Add checkpoint-11200

checkpoints/checkpoint-11200/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:33181192265ddb018675cce4ff0394c4745c28f62ed05859981bb2691fa55209
+oid sha256:66784a67a5984edb149f03db4ce5d207ce126c69cbc87032c24fccd8bb56563a
 size 4724960

checkpoints/checkpoint-11200/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8b22cc0170067d26011169d82830aeb78faf1f63e506cee5d85c540e0f4dccda
+oid sha256:536dcda99fcf7fb164cf62fcf6aa9b683d609a71e5deaea9a88a98e100e9d705
 size 2578187

checkpoints/checkpoint-11200/trainer_state.json

@@ -7431,422 +7431,422 @@
     },
     {
       "epoch": 1.5146507726899605,
-      "grad_norm": 1.098069429397583,
+      "grad_norm": 1.0981698036193848,
       "learning_rate": 9.913830040466556e-05,
-      "loss": 2.008710479736328,
+      "loss": 2.0087039947509764,
       "step": 10610
     },
     {
       "epoch": 1.5160783753881295,
-      "grad_norm": 1.5518226623535156,
+      "grad_norm": 1.5510938167572021,
       "learning_rate": 9.904308497976673e-05,
-      "loss": 2.1474123001098633,
+      "loss": 2.1474151611328125,
       "step": 10620
     },
     {
       "epoch": 1.5175059780862985,
-      "grad_norm": 1.2088004350662231,
+      "grad_norm": 1.2088773250579834,
       "learning_rate": 9.894786955486789e-05,
-      "loss": 2.0725215911865233,
+      "loss": 2.072478103637695,
       "step": 10630
     },
     {
       "epoch": 1.5189335807844677,
-      "grad_norm": 1.0846376419067383,
+      "grad_norm": 1.0845247507095337,
       "learning_rate": 9.885265412996906e-05,
-      "loss": 2.015618896484375,
+      "loss": 2.015526008605957,
       "step": 10640
     },
     {
       "epoch": 1.520361183482637,
-      "grad_norm": 1.201206088066101,
+      "grad_norm": 1.2015031576156616,
       "learning_rate": 9.875743870507023e-05,
-      "loss": 2.106952667236328,
+      "loss": 2.1070165634155273,
       "step": 10650
     },
     {
       "epoch": 1.521788786180806,
-      "grad_norm": 1.0636956691741943,
+      "grad_norm": 1.062964677810669,
       "learning_rate": 9.86622232801714e-05,
-      "loss": 2.0599929809570314,
+      "loss": 2.0601083755493166,
       "step": 10660
     },
     {
       "epoch": 1.523216388878975,
-      "grad_norm": 1.1827208995819092,
+      "grad_norm": 1.1827353239059448,
       "learning_rate": 9.856700785527256e-05,
-      "loss": 2.2304359436035157,
+      "loss": 2.230342674255371,
       "step": 10670
     },
     {
       "epoch": 1.5246439915771441,
-      "grad_norm": 1.2152652740478516,
+      "grad_norm": 1.2147736549377441,
       "learning_rate": 9.847179243037372e-05,
-      "loss": 2.083365631103516,
+      "loss": 2.083346366882324,
       "step": 10680
     },
     {
       "epoch": 1.5260715942753131,
-      "grad_norm": 1.0766453742980957,
+      "grad_norm": 1.0761597156524658,
       "learning_rate": 9.837657700547489e-05,
-      "loss": 2.119212341308594,
+      "loss": 2.1192899703979493,
       "step": 10690
     },
     {
       "epoch": 1.5274991969734821,
-      "grad_norm": 1.2730481624603271,
+      "grad_norm": 1.2726013660430908,
       "learning_rate": 9.828136158057606e-05,
-      "loss": 2.1518115997314453,
+      "loss": 2.1517145156860353,
       "step": 10700
     },
     {
       "epoch": 1.5289267996716513,
-      "grad_norm": 1.97091805934906,
+      "grad_norm": 1.9693447351455688,
       "learning_rate": 9.818614615567723e-05,
-      "loss": 1.9795692443847657,
+      "loss": 1.97952880859375,
       "step": 10710
     },
     {
       "epoch": 1.5303544023698206,
-      "grad_norm": 1.3302141427993774,
+      "grad_norm": 1.3304651975631714,
       "learning_rate": 9.809093073077838e-05,
-      "loss": 1.9785017013549804,
+      "loss": 1.9785858154296876,
       "step": 10720
     },
     {
       "epoch": 1.5317820050679896,
-      "grad_norm": 1.2388334274291992,
+      "grad_norm": 1.2396725416183472,
       "learning_rate": 9.799571530587955e-05,
-      "loss": 2.134817695617676,
+      "loss": 2.1346342086791994,
       "step": 10730
     },
     {
       "epoch": 1.5332096077661586,
-      "grad_norm": 1.4161068201065063,
+      "grad_norm": 1.413583517074585,
       "learning_rate": 9.790049988098072e-05,
-      "loss": 1.9905515670776368,
+      "loss": 1.9904579162597655,
       "step": 10740
     },
     {
       "epoch": 1.5346372104643278,
-      "grad_norm": 1.2048768997192383,
+      "grad_norm": 1.2050212621688843,
       "learning_rate": 9.780528445608189e-05,
-      "loss": 2.0450416564941407,
+      "loss": 2.045025444030762,
       "step": 10750
     },
     {
       "epoch": 1.536064813162497,
-      "grad_norm": 1.3425582647323608,
+      "grad_norm": 1.3423675298690796,
       "learning_rate": 9.771006903118306e-05,
-      "loss": 1.923790740966797,
+      "loss": 1.9239255905151367,
       "step": 10760
     },
     {
       "epoch": 1.537492415860666,
-      "grad_norm": 1.2138093709945679,
+      "grad_norm": 1.2138278484344482,
       "learning_rate": 9.761485360628421e-05,
-      "loss": 2.0516876220703124,
+      "loss": 2.0515163421630858,
       "step": 10770
     },
     {
       "epoch": 1.538920018558835,
-      "grad_norm": 1.098336100578308,
+      "grad_norm": 1.0984469652175903,
       "learning_rate": 9.751963818138538e-05,
-      "loss": 1.9940862655639648,
+      "loss": 1.9942876815795898,
       "step": 10780
     },
     {
       "epoch": 1.5403476212570042,
-      "grad_norm": 1.2795876264572144,
+      "grad_norm": 1.2791671752929688,
       "learning_rate": 9.742442275648655e-05,
-      "loss": 2.000428581237793,
+      "loss": 2.0003101348876955,
       "step": 10790
     },
     {
       "epoch": 1.5417752239551734,
-      "grad_norm": 1.0763850212097168,
+      "grad_norm": 1.0745598077774048,
       "learning_rate": 9.732920733158772e-05,
-      "loss": 1.9592803955078124,
+      "loss": 1.9592748641967774,
       "step": 10800
     },
     {
       "epoch": 1.5432028266533424,
-      "grad_norm": 1.1885327100753784,
+      "grad_norm": 1.1889386177062988,
       "learning_rate": 9.723399190668889e-05,
-      "loss": 1.9911989212036132,
+      "loss": 1.9912334442138673,
       "step": 10810
     },
     {
       "epoch": 1.5446304293515114,
-      "grad_norm": 1.2224466800689697,
+      "grad_norm": 1.2227346897125244,
       "learning_rate": 9.713877648179006e-05,
-      "loss": 2.0401016235351563,
+      "loss": 2.04000186920166,
       "step": 10820
     },
     {
       "epoch": 1.5460580320496806,
-      "grad_norm": 1.1554738283157349,
+      "grad_norm": 1.1555755138397217,
       "learning_rate": 9.704356105689123e-05,
-      "loss": 2.036018180847168,
+      "loss": 2.0359256744384764,
       "step": 10830
     },
     {
       "epoch": 1.5474856347478498,
-      "grad_norm": 1.1402208805084229,
+      "grad_norm": 1.141852617263794,
       "learning_rate": 9.694834563199238e-05,
-      "loss": 2.0756540298461914,
+      "loss": 2.0757726669311523,
       "step": 10840
     },
     {
       "epoch": 1.5489132374460186,
-      "grad_norm": 1.1714133024215698,
+      "grad_norm": 1.1708478927612305,
       "learning_rate": 9.685313020709355e-05,
-      "loss": 2.0774347305297853,
+      "loss": 2.077558708190918,
       "step": 10850
     },
     {
       "epoch": 1.5503408401441878,
-      "grad_norm": 1.3246486186981201,
+      "grad_norm": 1.3248668909072876,
       "learning_rate": 9.675791478219472e-05,
-      "loss": 2.141210174560547,
+      "loss": 2.1411231994628905,
       "step": 10860
     },
     {
       "epoch": 1.551768442842357,
-      "grad_norm": 1.1436152458190918,
+      "grad_norm": 1.1437745094299316,
       "learning_rate": 9.666269935729589e-05,
-      "loss": 2.2358598709106445,
+      "loss": 2.2358665466308594,
       "step": 10870
     },
     {
       "epoch": 1.553196045540526,
-      "grad_norm": 1.4468668699264526,
+      "grad_norm": 1.4474693536758423,
       "learning_rate": 9.656748393239706e-05,
-      "loss": 2.168443298339844,
+      "loss": 2.1683933258056642,
       "step": 10880
     },
     {
       "epoch": 1.554623648238695,
-      "grad_norm": 1.4564536809921265,
+      "grad_norm": 1.4569262266159058,
       "learning_rate": 9.647226850749823e-05,
-      "loss": 2.087565612792969,
+      "loss": 2.0875593185424806,
       "step": 10890
     },
     {
       "epoch": 1.5560512509368642,
-      "grad_norm": 1.234102487564087,
+      "grad_norm": 1.2359139919281006,
       "learning_rate": 9.63770530825994e-05,
-      "loss": 2.0447998046875,
+      "loss": 2.0447929382324217,
       "step": 10900
     },
     {
       "epoch": 1.5574788536350335,
-      "grad_norm": 1.1943862438201904,
+      "grad_norm": 1.1910909414291382,
       "learning_rate": 9.628183765770056e-05,
-      "loss": 1.9503068923950195,
+      "loss": 1.9503170013427735,
       "step": 10910
     },
     {
       "epoch": 1.5589064563332025,
-      "grad_norm": 1.2671523094177246,
+      "grad_norm": 1.267309546470642,
       "learning_rate": 9.618662223280172e-05,
-      "loss": 1.9227149963378907,
+      "loss": 1.9226907730102538,
       "step": 10920
     },
     {
       "epoch": 1.5603340590313715,
-      "grad_norm": 1.057041049003601,
+      "grad_norm": 1.0570237636566162,
       "learning_rate": 9.609140680790289e-05,
-      "loss": 2.067658233642578,
+      "loss": 2.0675920486450194,
       "step": 10930
     },
     {
       "epoch": 1.5617616617295407,
-      "grad_norm": 1.0804142951965332,
+      "grad_norm": 1.0798622369766235,
       "learning_rate": 9.599619138300406e-05,
-      "loss": 2.1086441040039063,
+      "loss": 2.1086353302001952,
       "step": 10940
     },
     {
       "epoch": 1.56318926442771,
-      "grad_norm": 0.9749084115028381,
+      "grad_norm": 0.9747142791748047,
       "learning_rate": 9.590097595810523e-05,
-      "loss": 2.0530927658081053,
+      "loss": 2.0529823303222656,
       "step": 10950
     },
     {
       "epoch": 1.5646168671258789,
-      "grad_norm": 1.1373926401138306,
+      "grad_norm": 1.1378403902053833,
       "learning_rate": 9.580576053320638e-05,
-      "loss": 2.195412254333496,
+      "loss": 2.1953744888305664,
       "step": 10960
     },
     {
       "epoch": 1.5660444698240479,
-      "grad_norm": 1.4987021684646606,
+      "grad_norm": 1.4991871118545532,
       "learning_rate": 9.571054510830755e-05,
-      "loss": 1.9444866180419922,
+      "loss": 1.94462890625,
       "step": 10970
     },
     {
       "epoch": 1.567472072522217,
-      "grad_norm": 1.005608081817627,
+      "grad_norm": 1.005947232246399,
       "learning_rate": 9.561532968340872e-05,
-      "loss": 2.1251079559326174,
+      "loss": 2.1250656127929686,
       "step": 10980
     },
     {
       "epoch": 1.5688996752203863,
-      "grad_norm": 1.2665334939956665,
+      "grad_norm": 1.265784502029419,
       "learning_rate": 9.552011425850989e-05,
-      "loss": 1.9724187850952148,
+      "loss": 1.9723976135253907,
       "step": 10990
     },
     {
       "epoch": 1.5703272779185553,
-      "grad_norm": 1.7481650114059448,
+      "grad_norm": 1.7473640441894531,
       "learning_rate": 9.542489883361106e-05,
       "loss": 2.0439002990722654,
       "step": 11000
     },
     {
       "epoch": 1.5717548806167243,
-      "grad_norm": 1.3757065534591675,
+      "grad_norm": 1.376124620437622,
       "learning_rate": 9.532968340871221e-05,
-      "loss": 2.13690299987793,
+      "loss": 2.1368425369262694,
       "step": 11010
     },
     {
       "epoch": 1.5731824833148935,
-      "grad_norm": 1.0686981678009033,
+      "grad_norm": 1.0686304569244385,
       "learning_rate": 9.523446798381338e-05,
-      "loss": 2.0076725006103517,
+      "loss": 2.0076366424560548,
       "step": 11020
     },
     {
       "epoch": 1.5746100860130625,
-      "grad_norm": 1.4125736951828003,
+      "grad_norm": 1.4122185707092285,
       "learning_rate": 9.513925255891455e-05,
-      "loss": 2.0957643508911135,
+      "loss": 2.095813751220703,
       "step": 11030
     },
     {
       "epoch": 1.5760376887112315,
-      "grad_norm": 1.2688685655593872,
+      "grad_norm": 1.26949143409729,
       "learning_rate": 9.504403713401572e-05,
-      "loss": 2.185018539428711,
+      "loss": 2.1850494384765624,
       "step": 11040
     },
     {
       "epoch": 1.5774652914094007,
-      "grad_norm": 1.1199735403060913,
+      "grad_norm": 1.120142936706543,
       "learning_rate": 9.494882170911689e-05,
-      "loss": 2.0660390853881836,
+      "loss": 2.0661577224731444,
       "step": 11050
     },
     {
       "epoch": 1.57889289410757,
-      "grad_norm": 1.2410614490509033,
+      "grad_norm": 1.2405672073364258,
       "learning_rate": 9.485360628421804e-05,
-      "loss": 1.9966764450073242,
+      "loss": 1.9966974258422852,
       "step": 11060
     },
     {
       "epoch": 1.580320496805739,
-      "grad_norm": 1.160320520401001,
+      "grad_norm": 1.159519076347351,
       "learning_rate": 9.475839085931921e-05,
-      "loss": 1.9303186416625977,
+      "loss": 1.9302923202514648,
       "step": 11070
     },
     {
       "epoch": 1.581748099503908,
-      "grad_norm": 1.2862969636917114,
+      "grad_norm": 1.2723621129989624,
       "learning_rate": 9.466317543442038e-05,
-      "loss": 1.8376413345336915,
+      "loss": 1.837651252746582,
       "step": 11080
     },
     {
       "epoch": 1.5831757022020772,
-      "grad_norm": 1.3544589281082153,
+      "grad_norm": 1.354519009590149,
       "learning_rate": 9.456796000952155e-05,
-      "loss": 2.001028060913086,
+      "loss": 2.001264953613281,
       "step": 11090
     },
     {
       "epoch": 1.5846033049002464,
-      "grad_norm": 1.2006914615631104,
+      "grad_norm": 1.2013628482818604,
       "learning_rate": 9.447274458462272e-05,
-      "loss": 2.006683921813965,
+      "loss": 2.0066423416137695,
       "step": 11100
     },
     {
       "epoch": 1.5860309075984154,
-      "grad_norm": 1.1594748497009277,
+      "grad_norm": 1.159398078918457,
       "learning_rate": 9.437752915972387e-05,
-      "loss": 2.008784294128418,
+      "loss": 2.0087814331054688,
       "step": 11110
     },
     {
       "epoch": 1.5874585102965844,
-      "grad_norm": 1.259328842163086,
+      "grad_norm": 1.2587201595306396,
       "learning_rate": 9.428231373482504e-05,
-      "loss": 1.977517318725586,
+      "loss": 1.9774051666259767,
       "step": 11120
     },
     {
       "epoch": 1.5888861129947536,
-      "grad_norm": 1.238774061203003,
+      "grad_norm": 1.2386575937271118,
       "learning_rate": 9.418709830992621e-05,
-      "loss": 2.0367507934570312,
+      "loss": 2.03675537109375,
       "step": 11130
     },
     {
       "epoch": 1.5903137156929228,
-      "grad_norm": 1.1448348760604858,
+      "grad_norm": 1.1446408033370972,
       "learning_rate": 9.409188288502738e-05,
-      "loss": 2.0270627975463866,
+      "loss": 2.0270097732543944,
       "step": 11140
     },
     {
       "epoch": 1.5917413183910918,
-      "grad_norm": 1.091199278831482,
+      "grad_norm": 1.090232491493225,
       "learning_rate": 9.399666746012853e-05,
-      "loss": 2.1191539764404297,
+      "loss": 2.1192419052124025,
       "step": 11150
     },
     {
       "epoch": 1.5931689210892608,
-      "grad_norm": 1.136646032333374,
+      "grad_norm": 1.137207269668579,
       "learning_rate": 9.39014520352297e-05,
-      "loss": 2.0415071487426757,
+      "loss": 2.041428565979004,
       "step": 11160
     },
     {
       "epoch": 1.59459652378743,
-      "grad_norm": 1.2094829082489014,
+      "grad_norm": 1.2093437910079956,
       "learning_rate": 9.380623661033087e-05,
-      "loss": 1.9333257675170898,
+      "loss": 1.9331945419311523,
       "step": 11170
     },
     {
       "epoch": 1.5960241264855992,
-      "grad_norm": 1.2323769330978394,
+      "grad_norm": 1.2326594591140747,
       "learning_rate": 9.371102118543204e-05,
-      "loss": 2.212533950805664,
+      "loss": 2.212518882751465,
       "step": 11180
     },
     {
       "epoch": 1.5974517291837682,
-      "grad_norm": 1.2837002277374268,
+      "grad_norm": 1.280666708946228,
       "learning_rate": 9.361580576053321e-05,
-      "loss": 2.067727470397949,
+      "loss": 2.067685127258301,
       "step": 11190
     },
     {
       "epoch": 1.5988793318819372,
-      "grad_norm": 1.3468828201293945,
+      "grad_norm": 1.3486379384994507,
       "learning_rate": 9.352059033563438e-05,
-      "loss": 1.9625717163085938,
+      "loss": 1.962606430053711,
       "step": 11200
     }
   ],