Upload folder using huggingface_hub

adapter_config.json

@@ -25,13 +25,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "o_proj",
-    "gate_proj",
-    "down_proj",
-    "q_proj",
     "v_proj",
     "k_proj",
-    "up_proj"
+    "up_proj",
+    "down_proj",
+    "o_proj",
+    "q_proj",
+    "gate_proj"
   ],
   "target_parameters": null,
   "task_type": "CAUSAL_LM",

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:22dc28c276a33a2fac1a0b854e47ef811350dcff438b52546f72b98758964aba
+oid sha256:b63a758a9b540c37ae331e4841cb911035c86022b292e0cdf572ab6479dc5a54
 size 323014168

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6bcd3e8a6755aadede004d6c2ca0671c37f040f18df035d7f55b83e92d3fd2fa
+oid sha256:c5a487189652f7d7b30832e815c941a95ce2e54fece9c54e14c2ea38393e2e73
 size 646164683

rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1ca5d1f8ee4908b1d9401855c7e1e111027cd0fdbc032882bcd05545f08deeed
+oid sha256:48f66984fb58b21aecf9514f446522dd48ab7147753a3099979fa3ec5c030ebb
 size 14645

trainer_state.json

@@ -11,799 +11,799 @@
   "log_history": [
     {
       "epoch": 0.060350030175015085,
-      "grad_norm": 0.21468117833137512,
+      "grad_norm": 0.2307848036289215,
       "learning_rate": 8.18181818181818e-05,
-      "loss": 1.8508,
-      "mean_token_accuracy": 0.6162685614824295,
-      "num_tokens": 157389.0,
+      "loss": 1.8669,
+      "mean_token_accuracy": 0.6139279180765151,
+      "num_tokens": 157037.0,
       "step": 25
     },
     {
       "epoch": 0.12070006035003017,
-      "grad_norm": 0.28003624081611633,
+      "grad_norm": 0.315602570772171,
       "learning_rate": 0.00016704545454545452,
-      "loss": 1.0701,
-      "mean_token_accuracy": 0.7399352079629898,
-      "num_tokens": 284181.0,
+      "loss": 1.0736,
+      "mean_token_accuracy": 0.7409161877632141,
+      "num_tokens": 282266.0,
       "step": 50
     },
     {
       "epoch": 0.18105009052504525,
-      "grad_norm": 0.226594477891922,
+      "grad_norm": 0.2508843243122101,
       "learning_rate": 0.0002522727272727273,
-      "loss": 0.6641,
-      "mean_token_accuracy": 0.8210779559612275,
-      "num_tokens": 442731.0,
+      "loss": 0.67,
+      "mean_token_accuracy": 0.819593402147293,
+      "num_tokens": 439795.0,
       "step": 75
     },
     {
       "epoch": 0.24140012070006034,
-      "grad_norm": 0.4477122724056244,
+      "grad_norm": 0.4052221179008484,
       "learning_rate": 0.0002999887132933212,
-      "loss": 0.5388,
-      "mean_token_accuracy": 0.8498140323162079,
-      "num_tokens": 570887.0,
+      "loss": 0.5476,
+      "mean_token_accuracy": 0.8481135076284408,
+      "num_tokens": 567017.0,
       "step": 100
     },
     {
       "epoch": 0.30175015087507545,
-      "grad_norm": 0.270370751619339,
+      "grad_norm": 0.28632256388664246,
       "learning_rate": 0.0002998791256978121,
-      "loss": 0.4147,
-      "mean_token_accuracy": 0.8810201096534729,
-      "num_tokens": 729150.0,
+      "loss": 0.4307,
+      "mean_token_accuracy": 0.8766800117492676,
+      "num_tokens": 723951.0,
       "step": 125
     },
     {
       "epoch": 0.3621001810500905,
-      "grad_norm": 0.3737528324127197,
+      "grad_norm": 0.37199538946151733,
       "learning_rate": 0.0002996530399366737,
-      "loss": 0.3783,
-      "mean_token_accuracy": 0.8932037615776062,
-      "num_tokens": 855423.0,
+      "loss": 0.3577,
+      "mean_token_accuracy": 0.8976909649372101,
+      "num_tokens": 850124.0,
       "step": 150
     },
     {
       "epoch": 0.4224502112251056,
-      "grad_norm": 0.32675644755363464,
+      "grad_norm": 0.24243904650211334,
       "learning_rate": 0.00029931063174202567,
-      "loss": 0.2747,
-      "mean_token_accuracy": 0.9195706987380982,
-      "num_tokens": 1013920.0,
+      "loss": 0.2541,
+      "mean_token_accuracy": 0.9251525086164475,
+      "num_tokens": 1007442.0,
       "step": 175
     },
     {
       "epoch": 0.4828002414001207,
-      "grad_norm": 0.40882647037506104,
+      "grad_norm": 0.395974338054657,
       "learning_rate": 0.00029885216726118104,
-      "loss": 0.2362,
-      "mean_token_accuracy": 0.9326811420917511,
-      "num_tokens": 1139058.0,
+      "loss": 0.2702,
+      "mean_token_accuracy": 0.923055305480957,
+      "num_tokens": 1134123.0,
       "step": 200
     },
     {
       "epoch": 0.5431502715751357,
-      "grad_norm": 0.2929173409938812,
+      "grad_norm": 0.2322138547897339,
       "learning_rate": 0.00029827800284977474,
-      "loss": 0.1925,
-      "mean_token_accuracy": 0.943121486902237,
-      "num_tokens": 1294639.0,
+      "loss": 0.1811,
+      "mean_token_accuracy": 0.9476957750320435,
+      "num_tokens": 1293162.0,
       "step": 225
     },
     {
       "epoch": 0.6035003017501509,
-      "grad_norm": 0.49980688095092773,
+      "grad_norm": 0.33882594108581543,
       "learning_rate": 0.00029758858479477575,
-      "loss": 0.1879,
-      "mean_token_accuracy": 0.9459474259614944,
-      "num_tokens": 1420667.0,
+      "loss": 0.1739,
+      "mean_token_accuracy": 0.949301148056984,
+      "num_tokens": 1420644.0,
       "step": 250
     },
     {
       "epoch": 0.663850331925166,
-      "grad_norm": 0.2513156533241272,
+      "grad_norm": 0.1778780221939087,
       "learning_rate": 0.0002967844489675963,
-      "loss": 0.1538,
-      "mean_token_accuracy": 0.9570270645618438,
-      "num_tokens": 1577420.0,
+      "loss": 0.1542,
+      "mean_token_accuracy": 0.9562543380260468,
+      "num_tokens": 1577978.0,
       "step": 275
     },
     {
       "epoch": 0.724200362100181,
-      "grad_norm": 0.48155277967453003,
+      "grad_norm": 0.24160577356815338,
       "learning_rate": 0.00029586622040756957,
-      "loss": 0.1349,
-      "mean_token_accuracy": 0.9620789134502411,
-      "num_tokens": 1705571.0,
+      "loss": 0.1321,
+      "mean_token_accuracy": 0.9632052791118622,
+      "num_tokens": 1704858.0,
       "step": 300
     },
     {
       "epoch": 0.7845503922751962,
-      "grad_norm": 0.16763296723365784,
+      "grad_norm": 0.21109254658222198,
       "learning_rate": 0.0002948346128361186,
-      "loss": 0.1231,
-      "mean_token_accuracy": 0.9658524990081787,
-      "num_tokens": 1863982.0,
+      "loss": 0.1185,
+      "mean_token_accuracy": 0.9662619507312775,
+      "num_tokens": 1862834.0,
       "step": 325
     },
     {
       "epoch": 0.8449004224502112,
-      "grad_norm": 0.3732389211654663,
+      "grad_norm": 0.3435733914375305,
       "learning_rate": 0.00029369042810199416,
-      "loss": 0.11,
-      "mean_token_accuracy": 0.9696171337366104,
-      "num_tokens": 1989622.0,
+      "loss": 0.1058,
+      "mean_token_accuracy": 0.9695043462514877,
+      "num_tokens": 1988425.0,
       "step": 350
     },
     {
       "epoch": 0.9052504526252263,
-      "grad_norm": 0.20290179550647736,
+      "grad_norm": 0.14936767518520355,
       "learning_rate": 0.0002924345555580135,
-      "loss": 0.0928,
-      "mean_token_accuracy": 0.9740620368719101,
-      "num_tokens": 2146763.0,
+      "loss": 0.0951,
+      "mean_token_accuracy": 0.9735645836591721,
+      "num_tokens": 2145635.0,
       "step": 375
     },
     {
       "epoch": 0.9656004828002414,
-      "grad_norm": 0.3237822651863098,
+      "grad_norm": 0.36445313692092896,
       "learning_rate": 0.000291067971369783,
-      "loss": 0.0932,
-      "mean_token_accuracy": 0.9736461997032165,
-      "num_tokens": 2273110.0,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9770607089996338,
+      "num_tokens": 2273525.0,
       "step": 400
     },
     {
       "epoch": 1.0,
-      "eval_loss": 0.08881417661905289,
-      "eval_mean_token_accuracy": 0.9754456977586489,
+      "eval_loss": 0.08547013998031616,
+      "eval_mean_token_accuracy": 0.9758709730328741,
       "eval_num_tokens": 2354180.0,
-      "eval_runtime": 62.77,
-      "eval_samples_per_second": 5.879,
-      "eval_steps_per_second": 2.947,
+      "eval_runtime": 62.7554,
+      "eval_samples_per_second": 5.88,
+      "eval_steps_per_second": 2.948,
       "step": 415
     },
     {
       "epoch": 1.024140012070006,
-      "grad_norm": 0.13449090719223022,
+      "grad_norm": 0.17265604436397552,
       "learning_rate": 0.0002895917377569438,
-      "loss": 0.0954,
-      "mean_token_accuracy": 0.9730558665757326,
-      "num_tokens": 2423130.0,
+      "loss": 0.0897,
+      "mean_token_accuracy": 0.9755592708735122,
+      "num_tokens": 2423409.0,
       "step": 425
     },
     {
       "epoch": 1.0844900422450212,
-      "grad_norm": 0.2197084277868271,
+      "grad_norm": 0.17761775851249695,
       "learning_rate": 0.00028800700216752875,
-      "loss": 0.0551,
-      "mean_token_accuracy": 0.9840189307928086,
-      "num_tokens": 2564094.0,
+      "loss": 0.0569,
+      "mean_token_accuracy": 0.9834855335950852,
+      "num_tokens": 2566528.0,
       "step": 450
     },
     {
       "epoch": 1.1448400724200363,
-      "grad_norm": 0.13322922587394714,
+      "grad_norm": 0.16882313787937164,
       "learning_rate": 0.00028631499638607285,
-      "loss": 0.0742,
-      "mean_token_accuracy": 0.979583665728569,
-      "num_tokens": 2704963.0,
+      "loss": 0.0793,
+      "mean_token_accuracy": 0.9781825703382492,
+      "num_tokens": 2709916.0,
       "step": 475
     },
     {
       "epoch": 1.2051901025950513,
-      "grad_norm": 0.18199768662452698,
+      "grad_norm": 0.12463518232107162,
       "learning_rate": 0.0002845170355761712,
-      "loss": 0.0578,
-      "mean_token_accuracy": 0.9836762601137161,
-      "num_tokens": 2847569.0,
+      "loss": 0.0574,
+      "mean_token_accuracy": 0.9840801757574081,
+      "num_tokens": 2854287.0,
       "step": 500
     },
     {
       "epoch": 1.2655401327700664,
-      "grad_norm": 0.14187701046466827,
+      "grad_norm": 0.12803447246551514,
       "learning_rate": 0.0002826145172582274,
-      "loss": 0.0688,
-      "mean_token_accuracy": 0.9806136053800583,
-      "num_tokens": 2989847.0,
+      "loss": 0.0645,
+      "mean_token_accuracy": 0.9821405208110809,
+      "num_tokens": 2996401.0,
       "step": 525
     },
     {
       "epoch": 1.3258901629450814,
-      "grad_norm": 0.3071412146091461,
+      "grad_norm": 0.10773979872465134,
       "learning_rate": 0.00028060892022318764,
-      "loss": 0.0552,
-      "mean_token_accuracy": 0.9846898633241653,
-      "num_tokens": 3134888.0,
+      "loss": 0.054,
+      "mean_token_accuracy": 0.9844699889421463,
+      "num_tokens": 3138381.0,
       "step": 550
     },
     {
       "epoch": 1.3862401931200965,
-      "grad_norm": 0.16651563346385956,
+      "grad_norm": 0.14240588247776031,
       "learning_rate": 0.0002785018033831051,
-      "loss": 0.063,
-      "mean_token_accuracy": 0.9824958252906799,
-      "num_tokens": 3277522.0,
+      "loss": 0.0744,
+      "mean_token_accuracy": 0.9793990004062653,
+      "num_tokens": 3279432.0,
       "step": 575
     },
     {
       "epoch": 1.4465902232951118,
-      "grad_norm": 0.08527401834726334,
+      "grad_norm": 0.12011445313692093,
       "learning_rate": 0.0002762948045594276,
-      "loss": 0.0445,
-      "mean_token_accuracy": 0.986923239827156,
-      "num_tokens": 3419296.0,
+      "loss": 0.0508,
+      "mean_token_accuracy": 0.985177715420723,
+      "num_tokens": 3422945.0,
       "step": 600
     },
     {
       "epoch": 1.5069402534701268,
-      "grad_norm": 0.09827699512243271,
+      "grad_norm": 0.07065067440271378,
       "learning_rate": 0.0002739896392099502,
-      "loss": 0.0683,
-      "mean_token_accuracy": 0.9812941312789917,
-      "num_tokens": 3561482.0,
+      "loss": 0.0672,
+      "mean_token_accuracy": 0.9815575838088989,
+      "num_tokens": 3565901.0,
       "step": 625
     },
     {
       "epoch": 1.567290283645142,
-      "grad_norm": 0.1221788227558136,
+      "grad_norm": 0.13502469658851624,
       "learning_rate": 0.00027158809909542307,
-      "loss": 0.0535,
-      "mean_token_accuracy": 0.98518315076828,
-      "num_tokens": 3704279.0,
+      "loss": 0.0451,
+      "mean_token_accuracy": 0.98676638007164,
+      "num_tokens": 3708558.0,
       "step": 650
     },
     {
       "epoch": 1.627640313820157,
-      "grad_norm": 0.10258518159389496,
+      "grad_norm": 0.08584199845790863,
       "learning_rate": 0.00026909205088685,
-      "loss": 0.0616,
-      "mean_token_accuracy": 0.9829519605636596,
-      "num_tokens": 3845569.0,
+      "loss": 0.062,
+      "mean_token_accuracy": 0.9832429319620133,
+      "num_tokens": 3849682.0,
       "step": 675
     },
     {
       "epoch": 1.687990343995172,
-      "grad_norm": 0.09674005955457687,
+      "grad_norm": 0.10930341482162476,
       "learning_rate": 0.0002665034347145612,
-      "loss": 0.0422,
-      "mean_token_accuracy": 0.987600001692772,
-      "num_tokens": 3986667.0,
+      "loss": 0.0464,
+      "mean_token_accuracy": 0.9868946731090545,
+      "num_tokens": 3993080.0,
       "step": 700
     },
     {
       "epoch": 1.748340374170187,
-      "grad_norm": 0.1226281225681305,
+      "grad_norm": 0.1063772663474083,
       "learning_rate": 0.000263824262660187,
-      "loss": 0.0595,
-      "mean_token_accuracy": 0.9835411328077316,
-      "num_tokens": 4126147.0,
+      "loss": 0.0611,
+      "mean_token_accuracy": 0.9831858760118485,
+      "num_tokens": 4135566.0,
       "step": 725
     },
     {
       "epoch": 1.8086904043452021,
-      "grad_norm": 0.16110199689865112,
+      "grad_norm": 0.11367379873991013,
       "learning_rate": 0.0002610566171927056,
-      "loss": 0.0399,
-      "mean_token_accuracy": 0.9882411390542984,
-      "num_tokens": 4268948.0,
+      "loss": 0.0426,
+      "mean_token_accuracy": 0.9874300253391266,
+      "num_tokens": 4278312.0,
       "step": 750
     },
     {
       "epoch": 1.8690404345202172,
-      "grad_norm": 0.14910632371902466,
+      "grad_norm": 0.09623493254184723,
       "learning_rate": 0.00025820264954977976,
-      "loss": 0.0627,
-      "mean_token_accuracy": 0.9824194663763046,
-      "num_tokens": 4410572.0,
+      "loss": 0.0594,
+      "mean_token_accuracy": 0.9828196185827255,
+      "num_tokens": 4417338.0,
       "step": 775
     },
     {
       "epoch": 1.9293904646952322,
-      "grad_norm": 0.1125180646777153,
+      "grad_norm": 0.1381269097328186,
       "learning_rate": 0.00025526457806564136,
-      "loss": 0.0425,
-      "mean_token_accuracy": 0.987920394539833,
-      "num_tokens": 4553968.0,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9886600559949875,
+      "num_tokens": 4556744.0,
       "step": 800
     },
     {
       "epoch": 1.9897404948702473,
-      "grad_norm": 0.11598383635282516,
+      "grad_norm": 0.10048148781061172,
       "learning_rate": 0.00025224468644682245,
-      "loss": 0.046,
-      "mean_token_accuracy": 0.98680981695652,
-      "num_tokens": 4687933.0,
+      "loss": 0.0436,
+      "mean_token_accuracy": 0.9878453743457795,
+      "num_tokens": 4688626.0,
       "step": 825
     },
     {
       "epoch": 2.0,
-      "eval_loss": 0.05183422192931175,
-      "eval_mean_token_accuracy": 0.9854742443239367,
+      "eval_loss": 0.05272972211241722,
+      "eval_mean_token_accuracy": 0.9853407392630706,
       "eval_num_tokens": 4708360.0,
-      "eval_runtime": 62.7258,
-      "eval_samples_per_second": 5.883,
-      "eval_steps_per_second": 2.949,
+      "eval_runtime": 62.7886,
+      "eval_samples_per_second": 5.877,
+      "eval_steps_per_second": 2.946,
       "step": 830
     },
     {
       "epoch": 2.048280024140012,
-      "grad_norm": 0.12139257043600082,
+      "grad_norm": 0.14225316047668457,
       "learning_rate": 0.00024914532199707444,
-      "loss": 0.0467,
-      "mean_token_accuracy": 0.9864320312578654,
-      "num_tokens": 4837944.0,
+      "loss": 0.0429,
+      "mean_token_accuracy": 0.9869677971318825,
+      "num_tokens": 4838704.0,
       "step": 850
     },
     {
       "epoch": 2.1086300543150274,
-      "grad_norm": 0.07626475393772125,
+      "grad_norm": 0.16189508140087128,
       "learning_rate": 0.00024596889379285353,
-      "loss": 0.0334,
-      "mean_token_accuracy": 0.9904208928346634,
-      "num_tokens": 4969799.0,
+      "loss": 0.0353,
+      "mean_token_accuracy": 0.9899542284011841,
+      "num_tokens": 4970478.0,
       "step": 875
     },
     {
       "epoch": 2.1689800844900424,
-      "grad_norm": 0.0758512020111084,
+      "grad_norm": 0.08687274158000946,
       "learning_rate": 0.00024271787081079228,
-      "loss": 0.044,
-      "mean_token_accuracy": 0.9870604687929153,
-      "num_tokens": 5120530.0,
+      "loss": 0.0464,
+      "mean_token_accuracy": 0.9863937604427337,
+      "num_tokens": 5120208.0,
       "step": 900
     },
     {
       "epoch": 2.2293301146650575,
-      "grad_norm": 0.11110670864582062,
+      "grad_norm": 0.10480177402496338,
       "learning_rate": 0.00023939478000861117,
-      "loss": 0.0318,
-      "mean_token_accuracy": 0.9904728019237519,
-      "num_tokens": 5253594.0,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9905521881580353,
+      "num_tokens": 5252759.0,
       "step": 925
     },
     {
       "epoch": 2.2896801448400725,
-      "grad_norm": 0.05050364509224892,
+      "grad_norm": 0.07012847810983658,
       "learning_rate": 0.00023600220436096318,
-      "loss": 0.0474,
-      "mean_token_accuracy": 0.9859576737880706,
-      "num_tokens": 5406515.0,
+      "loss": 0.0477,
+      "mean_token_accuracy": 0.98607047021389,
+      "num_tokens": 5405015.0,
       "step": 950
     },
     {
       "epoch": 2.3500301750150876,
-      "grad_norm": 0.10038283467292786,
+      "grad_norm": 0.23715409636497498,
       "learning_rate": 0.00023254278085173684,
-      "loss": 0.0317,
-      "mean_token_accuracy": 0.9904199486970902,
-      "num_tokens": 5540873.0,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9909250718355179,
+      "num_tokens": 5538251.0,
       "step": 975
     },
     {
       "epoch": 2.4103802051901027,
-      "grad_norm": 0.06138943135738373,
+      "grad_norm": 0.08109632134437561,
       "learning_rate": 0.00022901919842437972,
-      "loss": 0.0416,
-      "mean_token_accuracy": 0.9875155621767044,
-      "num_tokens": 5690736.0,
+      "loss": 0.0474,
+      "mean_token_accuracy": 0.9859592580795288,
+      "num_tokens": 5691272.0,
       "step": 1000
     },
     {
       "epoch": 2.4707302353651177,
-      "grad_norm": 0.053865499794483185,
+      "grad_norm": 0.09655272960662842,
       "learning_rate": 0.00022543419589183397,
-      "loss": 0.0279,
-      "mean_token_accuracy": 0.9913687032461166,
-      "num_tokens": 5822410.0,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9903653126955032,
+      "num_tokens": 5824972.0,
       "step": 1025
     },
     {
       "epoch": 2.5310802655401328,
-      "grad_norm": 0.06363498419523239,
+      "grad_norm": 0.052605971693992615,
       "learning_rate": 0.00022179055980770993,
-      "loss": 0.0438,
-      "mean_token_accuracy": 0.9867295175790787,
-      "num_tokens": 5973587.0,
+      "loss": 0.0412,
+      "mean_token_accuracy": 0.9876793098449707,
+      "num_tokens": 5975189.0,
       "step": 1050
     },
     {
       "epoch": 2.591430295715148,
-      "grad_norm": 0.09079308062791824,
+      "grad_norm": 0.08622315526008606,
       "learning_rate": 0.0002180911223003513,
-      "loss": 0.0283,
-      "mean_token_accuracy": 0.9913936233520508,
-      "num_tokens": 6106943.0,
+      "loss": 0.0295,
+      "mean_token_accuracy": 0.9909289944171905,
+      "num_tokens": 6108290.0,
       "step": 1075
     },
     {
       "epoch": 2.651780325890163,
-      "grad_norm": 0.06340842694044113,
+      "grad_norm": 0.07412232458591461,
       "learning_rate": 0.00021433875887147627,
-      "loss": 0.0426,
-      "mean_token_accuracy": 0.9875069695711136,
-      "num_tokens": 6259442.0,
+      "loss": 0.0468,
+      "mean_token_accuracy": 0.986087337732315,
+      "num_tokens": 6259582.0,
       "step": 1100
     },
     {
       "epoch": 2.712130356065178,
-      "grad_norm": 0.0556156300008297,
+      "grad_norm": 0.09526045620441437,
       "learning_rate": 0.00021053638616110525,
-      "loss": 0.029,
-      "mean_token_accuracy": 0.9911601048707962,
-      "num_tokens": 6392790.0,
+      "loss": 0.0292,
+      "mean_token_accuracy": 0.9911455798149109,
+      "num_tokens": 6392200.0,
       "step": 1125
     },
     {
       "epoch": 2.772480386240193,
-      "grad_norm": 0.07941101491451263,
+      "grad_norm": 0.07069120556116104,
       "learning_rate": 0.00020668695968051274,
-      "loss": 0.0435,
-      "mean_token_accuracy": 0.9869343960285186,
-      "num_tokens": 6545535.0,
+      "loss": 0.044,
+      "mean_token_accuracy": 0.9870589804649353,
+      "num_tokens": 6543588.0,
       "step": 1150
     },
     {
       "epoch": 2.832830416415208,
-      "grad_norm": 0.053617168217897415,
+      "grad_norm": 0.09000532329082489,
       "learning_rate": 0.00020279347151496482,
-      "loss": 0.0285,
-      "mean_token_accuracy": 0.9914081859588623,
-      "num_tokens": 6679597.0,
+      "loss": 0.0296,
+      "mean_token_accuracy": 0.9909429329633713,
+      "num_tokens": 6676376.0,
       "step": 1175
     },
     {
       "epoch": 2.8931804465902236,
-      "grad_norm": 0.059137940406799316,
+      "grad_norm": 0.05883787199854851,
       "learning_rate": 0.00019885894799802922,
-      "loss": 0.0439,
-      "mean_token_accuracy": 0.98672642827034,
-      "num_tokens": 6830721.0,
+      "loss": 0.0424,
+      "mean_token_accuracy": 0.9874570268392563,
+      "num_tokens": 6827932.0,
       "step": 1200
     },
     {
       "epoch": 2.9535304767652386,
-      "grad_norm": 0.054504215717315674,
+      "grad_norm": 0.13316047191619873,
       "learning_rate": 0.00019488644735926396,
-      "loss": 0.0289,
-      "mean_token_accuracy": 0.9915051358938217,
-      "num_tokens": 6962566.0,
+      "loss": 0.0295,
+      "mean_token_accuracy": 0.991017809510231,
+      "num_tokens": 6961443.0,
       "step": 1225
     },
     {
       "epoch": 3.0,
-      "eval_loss": 0.04319874569773674,
-      "eval_mean_token_accuracy": 0.9879072198996672,
+      "eval_loss": 0.044597771018743515,
+      "eval_mean_token_accuracy": 0.9876517611580926,
       "eval_num_tokens": 7062540.0,
-      "eval_runtime": 62.809,
-      "eval_samples_per_second": 5.875,
-      "eval_steps_per_second": 2.945,
+      "eval_runtime": 62.7187,
+      "eval_samples_per_second": 5.883,
+      "eval_steps_per_second": 2.95,
       "step": 1245
     },
     {
       "epoch": 3.012070006035003,
-      "grad_norm": 0.07366561144590378,
+      "grad_norm": 0.15881508588790894,
       "learning_rate": 0.00019087905734711452,
       "loss": 0.0369,
-      "mean_token_accuracy": 0.9888100163223817,
-      "num_tokens": 7098485.0,
+      "mean_token_accuracy": 0.9889105128258774,
+      "num_tokens": 7100107.0,
       "step": 1250
     },
     {
       "epoch": 3.0724200362100182,
-      "grad_norm": 0.04900413379073143,
+      "grad_norm": 0.04447310045361519,
       "learning_rate": 0.00018683989282886613,
-      "loss": 0.0243,
-      "mean_token_accuracy": 0.992353920340538,
-      "num_tokens": 7246823.0,
+      "loss": 0.0279,
+      "mean_token_accuracy": 0.9910174655914307,
+      "num_tokens": 7250123.0,
       "step": 1275
     },
     {
       "epoch": 3.1327700663850333,
-      "grad_norm": 0.11869508028030396,
+      "grad_norm": 0.05848820507526398,
       "learning_rate": 0.0001827720933695173,
-      "loss": 0.0293,
-      "mean_token_accuracy": 0.9912081670761108,
-      "num_tokens": 7381683.0,
+      "loss": 0.0283,
+      "mean_token_accuracy": 0.9908898574113846,
+      "num_tokens": 7386980.0,
       "step": 1300
     },
     {
       "epoch": 3.1931200965600484,
-      "grad_norm": 0.06294097751379013,
+      "grad_norm": 0.05700680986046791,
       "learning_rate": 0.00017867882079145627,
       "loss": 0.0268,
-      "mean_token_accuracy": 0.9917609655857086,
-      "num_tokens": 7528755.0,
+      "mean_token_accuracy": 0.9915226358175278,
+      "num_tokens": 7534097.0,
       "step": 1325
     },
     {
       "epoch": 3.2534701267350634,
-      "grad_norm": 0.1576467901468277,
+      "grad_norm": 0.06182008236646652,
       "learning_rate": 0.00017456325671683724,
-      "loss": 0.0286,
-      "mean_token_accuracy": 0.9911142766475678,
-      "num_tokens": 7663142.0,
+      "loss": 0.0293,
+      "mean_token_accuracy": 0.9909615200757981,
+      "num_tokens": 7670112.0,
       "step": 1350
     },
     {
       "epoch": 3.3138201569100785,
-      "grad_norm": 0.05235208570957184,
+      "grad_norm": 0.05402887985110283,
       "learning_rate": 0.00017042860009456638,
-      "loss": 0.0259,
-      "mean_token_accuracy": 0.991670835018158,
-      "num_tokens": 7809523.0,
+      "loss": 0.0269,
+      "mean_token_accuracy": 0.9914444983005524,
+      "num_tokens": 7819619.0,
       "step": 1375
     },
     {
       "epoch": 3.3741701870850935,
-      "grad_norm": 0.0751878172159195,
+      "grad_norm": 0.07528018951416016,
       "learning_rate": 0.00016627806471382066,
-      "loss": 0.0294,
-      "mean_token_accuracy": 0.9906228709220887,
-      "num_tokens": 7944654.0,
+      "loss": 0.0282,
+      "mean_token_accuracy": 0.9909832864999771,
+      "num_tokens": 7955967.0,
       "step": 1400
     },
     {
       "epoch": 3.4345202172601086,
-      "grad_norm": 0.03894800692796707,
+      "grad_norm": 0.05704252049326897,
       "learning_rate": 0.00016211487670603078,
-      "loss": 0.0269,
-      "mean_token_accuracy": 0.9913842624425888,
-      "num_tokens": 8092398.0,
+      "loss": 0.0259,
+      "mean_token_accuracy": 0.9916054201126099,
+      "num_tokens": 8102608.0,
       "step": 1425
     },
     {
       "epoch": 3.4948702474351236,
-      "grad_norm": 0.09787385165691376,
+      "grad_norm": 0.08405063301324844,
       "learning_rate": 0.0001579422720372715,
-      "loss": 0.0288,
-      "mean_token_accuracy": 0.9911862814426422,
-      "num_tokens": 8228938.0,
+      "loss": 0.0289,
+      "mean_token_accuracy": 0.9906303453445434,
+      "num_tokens": 8236645.0,
       "step": 1450
     },
     {
       "epoch": 3.5552202776101387,
-      "grad_norm": 0.05320660397410393,
+      "grad_norm": 0.03225329518318176,
       "learning_rate": 0.00015376349399300745,
-      "loss": 0.0291,
-      "mean_token_accuracy": 0.9911050695180893,
-      "num_tokens": 8378749.0,
+      "loss": 0.0261,
+      "mean_token_accuracy": 0.99147840321064,
+      "num_tokens": 8383287.0,
       "step": 1475
     },
     {
       "epoch": 3.6155703077851538,
-      "grad_norm": 0.08408358693122864,
+      "grad_norm": 0.09672655165195465,
       "learning_rate": 0.0001495817906571492,
-      "loss": 0.0312,
-      "mean_token_accuracy": 0.9901037472486496,
-      "num_tokens": 8516085.0,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9903279423713685,
+      "num_tokens": 8517618.0,
       "step": 1500
     },
     {
       "epoch": 3.675920337960169,
-      "grad_norm": 0.050662487745285034,
+      "grad_norm": 0.03548620641231537,
       "learning_rate": 0.00014540041238738055,
-      "loss": 0.0243,
-      "mean_token_accuracy": 0.9920293813943863,
-      "num_tokens": 8664227.0,
+      "loss": 0.0256,
+      "mean_token_accuracy": 0.9921301937103272,
+      "num_tokens": 8665120.0,
       "step": 1525
     },
     {
       "epoch": 3.736270368135184,
-      "grad_norm": 0.11257333308458328,
+      "grad_norm": 0.05516292154788971,
       "learning_rate": 0.00014122260928871734,
-      "loss": 0.0304,
-      "mean_token_accuracy": 0.990105789899826,
-      "num_tokens": 8802546.0,
+      "loss": 0.0282,
+      "mean_token_accuracy": 0.9913522702455521,
+      "num_tokens": 8800218.0,
       "step": 1550
     },
     {
       "epoch": 3.796620398310199,
-      "grad_norm": 0.05049088969826698,
+      "grad_norm": 0.05250876769423485,
       "learning_rate": 0.00013705162868726396,
-      "loss": 0.0236,
-      "mean_token_accuracy": 0.992456527352333,
-      "num_tokens": 8952407.0,
+      "loss": 0.025,
+      "mean_token_accuracy": 0.9920567196607589,
+      "num_tokens": 8948083.0,
       "step": 1575
     },
     {
       "epoch": 3.856970428485214,
-      "grad_norm": 0.07625644654035568,
+      "grad_norm": 0.06773823499679565,
       "learning_rate": 0.00013289071260612855,
-      "loss": 0.027,
-      "mean_token_accuracy": 0.991376177072525,
-      "num_tokens": 9087703.0,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9904211789369584,
+      "num_tokens": 9084691.0,
       "step": 1600
     },
     {
       "epoch": 3.9173204586602295,
-      "grad_norm": 0.03547623008489609,
+      "grad_norm": 0.041708026081323624,
       "learning_rate": 0.00012874309524546083,
-      "loss": 0.0255,
-      "mean_token_accuracy": 0.991724653840065,
-      "num_tokens": 9236298.0,
+      "loss": 0.0252,
+      "mean_token_accuracy": 0.9919029641151428,
+      "num_tokens": 9235980.0,
       "step": 1625
     },
     {
       "epoch": 3.9776704888352445,
-      "grad_norm": 0.056914571672677994,
+      "grad_norm": 0.04703540354967117,
       "learning_rate": 0.00012461200046857084,
-      "loss": 0.0258,
-      "mean_token_accuracy": 0.9917834293842316,
-      "num_tokens": 9369153.0,
+      "loss": 0.0262,
+      "mean_token_accuracy": 0.99182637155056,
+      "num_tokens": 9368839.0,
       "step": 1650
     },
     {
       "epoch": 4.0,
-      "eval_loss": 0.04025053605437279,
-      "eval_mean_token_accuracy": 0.9888840298394899,
+      "eval_loss": 0.040296006947755814,
+      "eval_mean_token_accuracy": 0.9890867027076515,
       "eval_num_tokens": 9416720.0,
-      "eval_runtime": 62.7742,
-      "eval_samples_per_second": 5.878,
-      "eval_steps_per_second": 2.947,
+      "eval_runtime": 62.7415,
+      "eval_samples_per_second": 5.881,
+      "eval_steps_per_second": 2.949,
       "step": 1660
     },
     {
       "epoch": 4.036210018105009,
-      "grad_norm": 0.03895945847034454,
+      "grad_norm": 0.05349954590201378,
       "learning_rate": 0.00012050063929608123,
-      "loss": 0.025,
-      "mean_token_accuracy": 0.9920010032113066,
-      "num_tokens": 9517464.0,
+      "loss": 0.0245,
+      "mean_token_accuracy": 0.9919631487315463,
+      "num_tokens": 9516874.0,
       "step": 1675
     },
     {
       "epoch": 4.096560048280024,
-      "grad_norm": 0.038321759551763535,
+      "grad_norm": 0.058036044239997864,
       "learning_rate": 0.0001164122074100633,
-      "loss": 0.0196,
-      "mean_token_accuracy": 0.9934940075874329,
-      "num_tokens": 9654511.0,
+      "loss": 0.0204,
+      "mean_token_accuracy": 0.9933006262779236,
+      "num_tokens": 9654378.0,
       "step": 1700
     },
     {
       "epoch": 4.15691007845504,
-      "grad_norm": 0.048036105930805206,
+      "grad_norm": 0.06282967329025269,
       "learning_rate": 0.00011234988267009415,
-      "loss": 0.0231,
-      "mean_token_accuracy": 0.9922078281641007,
-      "num_tokens": 9800672.0,
+      "loss": 0.0228,
+      "mean_token_accuracy": 0.9928084880113601,
+      "num_tokens": 9798671.0,
       "step": 1725
     },
     {
       "epoch": 4.217260108630055,
-      "grad_norm": 0.09493701159954071,
+      "grad_norm": 0.03295363485813141,
       "learning_rate": 0.00010831682264316787,
-      "loss": 0.0204,
-      "mean_token_accuracy": 0.993373184800148,
-      "num_tokens": 9938087.0,
+      "loss": 0.0199,
+      "mean_token_accuracy": 0.9936361283063888,
+      "num_tokens": 9936183.0,
       "step": 1750
     },
     {
       "epoch": 4.27761013880507,
-      "grad_norm": 0.056498508900403976,
+      "grad_norm": 0.07894692569971085,
       "learning_rate": 0.00010431616214937911,
-      "loss": 0.0227,
-      "mean_token_accuracy": 0.9926544779539108,
-      "num_tokens": 10084429.0,
+      "loss": 0.0248,
+      "mean_token_accuracy": 0.9920186513662338,
+      "num_tokens": 10083447.0,
       "step": 1775
     },
     {
       "epoch": 4.337960168980085,
-      "grad_norm": 0.04016057401895523,
+      "grad_norm": 0.052625857293605804,
       "learning_rate": 0.00010035101082528777,
-      "loss": 0.0197,
-      "mean_token_accuracy": 0.9933375012874603,
-      "num_tokens": 10221190.0,
+      "loss": 0.0198,
+      "mean_token_accuracy": 0.9934204763174057,
+      "num_tokens": 10222918.0,
       "step": 1800
     },
     {
       "epoch": 4.3983101991551,
-      "grad_norm": 0.06675443053245544,
+      "grad_norm": 0.05687833949923515,
       "learning_rate": 9.642445070685809e-05,
-      "loss": 0.0253,
-      "mean_token_accuracy": 0.992013863325119,
-      "num_tokens": 10367374.0,
+      "loss": 0.0245,
+      "mean_token_accuracy": 0.9919208544492721,
+      "num_tokens": 10371876.0,
       "step": 1825
     },
     {
       "epoch": 4.458660229330115,
-      "grad_norm": 0.06365606188774109,
+      "grad_norm": 0.036979444324970245,
       "learning_rate": 9.253953383385157e-05,
-      "loss": 0.0198,
-      "mean_token_accuracy": 0.9934730947017669,
-      "num_tokens": 10505133.0,
+      "loss": 0.0187,
+      "mean_token_accuracy": 0.9938018727302551,
+      "num_tokens": 10509835.0,
       "step": 1850
     },
     {
       "epoch": 4.51901025950513,
-      "grad_norm": 0.04390507563948631,
+      "grad_norm": 0.08300971984863281,
       "learning_rate": 8.869927987753459e-05,
-      "loss": 0.025,
-      "mean_token_accuracy": 0.9921487855911255,
-      "num_tokens": 10654059.0,
+      "loss": 0.0237,
+      "mean_token_accuracy": 0.9923136639595032,
+      "num_tokens": 10655174.0,
       "step": 1875
     },
     {
       "epoch": 4.579360289680145,
-      "grad_norm": 0.04688199982047081,
+      "grad_norm": 0.034062255173921585,
       "learning_rate": 8.490667379354661e-05,
-      "loss": 0.0193,
-      "mean_token_accuracy": 0.9935662603378296,
-      "num_tokens": 10791488.0,
+      "loss": 0.0192,
+      "mean_token_accuracy": 0.9937297248840332,
+      "num_tokens": 10792672.0,
       "step": 1900
     },
     {
       "epoch": 4.63971031985516,
-      "grad_norm": 0.048728689551353455,
+      "grad_norm": 0.09020201861858368,
       "learning_rate": 8.116466350175079e-05,
-      "loss": 0.0219,
-      "mean_token_accuracy": 0.9928493529558182,
-      "num_tokens": 10936918.0,
+      "loss": 0.0239,
+      "mean_token_accuracy": 0.9922147291898727,
+      "num_tokens": 10939329.0,
       "step": 1925
     },
     {
       "epoch": 4.700060350030175,
-      "grad_norm": 0.03529852256178856,
+      "grad_norm": 0.055495839565992355,
       "learning_rate": 7.747615759487304e-05,
-      "loss": 0.019,
-      "mean_token_accuracy": 0.9935548371076584,
-      "num_tokens": 11074592.0,
+      "loss": 0.0185,
+      "mean_token_accuracy": 0.9938122999668121,
+      "num_tokens": 11077006.0,
       "step": 1950
     },
     {
       "epoch": 4.76041038020519,
-      "grad_norm": 0.04538232088088989,
+      "grad_norm": 0.05608997866511345,
       "learning_rate": 7.38440230777085e-05,
-      "loss": 0.0231,
-      "mean_token_accuracy": 0.9924061894416809,
-      "num_tokens": 11221831.0,
+      "loss": 0.0255,
+      "mean_token_accuracy": 0.9921148890256881,
+      "num_tokens": 11222702.0,
       "step": 1975
     },
     {
       "epoch": 4.820760410380205,
-      "grad_norm": 0.05990992859005928,
+      "grad_norm": 0.05004817619919777,
       "learning_rate": 7.027108313865378e-05,
-      "loss": 0.018,
-      "mean_token_accuracy": 0.9938581752777099,
-      "num_tokens": 11360739.0,
+      "loss": 0.02,
+      "mean_token_accuracy": 0.9935011225938797,
+      "num_tokens": 11359725.0,
       "step": 2000
     },
     {
       "epoch": 4.88111044055522,
-      "grad_norm": 0.05956440791487694,
+      "grad_norm": 0.054554738104343414,
       "learning_rate": 6.676011495529687e-05,
-      "loss": 0.0242,
-      "mean_token_accuracy": 0.9921341562271118,
-      "num_tokens": 11507676.0,
+      "loss": 0.0231,
+      "mean_token_accuracy": 0.9923803770542144,
+      "num_tokens": 11506931.0,
       "step": 2025
     },
     {
       "epoch": 4.941460470730235,
-      "grad_norm": 0.03628522902727127,
+      "grad_norm": 0.042225901037454605,
       "learning_rate": 6.331384753577056e-05,
-      "loss": 0.0186,
-      "mean_token_accuracy": 0.9937015652656556,
-      "num_tokens": 11646256.0,
+      "loss": 0.0187,
+      "mean_token_accuracy": 0.9937945604324341,
+      "num_tokens": 11643259.0,
       "step": 2050
     },
     {
       "epoch": 5.0,
-      "grad_norm": 0.12821151316165924,
+      "grad_norm": 0.13635069131851196,
       "learning_rate": 5.993495959754631e-05,
-      "loss": 0.0215,
-      "mean_token_accuracy": 0.9930061482891595,
+      "loss": 0.0206,
+      "mean_token_accuracy": 0.9932199802595315,
       "num_tokens": 11770900.0,
       "step": 2075
     },
     {
       "epoch": 5.0,
-      "eval_loss": 0.0397900752723217,
-      "eval_mean_token_accuracy": 0.9896509357400842,
+      "eval_loss": 0.039575137197971344,
+      "eval_mean_token_accuracy": 0.9895688775423411,
       "eval_num_tokens": 11770900.0,
-      "eval_runtime": 62.8881,
-      "eval_samples_per_second": 5.868,
-      "eval_steps_per_second": 2.942,
+      "eval_runtime": 62.7395,
+      "eval_samples_per_second": 5.881,
+      "eval_steps_per_second": 2.949,
       "step": 2075
     }
   ],
@@ -824,7 +824,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 5.056626317959158e+17,
+  "total_flos": 5.056986746477353e+17,
   "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b7976387d4d77aa2bbba99fb0c9d3198ef81bd50d2a09bece4b0fceea7d5b7d7
+oid sha256:9b88f0eeb3177d75967c3f15be8bf9827d703a5ee133080d3079062fe228fa99
 size 5969