Upload folder using huggingface_hub

adapter_config.json

@@ -26,9 +26,9 @@
   "revision": null,
   "target_modules": [
     "k",
+    "q",
     "v",
-    "o",
-    "q"
+    "o"
   ],
   "target_parameters": null,
   "task_type": "SEQ_2_SEQ_LM",

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1c9be41482f0a62915c16fac214840b64998ad4e4963456fb13a51defeb9bc73
+oid sha256:797dda014330e240bbf2198fbfecf24161302013ea18d549054ff8acca7a03b0
 size 151086440

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7d2cd9547fae1ebf985668f0dc3f978ab6cac637ecea25cbe7be4a3500598ddf
+oid sha256:e01994dedaff9226af3026ef9a65b2747ea0b2bfeb83e960d7a79d2048512c6c
 size 302477347

rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f8cd94ba93502a1e3b3dbda1bb885a39125c1c2d94bcd2febfa4955fb9c3624e
+oid sha256:0ca140b3697ff3374a23729703049da223b8dbaa84a333abd9ccf3a5a80b6436
 size 14645

scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:97f6a9ab565c9d8de5b29c1536f49a6411be870ff0c5894d4016ce822ee32d94
+oid sha256:a78cb00c76b19fb842c2de927f65fd790fb1d7e6237168bbf05560ea87bc9210
 size 1465

trainer_state.json

@@ -1,1230 +1,1402 @@
 {
   "best_global_step": null,
-  "best_metric": 0.4038073718547821,
+  "best_metric": 0.3724241852760315,
   "best_model_checkpoint": null,
-  "epoch": 3.899815440331511,
+  "epoch": 3.9614956048037637,
   "eval_steps": 500,
-  "global_step": 14000,
+  "global_step": 16000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.0002785806316815823,
-      "grad_norm": 0.4645193815231323,
+      "epoch": 0.0002476166893648632,
+      "grad_norm": 0.5023800134658813,
       "learning_rate": 0.0,
-      "loss": 2.3986,
+      "loss": 3.0743,
       "step": 1
     },
     {
-      "epoch": 0.027858063168158233,
-      "grad_norm": 0.3897116482257843,
+      "epoch": 0.02476166893648632,
+      "grad_norm": 0.4190014898777008,
       "learning_rate": 1.9800000000000004e-05,
-      "loss": 2.4541,
+      "loss": 2.6667,
       "step": 100
     },
     {
-      "epoch": 0.055716126336316465,
-      "grad_norm": 0.9093355536460876,
+      "epoch": 0.04952333787297264,
+      "grad_norm": 0.7991347312927246,
       "learning_rate": 3.9800000000000005e-05,
-      "loss": 1.4843,
+      "loss": 1.4831,
       "step": 200
     },
     {
-      "epoch": 0.0835741895044747,
-      "grad_norm": 1.0977619886398315,
-      "learning_rate": 4.9826364280652023e-05,
-      "loss": 0.9759,
+      "epoch": 0.07428500680945896,
+      "grad_norm": 1.1581072807312012,
+      "learning_rate": 4.9877162196039115e-05,
+      "loss": 0.9423,
       "step": 300
     },
     {
-      "epoch": 0.11143225267263293,
-      "grad_norm": 1.0786027908325195,
-      "learning_rate": 4.9472005669737774e-05,
-      "loss": 0.8439,
+      "epoch": 0.09904667574594528,
+      "grad_norm": 1.1347203254699707,
+      "learning_rate": 4.962647280020055e-05,
+      "loss": 0.8084,
       "step": 400
     },
     {
-      "epoch": 0.13929031584079116,
-      "grad_norm": 1.2702925205230713,
-      "learning_rate": 4.911764705882353e-05,
-      "loss": 0.7883,
+      "epoch": 0.1238083446824316,
+      "grad_norm": 1.1398347616195679,
+      "learning_rate": 4.9375783404362e-05,
+      "loss": 0.7511,
       "step": 500
     },
     {
-      "epoch": 0.13929031584079116,
-      "eval_loss": 0.65428227186203,
-      "eval_runtime": 591.887,
-      "eval_samples_per_second": 48.518,
-      "eval_steps_per_second": 3.033,
+      "epoch": 0.1238083446824316,
+      "eval_loss": 0.6198802590370178,
+      "eval_runtime": 241.1394,
+      "eval_samples_per_second": 59.546,
+      "eval_steps_per_second": 3.724,
       "step": 500
     },
     {
-      "epoch": 0.1671483790089494,
-      "grad_norm": 1.3737353086471558,
-      "learning_rate": 4.876328844790929e-05,
-      "loss": 0.746,
+      "epoch": 0.14857001361891792,
+      "grad_norm": 1.1752710342407227,
+      "learning_rate": 4.912509400852344e-05,
+      "loss": 0.7039,
       "step": 600
     },
     {
-      "epoch": 0.19500644217710764,
-      "grad_norm": 1.3053338527679443,
-      "learning_rate": 4.840892983699504e-05,
-      "loss": 0.7122,
+      "epoch": 0.17333168255540424,
+      "grad_norm": 1.876571774482727,
+      "learning_rate": 4.887440461268489e-05,
+      "loss": 0.6632,
       "step": 700
     },
     {
-      "epoch": 0.22286450534526586,
-      "grad_norm": 1.814271092414856,
-      "learning_rate": 4.80545712260808e-05,
-      "loss": 0.6914,
+      "epoch": 0.19809335149189056,
+      "grad_norm": 1.3733500242233276,
+      "learning_rate": 4.8623715216846324e-05,
+      "loss": 0.6598,
       "step": 800
     },
     {
-      "epoch": 0.2507225685134241,
-      "grad_norm": 1.3019365072250366,
-      "learning_rate": 4.770021261516655e-05,
-      "loss": 0.6831,
+      "epoch": 0.22285502042837688,
+      "grad_norm": 1.16924250125885,
+      "learning_rate": 4.837302582100777e-05,
+      "loss": 0.6492,
       "step": 900
     },
     {
-      "epoch": 0.2785806316815823,
-      "grad_norm": 1.2910503149032593,
-      "learning_rate": 4.7345854004252306e-05,
-      "loss": 0.6714,
+      "epoch": 0.2476166893648632,
+      "grad_norm": 1.4239447116851807,
+      "learning_rate": 4.8122336425169215e-05,
+      "loss": 0.6285,
       "step": 1000
     },
     {
-      "epoch": 0.2785806316815823,
-      "eval_loss": 0.5705545544624329,
-      "eval_runtime": 592.3805,
-      "eval_samples_per_second": 48.477,
-      "eval_steps_per_second": 3.03,
+      "epoch": 0.2476166893648632,
+      "eval_loss": 0.5310020446777344,
+      "eval_runtime": 240.9166,
+      "eval_samples_per_second": 59.602,
+      "eval_steps_per_second": 3.727,
       "step": 1000
     },
     {
-      "epoch": 0.30643869484974057,
-      "grad_norm": 1.4649542570114136,
-      "learning_rate": 4.699149539333806e-05,
-      "loss": 0.6487,
+      "epoch": 0.2723783583013495,
+      "grad_norm": 1.2799785137176514,
+      "learning_rate": 4.787164702933066e-05,
+      "loss": 0.6147,
       "step": 1100
     },
     {
-      "epoch": 0.3342967580178988,
-      "grad_norm": 1.7467780113220215,
-      "learning_rate": 4.6637136782423815e-05,
-      "loss": 0.6456,
+      "epoch": 0.29714002723783584,
+      "grad_norm": 1.404555320739746,
+      "learning_rate": 4.7620957633492105e-05,
+      "loss": 0.6138,
       "step": 1200
     },
     {
-      "epoch": 0.362154821186057,
-      "grad_norm": 1.33451247215271,
-      "learning_rate": 4.628277817150957e-05,
-      "loss": 0.637,
+      "epoch": 0.3219016961743221,
+      "grad_norm": 1.3747936487197876,
+      "learning_rate": 4.737026823765355e-05,
+      "loss": 0.6068,
       "step": 1300
     },
     {
-      "epoch": 0.3900128843542153,
-      "grad_norm": 1.7314374446868896,
-      "learning_rate": 4.592841956059532e-05,
-      "loss": 0.6295,
+      "epoch": 0.3466633651108085,
+      "grad_norm": 1.4329625368118286,
+      "learning_rate": 4.7119578841814996e-05,
+      "loss": 0.5859,
       "step": 1400
     },
     {
-      "epoch": 0.4178709475223735,
-      "grad_norm": 1.344909906387329,
-      "learning_rate": 4.557406094968108e-05,
-      "loss": 0.621,
+      "epoch": 0.37142503404729477,
+      "grad_norm": 1.36211097240448,
+      "learning_rate": 4.686888944597644e-05,
+      "loss": 0.5769,
       "step": 1500
     },
     {
-      "epoch": 0.4178709475223735,
-      "eval_loss": 0.5306675434112549,
-      "eval_runtime": 592.5408,
-      "eval_samples_per_second": 48.464,
-      "eval_steps_per_second": 3.029,
+      "epoch": 0.37142503404729477,
+      "eval_loss": 0.4986073076725006,
+      "eval_runtime": 240.2236,
+      "eval_samples_per_second": 59.773,
+      "eval_steps_per_second": 3.738,
       "step": 1500
     },
     {
-      "epoch": 0.4457290106905317,
-      "grad_norm": 1.3075149059295654,
-      "learning_rate": 4.521970233876683e-05,
-      "loss": 0.6229,
+      "epoch": 0.3961867029837811,
+      "grad_norm": 1.0601556301116943,
+      "learning_rate": 4.6618200050137886e-05,
+      "loss": 0.5755,
       "step": 1600
     },
     {
-      "epoch": 0.47358707385869,
-      "grad_norm": 1.4076815843582153,
-      "learning_rate": 4.486534372785259e-05,
-      "loss": 0.6109,
+      "epoch": 0.4209483719202674,
+      "grad_norm": 1.286715030670166,
+      "learning_rate": 4.636751065429932e-05,
+      "loss": 0.5595,
       "step": 1700
     },
     {
-      "epoch": 0.5014451370268482,
-      "grad_norm": 1.3399609327316284,
-      "learning_rate": 4.451098511693834e-05,
-      "loss": 0.6024,
+      "epoch": 0.44571004085675375,
+      "grad_norm": 1.284048318862915,
+      "learning_rate": 4.611682125846077e-05,
+      "loss": 0.5558,
       "step": 1800
     },
     {
-      "epoch": 0.5293032001950064,
-      "grad_norm": 1.3271087408065796,
-      "learning_rate": 4.41566265060241e-05,
-      "loss": 0.5997,
+      "epoch": 0.47047170979324004,
+      "grad_norm": 1.2716588973999023,
+      "learning_rate": 4.586613186262221e-05,
+      "loss": 0.5504,
       "step": 1900
     },
     {
-      "epoch": 0.5571612633631646,
-      "grad_norm": 1.412111520767212,
-      "learning_rate": 4.3802267895109855e-05,
-      "loss": 0.5839,
+      "epoch": 0.4952333787297264,
+      "grad_norm": 1.1481940746307373,
+      "learning_rate": 4.561544246678366e-05,
+      "loss": 0.5519,
       "step": 2000
     },
     {
-      "epoch": 0.5571612633631646,
-      "eval_loss": 0.5096445679664612,
-      "eval_runtime": 594.5376,
-      "eval_samples_per_second": 48.301,
-      "eval_steps_per_second": 3.019,
+      "epoch": 0.4952333787297264,
+      "eval_loss": 0.476500004529953,
+      "eval_runtime": 240.9166,
+      "eval_samples_per_second": 59.602,
+      "eval_steps_per_second": 3.727,
       "step": 2000
     },
     {
-      "epoch": 0.5850193265313229,
-      "grad_norm": 1.2040321826934814,
-      "learning_rate": 4.3447909284195606e-05,
-      "loss": 0.5789,
+      "epoch": 0.5199950476662127,
+      "grad_norm": 1.0755871534347534,
+      "learning_rate": 4.5364753070945096e-05,
+      "loss": 0.5465,
       "step": 2100
     },
     {
-      "epoch": 0.6128773896994811,
-      "grad_norm": 1.2207297086715698,
-      "learning_rate": 4.3093550673281363e-05,
-      "loss": 0.5761,
+      "epoch": 0.544756716602699,
+      "grad_norm": 1.3121910095214844,
+      "learning_rate": 4.5114063675106544e-05,
+      "loss": 0.5438,
       "step": 2200
     },
     {
-      "epoch": 0.6407354528676393,
-      "grad_norm": 1.205450177192688,
-      "learning_rate": 4.2739192062367114e-05,
-      "loss": 0.5709,
+      "epoch": 0.5695183855391853,
+      "grad_norm": 1.2936429977416992,
+      "learning_rate": 4.4863374279267986e-05,
+      "loss": 0.5379,
       "step": 2300
     },
     {
-      "epoch": 0.6685935160357976,
-      "grad_norm": 1.3946952819824219,
-      "learning_rate": 4.238483345145287e-05,
-      "loss": 0.5696,
+      "epoch": 0.5942800544756717,
+      "grad_norm": 1.158836841583252,
+      "learning_rate": 4.4612684883429435e-05,
+      "loss": 0.5325,
       "step": 2400
     },
     {
-      "epoch": 0.6964515792039558,
-      "grad_norm": 1.1846948862075806,
-      "learning_rate": 4.203047484053862e-05,
-      "loss": 0.5552,
+      "epoch": 0.619041723412158,
+      "grad_norm": 1.0221409797668457,
+      "learning_rate": 4.436199548759088e-05,
+      "loss": 0.5313,
       "step": 2500
     },
     {
-      "epoch": 0.6964515792039558,
-      "eval_loss": 0.4957100749015808,
-      "eval_runtime": 593.2402,
-      "eval_samples_per_second": 48.407,
-      "eval_steps_per_second": 3.026,
+      "epoch": 0.619041723412158,
+      "eval_loss": 0.45980772376060486,
+      "eval_runtime": 240.2384,
+      "eval_samples_per_second": 59.77,
+      "eval_steps_per_second": 3.738,
       "step": 2500
     },
     {
-      "epoch": 0.724309642372114,
-      "grad_norm": 1.2646228075027466,
-      "learning_rate": 4.167611622962439e-05,
-      "loss": 0.5573,
+      "epoch": 0.6438033923486443,
+      "grad_norm": 1.4003841876983643,
+      "learning_rate": 4.411130609175232e-05,
+      "loss": 0.5246,
       "step": 2600
     },
     {
-      "epoch": 0.7521677055402723,
-      "grad_norm": 1.2027941942214966,
-      "learning_rate": 4.132175761871014e-05,
-      "loss": 0.5568,
+      "epoch": 0.6685650612851306,
+      "grad_norm": 1.0657808780670166,
+      "learning_rate": 4.386061669591377e-05,
+      "loss": 0.514,
       "step": 2700
     },
     {
-      "epoch": 0.7800257687084305,
-      "grad_norm": 1.2179590463638306,
-      "learning_rate": 4.0967399007795895e-05,
-      "loss": 0.5579,
+      "epoch": 0.693326730221617,
+      "grad_norm": 1.012877106666565,
+      "learning_rate": 4.360992730007521e-05,
+      "loss": 0.5117,
       "step": 2800
     },
     {
-      "epoch": 0.8078838318765887,
-      "grad_norm": 1.2543010711669922,
-      "learning_rate": 4.0613040396881646e-05,
-      "loss": 0.5632,
+      "epoch": 0.7180883991581033,
+      "grad_norm": 1.0176758766174316,
+      "learning_rate": 4.335923790423666e-05,
+      "loss": 0.5149,
       "step": 2900
     },
     {
-      "epoch": 0.835741895044747,
-      "grad_norm": 1.2597264051437378,
-      "learning_rate": 4.02586817859674e-05,
-      "loss": 0.545,
+      "epoch": 0.7428500680945895,
+      "grad_norm": 1.177230954170227,
+      "learning_rate": 4.310854850839809e-05,
+      "loss": 0.5074,
       "step": 3000
     },
     {
-      "epoch": 0.835741895044747,
-      "eval_loss": 0.4792587161064148,
-      "eval_runtime": 592.7585,
-      "eval_samples_per_second": 48.446,
-      "eval_steps_per_second": 3.028,
+      "epoch": 0.7428500680945895,
+      "eval_loss": 0.45146670937538147,
+      "eval_runtime": 241.1985,
+      "eval_samples_per_second": 59.532,
+      "eval_steps_per_second": 3.723,
       "step": 3000
     },
     {
-      "epoch": 0.8635999582129052,
-      "grad_norm": 1.3292843103408813,
-      "learning_rate": 3.9904323175053155e-05,
-      "loss": 0.5571,
+      "epoch": 0.7676117370310759,
+      "grad_norm": 1.0709675550460815,
+      "learning_rate": 4.285785911255954e-05,
+      "loss": 0.5095,
       "step": 3100
     },
     {
-      "epoch": 0.8914580213810634,
-      "grad_norm": 1.3142098188400269,
-      "learning_rate": 3.9549964564138906e-05,
-      "loss": 0.5397,
+      "epoch": 0.7923734059675622,
+      "grad_norm": 1.0372568368911743,
+      "learning_rate": 4.260716971672098e-05,
+      "loss": 0.4988,
       "step": 3200
     },
     {
-      "epoch": 0.9193160845492218,
-      "grad_norm": 1.389565348625183,
-      "learning_rate": 3.919560595322467e-05,
-      "loss": 0.5478,
+      "epoch": 0.8171350749040486,
+      "grad_norm": 1.3450722694396973,
+      "learning_rate": 4.235648032088243e-05,
+      "loss": 0.4972,
       "step": 3300
     },
     {
-      "epoch": 0.94717414771738,
-      "grad_norm": 1.1616095304489136,
-      "learning_rate": 3.884124734231042e-05,
-      "loss": 0.5396,
+      "epoch": 0.8418967438405348,
+      "grad_norm": 1.0184952020645142,
+      "learning_rate": 4.2105790925043874e-05,
+      "loss": 0.4986,
       "step": 3400
     },
     {
-      "epoch": 0.9750322108855382,
-      "grad_norm": 1.24950110912323,
-      "learning_rate": 3.848688873139618e-05,
-      "loss": 0.5337,
+      "epoch": 0.8666584127770212,
+      "grad_norm": 1.237382173538208,
+      "learning_rate": 4.1855101529205316e-05,
+      "loss": 0.4985,
       "step": 3500
     },
     {
-      "epoch": 0.9750322108855382,
-      "eval_loss": 0.4686237871646881,
-      "eval_runtime": 593.0319,
-      "eval_samples_per_second": 48.424,
-      "eval_steps_per_second": 3.027,
+      "epoch": 0.8666584127770212,
+      "eval_loss": 0.4388951063156128,
+      "eval_runtime": 240.9275,
+      "eval_samples_per_second": 59.599,
+      "eval_steps_per_second": 3.727,
       "step": 3500
     },
     {
-      "epoch": 1.0027858063168158,
-      "grad_norm": 1.3192429542541504,
-      "learning_rate": 3.813253012048193e-05,
-      "loss": 0.5301,
+      "epoch": 0.8914200817135075,
+      "grad_norm": 1.2948626279830933,
+      "learning_rate": 4.160441213336676e-05,
+      "loss": 0.4926,
       "step": 3600
     },
     {
-      "epoch": 1.030643869484974,
-      "grad_norm": 1.465832233428955,
-      "learning_rate": 3.777817150956769e-05,
-      "loss": 0.5272,
+      "epoch": 0.9161817506499939,
+      "grad_norm": 1.287011981010437,
+      "learning_rate": 4.1353722737528206e-05,
+      "loss": 0.4911,
       "step": 3700
     },
     {
-      "epoch": 1.0585019326531322,
-      "grad_norm": 1.4403393268585205,
-      "learning_rate": 3.742381289865344e-05,
-      "loss": 0.5236,
+      "epoch": 0.9409434195864801,
+      "grad_norm": 1.120126485824585,
+      "learning_rate": 4.110303334168965e-05,
+      "loss": 0.4995,
       "step": 3800
     },
     {
-      "epoch": 1.0863599958212906,
-      "grad_norm": 1.3801666498184204,
-      "learning_rate": 3.706945428773919e-05,
-      "loss": 0.5242,
+      "epoch": 0.9657050885229664,
+      "grad_norm": 0.967438817024231,
+      "learning_rate": 4.085234394585109e-05,
+      "loss": 0.4977,
       "step": 3900
     },
     {
-      "epoch": 1.1142180589894488,
-      "grad_norm": 1.076539397239685,
-      "learning_rate": 3.671509567682495e-05,
-      "loss": 0.5224,
+      "epoch": 0.9904667574594528,
+      "grad_norm": 1.091957926750183,
+      "learning_rate": 4.060165455001254e-05,
+      "loss": 0.4926,
       "step": 4000
     },
     {
-      "epoch": 1.1142180589894488,
-      "eval_loss": 0.46663618087768555,
-      "eval_runtime": 592.8183,
-      "eval_samples_per_second": 48.441,
-      "eval_steps_per_second": 3.028,
+      "epoch": 0.9904667574594528,
+      "eval_loss": 0.43045303225517273,
+      "eval_runtime": 240.1878,
+      "eval_samples_per_second": 59.782,
+      "eval_steps_per_second": 3.739,
       "step": 4000
     },
     {
-      "epoch": 1.142076122157607,
-      "grad_norm": 1.3829838037490845,
-      "learning_rate": 3.6360737065910704e-05,
-      "loss": 0.5217,
+      "epoch": 1.0151046180512566,
+      "grad_norm": 1.2335127592086792,
+      "learning_rate": 4.035096515417398e-05,
+      "loss": 0.4811,
       "step": 4100
     },
     {
-      "epoch": 1.1699341853257652,
-      "grad_norm": 1.098763108253479,
-      "learning_rate": 3.600637845499646e-05,
-      "loss": 0.5169,
+      "epoch": 1.0398662869877429,
+      "grad_norm": 1.196088194847107,
+      "learning_rate": 4.010027575833543e-05,
+      "loss": 0.4811,
       "step": 4200
     },
     {
-      "epoch": 1.1977922484939234,
-      "grad_norm": 1.260761022567749,
-      "learning_rate": 3.565201984408221e-05,
-      "loss": 0.5116,
+      "epoch": 1.0646279559242293,
+      "grad_norm": 0.9745426177978516,
+      "learning_rate": 3.9849586362496864e-05,
+      "loss": 0.4761,
       "step": 4300
     },
     {
-      "epoch": 1.2256503116620816,
-      "grad_norm": 1.3205015659332275,
-      "learning_rate": 3.529766123316797e-05,
-      "loss": 0.518,
+      "epoch": 1.0893896248607156,
+      "grad_norm": 1.0584094524383545,
+      "learning_rate": 3.959889696665831e-05,
+      "loss": 0.481,
       "step": 4400
     },
     {
-      "epoch": 1.2535083748302398,
-      "grad_norm": 1.3871066570281982,
-      "learning_rate": 3.494330262225372e-05,
-      "loss": 0.5181,
+      "epoch": 1.114151293797202,
+      "grad_norm": 0.9902614951133728,
+      "learning_rate": 3.9348207570819755e-05,
+      "loss": 0.4773,
       "step": 4500
     },
     {
-      "epoch": 1.2535083748302398,
-      "eval_loss": 0.45501282811164856,
-      "eval_runtime": 592.8496,
-      "eval_samples_per_second": 48.439,
-      "eval_steps_per_second": 3.028,
+      "epoch": 1.114151293797202,
+      "eval_loss": 0.4318952262401581,
+      "eval_runtime": 239.8947,
+      "eval_samples_per_second": 59.855,
+      "eval_steps_per_second": 3.743,
       "step": 4500
     },
     {
-      "epoch": 1.2813664379983982,
-      "grad_norm": 1.1054390668869019,
-      "learning_rate": 3.458894401133948e-05,
-      "loss": 0.5148,
+      "epoch": 1.1389129627336882,
+      "grad_norm": 0.948016881942749,
+      "learning_rate": 3.9097518174981204e-05,
+      "loss": 0.4763,
       "step": 4600
     },
     {
-      "epoch": 1.3092245011665564,
-      "grad_norm": 1.0034719705581665,
-      "learning_rate": 3.4234585400425235e-05,
-      "loss": 0.5109,
+      "epoch": 1.1636746316701745,
+      "grad_norm": 1.172553300857544,
+      "learning_rate": 3.8846828779142645e-05,
+      "loss": 0.4803,
       "step": 4700
     },
     {
-      "epoch": 1.3370825643347146,
-      "grad_norm": 1.417938232421875,
-      "learning_rate": 3.3880226789510986e-05,
-      "loss": 0.5104,
+      "epoch": 1.188436300606661,
+      "grad_norm": 0.985139012336731,
+      "learning_rate": 3.859613938330409e-05,
+      "loss": 0.4788,
       "step": 4800
     },
     {
-      "epoch": 1.3649406275028728,
-      "grad_norm": 1.243116855621338,
-      "learning_rate": 3.3525868178596744e-05,
-      "loss": 0.4988,
+      "epoch": 1.2131979695431472,
+      "grad_norm": 1.046460509300232,
+      "learning_rate": 3.834544998746553e-05,
+      "loss": 0.472,
       "step": 4900
     },
     {
-      "epoch": 1.392798690671031,
-      "grad_norm": 1.1632623672485352,
-      "learning_rate": 3.3171509567682495e-05,
-      "loss": 0.5072,
+      "epoch": 1.2379596384796336,
+      "grad_norm": 0.9939820766448975,
+      "learning_rate": 3.809476059162698e-05,
+      "loss": 0.4686,
       "step": 5000
     },
     {
-      "epoch": 1.392798690671031,
-      "eval_loss": 0.45067891478538513,
-      "eval_runtime": 593.6193,
-      "eval_samples_per_second": 48.376,
-      "eval_steps_per_second": 3.024,
+      "epoch": 1.2379596384796336,
+      "eval_loss": 0.42126885056495667,
+      "eval_runtime": 241.2272,
+      "eval_samples_per_second": 59.525,
+      "eval_steps_per_second": 3.723,
       "step": 5000
     },
     {
-      "epoch": 1.4206567538391894,
-      "grad_norm": 1.3165979385375977,
-      "learning_rate": 3.281715095676825e-05,
-      "loss": 0.4934,
+      "epoch": 1.2627213074161199,
+      "grad_norm": 1.1353379487991333,
+      "learning_rate": 3.784407119578842e-05,
+      "loss": 0.467,
       "step": 5100
     },
     {
-      "epoch": 1.4485148170073476,
-      "grad_norm": 1.3166348934173584,
-      "learning_rate": 3.2462792345854e-05,
-      "loss": 0.51,
+      "epoch": 1.2874829763526061,
+      "grad_norm": 1.0256654024124146,
+      "learning_rate": 3.759338179994986e-05,
+      "loss": 0.4652,
       "step": 5200
     },
     {
-      "epoch": 1.4763728801755058,
-      "grad_norm": 1.737769365310669,
-      "learning_rate": 3.210843373493976e-05,
-      "loss": 0.5079,
+      "epoch": 1.3122446452890926,
+      "grad_norm": 1.0825589895248413,
+      "learning_rate": 3.734269240411131e-05,
+      "loss": 0.4639,
       "step": 5300
     },
     {
-      "epoch": 1.504230943343664,
-      "grad_norm": 1.0135122537612915,
-      "learning_rate": 3.175407512402552e-05,
-      "loss": 0.4915,
+      "epoch": 1.3370063142255788,
+      "grad_norm": 1.0456293821334839,
+      "learning_rate": 3.709200300827275e-05,
+      "loss": 0.4716,
       "step": 5400
     },
     {
-      "epoch": 1.5320890065118222,
-      "grad_norm": 1.161383032798767,
-      "learning_rate": 3.139971651311127e-05,
-      "loss": 0.4972,
+      "epoch": 1.361767983162065,
+      "grad_norm": 0.9378584623336792,
+      "learning_rate": 3.6841313612434194e-05,
+      "loss": 0.4652,
       "step": 5500
     },
     {
-      "epoch": 1.5320890065118222,
-      "eval_loss": 0.4426087737083435,
-      "eval_runtime": 592.4939,
-      "eval_samples_per_second": 48.468,
-      "eval_steps_per_second": 3.03,
+      "epoch": 1.361767983162065,
+      "eval_loss": 0.41790884733200073,
+      "eval_runtime": 240.0061,
+      "eval_samples_per_second": 59.828,
+      "eval_steps_per_second": 3.742,
       "step": 5500
     },
     {
-      "epoch": 1.5599470696799806,
-      "grad_norm": 1.2716448307037354,
-      "learning_rate": 3.104535790219703e-05,
-      "loss": 0.4904,
+      "epoch": 1.3865296520985515,
+      "grad_norm": 0.9853220582008362,
+      "learning_rate": 3.6590624216595636e-05,
+      "loss": 0.4668,
       "step": 5600
     },
     {
-      "epoch": 1.5878051328481386,
-      "grad_norm": 1.2625107765197754,
-      "learning_rate": 3.069099929128278e-05,
-      "loss": 0.4943,
+      "epoch": 1.4112913210350377,
+      "grad_norm": 1.1101207733154297,
+      "learning_rate": 3.6339934820757084e-05,
+      "loss": 0.4682,
       "step": 5700
     },
     {
-      "epoch": 1.615663196016297,
-      "grad_norm": 1.2053842544555664,
-      "learning_rate": 3.0336640680368532e-05,
-      "loss": 0.5028,
+      "epoch": 1.436052989971524,
+      "grad_norm": 1.2333004474639893,
+      "learning_rate": 3.6089245424918526e-05,
+      "loss": 0.4552,
       "step": 5800
     },
     {
-      "epoch": 1.6435212591844552,
-      "grad_norm": 1.1495795249938965,
-      "learning_rate": 2.9982282069454286e-05,
-      "loss": 0.4929,
+      "epoch": 1.4608146589080104,
+      "grad_norm": 1.092253565788269,
+      "learning_rate": 3.5838556029079975e-05,
+      "loss": 0.4628,
       "step": 5900
     },
     {
-      "epoch": 1.6713793223526134,
-      "grad_norm": 1.1481202840805054,
-      "learning_rate": 2.9627923458540047e-05,
-      "loss": 0.4926,
+      "epoch": 1.4855763278444967,
+      "grad_norm": 1.0655156373977661,
+      "learning_rate": 3.558786663324142e-05,
+      "loss": 0.467,
       "step": 6000
     },
     {
-      "epoch": 1.6713793223526134,
-      "eval_loss": 0.444801926612854,
-      "eval_runtime": 592.9633,
-      "eval_samples_per_second": 48.43,
-      "eval_steps_per_second": 3.027,
+      "epoch": 1.4855763278444967,
+      "eval_loss": 0.4111788272857666,
+      "eval_runtime": 239.9577,
+      "eval_samples_per_second": 59.84,
+      "eval_steps_per_second": 3.742,
       "step": 6000
     },
     {
-      "epoch": 1.6992373855207716,
-      "grad_norm": 1.3731409311294556,
-      "learning_rate": 2.92735648476258e-05,
-      "loss": 0.5006,
+      "epoch": 1.510337996780983,
+      "grad_norm": 1.1807501316070557,
+      "learning_rate": 3.533717723740286e-05,
+      "loss": 0.4646,
       "step": 6100
     },
     {
-      "epoch": 1.7270954486889298,
-      "grad_norm": 1.0181461572647095,
-      "learning_rate": 2.8919206236711555e-05,
-      "loss": 0.4855,
+      "epoch": 1.5350996657174694,
+      "grad_norm": 0.9427251815795898,
+      "learning_rate": 3.50864878415643e-05,
+      "loss": 0.4584,
       "step": 6200
     },
     {
-      "epoch": 1.7549535118570883,
-      "grad_norm": 1.5491364002227783,
-      "learning_rate": 2.856484762579731e-05,
-      "loss": 0.4994,
+      "epoch": 1.5598613346539558,
+      "grad_norm": 1.1745707988739014,
+      "learning_rate": 3.483579844572575e-05,
+      "loss": 0.463,
       "step": 6300
     },
     {
-      "epoch": 1.7828115750252462,
-      "grad_norm": 1.149021029472351,
-      "learning_rate": 2.8210489014883064e-05,
-      "loss": 0.4918,
+      "epoch": 1.5846230035904418,
+      "grad_norm": 0.9908198714256287,
+      "learning_rate": 3.458510904988719e-05,
+      "loss": 0.4515,
       "step": 6400
     },
     {
-      "epoch": 1.8106696381934047,
-      "grad_norm": 1.0654308795928955,
-      "learning_rate": 2.7856130403968818e-05,
-      "loss": 0.487,
+      "epoch": 1.6093846725269283,
+      "grad_norm": 0.8306753635406494,
+      "learning_rate": 3.433441965404863e-05,
+      "loss": 0.4496,
       "step": 6500
     },
     {
-      "epoch": 1.8106696381934047,
-      "eval_loss": 0.4356010854244232,
-      "eval_runtime": 593.5091,
-      "eval_samples_per_second": 48.385,
-      "eval_steps_per_second": 3.024,
+      "epoch": 1.6093846725269283,
+      "eval_loss": 0.4100804626941681,
+      "eval_runtime": 239.8457,
+      "eval_samples_per_second": 59.868,
+      "eval_steps_per_second": 3.744,
       "step": 6500
     },
     {
-      "epoch": 1.8385277013615628,
-      "grad_norm": 1.2700181007385254,
-      "learning_rate": 2.750177179305457e-05,
-      "loss": 0.4869,
+      "epoch": 1.6341463414634148,
+      "grad_norm": 0.9754220247268677,
+      "learning_rate": 3.4083730258210075e-05,
+      "loss": 0.4598,
       "step": 6600
     },
     {
-      "epoch": 1.866385764529721,
-      "grad_norm": 1.0347015857696533,
-      "learning_rate": 2.714741318214033e-05,
-      "loss": 0.4861,
+      "epoch": 1.658908010399901,
+      "grad_norm": 1.0641754865646362,
+      "learning_rate": 3.3833040862371524e-05,
+      "loss": 0.4677,
       "step": 6700
     },
     {
-      "epoch": 1.8942438276978795,
-      "grad_norm": 1.08585786819458,
-      "learning_rate": 2.6793054571226084e-05,
-      "loss": 0.4884,
+      "epoch": 1.6836696793363872,
+      "grad_norm": 1.0019625425338745,
+      "learning_rate": 3.3582351466532965e-05,
+      "loss": 0.4555,
       "step": 6800
     },
     {
-      "epoch": 1.9221018908660374,
-      "grad_norm": 1.2477092742919922,
-      "learning_rate": 2.6438695960311838e-05,
-      "loss": 0.478,
+      "epoch": 1.7084313482728737,
+      "grad_norm": 0.932325541973114,
+      "learning_rate": 3.3331662070694414e-05,
+      "loss": 0.4587,
       "step": 6900
     },
     {
-      "epoch": 1.9499599540341959,
-      "grad_norm": 1.1088532209396362,
-      "learning_rate": 2.6084337349397592e-05,
-      "loss": 0.4919,
+      "epoch": 1.73319301720936,
+      "grad_norm": 0.9858091473579407,
+      "learning_rate": 3.3080972674855856e-05,
+      "loss": 0.4451,
       "step": 7000
     },
     {
-      "epoch": 1.9499599540341959,
-      "eval_loss": 0.43220457434654236,
-      "eval_runtime": 593.2416,
-      "eval_samples_per_second": 48.407,
-      "eval_steps_per_second": 3.026,
+      "epoch": 1.73319301720936,
+      "eval_loss": 0.4070571959018707,
+      "eval_runtime": 239.6611,
+      "eval_samples_per_second": 59.914,
+      "eval_steps_per_second": 3.747,
       "step": 7000
     },
     {
-      "epoch": 1.977818017202354,
-      "grad_norm": 1.0134484767913818,
-      "learning_rate": 2.5729978738483347e-05,
-      "loss": 0.4861,
+      "epoch": 1.7579546861458462,
+      "grad_norm": 1.025078535079956,
+      "learning_rate": 3.28302832790173e-05,
+      "loss": 0.4591,
       "step": 7100
     },
     {
-      "epoch": 2.0055716126336316,
-      "grad_norm": 1.185179591178894,
-      "learning_rate": 2.53756201275691e-05,
-      "loss": 0.4805,
+      "epoch": 1.7827163550823326,
+      "grad_norm": 1.0056676864624023,
+      "learning_rate": 3.2579593883178747e-05,
+      "loss": 0.4498,
       "step": 7200
     },
     {
-      "epoch": 2.03342967580179,
-      "grad_norm": 1.2651209831237793,
-      "learning_rate": 2.5021261516654855e-05,
-      "loss": 0.4785,
+      "epoch": 1.8074780240188189,
+      "grad_norm": 0.9538693428039551,
+      "learning_rate": 3.232890448734019e-05,
+      "loss": 0.4512,
       "step": 7300
     },
     {
-      "epoch": 2.061287738969948,
-      "grad_norm": 1.0275731086730957,
-      "learning_rate": 2.466690290574061e-05,
-      "loss": 0.4767,
+      "epoch": 1.832239692955305,
+      "grad_norm": 0.9744948148727417,
+      "learning_rate": 3.207821509150163e-05,
+      "loss": 0.444,
       "step": 7400
     },
     {
-      "epoch": 2.0891458021381064,
-      "grad_norm": 1.0938606262207031,
-      "learning_rate": 2.4312544294826363e-05,
-      "loss": 0.4682,
+      "epoch": 1.8570013618917915,
+      "grad_norm": 1.077823281288147,
+      "learning_rate": 3.182752569566307e-05,
+      "loss": 0.4539,
       "step": 7500
     },
     {
-      "epoch": 2.0891458021381064,
-      "eval_loss": 0.4292644262313843,
-      "eval_runtime": 592.5766,
-      "eval_samples_per_second": 48.461,
-      "eval_steps_per_second": 3.029,
+      "epoch": 1.8570013618917915,
+      "eval_loss": 0.40227407217025757,
+      "eval_runtime": 239.5404,
+      "eval_samples_per_second": 59.944,
+      "eval_steps_per_second": 3.749,
       "step": 7500
     },
     {
-      "epoch": 2.1170038653062644,
-      "grad_norm": 1.1097854375839233,
-      "learning_rate": 2.395818568391212e-05,
-      "loss": 0.4703,
+      "epoch": 1.8817630308282778,
+      "grad_norm": 0.9259219169616699,
+      "learning_rate": 3.157683629982452e-05,
+      "loss": 0.4545,
       "step": 7600
     },
     {
-      "epoch": 2.144861928474423,
-      "grad_norm": 1.3471382856369019,
-      "learning_rate": 2.3603827072997875e-05,
-      "loss": 0.4843,
+      "epoch": 1.906524699764764,
+      "grad_norm": 1.0984644889831543,
+      "learning_rate": 3.132614690398596e-05,
+      "loss": 0.4426,
       "step": 7700
     },
     {
-      "epoch": 2.1727199916425812,
-      "grad_norm": 1.351805329322815,
-      "learning_rate": 2.324946846208363e-05,
-      "loss": 0.4776,
+      "epoch": 1.9312863687012505,
+      "grad_norm": 0.9966897964477539,
+      "learning_rate": 3.1075457508147405e-05,
+      "loss": 0.4362,
       "step": 7800
     },
     {
-      "epoch": 2.200578054810739,
-      "grad_norm": 1.1416364908218384,
-      "learning_rate": 2.2895109851169387e-05,
-      "loss": 0.4714,
+      "epoch": 1.956048037637737,
+      "grad_norm": 1.1213327646255493,
+      "learning_rate": 3.0824768112308846e-05,
+      "loss": 0.4465,
       "step": 7900
     },
     {
-      "epoch": 2.2284361179788976,
-      "grad_norm": 1.1691741943359375,
-      "learning_rate": 2.254075124025514e-05,
-      "loss": 0.4734,
+      "epoch": 1.980809706574223,
+      "grad_norm": 0.9818247556686401,
+      "learning_rate": 3.0574078716470295e-05,
+      "loss": 0.4342,
       "step": 8000
     },
     {
-      "epoch": 2.2284361179788976,
-      "eval_loss": 0.4240116477012634,
-      "eval_runtime": 594.3246,
-      "eval_samples_per_second": 48.319,
-      "eval_steps_per_second": 3.02,
+      "epoch": 1.980809706574223,
+      "eval_loss": 0.4005620777606964,
+      "eval_runtime": 240.1057,
+      "eval_samples_per_second": 59.803,
+      "eval_steps_per_second": 3.74,
       "step": 8000
     },
     {
-      "epoch": 2.2562941811470556,
-      "grad_norm": 1.0872358083724976,
-      "learning_rate": 2.2186392629340892e-05,
-      "loss": 0.4711,
+      "epoch": 2.005447567166027,
+      "grad_norm": 1.226771593093872,
+      "learning_rate": 3.032338932063174e-05,
+      "loss": 0.439,
       "step": 8100
     },
     {
-      "epoch": 2.284152244315214,
-      "grad_norm": 1.166755199432373,
-      "learning_rate": 2.1832034018426646e-05,
-      "loss": 0.4674,
+      "epoch": 2.0302092361025132,
+      "grad_norm": 1.0219694375991821,
+      "learning_rate": 3.0072699924793186e-05,
+      "loss": 0.4369,
       "step": 8200
     },
     {
-      "epoch": 2.312010307483372,
-      "grad_norm": 1.397355556488037,
-      "learning_rate": 2.1477675407512404e-05,
-      "loss": 0.4741,
+      "epoch": 2.0549709050389997,
+      "grad_norm": 1.0891128778457642,
+      "learning_rate": 2.9822010528954624e-05,
+      "loss": 0.44,
       "step": 8300
     },
     {
-      "epoch": 2.3398683706515304,
-      "grad_norm": 1.0663005113601685,
-      "learning_rate": 2.1123316796598158e-05,
-      "loss": 0.4741,
+      "epoch": 2.0797325739754857,
+      "grad_norm": 0.9588236212730408,
+      "learning_rate": 2.957132113311607e-05,
+      "loss": 0.4354,
       "step": 8400
     },
     {
-      "epoch": 2.367726433819689,
-      "grad_norm": 1.2197699546813965,
-      "learning_rate": 2.0768958185683912e-05,
-      "loss": 0.4691,
+      "epoch": 2.104494242911972,
+      "grad_norm": 1.0548973083496094,
+      "learning_rate": 2.9320631737277515e-05,
+      "loss": 0.4424,
       "step": 8500
     },
     {
-      "epoch": 2.367726433819689,
-      "eval_loss": 0.4226415753364563,
-      "eval_runtime": 592.9032,
-      "eval_samples_per_second": 48.435,
-      "eval_steps_per_second": 3.027,
+      "epoch": 2.104494242911972,
+      "eval_loss": 0.3962107300758362,
+      "eval_runtime": 240.0503,
+      "eval_samples_per_second": 59.817,
+      "eval_steps_per_second": 3.741,
       "step": 8500
     },
     {
-      "epoch": 2.395584496987847,
-      "grad_norm": 1.4034844636917114,
-      "learning_rate": 2.041459957476967e-05,
-      "loss": 0.4749,
+      "epoch": 2.1292559118484586,
+      "grad_norm": 0.7916435599327087,
+      "learning_rate": 2.906994234143896e-05,
+      "loss": 0.4375,
       "step": 8600
     },
     {
-      "epoch": 2.4234425601560052,
-      "grad_norm": 1.2953102588653564,
-      "learning_rate": 2.0060240963855424e-05,
-      "loss": 0.4757,
+      "epoch": 2.154017580784945,
+      "grad_norm": 1.0228310823440552,
+      "learning_rate": 2.8819252945600402e-05,
+      "loss": 0.4376,
       "step": 8700
     },
     {
-      "epoch": 2.451300623324163,
-      "grad_norm": 1.3071800470352173,
-      "learning_rate": 1.9705882352941178e-05,
-      "loss": 0.4668,
+      "epoch": 2.178779249721431,
+      "grad_norm": 1.0908734798431396,
+      "learning_rate": 2.8568563549761847e-05,
+      "loss": 0.4276,
       "step": 8800
     },
     {
-      "epoch": 2.4791586864923216,
-      "grad_norm": 1.1789804697036743,
-      "learning_rate": 1.9351523742026932e-05,
-      "loss": 0.4729,
+      "epoch": 2.2035409186579176,
+      "grad_norm": 1.1127111911773682,
+      "learning_rate": 2.831787415392329e-05,
+      "loss": 0.43,
       "step": 8900
     },
     {
-      "epoch": 2.5070167496604796,
-      "grad_norm": 1.2676748037338257,
-      "learning_rate": 1.8997165131112687e-05,
-      "loss": 0.4679,
+      "epoch": 2.228302587594404,
+      "grad_norm": 1.2166775465011597,
+      "learning_rate": 2.8067184758084734e-05,
+      "loss": 0.4366,
       "step": 9000
     },
     {
-      "epoch": 2.5070167496604796,
-      "eval_loss": 0.4187828004360199,
-      "eval_runtime": 592.1014,
-      "eval_samples_per_second": 48.5,
-      "eval_steps_per_second": 3.032,
+      "epoch": 2.228302587594404,
+      "eval_loss": 0.39486321806907654,
+      "eval_runtime": 240.2249,
+      "eval_samples_per_second": 59.773,
+      "eval_steps_per_second": 3.738,
       "step": 9000
     },
     {
-      "epoch": 2.534874812828638,
-      "grad_norm": 1.3900854587554932,
-      "learning_rate": 1.864280652019844e-05,
-      "loss": 0.4677,
+      "epoch": 2.25306425653089,
+      "grad_norm": 0.9300484657287598,
+      "learning_rate": 2.781649536224618e-05,
+      "loss": 0.4414,
       "step": 9100
     },
     {
-      "epoch": 2.5627328759967964,
-      "grad_norm": 1.2864222526550293,
-      "learning_rate": 1.8288447909284195e-05,
-      "loss": 0.4584,
+      "epoch": 2.2778259254673765,
+      "grad_norm": 1.0898849964141846,
+      "learning_rate": 2.756580596640762e-05,
+      "loss": 0.4378,
       "step": 9200
     },
     {
-      "epoch": 2.5905909391649544,
-      "grad_norm": 1.0441097021102905,
-      "learning_rate": 1.7934089298369953e-05,
-      "loss": 0.4654,
+      "epoch": 2.302587594403863,
+      "grad_norm": 0.9797316789627075,
+      "learning_rate": 2.7315116570569067e-05,
+      "loss": 0.4207,
       "step": 9300
     },
     {
-      "epoch": 2.618449002333113,
-      "grad_norm": 1.1249713897705078,
-      "learning_rate": 1.7579730687455707e-05,
-      "loss": 0.4674,
+      "epoch": 2.327349263340349,
+      "grad_norm": 1.1121026277542114,
+      "learning_rate": 2.7064427174730512e-05,
+      "loss": 0.4301,
       "step": 9400
     },
     {
-      "epoch": 2.6463070655012713,
-      "grad_norm": 1.2885221242904663,
-      "learning_rate": 1.722537207654146e-05,
-      "loss": 0.4694,
+      "epoch": 2.3521109322768354,
+      "grad_norm": 1.113186001777649,
+      "learning_rate": 2.6813737778891957e-05,
+      "loss": 0.4277,
       "step": 9500
     },
     {
-      "epoch": 2.6463070655012713,
-      "eval_loss": 0.4162205457687378,
-      "eval_runtime": 593.1028,
-      "eval_samples_per_second": 48.418,
-      "eval_steps_per_second": 3.026,
+      "epoch": 2.3521109322768354,
+      "eval_loss": 0.3930341899394989,
+      "eval_runtime": 240.1255,
+      "eval_samples_per_second": 59.798,
+      "eval_steps_per_second": 3.74,
       "step": 9500
     },
     {
-      "epoch": 2.6741651286694292,
-      "grad_norm": 1.1028586626052856,
-      "learning_rate": 1.6871013465627215e-05,
-      "loss": 0.4573,
+      "epoch": 2.376872601213322,
+      "grad_norm": 1.1765657663345337,
+      "learning_rate": 2.6563048383053396e-05,
+      "loss": 0.4321,
       "step": 9600
     },
     {
-      "epoch": 2.7020231918375877,
-      "grad_norm": 1.3248066902160645,
-      "learning_rate": 1.651665485471297e-05,
-      "loss": 0.4609,
+      "epoch": 2.401634270149808,
+      "grad_norm": 1.239833116531372,
+      "learning_rate": 2.631235898721484e-05,
+      "loss": 0.4282,
       "step": 9700
     },
     {
-      "epoch": 2.7298812550057456,
-      "grad_norm": 1.073268175125122,
-      "learning_rate": 1.6162296243798724e-05,
-      "loss": 0.4543,
+      "epoch": 2.4263959390862944,
+      "grad_norm": 0.9866438508033752,
+      "learning_rate": 2.6061669591376286e-05,
+      "loss": 0.4338,
       "step": 9800
     },
     {
-      "epoch": 2.757739318173904,
-      "grad_norm": 1.0753856897354126,
-      "learning_rate": 1.5807937632884478e-05,
-      "loss": 0.4667,
+      "epoch": 2.451157608022781,
+      "grad_norm": 0.9813755750656128,
+      "learning_rate": 2.581098019553773e-05,
+      "loss": 0.4211,
       "step": 9900
     },
     {
-      "epoch": 2.785597381342062,
-      "grad_norm": 1.2315281629562378,
-      "learning_rate": 1.5453579021970235e-05,
-      "loss": 0.4573,
+      "epoch": 2.4759192769592673,
+      "grad_norm": 1.1001412868499756,
+      "learning_rate": 2.556029079969917e-05,
+      "loss": 0.4386,
       "step": 10000
     },
     {
-      "epoch": 2.785597381342062,
-      "eval_loss": 0.41505932807922363,
-      "eval_runtime": 592.7426,
-      "eval_samples_per_second": 48.448,
-      "eval_steps_per_second": 3.028,
+      "epoch": 2.4759192769592673,
+      "eval_loss": 0.38949617743492126,
+      "eval_runtime": 241.0988,
+      "eval_samples_per_second": 59.557,
+      "eval_steps_per_second": 3.725,
       "step": 10000
     },
     {
-      "epoch": 2.8134554445102204,
-      "grad_norm": 1.1854623556137085,
-      "learning_rate": 1.509922041105599e-05,
-      "loss": 0.4579,
+      "epoch": 2.5006809458957533,
+      "grad_norm": 1.2975023984909058,
+      "learning_rate": 2.5309601403860615e-05,
+      "loss": 0.4324,
       "step": 10100
     },
     {
-      "epoch": 2.841313507678379,
-      "grad_norm": 1.1797943115234375,
-      "learning_rate": 1.4744861800141744e-05,
-      "loss": 0.4512,
+      "epoch": 2.5254426148322398,
+      "grad_norm": 1.2328298091888428,
+      "learning_rate": 2.505891200802206e-05,
+      "loss": 0.4269,
       "step": 10200
     },
     {
-      "epoch": 2.869171570846537,
-      "grad_norm": 1.2852693796157837,
-      "learning_rate": 1.43905031892275e-05,
-      "loss": 0.4657,
+      "epoch": 2.5502042837687258,
+      "grad_norm": 1.065771222114563,
+      "learning_rate": 2.4808222612183506e-05,
+      "loss": 0.4283,
       "step": 10300
     },
     {
-      "epoch": 2.8970296340146953,
-      "grad_norm": 1.4903700351715088,
-      "learning_rate": 1.4036144578313254e-05,
-      "loss": 0.4565,
+      "epoch": 2.5749659527052122,
+      "grad_norm": 1.0811374187469482,
+      "learning_rate": 2.455753321634495e-05,
+      "loss": 0.4229,
       "step": 10400
     },
     {
-      "epoch": 2.9248876971828532,
-      "grad_norm": 1.1999119520187378,
-      "learning_rate": 1.3681785967399008e-05,
-      "loss": 0.4505,
+      "epoch": 2.5997276216416987,
+      "grad_norm": 1.0005619525909424,
+      "learning_rate": 2.4306843820506396e-05,
+      "loss": 0.4219,
       "step": 10500
     },
     {
-      "epoch": 2.9248876971828532,
-      "eval_loss": 0.41494467854499817,
-      "eval_runtime": 594.0624,
-      "eval_samples_per_second": 48.34,
-      "eval_steps_per_second": 3.022,
+      "epoch": 2.5997276216416987,
+      "eval_loss": 0.38891834020614624,
+      "eval_runtime": 240.2464,
+      "eval_samples_per_second": 59.768,
+      "eval_steps_per_second": 3.738,
       "step": 10500
     },
     {
-      "epoch": 2.9527457603510117,
-      "grad_norm": 1.1072437763214111,
-      "learning_rate": 1.3327427356484764e-05,
-      "loss": 0.4636,
+      "epoch": 2.624489290578185,
+      "grad_norm": 1.1370168924331665,
+      "learning_rate": 2.4056154424667838e-05,
+      "loss": 0.4292,
       "step": 10600
     },
     {
-      "epoch": 2.9806038235191696,
-      "grad_norm": 1.0771669149398804,
-      "learning_rate": 1.2973068745570518e-05,
-      "loss": 0.4666,
+      "epoch": 2.649250959514671,
+      "grad_norm": 1.0099055767059326,
+      "learning_rate": 2.3805465028829283e-05,
+      "loss": 0.4253,
       "step": 10700
     },
     {
-      "epoch": 3.0083574189504474,
-      "grad_norm": 1.1003024578094482,
-      "learning_rate": 1.2618710134656272e-05,
-      "loss": 0.4456,
+      "epoch": 2.6740126284511576,
+      "grad_norm": 0.7887821197509766,
+      "learning_rate": 2.3554775632990725e-05,
+      "loss": 0.4321,
       "step": 10800
     },
     {
-      "epoch": 3.036215482118606,
-      "grad_norm": 1.417353630065918,
-      "learning_rate": 1.2264351523742028e-05,
-      "loss": 0.4594,
+      "epoch": 2.698774297387644,
+      "grad_norm": 1.0963307619094849,
+      "learning_rate": 2.330408623715217e-05,
+      "loss": 0.4287,
       "step": 10900
     },
     {
-      "epoch": 3.064073545286764,
-      "grad_norm": 1.1961385011672974,
-      "learning_rate": 1.1909992912827782e-05,
-      "loss": 0.4478,
+      "epoch": 2.72353596632413,
+      "grad_norm": 0.9316207766532898,
+      "learning_rate": 2.3053396841313612e-05,
+      "loss": 0.4165,
       "step": 11000
     },
     {
-      "epoch": 3.064073545286764,
-      "eval_loss": 0.4131682813167572,
-      "eval_runtime": 592.6765,
-      "eval_samples_per_second": 48.453,
-      "eval_steps_per_second": 3.029,
+      "epoch": 2.72353596632413,
+      "eval_loss": 0.3874943256378174,
+      "eval_runtime": 240.3044,
+      "eval_samples_per_second": 59.753,
+      "eval_steps_per_second": 3.737,
       "step": 11000
     },
     {
-      "epoch": 3.091931608454922,
-      "grad_norm": 1.3566352128982544,
-      "learning_rate": 1.1555634301913537e-05,
-      "loss": 0.4416,
+      "epoch": 2.7482976352606165,
+      "grad_norm": 1.1644989252090454,
+      "learning_rate": 2.2802707445475058e-05,
+      "loss": 0.4253,
       "step": 11100
     },
     {
-      "epoch": 3.1197896716230806,
-      "grad_norm": 1.3073431253433228,
-      "learning_rate": 1.1201275690999293e-05,
-      "loss": 0.4509,
+      "epoch": 2.773059304197103,
+      "grad_norm": 1.0232603549957275,
+      "learning_rate": 2.25520180496365e-05,
+      "loss": 0.4305,
       "step": 11200
     },
     {
-      "epoch": 3.1476477347912386,
-      "grad_norm": 1.3190025091171265,
-      "learning_rate": 1.0846917080085047e-05,
-      "loss": 0.4513,
+      "epoch": 2.7978209731335895,
+      "grad_norm": 1.025208830833435,
+      "learning_rate": 2.2301328653797945e-05,
+      "loss": 0.4274,
       "step": 11300
     },
     {
-      "epoch": 3.175505797959397,
-      "grad_norm": 1.3781085014343262,
-      "learning_rate": 1.0492558469170801e-05,
-      "loss": 0.4437,
+      "epoch": 2.8225826420700755,
+      "grad_norm": 1.0372883081436157,
+      "learning_rate": 2.205063925795939e-05,
+      "loss": 0.4169,
       "step": 11400
     },
     {
-      "epoch": 3.203363861127555,
-      "grad_norm": 1.306618332862854,
-      "learning_rate": 1.0138199858256555e-05,
-      "loss": 0.4454,
+      "epoch": 2.847344311006562,
+      "grad_norm": 1.093565821647644,
+      "learning_rate": 2.1799949862120832e-05,
+      "loss": 0.4236,
       "step": 11500
     },
     {
-      "epoch": 3.203363861127555,
-      "eval_loss": 0.4108191132545471,
-      "eval_runtime": 593.9192,
-      "eval_samples_per_second": 48.352,
-      "eval_steps_per_second": 3.022,
+      "epoch": 2.847344311006562,
+      "eval_loss": 0.3843945562839508,
+      "eval_runtime": 239.7883,
+      "eval_samples_per_second": 59.882,
+      "eval_steps_per_second": 3.745,
       "step": 11500
     },
     {
-      "epoch": 3.2312219242957134,
-      "grad_norm": 1.4700162410736084,
-      "learning_rate": 9.783841247342311e-06,
-      "loss": 0.4471,
+      "epoch": 2.872105979943048,
+      "grad_norm": 1.328104853630066,
+      "learning_rate": 2.1549260466282277e-05,
+      "loss": 0.4233,
       "step": 11600
     },
     {
-      "epoch": 3.2590799874638714,
-      "grad_norm": 1.1304278373718262,
-      "learning_rate": 9.429482636428065e-06,
-      "loss": 0.4536,
+      "epoch": 2.8968676488795344,
+      "grad_norm": 1.0304131507873535,
+      "learning_rate": 2.1298571070443722e-05,
+      "loss": 0.4252,
       "step": 11700
     },
     {
-      "epoch": 3.28693805063203,
-      "grad_norm": 1.4289369583129883,
-      "learning_rate": 9.075124025513821e-06,
-      "loss": 0.4474,
+      "epoch": 2.921629317816021,
+      "grad_norm": 0.9143759608268738,
+      "learning_rate": 2.1047881674605168e-05,
+      "loss": 0.4198,
       "step": 11800
     },
     {
-      "epoch": 3.3147961138001882,
-      "grad_norm": 1.3017057180404663,
-      "learning_rate": 8.720765414599575e-06,
-      "loss": 0.4449,
+      "epoch": 2.9463909867525073,
+      "grad_norm": 1.0201221704483032,
+      "learning_rate": 2.079719227876661e-05,
+      "loss": 0.4205,
       "step": 11900
     },
     {
-      "epoch": 3.342654176968346,
-      "grad_norm": 1.4927443265914917,
-      "learning_rate": 8.36640680368533e-06,
-      "loss": 0.4557,
+      "epoch": 2.9711526556889933,
+      "grad_norm": 0.9955803751945496,
+      "learning_rate": 2.0546502882928055e-05,
+      "loss": 0.4199,
       "step": 12000
     },
     {
-      "epoch": 3.342654176968346,
-      "eval_loss": 0.4085528552532196,
-      "eval_runtime": 592.2863,
-      "eval_samples_per_second": 48.485,
-      "eval_steps_per_second": 3.031,
+      "epoch": 2.9711526556889933,
+      "eval_loss": 0.38319218158721924,
+      "eval_runtime": 240.3339,
+      "eval_samples_per_second": 59.746,
+      "eval_steps_per_second": 3.736,
       "step": 12000
     },
     {
-      "epoch": 3.3705122401365046,
-      "grad_norm": 1.4530751705169678,
-      "learning_rate": 8.012048192771085e-06,
-      "loss": 0.455,
+      "epoch": 2.99591432462548,
+      "grad_norm": 1.1340243816375732,
+      "learning_rate": 2.0295813487089497e-05,
+      "loss": 0.4178,
       "step": 12100
     },
     {
-      "epoch": 3.3983703033046626,
-      "grad_norm": 1.4134407043457031,
-      "learning_rate": 7.65768958185684e-06,
-      "loss": 0.4504,
+      "epoch": 3.0205521852172836,
+      "grad_norm": 1.0646172761917114,
+      "learning_rate": 2.0045124091250942e-05,
+      "loss": 0.4169,
       "step": 12200
     },
     {
-      "epoch": 3.426228366472821,
-      "grad_norm": 1.2031890153884888,
-      "learning_rate": 7.303330970942594e-06,
-      "loss": 0.4473,
+      "epoch": 3.04531385415377,
+      "grad_norm": 0.8297644853591919,
+      "learning_rate": 1.9794434695412384e-05,
+      "loss": 0.4114,
       "step": 12300
     },
     {
-      "epoch": 3.454086429640979,
-      "grad_norm": 1.3506805896759033,
-      "learning_rate": 6.94897236002835e-06,
-      "loss": 0.4452,
+      "epoch": 3.070075523090256,
+      "grad_norm": 0.8982422947883606,
+      "learning_rate": 1.954374529957383e-05,
+      "loss": 0.4074,
       "step": 12400
     },
     {
-      "epoch": 3.4819444928091374,
-      "grad_norm": 1.6460970640182495,
-      "learning_rate": 6.594613749114104e-06,
-      "loss": 0.4475,
+      "epoch": 3.0948371920267426,
+      "grad_norm": 1.0254546403884888,
+      "learning_rate": 1.9293055903735274e-05,
+      "loss": 0.4143,
       "step": 12500
     },
     {
-      "epoch": 3.4819444928091374,
-      "eval_loss": 0.40733858942985535,
-      "eval_runtime": 593.8096,
-      "eval_samples_per_second": 48.361,
-      "eval_steps_per_second": 3.023,
+      "epoch": 3.0948371920267426,
+      "eval_loss": 0.3798697888851166,
+      "eval_runtime": 240.4712,
+      "eval_samples_per_second": 59.712,
+      "eval_steps_per_second": 3.734,
       "step": 12500
     },
     {
-      "epoch": 3.509802555977296,
-      "grad_norm": 1.34632408618927,
-      "learning_rate": 6.240255138199858e-06,
-      "loss": 0.4472,
+      "epoch": 3.119598860963229,
+      "grad_norm": 0.9920561909675598,
+      "learning_rate": 1.9042366507896716e-05,
+      "loss": 0.4082,
       "step": 12600
     },
     {
-      "epoch": 3.537660619145454,
-      "grad_norm": 1.1388472318649292,
-      "learning_rate": 5.885896527285613e-06,
-      "loss": 0.4504,
+      "epoch": 3.144360529899715,
+      "grad_norm": 1.0367820262908936,
+      "learning_rate": 1.879167711205816e-05,
+      "loss": 0.4176,
       "step": 12700
     },
     {
-      "epoch": 3.5655186823136122,
-      "grad_norm": 1.3838173151016235,
-      "learning_rate": 5.531537916371368e-06,
-      "loss": 0.4509,
+      "epoch": 3.1691221988362015,
+      "grad_norm": 1.2170422077178955,
+      "learning_rate": 1.8540987716219603e-05,
+      "loss": 0.4149,
       "step": 12800
     },
     {
-      "epoch": 3.59337674548177,
-      "grad_norm": 1.2048169374465942,
-      "learning_rate": 5.1771793054571225e-06,
-      "loss": 0.4484,
+      "epoch": 3.193883867772688,
+      "grad_norm": 1.0519484281539917,
+      "learning_rate": 1.829029832038105e-05,
+      "loss": 0.4169,
       "step": 12900
     },
     {
-      "epoch": 3.6212348086499286,
-      "grad_norm": 1.337014079093933,
-      "learning_rate": 4.8228206945428775e-06,
-      "loss": 0.4514,
+      "epoch": 3.2186455367091744,
+      "grad_norm": 1.0347421169281006,
+      "learning_rate": 1.803960892454249e-05,
+      "loss": 0.4086,
       "step": 13000
     },
     {
-      "epoch": 3.6212348086499286,
-      "eval_loss": 0.40576237440109253,
-      "eval_runtime": 592.3205,
-      "eval_samples_per_second": 48.482,
-      "eval_steps_per_second": 3.03,
+      "epoch": 3.2186455367091744,
+      "eval_loss": 0.3794644773006439,
+      "eval_runtime": 240.0067,
+      "eval_samples_per_second": 59.827,
+      "eval_steps_per_second": 3.742,
       "step": 13000
     },
     {
-      "epoch": 3.6490928718180866,
-      "grad_norm": 1.2974600791931152,
-      "learning_rate": 4.4684620836286326e-06,
-      "loss": 0.4449,
+      "epoch": 3.2434072056456604,
+      "grad_norm": 1.1561747789382935,
+      "learning_rate": 1.7788919528703936e-05,
+      "loss": 0.4029,
       "step": 13100
     },
     {
-      "epoch": 3.676950934986245,
-      "grad_norm": 1.1322963237762451,
-      "learning_rate": 4.114103472714388e-06,
-      "loss": 0.4514,
+      "epoch": 3.268168874582147,
+      "grad_norm": 0.8857806324958801,
+      "learning_rate": 1.753823013286538e-05,
+      "loss": 0.4125,
       "step": 13200
     },
     {
-      "epoch": 3.7048089981544035,
-      "grad_norm": 1.2485069036483765,
-      "learning_rate": 3.7597448618001422e-06,
-      "loss": 0.441,
+      "epoch": 3.2929305435186333,
+      "grad_norm": 0.8242279887199402,
+      "learning_rate": 1.7287540737026826e-05,
+      "loss": 0.421,
       "step": 13300
     },
     {
-      "epoch": 3.7326670613225614,
-      "grad_norm": 1.3043463230133057,
-      "learning_rate": 3.405386250885897e-06,
-      "loss": 0.4429,
+      "epoch": 3.3176922124551194,
+      "grad_norm": 1.0788533687591553,
+      "learning_rate": 1.7036851341188268e-05,
+      "loss": 0.4186,
       "step": 13400
     },
     {
-      "epoch": 3.76052512449072,
-      "grad_norm": 1.22727632522583,
-      "learning_rate": 3.0510276399716515e-06,
-      "loss": 0.4461,
+      "epoch": 3.342453881391606,
+      "grad_norm": 0.9279969930648804,
+      "learning_rate": 1.6786161945349713e-05,
+      "loss": 0.4111,
       "step": 13500
     },
     {
-      "epoch": 3.76052512449072,
-      "eval_loss": 0.40465712547302246,
-      "eval_runtime": 593.0804,
-      "eval_samples_per_second": 48.42,
-      "eval_steps_per_second": 3.027,
+      "epoch": 3.342453881391606,
+      "eval_loss": 0.37995103001594543,
+      "eval_runtime": 240.9168,
+      "eval_samples_per_second": 59.601,
+      "eval_steps_per_second": 3.727,
       "step": 13500
     },
     {
-      "epoch": 3.7883831876588783,
-      "grad_norm": 1.322852611541748,
-      "learning_rate": 2.696669029057406e-06,
-      "loss": 0.4495,
+      "epoch": 3.3672155503280923,
+      "grad_norm": 0.9397678375244141,
+      "learning_rate": 1.6535472549511155e-05,
+      "loss": 0.4118,
       "step": 13600
     },
     {
-      "epoch": 3.8162412508270362,
-      "grad_norm": 1.18853759765625,
-      "learning_rate": 2.342310418143161e-06,
-      "loss": 0.4363,
+      "epoch": 3.3919772192645783,
+      "grad_norm": 1.1751211881637573,
+      "learning_rate": 1.62847831536726e-05,
+      "loss": 0.4124,
       "step": 13700
     },
     {
-      "epoch": 3.8440993139951942,
-      "grad_norm": 1.22146475315094,
-      "learning_rate": 1.987951807228916e-06,
-      "loss": 0.4393,
+      "epoch": 3.4167388882010647,
+      "grad_norm": 0.8791301250457764,
+      "learning_rate": 1.6034093757834046e-05,
+      "loss": 0.4038,
       "step": 13800
     },
     {
-      "epoch": 3.8719573771633526,
-      "grad_norm": 1.3995431661605835,
-      "learning_rate": 1.6335931963146704e-06,
-      "loss": 0.4475,
+      "epoch": 3.441500557137551,
+      "grad_norm": 1.1133136749267578,
+      "learning_rate": 1.5783404361995488e-05,
+      "loss": 0.4164,
       "step": 13900
     },
     {
-      "epoch": 3.899815440331511,
-      "grad_norm": 1.402491807937622,
-      "learning_rate": 1.2792345854004253e-06,
-      "loss": 0.445,
+      "epoch": 3.4662622260740372,
+      "grad_norm": 1.121056079864502,
+      "learning_rate": 1.5532714966156933e-05,
+      "loss": 0.41,
       "step": 14000
     },
     {
-      "epoch": 3.899815440331511,
-      "eval_loss": 0.4038073718547821,
-      "eval_runtime": 594.2622,
-      "eval_samples_per_second": 48.324,
-      "eval_steps_per_second": 3.021,
+      "epoch": 3.4662622260740372,
+      "eval_loss": 0.3772788941860199,
+      "eval_runtime": 239.4727,
+      "eval_samples_per_second": 59.961,
+      "eval_steps_per_second": 3.75,
       "step": 14000
+    },
+    {
+      "epoch": 3.4910238950105237,
+      "grad_norm": 0.973131537437439,
+      "learning_rate": 1.5282025570318375e-05,
+      "loss": 0.412,
+      "step": 14100
+    },
+    {
+      "epoch": 3.51578556394701,
+      "grad_norm": 1.1479246616363525,
+      "learning_rate": 1.503133617447982e-05,
+      "loss": 0.4102,
+      "step": 14200
+    },
+    {
+      "epoch": 3.5405472328834966,
+      "grad_norm": 0.9543820023536682,
+      "learning_rate": 1.4780646778641264e-05,
+      "loss": 0.4146,
+      "step": 14300
+    },
+    {
+      "epoch": 3.5653089018199826,
+      "grad_norm": 1.2635592222213745,
+      "learning_rate": 1.4529957382802709e-05,
+      "loss": 0.4077,
+      "step": 14400
+    },
+    {
+      "epoch": 3.590070570756469,
+      "grad_norm": 1.339470386505127,
+      "learning_rate": 1.4279267986964151e-05,
+      "loss": 0.4099,
+      "step": 14500
+    },
+    {
+      "epoch": 3.590070570756469,
+      "eval_loss": 0.37771478295326233,
+      "eval_runtime": 239.2354,
+      "eval_samples_per_second": 60.02,
+      "eval_steps_per_second": 3.754,
+      "step": 14500
+    },
+    {
+      "epoch": 3.614832239692955,
+      "grad_norm": 0.9866228699684143,
+      "learning_rate": 1.4028578591125596e-05,
+      "loss": 0.4067,
+      "step": 14600
+    },
+    {
+      "epoch": 3.6395939086294415,
+      "grad_norm": 1.1348681449890137,
+      "learning_rate": 1.3777889195287038e-05,
+      "loss": 0.405,
+      "step": 14700
+    },
+    {
+      "epoch": 3.664355577565928,
+      "grad_norm": 0.934289813041687,
+      "learning_rate": 1.3527199799448483e-05,
+      "loss": 0.4057,
+      "step": 14800
+    },
+    {
+      "epoch": 3.6891172465024145,
+      "grad_norm": 0.8867693543434143,
+      "learning_rate": 1.3276510403609929e-05,
+      "loss": 0.4145,
+      "step": 14900
+    },
+    {
+      "epoch": 3.7138789154389005,
+      "grad_norm": 1.0333330631256104,
+      "learning_rate": 1.3025821007771372e-05,
+      "loss": 0.4131,
+      "step": 15000
+    },
+    {
+      "epoch": 3.7138789154389005,
+      "eval_loss": 0.3742632269859314,
+      "eval_runtime": 241.2181,
+      "eval_samples_per_second": 59.527,
+      "eval_steps_per_second": 3.723,
+      "step": 15000
+    },
+    {
+      "epoch": 3.738640584375387,
+      "grad_norm": 1.0293641090393066,
+      "learning_rate": 1.2775131611932817e-05,
+      "loss": 0.4101,
+      "step": 15100
+    },
+    {
+      "epoch": 3.763402253311873,
+      "grad_norm": 1.0152819156646729,
+      "learning_rate": 1.252444221609426e-05,
+      "loss": 0.4116,
+      "step": 15200
+    },
+    {
+      "epoch": 3.7881639222483594,
+      "grad_norm": 1.165980577468872,
+      "learning_rate": 1.2273752820255703e-05,
+      "loss": 0.4071,
+      "step": 15300
+    },
+    {
+      "epoch": 3.812925591184846,
+      "grad_norm": 1.1100980043411255,
+      "learning_rate": 1.2023063424417148e-05,
+      "loss": 0.4178,
+      "step": 15400
+    },
+    {
+      "epoch": 3.8376872601213323,
+      "grad_norm": 1.141007900238037,
+      "learning_rate": 1.1772374028578592e-05,
+      "loss": 0.4074,
+      "step": 15500
+    },
+    {
+      "epoch": 3.8376872601213323,
+      "eval_loss": 0.3737417459487915,
+      "eval_runtime": 240.5046,
+      "eval_samples_per_second": 59.704,
+      "eval_steps_per_second": 3.734,
+      "step": 15500
+    },
+    {
+      "epoch": 3.8624489290578183,
+      "grad_norm": 0.9891005754470825,
+      "learning_rate": 1.1521684632740035e-05,
+      "loss": 0.398,
+      "step": 15600
+    },
+    {
+      "epoch": 3.887210597994305,
+      "grad_norm": 1.1525514125823975,
+      "learning_rate": 1.127099523690148e-05,
+      "loss": 0.4035,
+      "step": 15700
+    },
+    {
+      "epoch": 3.9119722669307913,
+      "grad_norm": 1.0974273681640625,
+      "learning_rate": 1.1020305841062924e-05,
+      "loss": 0.406,
+      "step": 15800
+    },
+    {
+      "epoch": 3.9367339358672773,
+      "grad_norm": 1.150168776512146,
+      "learning_rate": 1.0769616445224368e-05,
+      "loss": 0.4058,
+      "step": 15900
+    },
+    {
+      "epoch": 3.9614956048037637,
+      "grad_norm": 0.9373348951339722,
+      "learning_rate": 1.0518927049385811e-05,
+      "loss": 0.4047,
+      "step": 16000
+    },
+    {
+      "epoch": 3.9614956048037637,
+      "eval_loss": 0.3724241852760315,
+      "eval_runtime": 240.6291,
+      "eval_samples_per_second": 59.673,
+      "eval_steps_per_second": 3.732,
+      "step": 16000
     }
   ],
   "logging_steps": 100,
-  "max_steps": 14360,
+  "max_steps": 20195,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 4,
+  "num_train_epochs": 5,
   "save_steps": 500,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -1238,8 +1410,8 @@
       "attributes": {}
     }
   },
-  "total_flos": 4.932896695826283e+21,
-  "train_batch_size": 4,
+  "total_flos": 5.63732250154283e+21,
+  "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3ebac2dfd955401b10729cffe3d6f2cf3db168592c54addf497fd98fd6c104b7
+oid sha256:56e616dee530411c17bd34a10ca4deabc5f2ea69a17e48e65f54379e09072e10
 size 5905