Training in progress, step 200, checkpoint

checkpoint-200/adapter_config.json

@@ -24,13 +24,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "down_proj",
-    "v_proj",
-    "k_proj",
     "up_proj",
-    "q_proj",
+    "v_proj",
+    "down_proj",
+    "gate_proj",
     "o_proj",
-    "gate_proj"
+    "k_proj",
+    "q_proj"
   ],
   "task_type": "CAUSAL_LM",
   "trainable_token_indices": null,

checkpoint-200/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:58c3fcb913483de4b772aba663257cae72677fb93f598ef89f5b9e90fd964763
+oid sha256:c61392ceee0f17408472923a95258c9efc36e6ec7a27191e3d6143f7ab1117b3
 size 70430032

checkpoint-200/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fbe1e7c345467a06110401584beab1ce341ecafacb1d749d155f178672b9e3ba
+oid sha256:9a1461c7966846d5e49e767074b1272bafedc146c5365eb828fdd5c8ce324b87
 size 36136293

checkpoint-200/trainer_state.json

@@ -11,1434 +11,1434 @@
   "log_history": [
     {
       "epoch": 0.0025220680958385876,
-      "grad_norm": 83.50857543945312,
+      "grad_norm": 83.13754272460938,
       "learning_rate": 0.0,
-      "loss": 3.3887,
+      "loss": 3.3942,
       "step": 1
     },
     {
       "epoch": 0.005044136191677175,
-      "grad_norm": 71.31573486328125,
+      "grad_norm": 71.36713409423828,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 3.4305,
+      "loss": 3.4365,
       "step": 2
     },
     {
       "epoch": 0.007566204287515763,
-      "grad_norm": 66.55958557128906,
+      "grad_norm": 66.02528381347656,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 3.0425,
+      "loss": 3.0385,
       "step": 3
     },
     {
       "epoch": 0.01008827238335435,
-      "grad_norm": 67.16094207763672,
+      "grad_norm": 66.64842224121094,
       "learning_rate": 6e-06,
-      "loss": 3.1193,
+      "loss": 3.1228,
       "step": 4
     },
     {
       "epoch": 0.012610340479192938,
-      "grad_norm": 58.159088134765625,
+      "grad_norm": 58.996517181396484,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 2.7595,
+      "loss": 2.7637,
       "step": 5
     },
     {
       "epoch": 0.015132408575031526,
-      "grad_norm": 39.46601867675781,
+      "grad_norm": 39.48223114013672,
       "learning_rate": 1e-05,
-      "loss": 2.3184,
+      "loss": 2.3209,
       "step": 6
     },
     {
       "epoch": 0.017654476670870115,
-      "grad_norm": 45.5927734375,
+      "grad_norm": 46.47368240356445,
       "learning_rate": 9.987325728770595e-06,
-      "loss": 2.1763,
+      "loss": 2.1914,
       "step": 7
     },
     {
       "epoch": 0.0201765447667087,
-      "grad_norm": 20.798906326293945,
+      "grad_norm": 21.601478576660156,
       "learning_rate": 9.974651457541193e-06,
-      "loss": 2.1835,
+      "loss": 2.1892,
       "step": 8
     },
     {
       "epoch": 0.02269861286254729,
-      "grad_norm": 15.517325401306152,
+      "grad_norm": 15.485176086425781,
       "learning_rate": 9.961977186311787e-06,
-      "loss": 2.1354,
+      "loss": 2.1412,
       "step": 9
     },
     {
       "epoch": 0.025220680958385876,
-      "grad_norm": 15.667418479919434,
+      "grad_norm": 16.23291778564453,
       "learning_rate": 9.949302915082384e-06,
-      "loss": 1.9287,
+      "loss": 1.9318,
       "step": 10
     },
     {
       "epoch": 0.027742749054224466,
-      "grad_norm": 13.178864479064941,
+      "grad_norm": 13.485719680786133,
       "learning_rate": 9.93662864385298e-06,
-      "loss": 1.9898,
+      "loss": 1.9938,
       "step": 11
     },
     {
       "epoch": 0.03026481715006305,
-      "grad_norm": 10.053223609924316,
+      "grad_norm": 10.41800308227539,
       "learning_rate": 9.923954372623576e-06,
-      "loss": 2.0013,
+      "loss": 2.0039,
       "step": 12
     },
     {
       "epoch": 0.03278688524590164,
-      "grad_norm": 14.007344245910645,
+      "grad_norm": 13.47785472869873,
       "learning_rate": 9.91128010139417e-06,
-      "loss": 2.0559,
+      "loss": 2.0528,
       "step": 13
     },
     {
       "epoch": 0.03530895334174023,
-      "grad_norm": 11.14796257019043,
+      "grad_norm": 11.072226524353027,
       "learning_rate": 9.898605830164766e-06,
-      "loss": 2.0379,
+      "loss": 2.0423,
       "step": 14
     },
     {
       "epoch": 0.03783102143757881,
-      "grad_norm": 13.28112506866455,
+      "grad_norm": 13.525521278381348,
       "learning_rate": 9.885931558935362e-06,
-      "loss": 1.9854,
+      "loss": 1.9874,
       "step": 15
     },
     {
       "epoch": 0.0403530895334174,
-      "grad_norm": 11.45155143737793,
+      "grad_norm": 11.501786231994629,
       "learning_rate": 9.873257287705957e-06,
-      "loss": 2.1884,
+      "loss": 2.1873,
       "step": 16
     },
     {
       "epoch": 0.04287515762925599,
-      "grad_norm": 10.636513710021973,
+      "grad_norm": 10.845256805419922,
       "learning_rate": 9.860583016476553e-06,
-      "loss": 1.9676,
+      "loss": 1.9688,
       "step": 17
     },
     {
       "epoch": 0.04539722572509458,
-      "grad_norm": 10.700081825256348,
+      "grad_norm": 10.69442367553711,
       "learning_rate": 9.847908745247149e-06,
-      "loss": 1.9426,
+      "loss": 1.9445,
       "step": 18
     },
     {
       "epoch": 0.04791929382093316,
-      "grad_norm": 10.350312232971191,
+      "grad_norm": 10.206631660461426,
       "learning_rate": 9.835234474017745e-06,
-      "loss": 1.931,
+      "loss": 1.93,
       "step": 19
     },
     {
       "epoch": 0.05044136191677175,
-      "grad_norm": 11.60483455657959,
+      "grad_norm": 11.61585521697998,
       "learning_rate": 9.822560202788341e-06,
-      "loss": 1.9397,
+      "loss": 1.94,
       "step": 20
     },
     {
       "epoch": 0.05296343001261034,
-      "grad_norm": 10.923288345336914,
+      "grad_norm": 11.214790344238281,
       "learning_rate": 9.809885931558936e-06,
-      "loss": 1.8861,
+      "loss": 1.8872,
       "step": 21
     },
     {
       "epoch": 0.05548549810844893,
-      "grad_norm": 10.222322463989258,
+      "grad_norm": 10.26319694519043,
       "learning_rate": 9.797211660329532e-06,
-      "loss": 1.8154,
+      "loss": 1.8142,
       "step": 22
     },
     {
       "epoch": 0.058007566204287514,
-      "grad_norm": 9.665061950683594,
+      "grad_norm": 9.72568130493164,
       "learning_rate": 9.784537389100128e-06,
-      "loss": 1.9885,
+      "loss": 1.9886,
       "step": 23
     },
     {
       "epoch": 0.0605296343001261,
-      "grad_norm": 10.544888496398926,
+      "grad_norm": 10.376155853271484,
       "learning_rate": 9.771863117870724e-06,
-      "loss": 1.9728,
+      "loss": 1.9708,
       "step": 24
     },
     {
       "epoch": 0.06305170239596469,
-      "grad_norm": 9.924367904663086,
+      "grad_norm": 10.003466606140137,
       "learning_rate": 9.759188846641318e-06,
-      "loss": 1.9665,
+      "loss": 1.9703,
       "step": 25
     },
     {
       "epoch": 0.06557377049180328,
-      "grad_norm": 9.470415115356445,
+      "grad_norm": 9.423194885253906,
       "learning_rate": 9.746514575411914e-06,
-      "loss": 1.8872,
+      "loss": 1.8896,
       "step": 26
     },
     {
       "epoch": 0.06809583858764187,
-      "grad_norm": 9.330588340759277,
+      "grad_norm": 9.320464134216309,
       "learning_rate": 9.73384030418251e-06,
-      "loss": 1.9458,
+      "loss": 1.9495,
       "step": 27
     },
     {
       "epoch": 0.07061790668348046,
-      "grad_norm": 9.101158142089844,
+      "grad_norm": 9.116365432739258,
       "learning_rate": 9.721166032953105e-06,
-      "loss": 2.0452,
+      "loss": 2.0434,
       "step": 28
     },
     {
       "epoch": 0.07313997477931904,
-      "grad_norm": 9.63065242767334,
+      "grad_norm": 9.689815521240234,
       "learning_rate": 9.708491761723701e-06,
-      "loss": 1.9648,
+      "loss": 1.9642,
       "step": 29
     },
     {
       "epoch": 0.07566204287515763,
-      "grad_norm": 9.038743019104004,
+      "grad_norm": 9.166707992553711,
       "learning_rate": 9.695817490494297e-06,
-      "loss": 1.9469,
+      "loss": 1.9487,
       "step": 30
     },
     {
       "epoch": 0.07818411097099622,
-      "grad_norm": 9.375469207763672,
+      "grad_norm": 9.399063110351562,
       "learning_rate": 9.683143219264893e-06,
-      "loss": 2.0387,
+      "loss": 2.0389,
       "step": 31
     },
     {
       "epoch": 0.0807061790668348,
-      "grad_norm": 9.170827865600586,
+      "grad_norm": 9.226134300231934,
       "learning_rate": 9.670468948035488e-06,
-      "loss": 2.0396,
+      "loss": 2.0427,
       "step": 32
     },
     {
       "epoch": 0.0832282471626734,
-      "grad_norm": 9.155048370361328,
+      "grad_norm": 9.105801582336426,
       "learning_rate": 9.657794676806086e-06,
-      "loss": 2.1346,
+      "loss": 2.1308,
       "step": 33
     },
     {
       "epoch": 0.08575031525851198,
-      "grad_norm": 9.660133361816406,
+      "grad_norm": 9.654237747192383,
       "learning_rate": 9.64512040557668e-06,
-      "loss": 1.814,
+      "loss": 1.8129,
       "step": 34
     },
     {
       "epoch": 0.08827238335435057,
-      "grad_norm": 9.577939987182617,
+      "grad_norm": 9.531847953796387,
       "learning_rate": 9.632446134347276e-06,
-      "loss": 1.9353,
+      "loss": 1.9344,
       "step": 35
     },
     {
       "epoch": 0.09079445145018916,
-      "grad_norm": 9.347375869750977,
+      "grad_norm": 9.385769844055176,
       "learning_rate": 9.619771863117872e-06,
-      "loss": 1.8748,
+      "loss": 1.8751,
       "step": 36
     },
     {
       "epoch": 0.09331651954602774,
-      "grad_norm": 9.635443687438965,
+      "grad_norm": 9.698821067810059,
       "learning_rate": 9.607097591888467e-06,
-      "loss": 2.0299,
+      "loss": 2.029,
       "step": 37
     },
     {
       "epoch": 0.09583858764186633,
-      "grad_norm": 9.470548629760742,
+      "grad_norm": 9.558582305908203,
       "learning_rate": 9.594423320659063e-06,
-      "loss": 1.966,
+      "loss": 1.9676,
       "step": 38
     },
     {
       "epoch": 0.09836065573770492,
-      "grad_norm": 9.292247772216797,
+      "grad_norm": 9.290550231933594,
       "learning_rate": 9.581749049429659e-06,
-      "loss": 1.8779,
+      "loss": 1.8812,
       "step": 39
     },
     {
       "epoch": 0.1008827238335435,
-      "grad_norm": 9.822443962097168,
+      "grad_norm": 9.714218139648438,
       "learning_rate": 9.569074778200255e-06,
-      "loss": 1.8484,
+      "loss": 1.8447,
       "step": 40
     },
     {
       "epoch": 0.1034047919293821,
-      "grad_norm": 8.519198417663574,
+      "grad_norm": 8.576757431030273,
       "learning_rate": 9.55640050697085e-06,
-      "loss": 1.7622,
+      "loss": 1.7642,
       "step": 41
     },
     {
       "epoch": 0.10592686002522068,
-      "grad_norm": 9.220460891723633,
+      "grad_norm": 9.177382469177246,
       "learning_rate": 9.543726235741445e-06,
-      "loss": 1.8659,
+      "loss": 1.8653,
       "step": 42
     },
     {
       "epoch": 0.10844892812105927,
-      "grad_norm": 8.564704895019531,
+      "grad_norm": 8.541791915893555,
       "learning_rate": 9.531051964512042e-06,
-      "loss": 1.8478,
+      "loss": 1.8476,
       "step": 43
     },
     {
       "epoch": 0.11097099621689786,
-      "grad_norm": 9.696242332458496,
+      "grad_norm": 9.684398651123047,
       "learning_rate": 9.518377693282636e-06,
-      "loss": 1.8625,
+      "loss": 1.865,
       "step": 44
     },
     {
       "epoch": 0.11349306431273644,
-      "grad_norm": 9.121529579162598,
+      "grad_norm": 9.168283462524414,
       "learning_rate": 9.505703422053234e-06,
-      "loss": 1.9736,
+      "loss": 1.9742,
       "step": 45
     },
     {
       "epoch": 0.11601513240857503,
-      "grad_norm": 9.0239896774292,
+      "grad_norm": 9.015252113342285,
       "learning_rate": 9.493029150823828e-06,
-      "loss": 1.8687,
+      "loss": 1.8696,
       "step": 46
     },
     {
       "epoch": 0.11853720050441362,
-      "grad_norm": 8.566268920898438,
+      "grad_norm": 8.578283309936523,
       "learning_rate": 9.480354879594424e-06,
-      "loss": 1.7855,
+      "loss": 1.786,
       "step": 47
     },
     {
       "epoch": 0.1210592686002522,
-      "grad_norm": 9.728326797485352,
+      "grad_norm": 9.717804908752441,
       "learning_rate": 9.46768060836502e-06,
-      "loss": 2.0394,
+      "loss": 2.0357,
       "step": 48
     },
     {
       "epoch": 0.1235813366960908,
-      "grad_norm": 8.644201278686523,
+      "grad_norm": 8.676836013793945,
       "learning_rate": 9.455006337135616e-06,
-      "loss": 1.8265,
+      "loss": 1.8266,
       "step": 49
     },
     {
       "epoch": 0.12610340479192939,
-      "grad_norm": 9.00877857208252,
+      "grad_norm": 9.012295722961426,
       "learning_rate": 9.442332065906211e-06,
       "loss": 1.9655,
       "step": 50
     },
     {
       "epoch": 0.12610340479192939,
-      "eval_loss": 1.8593257665634155,
-      "eval_runtime": 17.8929,
-      "eval_samples_per_second": 39.401,
-      "eval_steps_per_second": 19.729,
+      "eval_loss": 1.8593026399612427,
+      "eval_runtime": 12.489,
+      "eval_samples_per_second": 56.45,
+      "eval_steps_per_second": 28.265,
       "step": 50
     },
     {
       "epoch": 0.12862547288776796,
-      "grad_norm": 8.626359939575195,
+      "grad_norm": 8.632142066955566,
       "learning_rate": 9.429657794676807e-06,
-      "loss": 1.8767,
+      "loss": 1.877,
       "step": 51
     },
     {
       "epoch": 0.13114754098360656,
-      "grad_norm": 8.989786148071289,
+      "grad_norm": 9.000449180603027,
       "learning_rate": 9.416983523447403e-06,
-      "loss": 1.9197,
+      "loss": 1.9196,
       "step": 52
     },
     {
       "epoch": 0.13366960907944514,
-      "grad_norm": 8.265259742736816,
+      "grad_norm": 8.286824226379395,
       "learning_rate": 9.404309252217997e-06,
-      "loss": 1.7882,
+      "loss": 1.7899,
       "step": 53
     },
     {
       "epoch": 0.13619167717528374,
-      "grad_norm": 9.578763008117676,
+      "grad_norm": 9.664816856384277,
       "learning_rate": 9.391634980988594e-06,
-      "loss": 2.0616,
+      "loss": 2.0637,
       "step": 54
     },
     {
       "epoch": 0.13871374527112232,
-      "grad_norm": 9.27950382232666,
+      "grad_norm": 9.408864974975586,
       "learning_rate": 9.37896070975919e-06,
-      "loss": 1.9043,
+      "loss": 1.9069,
       "step": 55
     },
     {
       "epoch": 0.14123581336696092,
-      "grad_norm": 8.789308547973633,
+      "grad_norm": 8.858333587646484,
       "learning_rate": 9.366286438529786e-06,
-      "loss": 1.8745,
+      "loss": 1.8742,
       "step": 56
     },
     {
       "epoch": 0.1437578814627995,
-      "grad_norm": 8.460257530212402,
+      "grad_norm": 8.489448547363281,
       "learning_rate": 9.35361216730038e-06,
-      "loss": 1.9719,
+      "loss": 1.9704,
       "step": 57
     },
     {
       "epoch": 0.14627994955863807,
-      "grad_norm": 8.686505317687988,
+      "grad_norm": 8.65485668182373,
       "learning_rate": 9.340937896070976e-06,
       "loss": 1.8288,
       "step": 58
     },
     {
       "epoch": 0.14880201765447668,
-      "grad_norm": 9.074113845825195,
+      "grad_norm": 9.034996032714844,
       "learning_rate": 9.328263624841572e-06,
-      "loss": 1.8212,
+      "loss": 1.8237,
       "step": 59
     },
     {
       "epoch": 0.15132408575031525,
-      "grad_norm": 8.919495582580566,
+      "grad_norm": 8.894627571105957,
       "learning_rate": 9.315589353612169e-06,
-      "loss": 1.9267,
+      "loss": 1.9284,
       "step": 60
     },
     {
       "epoch": 0.15384615384615385,
-      "grad_norm": 9.349574089050293,
+      "grad_norm": 9.375040054321289,
       "learning_rate": 9.302915082382765e-06,
-      "loss": 1.8566,
+      "loss": 1.8564,
       "step": 61
     },
     {
       "epoch": 0.15636822194199243,
-      "grad_norm": 10.293651580810547,
+      "grad_norm": 10.312094688415527,
       "learning_rate": 9.290240811153359e-06,
-      "loss": 1.7412,
+      "loss": 1.7402,
       "step": 62
     },
     {
       "epoch": 0.15889029003783103,
-      "grad_norm": 9.373260498046875,
+      "grad_norm": 9.399353981018066,
       "learning_rate": 9.277566539923955e-06,
-      "loss": 1.9118,
+      "loss": 1.9146,
       "step": 63
     },
     {
       "epoch": 0.1614123581336696,
-      "grad_norm": 9.066275596618652,
+      "grad_norm": 9.056628227233887,
       "learning_rate": 9.264892268694551e-06,
-      "loss": 1.89,
+      "loss": 1.8897,
       "step": 64
     },
     {
       "epoch": 0.16393442622950818,
-      "grad_norm": 9.459461212158203,
+      "grad_norm": 9.38147258758545,
       "learning_rate": 9.252217997465146e-06,
-      "loss": 1.8408,
+      "loss": 1.8415,
       "step": 65
     },
     {
       "epoch": 0.1664564943253468,
-      "grad_norm": 9.49438190460205,
+      "grad_norm": 9.386856079101562,
       "learning_rate": 9.239543726235742e-06,
-      "loss": 1.9826,
+      "loss": 1.982,
       "step": 66
     },
     {
       "epoch": 0.16897856242118536,
-      "grad_norm": 9.088374137878418,
+      "grad_norm": 9.0509614944458,
       "learning_rate": 9.226869455006338e-06,
-      "loss": 1.8442,
+      "loss": 1.8456,
       "step": 67
     },
     {
       "epoch": 0.17150063051702397,
-      "grad_norm": 8.981496810913086,
+      "grad_norm": 8.862478256225586,
       "learning_rate": 9.214195183776934e-06,
-      "loss": 1.8115,
+      "loss": 1.8085,
       "step": 68
     },
     {
       "epoch": 0.17402269861286254,
-      "grad_norm": 9.342341423034668,
+      "grad_norm": 9.258233070373535,
       "learning_rate": 9.201520912547528e-06,
-      "loss": 1.849,
+      "loss": 1.8478,
       "step": 69
     },
     {
       "epoch": 0.17654476670870115,
-      "grad_norm": 9.18710708618164,
+      "grad_norm": 9.156318664550781,
       "learning_rate": 9.188846641318126e-06,
-      "loss": 1.7475,
+      "loss": 1.748,
       "step": 70
     },
     {
       "epoch": 0.17906683480453972,
-      "grad_norm": 9.204767227172852,
+      "grad_norm": 9.234468460083008,
       "learning_rate": 9.17617237008872e-06,
-      "loss": 1.8912,
+      "loss": 1.8946,
       "step": 71
     },
     {
       "epoch": 0.18158890290037832,
-      "grad_norm": 9.105266571044922,
+      "grad_norm": 9.102885246276855,
       "learning_rate": 9.163498098859315e-06,
-      "loss": 1.8394,
+      "loss": 1.8383,
       "step": 72
     },
     {
       "epoch": 0.1841109709962169,
-      "grad_norm": 9.192793846130371,
+      "grad_norm": 9.106884002685547,
       "learning_rate": 9.150823827629913e-06,
-      "loss": 1.9242,
+      "loss": 1.9243,
       "step": 73
     },
     {
       "epoch": 0.18663303909205547,
-      "grad_norm": 8.911624908447266,
+      "grad_norm": 8.918400764465332,
       "learning_rate": 9.138149556400507e-06,
-      "loss": 1.9121,
+      "loss": 1.9136,
       "step": 74
     },
     {
       "epoch": 0.18915510718789408,
-      "grad_norm": 9.20001220703125,
+      "grad_norm": 9.20316219329834,
       "learning_rate": 9.125475285171103e-06,
-      "loss": 1.9752,
+      "loss": 1.9757,
       "step": 75
     },
     {
       "epoch": 0.19167717528373265,
-      "grad_norm": 8.967398643493652,
+      "grad_norm": 8.938888549804688,
       "learning_rate": 9.1128010139417e-06,
-      "loss": 1.7422,
+      "loss": 1.744,
       "step": 76
     },
     {
       "epoch": 0.19419924337957126,
-      "grad_norm": 8.867136001586914,
+      "grad_norm": 8.884087562561035,
       "learning_rate": 9.100126742712296e-06,
-      "loss": 1.7391,
+      "loss": 1.741,
       "step": 77
     },
     {
       "epoch": 0.19672131147540983,
-      "grad_norm": 9.721291542053223,
+      "grad_norm": 9.786199569702148,
       "learning_rate": 9.08745247148289e-06,
-      "loss": 1.7955,
+      "loss": 1.795,
       "step": 78
     },
     {
       "epoch": 0.19924337957124844,
-      "grad_norm": 8.936488151550293,
+      "grad_norm": 8.94467830657959,
       "learning_rate": 9.074778200253486e-06,
-      "loss": 1.602,
+      "loss": 1.6029,
       "step": 79
     },
     {
       "epoch": 0.201765447667087,
-      "grad_norm": 8.965775489807129,
+      "grad_norm": 8.95161247253418,
       "learning_rate": 9.062103929024082e-06,
-      "loss": 1.9918,
+      "loss": 1.9906,
       "step": 80
     },
     {
       "epoch": 0.2042875157629256,
-      "grad_norm": 9.03048324584961,
+      "grad_norm": 8.998520851135254,
       "learning_rate": 9.049429657794677e-06,
-      "loss": 1.8228,
+      "loss": 1.8235,
       "step": 81
     },
     {
       "epoch": 0.2068095838587642,
-      "grad_norm": 9.195337295532227,
+      "grad_norm": 9.164076805114746,
       "learning_rate": 9.036755386565273e-06,
-      "loss": 1.9259,
+      "loss": 1.9275,
       "step": 82
     },
     {
       "epoch": 0.20933165195460277,
-      "grad_norm": 9.696627616882324,
+      "grad_norm": 9.689983367919922,
       "learning_rate": 9.024081115335869e-06,
-      "loss": 1.8497,
+      "loss": 1.8506,
       "step": 83
     },
     {
       "epoch": 0.21185372005044137,
-      "grad_norm": 9.298614501953125,
+      "grad_norm": 9.274272918701172,
       "learning_rate": 9.011406844106465e-06,
-      "loss": 1.8685,
+      "loss": 1.8702,
       "step": 84
     },
     {
       "epoch": 0.21437578814627994,
-      "grad_norm": 8.810253143310547,
+      "grad_norm": 8.776298522949219,
       "learning_rate": 8.998732572877061e-06,
-      "loss": 1.7147,
+      "loss": 1.7139,
       "step": 85
     },
     {
       "epoch": 0.21689785624211855,
-      "grad_norm": 8.825125694274902,
+      "grad_norm": 8.801532745361328,
       "learning_rate": 8.986058301647655e-06,
-      "loss": 1.8466,
+      "loss": 1.8459,
       "step": 86
     },
     {
       "epoch": 0.21941992433795712,
-      "grad_norm": 8.869908332824707,
+      "grad_norm": 8.84094524383545,
       "learning_rate": 8.973384030418252e-06,
-      "loss": 1.9044,
+      "loss": 1.9022,
       "step": 87
     },
     {
       "epoch": 0.22194199243379573,
-      "grad_norm": 9.151841163635254,
+      "grad_norm": 9.14400863647461,
       "learning_rate": 8.960709759188848e-06,
-      "loss": 1.8922,
+      "loss": 1.8897,
       "step": 88
     },
     {
       "epoch": 0.2244640605296343,
-      "grad_norm": 9.193544387817383,
+      "grad_norm": 9.148355484008789,
       "learning_rate": 8.948035487959444e-06,
-      "loss": 1.8199,
+      "loss": 1.8187,
       "step": 89
     },
     {
       "epoch": 0.22698612862547288,
-      "grad_norm": 8.479183197021484,
+      "grad_norm": 8.494791984558105,
       "learning_rate": 8.935361216730038e-06,
-      "loss": 1.847,
+      "loss": 1.8459,
       "step": 90
     },
     {
       "epoch": 0.22950819672131148,
-      "grad_norm": 9.11871337890625,
+      "grad_norm": 9.062163352966309,
       "learning_rate": 8.922686945500634e-06,
-      "loss": 1.8631,
+      "loss": 1.8613,
       "step": 91
     },
     {
       "epoch": 0.23203026481715006,
-      "grad_norm": 9.080836296081543,
+      "grad_norm": 9.061888694763184,
       "learning_rate": 8.91001267427123e-06,
-      "loss": 1.9849,
+      "loss": 1.9857,
       "step": 92
     },
     {
       "epoch": 0.23455233291298866,
-      "grad_norm": 10.182774543762207,
+      "grad_norm": 10.139209747314453,
       "learning_rate": 8.897338403041825e-06,
-      "loss": 1.933,
+      "loss": 1.9318,
       "step": 93
     },
     {
       "epoch": 0.23707440100882723,
-      "grad_norm": 9.077899932861328,
+      "grad_norm": 9.2800931930542,
       "learning_rate": 8.884664131812421e-06,
-      "loss": 1.7692,
+      "loss": 1.7709,
       "step": 94
     },
     {
       "epoch": 0.23959646910466584,
-      "grad_norm": 8.86284065246582,
+      "grad_norm": 8.862195014953613,
       "learning_rate": 8.871989860583017e-06,
-      "loss": 1.7995,
+      "loss": 1.7987,
       "step": 95
     },
     {
       "epoch": 0.2421185372005044,
-      "grad_norm": 9.612898826599121,
+      "grad_norm": 9.58297348022461,
       "learning_rate": 8.859315589353613e-06,
-      "loss": 2.045,
+      "loss": 2.0442,
       "step": 96
     },
     {
       "epoch": 0.244640605296343,
-      "grad_norm": 9.372990608215332,
+      "grad_norm": 9.38765811920166,
       "learning_rate": 8.846641318124208e-06,
-      "loss": 1.8652,
+      "loss": 1.8665,
       "step": 97
     },
     {
       "epoch": 0.2471626733921816,
-      "grad_norm": 9.045456886291504,
+      "grad_norm": 9.039520263671875,
       "learning_rate": 8.833967046894805e-06,
-      "loss": 1.8405,
+      "loss": 1.8397,
       "step": 98
     },
     {
       "epoch": 0.24968474148802017,
-      "grad_norm": 8.836333274841309,
+      "grad_norm": 8.83530330657959,
       "learning_rate": 8.8212927756654e-06,
-      "loss": 1.7977,
+      "loss": 1.7988,
       "step": 99
     },
     {
       "epoch": 0.25220680958385877,
-      "grad_norm": 8.542088508605957,
+      "grad_norm": 8.552284240722656,
       "learning_rate": 8.808618504435996e-06,
-      "loss": 1.7449,
+      "loss": 1.7445,
       "step": 100
     },
     {
       "epoch": 0.25220680958385877,
-      "eval_loss": 1.821334719657898,
-      "eval_runtime": 16.1546,
-      "eval_samples_per_second": 43.641,
-      "eval_steps_per_second": 21.851,
+      "eval_loss": 1.821302890777588,
+      "eval_runtime": 12.7293,
+      "eval_samples_per_second": 55.384,
+      "eval_steps_per_second": 27.731,
       "step": 100
     },
     {
       "epoch": 0.2547288776796974,
-      "grad_norm": 9.596298217773438,
+      "grad_norm": 9.627294540405273,
       "learning_rate": 8.795944233206592e-06,
-      "loss": 1.8074,
+      "loss": 1.8091,
       "step": 101
     },
     {
       "epoch": 0.2572509457755359,
-      "grad_norm": 8.635272026062012,
+      "grad_norm": 8.704631805419922,
       "learning_rate": 8.783269961977186e-06,
-      "loss": 2.0499,
+      "loss": 2.0544,
       "step": 102
     },
     {
       "epoch": 0.2597730138713745,
-      "grad_norm": 8.884391784667969,
+      "grad_norm": 8.846882820129395,
       "learning_rate": 8.770595690747782e-06,
-      "loss": 1.8544,
+      "loss": 1.8539,
       "step": 103
     },
     {
       "epoch": 0.26229508196721313,
-      "grad_norm": 8.27906322479248,
+      "grad_norm": 8.313948631286621,
       "learning_rate": 8.757921419518379e-06,
-      "loss": 1.6638,
+      "loss": 1.6635,
       "step": 104
     },
     {
       "epoch": 0.2648171500630517,
-      "grad_norm": 8.792101860046387,
+      "grad_norm": 8.77181625366211,
       "learning_rate": 8.745247148288975e-06,
-      "loss": 1.8727,
+      "loss": 1.8712,
       "step": 105
     },
     {
       "epoch": 0.2673392181588903,
-      "grad_norm": 8.464868545532227,
+      "grad_norm": 8.449075698852539,
       "learning_rate": 8.732572877059569e-06,
-      "loss": 1.7947,
+      "loss": 1.7943,
       "step": 106
     },
     {
       "epoch": 0.2698612862547289,
-      "grad_norm": 9.57347583770752,
+      "grad_norm": 9.596981048583984,
       "learning_rate": 8.719898605830165e-06,
       "loss": 2.0333,
       "step": 107
     },
     {
       "epoch": 0.2723833543505675,
-      "grad_norm": 8.682156562805176,
+      "grad_norm": 8.669624328613281,
       "learning_rate": 8.707224334600761e-06,
-      "loss": 1.7808,
+      "loss": 1.7816,
       "step": 108
     },
     {
       "epoch": 0.27490542244640603,
-      "grad_norm": 9.720832824707031,
+      "grad_norm": 9.708998680114746,
       "learning_rate": 8.694550063371356e-06,
-      "loss": 1.9601,
+      "loss": 1.9616,
       "step": 109
     },
     {
       "epoch": 0.27742749054224464,
-      "grad_norm": 8.890487670898438,
+      "grad_norm": 8.8702974319458,
       "learning_rate": 8.681875792141954e-06,
-      "loss": 1.9123,
+      "loss": 1.9119,
       "step": 110
     },
     {
       "epoch": 0.27994955863808324,
-      "grad_norm": 9.121131896972656,
+      "grad_norm": 9.147343635559082,
       "learning_rate": 8.669201520912548e-06,
-      "loss": 1.8399,
+      "loss": 1.8413,
       "step": 111
     },
     {
       "epoch": 0.28247162673392184,
-      "grad_norm": 8.931142807006836,
+      "grad_norm": 8.951756477355957,
       "learning_rate": 8.656527249683144e-06,
-      "loss": 1.88,
+      "loss": 1.8813,
       "step": 112
     },
     {
       "epoch": 0.2849936948297604,
-      "grad_norm": 9.079472541809082,
+      "grad_norm": 9.136735916137695,
       "learning_rate": 8.64385297845374e-06,
-      "loss": 1.7766,
+      "loss": 1.7765,
       "step": 113
     },
     {
       "epoch": 0.287515762925599,
-      "grad_norm": 9.532262802124023,
+      "grad_norm": 9.53905963897705,
       "learning_rate": 8.631178707224336e-06,
-      "loss": 1.7715,
+      "loss": 1.7704,
       "step": 114
     },
     {
       "epoch": 0.2900378310214376,
-      "grad_norm": 8.928796768188477,
+      "grad_norm": 8.893516540527344,
       "learning_rate": 8.61850443599493e-06,
-      "loss": 1.6179,
+      "loss": 1.6198,
       "step": 115
     },
     {
       "epoch": 0.29255989911727615,
-      "grad_norm": 8.79500961303711,
+      "grad_norm": 8.778462409973145,
       "learning_rate": 8.605830164765527e-06,
-      "loss": 1.7653,
+      "loss": 1.7663,
       "step": 116
     },
     {
       "epoch": 0.29508196721311475,
-      "grad_norm": 9.038860321044922,
+      "grad_norm": 9.09292984008789,
       "learning_rate": 8.593155893536123e-06,
       "loss": 1.8273,
       "step": 117
     },
     {
       "epoch": 0.29760403530895335,
-      "grad_norm": 9.18289852142334,
+      "grad_norm": 9.203009605407715,
       "learning_rate": 8.580481622306717e-06,
-      "loss": 1.945,
+      "loss": 1.9456,
       "step": 118
     },
     {
       "epoch": 0.30012610340479196,
-      "grad_norm": 9.104387283325195,
+      "grad_norm": 9.124825477600098,
       "learning_rate": 8.567807351077313e-06,
-      "loss": 1.8444,
+      "loss": 1.844,
       "step": 119
     },
     {
       "epoch": 0.3026481715006305,
-      "grad_norm": 9.251449584960938,
+      "grad_norm": 9.237686157226562,
       "learning_rate": 8.55513307984791e-06,
-      "loss": 1.7502,
+      "loss": 1.7495,
       "step": 120
     },
     {
       "epoch": 0.3051702395964691,
-      "grad_norm": 9.526406288146973,
+      "grad_norm": 9.535928726196289,
       "learning_rate": 8.542458808618506e-06,
-      "loss": 1.8277,
+      "loss": 1.828,
       "step": 121
     },
     {
       "epoch": 0.3076923076923077,
-      "grad_norm": 9.494799613952637,
+      "grad_norm": 9.44102954864502,
       "learning_rate": 8.5297845373891e-06,
-      "loss": 1.9802,
+      "loss": 1.9785,
       "step": 122
     },
     {
       "epoch": 0.31021437578814626,
-      "grad_norm": 9.093886375427246,
+      "grad_norm": 9.092142105102539,
       "learning_rate": 8.517110266159696e-06,
-      "loss": 1.7598,
+      "loss": 1.7611,
       "step": 123
     },
     {
       "epoch": 0.31273644388398486,
-      "grad_norm": 9.275187492370605,
+      "grad_norm": 9.274849891662598,
       "learning_rate": 8.504435994930292e-06,
-      "loss": 1.7269,
+      "loss": 1.7283,
       "step": 124
     },
     {
       "epoch": 0.31525851197982346,
-      "grad_norm": 8.311441421508789,
+      "grad_norm": 8.317136764526367,
       "learning_rate": 8.491761723700888e-06,
-      "loss": 1.8353,
+      "loss": 1.8366,
       "step": 125
     },
     {
       "epoch": 0.31778058007566207,
-      "grad_norm": 8.989309310913086,
+      "grad_norm": 8.939324378967285,
       "learning_rate": 8.479087452471484e-06,
-      "loss": 1.8799,
+      "loss": 1.878,
       "step": 126
     },
     {
       "epoch": 0.3203026481715006,
-      "grad_norm": 9.124334335327148,
+      "grad_norm": 9.072879791259766,
       "learning_rate": 8.466413181242079e-06,
-      "loss": 1.8709,
+      "loss": 1.8724,
       "step": 127
     },
     {
       "epoch": 0.3228247162673392,
-      "grad_norm": 9.008078575134277,
+      "grad_norm": 8.94404411315918,
       "learning_rate": 8.453738910012675e-06,
-      "loss": 1.7002,
+      "loss": 1.7007,
       "step": 128
     },
     {
       "epoch": 0.3253467843631778,
-      "grad_norm": 8.43070125579834,
+      "grad_norm": 8.423967361450195,
       "learning_rate": 8.441064638783271e-06,
-      "loss": 1.8012,
+      "loss": 1.8021,
       "step": 129
     },
     {
       "epoch": 0.32786885245901637,
-      "grad_norm": 9.406371116638184,
+      "grad_norm": 9.372537612915039,
       "learning_rate": 8.428390367553865e-06,
-      "loss": 1.8611,
+      "loss": 1.8622,
       "step": 130
     },
     {
       "epoch": 0.33039092055485497,
-      "grad_norm": 8.853006362915039,
+      "grad_norm": 8.834014892578125,
       "learning_rate": 8.415716096324462e-06,
-      "loss": 1.7709,
+      "loss": 1.7708,
       "step": 131
     },
     {
       "epoch": 0.3329129886506936,
-      "grad_norm": 9.170083045959473,
+      "grad_norm": 9.13171100616455,
       "learning_rate": 8.403041825095058e-06,
-      "loss": 2.0669,
+      "loss": 2.0644,
       "step": 132
     },
     {
       "epoch": 0.3354350567465322,
-      "grad_norm": 9.382063865661621,
+      "grad_norm": 9.38616943359375,
       "learning_rate": 8.390367553865654e-06,
       "loss": 1.8012,
       "step": 133
     },
     {
       "epoch": 0.3379571248423707,
-      "grad_norm": 9.157647132873535,
+      "grad_norm": 9.134305953979492,
       "learning_rate": 8.377693282636248e-06,
-      "loss": 1.8584,
+      "loss": 1.8579,
       "step": 134
     },
     {
       "epoch": 0.34047919293820933,
-      "grad_norm": 9.179341316223145,
+      "grad_norm": 9.147695541381836,
       "learning_rate": 8.365019011406846e-06,
-      "loss": 1.8589,
+      "loss": 1.8561,
       "step": 135
     },
     {
       "epoch": 0.34300126103404793,
-      "grad_norm": 9.133096694946289,
+      "grad_norm": 9.097079277038574,
       "learning_rate": 8.35234474017744e-06,
-      "loss": 1.9222,
+      "loss": 1.9241,
       "step": 136
     },
     {
       "epoch": 0.3455233291298865,
-      "grad_norm": 8.78919792175293,
+      "grad_norm": 8.7761812210083,
       "learning_rate": 8.339670468948035e-06,
-      "loss": 1.839,
+      "loss": 1.8393,
       "step": 137
     },
     {
       "epoch": 0.3480453972257251,
-      "grad_norm": 8.94666576385498,
+      "grad_norm": 8.91004467010498,
       "learning_rate": 8.326996197718633e-06,
-      "loss": 1.7538,
+      "loss": 1.7533,
       "step": 138
     },
     {
       "epoch": 0.3505674653215637,
-      "grad_norm": 9.572528839111328,
+      "grad_norm": 9.649863243103027,
       "learning_rate": 8.314321926489227e-06,
-      "loss": 1.9778,
+      "loss": 1.9777,
       "step": 139
     },
     {
       "epoch": 0.3530895334174023,
-      "grad_norm": 9.474174499511719,
+      "grad_norm": 9.433012008666992,
       "learning_rate": 8.301647655259823e-06,
-      "loss": 1.9579,
+      "loss": 1.959,
       "step": 140
     },
     {
       "epoch": 0.35561160151324084,
-      "grad_norm": 9.462223052978516,
+      "grad_norm": 9.400664329528809,
       "learning_rate": 8.28897338403042e-06,
-      "loss": 1.7689,
+      "loss": 1.7688,
       "step": 141
     },
     {
       "epoch": 0.35813366960907944,
-      "grad_norm": 8.904463768005371,
+      "grad_norm": 8.890508651733398,
       "learning_rate": 8.276299112801015e-06,
-      "loss": 1.7506,
+      "loss": 1.7521,
       "step": 142
     },
     {
       "epoch": 0.36065573770491804,
-      "grad_norm": 9.692485809326172,
+      "grad_norm": 9.65843391418457,
       "learning_rate": 8.26362484157161e-06,
-      "loss": 2.0011,
+      "loss": 1.9993,
       "step": 143
     },
     {
       "epoch": 0.36317780580075665,
-      "grad_norm": 8.99770450592041,
+      "grad_norm": 8.980792045593262,
       "learning_rate": 8.250950570342206e-06,
-      "loss": 1.7689,
+      "loss": 1.7685,
       "step": 144
     },
     {
       "epoch": 0.3656998738965952,
-      "grad_norm": 8.849214553833008,
+      "grad_norm": 8.810537338256836,
       "learning_rate": 8.238276299112802e-06,
-      "loss": 1.845,
+      "loss": 1.8443,
       "step": 145
     },
     {
       "epoch": 0.3682219419924338,
-      "grad_norm": 9.16205883026123,
+      "grad_norm": 9.117584228515625,
       "learning_rate": 8.225602027883396e-06,
-      "loss": 1.9285,
+      "loss": 1.9269,
       "step": 146
     },
     {
       "epoch": 0.3707440100882724,
-      "grad_norm": 9.169661521911621,
+      "grad_norm": 9.197243690490723,
       "learning_rate": 8.212927756653993e-06,
-      "loss": 1.8835,
+      "loss": 1.8833,
       "step": 147
     },
     {
       "epoch": 0.37326607818411095,
-      "grad_norm": 9.326395988464355,
+      "grad_norm": 9.297012329101562,
       "learning_rate": 8.200253485424589e-06,
-      "loss": 1.8162,
+      "loss": 1.8194,
       "step": 148
     },
     {
       "epoch": 0.37578814627994955,
-      "grad_norm": 8.962848663330078,
+      "grad_norm": 8.982695579528809,
       "learning_rate": 8.187579214195185e-06,
-      "loss": 1.8775,
+      "loss": 1.8801,
       "step": 149
     },
     {
       "epoch": 0.37831021437578816,
-      "grad_norm": 8.53931713104248,
+      "grad_norm": 8.55199909210205,
       "learning_rate": 8.17490494296578e-06,
-      "loss": 1.7893,
+      "loss": 1.7882,
       "step": 150
     },
     {
       "epoch": 0.37831021437578816,
-      "eval_loss": 1.8017936944961548,
-      "eval_runtime": 17.1952,
-      "eval_samples_per_second": 41.0,
-      "eval_steps_per_second": 20.529,
+      "eval_loss": 1.801971435546875,
+      "eval_runtime": 12.754,
+      "eval_samples_per_second": 55.277,
+      "eval_steps_per_second": 27.678,
       "step": 150
     },
     {
       "epoch": 0.38083228247162676,
-      "grad_norm": 8.814638137817383,
+      "grad_norm": 8.808869361877441,
       "learning_rate": 8.162230671736375e-06,
       "loss": 1.7401,
       "step": 151
     },
     {
       "epoch": 0.3833543505674653,
-      "grad_norm": 9.316661834716797,
+      "grad_norm": 9.283302307128906,
       "learning_rate": 8.149556400506971e-06,
-      "loss": 1.9799,
+      "loss": 1.9796,
       "step": 152
     },
     {
       "epoch": 0.3858764186633039,
-      "grad_norm": 9.053354263305664,
+      "grad_norm": 9.003050804138184,
       "learning_rate": 8.136882129277567e-06,
-      "loss": 1.6721,
+      "loss": 1.6729,
       "step": 153
     },
     {
       "epoch": 0.3883984867591425,
-      "grad_norm": 9.15576457977295,
+      "grad_norm": 9.130897521972656,
       "learning_rate": 8.124207858048164e-06,
-      "loss": 1.9065,
+      "loss": 1.9085,
       "step": 154
     },
     {
       "epoch": 0.39092055485498106,
-      "grad_norm": 8.911511421203613,
+      "grad_norm": 8.871198654174805,
       "learning_rate": 8.111533586818758e-06,
-      "loss": 1.9229,
+      "loss": 1.9218,
       "step": 155
     },
     {
       "epoch": 0.39344262295081966,
-      "grad_norm": 8.312164306640625,
+      "grad_norm": 8.295755386352539,
       "learning_rate": 8.098859315589354e-06,
-      "loss": 1.7699,
+      "loss": 1.7709,
       "step": 156
     },
     {
       "epoch": 0.39596469104665827,
-      "grad_norm": 9.155098915100098,
+      "grad_norm": 9.15046501159668,
       "learning_rate": 8.08618504435995e-06,
-      "loss": 1.9292,
+      "loss": 1.9304,
       "step": 157
     },
     {
       "epoch": 0.39848675914249687,
-      "grad_norm": 8.932412147521973,
+      "grad_norm": 8.887716293334961,
       "learning_rate": 8.073510773130546e-06,
-      "loss": 1.8102,
+      "loss": 1.8097,
       "step": 158
     },
     {
       "epoch": 0.4010088272383354,
-      "grad_norm": 8.135700225830078,
+      "grad_norm": 8.1146821975708,
       "learning_rate": 8.06083650190114e-06,
-      "loss": 1.72,
+      "loss": 1.7202,
       "step": 159
     },
     {
       "epoch": 0.403530895334174,
-      "grad_norm": 9.693431854248047,
+      "grad_norm": 9.62702465057373,
       "learning_rate": 8.048162230671737e-06,
-      "loss": 1.8558,
+      "loss": 1.8547,
       "step": 160
     },
     {
       "epoch": 0.4060529634300126,
-      "grad_norm": 8.98935317993164,
+      "grad_norm": 8.954533576965332,
       "learning_rate": 8.035487959442333e-06,
-      "loss": 1.9407,
+      "loss": 1.9418,
       "step": 161
     },
     {
       "epoch": 0.4085750315258512,
-      "grad_norm": 8.735937118530273,
+      "grad_norm": 8.709664344787598,
       "learning_rate": 8.022813688212927e-06,
-      "loss": 1.8333,
+      "loss": 1.8355,
       "step": 162
     },
     {
       "epoch": 0.4110970996216898,
-      "grad_norm": 9.313026428222656,
+      "grad_norm": 9.2684907913208,
       "learning_rate": 8.010139416983525e-06,
-      "loss": 1.9925,
+      "loss": 1.9908,
       "step": 163
     },
     {
       "epoch": 0.4136191677175284,
-      "grad_norm": 9.607177734375,
+      "grad_norm": 9.581305503845215,
       "learning_rate": 7.99746514575412e-06,
-      "loss": 1.8164,
+      "loss": 1.8155,
       "step": 164
     },
     {
       "epoch": 0.416141235813367,
-      "grad_norm": 9.086844444274902,
+      "grad_norm": 9.040098190307617,
       "learning_rate": 7.984790874524716e-06,
-      "loss": 1.8585,
+      "loss": 1.8591,
       "step": 165
     },
     {
       "epoch": 0.41866330390920553,
-      "grad_norm": 9.212312698364258,
+      "grad_norm": 9.17188549041748,
       "learning_rate": 7.972116603295312e-06,
       "loss": 1.8332,
       "step": 166
     },
     {
       "epoch": 0.42118537200504413,
-      "grad_norm": 9.246332168579102,
+      "grad_norm": 9.190821647644043,
       "learning_rate": 7.959442332065906e-06,
       "loss": 1.8488,
       "step": 167
     },
     {
       "epoch": 0.42370744010088274,
-      "grad_norm": 9.077427864074707,
+      "grad_norm": 9.045573234558105,
       "learning_rate": 7.946768060836502e-06,
-      "loss": 1.7442,
+      "loss": 1.7428,
       "step": 168
     },
     {
       "epoch": 0.4262295081967213,
-      "grad_norm": 9.019879341125488,
+      "grad_norm": 8.98148250579834,
       "learning_rate": 7.934093789607098e-06,
-      "loss": 1.9638,
+      "loss": 1.965,
       "step": 169
     },
     {
       "epoch": 0.4287515762925599,
-      "grad_norm": 8.645258903503418,
+      "grad_norm": 8.625720024108887,
       "learning_rate": 7.921419518377694e-06,
-      "loss": 1.8532,
+      "loss": 1.8538,
       "step": 170
     },
     {
       "epoch": 0.4312736443883985,
-      "grad_norm": 8.863842010498047,
+      "grad_norm": 8.84864330291748,
       "learning_rate": 7.908745247148289e-06,
-      "loss": 1.7754,
+      "loss": 1.7758,
       "step": 171
     },
     {
       "epoch": 0.4337957124842371,
-      "grad_norm": 9.303889274597168,
+      "grad_norm": 9.253456115722656,
       "learning_rate": 7.896070975918885e-06,
-      "loss": 1.774,
+      "loss": 1.7754,
       "step": 172
     },
     {
       "epoch": 0.43631778058007564,
-      "grad_norm": 9.072324752807617,
+      "grad_norm": 9.059581756591797,
       "learning_rate": 7.883396704689481e-06,
-      "loss": 1.9916,
+      "loss": 1.9926,
       "step": 173
     },
     {
       "epoch": 0.43883984867591425,
-      "grad_norm": 8.935220718383789,
+      "grad_norm": 8.904810905456543,
       "learning_rate": 7.870722433460076e-06,
-      "loss": 1.7988,
+      "loss": 1.7971,
       "step": 174
     },
     {
       "epoch": 0.44136191677175285,
-      "grad_norm": 8.469002723693848,
+      "grad_norm": 8.4295654296875,
       "learning_rate": 7.858048162230673e-06,
-      "loss": 1.7943,
+      "loss": 1.7947,
       "step": 175
     },
     {
       "epoch": 0.44388398486759145,
-      "grad_norm": 8.663657188415527,
+      "grad_norm": 8.664776802062988,
       "learning_rate": 7.845373891001268e-06,
       "loss": 1.9424,
       "step": 176
     },
     {
       "epoch": 0.44640605296343,
-      "grad_norm": 9.067521095275879,
+      "grad_norm": 9.025026321411133,
       "learning_rate": 7.832699619771864e-06,
-      "loss": 1.8128,
+      "loss": 1.8143,
       "step": 177
     },
     {
       "epoch": 0.4489281210592686,
-      "grad_norm": 8.450661659240723,
+      "grad_norm": 8.458661079406738,
       "learning_rate": 7.82002534854246e-06,
-      "loss": 1.7287,
+      "loss": 1.7291,
       "step": 178
     },
     {
       "epoch": 0.4514501891551072,
-      "grad_norm": 8.83759593963623,
+      "grad_norm": 8.800137519836426,
       "learning_rate": 7.807351077313056e-06,
-      "loss": 1.7602,
+      "loss": 1.759,
       "step": 179
     },
     {
       "epoch": 0.45397225725094575,
-      "grad_norm": 8.693680763244629,
+      "grad_norm": 8.654534339904785,
       "learning_rate": 7.79467680608365e-06,
-      "loss": 2.0174,
+      "loss": 2.0166,
       "step": 180
     },
     {
       "epoch": 0.45649432534678436,
-      "grad_norm": 9.432674407958984,
+      "grad_norm": 9.417305946350098,
       "learning_rate": 7.782002534854247e-06,
-      "loss": 1.9696,
+      "loss": 1.9709,
       "step": 181
     },
     {
       "epoch": 0.45901639344262296,
-      "grad_norm": 8.901416778564453,
+      "grad_norm": 8.89673137664795,
       "learning_rate": 7.769328263624843e-06,
-      "loss": 1.8075,
+      "loss": 1.8067,
       "step": 182
     },
     {
       "epoch": 0.46153846153846156,
-      "grad_norm": 8.675703048706055,
+      "grad_norm": 8.633663177490234,
       "learning_rate": 7.756653992395437e-06,
-      "loss": 1.8696,
+      "loss": 1.8676,
       "step": 183
     },
     {
       "epoch": 0.4640605296343001,
-      "grad_norm": 9.176361083984375,
+      "grad_norm": 9.134156227111816,
       "learning_rate": 7.743979721166033e-06,
-      "loss": 1.8572,
+      "loss": 1.857,
       "step": 184
     },
     {
       "epoch": 0.4665825977301387,
-      "grad_norm": 8.568829536437988,
+      "grad_norm": 8.542975425720215,
       "learning_rate": 7.73130544993663e-06,
       "loss": 1.638,
       "step": 185
     },
     {
       "epoch": 0.4691046658259773,
-      "grad_norm": 8.751213073730469,
+      "grad_norm": 8.729076385498047,
       "learning_rate": 7.718631178707225e-06,
-      "loss": 1.7869,
+      "loss": 1.7845,
       "step": 186
     },
     {
       "epoch": 0.47162673392181587,
-      "grad_norm": 8.986238479614258,
+      "grad_norm": 8.95164680480957,
       "learning_rate": 7.70595690747782e-06,
-      "loss": 1.831,
+      "loss": 1.8296,
       "step": 187
     },
     {
       "epoch": 0.47414880201765447,
-      "grad_norm": 9.22641372680664,
+      "grad_norm": 9.210955619812012,
       "learning_rate": 7.693282636248416e-06,
-      "loss": 1.7918,
+      "loss": 1.7914,
       "step": 188
     },
     {
       "epoch": 0.4766708701134931,
-      "grad_norm": 9.145533561706543,
+      "grad_norm": 9.148017883300781,
       "learning_rate": 7.680608365019012e-06,
-      "loss": 1.6651,
+      "loss": 1.6624,
       "step": 189
     },
     {
       "epoch": 0.4791929382093317,
-      "grad_norm": 8.419795989990234,
+      "grad_norm": 8.401142120361328,
       "learning_rate": 7.667934093789608e-06,
-      "loss": 1.6928,
+      "loss": 1.691,
       "step": 190
     },
     {
       "epoch": 0.4817150063051702,
-      "grad_norm": 9.782970428466797,
+      "grad_norm": 9.830410957336426,
       "learning_rate": 7.655259822560204e-06,
-      "loss": 1.9865,
+      "loss": 1.9906,
       "step": 191
     },
     {
       "epoch": 0.4842370744010088,
-      "grad_norm": 8.885458946228027,
+      "grad_norm": 8.865063667297363,
       "learning_rate": 7.642585551330799e-06,
-      "loss": 1.9461,
+      "loss": 1.9455,
       "step": 192
     },
     {
       "epoch": 0.48675914249684743,
-      "grad_norm": 10.114888191223145,
+      "grad_norm": 10.0823335647583,
       "learning_rate": 7.629911280101395e-06,
-      "loss": 1.6703,
+      "loss": 1.6741,
       "step": 193
     },
     {
       "epoch": 0.489281210592686,
-      "grad_norm": 9.351268768310547,
+      "grad_norm": 9.399166107177734,
       "learning_rate": 7.617237008871991e-06,
-      "loss": 1.8674,
+      "loss": 1.8719,
       "step": 194
     },
     {
       "epoch": 0.4918032786885246,
-      "grad_norm": 9.144396781921387,
+      "grad_norm": 9.091114044189453,
       "learning_rate": 7.604562737642585e-06,
-      "loss": 2.0062,
+      "loss": 2.0057,
       "step": 195
     },
     {
       "epoch": 0.4943253467843632,
-      "grad_norm": 9.084301948547363,
+      "grad_norm": 9.053544044494629,
       "learning_rate": 7.591888466413182e-06,
-      "loss": 2.015,
+      "loss": 2.0169,
       "step": 196
     },
     {
       "epoch": 0.4968474148802018,
-      "grad_norm": 8.582064628601074,
+      "grad_norm": 8.5532808303833,
       "learning_rate": 7.5792141951837775e-06,
-      "loss": 1.8281,
+      "loss": 1.8269,
       "step": 197
     },
     {
       "epoch": 0.49936948297604034,
-      "grad_norm": 8.800094604492188,
+      "grad_norm": 8.772011756896973,
       "learning_rate": 7.566539923954374e-06,
-      "loss": 1.7135,
+      "loss": 1.7108,
       "step": 198
     },
     {
       "epoch": 0.501891551071879,
-      "grad_norm": 9.495306015014648,
+      "grad_norm": 9.473034858703613,
       "learning_rate": 7.553865652724969e-06,
       "loss": 1.8995,
       "step": 199
     },
     {
       "epoch": 0.5044136191677175,
-      "grad_norm": 9.089255332946777,
+      "grad_norm": 9.129395484924316,
       "learning_rate": 7.541191381495565e-06,
-      "loss": 1.8346,
+      "loss": 1.8357,
       "step": 200
     },
     {
       "epoch": 0.5044136191677175,
-      "eval_loss": 1.7910906076431274,
-      "eval_runtime": 16.2728,
-      "eval_samples_per_second": 43.324,
-      "eval_steps_per_second": 21.693,
+      "eval_loss": 1.7911782264709473,
+      "eval_runtime": 12.345,
+      "eval_samples_per_second": 57.108,
+      "eval_steps_per_second": 28.595,
       "step": 200
     }
   ],

checkpoint-200/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:71daefb40029d746ea8c935985a9ef6300d5a3ab9638c111b5d4090b488a0f6f
+oid sha256:e3f78bb4001883b46b552ecb677bd27981f7002004fb9b300c25d87134402e93
 size 6033