
+  "best_global_step": 10750,
+  "best_metric": 0.28601643443107605,
+  "best_model_checkpoint": "./lfm25-kodcode-lora/checkpoint-10750",
+  "epoch": 0.8144093638137085,
   "eval_steps": 250,
-  "global_step": 13000,
+  "global_step": 10750,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -11231,2355 +11231,6 @@
       "eval_samples_per_second": 5.851,
       "eval_steps_per_second": 1.464,
       "step": 10750
-    },
-    {
-      "entropy": 0.2615930700674653,
-      "epoch": 0.8151669539195818,
-      "grad_norm": 0.4948700964450836,
-      "learning_rate": 8.70287588830654e-06,
-      "loss": 0.30023128986358644,
-      "mean_token_accuracy": 0.9172126367688179,
-      "num_tokens": 95394.0,
-      "step": 10760
-    },
-    {
-      "entropy": 0.26154228337109087,
-      "epoch": 0.815924544025455,
-      "grad_norm": 0.3987369239330292,
-      "learning_rate": 8.633838805669543e-06,
-      "loss": 0.3083802223205566,
-      "mean_token_accuracy": 0.916927844285965,
-      "num_tokens": 198965.0,
-      "step": 10770
-    },
-    {
-      "entropy": 0.29399398546665906,
-      "epoch": 0.8166821341313283,
-      "grad_norm": 0.4157302677631378,
-      "learning_rate": 8.565050754034847e-06,
-      "loss": 0.330983567237854,
-      "mean_token_accuracy": 0.907181803882122,
-      "num_tokens": 300936.0,
-      "step": 10780
-    },
-    {
-      "entropy": 0.2811663061380386,
-      "epoch": 0.8174397242372015,
-      "grad_norm": 0.41286197304725647,
-      "learning_rate": 8.496512147517688e-06,
-      "loss": 0.3124821662902832,
-      "mean_token_accuracy": 0.9098848760128021,
-      "num_tokens": 407545.0,
-      "step": 10790
-    },
-    {
-      "entropy": 0.29347006008028986,
-      "epoch": 0.8181973143430746,
-      "grad_norm": 0.4638175666332245,
-      "learning_rate": 8.428223398731588e-06,
-      "loss": 0.32779746055603026,
-      "mean_token_accuracy": 0.9062278017401695,
-      "num_tokens": 511306.0,
-      "step": 10800
-    },
-    {
-      "entropy": 0.3169729994609952,
-      "epoch": 0.8189549044489479,
-      "grad_norm": 0.6172231435775757,
-      "learning_rate": 8.360184918785902e-06,
-      "loss": 0.3641566038131714,
-      "mean_token_accuracy": 0.8966098591685295,
-      "num_tokens": 615936.0,
-      "step": 10810
-    },
-    {
-      "entropy": 0.28694524206221106,
-      "epoch": 0.8197124945548211,
-      "grad_norm": 0.5250155925750732,
-      "learning_rate": 8.292397117283318e-06,
-      "loss": 0.3268028974533081,
-      "mean_token_accuracy": 0.9117160573601722,
-      "num_tokens": 707912.0,
-      "step": 10820
-    },
-    {
-      "entropy": 0.2993797710165381,
-      "epoch": 0.8204700846606944,
-      "grad_norm": 0.450602650642395,
-      "learning_rate": 8.224860402317398e-06,
-      "loss": 0.3751563549041748,
-      "mean_token_accuracy": 0.9021484076976776,
-      "num_tokens": 797927.0,
-      "step": 10830
-    },
-    {
-      "entropy": 0.2887501694262028,
-      "epoch": 0.8212276747665676,
-      "grad_norm": 0.3982142210006714,
-      "learning_rate": 8.157575180470139e-06,
-      "loss": 0.3479593276977539,
-      "mean_token_accuracy": 0.9077468276023865,
-      "num_tokens": 904264.0,
-      "step": 10840
-    },
-    {
-      "entropy": 0.30117298644036056,
-      "epoch": 0.8219852648724407,
-      "grad_norm": 0.4184916913509369,
-      "learning_rate": 8.090541856809465e-06,
-      "loss": 0.33893485069274903,
-      "mean_token_accuracy": 0.9050878256559372,
-      "num_tokens": 1007433.0,
-      "step": 10850
-    },
-    {
-      "entropy": 0.30625202376395466,
-      "epoch": 0.822742854978314,
-      "grad_norm": 0.43937817215919495,
-      "learning_rate": 8.02376083488688e-06,
-      "loss": 0.34802818298339844,
-      "mean_token_accuracy": 0.9029570311307907,
-      "num_tokens": 1106432.0,
-      "step": 10860
-    },
-    {
-      "entropy": 0.29251224398612974,
-      "epoch": 0.8235004450841872,
-      "grad_norm": 0.4894697964191437,
-      "learning_rate": 7.957232516734963e-06,
-      "loss": 0.35279672145843505,
-      "mean_token_accuracy": 0.9066073164343834,
-      "num_tokens": 1204980.0,
-      "step": 10870
-    },
-    {
-      "entropy": 0.3121814550831914,
-      "epoch": 0.8242580351900605,
-      "grad_norm": 0.451427698135376,
-      "learning_rate": 7.890957302865e-06,
-      "loss": 0.37927200794219973,
-      "mean_token_accuracy": 0.8981422513723374,
-      "num_tokens": 1305202.0,
-      "step": 10880
-    },
-    {
-      "entropy": 0.29268699828535316,
-      "epoch": 0.8250156252959336,
-      "grad_norm": 0.3992355465888977,
-      "learning_rate": 7.82493559226451e-06,
-      "loss": 0.3250342607498169,
-      "mean_token_accuracy": 0.9062216132879257,
-      "num_tokens": 1407868.0,
-      "step": 10890
-    },
-    {
-      "entropy": 0.36002040691673753,
-      "epoch": 0.8257732154018068,
-      "grad_norm": 0.4525466561317444,
-      "learning_rate": 7.759167782394916e-06,
-      "loss": 0.4119568824768066,
-      "mean_token_accuracy": 0.8839563131332397,
-      "num_tokens": 1513950.0,
-      "step": 10900
-    },
-    {
-      "entropy": 0.30947884283959864,
-      "epoch": 0.8265308055076801,
-      "grad_norm": 0.4585402309894562,
-      "learning_rate": 7.69365426918911e-06,
-      "loss": 0.3610663414001465,
-      "mean_token_accuracy": 0.9016437903046608,
-      "num_tokens": 1613270.0,
-      "step": 10910
-    },
-    {
-      "entropy": 0.29806773774325845,
-      "epoch": 0.8272883956135533,
-      "grad_norm": 0.43951550126075745,
-      "learning_rate": 7.628395447049058e-06,
-      "loss": 0.34964966773986816,
-      "mean_token_accuracy": 0.9043863728642464,
-      "num_tokens": 1710564.0,
-      "step": 10920
-    },
-    {
-      "entropy": 0.27706831842660906,
-      "epoch": 0.8280459857194266,
-      "grad_norm": 0.43634042143821716,
-      "learning_rate": 7.563391708843465e-06,
-      "loss": 0.31148567199707033,
-      "mean_token_accuracy": 0.9122755557298661,
-      "num_tokens": 1804996.0,
-      "step": 10930
-    },
-    {
-      "entropy": 0.33176610227674247,
-      "epoch": 0.8288035758252997,
-      "grad_norm": 0.39294061064720154,
-      "learning_rate": 7.498643445905385e-06,
-      "loss": 0.37499704360961916,
-      "mean_token_accuracy": 0.8953326553106308,
-      "num_tokens": 1914019.0,
-      "step": 10940
-    },
-    {
-      "entropy": 0.2956154864281416,
-      "epoch": 0.8295611659311729,
-      "grad_norm": 0.4650717079639435,
-      "learning_rate": 7.434151048029858e-06,
-      "loss": 0.3409877777099609,
-      "mean_token_accuracy": 0.9076192870736122,
-      "num_tokens": 2009215.0,
-      "step": 10950
-    },
-    {
-      "entropy": 0.30589213948696853,
-      "epoch": 0.8303187560370462,
-      "grad_norm": 0.422702819108963,
-      "learning_rate": 7.369914903471592e-06,
-      "loss": 0.3454046487808228,
-      "mean_token_accuracy": 0.9033974334597588,
-      "num_tokens": 2113318.0,
-      "step": 10960
-    },
-    {
-      "entropy": 0.2733181806281209,
-      "epoch": 0.8310763461429194,
-      "grad_norm": 0.47341540455818176,
-      "learning_rate": 7.305935398942598e-06,
-      "loss": 0.3034660816192627,
-      "mean_token_accuracy": 0.9125015497207641,
-      "num_tokens": 2212212.0,
-      "step": 10970
-    },
-    {
-      "entropy": 0.32005896270275114,
-      "epoch": 0.8318339362487925,
-      "grad_norm": 0.4640633761882782,
-      "learning_rate": 7.2422129196098695e-06,
-      "loss": 0.3682070255279541,
-      "mean_token_accuracy": 0.8993272170424461,
-      "num_tokens": 2309913.0,
-      "step": 10980
-    },
-    {
-      "entropy": 0.29598398655653,
-      "epoch": 0.8325915263546658,
-      "grad_norm": 0.43602728843688965,
-      "learning_rate": 7.178747849093093e-06,
-      "loss": 0.34798197746276854,
-      "mean_token_accuracy": 0.9048758611083031,
-      "num_tokens": 2406011.0,
-      "step": 10990
-    },
-    {
-      "entropy": 0.2882571941241622,
-      "epoch": 0.833349116460539,
-      "grad_norm": 0.3429960608482361,
-      "learning_rate": 7.115540569462254e-06,
-      "loss": 0.31615209579467773,
-      "mean_token_accuracy": 0.9083688676357269,
-      "num_tokens": 2498979.0,
-      "step": 11000
-    },
-    {
-      "epoch": 0.833349116460539,
-      "eval_entropy": 0.2886375365185827,
-      "eval_loss": 0.28569018840789795,
-      "eval_mean_token_accuracy": 0.9073382049017631,
-      "eval_num_tokens": 2498979.0,
-      "eval_runtime": 363.4037,
-      "eval_samples_per_second": 5.872,
-      "eval_steps_per_second": 1.469,
-      "step": 11000
-    },
-    {
-      "entropy": 0.32097633071243764,
-      "epoch": 0.8341067065664123,
-      "grad_norm": 0.4701482057571411,
-      "learning_rate": 7.052591461235453e-06,
-      "loss": 0.37491984367370607,
-      "mean_token_accuracy": 0.8957624956965446,
-      "num_tokens": 2599721.0,
-      "step": 11010
-    },
-    {
-      "entropy": 0.3161700405180454,
-      "epoch": 0.8348642966722855,
-      "grad_norm": 0.37062007188796997,
-      "learning_rate": 6.989900903376534e-06,
-      "loss": 0.37365403175354006,
-      "mean_token_accuracy": 0.8983507692813874,
-      "num_tokens": 2704511.0,
-      "step": 11020
-    },
-    {
-      "entropy": 0.3192948242649436,
-      "epoch": 0.8356218867781586,
-      "grad_norm": 0.4017787277698517,
-      "learning_rate": 6.927469273292819e-06,
-      "loss": 0.3574330806732178,
-      "mean_token_accuracy": 0.8968805193901062,
-      "num_tokens": 2810832.0,
-      "step": 11030
-    },
-    {
-      "entropy": 0.3310425169765949,
-      "epoch": 0.8363794768840319,
-      "grad_norm": 0.466722309589386,
-      "learning_rate": 6.865296946832866e-06,
-      "loss": 0.3679776668548584,
-      "mean_token_accuracy": 0.8972247004508972,
-      "num_tokens": 2913394.0,
-      "step": 11040
-    },
-    {
-      "entropy": 0.3021887818351388,
-      "epoch": 0.8371370669899051,
-      "grad_norm": 0.33357468247413635,
-      "learning_rate": 6.8033842982841545e-06,
-      "loss": 0.3556036949157715,
-      "mean_token_accuracy": 0.9023103356361389,
-      "num_tokens": 3011528.0,
-      "step": 11050
-    },
-    {
-      "entropy": 0.3188148533925414,
-      "epoch": 0.8378946570957784,
-      "grad_norm": 0.5248581767082214,
-      "learning_rate": 6.741731700370885e-06,
-      "loss": 0.3591402530670166,
-      "mean_token_accuracy": 0.8997363194823265,
-      "num_tokens": 3114520.0,
-      "step": 11060
-    },
-    {
-      "entropy": 0.28566175922751424,
-      "epoch": 0.8386522472016515,
-      "grad_norm": 0.4290429651737213,
-      "learning_rate": 6.680339524251694e-06,
-      "loss": 0.33682057857513426,
-      "mean_token_accuracy": 0.9096178874373436,
-      "num_tokens": 3220100.0,
-      "step": 11070
-    },
-    {
-      "entropy": 0.2704159118235111,
-      "epoch": 0.8394098373075247,
-      "grad_norm": 0.5517525672912598,
-      "learning_rate": 6.619208139517441e-06,
-      "loss": 0.3258359909057617,
-      "mean_token_accuracy": 0.9156906500458717,
-      "num_tokens": 3311139.0,
-      "step": 11080
-    },
-    {
-      "entropy": 0.2802377393469214,
-      "epoch": 0.840167427413398,
-      "grad_norm": 0.43798232078552246,
-      "learning_rate": 6.558337914188989e-06,
-      "loss": 0.31688680648803713,
-      "mean_token_accuracy": 0.9112754926085472,
-      "num_tokens": 3408982.0,
-      "step": 11090
-    },
-    {
-      "entropy": 0.2785337887704372,
-      "epoch": 0.8409250175192712,
-      "grad_norm": 0.48913341760635376,
-      "learning_rate": 6.497729214714959e-06,
-      "loss": 0.33413968086242674,
-      "mean_token_accuracy": 0.9114461779594422,
-      "num_tokens": 3506297.0,
-      "step": 11100
-    },
-    {
-      "entropy": 0.25448127966374157,
-      "epoch": 0.8416826076251445,
-      "grad_norm": 0.6489408612251282,
-      "learning_rate": 6.437382405969572e-06,
-      "loss": 0.3060408115386963,
-      "mean_token_accuracy": 0.9170120373368263,
-      "num_tokens": 3599883.0,
-      "step": 11110
-    },
-    {
-      "entropy": 0.2931193061172962,
-      "epoch": 0.8424401977310176,
-      "grad_norm": 0.39941656589508057,
-      "learning_rate": 6.377297851250396e-06,
-      "loss": 0.34188551902770997,
-      "mean_token_accuracy": 0.9045077741146088,
-      "num_tokens": 3695770.0,
-      "step": 11120
-    },
-    {
-      "entropy": 0.28530145399272444,
-      "epoch": 0.8431977878368908,
-      "grad_norm": 0.5832393169403076,
-      "learning_rate": 6.317475912276205e-06,
-      "loss": 0.3301248073577881,
-      "mean_token_accuracy": 0.9066288366913795,
-      "num_tokens": 3792849.0,
-      "step": 11130
-    },
-    {
-      "entropy": 0.28422479014843705,
-      "epoch": 0.8439553779427641,
-      "grad_norm": 0.5767186284065247,
-      "learning_rate": 6.257916949184772e-06,
-      "loss": 0.34296748638153074,
-      "mean_token_accuracy": 0.9086409032344818,
-      "num_tokens": 3886352.0,
-      "step": 11140
-    },
-    {
-      "entropy": 0.33191717118024827,
-      "epoch": 0.8447129680486373,
-      "grad_norm": 0.31320858001708984,
-      "learning_rate": 6.1986213205307285e-06,
-      "loss": 0.38002519607543944,
-      "mean_token_accuracy": 0.8913838744163514,
-      "num_tokens": 3990242.0,
-      "step": 11150
-    },
-    {
-      "entropy": 0.27569183316081763,
-      "epoch": 0.8454705581545106,
-      "grad_norm": 0.35137414932250977,
-      "learning_rate": 6.139589383283367e-06,
-      "loss": 0.3208441734313965,
-      "mean_token_accuracy": 0.9123941317200661,
-      "num_tokens": 4090031.0,
-      "step": 11160
-    },
-    {
-      "entropy": 0.26733697205781937,
-      "epoch": 0.8462281482603837,
-      "grad_norm": 0.4320410192012787,
-      "learning_rate": 6.08082149282454e-06,
-      "loss": 0.3125286102294922,
-      "mean_token_accuracy": 0.914748401939869,
-      "num_tokens": 4185773.0,
-      "step": 11170
-    },
-    {
-      "entropy": 0.23407292291522025,
-      "epoch": 0.8469857383662569,
-      "grad_norm": 0.48707106709480286,
-      "learning_rate": 6.022318002946492e-06,
-      "loss": 0.2711918830871582,
-      "mean_token_accuracy": 0.9267101764678956,
-      "num_tokens": 4278787.0,
-      "step": 11180
-    },
-    {
-      "entropy": 0.316350325755775,
-      "epoch": 0.8477433284721302,
-      "grad_norm": 0.3188752830028534,
-      "learning_rate": 5.964079265849698e-06,
-      "loss": 0.3772554397583008,
-      "mean_token_accuracy": 0.8990112885832786,
-      "num_tokens": 4385286.0,
-      "step": 11190
-    },
-    {
-      "entropy": 0.28382357601076363,
-      "epoch": 0.8485009185780034,
-      "grad_norm": 0.5237545371055603,
-      "learning_rate": 5.906105632140818e-06,
-      "loss": 0.33560957908630373,
-      "mean_token_accuracy": 0.9108323022723198,
-      "num_tokens": 4478235.0,
-      "step": 11200
-    },
-    {
-      "entropy": 0.29501939918845893,
-      "epoch": 0.8492585086838765,
-      "grad_norm": 0.4553075134754181,
-      "learning_rate": 5.848397450830539e-06,
-      "loss": 0.3474492073059082,
-      "mean_token_accuracy": 0.9073790535330772,
-      "num_tokens": 4574842.0,
-      "step": 11210
-    },
-    {
-      "entropy": 0.2461070492863655,
-      "epoch": 0.8500160987897498,
-      "grad_norm": 0.3291778564453125,
-      "learning_rate": 5.7909550693314665e-06,
-      "loss": 0.2918540477752686,
-      "mean_token_accuracy": 0.9265199154615402,
-      "num_tokens": 4661233.0,
-      "step": 11220
-    },
-    {
-      "entropy": 0.2952361168339849,
-      "epoch": 0.850773688895623,
-      "grad_norm": 0.5359086990356445,
-      "learning_rate": 5.73377883345606e-06,
-      "loss": 0.3461754560470581,
-      "mean_token_accuracy": 0.9084222391247749,
-      "num_tokens": 4753367.0,
-      "step": 11230
-    },
-    {
-      "entropy": 0.28638926018029454,
-      "epoch": 0.8515312790014963,
-      "grad_norm": 0.4929027855396271,
-      "learning_rate": 5.676869087414544e-06,
-      "loss": 0.3351025104522705,
-      "mean_token_accuracy": 0.9095506638288497,
-      "num_tokens": 4846891.0,
-      "step": 11240
-    },
-    {
-      "entropy": 0.2919112669304013,
-      "epoch": 0.8522888691073695,
-      "grad_norm": 0.43144306540489197,
-      "learning_rate": 5.6202261738128145e-06,
-      "loss": 0.3436851739883423,
-      "mean_token_accuracy": 0.908247809112072,
-      "num_tokens": 4945534.0,
-      "step": 11250
-    },
-    {
-      "epoch": 0.8522888691073695,
-      "eval_entropy": 0.28951325608689465,
-      "eval_loss": 0.28531700372695923,
-      "eval_mean_token_accuracy": 0.907498625333836,
-      "eval_num_tokens": 4945534.0,
-      "eval_runtime": 363.816,
-      "eval_samples_per_second": 5.866,
-      "eval_steps_per_second": 1.468,
-      "step": 11250
-    },
-    {
-      "entropy": 0.3246528644114733,
-      "epoch": 0.8530464592132426,
-      "grad_norm": 0.5182899832725525,
-      "learning_rate": 5.563850433650408e-06,
-      "loss": 0.3830410003662109,
-      "mean_token_accuracy": 0.8927986115217209,
-      "num_tokens": 5053848.0,
-      "step": 11260
-    },
-    {
-      "entropy": 0.2774768492206931,
-      "epoch": 0.8538040493191159,
-      "grad_norm": 0.45578455924987793,
-      "learning_rate": 5.507742206318422e-06,
-      "loss": 0.3247124195098877,
-      "mean_token_accuracy": 0.9107116803526878,
-      "num_tokens": 5150160.0,
-      "step": 11270
-    },
-    {
-      "entropy": 0.2593419700860977,
-      "epoch": 0.8545616394249891,
-      "grad_norm": 0.4656168818473816,
-      "learning_rate": 5.4519018295974945e-06,
-      "loss": 0.29569401741027834,
-      "mean_token_accuracy": 0.9186507374048233,
-      "num_tokens": 5245280.0,
-      "step": 11280
-    },
-    {
-      "entropy": 0.2584766630083323,
-      "epoch": 0.8553192295308624,
-      "grad_norm": 0.36138275265693665,
-      "learning_rate": 5.396329639655756e-06,
-      "loss": 0.2868622064590454,
-      "mean_token_accuracy": 0.9181327551603318,
-      "num_tokens": 5345185.0,
-      "step": 11290
-    },
-    {
-      "entropy": 0.29062068052589896,
-      "epoch": 0.8560768196367355,
-      "grad_norm": 0.3519483804702759,
-      "learning_rate": 5.341025971046798e-06,
-      "loss": 0.3423572540283203,
-      "mean_token_accuracy": 0.9081257462501526,
-      "num_tokens": 5445289.0,
-      "step": 11300
-    },
-    {
-      "entropy": 0.27519712056964635,
-      "epoch": 0.8568344097426087,
-      "grad_norm": 0.3887447416782379,
-      "learning_rate": 5.285991156707681e-06,
-      "loss": 0.317475700378418,
-      "mean_token_accuracy": 0.9138319388031959,
-      "num_tokens": 5544079.0,
-      "step": 11310
-    },
-    {
-      "entropy": 0.2655044592916965,
-      "epoch": 0.857591999848482,
-      "grad_norm": 0.5422892570495605,
-      "learning_rate": 5.2312255279569236e-06,
-      "loss": 0.3148800373077393,
-      "mean_token_accuracy": 0.9163942500948906,
-      "num_tokens": 5636934.0,
-      "step": 11320
-    },
-    {
-      "entropy": 0.26932154782116413,
-      "epoch": 0.8583495899543552,
-      "grad_norm": 0.3854497969150543,
-      "learning_rate": 5.176729414492493e-06,
-      "loss": 0.32476861476898194,
-      "mean_token_accuracy": 0.9170946836471557,
-      "num_tokens": 5734349.0,
-      "step": 11330
-    },
-    {
-      "entropy": 0.2900954818353057,
-      "epoch": 0.8591071800602285,
-      "grad_norm": 0.40638411045074463,
-      "learning_rate": 5.122503144389834e-06,
-      "loss": 0.33924498558044436,
-      "mean_token_accuracy": 0.9079746246337891,
-      "num_tokens": 5835640.0,
-      "step": 11340
-    },
-    {
-      "entropy": 0.3010376073420048,
-      "epoch": 0.8598647701661016,
-      "grad_norm": 0.44111737608909607,
-      "learning_rate": 5.068547044099891e-06,
-      "loss": 0.31704282760620117,
-      "mean_token_accuracy": 0.9046201959252358,
-      "num_tokens": 5934993.0,
-      "step": 11350
-    },
-    {
-      "entropy": 0.3023892106488347,
-      "epoch": 0.8606223602719748,
-      "grad_norm": 0.37891948223114014,
-      "learning_rate": 5.014861438447144e-06,
-      "loss": 0.3835698366165161,
-      "mean_token_accuracy": 0.9033500358462334,
-      "num_tokens": 6038266.0,
-      "step": 11360
-    },
-    {
-      "entropy": 0.2630012549459934,
-      "epoch": 0.8613799503778481,
-      "grad_norm": 0.38084009289741516,
-      "learning_rate": 4.961446650627644e-06,
-      "loss": 0.30712451934814455,
-      "mean_token_accuracy": 0.917777256667614,
-      "num_tokens": 6141215.0,
-      "step": 11370
-    },
-    {
-      "entropy": 0.30645178388804195,
-      "epoch": 0.8621375404837213,
-      "grad_norm": 0.4121268093585968,
-      "learning_rate": 4.9083030022070985e-06,
-      "loss": 0.34171285629272463,
-      "mean_token_accuracy": 0.9038849920034409,
-      "num_tokens": 6235571.0,
-      "step": 11380
-    },
-    {
-      "entropy": 0.2727683449164033,
-      "epoch": 0.8628951305895946,
-      "grad_norm": 0.364166796207428,
-      "learning_rate": 4.855430813118872e-06,
-      "loss": 0.31318480968475343,
-      "mean_token_accuracy": 0.915680679678917,
-      "num_tokens": 6333413.0,
-      "step": 11390
-    },
-    {
-      "entropy": 0.3058678261935711,
-      "epoch": 0.8636527206954677,
-      "grad_norm": 0.37689122557640076,
-      "learning_rate": 4.802830401662129e-06,
-      "loss": 0.3599482297897339,
-      "mean_token_accuracy": 0.9037078693509102,
-      "num_tokens": 6435643.0,
-      "step": 11400
-    },
-    {
-      "entropy": 0.28505960423499344,
-      "epoch": 0.8644103108013409,
-      "grad_norm": 0.6746123433113098,
-      "learning_rate": 4.750502084499886e-06,
-      "loss": 0.3244886875152588,
-      "mean_token_accuracy": 0.9097675696015358,
-      "num_tokens": 6534257.0,
-      "step": 11410
-    },
-    {
-      "entropy": 0.3098301423713565,
-      "epoch": 0.8651679009072142,
-      "grad_norm": 0.500696063041687,
-      "learning_rate": 4.6984461766570955e-06,
-      "loss": 0.33419573307037354,
-      "mean_token_accuracy": 0.8996593952178955,
-      "num_tokens": 6643378.0,
-      "step": 11420
-    },
-    {
-      "entropy": 0.2496833168901503,
-      "epoch": 0.8659254910130874,
-      "grad_norm": 0.4080503582954407,
-      "learning_rate": 4.646662991518775e-06,
-      "loss": 0.26940312385559084,
-      "mean_token_accuracy": 0.9210411250591278,
-      "num_tokens": 6734073.0,
-      "step": 11430
-    },
-    {
-      "entropy": 0.3179155085235834,
-      "epoch": 0.8666830811189606,
-      "grad_norm": 0.35618963837623596,
-      "learning_rate": 4.595152840828099e-06,
-      "loss": 0.37588114738464357,
-      "mean_token_accuracy": 0.8984893754124641,
-      "num_tokens": 6843787.0,
-      "step": 11440
-    },
-    {
-      "entropy": 0.28430836256593467,
-      "epoch": 0.8674406712248338,
-      "grad_norm": 0.4345318675041199,
-      "learning_rate": 4.543916034684531e-06,
-      "loss": 0.33840193748474123,
-      "mean_token_accuracy": 0.9102910846471787,
-      "num_tokens": 6943533.0,
-      "step": 11450
-    },
-    {
-      "entropy": 0.28566361106932164,
-      "epoch": 0.868198261330707,
-      "grad_norm": 0.3477196991443634,
-      "learning_rate": 4.492952881541962e-06,
-      "loss": 0.34356863498687745,
-      "mean_token_accuracy": 0.9089460060000419,
-      "num_tokens": 7046514.0,
-      "step": 11460
-    },
-    {
-      "entropy": 0.2749630717560649,
-      "epoch": 0.8689558514365803,
-      "grad_norm": 0.44582805037498474,
-      "learning_rate": 4.442263688206838e-06,
-      "loss": 0.3119127988815308,
-      "mean_token_accuracy": 0.9123926997184754,
-      "num_tokens": 7147898.0,
-      "step": 11470
-    },
-    {
-      "entropy": 0.2725351193919778,
-      "epoch": 0.8697134415424534,
-      "grad_norm": 0.5893859267234802,
-      "learning_rate": 4.391848759836331e-06,
-      "loss": 0.30458645820617675,
-      "mean_token_accuracy": 0.9120975598692894,
-      "num_tokens": 7238758.0,
-      "step": 11480
-    },
-    {
-      "entropy": 0.2887442596256733,
-      "epoch": 0.8704710316483266,
-      "grad_norm": 0.38634830713272095,
-      "learning_rate": 4.341708399936484e-06,
-      "loss": 0.34334683418273926,
-      "mean_token_accuracy": 0.9080741614103317,
-      "num_tokens": 7332778.0,
-      "step": 11490
-    },
-    {
-      "entropy": 0.3163892447948456,
-      "epoch": 0.8712286217541999,
-      "grad_norm": 0.4110091030597687,
-      "learning_rate": 4.291842910360411e-06,
-      "loss": 0.3650186061859131,
-      "mean_token_accuracy": 0.8998959764838219,
-      "num_tokens": 7435633.0,
-      "step": 11500
-    },
-    {
-      "epoch": 0.8712286217541999,
-      "eval_entropy": 0.28902850807940916,
-      "eval_loss": 0.28520920872688293,
-      "eval_mean_token_accuracy": 0.9075907013389501,
-      "eval_num_tokens": 7435633.0,
-      "eval_runtime": 362.4662,
-      "eval_samples_per_second": 5.887,
-      "eval_steps_per_second": 1.473,
-      "step": 11500
-    },
-    {
-      "entropy": 0.3062978575006127,
-      "epoch": 0.8719862118600731,
-      "grad_norm": 0.523257315158844,
-      "learning_rate": 4.242252591306439e-06,
-      "loss": 0.35051255226135253,
-      "mean_token_accuracy": 0.9012857481837273,
-      "num_tokens": 7532984.0,
-      "step": 11510
-    },
-    {
-      "entropy": 0.27845307141542436,
-      "epoch": 0.8727438019659464,
-      "grad_norm": 0.41053506731987,
-      "learning_rate": 4.1929377413163386e-06,
-      "loss": 0.33397209644317627,
-      "mean_token_accuracy": 0.9132984727621078,
-      "num_tokens": 7630708.0,
-      "step": 11520
-    },
-    {
-      "entropy": 0.2766347942873836,
-      "epoch": 0.8735013920718195,
-      "grad_norm": 0.3962479829788208,
-      "learning_rate": 4.143898657273509e-06,
-      "loss": 0.32157082557678224,
-      "mean_token_accuracy": 0.9120013162493705,
-      "num_tokens": 7731367.0,
-      "step": 11530
-    },
-    {
-      "entropy": 0.30036948043853046,
-      "epoch": 0.8742589821776927,
-      "grad_norm": 0.3602987825870514,
-      "learning_rate": 4.095135634401198e-06,
-      "loss": 0.3608391284942627,
-      "mean_token_accuracy": 0.9042830646038056,
-      "num_tokens": 7828677.0,
-      "step": 11540
-    },
-    {
-      "entropy": 0.28864908814430235,
-      "epoch": 0.875016572283566,
-      "grad_norm": 0.4127624034881592,
-      "learning_rate": 4.046648966260708e-06,
-      "loss": 0.33952603340148924,
-      "mean_token_accuracy": 0.9077737912535667,
-      "num_tokens": 7926097.0,
-      "step": 11550
-    },
-    {
-      "entropy": 0.28967538885772226,
-      "epoch": 0.8757741623894392,
-      "grad_norm": 0.3790387511253357,
-      "learning_rate": 3.99843894474965e-06,
-      "loss": 0.3468469619750977,
-      "mean_token_accuracy": 0.9093071848154068,
-      "num_tokens": 8024723.0,
-      "step": 11560
-    },
-    {
-      "entropy": 0.28250670302659275,
-      "epoch": 0.8765317524953125,
-      "grad_norm": 0.43996334075927734,
-      "learning_rate": 3.950505860100173e-06,
-      "loss": 0.31765334606170653,
-      "mean_token_accuracy": 0.9129352971911431,
-      "num_tokens": 8121274.0,
-      "step": 11570
-    },
-    {
-      "entropy": 0.2830861508846283,
-      "epoch": 0.8772893426011856,
-      "grad_norm": 0.49645763635635376,
-      "learning_rate": 3.902850000877228e-06,
-      "loss": 0.34697833061218264,
-      "mean_token_accuracy": 0.9117490366101265,
-      "num_tokens": 8214273.0,
-      "step": 11580
-    },
-    {
-      "entropy": 0.2739479450508952,
-      "epoch": 0.8780469327070588,
-      "grad_norm": 0.40956977009773254,
-      "learning_rate": 3.855471653976833e-06,
-      "loss": 0.31866281032562255,
-      "mean_token_accuracy": 0.9111287161707878,
-      "num_tokens": 8311017.0,
-      "step": 11590
-    },
-    {
-      "entropy": 0.24923466965556146,
-      "epoch": 0.8788045228129321,
-      "grad_norm": 0.46503928303718567,
-      "learning_rate": 3.8083711046242976e-06,
-      "loss": 0.29086575508117674,
-      "mean_token_accuracy": 0.9209815084934234,
-      "num_tokens": 8398430.0,
-      "step": 11600
-    },
-    {
-      "entropy": 0.2712684884667397,
-      "epoch": 0.8795621129188053,
-      "grad_norm": 0.5817049145698547,
-      "learning_rate": 3.7615486363725862e-06,
-      "loss": 0.317930793762207,
-      "mean_token_accuracy": 0.9129366233944893,
-      "num_tokens": 8499764.0,
-      "step": 11610
-    },
-    {
-      "entropy": 0.2838691871613264,
-      "epoch": 0.8803197030246785,
-      "grad_norm": 0.4317021667957306,
-      "learning_rate": 3.715004531100552e-06,
-      "loss": 0.33291192054748536,
-      "mean_token_accuracy": 0.9116261765360832,
-      "num_tokens": 8597456.0,
-      "step": 11620
-    },
-    {
-      "entropy": 0.3134252499789,
-      "epoch": 0.8810772931305517,
-      "grad_norm": 0.39608034491539,
-      "learning_rate": 3.668739069011262e-06,
-      "loss": 0.35954484939575193,
-      "mean_token_accuracy": 0.9001775801181793,
-      "num_tokens": 8695484.0,
-      "step": 11630
-    },
-    {
-      "entropy": 0.30583129059523345,
-      "epoch": 0.8818348832364249,
-      "grad_norm": 0.39231637120246887,
-      "learning_rate": 3.6227525286303077e-06,
-      "loss": 0.3442234516143799,
-      "mean_token_accuracy": 0.9026809856295586,
-      "num_tokens": 8797939.0,
-      "step": 11640
-    },
-    {
-      "entropy": 0.2677101884037256,
-      "epoch": 0.8825924733422982,
-      "grad_norm": 0.4724090099334717,
-      "learning_rate": 3.5770451868041176e-06,
-      "loss": 0.3116743087768555,
-      "mean_token_accuracy": 0.9159694716334343,
-      "num_tokens": 8888698.0,
-      "step": 11650
-    },
-    {
-      "entropy": 0.2846052870154381,
-      "epoch": 0.8833500634481714,
-      "grad_norm": 0.4535260796546936,
-      "learning_rate": 3.5316173186983047e-06,
-      "loss": 0.3254629850387573,
-      "mean_token_accuracy": 0.9086816072463989,
-      "num_tokens": 8984200.0,
-      "step": 11660
-    },
-    {
-      "entropy": 0.2747473681345582,
-      "epoch": 0.8841076535540446,
-      "grad_norm": 0.4237743616104126,
-      "learning_rate": 3.486469197796016e-06,
-      "loss": 0.3034130811691284,
-      "mean_token_accuracy": 0.91242136657238,
-      "num_tokens": 9081499.0,
-      "step": 11670
-    },
-    {
-      "entropy": 0.29838810246437786,
-      "epoch": 0.8848652436599178,
-      "grad_norm": 0.46438512206077576,
-      "learning_rate": 3.4416010958962575e-06,
-      "loss": 0.3530242443084717,
-      "mean_token_accuracy": 0.903732942044735,
-      "num_tokens": 9180324.0,
-      "step": 11680
-    },
-    {
-      "entropy": 0.2972015455365181,
-      "epoch": 0.885622833765791,
-      "grad_norm": 0.5626263618469238,
-      "learning_rate": 3.3970132831122825e-06,
-      "loss": 0.3368989944458008,
-      "mean_token_accuracy": 0.9042138263583184,
-      "num_tokens": 9281689.0,
-      "step": 11690
-    },
-    {
-      "entropy": 0.28716162145137786,
-      "epoch": 0.8863804238716643,
-      "grad_norm": 0.4129055440425873,
-      "learning_rate": 3.3527060278699563e-06,
-      "loss": 0.333894681930542,
-      "mean_token_accuracy": 0.9115474045276641,
-      "num_tokens": 9373442.0,
-      "step": 11700
-    },
-    {
-      "entropy": 0.3078194234520197,
-      "epoch": 0.8871380139775374,
-      "grad_norm": 0.3562985956668854,
-      "learning_rate": 3.3086795969061557e-06,
-      "loss": 0.3490109443664551,
-      "mean_token_accuracy": 0.9009092181921006,
-      "num_tokens": 9474012.0,
-      "step": 11710
-    },
-    {
-      "entropy": 0.32325697001069786,
-      "epoch": 0.8878956040834106,
-      "grad_norm": 0.4092463552951813,
-      "learning_rate": 3.264934255267127e-06,
-      "loss": 0.38159396648406985,
-      "mean_token_accuracy": 0.894842478632927,
-      "num_tokens": 9584832.0,
-      "step": 11720
-    },
-    {
-      "entropy": 0.24773544836789368,
-      "epoch": 0.8886531941892839,
-      "grad_norm": 0.43235835433006287,
-      "learning_rate": 3.2214702663069308e-06,
-      "loss": 0.2853692054748535,
-      "mean_token_accuracy": 0.9219878166913986,
-      "num_tokens": 9685029.0,
-      "step": 11730
-    },
-    {
-      "entropy": 0.2398093307390809,
-      "epoch": 0.8894107842951571,
-      "grad_norm": 0.35412466526031494,
-      "learning_rate": 3.1782878916858273e-06,
-      "loss": 0.29156792163848877,
-      "mean_token_accuracy": 0.9237233802676201,
-      "num_tokens": 9781772.0,
-      "step": 11740
-    },
-    {
-      "entropy": 0.2823848981410265,
-      "epoch": 0.8901683744010304,
-      "grad_norm": 0.4498985707759857,
-      "learning_rate": 3.1353873913687327e-06,
-      "loss": 0.3437998294830322,
-      "mean_token_accuracy": 0.907787038385868,
-      "num_tokens": 9874032.0,
-      "step": 11750
-    },
-    {
-      "epoch": 0.8901683744010304,
-      "eval_entropy": 0.2877783386964588,
-      "eval_loss": 0.28504320979118347,
-      "eval_mean_token_accuracy": 0.907661068528779,
-      "eval_num_tokens": 9874032.0,
-      "eval_runtime": 364.3306,
-      "eval_samples_per_second": 5.857,
-      "eval_steps_per_second": 1.466,
-      "step": 11750
-    },
-    {
-      "entropy": 0.26369482185691595,
-      "epoch": 0.8909259645069035,
-      "grad_norm": 0.5122780203819275,
-      "learning_rate": 3.092769023623604e-06,
-      "loss": 0.3001707077026367,
-      "mean_token_accuracy": 0.9151472270488739,
-      "num_tokens": 9969565.0,
-      "step": 11760
-    },
-    {
-      "entropy": 0.2945669298991561,
-      "epoch": 0.8916835546127767,
-      "grad_norm": 0.44441863894462585,
-      "learning_rate": 3.050433045019935e-06,
-      "loss": 0.34853763580322267,
-      "mean_token_accuracy": 0.9060156434774399,
-      "num_tokens": 10073446.0,
-      "step": 11770
-    },
-    {
-      "entropy": 0.28827795311808585,
-      "epoch": 0.89244114471865,
-      "grad_norm": 0.44912177324295044,
-      "learning_rate": 3.008379710427184e-06,
-      "loss": 0.3416128635406494,
-      "mean_token_accuracy": 0.9073840662837028,
-      "num_tokens": 10178033.0,
-      "step": 11780
-    },
-    {
-      "entropy": 0.29140637572854755,
-      "epoch": 0.8931987348245232,
-      "grad_norm": 0.42318063974380493,
-      "learning_rate": 2.9666092730132546e-06,
-      "loss": 0.32835893630981444,
-      "mean_token_accuracy": 0.9083481669425965,
-      "num_tokens": 10279691.0,
-      "step": 11790
-    },
-    {
-      "entropy": 0.2990898171439767,
-      "epoch": 0.8939563249303965,
-      "grad_norm": 0.5038789510726929,
-      "learning_rate": 2.925121984242951e-06,
-      "loss": 0.34421396255493164,
-      "mean_token_accuracy": 0.9057884082198143,
-      "num_tokens": 10375462.0,
-      "step": 11800
-    },
-    {
-      "entropy": 0.26596365328878163,
-      "epoch": 0.8947139150362696,
-      "grad_norm": 0.5523534417152405,
-      "learning_rate": 2.8839180938764863e-06,
-      "loss": 0.31238632202148436,
-      "mean_token_accuracy": 0.9166907787322998,
-      "num_tokens": 10468250.0,
-      "step": 11810
-    },
-    {
-      "entropy": 0.26575090661644934,
-      "epoch": 0.8954715051421428,
-      "grad_norm": 0.4703275263309479,
-      "learning_rate": 2.842997849967971e-06,
-      "loss": 0.29630722999572756,
-      "mean_token_accuracy": 0.9156646102666854,
-      "num_tokens": 10563173.0,
-      "step": 11820
-    },
-    {
-      "entropy": 0.2932529145851731,
-      "epoch": 0.8962290952480161,
-      "grad_norm": 0.3833015263080597,
-      "learning_rate": 2.8023614988639037e-06,
-      "loss": 0.3358645439147949,
-      "mean_token_accuracy": 0.9068357229232789,
-      "num_tokens": 10664872.0,
-      "step": 11830
-    },
-    {
-      "entropy": 0.3053524810820818,
-      "epoch": 0.8969866853538893,
-      "grad_norm": 0.48305702209472656,
-      "learning_rate": 2.7620092852017163e-06,
-      "loss": 0.3639130830764771,
-      "mean_token_accuracy": 0.8995088264346123,
-      "num_tokens": 10765468.0,
-      "step": 11840
-    },
-    {
-      "entropy": 0.27784552425146103,
-      "epoch": 0.8977442754597625,
-      "grad_norm": 0.7603539228439331,
-      "learning_rate": 2.721941451908283e-06,
-      "loss": 0.3374777317047119,
-      "mean_token_accuracy": 0.9113835170865059,
-      "num_tokens": 10860042.0,
-      "step": 11850
-    },
-    {
-      "entropy": 0.2647241912782192,
-      "epoch": 0.8985018655656357,
-      "grad_norm": 0.41474252939224243,
-      "learning_rate": 2.6821582401984624e-06,
-      "loss": 0.3280783653259277,
-      "mean_token_accuracy": 0.9178823977708817,
-      "num_tokens": 10956561.0,
-      "step": 11860
-    },
-    {
-      "entropy": 0.29947108291089536,
-      "epoch": 0.8992594556715089,
-      "grad_norm": 0.7285243272781372,
-      "learning_rate": 2.6426598895736397e-06,
-      "loss": 0.3578646659851074,
-      "mean_token_accuracy": 0.9039269968867302,
-      "num_tokens": 11052288.0,
-      "step": 11870
-    },
-    {
-      "entropy": 0.2672554237768054,
-      "epoch": 0.9000170457773822,
-      "grad_norm": 0.3953443765640259,
-      "learning_rate": 2.6034466378203006e-06,
-      "loss": 0.32422146797180174,
-      "mean_token_accuracy": 0.9144565418362618,
-      "num_tokens": 11148839.0,
-      "step": 11880
-    },
-    {
-      "entropy": 0.31262458153069017,
-      "epoch": 0.9007746358832553,
-      "grad_norm": 0.41961008310317993,
-      "learning_rate": 2.56451872100858e-06,
-      "loss": 0.36020898818969727,
-      "mean_token_accuracy": 0.8983094453811645,
-      "num_tokens": 11249056.0,
-      "step": 11890
-    },
-    {
-      "entropy": 0.2922690697014332,
-      "epoch": 0.9015322259891286,
-      "grad_norm": 0.435486763715744,
-      "learning_rate": 2.5258763734908543e-06,
-      "loss": 0.33933568000793457,
-      "mean_token_accuracy": 0.9072558209300041,
-      "num_tokens": 11350465.0,
-      "step": 11900
-    },
-    {
-      "entropy": 0.2990702610462904,
-      "epoch": 0.9022898160950018,
-      "grad_norm": 0.3435786962509155,
-      "learning_rate": 2.4875198279003197e-06,
-      "loss": 0.32688493728637696,
-      "mean_token_accuracy": 0.9039709404110908,
-      "num_tokens": 11461689.0,
-      "step": 11910
-    },
-    {
-      "entropy": 0.2897825580090284,
-      "epoch": 0.903047406200875,
-      "grad_norm": 0.43031200766563416,
-      "learning_rate": 2.4494493151496102e-06,
-      "loss": 0.3513793468475342,
-      "mean_token_accuracy": 0.9052729144692421,
-      "num_tokens": 11552766.0,
-      "step": 11920
-    },
-    {
-      "entropy": 0.3014989258721471,
-      "epoch": 0.9038049963067483,
-      "grad_norm": 0.5037825703620911,
-      "learning_rate": 2.4116650644293915e-06,
-      "loss": 0.34794044494628906,
-      "mean_token_accuracy": 0.9035784974694252,
-      "num_tokens": 11649054.0,
-      "step": 11930
-    },
-    {
-      "entropy": 0.26335182003676894,
-      "epoch": 0.9045625864126214,
-      "grad_norm": 0.4968317449092865,
-      "learning_rate": 2.3741673032069756e-06,
-      "loss": 0.29862327575683595,
-      "mean_token_accuracy": 0.916699244081974,
-      "num_tokens": 11745904.0,
-      "step": 11940
-    },
-    {
-      "entropy": 0.27952721528708935,
-      "epoch": 0.9053201765184947,
-      "grad_norm": 0.4142918586730957,
-      "learning_rate": 2.336956257224976e-06,
-      "loss": 0.3438222646713257,
-      "mean_token_accuracy": 0.9101695835590362,
-      "num_tokens": 11848622.0,
-      "step": 11950
-    },
-    {
-      "entropy": 0.2757990902289748,
-      "epoch": 0.9060777666243679,
-      "grad_norm": 0.6232902407646179,
-      "learning_rate": 2.3000321504999256e-06,
-      "loss": 0.301646089553833,
-      "mean_token_accuracy": 0.9144108861684799,
-      "num_tokens": 11948966.0,
-      "step": 11960
-    },
-    {
-      "entropy": 0.29781850054860115,
-      "epoch": 0.9068353567302411,
-      "grad_norm": 0.5357040166854858,
-      "learning_rate": 2.2633952053209396e-06,
-      "loss": 0.3362586498260498,
-      "mean_token_accuracy": 0.9043088600039482,
-      "num_tokens": 12050900.0,
-      "step": 11970
-    },
-    {
-      "entropy": 0.3077027928084135,
-      "epoch": 0.9075929468361144,
-      "grad_norm": 0.3803006708621979,
-      "learning_rate": 2.2270456422483655e-06,
-      "loss": 0.36307554244995116,
-      "mean_token_accuracy": 0.8973830282688141,
-      "num_tokens": 12147023.0,
-      "step": 11980
-    },
-    {
-      "entropy": 0.2707913937047124,
-      "epoch": 0.9083505369419875,
-      "grad_norm": 0.47196221351623535,
-      "learning_rate": 2.190983680112485e-06,
-      "loss": 0.31296596527099607,
-      "mean_token_accuracy": 0.9124700129032135,
-      "num_tokens": 12235151.0,
-      "step": 11990
-    },
-    {
-      "entropy": 0.2599682783707976,
-      "epoch": 0.9091081270478607,
-      "grad_norm": 0.49874573945999146,
-      "learning_rate": 2.155209536012148e-06,
-      "loss": 0.33502559661865233,
-      "mean_token_accuracy": 0.9186246544122696,
-      "num_tokens": 12323993.0,
-      "step": 12000
-    },
-    {
-      "epoch": 0.9091081270478607,
-      "eval_entropy": 0.2891411053553949,
-      "eval_loss": 0.28493180871009827,
-      "eval_mean_token_accuracy": 0.907693214184336,
-      "eval_num_tokens": 12323993.0,
-      "eval_runtime": 362.6666,
-      "eval_samples_per_second": 5.884,
-      "eval_steps_per_second": 1.472,
-      "step": 12000
-    },
-    {
-      "entropy": 0.3315456384792924,
-      "epoch": 0.909865717153734,
-      "grad_norm": 0.36325132846832275,
-      "learning_rate": 2.119723425313508e-06,
-      "loss": 0.3699875593185425,
-      "mean_token_accuracy": 0.8940782964229583,
-      "num_tokens": 12430861.0,
-      "step": 12010
-    },
-    {
-      "entropy": 0.3365255404263735,
-      "epoch": 0.9106233072596072,
-      "grad_norm": 0.4198094308376312,
-      "learning_rate": 2.0845255616487146e-06,
-      "loss": 0.39371466636657715,
-      "mean_token_accuracy": 0.8894189238548279,
-      "num_tokens": 12538044.0,
-      "step": 12020
-    },
-    {
-      "entropy": 0.28438792023807763,
-      "epoch": 0.9113808973654804,
-      "grad_norm": 0.39958682656288147,
-      "learning_rate": 2.049616156914619e-06,
-      "loss": 0.32168402671813967,
-      "mean_token_accuracy": 0.9102310806512832,
-      "num_tokens": 12639720.0,
-      "step": 12030
-    },
-    {
-      "entropy": 0.2975167462602258,
-      "epoch": 0.9121384874713536,
-      "grad_norm": 0.43647003173828125,
-      "learning_rate": 2.014995421271498e-06,
-      "loss": 0.34835057258605956,
-      "mean_token_accuracy": 0.9067213013768196,
-      "num_tokens": 12743172.0,
-      "step": 12040
-    },
-    {
-      "entropy": 0.28764260224997995,
-      "epoch": 0.9128960775772268,
-      "grad_norm": 0.3330994248390198,
-      "learning_rate": 1.9806635631418048e-06,
-      "loss": 0.3269298315048218,
-      "mean_token_accuracy": 0.9144092231988907,
-      "num_tokens": 12843459.0,
-      "step": 12050
-    },
-    {
-      "entropy": 0.28370366655290125,
-      "epoch": 0.9136536676831001,
-      "grad_norm": 0.46237102150917053,
-      "learning_rate": 1.9466207892088805e-06,
-      "loss": 0.32587761878967286,
-      "mean_token_accuracy": 0.9086109101772308,
-      "num_tokens": 12949924.0,
-      "step": 12060
-    },
-    {
-      "entropy": 0.31762309623882173,
-      "epoch": 0.9144112577889733,
-      "grad_norm": 0.4568561315536499,
-      "learning_rate": 1.912867304415761e-06,
-      "loss": 0.37861361503601076,
-      "mean_token_accuracy": 0.898035791516304,
-      "num_tokens": 13054862.0,
-      "step": 12070
-    },
-    {
-      "entropy": 0.29679893162101506,
-      "epoch": 0.9151688478948465,
-      "grad_norm": 0.4890940487384796,
-      "learning_rate": 1.8794033119638953e-06,
-      "loss": 0.3435173988342285,
-      "mean_token_accuracy": 0.9060111805796623,
-      "num_tokens": 13159877.0,
-      "step": 12080
-    },
-    {
-      "entropy": 0.26280158162117007,
-      "epoch": 0.9159264380007197,
-      "grad_norm": 0.48253440856933594,
-      "learning_rate": 1.846229013311951e-06,
-      "loss": 0.29681944847106934,
-      "mean_token_accuracy": 0.9169836983084678,
-      "num_tokens": 13250560.0,
-      "step": 12090
-    },
-    {
-      "entropy": 0.3261014189571142,
-      "epoch": 0.9166840281065929,
-      "grad_norm": 0.4755929708480835,
-      "learning_rate": 1.8133446081745886e-06,
-      "loss": 0.36304349899291993,
-      "mean_token_accuracy": 0.8956771790981293,
-      "num_tokens": 13352805.0,
-      "step": 12100
-    },
-    {
-      "entropy": 0.27531050983816385,
-      "epoch": 0.9174416182124662,
-      "grad_norm": 0.4745216965675354,
-      "learning_rate": 1.7807502945212605e-06,
-      "loss": 0.3325106859207153,
-      "mean_token_accuracy": 0.9112949371337891,
-      "num_tokens": 13447502.0,
-      "step": 12110
-    },
-    {
-      "entropy": 0.26114649400115014,
-      "epoch": 0.9181992083183393,
-      "grad_norm": 0.4469934403896332,
-      "learning_rate": 1.748446268575027e-06,
-      "loss": 0.30356202125549314,
-      "mean_token_accuracy": 0.916273832321167,
-      "num_tokens": 13542441.0,
-      "step": 12120
-    },
-    {
-      "entropy": 0.2954039676114917,
-      "epoch": 0.9189567984242126,
-      "grad_norm": 0.4029696583747864,
-      "learning_rate": 1.7164327248113622e-06,
-      "loss": 0.33674476146697996,
-      "mean_token_accuracy": 0.906862536072731,
-      "num_tokens": 13650072.0,
-      "step": 12130
-    },
-    {
-      "entropy": 0.24167130887508392,
-      "epoch": 0.9197143885300858,
-      "grad_norm": 0.44493207335472107,
-      "learning_rate": 1.6847098559569952e-06,
-      "loss": 0.26566371917724607,
-      "mean_token_accuracy": 0.923855085670948,
-      "num_tokens": 13744399.0,
-      "step": 12140
-    },
-    {
-      "entropy": 0.2633110383525491,
-      "epoch": 0.920471978635959,
-      "grad_norm": 0.451709121465683,
-      "learning_rate": 1.653277852988744e-06,
-      "loss": 0.313076114654541,
-      "mean_token_accuracy": 0.9192310333251953,
-      "num_tokens": 13837719.0,
-      "step": 12150
-    },
-    {
-      "entropy": 0.29435538705438374,
-      "epoch": 0.9212295687418323,
-      "grad_norm": 0.39204367995262146,
-      "learning_rate": 1.6221369051323655e-06,
-      "loss": 0.3549853563308716,
-      "mean_token_accuracy": 0.9049896076321602,
-      "num_tokens": 13931355.0,
-      "step": 12160
-    },
-    {
-      "entropy": 0.29209454338997604,
-      "epoch": 0.9219871588477054,
-      "grad_norm": 0.38774025440216064,
-      "learning_rate": 1.5912871998614187e-06,
-      "loss": 0.34168984889984133,
-      "mean_token_accuracy": 0.9070258140563965,
-      "num_tokens": 14039359.0,
-      "step": 12170
-    },
-    {
-      "entropy": 0.31331349443644285,
-      "epoch": 0.9227447489535787,
-      "grad_norm": 0.41252318024635315,
-      "learning_rate": 1.5607289228961374e-06,
-      "loss": 0.3593226432800293,
-      "mean_token_accuracy": 0.8983490958809852,
-      "num_tokens": 14137741.0,
-      "step": 12180
-    },
-    {
-      "entropy": 0.28924416061490776,
-      "epoch": 0.9235023390594519,
-      "grad_norm": 0.5740947127342224,
-      "learning_rate": 1.5304622582023031e-06,
-      "loss": 0.3327749252319336,
-      "mean_token_accuracy": 0.9081077992916107,
-      "num_tokens": 14232465.0,
-      "step": 12190
-    },
-    {
-      "entropy": 0.2850802129134536,
-      "epoch": 0.9242599291653251,
-      "grad_norm": 0.4056800305843353,
-      "learning_rate": 1.5004873879901571e-06,
-      "loss": 0.34341509342193605,
-      "mean_token_accuracy": 0.9104896888136864,
-      "num_tokens": 14326738.0,
-      "step": 12200
-    },
-    {
-      "entropy": 0.2863912934437394,
-      "epoch": 0.9250175192711984,
-      "grad_norm": 0.3346513509750366,
-      "learning_rate": 1.4708044927132737e-06,
-      "loss": 0.35355463027954104,
-      "mean_token_accuracy": 0.9096838116645813,
-      "num_tokens": 14424392.0,
-      "step": 12210
-    },
-    {
-      "entropy": 0.3078114558942616,
-      "epoch": 0.9257751093770715,
-      "grad_norm": 0.38695788383483887,
-      "learning_rate": 1.441413751067494e-06,
-      "loss": 0.367154598236084,
-      "mean_token_accuracy": 0.9017126500606537,
-      "num_tokens": 14534354.0,
-      "step": 12220
-    },
-    {
-      "entropy": 0.32155216187238694,
-      "epoch": 0.9265326994829447,
-      "grad_norm": 0.5330353379249573,
-      "learning_rate": 1.4123153399898614e-06,
-      "loss": 0.36757822036743165,
-      "mean_token_accuracy": 0.8965023845434189,
-      "num_tokens": 14639036.0,
-      "step": 12230
-    },
-    {
-      "entropy": 0.2916434150189161,
-      "epoch": 0.927290289588818,
-      "grad_norm": 0.46392399072647095,
-      "learning_rate": 1.3835094346575262e-06,
-      "loss": 0.33872079849243164,
-      "mean_token_accuracy": 0.9077073231339454,
-      "num_tokens": 14740021.0,
-      "step": 12240
-    },
-    {
-      "entropy": 0.2472118590027094,
-      "epoch": 0.9280478796946912,
-      "grad_norm": 0.37260034680366516,
-      "learning_rate": 1.3549962084867197e-06,
-      "loss": 0.29043848514556886,
-      "mean_token_accuracy": 0.9208429232239723,
-      "num_tokens": 14830662.0,
-      "step": 12250
-    },
-    {
-      "epoch": 0.9280478796946912,
-      "eval_entropy": 0.2879207830890064,
-      "eval_loss": 0.2848566472530365,
-      "eval_mean_token_accuracy": 0.9077223223470123,
-      "eval_num_tokens": 14830662.0,
-      "eval_runtime": 362.6256,
-      "eval_samples_per_second": 5.885,
-      "eval_steps_per_second": 1.473,
-      "step": 12250
-    },
-    {
-      "entropy": 0.27810837496072055,
-      "epoch": 0.9288054698005644,
-      "grad_norm": 0.4152062237262726,
-      "learning_rate": 1.326775833131677e-06,
-      "loss": 0.34612598419189455,
-      "mean_token_accuracy": 0.9133247256278991,
-      "num_tokens": 14925605.0,
-      "step": 12260
-    },
-    {
-      "entropy": 0.2930800588801503,
-      "epoch": 0.9295630599064376,
-      "grad_norm": 0.3420519530773163,
-      "learning_rate": 1.2988484784836497e-06,
-      "loss": 0.3541224956512451,
-      "mean_token_accuracy": 0.9066302254796028,
-      "num_tokens": 15025948.0,
-      "step": 12270
-    },
-    {
-      "entropy": 0.34085101950913665,
-      "epoch": 0.9303206500123108,
-      "grad_norm": 0.42782142758369446,
-      "learning_rate": 1.2712143126698384e-06,
-      "loss": 0.40104336738586427,
-      "mean_token_accuracy": 0.8878615260124206,
-      "num_tokens": 15137971.0,
-      "step": 12280
-    },
-    {
-      "entropy": 0.24184838123619556,
-      "epoch": 0.9310782401181841,
-      "grad_norm": 0.4608571231365204,
-      "learning_rate": 1.2438735020524062e-06,
-      "loss": 0.2901769161224365,
-      "mean_token_accuracy": 0.9232783406972885,
-      "num_tokens": 15234313.0,
-      "step": 12290
-    },
-    {
-      "entropy": 0.2717073464766145,
-      "epoch": 0.9318358302240572,
-      "grad_norm": 0.46019455790519714,
-      "learning_rate": 1.2168262112274676e-06,
-      "loss": 0.3081354141235352,
-      "mean_token_accuracy": 0.9139297619462013,
-      "num_tokens": 15331902.0,
-      "step": 12300
-    },
-    {
-      "entropy": 0.3171485211700201,
-      "epoch": 0.9325934203299305,
-      "grad_norm": 0.44192010164260864,
-      "learning_rate": 1.1900726030241005e-06,
-      "loss": 0.359798789024353,
-      "mean_token_accuracy": 0.899483498930931,
-      "num_tokens": 15431160.0,
-      "step": 12310
-    },
-    {
-      "entropy": 0.29771914295852187,
-      "epoch": 0.9333510104358037,
-      "grad_norm": 0.35575592517852783,
-      "learning_rate": 1.1636128385033696e-06,
-      "loss": 0.32083702087402344,
-      "mean_token_accuracy": 0.9065295130014419,
-      "num_tokens": 15528945.0,
-      "step": 12320
-    },
-    {
-      "entropy": 0.328072508983314,
-      "epoch": 0.9341086005416769,
-      "grad_norm": 0.37298858165740967,
-      "learning_rate": 1.1374470769573541e-06,
-      "loss": 0.3967781066894531,
-      "mean_token_accuracy": 0.8941646784543991,
-      "num_tokens": 15630573.0,
-      "step": 12330
-    },
-    {
-      "entropy": 0.30953355710953473,
-      "epoch": 0.9348661906475502,
-      "grad_norm": 0.4739098846912384,
-      "learning_rate": 1.1115754759081777e-06,
-      "loss": 0.3533921480178833,
-      "mean_token_accuracy": 0.8996481195092201,
-      "num_tokens": 15734620.0,
-      "step": 12340
-    },
-    {
-      "entropy": 0.30768233593553307,
-      "epoch": 0.9356237807534233,
-      "grad_norm": 0.4143330156803131,
-      "learning_rate": 1.0859981911070793e-06,
-      "loss": 0.33932952880859374,
-      "mean_token_accuracy": 0.9027550190687179,
-      "num_tokens": 15833310.0,
-      "step": 12350
-    },
-    {
-      "entropy": 0.2843058746308088,
-      "epoch": 0.9363813708592966,
-      "grad_norm": 0.4079853296279907,
-      "learning_rate": 1.060715376533472e-06,
-      "loss": 0.3476103782653809,
-      "mean_token_accuracy": 0.9069626480340958,
-      "num_tokens": 15933036.0,
-      "step": 12360
-    },
-    {
-      "entropy": 0.2918760960921645,
-      "epoch": 0.9371389609651698,
-      "grad_norm": 0.3709755539894104,
-      "learning_rate": 1.0357271843939864e-06,
-      "loss": 0.32259211540222166,
-      "mean_token_accuracy": 0.9056548669934272,
-      "num_tokens": 16035404.0,
-      "step": 12370
-    },
-    {
-      "entropy": 0.2681634811684489,
-      "epoch": 0.937896551071043,
-      "grad_norm": 0.5038941502571106,
-      "learning_rate": 1.0110337651215995e-06,
-      "loss": 0.31412277221679685,
-      "mean_token_accuracy": 0.915467070043087,
-      "num_tokens": 16132094.0,
-      "step": 12380
-    },
-    {
-      "entropy": 0.311520766839385,
-      "epoch": 0.9386541411769163,
-      "grad_norm": 0.42714932560920715,
-      "learning_rate": 9.866352673747025e-07,
-      "loss": 0.3627468109130859,
-      "mean_token_accuracy": 0.9007305800914764,
-      "num_tokens": 16224173.0,
-      "step": 12390
-    },
-    {
-      "entropy": 0.24273829255253077,
-      "epoch": 0.9394117312827894,
-      "grad_norm": 0.4695494472980499,
-      "learning_rate": 9.62531838036218e-07,
-      "loss": 0.2682880163192749,
-      "mean_token_accuracy": 0.9242408692836761,
-      "num_tokens": 16318335.0,
-      "step": 12400
-    },
-    {
-      "entropy": 0.2871857823804021,
-      "epoch": 0.9401693213886627,
-      "grad_norm": 0.3833423852920532,
-      "learning_rate": 9.387236222126894e-07,
-      "loss": 0.32558517456054686,
-      "mean_token_accuracy": 0.9094377249479294,
-      "num_tokens": 16417299.0,
-      "step": 12410
-    },
-    {
-      "entropy": 0.3126089680939913,
-      "epoch": 0.9409269114945359,
-      "grad_norm": 0.41594237089157104,
-      "learning_rate": 9.152107632334428e-07,
-      "loss": 0.3656469345092773,
-      "mean_token_accuracy": 0.8993785634636879,
-      "num_tokens": 16516774.0,
-      "step": 12420
-    },
-    {
-      "entropy": 0.2751727135851979,
-      "epoch": 0.9416845016004091,
-      "grad_norm": 0.5532023310661316,
-      "learning_rate": 8.9199340264971e-07,
-      "loss": 0.3194823503494263,
-      "mean_token_accuracy": 0.9125665321946144,
-      "num_tokens": 16618168.0,
-      "step": 12430
-    },
-    {
-      "entropy": 0.2662426060065627,
-      "epoch": 0.9424420917062823,
-      "grad_norm": 0.42721033096313477,
-      "learning_rate": 8.690716802337628e-07,
-      "loss": 0.32259302139282225,
-      "mean_token_accuracy": 0.9164244115352631,
-      "num_tokens": 16717553.0,
-      "step": 12440
-    },
-    {
-      "entropy": 0.2952677818015218,
-      "epoch": 0.9431996818121555,
-      "grad_norm": 0.5415988564491272,
-      "learning_rate": 8.464457339780907e-07,
-      "loss": 0.35180134773254396,
-      "mean_token_accuracy": 0.9077468425035476,
-      "num_tokens": 16803125.0,
-      "step": 12450
-    },
-    {
-      "entropy": 0.27926060985773804,
-      "epoch": 0.9439572719180287,
-      "grad_norm": 0.5080968141555786,
-      "learning_rate": 8.241157000945631e-07,
-      "loss": 0.31758029460906984,
-      "mean_token_accuracy": 0.9103856220841408,
-      "num_tokens": 16902308.0,
-      "step": 12460
-    },
-    {
-      "entropy": 0.29949437715113164,
-      "epoch": 0.944714862023902,
-      "grad_norm": 0.5077760815620422,
-      "learning_rate": 8.020817130136083e-07,
-      "loss": 0.3514946460723877,
-      "mean_token_accuracy": 0.9059733942151069,
-      "num_tokens": 17003776.0,
-      "step": 12470
-    },
-    {
-      "entropy": 0.30533884670585393,
-      "epoch": 0.9454724521297752,
-      "grad_norm": 0.47848737239837646,
-      "learning_rate": 7.803439053833961e-07,
-      "loss": 0.36388406753540037,
-      "mean_token_accuracy": 0.9001722037792206,
-      "num_tokens": 17104375.0,
-      "step": 12480
-    },
-    {
-      "entropy": 0.32481123935431244,
-      "epoch": 0.9462300422356484,
-      "grad_norm": 0.3776695728302002,
-      "learning_rate": 7.589024080690621e-07,
-      "loss": 0.3672971725463867,
-      "mean_token_accuracy": 0.8965393304824829,
-      "num_tokens": 17210665.0,
-      "step": 12490
-    },
-    {
-      "entropy": 0.3082854451611638,
-      "epoch": 0.9469876323415216,
-      "grad_norm": 0.3830999732017517,
-      "learning_rate": 7.377573501519019e-07,
-      "loss": 0.34308252334594724,
-      "mean_token_accuracy": 0.9049276232719421,
-      "num_tokens": 17308912.0,
-      "step": 12500
-    },
-    {
-      "epoch": 0.9469876323415216,
-      "eval_entropy": 0.28848998702030054,
-      "eval_loss": 0.2848212718963623,
-      "eval_mean_token_accuracy": 0.9077069452416138,
-      "eval_num_tokens": 17308912.0,
-      "eval_runtime": 363.3483,
-      "eval_samples_per_second": 5.873,
-      "eval_steps_per_second": 1.47,
-      "step": 12500
-    },
-    {
-      "entropy": 0.2735982878133655,
-      "epoch": 0.9477452224473948,
-      "grad_norm": 0.452688068151474,
-      "learning_rate": 7.169088589285944e-07,
-      "loss": 0.33200070858001707,
-      "mean_token_accuracy": 0.9136006936430932,
-      "num_tokens": 17411212.0,
-      "step": 12510
-    },
-    {
-      "entropy": 0.2732477217912674,
-      "epoch": 0.9485028125532681,
-      "grad_norm": 0.6040942668914795,
-      "learning_rate": 6.963570599104297e-07,
-      "loss": 0.32123594284057616,
-      "mean_token_accuracy": 0.9131623834371567,
-      "num_tokens": 17501510.0,
-      "step": 12520
-    },
-    {
-      "entropy": 0.30350907929241655,
-      "epoch": 0.9492604026591412,
-      "grad_norm": 0.36344295740127563,
-      "learning_rate": 6.761020768225823e-07,
-      "loss": 0.3587135076522827,
-      "mean_token_accuracy": 0.9031198397278786,
-      "num_tokens": 17609757.0,
-      "step": 12530
-    },
-    {
-      "entropy": 0.29679438304156064,
-      "epoch": 0.9500179927650145,
-      "grad_norm": 0.6506656408309937,
-      "learning_rate": 6.561440316033284e-07,
-      "loss": 0.3296821594238281,
-      "mean_token_accuracy": 0.9067649766802788,
-      "num_tokens": 17714241.0,
-      "step": 12540
-    },
-    {
-      "entropy": 0.28785474970936775,
-      "epoch": 0.9507755828708877,
-      "grad_norm": 0.4117507040500641,
-      "learning_rate": 6.364830444033465e-07,
-      "loss": 0.32306294441223143,
-      "mean_token_accuracy": 0.9076434358954429,
-      "num_tokens": 17816018.0,
-      "step": 12550
-    },
-    {
-      "entropy": 0.29005246944725516,
-      "epoch": 0.9515331729767609,
-      "grad_norm": 0.3879031538963318,
-      "learning_rate": 6.171192335849674e-07,
-      "loss": 0.34095020294189454,
-      "mean_token_accuracy": 0.9068945035338402,
-      "num_tokens": 17917000.0,
-      "step": 12560
-    },
-    {
-      "entropy": 0.26332274451851845,
-      "epoch": 0.9522907630826342,
-      "grad_norm": 0.4287002682685852,
-      "learning_rate": 5.980527157214755e-07,
-      "loss": 0.2998778820037842,
-      "mean_token_accuracy": 0.9176961481571198,
-      "num_tokens": 18011204.0,
-      "step": 12570
-    },
-    {
-      "entropy": 0.2727514169178903,
-      "epoch": 0.9530483531885073,
-      "grad_norm": 0.5085830688476562,
-      "learning_rate": 5.792836055963979e-07,
-      "loss": 0.2996708393096924,
-      "mean_token_accuracy": 0.9111582413315773,
-      "num_tokens": 18109633.0,
-      "step": 12580
-    },
-    {
-      "entropy": 0.32051171325147154,
-      "epoch": 0.9538059432943806,
-      "grad_norm": 0.45694804191589355,
-      "learning_rate": 5.608120162028274e-07,
-      "loss": 0.37244515419006347,
-      "mean_token_accuracy": 0.8994297966361046,
-      "num_tokens": 18208047.0,
-      "step": 12590
-    },
-    {
-      "entropy": 0.3029838014394045,
-      "epoch": 0.9545635334002538,
-      "grad_norm": 0.4092998802661896,
-      "learning_rate": 5.426380587427282e-07,
-      "loss": 0.3454578399658203,
-      "mean_token_accuracy": 0.9031105905771255,
-      "num_tokens": 18312593.0,
-      "step": 12600
-    },
-    {
-      "entropy": 0.2497973307967186,
-      "epoch": 0.955321123506127,
-      "grad_norm": 0.4349222481250763,
-      "learning_rate": 5.247618426262701e-07,
-      "loss": 0.2817720413208008,
-      "mean_token_accuracy": 0.9180493727326393,
-      "num_tokens": 18404504.0,
-      "step": 12610
-    },
-    {
-      "entropy": 0.27505169734358786,
-      "epoch": 0.9560787136120003,
-      "grad_norm": 0.522512674331665,
-      "learning_rate": 5.071834754711735e-07,
-      "loss": 0.3576214790344238,
-      "mean_token_accuracy": 0.9146435871720314,
-      "num_tokens": 18505157.0,
-      "step": 12620
-    },
-    {
-      "entropy": 0.3274114767089486,
-      "epoch": 0.9568363037178734,
-      "grad_norm": 0.33777111768722534,
-      "learning_rate": 4.899030631020651e-07,
-      "loss": 0.38010802268981936,
-      "mean_token_accuracy": 0.8943869322538376,
-      "num_tokens": 18614727.0,
-      "step": 12630
-    },
-    {
-      "entropy": 0.272926408611238,
-      "epoch": 0.9575938938237467,
-      "grad_norm": 0.4114817976951599,
-      "learning_rate": 4.7292070954983445e-07,
-      "loss": 0.31340248584747316,
-      "mean_token_accuracy": 0.9126570343971252,
-      "num_tokens": 18704775.0,
-      "step": 12640
-    },
-    {
-      "entropy": 0.24151614923030137,
-      "epoch": 0.9583514839296199,
-      "grad_norm": 0.43427976965904236,
-      "learning_rate": 4.5623651705099523e-07,
-      "loss": 0.29600989818573,
-      "mean_token_accuracy": 0.9213308230042457,
-      "num_tokens": 18794797.0,
-      "step": 12650
-    },
-    {
-      "entropy": 0.3089570691809058,
-      "epoch": 0.9591090740354931,
-      "grad_norm": 0.37103182077407837,
-      "learning_rate": 4.398505860471081e-07,
-      "loss": 0.36256885528564453,
-      "mean_token_accuracy": 0.8995309889316558,
-      "num_tokens": 18893714.0,
-      "step": 12660
-    },
-    {
-      "entropy": 0.3140263443812728,
-      "epoch": 0.9598666641413663,
-      "grad_norm": 0.38387659192085266,
-      "learning_rate": 4.237630151841199e-07,
-      "loss": 0.36608157157897947,
-      "mean_token_accuracy": 0.899445253610611,
-      "num_tokens": 19002857.0,
-      "step": 12670
-    },
-    {
-      "entropy": 0.26929782778024675,
-      "epoch": 0.9606242542472395,
-      "grad_norm": 0.4742204248905182,
-      "learning_rate": 4.079739013118256e-07,
-      "loss": 0.2934380531311035,
-      "mean_token_accuracy": 0.9163713708519936,
-      "num_tokens": 19100480.0,
-      "step": 12680
-    },
-    {
-      "entropy": 0.2855357639491558,
-      "epoch": 0.9613818443531128,
-      "grad_norm": 0.4664275348186493,
-      "learning_rate": 3.924833394832517e-07,
-      "loss": 0.3293171405792236,
-      "mean_token_accuracy": 0.906791627407074,
-      "num_tokens": 19194946.0,
-      "step": 12690
-    },
-    {
-      "entropy": 0.3293551471084356,
-      "epoch": 0.962139434458986,
-      "grad_norm": 0.6117649674415588,
-      "learning_rate": 3.7729142295409583e-07,
-      "loss": 0.38983731269836425,
-      "mean_token_accuracy": 0.8953660592436791,
-      "num_tokens": 19293533.0,
-      "step": 12700
-    },
-    {
-      "entropy": 0.2702875768765807,
-      "epoch": 0.9628970245648591,
-      "grad_norm": 0.4539230465888977,
-      "learning_rate": 3.623982431821604e-07,
-      "loss": 0.3196955442428589,
-      "mean_token_accuracy": 0.9138116016983986,
-      "num_tokens": 19389427.0,
-      "step": 12710
-    },
-    {
-      "entropy": 0.3204577224329114,
-      "epoch": 0.9636546146707324,
-      "grad_norm": 0.5138148069381714,
-      "learning_rate": 3.478038898268032e-07,
-      "loss": 0.3672304630279541,
-      "mean_token_accuracy": 0.8978322297334671,
-      "num_tokens": 19495598.0,
-      "step": 12720
-    },
-    {
-      "entropy": 0.31709345784038306,
-      "epoch": 0.9644122047766056,
-      "grad_norm": 0.358312726020813,
-      "learning_rate": 3.335084507483932e-07,
-      "loss": 0.387695837020874,
-      "mean_token_accuracy": 0.8973437681794166,
-      "num_tokens": 19592196.0,
-      "step": 12730
-    },
-    {
-      "entropy": 0.25532410480082035,
-      "epoch": 0.9651697948824788,
-      "grad_norm": 0.40943631529808044,
-      "learning_rate": 3.1951201200780543e-07,
-      "loss": 0.2958423376083374,
-      "mean_token_accuracy": 0.9203238844871521,
-      "num_tokens": 19697481.0,
-      "step": 12740
-    },
-    {
-      "entropy": 0.2499313147738576,
-      "epoch": 0.9659273849883521,
-      "grad_norm": 0.4138835668563843,
-      "learning_rate": 3.058146578658605e-07,
-      "loss": 0.2831275224685669,
-      "mean_token_accuracy": 0.9220834314823151,
-      "num_tokens": 19784316.0,
-      "step": 12750
-    },
-    {
-      "epoch": 0.9659273849883521,
-      "eval_entropy": 0.2880412316235822,
-      "eval_loss": 0.2847772538661957,
-      "eval_mean_token_accuracy": 0.9076900969953573,
-      "eval_num_tokens": 19784316.0,
-      "eval_runtime": 362.2827,
-      "eval_samples_per_second": 5.89,
-      "eval_steps_per_second": 1.474,
-      "step": 12750
-    },
-    {
-      "entropy": 0.2813654486089945,
-      "epoch": 0.9666849750942252,
-      "grad_norm": 0.37392255663871765,
-      "learning_rate": 2.9241647078285805e-07,
-      "loss": 0.3264885902404785,
-      "mean_token_accuracy": 0.9118224963545799,
-      "num_tokens": 19880078.0,
-      "step": 12760
-    },
-    {
-      "entropy": 0.3182709474116564,
-      "epoch": 0.9674425652000985,
-      "grad_norm": 0.3846786618232727,
-      "learning_rate": 2.793175314180607e-07,
-      "loss": 0.37220053672790526,
-      "mean_token_accuracy": 0.9005269780755043,
-      "num_tokens": 19979338.0,
-      "step": 12770
-    },
-    {
-      "entropy": 0.3049146421253681,
-      "epoch": 0.9682001553059717,
-      "grad_norm": 0.377900093793869,
-      "learning_rate": 2.665179186292055e-07,
-      "loss": 0.33737356662750245,
-      "mean_token_accuracy": 0.9025636494159699,
-      "num_tokens": 20085673.0,
-      "step": 12780
-    },
-    {
-      "entropy": 0.28332545049488544,
-      "epoch": 0.9689577454118449,
-      "grad_norm": 0.5380370616912842,
-      "learning_rate": 2.5401770947203753e-07,
-      "loss": 0.316477108001709,
-      "mean_token_accuracy": 0.9086288943886757,
-      "num_tokens": 20174485.0,
-      "step": 12790
-    },
-    {
-      "entropy": 0.30698223169893024,
-      "epoch": 0.9697153355177182,
-      "grad_norm": 0.42323195934295654,
-      "learning_rate": 2.4181697919984925e-07,
-      "loss": 0.33313066959381105,
-      "mean_token_accuracy": 0.9010374918580055,
-      "num_tokens": 20265008.0,
-      "step": 12800
-    },
-    {
-      "entropy": 0.29079750888049605,
-      "epoch": 0.9704729256235913,
-      "grad_norm": 0.5054340958595276,
-      "learning_rate": 2.2991580126301426e-07,
-      "loss": 0.3235260725021362,
-      "mean_token_accuracy": 0.9071223631501197,
-      "num_tokens": 20360911.0,
-      "step": 12810
-    },
-    {
-      "entropy": 0.2743817076086998,
-      "epoch": 0.9712305157294646,
-      "grad_norm": 0.5243687629699707,
-      "learning_rate": 2.1831424730854867e-07,
-      "loss": 0.2899976968765259,
-      "mean_token_accuracy": 0.913919959962368,
-      "num_tokens": 20458621.0,
-      "step": 12820
-    },
-    {
-      "entropy": 0.2847770480439067,
-      "epoch": 0.9719881058353378,
-      "grad_norm": 0.4167582392692566,
-      "learning_rate": 2.0701238717969473e-07,
-      "loss": 0.32479381561279297,
-      "mean_token_accuracy": 0.9088508263230324,
-      "num_tokens": 20549278.0,
-      "step": 12830
-    },
-    {
-      "entropy": 0.27718366011977197,
-      "epoch": 0.972745695941211,
-      "grad_norm": 0.42621564865112305,
-      "learning_rate": 1.9601028891548245e-07,
-      "loss": 0.3166377067565918,
-      "mean_token_accuracy": 0.9102173432707786,
-      "num_tokens": 20650288.0,
-      "step": 12840
-    },
-    {
-      "entropy": 0.3363506423309445,
-      "epoch": 0.9735032860470842,
-      "grad_norm": 0.4228529632091522,
-      "learning_rate": 1.8530801875033532e-07,
-      "loss": 0.39524827003479,
-      "mean_token_accuracy": 0.8900739133358002,
-      "num_tokens": 20752131.0,
-      "step": 12850
-    },
-    {
-      "entropy": 0.2835741061717272,
-      "epoch": 0.9742608761529574,
-      "grad_norm": 0.4159538447856903,
-      "learning_rate": 1.7490564111364293e-07,
-      "loss": 0.33005437850952146,
-      "mean_token_accuracy": 0.909838430583477,
-      "num_tokens": 20857249.0,
-      "step": 12860
-    },
-    {
-      "entropy": 0.2822339989244938,
-      "epoch": 0.9750184662588307,
-      "grad_norm": 0.4528020918369293,
-      "learning_rate": 1.648032186294224e-07,
-      "loss": 0.3275020599365234,
-      "mean_token_accuracy": 0.9100241079926491,
-      "num_tokens": 20964081.0,
-      "step": 12870
-    },
-    {
-      "entropy": 0.2959194635972381,
-      "epoch": 0.9757760563647039,
-      "grad_norm": 0.3836463391780853,
-      "learning_rate": 1.5500081211588523e-07,
-      "loss": 0.3620389461517334,
-      "mean_token_accuracy": 0.9032949566841125,
-      "num_tokens": 21068983.0,
-      "step": 12880
-    },
-    {
-      "entropy": 0.2801917511969805,
-      "epoch": 0.976533646470577,
-      "grad_norm": 0.41730958223342896,
-      "learning_rate": 1.4549848058511008e-07,
-      "loss": 0.3333770751953125,
-      "mean_token_accuracy": 0.9115513071417809,
-      "num_tokens": 21168935.0,
-      "step": 12890
-    },
-    {
-      "entropy": 0.2623455310240388,
-      "epoch": 0.9772912365764503,
-      "grad_norm": 0.3546130657196045,
-      "learning_rate": 1.3629628124268156e-07,
-      "loss": 0.2910778999328613,
-      "mean_token_accuracy": 0.9170158222317696,
-      "num_tokens": 21267005.0,
-      "step": 12900
-    },
-    {
-      "entropy": 0.2856149373576045,
-      "epoch": 0.9780488266823235,
-      "grad_norm": 0.3981967866420746,
-      "learning_rate": 1.2739426948732424e-07,
-      "loss": 0.2948747634887695,
-      "mean_token_accuracy": 0.9070953279733658,
-      "num_tokens": 21367064.0,
-      "step": 12910
-    },
-    {
-      "entropy": 0.2606488248333335,
-      "epoch": 0.9788064167881968,
-      "grad_norm": 0.4281143248081207,
-      "learning_rate": 1.1879249891059707e-07,
-      "loss": 0.29581537246704104,
-      "mean_token_accuracy": 0.9164496824145317,
-      "num_tokens": 21464079.0,
-      "step": 12920
-    },
-    {
-      "entropy": 0.29077529795467855,
-      "epoch": 0.97956400689407,
-      "grad_norm": 0.4827267825603485,
-      "learning_rate": 1.1049102129655487e-07,
-      "loss": 0.34884982109069823,
-      "mean_token_accuracy": 0.9087532594799995,
-      "num_tokens": 21567841.0,
-      "step": 12930
-    },
-    {
-      "entropy": 0.28863470936194063,
-      "epoch": 0.9803215969999431,
-      "grad_norm": 0.5885609984397888,
-      "learning_rate": 1.0248988662144299e-07,
-      "loss": 0.33504061698913573,
-      "mean_token_accuracy": 0.9087083443999291,
-      "num_tokens": 21666124.0,
-      "step": 12940
-    },
-    {
-      "entropy": 0.32063686568289995,
-      "epoch": 0.9810791871058164,
-      "grad_norm": 0.4309108257293701,
-      "learning_rate": 9.478914305338093e-08,
-      "loss": 0.36399550437927247,
-      "mean_token_accuracy": 0.8976536452770233,
-      "num_tokens": 21775070.0,
-      "step": 12950
-    },
-    {
-      "entropy": 0.2755927896127105,
-      "epoch": 0.9818367772116896,
-      "grad_norm": 0.4799281358718872,
-      "learning_rate": 8.738883695210698e-08,
-      "loss": 0.32933473587036133,
-      "mean_token_accuracy": 0.9116968855261802,
-      "num_tokens": 21870056.0,
-      "step": 12960
-    },
-    {
-      "entropy": 0.29384942166507244,
-      "epoch": 0.9825943673175628,
-      "grad_norm": 0.5525132417678833,
-      "learning_rate": 8.02890128686562e-08,
-      "loss": 0.3565049648284912,
-      "mean_token_accuracy": 0.9026440948247909,
-      "num_tokens": 21968437.0,
-      "step": 12970
-    },
-    {
-      "entropy": 0.32599514480680225,
-      "epoch": 0.9833519574234361,
-      "grad_norm": 0.3205714523792267,
-      "learning_rate": 7.34897135451218e-08,
-      "loss": 0.3695268392562866,
-      "mean_token_accuracy": 0.8950692549347877,
-      "num_tokens": 22079777.0,
-      "step": 12980
-    },
-    {
-      "entropy": 0.2817994005978107,
-      "epoch": 0.9841095475293092,
-      "grad_norm": 0.40537744760513306,
-      "learning_rate": 6.699097991438863e-08,
-      "loss": 0.34434051513671876,
-      "mean_token_accuracy": 0.9097468137741089,
-      "num_tokens": 22178916.0,
-      "step": 12990
-    },
-    {
-      "entropy": 0.25305229146033525,
-      "epoch": 0.9848671376351825,
-      "grad_norm": 0.5301713943481445,
-      "learning_rate": 6.079285109989452e-08,
-      "loss": 0.2990108013153076,
-      "mean_token_accuracy": 0.9184802591800689,
-      "num_tokens": 22276240.0,
-      "step": 13000
-    },
-    {
-      "epoch": 0.9848671376351825,
-      "eval_entropy": 0.2880801955370532,
-      "eval_loss": 0.2847670912742615,
-      "eval_mean_token_accuracy": 0.9077237584394462,
-      "eval_num_tokens": 22276240.0,
-      "eval_runtime": 357.9585,
-      "eval_samples_per_second": 5.962,
-      "eval_steps_per_second": 1.492,
-      "step": 13000
     }
   ],
   "logging_steps": 10,
@@ -13599,7 +11250,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.1281853685296005e+18,
+  "total_flos": 9.324210024475525e+17,
   "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null