Initial model upload

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cab5310cae32d01989e1f9ef72af4fea6df0190400e5c53d3e17161d600e5dfb
-size 7340032
+oid sha256:e0fb0c8a0509cdc3503852b0cd47bb33125d6de395cc17fead8a9e08084f9e69
+size 54222228

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b8df622ba3d6ea956be7164d00da5e75a1c29dc82ecdf21403a9d15437ffb728
-size 7340032
+oid sha256:2d8224815f4f53aa034cdfa633e01f8252201006e80d3c960f938293e7cb29ab
+size 108569611

rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8d3f19bc849d895ca9a0758650109d5bd3e5979db276cf56daab9cbeec109fed
+oid sha256:aadc7a9ad2323cee9959bc1b037df67cde7e3f6e374e424d79c78f13556f8e69
 size 14645

trainer_state.json

@@ -11,723 +11,723 @@
   "log_history": [
     {
       "epoch": 0.029123951537744643,
-      "grad_norm": 8.23077392578125,
+      "grad_norm": 7.719064712524414,
       "learning_rate": 4.951557160608885e-05,
-      "loss": 0.8096,
+      "loss": 0.8506,
       "step": 500
     },
     {
       "epoch": 0.058247903075489285,
-      "grad_norm": 7.531521797180176,
+      "grad_norm": 5.694650650024414,
       "learning_rate": 4.9030172413793105e-05,
-      "loss": 0.5976,
+      "loss": 0.685,
       "step": 1000
     },
     {
       "epoch": 0.08737185461323392,
-      "grad_norm": 7.849789142608643,
+      "grad_norm": 5.49603271484375,
       "learning_rate": 4.854477322149736e-05,
-      "loss": 0.5605,
+      "loss": 0.6337,
       "step": 1500
     },
     {
       "epoch": 0.11649580615097857,
-      "grad_norm": 10.54165267944336,
+      "grad_norm": 5.456332683563232,
       "learning_rate": 4.805937402920162e-05,
-      "loss": 0.5217,
+      "loss": 0.6202,
       "step": 2000
     },
     {
       "epoch": 0.1456197576887232,
-      "grad_norm": 5.914309024810791,
+      "grad_norm": 7.232175350189209,
       "learning_rate": 4.757397483690587e-05,
-      "loss": 0.5199,
+      "loss": 0.5985,
       "step": 2500
     },
     {
       "epoch": 0.17474370922646784,
-      "grad_norm": 4.58743143081665,
+      "grad_norm": 5.183197975158691,
       "learning_rate": 4.7088575644610125e-05,
-      "loss": 0.5066,
+      "loss": 0.5673,
       "step": 3000
     },
     {
       "epoch": 0.20386766076421248,
-      "grad_norm": 7.739262104034424,
+      "grad_norm": 7.102158546447754,
       "learning_rate": 4.660317645231438e-05,
-      "loss": 0.4885,
+      "loss": 0.5646,
       "step": 3500
     },
     {
       "epoch": 0.23299161230195714,
-      "grad_norm": 7.6342034339904785,
+      "grad_norm": 5.905701160430908,
       "learning_rate": 4.611777726001864e-05,
-      "loss": 0.4702,
+      "loss": 0.5482,
       "step": 4000
     },
     {
       "epoch": 0.26211556383970175,
-      "grad_norm": 4.743926525115967,
+      "grad_norm": 6.96270227432251,
       "learning_rate": 4.56323780677229e-05,
-      "loss": 0.4668,
+      "loss": 0.5398,
       "step": 4500
     },
     {
       "epoch": 0.2912395153774464,
-      "grad_norm": 6.3329596519470215,
+      "grad_norm": 7.838745594024658,
       "learning_rate": 4.514697887542715e-05,
-      "loss": 0.4621,
+      "loss": 0.5399,
       "step": 5000
     },
     {
       "epoch": 0.3203634669151911,
-      "grad_norm": 6.19777774810791,
+      "grad_norm": 5.497215270996094,
       "learning_rate": 4.466157968313141e-05,
-      "loss": 0.4548,
+      "loss": 0.5361,
       "step": 5500
     },
     {
       "epoch": 0.3494874184529357,
-      "grad_norm": 5.018973350524902,
+      "grad_norm": 6.11887264251709,
       "learning_rate": 4.417618049083567e-05,
-      "loss": 0.4546,
+      "loss": 0.5256,
       "step": 6000
     },
     {
       "epoch": 0.37861136999068035,
-      "grad_norm": 8.148200035095215,
+      "grad_norm": 5.674872398376465,
       "learning_rate": 4.369078129853993e-05,
-      "loss": 0.4488,
+      "loss": 0.5187,
       "step": 6500
     },
     {
       "epoch": 0.40773532152842495,
-      "grad_norm": 5.203968524932861,
+      "grad_norm": 10.958036422729492,
       "learning_rate": 4.320538210624418e-05,
-      "loss": 0.4359,
+      "loss": 0.525,
       "step": 7000
     },
     {
       "epoch": 0.4368592730661696,
-      "grad_norm": 5.1505231857299805,
+      "grad_norm": 4.532156944274902,
       "learning_rate": 4.271998291394843e-05,
-      "loss": 0.4356,
+      "loss": 0.5192,
       "step": 7500
     },
     {
       "epoch": 0.4659832246039143,
-      "grad_norm": 7.869897365570068,
+      "grad_norm": 6.108493804931641,
       "learning_rate": 4.223458372165269e-05,
-      "loss": 0.4297,
+      "loss": 0.52,
       "step": 8000
     },
     {
       "epoch": 0.4951071761416589,
-      "grad_norm": 5.874831676483154,
+      "grad_norm": 4.12423849105835,
       "learning_rate": 4.1749184529356947e-05,
-      "loss": 0.4227,
+      "loss": 0.4964,
       "step": 8500
     },
     {
       "epoch": 0.5242311276794035,
-      "grad_norm": 3.657247543334961,
+      "grad_norm": 5.869413375854492,
       "learning_rate": 4.12637853370612e-05,
-      "loss": 0.4304,
+      "loss": 0.5063,
       "step": 9000
     },
     {
       "epoch": 0.5533550792171482,
-      "grad_norm": 5.709639072418213,
+      "grad_norm": 5.091358184814453,
       "learning_rate": 4.0778386144765457e-05,
-      "loss": 0.4307,
+      "loss": 0.5067,
       "step": 9500
     },
     {
       "epoch": 0.5824790307548928,
-      "grad_norm": 7.34050178527832,
+      "grad_norm": 5.510972023010254,
       "learning_rate": 4.0292986952469715e-05,
-      "loss": 0.4188,
+      "loss": 0.5142,
       "step": 10000
     },
     {
       "epoch": 0.6116029822926374,
-      "grad_norm": 4.709152698516846,
+      "grad_norm": 4.2776031494140625,
       "learning_rate": 3.980758776017397e-05,
-      "loss": 0.4218,
+      "loss": 0.4946,
       "step": 10500
     },
     {
       "epoch": 0.6407269338303822,
-      "grad_norm": 4.877218246459961,
+      "grad_norm": 4.187364101409912,
       "learning_rate": 3.9322188567878225e-05,
-      "loss": 0.4139,
+      "loss": 0.494,
       "step": 11000
     },
     {
       "epoch": 0.6698508853681268,
-      "grad_norm": 5.153838634490967,
+      "grad_norm": 5.2663774490356445,
       "learning_rate": 3.883678937558248e-05,
-      "loss": 0.4214,
+      "loss": 0.487,
       "step": 11500
     },
     {
       "epoch": 0.6989748369058714,
-      "grad_norm": 6.150764465332031,
+      "grad_norm": 5.077462196350098,
       "learning_rate": 3.835139018328674e-05,
-      "loss": 0.4042,
+      "loss": 0.491,
       "step": 12000
     },
     {
       "epoch": 0.7280987884436161,
-      "grad_norm": 6.341733932495117,
+      "grad_norm": 5.158447265625,
       "learning_rate": 3.786599099099099e-05,
-      "loss": 0.4063,
+      "loss": 0.4943,
       "step": 12500
     },
     {
       "epoch": 0.7572227399813607,
-      "grad_norm": 4.536073684692383,
+      "grad_norm": 4.602211952209473,
       "learning_rate": 3.7380591798695245e-05,
-      "loss": 0.4016,
+      "loss": 0.4952,
       "step": 13000
     },
     {
       "epoch": 0.7863466915191053,
-      "grad_norm": 3.744781255722046,
+      "grad_norm": 5.792430877685547,
       "learning_rate": 3.68951926063995e-05,
-      "loss": 0.4055,
+      "loss": 0.4751,
       "step": 13500
     },
     {
       "epoch": 0.8154706430568499,
-      "grad_norm": 6.1853132247924805,
+      "grad_norm": 6.563935279846191,
       "learning_rate": 3.640979341410376e-05,
-      "loss": 0.3963,
+      "loss": 0.4735,
       "step": 14000
     },
     {
       "epoch": 0.8445945945945946,
-      "grad_norm": 7.477056503295898,
+      "grad_norm": 6.827334403991699,
       "learning_rate": 3.592439422180802e-05,
-      "loss": 0.4004,
+      "loss": 0.48,
       "step": 14500
     },
     {
       "epoch": 0.8737185461323392,
-      "grad_norm": 5.708552837371826,
+      "grad_norm": 3.2044920921325684,
       "learning_rate": 3.543899502951227e-05,
-      "loss": 0.3987,
+      "loss": 0.4797,
       "step": 15000
     },
     {
       "epoch": 0.9028424976700838,
-      "grad_norm": 4.550114154815674,
+      "grad_norm": 6.916011810302734,
       "learning_rate": 3.495359583721653e-05,
-      "loss": 0.4115,
+      "loss": 0.4763,
       "step": 15500
     },
     {
       "epoch": 0.9319664492078286,
-      "grad_norm": 5.300821781158447,
+      "grad_norm": 4.372725009918213,
       "learning_rate": 3.446819664492079e-05,
-      "loss": 0.3934,
+      "loss": 0.4662,
       "step": 16000
     },
     {
       "epoch": 0.9610904007455732,
-      "grad_norm": 4.5532402992248535,
+      "grad_norm": 6.041025638580322,
       "learning_rate": 3.398279745262504e-05,
-      "loss": 0.3985,
+      "loss": 0.4726,
       "step": 16500
     },
     {
       "epoch": 0.9902143522833178,
-      "grad_norm": 6.533886909484863,
+      "grad_norm": 5.349719524383545,
       "learning_rate": 3.34973982603293e-05,
-      "loss": 0.3987,
+      "loss": 0.4631,
       "step": 17000
     },
     {
       "epoch": 1.0193383038210624,
-      "grad_norm": 8.014822959899902,
+      "grad_norm": 8.095736503601074,
       "learning_rate": 3.301199906803355e-05,
-      "loss": 0.365,
+      "loss": 0.4372,
       "step": 17500
     },
     {
       "epoch": 1.048462255358807,
-      "grad_norm": 6.704738140106201,
+      "grad_norm": 6.805840969085693,
       "learning_rate": 3.252659987573781e-05,
-      "loss": 0.3469,
+      "loss": 0.432,
       "step": 18000
     },
     {
       "epoch": 1.0775862068965518,
-      "grad_norm": 4.034560680389404,
+      "grad_norm": 5.601711273193359,
       "learning_rate": 3.204120068344206e-05,
-      "loss": 0.3492,
+      "loss": 0.4127,
       "step": 18500
     },
     {
       "epoch": 1.1067101584342964,
-      "grad_norm": 4.0876665115356445,
+      "grad_norm": 5.928903102874756,
       "learning_rate": 3.155580149114632e-05,
-      "loss": 0.3514,
+      "loss": 0.434,
       "step": 19000
     },
     {
       "epoch": 1.135834109972041,
-      "grad_norm": 5.907679557800293,
+      "grad_norm": 12.023369789123535,
       "learning_rate": 3.1070402298850576e-05,
-      "loss": 0.3524,
+      "loss": 0.4229,
       "step": 19500
     },
     {
       "epoch": 1.1649580615097856,
-      "grad_norm": 4.2778520584106445,
+      "grad_norm": 7.795078754425049,
       "learning_rate": 3.0585003106554835e-05,
-      "loss": 0.3572,
+      "loss": 0.431,
       "step": 20000
     },
     {
       "epoch": 1.1940820130475303,
-      "grad_norm": 7.92435884475708,
+      "grad_norm": 6.34676456451416,
       "learning_rate": 3.009960391425909e-05,
-      "loss": 0.3504,
+      "loss": 0.4395,
       "step": 20500
     },
     {
       "epoch": 1.2232059645852749,
-      "grad_norm": 4.7413740158081055,
+      "grad_norm": 4.763006210327148,
       "learning_rate": 2.9614204721963345e-05,
-      "loss": 0.3529,
+      "loss": 0.4179,
       "step": 21000
     },
     {
       "epoch": 1.2523299161230197,
-      "grad_norm": 5.323243141174316,
+      "grad_norm": 6.222995758056641,
       "learning_rate": 2.9128805529667603e-05,
-      "loss": 0.3496,
+      "loss": 0.418,
       "step": 21500
     },
     {
       "epoch": 1.281453867660764,
-      "grad_norm": 6.30069637298584,
+      "grad_norm": 3.544293165206909,
       "learning_rate": 2.8643406337371858e-05,
-      "loss": 0.3496,
+      "loss": 0.4212,
       "step": 22000
     },
     {
       "epoch": 1.310577819198509,
-      "grad_norm": 6.415484428405762,
+      "grad_norm": 7.1324896812438965,
       "learning_rate": 2.815800714507611e-05,
-      "loss": 0.3497,
+      "loss": 0.4267,
       "step": 22500
     },
     {
       "epoch": 1.3397017707362535,
-      "grad_norm": 7.066019058227539,
+      "grad_norm": 4.74523401260376,
       "learning_rate": 2.7672607952780365e-05,
-      "loss": 0.3469,
+      "loss": 0.42,
       "step": 23000
     },
     {
       "epoch": 1.3688257222739981,
-      "grad_norm": 6.447554111480713,
+      "grad_norm": 4.924070835113525,
       "learning_rate": 2.7187208760484623e-05,
-      "loss": 0.3519,
+      "loss": 0.4211,
       "step": 23500
     },
     {
       "epoch": 1.3979496738117427,
-      "grad_norm": 6.272543430328369,
+      "grad_norm": 6.527987957000732,
       "learning_rate": 2.6701809568188878e-05,
-      "loss": 0.3416,
+      "loss": 0.4215,
       "step": 24000
     },
     {
       "epoch": 1.4270736253494873,
-      "grad_norm": 4.0889668464660645,
+      "grad_norm": 6.442528247833252,
       "learning_rate": 2.6216410375893136e-05,
-      "loss": 0.3432,
+      "loss": 0.4214,
       "step": 24500
     },
     {
       "epoch": 1.4561975768872322,
-      "grad_norm": 4.522316932678223,
+      "grad_norm": 8.452723503112793,
       "learning_rate": 2.573101118359739e-05,
-      "loss": 0.3457,
+      "loss": 0.4179,
       "step": 25000
     },
     {
       "epoch": 1.4853215284249768,
-      "grad_norm": 6.167509078979492,
+      "grad_norm": 8.910780906677246,
       "learning_rate": 2.524561199130165e-05,
-      "loss": 0.3486,
+      "loss": 0.4109,
       "step": 25500
     },
     {
       "epoch": 1.5144454799627214,
-      "grad_norm": 4.351974964141846,
+      "grad_norm": 7.066174507141113,
       "learning_rate": 2.47602127990059e-05,
-      "loss": 0.3534,
+      "loss": 0.4258,
       "step": 26000
     },
     {
       "epoch": 1.543569431500466,
-      "grad_norm": 6.637437343597412,
+      "grad_norm": 5.441757678985596,
       "learning_rate": 2.427481360671016e-05,
-      "loss": 0.3404,
+      "loss": 0.4122,
       "step": 26500
     },
     {
       "epoch": 1.5726933830382106,
-      "grad_norm": 8.26878833770752,
+      "grad_norm": 3.0415217876434326,
       "learning_rate": 2.3789414414414415e-05,
-      "loss": 0.3406,
+      "loss": 0.4205,
       "step": 27000
     },
     {
       "epoch": 1.6018173345759554,
-      "grad_norm": 3.1995205879211426,
+      "grad_norm": 5.961942672729492,
       "learning_rate": 2.3304015222118673e-05,
-      "loss": 0.3494,
+      "loss": 0.4115,
       "step": 27500
     },
     {
       "epoch": 1.6309412861136998,
-      "grad_norm": 8.005794525146484,
+      "grad_norm": 5.98954963684082,
       "learning_rate": 2.2818616029822928e-05,
-      "loss": 0.3471,
+      "loss": 0.4285,
       "step": 28000
     },
     {
       "epoch": 1.6600652376514446,
-      "grad_norm": 3.5494773387908936,
+      "grad_norm": 4.108989715576172,
       "learning_rate": 2.2333216837527183e-05,
-      "loss": 0.3422,
+      "loss": 0.4105,
       "step": 28500
     },
     {
       "epoch": 1.689189189189189,
-      "grad_norm": 6.0129780769348145,
+      "grad_norm": 4.233526706695557,
       "learning_rate": 2.1847817645231438e-05,
-      "loss": 0.352,
+      "loss": 0.4134,
       "step": 29000
     },
     {
       "epoch": 1.7183131407269339,
-      "grad_norm": 5.822954177856445,
+      "grad_norm": 3.5803985595703125,
       "learning_rate": 2.1362418452935696e-05,
-      "loss": 0.3439,
+      "loss": 0.4139,
       "step": 29500
     },
     {
       "epoch": 1.7474370922646785,
-      "grad_norm": 6.08833646774292,
+      "grad_norm": 5.007371425628662,
       "learning_rate": 2.087701926063995e-05,
-      "loss": 0.3458,
+      "loss": 0.4108,
       "step": 30000
     },
     {
       "epoch": 1.776561043802423,
-      "grad_norm": 4.251603603363037,
+      "grad_norm": 3.3870747089385986,
       "learning_rate": 2.039162006834421e-05,
-      "loss": 0.3445,
+      "loss": 0.4106,
       "step": 30500
     },
     {
       "epoch": 1.805684995340168,
-      "grad_norm": 3.8352158069610596,
+      "grad_norm": 6.3845109939575195,
       "learning_rate": 1.990622087604846e-05,
-      "loss": 0.3367,
+      "loss": 0.4122,
       "step": 31000
     },
     {
       "epoch": 1.8348089468779123,
-      "grad_norm": 3.8754615783691406,
+      "grad_norm": 4.464995861053467,
       "learning_rate": 1.942082168375272e-05,
-      "loss": 0.3318,
+      "loss": 0.4157,
       "step": 31500
     },
     {
       "epoch": 1.8639328984156571,
-      "grad_norm": 6.027359485626221,
+      "grad_norm": 4.677614688873291,
       "learning_rate": 1.8935422491456975e-05,
-      "loss": 0.3392,
+      "loss": 0.4035,
       "step": 32000
     },
     {
       "epoch": 1.8930568499534017,
-      "grad_norm": 3.559069871902466,
+      "grad_norm": 3.4746272563934326,
       "learning_rate": 1.8450023299161233e-05,
-      "loss": 0.3364,
+      "loss": 0.4122,
       "step": 32500
     },
     {
       "epoch": 1.9221808014911463,
-      "grad_norm": 4.697865962982178,
+      "grad_norm": 5.925264358520508,
       "learning_rate": 1.7964624106865488e-05,
-      "loss": 0.3403,
+      "loss": 0.4049,
       "step": 33000
     },
     {
       "epoch": 1.951304753028891,
-      "grad_norm": 9.035033226013184,
+      "grad_norm": 5.253298759460449,
       "learning_rate": 1.7479224914569743e-05,
-      "loss": 0.3414,
+      "loss": 0.4044,
       "step": 33500
     },
     {
       "epoch": 1.9804287045666356,
-      "grad_norm": 6.350938320159912,
+      "grad_norm": 6.474035263061523,
       "learning_rate": 1.6993825722273998e-05,
-      "loss": 0.3359,
+      "loss": 0.4122,
       "step": 34000
     },
     {
       "epoch": 2.0095526561043804,
-      "grad_norm": 3.888198137283325,
+      "grad_norm": 6.488181114196777,
       "learning_rate": 1.6508426529978256e-05,
-      "loss": 0.3263,
+      "loss": 0.3828,
       "step": 34500
     },
     {
       "epoch": 2.0386766076421248,
-      "grad_norm": 11.91583251953125,
+      "grad_norm": 7.312210559844971,
       "learning_rate": 1.602302733768251e-05,
-      "loss": 0.2927,
+      "loss": 0.3675,
       "step": 35000
     },
     {
       "epoch": 2.0678005591798696,
-      "grad_norm": 5.267765045166016,
+      "grad_norm": 4.655118942260742,
       "learning_rate": 1.5537628145386766e-05,
-      "loss": 0.3013,
+      "loss": 0.3558,
       "step": 35500
     },
     {
       "epoch": 2.096924510717614,
-      "grad_norm": 5.365530967712402,
+      "grad_norm": 7.048787593841553,
       "learning_rate": 1.5052228953091021e-05,
-      "loss": 0.299,
+      "loss": 0.3596,
       "step": 36000
     },
     {
       "epoch": 2.126048462255359,
-      "grad_norm": 3.0507187843322754,
+      "grad_norm": 6.153940677642822,
       "learning_rate": 1.4566829760795278e-05,
-      "loss": 0.3068,
+      "loss": 0.3729,
       "step": 36500
     },
     {
       "epoch": 2.1551724137931036,
-      "grad_norm": 3.4967715740203857,
+      "grad_norm": 7.460168361663818,
       "learning_rate": 1.4081430568499535e-05,
-      "loss": 0.3012,
+      "loss": 0.3689,
       "step": 37000
     },
     {
       "epoch": 2.184296365330848,
-      "grad_norm": 7.34511137008667,
+      "grad_norm": 4.816644191741943,
       "learning_rate": 1.3596031376203791e-05,
-      "loss": 0.2981,
+      "loss": 0.3644,
       "step": 37500
     },
     {
       "epoch": 2.213420316868593,
-      "grad_norm": 4.908951759338379,
+      "grad_norm": 8.305159568786621,
       "learning_rate": 1.3110632183908048e-05,
-      "loss": 0.3,
+      "loss": 0.3664,
       "step": 38000
     },
     {
       "epoch": 2.2425442684063372,
-      "grad_norm": 4.048547744750977,
+      "grad_norm": 5.917420864105225,
       "learning_rate": 1.2625232991612301e-05,
-      "loss": 0.2918,
+      "loss": 0.3733,
       "step": 38500
     },
     {
       "epoch": 2.271668219944082,
-      "grad_norm": 6.06612491607666,
+      "grad_norm": 5.741204738616943,
       "learning_rate": 1.2139833799316558e-05,
-      "loss": 0.2901,
+      "loss": 0.3677,
       "step": 39000
     },
     {
       "epoch": 2.3007921714818265,
-      "grad_norm": 3.469628095626831,
+      "grad_norm": 3.565974712371826,
       "learning_rate": 1.1654434607020815e-05,
-      "loss": 0.304,
+      "loss": 0.3617,
       "step": 39500
     },
     {
       "epoch": 2.3299161230195713,
-      "grad_norm": 8.2776460647583,
+      "grad_norm": 8.44483757019043,
       "learning_rate": 1.1169035414725071e-05,
-      "loss": 0.2986,
+      "loss": 0.3661,
       "step": 40000
     },
     {
       "epoch": 2.359040074557316,
-      "grad_norm": 6.029278755187988,
+      "grad_norm": 6.406001091003418,
       "learning_rate": 1.0683636222429326e-05,
-      "loss": 0.2981,
+      "loss": 0.3613,
       "step": 40500
     },
     {
       "epoch": 2.3881640260950605,
-      "grad_norm": 5.872582912445068,
+      "grad_norm": 6.529003143310547,
       "learning_rate": 1.0198237030133583e-05,
-      "loss": 0.298,
+      "loss": 0.3685,
       "step": 41000
     },
     {
       "epoch": 2.4172879776328053,
-      "grad_norm": 3.5587821006774902,
+      "grad_norm": 6.799564838409424,
       "learning_rate": 9.712837837837838e-06,
-      "loss": 0.2836,
+      "loss": 0.3571,
       "step": 41500
     },
     {
       "epoch": 2.4464119291705497,
-      "grad_norm": 3.2673676013946533,
+      "grad_norm": 8.416892051696777,
       "learning_rate": 9.227438645542095e-06,
-      "loss": 0.293,
+      "loss": 0.3613,
       "step": 42000
     },
     {
       "epoch": 2.4755358807082946,
-      "grad_norm": 2.4192936420440674,
+      "grad_norm": 6.679762363433838,
       "learning_rate": 8.742039453246351e-06,
-      "loss": 0.3051,
+      "loss": 0.354,
       "step": 42500
     },
     {
       "epoch": 2.5046598322460394,
-      "grad_norm": 2.684082269668579,
+      "grad_norm": 6.462404727935791,
       "learning_rate": 8.256640260950606e-06,
-      "loss": 0.2865,
+      "loss": 0.3608,
       "step": 43000
     },
     {
       "epoch": 2.5337837837837838,
-      "grad_norm": 7.754036903381348,
+      "grad_norm": 8.391572952270508,
       "learning_rate": 7.771241068654863e-06,
-      "loss": 0.2844,
+      "loss": 0.3714,
       "step": 43500
     },
     {
       "epoch": 2.562907735321528,
-      "grad_norm": 7.571073532104492,
+      "grad_norm": 6.29732084274292,
       "learning_rate": 7.285841876359117e-06,
-      "loss": 0.2867,
+      "loss": 0.3645,
       "step": 44000
     },
     {
       "epoch": 2.592031686859273,
-      "grad_norm": 4.301662921905518,
+      "grad_norm": 5.085122585296631,
       "learning_rate": 6.800442684063374e-06,
-      "loss": 0.2909,
+      "loss": 0.3555,
       "step": 44500
     },
     {
       "epoch": 2.621155638397018,
-      "grad_norm": 9.151103019714355,
+      "grad_norm": 6.8417534828186035,
       "learning_rate": 6.31504349176763e-06,
-      "loss": 0.2953,
+      "loss": 0.3609,
       "step": 45000
     },
     {
       "epoch": 2.650279589934762,
-      "grad_norm": 2.850679636001587,
+      "grad_norm": 2.261542797088623,
       "learning_rate": 5.829644299471886e-06,
-      "loss": 0.2953,
+      "loss": 0.3633,
       "step": 45500
     },
     {
       "epoch": 2.679403541472507,
-      "grad_norm": 6.948458194732666,
+      "grad_norm": 5.810998916625977,
       "learning_rate": 5.344245107176142e-06,
-      "loss": 0.2959,
+      "loss": 0.3658,
       "step": 46000
     },
     {
       "epoch": 2.7085274930102514,
-      "grad_norm": 2.6653831005096436,
+      "grad_norm": 4.694461345672607,
       "learning_rate": 4.858845914880398e-06,
-      "loss": 0.303,
+      "loss": 0.3551,
       "step": 46500
     },
     {
       "epoch": 2.7376514445479962,
-      "grad_norm": 6.758817195892334,
+      "grad_norm": 5.589244842529297,
       "learning_rate": 4.373446722584654e-06,
-      "loss": 0.2956,
+      "loss": 0.3526,
       "step": 47000
     },
     {
       "epoch": 2.766775396085741,
-      "grad_norm": 4.432290077209473,
+      "grad_norm": 9.656843185424805,
       "learning_rate": 3.8880475302889095e-06,
-      "loss": 0.2906,
+      "loss": 0.3529,
       "step": 47500
     },
     {
       "epoch": 2.7958993476234855,
-      "grad_norm": 2.109107255935669,
+      "grad_norm": 4.115970611572266,
       "learning_rate": 3.402648337993166e-06,
-      "loss": 0.2934,
+      "loss": 0.3593,
       "step": 48000
     },
     {
       "epoch": 2.8250232991612303,
-      "grad_norm": 7.6626715660095215,
+      "grad_norm": 4.918366432189941,
       "learning_rate": 2.917249145697422e-06,
-      "loss": 0.2979,
+      "loss": 0.3602,
       "step": 48500
     },
     {
       "epoch": 2.8541472506989747,
-      "grad_norm": 4.676031112670898,
+      "grad_norm": 4.475345134735107,
       "learning_rate": 2.431849953401678e-06,
-      "loss": 0.2898,
+      "loss": 0.3586,
       "step": 49000
     },
     {
       "epoch": 2.8832712022367195,
-      "grad_norm": 7.312727928161621,
+      "grad_norm": 5.438355445861816,
       "learning_rate": 1.9464507611059337e-06,
-      "loss": 0.2887,
+      "loss": 0.3619,
       "step": 49500
     },
     {
       "epoch": 2.9123951537744643,
-      "grad_norm": 7.303893089294434,
+      "grad_norm": 7.748974323272705,
       "learning_rate": 1.4610515688101895e-06,
-      "loss": 0.2888,
+      "loss": 0.3612,
       "step": 50000
     },
     {
       "epoch": 2.9415191053122087,
-      "grad_norm": 7.751052379608154,
+      "grad_norm": 6.27733850479126,
       "learning_rate": 9.756523765144455e-07,
-      "loss": 0.2909,
+      "loss": 0.3609,
       "step": 50500
     },
     {
       "epoch": 2.9706430568499536,
-      "grad_norm": 6.1867756843566895,
+      "grad_norm": 7.872511386871338,
       "learning_rate": 4.902531842187015e-07,
-      "loss": 0.2964,
+      "loss": 0.3565,
       "step": 51000
     },
     {
       "epoch": 2.999767008387698,
-      "grad_norm": 6.950267314910889,
+      "grad_norm": 6.070102691650391,
       "learning_rate": 4.85399192295744e-09,
-      "loss": 0.2921,
+      "loss": 0.3606,
       "step": 51500
     }
   ],
@@ -748,7 +748,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.2121938098975232e+16,
+  "total_flos": 1.2121960164297984e+16,
   "train_batch_size": 32,
   "trial_name": null,
   "trial_params": null

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7716ebf599292f09f6ed7bc5ff274b4125ac052ae0418ba0b28651779606bf59
+oid sha256:f3e4c250ddedede950a8a80e8efc48ef5514393ffdb76e81cb7f9cdd0c1cfde5
 size 5841