update file

adapter_config.json

@@ -20,12 +20,12 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "encoder.block.0.layer.0.SelfAttention.q",
+    "decoder.block.0.layer.0.SelfAttention.v",
     "decoder.block.0.layer.0.SelfAttention.q",
+    "encoder.block.0.layer.0.SelfAttention.k",
     "encoder.block.0.layer.0.SelfAttention.v",
     "decoder.block.0.layer.0.SelfAttention.k",
-    "encoder.block.0.layer.0.SelfAttention.k",
-    "decoder.block.0.layer.0.SelfAttention.v"
+    "encoder.block.0.layer.0.SelfAttention.q"
   ],
   "task_type": "SEQ2SEQ_LM",
   "use_dora": false,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:65888f0de5ec92baef29077564e77fedd9844936f582be94e36eaf7a08fa51a1
+oid sha256:7804b95cebed4f6f8db8baaf5f979f72a108aa34b20499c4341cebfb6f67cfe2
 size 1181328

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8d99aa47f1373e797a786636450d11ce1bedf1e455ecc4f1adb13558d091ec34
+oid sha256:f872f390b056e639ab5bbf69899d835411f08590b0b39576bf331a5b98b96c9b
 size 2366982

rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:888d142298de620b1c05abd3048212502b637c5c30e2c251c2040a31bcf632ab
+oid sha256:00eadd813b6ef57e893ab6475e3a551bc56d211d5c977084ebbb2c8bcff789b1
 size 14244

scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:313d9e6a2e2433f67a0da276d33bd1e89d9178f1f5f7d57255849b717b7138ef
+oid sha256:d2400a287bc72199e423eecabbc1d481a4ec8fef2fda26a5cc179ecdd2357caf
 size 1064

trainer_state.json

@@ -3,616 +3,1218 @@
   "best_model_checkpoint": null,
   "epoch": 4.0,
   "eval_steps": 500,
-  "global_step": 8668,
+  "global_step": 17216,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.046146746654360866,
-      "grad_norm": 0.21824534237384796,
-      "learning_rate": 2.23760092272203e-06,
-      "loss": 3.136,
+      "epoch": 0.023234200743494422,
+      "grad_norm": 0.33875614404678345,
+      "learning_rate": 1.1149825783972125e-06,
+      "loss": 3.2044,
       "step": 100
     },
     {
-      "epoch": 0.09229349330872173,
-      "grad_norm": 0.2850794494152069,
-      "learning_rate": 4.544405997693195e-06,
-      "loss": 3.1073,
+      "epoch": 0.046468401486988845,
+      "grad_norm": 0.09059225022792816,
+      "learning_rate": 2.2764227642276426e-06,
+      "loss": 3.1582,
       "step": 200
     },
     {
-      "epoch": 0.1384402399630826,
-      "grad_norm": 0.36149144172668457,
-      "learning_rate": 6.828143021914648e-06,
-      "loss": 3.1035,
+      "epoch": 0.06970260223048327,
+      "grad_norm": 0.24917162954807281,
+      "learning_rate": 3.4378629500580724e-06,
+      "loss": 3.1608,
       "step": 300
     },
     {
-      "epoch": 0.18458698661744347,
-      "grad_norm": 0.5060057640075684,
-      "learning_rate": 9.134948096885815e-06,
-      "loss": 3.106,
+      "epoch": 0.09293680297397769,
+      "grad_norm": 0.465605229139328,
+      "learning_rate": 4.599303135888502e-06,
+      "loss": 3.1981,
       "step": 400
     },
     {
-      "epoch": 0.23073373327180433,
-      "grad_norm": 0.3851727247238159,
-      "learning_rate": 1.144175317185698e-05,
-      "loss": 3.0204,
+      "epoch": 0.11617100371747212,
+      "grad_norm": 0.27495619654655457,
+      "learning_rate": 5.7607433217189324e-06,
+      "loss": 3.1815,
       "step": 500
     },
     {
-      "epoch": 0.2768804799261652,
-      "grad_norm": 0.30784064531326294,
-      "learning_rate": 1.3748558246828143e-05,
-      "loss": 2.9708,
+      "epoch": 0.13940520446096655,
+      "grad_norm": 0.19188807904720306,
+      "learning_rate": 6.922183507549362e-06,
+      "loss": 3.1294,
       "step": 600
     },
     {
-      "epoch": 0.3230272265805261,
-      "grad_norm": 0.6206746697425842,
-      "learning_rate": 1.605536332179931e-05,
-      "loss": 2.8294,
+      "epoch": 0.16263940520446096,
+      "grad_norm": 0.5246957540512085,
+      "learning_rate": 8.083623693379791e-06,
+      "loss": 3.0677,
       "step": 700
     },
     {
-      "epoch": 0.36917397323488693,
-      "grad_norm": 0.47670331597328186,
-      "learning_rate": 1.8362168396770474e-05,
-      "loss": 2.6824,
+      "epoch": 0.18587360594795538,
+      "grad_norm": 0.258408784866333,
+      "learning_rate": 9.24506387921022e-06,
+      "loss": 2.9832,
       "step": 800
     },
     {
-      "epoch": 0.4153207198892478,
-      "grad_norm": 0.4056473672389984,
-      "learning_rate": 1.9925650557620818e-05,
-      "loss": 2.4481,
+      "epoch": 0.20910780669144982,
+      "grad_norm": 0.31014084815979004,
+      "learning_rate": 1.0406504065040652e-05,
+      "loss": 2.9743,
       "step": 900
     },
     {
-      "epoch": 0.46146746654360865,
-      "grad_norm": 0.8275523781776428,
-      "learning_rate": 1.96692731701064e-05,
-      "loss": 2.0463,
+      "epoch": 0.23234200743494424,
+      "grad_norm": 0.4873325824737549,
+      "learning_rate": 1.1567944250871081e-05,
+      "loss": 2.8721,
       "step": 1000
     },
     {
-      "epoch": 0.5076142131979695,
-      "grad_norm": 0.5123036503791809,
-      "learning_rate": 1.9412895782591976e-05,
-      "loss": 1.9237,
+      "epoch": 0.2555762081784387,
+      "grad_norm": 0.7442412972450256,
+      "learning_rate": 1.272938443670151e-05,
+      "loss": 2.7949,
       "step": 1100
     },
     {
-      "epoch": 0.5537609598523304,
-      "grad_norm": 0.7184427976608276,
-      "learning_rate": 1.9156518395077554e-05,
-      "loss": 1.881,
+      "epoch": 0.2788104089219331,
+      "grad_norm": 0.6129536628723145,
+      "learning_rate": 1.389082462253194e-05,
+      "loss": 2.6381,
       "step": 1200
     },
     {
-      "epoch": 0.5999077065066912,
-      "grad_norm": 1.0534940958023071,
-      "learning_rate": 1.8900141007563134e-05,
-      "loss": 1.8219,
+      "epoch": 0.3020446096654275,
+      "grad_norm": 0.5687291026115417,
+      "learning_rate": 1.5052264808362371e-05,
+      "loss": 2.4031,
       "step": 1300
     },
     {
-      "epoch": 0.6460544531610521,
-      "grad_norm": 0.9063017964363098,
-      "learning_rate": 1.8646327393923858e-05,
-      "loss": 1.7868,
+      "epoch": 0.3252788104089219,
+      "grad_norm": 0.6154528856277466,
+      "learning_rate": 1.62137049941928e-05,
+      "loss": 2.107,
       "step": 1400
     },
     {
-      "epoch": 0.6922011998154131,
-      "grad_norm": 0.7229586839675903,
-      "learning_rate": 1.8389950006409436e-05,
-      "loss": 1.7581,
+      "epoch": 0.34851301115241634,
+      "grad_norm": 0.8730382323265076,
+      "learning_rate": 1.7375145180023228e-05,
+      "loss": 1.981,
       "step": 1500
     },
     {
-      "epoch": 0.7383479464697739,
-      "grad_norm": 0.6040017604827881,
-      "learning_rate": 1.8133572618895013e-05,
-      "loss": 1.7096,
+      "epoch": 0.37174721189591076,
+      "grad_norm": 0.6668545603752136,
+      "learning_rate": 1.8536585365853663e-05,
+      "loss": 1.9311,
       "step": 1600
     },
     {
-      "epoch": 0.7844946931241348,
-      "grad_norm": 0.7014355659484863,
-      "learning_rate": 1.7877195231380594e-05,
-      "loss": 1.6372,
+      "epoch": 0.3949814126394052,
+      "grad_norm": 0.6021186709403992,
+      "learning_rate": 1.969802555168409e-05,
+      "loss": 1.8733,
       "step": 1700
     },
     {
-      "epoch": 0.8306414397784956,
-      "grad_norm": 0.45642441511154175,
-      "learning_rate": 1.762081784386617e-05,
-      "loss": 1.5638,
+      "epoch": 0.41821561338289964,
+      "grad_norm": 0.8815124034881592,
+      "learning_rate": 1.9904479153220604e-05,
+      "loss": 1.8434,
       "step": 1800
     },
     {
-      "epoch": 0.8767881864328565,
-      "grad_norm": 0.7556698322296143,
-      "learning_rate": 1.7364440456351752e-05,
-      "loss": 1.5481,
+      "epoch": 0.44144981412639406,
+      "grad_norm": 1.1727079153060913,
+      "learning_rate": 1.9775396927843037e-05,
+      "loss": 1.8051,
       "step": 1900
     },
     {
-      "epoch": 0.9229349330872173,
-      "grad_norm": 0.48637843132019043,
-      "learning_rate": 1.710806306883733e-05,
-      "loss": 1.4976,
+      "epoch": 0.4646840148698885,
+      "grad_norm": 1.1215996742248535,
+      "learning_rate": 1.964631470246547e-05,
+      "loss": 1.733,
       "step": 2000
     },
     {
-      "epoch": 0.9690816797415782,
-      "grad_norm": 0.6813339591026306,
-      "learning_rate": 1.685168568132291e-05,
-      "loss": 1.4417,
+      "epoch": 0.4879182156133829,
+      "grad_norm": 1.1965365409851074,
+      "learning_rate": 1.9517232477087907e-05,
+      "loss": 1.6994,
       "step": 2100
     },
     {
-      "epoch": 1.015228426395939,
-      "grad_norm": 1.094603180885315,
-      "learning_rate": 1.6595308293808488e-05,
-      "loss": 1.3938,
+      "epoch": 0.5111524163568774,
+      "grad_norm": 1.2489936351776123,
+      "learning_rate": 1.938815025171034e-05,
+      "loss": 1.6529,
       "step": 2200
     },
     {
-      "epoch": 1.0613751730503,
-      "grad_norm": 0.37147483229637146,
-      "learning_rate": 1.6338930906294065e-05,
-      "loss": 1.3497,
+      "epoch": 0.5343866171003717,
+      "grad_norm": 1.5988222360610962,
+      "learning_rate": 1.9259068026332776e-05,
+      "loss": 1.5897,
       "step": 2300
     },
     {
-      "epoch": 1.1075219197046609,
-      "grad_norm": 0.2780097424983978,
-      "learning_rate": 1.6082553518779646e-05,
-      "loss": 1.3099,
+      "epoch": 0.5576208178438662,
+      "grad_norm": 0.6558517217636108,
+      "learning_rate": 1.912998580095521e-05,
+      "loss": 1.5099,
       "step": 2400
     },
     {
-      "epoch": 1.1536686663590217,
-      "grad_norm": 0.271342933177948,
-      "learning_rate": 1.5826176131265223e-05,
-      "loss": 1.285,
+      "epoch": 0.5808550185873605,
+      "grad_norm": 0.7629631757736206,
+      "learning_rate": 1.900219439783142e-05,
+      "loss": 1.4466,
       "step": 2500
     },
     {
-      "epoch": 1.1998154130133827,
-      "grad_norm": 0.27299413084983826,
-      "learning_rate": 1.55697987437508e-05,
-      "loss": 1.2614,
+      "epoch": 0.604089219330855,
+      "grad_norm": 0.9707331657409668,
+      "learning_rate": 1.8873112172453855e-05,
+      "loss": 1.3949,
       "step": 2600
     },
     {
-      "epoch": 1.2459621596677435,
-      "grad_norm": 0.29231297969818115,
-      "learning_rate": 1.531342135623638e-05,
-      "loss": 1.2308,
+      "epoch": 0.6273234200743495,
+      "grad_norm": 0.849176287651062,
+      "learning_rate": 1.874402994707629e-05,
+      "loss": 1.3449,
       "step": 2700
     },
     {
-      "epoch": 1.2921089063221043,
-      "grad_norm": 0.22232797741889954,
-      "learning_rate": 1.505704396872196e-05,
-      "loss": 1.1944,
+      "epoch": 0.6505576208178439,
+      "grad_norm": 0.460151731967926,
+      "learning_rate": 1.8614947721698724e-05,
+      "loss": 1.3182,
       "step": 2800
     },
     {
-      "epoch": 1.338255652976465,
-      "grad_norm": 0.9203324913978577,
-      "learning_rate": 1.480066658120754e-05,
-      "loss": 1.1544,
+      "epoch": 0.6737918215613383,
+      "grad_norm": 0.652923047542572,
+      "learning_rate": 1.8485865496321157e-05,
+      "loss": 1.2623,
       "step": 2900
     },
     {
-      "epoch": 1.384402399630826,
-      "grad_norm": 0.19580288231372833,
-      "learning_rate": 1.4544289193693117e-05,
-      "loss": 1.0996,
+      "epoch": 0.6970260223048327,
+      "grad_norm": 0.5269683599472046,
+      "learning_rate": 1.8356783270943594e-05,
+      "loss": 1.2059,
       "step": 3000
     },
     {
-      "epoch": 1.430549146285187,
-      "grad_norm": 0.42558759450912476,
-      "learning_rate": 1.4287911806178696e-05,
-      "loss": 1.0759,
+      "epoch": 0.7202602230483272,
+      "grad_norm": 0.6761623024940491,
+      "learning_rate": 1.8227701045566027e-05,
+      "loss": 1.1477,
       "step": 3100
     },
     {
-      "epoch": 1.4766958929395477,
-      "grad_norm": 0.24304209649562836,
-      "learning_rate": 1.4031534418664275e-05,
-      "loss": 1.0637,
+      "epoch": 0.7434944237918215,
+      "grad_norm": 0.4611155390739441,
+      "learning_rate": 1.809861882018846e-05,
+      "loss": 1.1063,
       "step": 3200
     },
     {
-      "epoch": 1.5228426395939088,
-      "grad_norm": 0.25986990332603455,
-      "learning_rate": 1.3775157031149852e-05,
-      "loss": 1.0446,
+      "epoch": 0.766728624535316,
+      "grad_norm": 1.20090913772583,
+      "learning_rate": 1.7969536594810896e-05,
+      "loss": 1.0812,
       "step": 3300
     },
     {
-      "epoch": 1.5689893862482696,
-      "grad_norm": 0.20725102722644806,
-      "learning_rate": 1.3518779643635433e-05,
-      "loss": 1.0384,
+      "epoch": 0.7899628252788105,
+      "grad_norm": 0.5198754072189331,
+      "learning_rate": 1.7840454369433332e-05,
+      "loss": 1.0637,
       "step": 3400
     },
     {
-      "epoch": 1.6151361329026304,
-      "grad_norm": 0.18936870992183685,
-      "learning_rate": 1.3262402256121012e-05,
-      "loss": 1.0139,
+      "epoch": 0.8131970260223048,
+      "grad_norm": 0.7287588119506836,
+      "learning_rate": 1.7711372144055765e-05,
+      "loss": 1.0311,
       "step": 3500
     },
     {
-      "epoch": 1.6612828795569912,
-      "grad_norm": 0.19966499507427216,
-      "learning_rate": 1.300602486860659e-05,
-      "loss": 1.0058,
+      "epoch": 0.8364312267657993,
+      "grad_norm": 0.850121021270752,
+      "learning_rate": 1.75822899186782e-05,
+      "loss": 0.9687,
       "step": 3600
     },
     {
-      "epoch": 1.707429626211352,
-      "grad_norm": 0.30528759956359863,
-      "learning_rate": 1.2749647481092169e-05,
-      "loss": 0.9838,
+      "epoch": 0.8596654275092936,
+      "grad_norm": 0.5256717801094055,
+      "learning_rate": 1.7453207693300635e-05,
+      "loss": 0.8706,
       "step": 3700
     },
     {
-      "epoch": 1.753576372865713,
-      "grad_norm": 0.2316664308309555,
-      "learning_rate": 1.2493270093577748e-05,
-      "loss": 0.9904,
+      "epoch": 0.8828996282527881,
+      "grad_norm": 0.6515185236930847,
+      "learning_rate": 1.7324125467923068e-05,
+      "loss": 0.8474,
       "step": 3800
     },
     {
-      "epoch": 1.7997231195200738,
-      "grad_norm": 0.2217002511024475,
-      "learning_rate": 1.2236892706063325e-05,
-      "loss": 0.9735,
+      "epoch": 0.9061338289962825,
+      "grad_norm": 0.8604176640510559,
+      "learning_rate": 1.7195043242545504e-05,
+      "loss": 0.8302,
       "step": 3900
     },
     {
-      "epoch": 1.8458698661744348,
-      "grad_norm": 0.2654038369655609,
-      "learning_rate": 1.1980515318548904e-05,
-      "loss": 0.9804,
+      "epoch": 0.929368029739777,
+      "grad_norm": 0.3369189202785492,
+      "learning_rate": 1.7065961017167937e-05,
+      "loss": 0.7959,
       "step": 4000
     },
     {
-      "epoch": 1.8920166128287956,
-      "grad_norm": 0.20543397963047028,
-      "learning_rate": 1.1724137931034483e-05,
-      "loss": 0.9424,
+      "epoch": 0.9526022304832714,
+      "grad_norm": 0.4804532527923584,
+      "learning_rate": 1.6936878791790373e-05,
+      "loss": 0.7945,
       "step": 4100
     },
     {
-      "epoch": 1.9381633594831564,
-      "grad_norm": 0.24414564669132233,
-      "learning_rate": 1.1467760543520064e-05,
-      "loss": 0.9353,
+      "epoch": 0.9758364312267658,
+      "grad_norm": 0.3839660882949829,
+      "learning_rate": 1.6807796566412806e-05,
+      "loss": 0.7975,
       "step": 4200
     },
     {
-      "epoch": 1.9843101061375172,
-      "grad_norm": 0.19333013892173767,
-      "learning_rate": 1.1211383156005641e-05,
-      "loss": 0.9374,
+      "epoch": 0.9990706319702602,
+      "grad_norm": 0.31136325001716614,
+      "learning_rate": 1.667871434103524e-05,
+      "loss": 0.7804,
       "step": 4300
     },
     {
-      "epoch": 2.030456852791878,
-      "grad_norm": 0.20063996315002441,
-      "learning_rate": 1.095500576849122e-05,
-      "loss": 0.9409,
+      "epoch": 1.0223048327137547,
+      "grad_norm": 0.2822754681110382,
+      "learning_rate": 1.6549632115657676e-05,
+      "loss": 0.7502,
       "step": 4400
     },
     {
-      "epoch": 2.076603599446239,
-      "grad_norm": 0.4319429397583008,
-      "learning_rate": 1.0698628380976798e-05,
-      "loss": 0.93,
+      "epoch": 1.045539033457249,
+      "grad_norm": 0.3364527225494385,
+      "learning_rate": 1.6420549890280112e-05,
+      "loss": 0.747,
       "step": 4500
     },
     {
-      "epoch": 2.1227503461006,
-      "grad_norm": 0.21358811855316162,
-      "learning_rate": 1.0442250993462377e-05,
-      "loss": 0.9231,
+      "epoch": 1.0687732342007434,
+      "grad_norm": 0.45242545008659363,
+      "learning_rate": 1.6291467664902545e-05,
+      "loss": 0.7263,
       "step": 4600
     },
     {
-      "epoch": 2.168897092754961,
-      "grad_norm": 0.2252470701932907,
-      "learning_rate": 1.0185873605947956e-05,
-      "loss": 0.9267,
+      "epoch": 1.092007434944238,
+      "grad_norm": 0.2541595995426178,
+      "learning_rate": 1.6162385439524978e-05,
+      "loss": 0.7311,
       "step": 4700
     },
     {
-      "epoch": 2.2150438394093217,
-      "grad_norm": 0.6058911681175232,
-      "learning_rate": 9.929496218433535e-06,
-      "loss": 0.902,
+      "epoch": 1.1152416356877324,
+      "grad_norm": 0.32410866022109985,
+      "learning_rate": 1.6033303214147415e-05,
+      "loss": 0.7213,
       "step": 4800
     },
     {
-      "epoch": 2.2611905860636825,
-      "grad_norm": 0.27027812600135803,
-      "learning_rate": 9.673118830919114e-06,
-      "loss": 0.8908,
+      "epoch": 1.1384758364312269,
+      "grad_norm": 0.28702208399772644,
+      "learning_rate": 1.5904220988769848e-05,
+      "loss": 0.7103,
       "step": 4900
     },
     {
-      "epoch": 2.3073373327180433,
-      "grad_norm": 0.3116415739059448,
-      "learning_rate": 9.416741443404692e-06,
-      "loss": 0.8971,
+      "epoch": 1.161710037174721,
+      "grad_norm": 0.2637524902820587,
+      "learning_rate": 1.577513876339228e-05,
+      "loss": 0.7033,
       "step": 5000
     },
     {
-      "epoch": 2.353484079372404,
-      "grad_norm": 0.2324889898300171,
-      "learning_rate": 9.160364055890272e-06,
-      "loss": 0.8927,
+      "epoch": 1.1849442379182156,
+      "grad_norm": 0.38048645853996277,
+      "learning_rate": 1.5646056538014717e-05,
+      "loss": 0.7111,
       "step": 5100
     },
     {
-      "epoch": 2.3996308260267654,
-      "grad_norm": 0.18322697281837463,
-      "learning_rate": 8.90398666837585e-06,
-      "loss": 0.8844,
+      "epoch": 1.20817843866171,
+      "grad_norm": 0.22926197946071625,
+      "learning_rate": 1.5516974312637153e-05,
+      "loss": 0.7053,
       "step": 5200
     },
     {
-      "epoch": 2.445777572681126,
-      "grad_norm": 1.4241108894348145,
-      "learning_rate": 8.650173054736572e-06,
-      "loss": 0.883,
+      "epoch": 1.2314126394052045,
+      "grad_norm": 0.2666023373603821,
+      "learning_rate": 1.5387892087259586e-05,
+      "loss": 0.6915,
       "step": 5300
     },
     {
-      "epoch": 2.491924319335487,
-      "grad_norm": 0.22246557474136353,
-      "learning_rate": 8.393795667222153e-06,
-      "loss": 0.8841,
+      "epoch": 1.2546468401486988,
+      "grad_norm": 0.2618410587310791,
+      "learning_rate": 1.525880986188202e-05,
+      "loss": 0.6843,
       "step": 5400
     },
     {
-      "epoch": 2.5380710659898478,
-      "grad_norm": 0.19322210550308228,
-      "learning_rate": 8.13741827970773e-06,
-      "loss": 0.89,
+      "epoch": 1.2778810408921932,
+      "grad_norm": 0.24479706585407257,
+      "learning_rate": 1.5129727636504454e-05,
+      "loss": 0.6775,
       "step": 5500
     },
     {
-      "epoch": 2.5842178126442086,
-      "grad_norm": 0.20664915442466736,
-      "learning_rate": 7.881040892193309e-06,
-      "loss": 0.8762,
+      "epoch": 1.3011152416356877,
+      "grad_norm": 0.19555561244487762,
+      "learning_rate": 1.5000645411126889e-05,
+      "loss": 0.6601,
       "step": 5600
     },
     {
-      "epoch": 2.6303645592985694,
-      "grad_norm": 0.19776581227779388,
-      "learning_rate": 7.624663504678887e-06,
-      "loss": 0.8649,
+      "epoch": 1.3243494423791822,
+      "grad_norm": 0.2121550738811493,
+      "learning_rate": 1.4871563185749323e-05,
+      "loss": 0.6625,
       "step": 5700
     },
     {
-      "epoch": 2.67651130595293,
-      "grad_norm": 0.7888526916503906,
-      "learning_rate": 7.368286117164467e-06,
-      "loss": 0.8678,
+      "epoch": 1.3475836431226766,
+      "grad_norm": 0.36492133140563965,
+      "learning_rate": 1.474248096037176e-05,
+      "loss": 0.6567,
       "step": 5800
     },
     {
-      "epoch": 2.722658052607291,
-      "grad_norm": 0.7994652390480042,
-      "learning_rate": 7.1119087296500455e-06,
-      "loss": 0.7896,
+      "epoch": 1.370817843866171,
+      "grad_norm": 0.28411343693733215,
+      "learning_rate": 1.4613398734994193e-05,
+      "loss": 0.6424,
       "step": 5900
     },
     {
-      "epoch": 2.768804799261652,
-      "grad_norm": 0.2278624325990677,
-      "learning_rate": 6.855531342135624e-06,
-      "loss": 0.7324,
+      "epoch": 1.3940520446096654,
+      "grad_norm": 0.3487832248210907,
+      "learning_rate": 1.4484316509616627e-05,
+      "loss": 0.6508,
       "step": 6000
     },
     {
-      "epoch": 2.814951545916013,
-      "grad_norm": 0.24736915528774261,
-      "learning_rate": 6.599153954621203e-06,
-      "loss": 0.7139,
+      "epoch": 1.4172862453531598,
+      "grad_norm": 0.4025629758834839,
+      "learning_rate": 1.4355234284239062e-05,
+      "loss": 0.6374,
       "step": 6100
     },
     {
-      "epoch": 2.861098292570374,
-      "grad_norm": 0.22141049802303314,
-      "learning_rate": 6.342776567106782e-06,
-      "loss": 0.7116,
+      "epoch": 1.4405204460966543,
+      "grad_norm": 0.31936919689178467,
+      "learning_rate": 1.4226152058861495e-05,
+      "loss": 0.6462,
       "step": 6200
     },
     {
-      "epoch": 2.9072450392247347,
-      "grad_norm": 0.21581608057022095,
-      "learning_rate": 6.08639917959236e-06,
-      "loss": 0.7185,
+      "epoch": 1.4637546468401488,
+      "grad_norm": 0.27360206842422485,
+      "learning_rate": 1.409706983348393e-05,
+      "loss": 0.6382,
       "step": 6300
     },
     {
-      "epoch": 2.9533917858790955,
-      "grad_norm": 0.27274981141090393,
-      "learning_rate": 5.830021792077939e-06,
-      "loss": 0.7114,
+      "epoch": 1.486988847583643,
+      "grad_norm": 0.35483697056770325,
+      "learning_rate": 1.3967987608106366e-05,
+      "loss": 0.6274,
       "step": 6400
     },
     {
-      "epoch": 2.9995385325334563,
-      "grad_norm": 0.4279099106788635,
-      "learning_rate": 5.5736444045635175e-06,
-      "loss": 0.6974,
+      "epoch": 1.5102230483271375,
+      "grad_norm": 0.30311813950538635,
+      "learning_rate": 1.38389053827288e-05,
+      "loss": 0.6258,
       "step": 6500
     },
     {
-      "epoch": 3.045685279187817,
-      "grad_norm": 0.4010777473449707,
-      "learning_rate": 5.3172670170490966e-06,
-      "loss": 0.7089,
+      "epoch": 1.533457249070632,
+      "grad_norm": 0.3184954524040222,
+      "learning_rate": 1.3709823157351234e-05,
+      "loss": 0.6313,
       "step": 6600
     },
     {
-      "epoch": 3.0918320258421783,
-      "grad_norm": 0.4470697343349457,
-      "learning_rate": 5.060889629534676e-06,
-      "loss": 0.6895,
+      "epoch": 1.5566914498141264,
+      "grad_norm": 0.2632908821105957,
+      "learning_rate": 1.3580740931973668e-05,
+      "loss": 0.6217,
       "step": 6700
     },
     {
-      "epoch": 3.137978772496539,
-      "grad_norm": 0.534737229347229,
-      "learning_rate": 4.804512242020255e-06,
-      "loss": 0.709,
+      "epoch": 1.579925650557621,
+      "grad_norm": 0.22145096957683563,
+      "learning_rate": 1.3451658706596103e-05,
+      "loss": 0.6245,
       "step": 6800
     },
     {
-      "epoch": 3.1841255191509,
-      "grad_norm": 0.3858148157596588,
-      "learning_rate": 4.548134854505833e-06,
-      "loss": 0.7008,
+      "epoch": 1.6031598513011152,
+      "grad_norm": 0.5008528828620911,
+      "learning_rate": 1.3322576481218536e-05,
+      "loss": 0.6187,
       "step": 6900
     },
     {
-      "epoch": 3.2302722658052607,
-      "grad_norm": 4.884620189666748,
-      "learning_rate": 4.291757466991412e-06,
-      "loss": 0.6926,
+      "epoch": 1.6263940520446096,
+      "grad_norm": 0.25452372431755066,
+      "learning_rate": 1.3193494255840972e-05,
+      "loss": 0.6084,
       "step": 7000
     },
     {
-      "epoch": 3.2764190124596215,
-      "grad_norm": 0.44127726554870605,
-      "learning_rate": 4.03538007947699e-06,
-      "loss": 0.6986,
+      "epoch": 1.649628252788104,
+      "grad_norm": 0.3917735815048218,
+      "learning_rate": 1.3064412030463407e-05,
+      "loss": 0.6088,
       "step": 7100
     },
     {
-      "epoch": 3.3225657591139823,
-      "grad_norm": 1.2448310852050781,
-      "learning_rate": 3.7790026919625694e-06,
-      "loss": 0.6884,
+      "epoch": 1.6728624535315983,
+      "grad_norm": 0.28736940026283264,
+      "learning_rate": 1.2935329805085842e-05,
+      "loss": 0.6084,
       "step": 7200
     },
     {
-      "epoch": 3.368712505768343,
-      "grad_norm": 0.8101204633712769,
-      "learning_rate": 3.522625304448148e-06,
-      "loss": 0.6851,
+      "epoch": 1.696096654275093,
+      "grad_norm": 0.3900860548019409,
+      "learning_rate": 1.2807538401962051e-05,
+      "loss": 0.6017,
       "step": 7300
     },
     {
-      "epoch": 3.4148592524227044,
-      "grad_norm": 0.5153388381004333,
-      "learning_rate": 3.2662479169337267e-06,
-      "loss": 0.6938,
+      "epoch": 1.7193308550185873,
+      "grad_norm": 0.2482582926750183,
+      "learning_rate": 1.2678456176584486e-05,
+      "loss": 0.5964,
       "step": 7400
     },
     {
-      "epoch": 3.461005999077065,
-      "grad_norm": 0.33079493045806885,
-      "learning_rate": 3.0098705294193053e-06,
-      "loss": 0.6948,
+      "epoch": 1.7425650557620818,
+      "grad_norm": 0.2464774250984192,
+      "learning_rate": 1.254937395120692e-05,
+      "loss": 0.5929,
       "step": 7500
     },
     {
-      "epoch": 3.507152745731426,
-      "grad_norm": 1.0203328132629395,
-      "learning_rate": 2.7534931419048844e-06,
-      "loss": 0.6855,
+      "epoch": 1.7657992565055762,
+      "grad_norm": 0.36112162470817566,
+      "learning_rate": 1.2420291725829354e-05,
+      "loss": 0.5913,
       "step": 7600
     },
     {
-      "epoch": 3.553299492385787,
-      "grad_norm": 0.3520820140838623,
-      "learning_rate": 2.4971157543904627e-06,
-      "loss": 0.6849,
+      "epoch": 1.7890334572490705,
+      "grad_norm": 0.30204829573631287,
+      "learning_rate": 1.2291209500451788e-05,
+      "loss": 0.5804,
       "step": 7700
     },
     {
-      "epoch": 3.5994462390401476,
-      "grad_norm": 0.28180956840515137,
-      "learning_rate": 2.2407383668760417e-06,
-      "loss": 0.699,
+      "epoch": 1.8122676579925652,
+      "grad_norm": 0.2731075584888458,
+      "learning_rate": 1.2162127275074223e-05,
+      "loss": 0.5881,
       "step": 7800
     },
     {
-      "epoch": 3.6455929856945084,
-      "grad_norm": 0.33973556756973267,
-      "learning_rate": 1.9843609793616204e-06,
-      "loss": 0.6756,
+      "epoch": 1.8355018587360594,
+      "grad_norm": 0.24604862928390503,
+      "learning_rate": 1.2033045049696656e-05,
+      "loss": 0.5679,
       "step": 7900
     },
     {
-      "epoch": 3.6917397323488697,
-      "grad_norm": 0.3416615128517151,
-      "learning_rate": 1.7279835918471993e-06,
-      "loss": 0.685,
+      "epoch": 1.858736059479554,
+      "grad_norm": 0.3449194133281708,
+      "learning_rate": 1.1903962824319092e-05,
+      "loss": 0.582,
       "step": 8000
     },
     {
-      "epoch": 3.7378864790032305,
-      "grad_norm": 0.7213825583457947,
-      "learning_rate": 1.4716062043327781e-06,
-      "loss": 0.6764,
+      "epoch": 1.8819702602230484,
+      "grad_norm": 0.310375452041626,
+      "learning_rate": 1.1774880598941527e-05,
+      "loss": 0.575,
       "step": 8100
     },
     {
-      "epoch": 3.7840332256575913,
-      "grad_norm": 0.5637441873550415,
-      "learning_rate": 1.2152288168183566e-06,
-      "loss": 0.6896,
+      "epoch": 1.9052044609665426,
+      "grad_norm": 0.28315114974975586,
+      "learning_rate": 1.1645798373563962e-05,
+      "loss": 0.5722,
       "step": 8200
     },
     {
-      "epoch": 3.830179972311952,
-      "grad_norm": 0.4190536141395569,
-      "learning_rate": 9.588514293039355e-07,
-      "loss": 0.675,
+      "epoch": 1.9284386617100373,
+      "grad_norm": 0.3091906011104584,
+      "learning_rate": 1.1516716148186395e-05,
+      "loss": 0.5533,
       "step": 8300
     },
     {
-      "epoch": 3.876326718966313,
-      "grad_norm": 0.37957823276519775,
-      "learning_rate": 7.024740417895142e-07,
-      "loss": 0.6767,
+      "epoch": 1.9516728624535316,
+      "grad_norm": 0.28990840911865234,
+      "learning_rate": 1.138763392280883e-05,
+      "loss": 0.5724,
       "step": 8400
     },
     {
-      "epoch": 3.9224734656206737,
-      "grad_norm": 0.5404504537582397,
-      "learning_rate": 4.4609665427509294e-07,
-      "loss": 0.695,
+      "epoch": 1.974907063197026,
+      "grad_norm": 0.44591304659843445,
+      "learning_rate": 1.1258551697431264e-05,
+      "loss": 0.5701,
       "step": 8500
     },
     {
-      "epoch": 3.9686202122750345,
-      "grad_norm": 14.911314964294434,
-      "learning_rate": 1.8971926676067174e-07,
-      "loss": 0.6794,
+      "epoch": 1.9981412639405205,
+      "grad_norm": 0.26404786109924316,
+      "learning_rate": 1.11294694720537e-05,
+      "loss": 0.553,
       "step": 8600
+    },
+    {
+      "epoch": 2.0213754646840147,
+      "grad_norm": 0.2843058705329895,
+      "learning_rate": 1.1000387246676133e-05,
+      "loss": 0.5631,
+      "step": 8700
+    },
+    {
+      "epoch": 2.0446096654275094,
+      "grad_norm": 0.20029422640800476,
+      "learning_rate": 1.0871305021298568e-05,
+      "loss": 0.5495,
+      "step": 8800
+    },
+    {
+      "epoch": 2.0678438661710037,
+      "grad_norm": 0.26215997338294983,
+      "learning_rate": 1.0742222795921003e-05,
+      "loss": 0.5562,
+      "step": 8900
+    },
+    {
+      "epoch": 2.091078066914498,
+      "grad_norm": 0.29611942172050476,
+      "learning_rate": 1.0613140570543436e-05,
+      "loss": 0.5541,
+      "step": 9000
+    },
+    {
+      "epoch": 2.1143122676579926,
+      "grad_norm": 0.2809213697910309,
+      "learning_rate": 1.048405834516587e-05,
+      "loss": 0.5429,
+      "step": 9100
+    },
+    {
+      "epoch": 2.137546468401487,
+      "grad_norm": 0.4684973657131195,
+      "learning_rate": 1.0354976119788307e-05,
+      "loss": 0.5518,
+      "step": 9200
+    },
+    {
+      "epoch": 2.1607806691449816,
+      "grad_norm": 0.2790776193141937,
+      "learning_rate": 1.0225893894410741e-05,
+      "loss": 0.5485,
+      "step": 9300
+    },
+    {
+      "epoch": 2.184014869888476,
+      "grad_norm": 0.24624982476234436,
+      "learning_rate": 1.0096811669033174e-05,
+      "loss": 0.5434,
+      "step": 9400
+    },
+    {
+      "epoch": 2.20724907063197,
+      "grad_norm": 0.27161070704460144,
+      "learning_rate": 9.967729443655609e-06,
+      "loss": 0.5503,
+      "step": 9500
+    },
+    {
+      "epoch": 2.2304832713754648,
+      "grad_norm": 0.2635902166366577,
+      "learning_rate": 9.838647218278044e-06,
+      "loss": 0.538,
+      "step": 9600
+    },
+    {
+      "epoch": 2.253717472118959,
+      "grad_norm": 0.35729700326919556,
+      "learning_rate": 9.709564992900478e-06,
+      "loss": 0.5376,
+      "step": 9700
+    },
+    {
+      "epoch": 2.2769516728624537,
+      "grad_norm": 0.224281907081604,
+      "learning_rate": 9.580482767522913e-06,
+      "loss": 0.5423,
+      "step": 9800
+    },
+    {
+      "epoch": 2.300185873605948,
+      "grad_norm": 0.2016523778438568,
+      "learning_rate": 9.451400542145348e-06,
+      "loss": 0.54,
+      "step": 9900
+    },
+    {
+      "epoch": 2.323420074349442,
+      "grad_norm": 0.3719424605369568,
+      "learning_rate": 9.322318316767782e-06,
+      "loss": 0.5326,
+      "step": 10000
+    },
+    {
+      "epoch": 2.346654275092937,
+      "grad_norm": 0.22268572449684143,
+      "learning_rate": 9.193236091390217e-06,
+      "loss": 0.5379,
+      "step": 10100
+    },
+    {
+      "epoch": 2.369888475836431,
+      "grad_norm": 0.3181590735912323,
+      "learning_rate": 9.06415386601265e-06,
+      "loss": 0.5328,
+      "step": 10200
+    },
+    {
+      "epoch": 2.393122676579926,
+      "grad_norm": 0.2703763246536255,
+      "learning_rate": 8.935071640635087e-06,
+      "loss": 0.5276,
+      "step": 10300
+    },
+    {
+      "epoch": 2.41635687732342,
+      "grad_norm": 0.2698732912540436,
+      "learning_rate": 8.80598941525752e-06,
+      "loss": 0.5338,
+      "step": 10400
+    },
+    {
+      "epoch": 2.4395910780669143,
+      "grad_norm": 0.2765790820121765,
+      "learning_rate": 8.676907189879954e-06,
+      "loss": 0.5418,
+      "step": 10500
+    },
+    {
+      "epoch": 2.462825278810409,
+      "grad_norm": 0.36516493558883667,
+      "learning_rate": 8.547824964502389e-06,
+      "loss": 0.5249,
+      "step": 10600
+    },
+    {
+      "epoch": 2.4860594795539033,
+      "grad_norm": 0.23371903598308563,
+      "learning_rate": 8.418742739124824e-06,
+      "loss": 0.5318,
+      "step": 10700
+    },
+    {
+      "epoch": 2.5092936802973975,
+      "grad_norm": 0.23883387446403503,
+      "learning_rate": 8.289660513747258e-06,
+      "loss": 0.5336,
+      "step": 10800
+    },
+    {
+      "epoch": 2.532527881040892,
+      "grad_norm": 0.23600026965141296,
+      "learning_rate": 8.160578288369693e-06,
+      "loss": 0.5207,
+      "step": 10900
+    },
+    {
+      "epoch": 2.5557620817843865,
+      "grad_norm": 0.22283987700939178,
+      "learning_rate": 8.031496062992128e-06,
+      "loss": 0.5261,
+      "step": 11000
+    },
+    {
+      "epoch": 2.578996282527881,
+      "grad_norm": 0.3077383041381836,
+      "learning_rate": 7.90241383761456e-06,
+      "loss": 0.5117,
+      "step": 11100
+    },
+    {
+      "epoch": 2.6022304832713754,
+      "grad_norm": 0.24372899532318115,
+      "learning_rate": 7.773331612236995e-06,
+      "loss": 0.5251,
+      "step": 11200
+    },
+    {
+      "epoch": 2.6254646840148697,
+      "grad_norm": 0.3168962001800537,
+      "learning_rate": 7.64424938685943e-06,
+      "loss": 0.5238,
+      "step": 11300
+    },
+    {
+      "epoch": 2.6486988847583643,
+      "grad_norm": 0.2522094249725342,
+      "learning_rate": 7.515167161481865e-06,
+      "loss": 0.5141,
+      "step": 11400
+    },
+    {
+      "epoch": 2.6719330855018586,
+      "grad_norm": 0.4139024317264557,
+      "learning_rate": 7.3860849361042984e-06,
+      "loss": 0.5185,
+      "step": 11500
+    },
+    {
+      "epoch": 2.6951672862453533,
+      "grad_norm": 0.2781153619289398,
+      "learning_rate": 7.257002710726734e-06,
+      "loss": 0.5121,
+      "step": 11600
+    },
+    {
+      "epoch": 2.7184014869888475,
+      "grad_norm": 0.38515913486480713,
+      "learning_rate": 7.127920485349168e-06,
+      "loss": 0.5178,
+      "step": 11700
+    },
+    {
+      "epoch": 2.741635687732342,
+      "grad_norm": 0.33289971947669983,
+      "learning_rate": 6.998838259971602e-06,
+      "loss": 0.5124,
+      "step": 11800
+    },
+    {
+      "epoch": 2.7648698884758365,
+      "grad_norm": 0.36876046657562256,
+      "learning_rate": 6.871046856847813e-06,
+      "loss": 0.5137,
+      "step": 11900
+    },
+    {
+      "epoch": 2.7881040892193307,
+      "grad_norm": 0.28098130226135254,
+      "learning_rate": 6.7419646314702466e-06,
+      "loss": 0.509,
+      "step": 12000
+    },
+    {
+      "epoch": 2.8113382899628254,
+      "grad_norm": 0.32521939277648926,
+      "learning_rate": 6.612882406092681e-06,
+      "loss": 0.512,
+      "step": 12100
+    },
+    {
+      "epoch": 2.8345724907063197,
+      "grad_norm": 0.23627902567386627,
+      "learning_rate": 6.483800180715116e-06,
+      "loss": 0.5084,
+      "step": 12200
+    },
+    {
+      "epoch": 2.857806691449814,
+      "grad_norm": 0.23111554980278015,
+      "learning_rate": 6.354717955337551e-06,
+      "loss": 0.517,
+      "step": 12300
+    },
+    {
+      "epoch": 2.8810408921933086,
+      "grad_norm": 0.3062553107738495,
+      "learning_rate": 6.2256357299599844e-06,
+      "loss": 0.5063,
+      "step": 12400
+    },
+    {
+      "epoch": 2.904275092936803,
+      "grad_norm": 0.3274383842945099,
+      "learning_rate": 6.09655350458242e-06,
+      "loss": 0.5066,
+      "step": 12500
+    },
+    {
+      "epoch": 2.9275092936802976,
+      "grad_norm": 0.25803956389427185,
+      "learning_rate": 5.967471279204854e-06,
+      "loss": 0.5064,
+      "step": 12600
+    },
+    {
+      "epoch": 2.950743494423792,
+      "grad_norm": 0.29026666283607483,
+      "learning_rate": 5.838389053827288e-06,
+      "loss": 0.5088,
+      "step": 12700
+    },
+    {
+      "epoch": 2.973977695167286,
+      "grad_norm": 0.36228805780410767,
+      "learning_rate": 5.709306828449723e-06,
+      "loss": 0.507,
+      "step": 12800
+    },
+    {
+      "epoch": 2.9972118959107807,
+      "grad_norm": 0.2669726014137268,
+      "learning_rate": 5.580224603072157e-06,
+      "loss": 0.4934,
+      "step": 12900
+    },
+    {
+      "epoch": 3.020446096654275,
+      "grad_norm": 0.24396216869354248,
+      "learning_rate": 5.451142377694592e-06,
+      "loss": 0.5099,
+      "step": 13000
+    },
+    {
+      "epoch": 3.0436802973977697,
+      "grad_norm": 0.25540581345558167,
+      "learning_rate": 5.322060152317027e-06,
+      "loss": 0.5037,
+      "step": 13100
+    },
+    {
+      "epoch": 3.066914498141264,
+      "grad_norm": 0.1964583396911621,
+      "learning_rate": 5.192977926939461e-06,
+      "loss": 0.5055,
+      "step": 13200
+    },
+    {
+      "epoch": 3.090148698884758,
+      "grad_norm": 0.2318154275417328,
+      "learning_rate": 5.063895701561895e-06,
+      "loss": 0.5041,
+      "step": 13300
+    },
+    {
+      "epoch": 3.113382899628253,
+      "grad_norm": 0.28110265731811523,
+      "learning_rate": 4.9348134761843295e-06,
+      "loss": 0.5043,
+      "step": 13400
+    },
+    {
+      "epoch": 3.136617100371747,
+      "grad_norm": 0.3360753357410431,
+      "learning_rate": 4.805731250806764e-06,
+      "loss": 0.4915,
+      "step": 13500
+    },
+    {
+      "epoch": 3.159851301115242,
+      "grad_norm": 0.3044135868549347,
+      "learning_rate": 4.676649025429199e-06,
+      "loss": 0.499,
+      "step": 13600
+    },
+    {
+      "epoch": 3.183085501858736,
+      "grad_norm": 0.28163620829582214,
+      "learning_rate": 4.547566800051634e-06,
+      "loss": 0.4996,
+      "step": 13700
+    },
+    {
+      "epoch": 3.2063197026022303,
+      "grad_norm": 0.23853909969329834,
+      "learning_rate": 4.418484574674068e-06,
+      "loss": 0.5073,
+      "step": 13800
+    },
+    {
+      "epoch": 3.229553903345725,
+      "grad_norm": 0.25510174036026,
+      "learning_rate": 4.289402349296502e-06,
+      "loss": 0.4988,
+      "step": 13900
+    },
+    {
+      "epoch": 3.2527881040892193,
+      "grad_norm": 0.650174081325531,
+      "learning_rate": 4.160320123918937e-06,
+      "loss": 0.5024,
+      "step": 14000
+    },
+    {
+      "epoch": 3.276022304832714,
+      "grad_norm": 0.36293137073516846,
+      "learning_rate": 4.0312378985413715e-06,
+      "loss": 0.4913,
+      "step": 14100
+    },
+    {
+      "epoch": 3.299256505576208,
+      "grad_norm": 0.35399818420410156,
+      "learning_rate": 3.902155673163805e-06,
+      "loss": 0.4993,
+      "step": 14200
+    },
+    {
+      "epoch": 3.3224907063197024,
+      "grad_norm": 0.2553289830684662,
+      "learning_rate": 3.7730734477862404e-06,
+      "loss": 0.5017,
+      "step": 14300
+    },
+    {
+      "epoch": 3.345724907063197,
+      "grad_norm": 0.25535061955451965,
+      "learning_rate": 3.643991222408675e-06,
+      "loss": 0.4895,
+      "step": 14400
+    },
+    {
+      "epoch": 3.3689591078066914,
+      "grad_norm": 0.2772742509841919,
+      "learning_rate": 3.514908997031109e-06,
+      "loss": 0.4954,
+      "step": 14500
+    },
+    {
+      "epoch": 3.392193308550186,
+      "grad_norm": 0.26105812191963196,
+      "learning_rate": 3.387117593907319e-06,
+      "loss": 0.4964,
+      "step": 14600
+    },
+    {
+      "epoch": 3.4154275092936803,
+      "grad_norm": 0.2538992166519165,
+      "learning_rate": 3.258035368529754e-06,
+      "loss": 0.4985,
+      "step": 14700
+    },
+    {
+      "epoch": 3.4386617100371746,
+      "grad_norm": 0.2889178693294525,
+      "learning_rate": 3.128953143152188e-06,
+      "loss": 0.4969,
+      "step": 14800
+    },
+    {
+      "epoch": 3.4618959107806693,
+      "grad_norm": 0.28792130947113037,
+      "learning_rate": 2.9998709177746228e-06,
+      "loss": 0.4985,
+      "step": 14900
+    },
+    {
+      "epoch": 3.4851301115241635,
+      "grad_norm": 0.36826494336128235,
+      "learning_rate": 2.8707886923970575e-06,
+      "loss": 0.4937,
+      "step": 15000
+    },
+    {
+      "epoch": 3.508364312267658,
+      "grad_norm": 0.24432937800884247,
+      "learning_rate": 2.7417064670194917e-06,
+      "loss": 0.4892,
+      "step": 15100
+    },
+    {
+      "epoch": 3.5315985130111525,
+      "grad_norm": 0.36436623334884644,
+      "learning_rate": 2.6126242416419264e-06,
+      "loss": 0.5029,
+      "step": 15200
+    },
+    {
+      "epoch": 3.5548327137546467,
+      "grad_norm": 0.3257830739021301,
+      "learning_rate": 2.4835420162643606e-06,
+      "loss": 0.484,
+      "step": 15300
+    },
+    {
+      "epoch": 3.5780669144981414,
+      "grad_norm": 0.20910651981830597,
+      "learning_rate": 2.354459790886795e-06,
+      "loss": 0.4934,
+      "step": 15400
+    },
+    {
+      "epoch": 3.6013011152416357,
+      "grad_norm": 0.27706313133239746,
+      "learning_rate": 2.2253775655092296e-06,
+      "loss": 0.4972,
+      "step": 15500
+    },
+    {
+      "epoch": 3.6245353159851303,
+      "grad_norm": 0.28043028712272644,
+      "learning_rate": 2.0962953401316643e-06,
+      "loss": 0.4878,
+      "step": 15600
+    },
+    {
+      "epoch": 3.6477695167286246,
+      "grad_norm": 0.34835153818130493,
+      "learning_rate": 1.9672131147540985e-06,
+      "loss": 0.4954,
+      "step": 15700
+    },
+    {
+      "epoch": 3.671003717472119,
+      "grad_norm": 0.3561202585697174,
+      "learning_rate": 1.838130889376533e-06,
+      "loss": 0.4992,
+      "step": 15800
+    },
+    {
+      "epoch": 3.6942379182156135,
+      "grad_norm": 0.2767621576786041,
+      "learning_rate": 1.7090486639989677e-06,
+      "loss": 0.4982,
+      "step": 15900
+    },
+    {
+      "epoch": 3.717472118959108,
+      "grad_norm": 0.22851090133190155,
+      "learning_rate": 1.579966438621402e-06,
+      "loss": 0.498,
+      "step": 16000
+    },
+    {
+      "epoch": 3.7407063197026025,
+      "grad_norm": 0.28282201290130615,
+      "learning_rate": 1.4508842132438364e-06,
+      "loss": 0.4898,
+      "step": 16100
+    },
+    {
+      "epoch": 3.7639405204460967,
+      "grad_norm": 0.24474182724952698,
+      "learning_rate": 1.3218019878662709e-06,
+      "loss": 0.501,
+      "step": 16200
+    },
+    {
+      "epoch": 3.787174721189591,
+      "grad_norm": 0.27427938580513,
+      "learning_rate": 1.1927197624887055e-06,
+      "loss": 0.4966,
+      "step": 16300
+    },
+    {
+      "epoch": 3.8104089219330852,
+      "grad_norm": 0.38391393423080444,
+      "learning_rate": 1.0636375371111398e-06,
+      "loss": 0.4941,
+      "step": 16400
+    },
+    {
+      "epoch": 3.83364312267658,
+      "grad_norm": 0.3098974823951721,
+      "learning_rate": 9.345553117335744e-07,
+      "loss": 0.4879,
+      "step": 16500
+    },
+    {
+      "epoch": 3.8568773234200746,
+      "grad_norm": 0.2817577123641968,
+      "learning_rate": 8.054730863560088e-07,
+      "loss": 0.4925,
+      "step": 16600
+    },
+    {
+      "epoch": 3.880111524163569,
+      "grad_norm": 0.3037372827529907,
+      "learning_rate": 6.763908609784433e-07,
+      "loss": 0.4927,
+      "step": 16700
+    },
+    {
+      "epoch": 3.903345724907063,
+      "grad_norm": 0.2850995659828186,
+      "learning_rate": 5.473086356008779e-07,
+      "loss": 0.4909,
+      "step": 16800
+    },
+    {
+      "epoch": 3.9265799256505574,
+      "grad_norm": 0.25115731358528137,
+      "learning_rate": 4.182264102233123e-07,
+      "loss": 0.5,
+      "step": 16900
+    },
+    {
+      "epoch": 3.949814126394052,
+      "grad_norm": 0.4323899745941162,
+      "learning_rate": 2.8914418484574677e-07,
+      "loss": 0.4861,
+      "step": 17000
+    },
+    {
+      "epoch": 3.9730483271375467,
+      "grad_norm": 0.30076873302459717,
+      "learning_rate": 1.6006195946818127e-07,
+      "loss": 0.4855,
+      "step": 17100
+    },
+    {
+      "epoch": 3.996282527881041,
+      "grad_norm": 0.2874129116535187,
+      "learning_rate": 3.097973409061573e-08,
+      "loss": 0.4957,
+      "step": 17200
     }
   ],
   "logging_steps": 100,
-  "max_steps": 8668,
+  "max_steps": 17216,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 4,
   "save_steps": 100,
@@ -628,7 +1230,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.884583671986995e+16,
+  "total_flos": 3.805111076121907e+16,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0f25f4c63676c8f67c641355c20226b66c56f1324579ddcaea18b93da2d7de52
+oid sha256:11a7f882a57ed2395a4921dd13204e513e9cf0ef0b5aff13aea0e0bf009fa0ce
 size 5432