diff --git "a/checkpoints/checkpoint-170/trainer_state.json" "b/checkpoints/checkpoint-170/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoints/checkpoint-170/trainer_state.json"
@@ -0,0 +1,3434 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.34,
+  "eval_steps": 500,
+  "global_step": 170,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002,
+      "gl_detached_lm_loss": 0.4408648908138275,
+      "gl_distil_loss": 0.0004234774096403271,
+      "gl_dms_closed_frac": 0.002307269489392638,
+      "gl_dms_cr": 1.0023125410079956,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.0,
+      "gl_dms_target_frac": 0.0,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0004234774096403271,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 0.08518081903457642,
+      "learning_rate": 3e-05,
+      "loss": 0.0004,
+      "step": 1
+    },
+    {
+      "epoch": 0.004,
+      "gl_detached_lm_loss": 0.47212445735931396,
+      "gl_distil_loss": 0.0008152268128469586,
+      "gl_dms_closed_frac": 0.0022805265616625547,
+      "gl_dms_cr": 1.0022858381271362,
+      "gl_dms_loss": 0.026290902867913246,
+      "gl_dms_target_cr": 1.029411792755127,
+      "gl_dms_target_frac": 0.02857142873108387,
+      "gl_eos_tokens": 12.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.02710612863302231,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.5,
+      "gl_positions_without_loss_calculation": 12.5,
+      "grad_norm": 2.49080753326416,
+      "learning_rate": 3e-05,
+      "loss": 0.0271,
+      "step": 2
+    },
+    {
+      "epoch": 0.006,
+      "gl_detached_lm_loss": 0.4552858769893646,
+      "gl_distil_loss": 0.0005731006385758519,
+      "gl_dms_closed_frac": 0.011005931533873081,
+      "gl_dms_cr": 1.0111286640167236,
+      "gl_dms_loss": 0.04454962536692619,
+      "gl_dms_target_cr": 1.058823585510254,
+      "gl_dms_target_frac": 0.0555555559694767,
+      "gl_eos_tokens": 10.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.045122724026441574,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.5,
+      "gl_positions_without_loss_calculation": 10.5,
+      "grad_norm": 5.462989330291748,
+      "learning_rate": 3e-05,
+      "loss": 0.0451,
+      "step": 3
+    },
+    {
+      "epoch": 0.008,
+      "gl_detached_lm_loss": 0.4574431777000427,
+      "gl_distil_loss": 0.0012222880031913519,
+      "gl_dms_closed_frac": 0.04424896836280823,
+      "gl_dms_cr": 1.0463114976882935,
+      "gl_dms_loss": 0.036832112818956375,
+      "gl_dms_target_cr": 1.0882352590560913,
+      "gl_dms_target_frac": 0.0810810849070549,
+      "gl_eos_tokens": 9.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.03805439546704292,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.5,
+      "gl_positions_without_loss_calculation": 9.5,
+      "grad_norm": 15.773651123046875,
+      "learning_rate": 3e-05,
+      "loss": 0.0381,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "gl_detached_lm_loss": 0.48430225253105164,
+      "gl_distil_loss": 0.0016105370596051216,
+      "gl_dms_closed_frac": 0.1391979455947876,
+      "gl_dms_cr": 1.1620138883590698,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.1176470518112183,
+      "gl_dms_target_frac": 0.10526316612958908,
+      "gl_eos_tokens": 13.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0016105370596051216,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32754.75,
+      "gl_positions_without_loss_calculation": 13.25,
+      "grad_norm": 1.219625473022461,
+      "learning_rate": 3e-05,
+      "loss": 0.0016,
+      "step": 5
+    },
+    {
+      "epoch": 0.012,
+      "gl_detached_lm_loss": 0.45663878321647644,
+      "gl_distil_loss": 0.003948591183871031,
+      "gl_dms_closed_frac": 0.2567480504512787,
+      "gl_dms_cr": 1.3462255001068115,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.1470588445663452,
+      "gl_dms_target_frac": 0.12820513546466827,
+      "gl_eos_tokens": 13.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.003948591183871031,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32754.25,
+      "gl_positions_without_loss_calculation": 13.75,
+      "grad_norm": 6.573792457580566,
+      "learning_rate": 3e-05,
+      "loss": 0.0039,
+      "step": 6
+    },
+    {
+      "epoch": 0.014,
+      "gl_detached_lm_loss": 0.4690469801425934,
+      "gl_distil_loss": 0.005540270358324051,
+      "gl_dms_closed_frac": 0.34524545073509216,
+      "gl_dms_cr": 1.5293982028961182,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.1764706373214722,
+      "gl_dms_target_frac": 0.14999999105930328,
+      "gl_eos_tokens": 10.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005540270358324051,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.25,
+      "gl_positions_without_loss_calculation": 10.75,
+      "grad_norm": 4.326766014099121,
+      "learning_rate": 3e-05,
+      "loss": 0.0055,
+      "step": 7
+    },
+    {
+      "epoch": 0.016,
+      "gl_detached_lm_loss": 0.4756607413291931,
+      "gl_distil_loss": 0.00488277105614543,
+      "gl_dms_closed_frac": 0.41206902265548706,
+      "gl_dms_cr": 1.7022228240966797,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.20588219165802,
+      "gl_dms_target_frac": 0.17073169350624084,
+      "gl_eos_tokens": 11.625,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.00488277105614543,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.375,
+      "gl_positions_without_loss_calculation": 11.625,
+      "grad_norm": 2.491706132888794,
+      "learning_rate": 3e-05,
+      "loss": 0.0049,
+      "step": 8
+    },
+    {
+      "epoch": 0.018,
+      "gl_detached_lm_loss": 0.47128596901893616,
+      "gl_distil_loss": 0.004345947876572609,
+      "gl_dms_closed_frac": 0.43498629331588745,
+      "gl_dms_cr": 1.7745075225830078,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.235294222831726,
+      "gl_dms_target_frac": 0.190476194024086,
+      "gl_eos_tokens": 10.625,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.004345947876572609,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.375,
+      "gl_positions_without_loss_calculation": 10.625,
+      "grad_norm": 4.466060161590576,
+      "learning_rate": 3e-05,
+      "loss": 0.0043,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "gl_detached_lm_loss": 0.47395652532577515,
+      "gl_distil_loss": 0.005074856802821159,
+      "gl_dms_closed_frac": 0.43777039647102356,
+      "gl_dms_cr": 1.780246376991272,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.264705777168274,
+      "gl_dms_target_frac": 0.20930232107639313,
+      "gl_eos_tokens": 10.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005074856802821159,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.25,
+      "gl_positions_without_loss_calculation": 10.75,
+      "grad_norm": 7.389776229858398,
+      "learning_rate": 3e-05,
+      "loss": 0.0051,
+      "step": 10
+    },
+    {
+      "epoch": 0.022,
+      "gl_detached_lm_loss": 0.4599134922027588,
+      "gl_distil_loss": 0.004142467863857746,
+      "gl_dms_closed_frac": 0.4546947777271271,
+      "gl_dms_cr": 1.8353639841079712,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.29411780834198,
+      "gl_dms_target_frac": 0.22727273404598236,
+      "gl_eos_tokens": 10.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.004142467863857746,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.75,
+      "gl_positions_without_loss_calculation": 10.25,
+      "grad_norm": 5.261406421661377,
+      "learning_rate": 3e-05,
+      "loss": 0.0041,
+      "step": 11
+    },
+    {
+      "epoch": 0.024,
+      "gl_detached_lm_loss": 0.47876226902008057,
+      "gl_distil_loss": 0.0035050029400736094,
+      "gl_dms_closed_frac": 0.43767115473747253,
+      "gl_dms_cr": 1.779110312461853,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.3235293626785278,
+      "gl_dms_target_frac": 0.24444445967674255,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0035050029400736094,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 1.726235032081604,
+      "learning_rate": 3e-05,
+      "loss": 0.0035,
+      "step": 12
+    },
+    {
+      "epoch": 0.026,
+      "gl_detached_lm_loss": 0.47314175963401794,
+      "gl_distil_loss": 0.00313362805172801,
+      "gl_dms_closed_frac": 0.42793411016464233,
+      "gl_dms_cr": 1.7503966093063354,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.3529412746429443,
+      "gl_dms_target_frac": 0.2608695328235626,
+      "gl_eos_tokens": 9.125,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.00313362805172801,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.875,
+      "gl_positions_without_loss_calculation": 9.125,
+      "grad_norm": 1.3173593282699585,
+      "learning_rate": 3e-05,
+      "loss": 0.0031,
+      "step": 13
+    },
+    {
+      "epoch": 0.028,
+      "gl_detached_lm_loss": 0.46165570616722107,
+      "gl_distil_loss": 0.0026301750913262367,
+      "gl_dms_closed_frac": 0.4108446538448334,
+      "gl_dms_cr": 1.6978795528411865,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.3823529481887817,
+      "gl_dms_target_frac": 0.27659574151039124,
+      "gl_eos_tokens": 12.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0026301750913262367,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.75,
+      "gl_positions_without_loss_calculation": 12.25,
+      "grad_norm": 2.1748335361480713,
+      "learning_rate": 3e-05,
+      "loss": 0.0026,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "gl_detached_lm_loss": 0.47738948464393616,
+      "gl_distil_loss": 0.00245979567989707,
+      "gl_dms_closed_frac": 0.3945341110229492,
+      "gl_dms_cr": 1.6518938541412354,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.4117648601531982,
+      "gl_dms_target_frac": 0.2916666567325592,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.00245979567989707,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 1.0410534143447876,
+      "learning_rate": 3e-05,
+      "loss": 0.0025,
+      "step": 15
+    },
+    {
+      "epoch": 0.032,
+      "gl_detached_lm_loss": 0.4363529086112976,
+      "gl_distil_loss": 0.002206941368058324,
+      "gl_dms_closed_frac": 0.3827410936355591,
+      "gl_dms_cr": 1.6204078197479248,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.441176414489746,
+      "gl_dms_target_frac": 0.30612245202064514,
+      "gl_eos_tokens": 10.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.002206941368058324,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.75,
+      "gl_positions_without_loss_calculation": 10.25,
+      "grad_norm": 1.933333396911621,
+      "learning_rate": 3e-05,
+      "loss": 0.0022,
+      "step": 16
+    },
+    {
+      "epoch": 0.034,
+      "gl_detached_lm_loss": 0.4372034966945648,
+      "gl_distil_loss": 0.0019222048576921225,
+      "gl_dms_closed_frac": 0.3600853383541107,
+      "gl_dms_cr": 1.563251256942749,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.470588207244873,
+      "gl_dms_target_frac": 0.31999996304512024,
+      "gl_eos_tokens": 10.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0019222048576921225,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.0,
+      "gl_positions_without_loss_calculation": 10.0,
+      "grad_norm": 1.3154582977294922,
+      "learning_rate": 3e-05,
+      "loss": 0.0019,
+      "step": 17
+    },
+    {
+      "epoch": 0.036,
+      "gl_detached_lm_loss": 0.4272920787334442,
+      "gl_distil_loss": 0.0017361550126224756,
+      "gl_dms_closed_frac": 0.3457123935222626,
+      "gl_dms_cr": 1.5294559001922607,
+      "gl_dms_loss": 0.003807865083217621,
+      "gl_dms_target_cr": 1.5,
+      "gl_dms_target_frac": 0.3333333134651184,
+      "gl_eos_tokens": 9.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.00554402032867074,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.25,
+      "gl_positions_without_loss_calculation": 9.75,
+      "grad_norm": 3.785482168197632,
+      "learning_rate": 3e-05,
+      "loss": 0.0055,
+      "step": 18
+    },
+    {
+      "epoch": 0.038,
+      "gl_detached_lm_loss": 0.4550657570362091,
+      "gl_distil_loss": 0.0017214431427419186,
+      "gl_dms_closed_frac": 0.3477359712123871,
+      "gl_dms_cr": 1.533597707748413,
+      "gl_dms_loss": 0.0037946999073028564,
+      "gl_dms_target_cr": 1.529411792755127,
+      "gl_dms_target_frac": 0.3461538553237915,
+      "gl_eos_tokens": 12.625,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005516142584383488,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.375,
+      "gl_positions_without_loss_calculation": 12.625,
+      "grad_norm": 7.753348350524902,
+      "learning_rate": 3e-05,
+      "loss": 0.0055,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "gl_detached_lm_loss": 0.4646049737930298,
+      "gl_distil_loss": 0.0019962999504059553,
+      "gl_dms_closed_frac": 0.36188656091690063,
+      "gl_dms_cr": 1.5680888891220093,
+      "gl_dms_loss": 0.0048120878636837006,
+      "gl_dms_target_cr": 1.558823585510254,
+      "gl_dms_target_frac": 0.358490526676178,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0068083880469202995,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 15.435755729675293,
+      "learning_rate": 3e-05,
+      "loss": 0.0068,
+      "step": 20
+    },
+    {
+      "epoch": 0.042,
+      "gl_detached_lm_loss": 0.45961058139801025,
+      "gl_distil_loss": 0.001984530594199896,
+      "gl_dms_closed_frac": 0.395085871219635,
+      "gl_dms_cr": 1.653983473777771,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.5882351398468018,
+      "gl_dms_target_frac": 0.37037035822868347,
+      "gl_eos_tokens": 10.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.001984530594199896,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.25,
+      "gl_positions_without_loss_calculation": 10.75,
+      "grad_norm": 0.7058410048484802,
+      "learning_rate": 3e-05,
+      "loss": 0.002,
+      "step": 21
+    },
+    {
+      "epoch": 0.044,
+      "gl_detached_lm_loss": 0.4503985643386841,
+      "gl_distil_loss": 0.002159471856430173,
+      "gl_dms_closed_frac": 0.4215147793292999,
+      "gl_dms_cr": 1.7290366888046265,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.6176470518112183,
+      "gl_dms_target_frac": 0.38181817531585693,
+      "gl_eos_tokens": 10.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.002159471856430173,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.0,
+      "gl_positions_without_loss_calculation": 10.0,
+      "grad_norm": 1.5459787845611572,
+      "learning_rate": 3e-05,
+      "loss": 0.0022,
+      "step": 22
+    },
+    {
+      "epoch": 0.046,
+      "gl_detached_lm_loss": 0.5068718194961548,
+      "gl_distil_loss": 0.0023190875072032213,
+      "gl_dms_closed_frac": 0.4291229844093323,
+      "gl_dms_cr": 1.7522425651550293,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.6470587253570557,
+      "gl_dms_target_frac": 0.3928571045398712,
+      "gl_eos_tokens": 12.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0023190875072032213,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.5,
+      "gl_positions_without_loss_calculation": 12.5,
+      "grad_norm": 1.3756858110427856,
+      "learning_rate": 3e-05,
+      "loss": 0.0023,
+      "step": 23
+    },
+    {
+      "epoch": 0.048,
+      "gl_detached_lm_loss": 0.4678501486778259,
+      "gl_distil_loss": 0.0021665464155375957,
+      "gl_dms_closed_frac": 0.44796285033226013,
+      "gl_dms_cr": 1.8125988245010376,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.6764706373214722,
+      "gl_dms_target_frac": 0.403508722782135,
+      "gl_eos_tokens": 11.375,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0021665464155375957,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.625,
+      "gl_positions_without_loss_calculation": 11.375,
+      "grad_norm": 0.5030384063720703,
+      "learning_rate": 3e-05,
+      "loss": 0.0022,
+      "step": 24
+    },
+    {
+      "epoch": 0.05,
+      "gl_detached_lm_loss": 0.46251434087753296,
+      "gl_distil_loss": 0.0021796354558318853,
+      "gl_dms_closed_frac": 0.4480258822441101,
+      "gl_dms_cr": 1.8118644952774048,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.70588219165802,
+      "gl_dms_target_frac": 0.4137931168079376,
+      "gl_eos_tokens": 11.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0021796354558318853,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.25,
+      "gl_positions_without_loss_calculation": 11.75,
+      "grad_norm": 0.4940311908721924,
+      "learning_rate": 3e-05,
+      "loss": 0.0022,
+      "step": 25
+    },
+    {
+      "epoch": 0.052,
+      "gl_detached_lm_loss": 0.5336480736732483,
+      "gl_distil_loss": 0.0026741931214928627,
+      "gl_dms_closed_frac": 0.4484004080295563,
+      "gl_dms_cr": 1.8130683898925781,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.735294222831726,
+      "gl_dms_target_frac": 0.423728883266449,
+      "gl_eos_tokens": 10.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0026741931214928627,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.25,
+      "gl_positions_without_loss_calculation": 10.75,
+      "grad_norm": 1.8604220151901245,
+      "learning_rate": 3e-05,
+      "loss": 0.0027,
+      "step": 26
+    },
+    {
+      "epoch": 0.054,
+      "gl_detached_lm_loss": 0.49125808477401733,
+      "gl_distil_loss": 0.0021746917627751827,
+      "gl_dms_closed_frac": 0.45190396904945374,
+      "gl_dms_cr": 1.825452208518982,
+      "gl_dms_loss": 0.0002978183329105377,
+      "gl_dms_target_cr": 1.764705777168274,
+      "gl_dms_target_frac": 0.4333333671092987,
+      "gl_eos_tokens": 10.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0024725100956857204,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.0,
+      "gl_positions_without_loss_calculation": 10.0,
+      "grad_norm": 3.132777452468872,
+      "learning_rate": 3e-05,
+      "loss": 0.0025,
+      "step": 27
+    },
+    {
+      "epoch": 0.056,
+      "gl_detached_lm_loss": 0.44300004839897156,
+      "gl_distil_loss": 0.002110114088281989,
+      "gl_dms_closed_frac": 0.4698338508605957,
+      "gl_dms_cr": 1.8869034051895142,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.79411780834198,
+      "gl_dms_target_frac": 0.44262292981147766,
+      "gl_eos_tokens": 10.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.002110114088281989,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.0,
+      "gl_positions_without_loss_calculation": 10.0,
+      "grad_norm": 0.48894694447517395,
+      "learning_rate": 3e-05,
+      "loss": 0.0021,
+      "step": 28
+    },
+    {
+      "epoch": 0.058,
+      "gl_detached_lm_loss": 0.4883745610713959,
+      "gl_distil_loss": 0.0021808757446706295,
+      "gl_dms_closed_frac": 0.4731365442276001,
+      "gl_dms_cr": 1.8983519077301025,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.8235293626785278,
+      "gl_dms_target_frac": 0.4516129195690155,
+      "gl_eos_tokens": 11.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0021808757446706295,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.25,
+      "gl_positions_without_loss_calculation": 11.75,
+      "grad_norm": 0.7062388062477112,
+      "learning_rate": 3e-05,
+      "loss": 0.0022,
+      "step": 29
+    },
+    {
+      "epoch": 0.06,
+      "gl_detached_lm_loss": 0.46343907713890076,
+      "gl_distil_loss": 0.002232843078672886,
+      "gl_dms_closed_frac": 0.48099470138549805,
+      "gl_dms_cr": 1.92780601978302,
+      "gl_dms_loss": 0.00032526254653930664,
+      "gl_dms_target_cr": 1.8529412746429443,
+      "gl_dms_target_frac": 0.4603174328804016,
+      "gl_eos_tokens": 11.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0025581056252121925,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.0,
+      "gl_positions_without_loss_calculation": 11.0,
+      "grad_norm": 3.083817958831787,
+      "learning_rate": 3e-05,
+      "loss": 0.0026,
+      "step": 30
+    },
+    {
+      "epoch": 0.062,
+      "gl_detached_lm_loss": 0.4522785544395447,
+      "gl_distil_loss": 0.0021007806062698364,
+      "gl_dms_closed_frac": 0.4981619715690613,
+      "gl_dms_cr": 1.9931349754333496,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.8823529481887817,
+      "gl_dms_target_frac": 0.46875,
+      "gl_eos_tokens": 11.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0021007806062698364,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.75,
+      "gl_positions_without_loss_calculation": 11.25,
+      "grad_norm": 0.6077609062194824,
+      "learning_rate": 3e-05,
+      "loss": 0.0021,
+      "step": 31
+    },
+    {
+      "epoch": 0.064,
+      "gl_detached_lm_loss": 0.4637286067008972,
+      "gl_distil_loss": 0.002412236761301756,
+      "gl_dms_closed_frac": 0.5112192034721375,
+      "gl_dms_cr": 2.046431064605713,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.9117648601531982,
+      "gl_dms_target_frac": 0.47692304849624634,
+      "gl_eos_tokens": 10.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.002412236761301756,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.5,
+      "gl_positions_without_loss_calculation": 10.5,
+      "grad_norm": 0.4839731752872467,
+      "learning_rate": 3e-05,
+      "loss": 0.0024,
+      "step": 32
+    },
+    {
+      "epoch": 0.066,
+      "gl_detached_lm_loss": 0.48589426279067993,
+      "gl_distil_loss": 0.0024046343751251698,
+      "gl_dms_closed_frac": 0.5115869045257568,
+      "gl_dms_cr": 2.0482912063598633,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.941176414489746,
+      "gl_dms_target_frac": 0.4848484992980957,
+      "gl_eos_tokens": 11.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0024046343751251698,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.0,
+      "gl_positions_without_loss_calculation": 11.0,
+      "grad_norm": 0.4125555157661438,
+      "learning_rate": 3e-05,
+      "loss": 0.0024,
+      "step": 33
+    },
+    {
+      "epoch": 0.068,
+      "gl_detached_lm_loss": 0.4837048351764679,
+      "gl_distil_loss": 0.0023821976501494646,
+      "gl_dms_closed_frac": 0.5104748606681824,
+      "gl_dms_cr": 2.0433568954467773,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 1.9705884456634521,
+      "gl_dms_target_frac": 0.4925372898578644,
+      "gl_eos_tokens": 12.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0023821976501494646,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.0,
+      "gl_positions_without_loss_calculation": 12.0,
+      "grad_norm": 0.4393292963504791,
+      "learning_rate": 3e-05,
+      "loss": 0.0024,
+      "step": 34
+    },
+    {
+      "epoch": 0.07,
+      "gl_detached_lm_loss": 0.45272281765937805,
+      "gl_distil_loss": 0.002244414296001196,
+      "gl_dms_closed_frac": 0.5154764652252197,
+      "gl_dms_cr": 2.0640711784362793,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.0,
+      "gl_dms_target_frac": 0.5,
+      "gl_eos_tokens": 12.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.002244414296001196,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.0,
+      "gl_positions_without_loss_calculation": 12.0,
+      "grad_norm": 0.6267670392990112,
+      "learning_rate": 3e-05,
+      "loss": 0.0022,
+      "step": 35
+    },
+    {
+      "epoch": 0.072,
+      "gl_detached_lm_loss": 0.4794645309448242,
+      "gl_distil_loss": 0.0021888441406190395,
+      "gl_dms_closed_frac": 0.5047264099121094,
+      "gl_dms_cr": 2.0192503929138184,
+      "gl_dms_loss": 0.00308087095618248,
+      "gl_dms_target_cr": 2.029411554336548,
+      "gl_dms_target_frac": 0.5072464346885681,
+      "gl_eos_tokens": 9.875,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005269715096801519,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.125,
+      "gl_positions_without_loss_calculation": 9.875,
+      "grad_norm": 17.163253784179688,
+      "learning_rate": 3e-05,
+      "loss": 0.0053,
+      "step": 36
+    },
+    {
+      "epoch": 0.074,
+      "gl_detached_lm_loss": 0.4689008891582489,
+      "gl_distil_loss": 0.0023856312036514282,
+      "gl_dms_closed_frac": 0.5187268257141113,
+      "gl_dms_cr": 2.077929973602295,
+      "gl_dms_loss": 0.0002900660037994385,
+      "gl_dms_target_cr": 2.058823585510254,
+      "gl_dms_target_frac": 0.5142857432365417,
+      "gl_eos_tokens": 10.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0026756974402815104,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.75,
+      "gl_positions_without_loss_calculation": 10.25,
+      "grad_norm": 2.661439895629883,
+      "learning_rate": 3e-05,
+      "loss": 0.0027,
+      "step": 37
+    },
+    {
+      "epoch": 0.076,
+      "gl_detached_lm_loss": 0.44916558265686035,
+      "gl_distil_loss": 0.0023738506715744734,
+      "gl_dms_closed_frac": 0.538723349571228,
+      "gl_dms_cr": 2.168072462081909,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.08823561668396,
+      "gl_dms_target_frac": 0.5211267471313477,
+      "gl_eos_tokens": 10.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0023738506715744734,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.75,
+      "gl_positions_without_loss_calculation": 10.25,
+      "grad_norm": 0.7621412873268127,
+      "learning_rate": 3e-05,
+      "loss": 0.0024,
+      "step": 38
+    },
+    {
+      "epoch": 0.078,
+      "gl_detached_lm_loss": 0.464703232049942,
+      "gl_distil_loss": 0.0026064212433993816,
+      "gl_dms_closed_frac": 0.5474792122840881,
+      "gl_dms_cr": 2.210324764251709,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.117647171020508,
+      "gl_dms_target_frac": 0.5277777314186096,
+      "gl_eos_tokens": 12.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0026064212433993816,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.0,
+      "gl_positions_without_loss_calculation": 12.0,
+      "grad_norm": 1.6320092678070068,
+      "learning_rate": 3e-05,
+      "loss": 0.0026,
+      "step": 39
+    },
+    {
+      "epoch": 0.08,
+      "gl_detached_lm_loss": 0.4954476058483124,
+      "gl_distil_loss": 0.002672001253813505,
+      "gl_dms_closed_frac": 0.5522975921630859,
+      "gl_dms_cr": 2.233839273452759,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.1470587253570557,
+      "gl_dms_target_frac": 0.5342465043067932,
+      "gl_eos_tokens": 12.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.002672001253813505,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.75,
+      "gl_positions_without_loss_calculation": 12.25,
+      "grad_norm": 0.5290228128433228,
+      "learning_rate": 3e-05,
+      "loss": 0.0027,
+      "step": 40
+    },
+    {
+      "epoch": 0.082,
+      "gl_detached_lm_loss": 0.46069520711898804,
+      "gl_distil_loss": 0.0026594277005642653,
+      "gl_dms_closed_frac": 0.5577877759933472,
+      "gl_dms_cr": 2.261887788772583,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.1764705181121826,
+      "gl_dms_target_frac": 0.5405405163764954,
+      "gl_eos_tokens": 9.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0026594277005642653,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.25,
+      "gl_positions_without_loss_calculation": 9.75,
+      "grad_norm": 0.593712329864502,
+      "learning_rate": 3e-05,
+      "loss": 0.0027,
+      "step": 41
+    },
+    {
+      "epoch": 0.084,
+      "gl_detached_lm_loss": 0.4498485028743744,
+      "gl_distil_loss": 0.0026302323676645756,
+      "gl_dms_closed_frac": 0.5595740675926208,
+      "gl_dms_cr": 2.2711756229400635,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.2058823108673096,
+      "gl_dms_target_frac": 0.54666668176651,
+      "gl_eos_tokens": 11.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0026302323676645756,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.25,
+      "gl_positions_without_loss_calculation": 11.75,
+      "grad_norm": 1.180981159210205,
+      "learning_rate": 3e-05,
+      "loss": 0.0026,
+      "step": 42
+    },
+    {
+      "epoch": 0.086,
+      "gl_detached_lm_loss": 0.44525861740112305,
+      "gl_distil_loss": 0.0024244743399322033,
+      "gl_dms_closed_frac": 0.5586432814598083,
+      "gl_dms_cr": 2.266235828399658,
+      "gl_dms_loss": 0.0005865171551704407,
+      "gl_dms_target_cr": 2.2352941036224365,
+      "gl_dms_target_frac": 0.5526315569877625,
+      "gl_eos_tokens": 11.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.003010991495102644,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.25,
+      "gl_positions_without_loss_calculation": 11.75,
+      "grad_norm": 2.4405698776245117,
+      "learning_rate": 3e-05,
+      "loss": 0.003,
+      "step": 43
+    },
+    {
+      "epoch": 0.088,
+      "gl_detached_lm_loss": 0.45686984062194824,
+      "gl_distil_loss": 0.002607216825708747,
+      "gl_dms_closed_frac": 0.5671321749687195,
+      "gl_dms_cr": 2.3103342056274414,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.2647058963775635,
+      "gl_dms_target_frac": 0.5584415793418884,
+      "gl_eos_tokens": 11.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.002607216825708747,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.0,
+      "gl_positions_without_loss_calculation": 11.0,
+      "grad_norm": 1.1561832427978516,
+      "learning_rate": 3e-05,
+      "loss": 0.0026,
+      "step": 44
+    },
+    {
+      "epoch": 0.09,
+      "gl_detached_lm_loss": 0.47623857855796814,
+      "gl_distil_loss": 0.0026032254099845886,
+      "gl_dms_closed_frac": 0.5701911449432373,
+      "gl_dms_cr": 2.326831102371216,
+      "gl_dms_loss": 8.52048397064209e-05,
+      "gl_dms_target_cr": 2.2941176891326904,
+      "gl_dms_target_frac": 0.5641025900840759,
+      "gl_eos_tokens": 10.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0026884302496910095,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.0,
+      "gl_positions_without_loss_calculation": 10.0,
+      "grad_norm": 2.5896213054656982,
+      "learning_rate": 3e-05,
+      "loss": 0.0027,
+      "step": 45
+    },
+    {
+      "epoch": 0.092,
+      "gl_detached_lm_loss": 0.4935804307460785,
+      "gl_distil_loss": 0.002987999701872468,
+      "gl_dms_closed_frac": 0.581356406211853,
+      "gl_dms_cr": 2.389448881149292,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.3235294818878174,
+      "gl_dms_target_frac": 0.5696201920509338,
+      "gl_eos_tokens": 10.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.002987999701872468,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.5,
+      "gl_positions_without_loss_calculation": 10.5,
+      "grad_norm": 0.5520856380462646,
+      "learning_rate": 3e-05,
+      "loss": 0.003,
+      "step": 46
+    },
+    {
+      "epoch": 0.094,
+      "gl_detached_lm_loss": 0.47648417949676514,
+      "gl_distil_loss": 0.0030070601496845484,
+      "gl_dms_closed_frac": 0.5908368229866028,
+      "gl_dms_cr": 2.444538116455078,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.3529412746429443,
+      "gl_dms_target_frac": 0.574999988079071,
+      "gl_eos_tokens": 11.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0030070601496845484,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.0,
+      "gl_positions_without_loss_calculation": 11.0,
+      "grad_norm": 0.5767875909805298,
+      "learning_rate": 3e-05,
+      "loss": 0.003,
+      "step": 47
+    },
+    {
+      "epoch": 0.096,
+      "gl_detached_lm_loss": 0.4622802734375,
+      "gl_distil_loss": 0.002870834432542324,
+      "gl_dms_closed_frac": 0.5926159620285034,
+      "gl_dms_cr": 2.454831600189209,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.382352828979492,
+      "gl_dms_target_frac": 0.5802469253540039,
+      "gl_eos_tokens": 12.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.002870834432542324,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.75,
+      "gl_positions_without_loss_calculation": 12.25,
+      "grad_norm": 1.0462926626205444,
+      "learning_rate": 3e-05,
+      "loss": 0.0029,
+      "step": 48
+    },
+    {
+      "epoch": 0.098,
+      "gl_detached_lm_loss": 0.4593689739704132,
+      "gl_distil_loss": 0.0029501053504645824,
+      "gl_dms_closed_frac": 0.5951315760612488,
+      "gl_dms_cr": 2.4700350761413574,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.41176438331604,
+      "gl_dms_target_frac": 0.5853658318519592,
+      "gl_eos_tokens": 12.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0029501053504645824,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.25,
+      "gl_positions_without_loss_calculation": 12.75,
+      "grad_norm": 0.8309232592582703,
+      "learning_rate": 3e-05,
+      "loss": 0.003,
+      "step": 49
+    },
+    {
+      "epoch": 0.1,
+      "gl_detached_lm_loss": 0.46503978967666626,
+      "gl_distil_loss": 0.002882842207327485,
+      "gl_dms_closed_frac": 0.5900493860244751,
+      "gl_dms_cr": 2.4397754669189453,
+      "gl_dms_loss": 0.0024773404002189636,
+      "gl_dms_target_cr": 2.441176414489746,
+      "gl_dms_target_frac": 0.5903614163398743,
+      "gl_eos_tokens": 11.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005360182840377092,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.0,
+      "gl_positions_without_loss_calculation": 11.0,
+      "grad_norm": 9.821179389953613,
+      "learning_rate": 3e-05,
+      "loss": 0.0054,
+      "step": 50
+    },
+    {
+      "epoch": 0.102,
+      "gl_detached_lm_loss": 0.47805801033973694,
+      "gl_distil_loss": 0.0031700131949037313,
+      "gl_dms_closed_frac": 0.5995256304740906,
+      "gl_dms_cr": 2.4975829124450684,
+      "gl_dms_loss": 0.0009735524654388428,
+      "gl_dms_target_cr": 2.470588445663452,
+      "gl_dms_target_frac": 0.5952381491661072,
+      "gl_eos_tokens": 9.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.004143565893173218,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32759.0,
+      "gl_positions_without_loss_calculation": 9.0,
+      "grad_norm": 5.710433006286621,
+      "learning_rate": 3e-05,
+      "loss": 0.0041,
+      "step": 51
+    },
+    {
+      "epoch": 0.104,
+      "gl_detached_lm_loss": 0.4751797914505005,
+      "gl_distil_loss": 0.0030524739995598793,
+      "gl_dms_closed_frac": 0.612511157989502,
+      "gl_dms_cr": 2.581015110015869,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.5,
+      "gl_dms_target_frac": 0.5999999642372131,
+      "gl_eos_tokens": 10.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0030524739995598793,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.5,
+      "gl_positions_without_loss_calculation": 10.5,
+      "grad_norm": 0.5378996133804321,
+      "learning_rate": 3e-05,
+      "loss": 0.0031,
+      "step": 52
+    },
+    {
+      "epoch": 0.106,
+      "gl_detached_lm_loss": 0.47998565435409546,
+      "gl_distil_loss": 0.0036123180761933327,
+      "gl_dms_closed_frac": 0.6233031749725342,
+      "gl_dms_cr": 2.655303955078125,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.529411554336548,
+      "gl_dms_target_frac": 0.604651153087616,
+      "gl_eos_tokens": 9.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0036123180761933327,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.25,
+      "gl_positions_without_loss_calculation": 9.75,
+      "grad_norm": 0.7021176218986511,
+      "learning_rate": 3e-05,
+      "loss": 0.0036,
+      "step": 53
+    },
+    {
+      "epoch": 0.108,
+      "gl_detached_lm_loss": 0.4588249623775482,
+      "gl_distil_loss": 0.0031498277094215155,
+      "gl_dms_closed_frac": 0.6244292855262756,
+      "gl_dms_cr": 2.6633052825927734,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.558823585510254,
+      "gl_dms_target_frac": 0.6091954112052917,
+      "gl_eos_tokens": 13.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0031498277094215155,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.0,
+      "gl_positions_without_loss_calculation": 13.0,
+      "grad_norm": 0.5357586741447449,
+      "learning_rate": 3e-05,
+      "loss": 0.0031,
+      "step": 54
+    },
+    {
+      "epoch": 0.11,
+      "gl_detached_lm_loss": 0.45006063580513,
+      "gl_distil_loss": 0.0030964650213718414,
+      "gl_dms_closed_frac": 0.6261225342750549,
+      "gl_dms_cr": 2.6749215126037598,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.58823561668396,
+      "gl_dms_target_frac": 0.6136364340782166,
+      "gl_eos_tokens": 9.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0030964650213718414,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.25,
+      "gl_positions_without_loss_calculation": 9.75,
+      "grad_norm": 0.5736832022666931,
+      "learning_rate": 3e-05,
+      "loss": 0.0031,
+      "step": 55
+    },
+    {
+      "epoch": 0.112,
+      "gl_detached_lm_loss": 0.4804791212081909,
+      "gl_distil_loss": 0.003160958644002676,
+      "gl_dms_closed_frac": 0.6197777390480042,
+      "gl_dms_cr": 2.6310806274414062,
+      "gl_dms_loss": 0.002230919897556305,
+      "gl_dms_target_cr": 2.617647171020508,
+      "gl_dms_target_frac": 0.6179775595664978,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005391878541558981,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 9.701092720031738,
+      "learning_rate": 3e-05,
+      "loss": 0.0054,
+      "step": 56
+    },
+    {
+      "epoch": 0.114,
+      "gl_detached_lm_loss": 0.48314258456230164,
+      "gl_distil_loss": 0.003439662978053093,
+      "gl_dms_closed_frac": 0.6313677430152893,
+      "gl_dms_cr": 2.7136151790618896,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.6470587253570557,
+      "gl_dms_target_frac": 0.6222222447395325,
+      "gl_eos_tokens": 10.875,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.003439662978053093,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.125,
+      "gl_positions_without_loss_calculation": 10.875,
+      "grad_norm": 0.7246235013008118,
+      "learning_rate": 3e-05,
+      "loss": 0.0034,
+      "step": 57
+    },
+    {
+      "epoch": 0.116,
+      "gl_detached_lm_loss": 0.44782233238220215,
+      "gl_distil_loss": 0.0031148388516157866,
+      "gl_dms_closed_frac": 0.6334483623504639,
+      "gl_dms_cr": 2.7288801670074463,
+      "gl_dms_loss": 0.000607810914516449,
+      "gl_dms_target_cr": 2.6764702796936035,
+      "gl_dms_target_frac": 0.6263737082481384,
+      "gl_eos_tokens": 10.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0037226497661322355,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.0,
+      "gl_positions_without_loss_calculation": 10.0,
+      "grad_norm": 2.5122969150543213,
+      "learning_rate": 3e-05,
+      "loss": 0.0037,
+      "step": 58
+    },
+    {
+      "epoch": 0.118,
+      "gl_detached_lm_loss": 0.4750285744667053,
+      "gl_distil_loss": 0.003389698453247547,
+      "gl_dms_closed_frac": 0.6386871337890625,
+      "gl_dms_cr": 2.7682950496673584,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.7058825492858887,
+      "gl_dms_target_frac": 0.6304348111152649,
+      "gl_eos_tokens": 10.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.003389698453247547,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.25,
+      "gl_positions_without_loss_calculation": 10.75,
+      "grad_norm": 0.5368403792381287,
+      "learning_rate": 3e-05,
+      "loss": 0.0034,
+      "step": 59
+    },
+    {
+      "epoch": 0.12,
+      "gl_detached_lm_loss": 0.46681150794029236,
+      "gl_distil_loss": 0.0033188825473189354,
+      "gl_dms_closed_frac": 0.6449574828147888,
+      "gl_dms_cr": 2.817054033279419,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.7352941036224365,
+      "gl_dms_target_frac": 0.6344085335731506,
+      "gl_eos_tokens": 10.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0033188825473189354,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.75,
+      "gl_positions_without_loss_calculation": 10.25,
+      "grad_norm": 0.4873170554637909,
+      "learning_rate": 3e-05,
+      "loss": 0.0033,
+      "step": 60
+    },
+    {
+      "epoch": 0.122,
+      "gl_detached_lm_loss": 0.45119157433509827,
+      "gl_distil_loss": 0.0035455345641821623,
+      "gl_dms_closed_frac": 0.6473181843757629,
+      "gl_dms_cr": 2.8355660438537598,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.7647058963775635,
+      "gl_dms_target_frac": 0.6382978558540344,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0035455345641821623,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 0.7403936982154846,
+      "learning_rate": 3e-05,
+      "loss": 0.0035,
+      "step": 61
+    },
+    {
+      "epoch": 0.124,
+      "gl_detached_lm_loss": 0.47377100586891174,
+      "gl_distil_loss": 0.003390070516616106,
+      "gl_dms_closed_frac": 0.6446278691291809,
+      "gl_dms_cr": 2.814770221710205,
+      "gl_dms_loss": 0.0015726163983345032,
+      "gl_dms_target_cr": 2.7941174507141113,
+      "gl_dms_target_frac": 0.6421052813529968,
+      "gl_eos_tokens": 9.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.004962686914950609,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.5,
+      "gl_positions_without_loss_calculation": 9.5,
+      "grad_norm": 4.738519191741943,
+      "learning_rate": 3e-05,
+      "loss": 0.005,
+      "step": 62
+    },
+    {
+      "epoch": 0.126,
+      "gl_detached_lm_loss": 0.480397492647171,
+      "gl_distil_loss": 0.0035927374847233295,
+      "gl_dms_closed_frac": 0.650998592376709,
+      "gl_dms_cr": 2.86590838432312,
+      "gl_dms_loss": 0.0002143457531929016,
+      "gl_dms_target_cr": 2.8235297203063965,
+      "gl_dms_target_frac": 0.6458333134651184,
+      "gl_eos_tokens": 10.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.003807083237916231,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.75,
+      "gl_positions_without_loss_calculation": 10.25,
+      "grad_norm": 4.895111560821533,
+      "learning_rate": 3e-05,
+      "loss": 0.0038,
+      "step": 63
+    },
+    {
+      "epoch": 0.128,
+      "gl_detached_lm_loss": 0.45455634593963623,
+      "gl_distil_loss": 0.003494451055303216,
+      "gl_dms_closed_frac": 0.6617327928543091,
+      "gl_dms_cr": 2.95637845993042,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.8529412746429443,
+      "gl_dms_target_frac": 0.6494845747947693,
+      "gl_eos_tokens": 12.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.003494451055303216,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.25,
+      "gl_positions_without_loss_calculation": 12.75,
+      "grad_norm": 0.656202495098114,
+      "learning_rate": 3e-05,
+      "loss": 0.0035,
+      "step": 64
+    },
+    {
+      "epoch": 0.13,
+      "gl_detached_lm_loss": 0.4529498219490051,
+      "gl_distil_loss": 0.004102835897356272,
+      "gl_dms_closed_frac": 0.6730498671531677,
+      "gl_dms_cr": 3.058912754058838,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.882352828979492,
+      "gl_dms_target_frac": 0.6530612111091614,
+      "gl_eos_tokens": 11.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.004102835897356272,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.0,
+      "gl_positions_without_loss_calculation": 11.0,
+      "grad_norm": 2.2566981315612793,
+      "learning_rate": 3e-05,
+      "loss": 0.0041,
+      "step": 65
+    },
+    {
+      "epoch": 0.132,
+      "gl_detached_lm_loss": 0.4596954882144928,
+      "gl_distil_loss": 0.0038116772193461657,
+      "gl_dms_closed_frac": 0.6781314015388489,
+      "gl_dms_cr": 3.107957363128662,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.91176438331604,
+      "gl_dms_target_frac": 0.6565656661987305,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0038116772193461657,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 1.5338054895401,
+      "learning_rate": 3e-05,
+      "loss": 0.0038,
+      "step": 66
+    },
+    {
+      "epoch": 0.134,
+      "gl_detached_lm_loss": 0.4333009123802185,
+      "gl_distil_loss": 0.0037192029412835836,
+      "gl_dms_closed_frac": 0.6820775866508484,
+      "gl_dms_cr": 3.1470866203308105,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.941176414489746,
+      "gl_dms_target_frac": 0.6600000262260437,
+      "gl_eos_tokens": 10.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0037192029412835836,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.75,
+      "gl_positions_without_loss_calculation": 10.25,
+      "grad_norm": 0.676351010799408,
+      "learning_rate": 3e-05,
+      "loss": 0.0037,
+      "step": 67
+    },
+    {
+      "epoch": 0.136,
+      "gl_detached_lm_loss": 0.47546181082725525,
+      "gl_distil_loss": 0.004121327772736549,
+      "gl_dms_closed_frac": 0.6781496405601501,
+      "gl_dms_cr": 3.1076772212982178,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 2.970588445663452,
+      "gl_dms_target_frac": 0.6633663177490234,
+      "gl_eos_tokens": 12.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.004121327772736549,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.0,
+      "gl_positions_without_loss_calculation": 12.0,
+      "grad_norm": 1.0178581476211548,
+      "learning_rate": 3e-05,
+      "loss": 0.0041,
+      "step": 68
+    },
+    {
+      "epoch": 0.138,
+      "gl_detached_lm_loss": 0.48524773120880127,
+      "gl_distil_loss": 0.0038435542955994606,
+      "gl_dms_closed_frac": 0.6703423261642456,
+      "gl_dms_cr": 3.0347557067871094,
+      "gl_dms_loss": 0.001511596143245697,
+      "gl_dms_target_cr": 3.0,
+      "gl_dms_target_frac": 0.6666666269302368,
+      "gl_eos_tokens": 10.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005355150438845158,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.5,
+      "gl_positions_without_loss_calculation": 10.5,
+      "grad_norm": 6.971191883087158,
+      "learning_rate": 3e-05,
+      "loss": 0.0054,
+      "step": 69
+    },
+    {
+      "epoch": 0.14,
+      "gl_detached_lm_loss": 0.47538912296295166,
+      "gl_distil_loss": 0.003917126450687647,
+      "gl_dms_closed_frac": 0.6765962839126587,
+      "gl_dms_cr": 3.0925676822662354,
+      "gl_dms_loss": 0.0001483336091041565,
+      "gl_dms_target_cr": 3.029411554336548,
+      "gl_dms_target_frac": 0.6699028611183167,
+      "gl_eos_tokens": 11.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.004065460059791803,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.75,
+      "gl_positions_without_loss_calculation": 11.25,
+      "grad_norm": 2.054696798324585,
+      "learning_rate": 3e-05,
+      "loss": 0.0041,
+      "step": 70
+    },
+    {
+      "epoch": 0.142,
+      "gl_detached_lm_loss": 0.45387187600135803,
+      "gl_distil_loss": 0.003971648868173361,
+      "gl_dms_closed_frac": 0.689899206161499,
+      "gl_dms_cr": 3.2253713607788086,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.058823585510254,
+      "gl_dms_target_frac": 0.6730769872665405,
+      "gl_eos_tokens": 11.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.003971648868173361,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.0,
+      "gl_positions_without_loss_calculation": 11.0,
+      "grad_norm": 0.9588703513145447,
+      "learning_rate": 3e-05,
+      "loss": 0.004,
+      "step": 71
+    },
+    {
+      "epoch": 0.144,
+      "gl_detached_lm_loss": 0.4384338855743408,
+      "gl_distil_loss": 0.0038274391554296017,
+      "gl_dms_closed_frac": 0.6943807601928711,
+      "gl_dms_cr": 3.272785186767578,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.08823561668396,
+      "gl_dms_target_frac": 0.6761904358863831,
+      "gl_eos_tokens": 12.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0038274391554296017,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.75,
+      "gl_positions_without_loss_calculation": 12.25,
+      "grad_norm": 0.6780281662940979,
+      "learning_rate": 3e-05,
+      "loss": 0.0038,
+      "step": 72
+    },
+    {
+      "epoch": 0.146,
+      "gl_detached_lm_loss": 0.44365930557250977,
+      "gl_distil_loss": 0.003922718111425638,
+      "gl_dms_closed_frac": 0.6959836483001709,
+      "gl_dms_cr": 3.2900500297546387,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.117647171020508,
+      "gl_dms_target_frac": 0.6792453527450562,
+      "gl_eos_tokens": 11.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.003922718111425638,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.75,
+      "gl_positions_without_loss_calculation": 11.25,
+      "grad_norm": 0.5752905011177063,
+      "learning_rate": 3e-05,
+      "loss": 0.0039,
+      "step": 73
+    },
+    {
+      "epoch": 0.148,
+      "gl_detached_lm_loss": 0.4326343834400177,
+      "gl_distil_loss": 0.003944290801882744,
+      "gl_dms_closed_frac": 0.696725070476532,
+      "gl_dms_cr": 3.2982115745544434,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.1470587253570557,
+      "gl_dms_target_frac": 0.6822429299354553,
+      "gl_eos_tokens": 10.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.003944290801882744,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.5,
+      "gl_positions_without_loss_calculation": 10.5,
+      "grad_norm": 0.799717366695404,
+      "learning_rate": 3e-05,
+      "loss": 0.0039,
+      "step": 74
+    },
+    {
+      "epoch": 0.15,
+      "gl_detached_lm_loss": 0.486849308013916,
+      "gl_distil_loss": 0.004036651458591223,
+      "gl_dms_closed_frac": 0.6863027215003967,
+      "gl_dms_cr": 3.1896045207977295,
+      "gl_dms_loss": 0.0025970041751861572,
+      "gl_dms_target_cr": 3.1764702796936035,
+      "gl_dms_target_frac": 0.6851851940155029,
+      "gl_eos_tokens": 12.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.00663365563377738,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.0,
+      "gl_positions_without_loss_calculation": 12.0,
+      "grad_norm": 14.300118446350098,
+      "learning_rate": 3e-05,
+      "loss": 0.0066,
+      "step": 75
+    },
+    {
+      "epoch": 0.152,
+      "gl_detached_lm_loss": 0.4877408444881439,
+      "gl_distil_loss": 0.004087416920810938,
+      "gl_dms_closed_frac": 0.688453733921051,
+      "gl_dms_cr": 3.2113051414489746,
+      "gl_dms_loss": 0.0024287551641464233,
+      "gl_dms_target_cr": 3.2058825492858887,
+      "gl_dms_target_frac": 0.6880733966827393,
+      "gl_eos_tokens": 11.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0065161725506186485,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.75,
+      "gl_positions_without_loss_calculation": 11.25,
+      "grad_norm": 13.963382720947266,
+      "learning_rate": 3e-05,
+      "loss": 0.0065,
+      "step": 76
+    },
+    {
+      "epoch": 0.154,
+      "gl_detached_lm_loss": 0.44424474239349365,
+      "gl_distil_loss": 0.0036926830653101206,
+      "gl_dms_closed_frac": 0.6956153512001038,
+      "gl_dms_cr": 3.2859933376312256,
+      "gl_dms_loss": 0.0003067776560783386,
+      "gl_dms_target_cr": 3.2352941036224365,
+      "gl_dms_target_frac": 0.6909090280532837,
+      "gl_eos_tokens": 11.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0039994604885578156,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.25,
+      "gl_positions_without_loss_calculation": 11.75,
+      "grad_norm": 2.204572916030884,
+      "learning_rate": 3e-05,
+      "loss": 0.004,
+      "step": 77
+    },
+    {
+      "epoch": 0.156,
+      "gl_detached_lm_loss": 0.49248582124710083,
+      "gl_distil_loss": 0.00488574244081974,
+      "gl_dms_closed_frac": 0.7086416482925415,
+      "gl_dms_cr": 3.4329066276550293,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.2647058963775635,
+      "gl_dms_target_frac": 0.6936936974525452,
+      "gl_eos_tokens": 11.375,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.00488574244081974,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.625,
+      "gl_positions_without_loss_calculation": 11.375,
+      "grad_norm": 0.9433761835098267,
+      "learning_rate": 3e-05,
+      "loss": 0.0049,
+      "step": 78
+    },
+    {
+      "epoch": 0.158,
+      "gl_detached_lm_loss": 0.47902607917785645,
+      "gl_distil_loss": 0.004779959097504616,
+      "gl_dms_closed_frac": 0.7143765091896057,
+      "gl_dms_cr": 3.503458023071289,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.2941174507141113,
+      "gl_dms_target_frac": 0.696428656578064,
+      "gl_eos_tokens": 10.375,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.004779959097504616,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.625,
+      "gl_positions_without_loss_calculation": 10.375,
+      "grad_norm": 1.8233813047409058,
+      "learning_rate": 3e-05,
+      "loss": 0.0048,
+      "step": 79
+    },
+    {
+      "epoch": 0.16,
+      "gl_detached_lm_loss": 0.45535245537757874,
+      "gl_distil_loss": 0.004689468070864677,
+      "gl_dms_closed_frac": 0.7179027795791626,
+      "gl_dms_cr": 3.545504093170166,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.3235297203063965,
+      "gl_dms_target_frac": 0.6991150379180908,
+      "gl_eos_tokens": 10.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.004689468070864677,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.25,
+      "gl_positions_without_loss_calculation": 10.75,
+      "grad_norm": 0.7161194086074829,
+      "learning_rate": 3e-05,
+      "loss": 0.0047,
+      "step": 80
+    },
+    {
+      "epoch": 0.162,
+      "gl_detached_lm_loss": 0.4955553412437439,
+      "gl_distil_loss": 0.005196040961891413,
+      "gl_dms_closed_frac": 0.7184000015258789,
+      "gl_dms_cr": 3.553600788116455,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.3529412746429443,
+      "gl_dms_target_frac": 0.7017544507980347,
+      "gl_eos_tokens": 10.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005196040961891413,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.25,
+      "gl_positions_without_loss_calculation": 10.75,
+      "grad_norm": 1.219171404838562,
+      "learning_rate": 3e-05,
+      "loss": 0.0052,
+      "step": 81
+    },
+    {
+      "epoch": 0.164,
+      "gl_detached_lm_loss": 0.4791584610939026,
+      "gl_distil_loss": 0.004651572089642286,
+      "gl_dms_closed_frac": 0.7162825465202332,
+      "gl_dms_cr": 3.5258140563964844,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.382352828979492,
+      "gl_dms_target_frac": 0.7043478488922119,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.004651572089642286,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 0.9896210432052612,
+      "learning_rate": 3e-05,
+      "loss": 0.0047,
+      "step": 82
+    },
+    {
+      "epoch": 0.166,
+      "gl_detached_lm_loss": 0.47231408953666687,
+      "gl_distil_loss": 0.004033881705254316,
+      "gl_dms_closed_frac": 0.7049646973609924,
+      "gl_dms_cr": 3.3899085521698,
+      "gl_dms_loss": 0.0026998519897460938,
+      "gl_dms_target_cr": 3.41176438331604,
+      "gl_dms_target_frac": 0.7068965435028076,
+      "gl_eos_tokens": 11.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.00673373369500041,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.75,
+      "gl_positions_without_loss_calculation": 11.25,
+      "grad_norm": 13.66180419921875,
+      "learning_rate": 3e-05,
+      "loss": 0.0067,
+      "step": 83
+    },
+    {
+      "epoch": 0.168,
+      "gl_detached_lm_loss": 0.46945416927337646,
+      "gl_distil_loss": 0.00402010977268219,
+      "gl_dms_closed_frac": 0.7052032351493835,
+      "gl_dms_cr": 3.392927408218384,
+      "gl_dms_loss": 0.004722759127616882,
+      "gl_dms_target_cr": 3.4411768913269043,
+      "gl_dms_target_frac": 0.7094016671180725,
+      "gl_eos_tokens": 12.625,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.008742868900299072,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.375,
+      "gl_positions_without_loss_calculation": 12.625,
+      "grad_norm": 16.05010414123535,
+      "learning_rate": 3e-05,
+      "loss": 0.0087,
+      "step": 84
+    },
+    {
+      "epoch": 0.17,
+      "gl_detached_lm_loss": 0.44967958331108093,
+      "gl_distil_loss": 0.003981797024607658,
+      "gl_dms_closed_frac": 0.7158433198928833,
+      "gl_dms_cr": 3.520792245864868,
+      "gl_dms_loss": 0.0010783001780509949,
+      "gl_dms_target_cr": 3.470588445663452,
+      "gl_dms_target_frac": 0.7118644118309021,
+      "gl_eos_tokens": 12.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005060096736997366,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.0,
+      "gl_positions_without_loss_calculation": 12.0,
+      "grad_norm": 6.453094482421875,
+      "learning_rate": 3e-05,
+      "loss": 0.0051,
+      "step": 85
+    },
+    {
+      "epoch": 0.172,
+      "gl_detached_lm_loss": 0.4791140556335449,
+      "gl_distil_loss": 0.00494480412453413,
+      "gl_dms_closed_frac": 0.7277750372886658,
+      "gl_dms_cr": 3.6741161346435547,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.5,
+      "gl_dms_target_frac": 0.7142857909202576,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.00494480412453413,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 0.9170877933502197,
+      "learning_rate": 3e-05,
+      "loss": 0.0049,
+      "step": 86
+    },
+    {
+      "epoch": 0.174,
+      "gl_detached_lm_loss": 0.47055312991142273,
+      "gl_distil_loss": 0.005052007734775543,
+      "gl_dms_closed_frac": 0.7355045080184937,
+      "gl_dms_cr": 3.781733751296997,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.529411554336548,
+      "gl_dms_target_frac": 0.7166666388511658,
+      "gl_eos_tokens": 10.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005052007734775543,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.25,
+      "gl_positions_without_loss_calculation": 10.75,
+      "grad_norm": 0.9470101594924927,
+      "learning_rate": 3e-05,
+      "loss": 0.0051,
+      "step": 87
+    },
+    {
+      "epoch": 0.176,
+      "gl_detached_lm_loss": 0.4664342999458313,
+      "gl_distil_loss": 0.004945836495608091,
+      "gl_dms_closed_frac": 0.7362442016601562,
+      "gl_dms_cr": 3.79286789894104,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.5588231086730957,
+      "gl_dms_target_frac": 0.7190083265304565,
+      "gl_eos_tokens": 10.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.004945836495608091,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.0,
+      "gl_positions_without_loss_calculation": 10.0,
+      "grad_norm": 0.7975891828536987,
+      "learning_rate": 3e-05,
+      "loss": 0.0049,
+      "step": 88
+    },
+    {
+      "epoch": 0.178,
+      "gl_detached_lm_loss": 0.45679762959480286,
+      "gl_distil_loss": 0.004826837219297886,
+      "gl_dms_closed_frac": 0.7391374111175537,
+      "gl_dms_cr": 3.8347959518432617,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.58823561668396,
+      "gl_dms_target_frac": 0.7213115096092224,
+      "gl_eos_tokens": 12.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.004826837219297886,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.75,
+      "gl_positions_without_loss_calculation": 12.25,
+      "grad_norm": 0.7645382881164551,
+      "learning_rate": 3e-05,
+      "loss": 0.0048,
+      "step": 89
+    },
+    {
+      "epoch": 0.18,
+      "gl_detached_lm_loss": 0.4502984285354614,
+      "gl_distil_loss": 0.004537405911833048,
+      "gl_dms_closed_frac": 0.7365341782569885,
+      "gl_dms_cr": 3.797346830368042,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.617647171020508,
+      "gl_dms_target_frac": 0.7235772609710693,
+      "gl_eos_tokens": 12.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.004537405911833048,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.25,
+      "gl_positions_without_loss_calculation": 12.75,
+      "grad_norm": 2.0026698112487793,
+      "learning_rate": 3e-05,
+      "loss": 0.0045,
+      "step": 90
+    },
+    {
+      "epoch": 0.182,
+      "gl_detached_lm_loss": 0.46442925930023193,
+      "gl_distil_loss": 0.0043852319940924644,
+      "gl_dms_closed_frac": 0.727599561214447,
+      "gl_dms_cr": 3.6718647480010986,
+      "gl_dms_loss": 0.0010252445936203003,
+      "gl_dms_target_cr": 3.6470587253570557,
+      "gl_dms_target_frac": 0.725806474685669,
+      "gl_eos_tokens": 11.875,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005410476587712765,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.125,
+      "gl_positions_without_loss_calculation": 11.875,
+      "grad_norm": 7.0575032234191895,
+      "learning_rate": 3e-05,
+      "loss": 0.0054,
+      "step": 91
+    },
+    {
+      "epoch": 0.184,
+      "gl_detached_lm_loss": 0.4796100854873657,
+      "gl_distil_loss": 0.004694662988185883,
+      "gl_dms_closed_frac": 0.7282149791717529,
+      "gl_dms_cr": 3.6801345348358154,
+      "gl_dms_loss": 0.0016043484210968018,
+      "gl_dms_target_cr": 3.6764702796936035,
+      "gl_dms_target_frac": 0.7280000448226929,
+      "gl_eos_tokens": 10.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.006299011409282684,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.25,
+      "gl_positions_without_loss_calculation": 10.75,
+      "grad_norm": 6.339512825012207,
+      "learning_rate": 3e-05,
+      "loss": 0.0063,
+      "step": 92
+    },
+    {
+      "epoch": 0.186,
+      "gl_detached_lm_loss": 0.45466843247413635,
+      "gl_distil_loss": 0.0042836032807827,
+      "gl_dms_closed_frac": 0.7350358963012695,
+      "gl_dms_cr": 3.775805950164795,
+      "gl_dms_loss": 0.0011206194758415222,
+      "gl_dms_target_cr": 3.7058825492858887,
+      "gl_dms_target_frac": 0.7301587462425232,
+      "gl_eos_tokens": 12.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005404223222285509,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.5,
+      "gl_positions_without_loss_calculation": 12.5,
+      "grad_norm": 4.007299900054932,
+      "learning_rate": 3e-05,
+      "loss": 0.0054,
+      "step": 93
+    },
+    {
+      "epoch": 0.188,
+      "gl_detached_lm_loss": 0.4374981224536896,
+      "gl_distil_loss": 0.005377020221203566,
+      "gl_dms_closed_frac": 0.7533713579177856,
+      "gl_dms_cr": 4.057657241821289,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.7352941036224365,
+      "gl_dms_target_frac": 0.7322835326194763,
+      "gl_eos_tokens": 12.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005377020221203566,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.0,
+      "gl_positions_without_loss_calculation": 12.0,
+      "grad_norm": 2.0438311100006104,
+      "learning_rate": 3e-05,
+      "loss": 0.0054,
+      "step": 94
+    },
+    {
+      "epoch": 0.19,
+      "gl_detached_lm_loss": 0.46797192096710205,
+      "gl_distil_loss": 0.005093785934150219,
+      "gl_dms_closed_frac": 0.7516717314720154,
+      "gl_dms_cr": 4.029850482940674,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.7647058963775635,
+      "gl_dms_target_frac": 0.734375,
+      "gl_eos_tokens": 10.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005093785934150219,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.25,
+      "gl_positions_without_loss_calculation": 10.75,
+      "grad_norm": 1.0185790061950684,
+      "learning_rate": 3e-05,
+      "loss": 0.0051,
+      "step": 95
+    },
+    {
+      "epoch": 0.192,
+      "gl_detached_lm_loss": 0.47336265444755554,
+      "gl_distil_loss": 0.005353494081646204,
+      "gl_dms_closed_frac": 0.7560113072395325,
+      "gl_dms_cr": 4.101371765136719,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.7941174507141113,
+      "gl_dms_target_frac": 0.7364340424537659,
+      "gl_eos_tokens": 10.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005353494081646204,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.5,
+      "gl_positions_without_loss_calculation": 10.5,
+      "grad_norm": 1.1575500965118408,
+      "learning_rate": 3e-05,
+      "loss": 0.0054,
+      "step": 96
+    },
+    {
+      "epoch": 0.194,
+      "gl_detached_lm_loss": 0.4636452794075012,
+      "gl_distil_loss": 0.005527734290808439,
+      "gl_dms_closed_frac": 0.7628649473190308,
+      "gl_dms_cr": 4.217801094055176,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.8235297203063965,
+      "gl_dms_target_frac": 0.7384615540504456,
+      "gl_eos_tokens": 10.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005527734290808439,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.5,
+      "gl_positions_without_loss_calculation": 10.5,
+      "grad_norm": 1.2612135410308838,
+      "learning_rate": 3e-05,
+      "loss": 0.0055,
+      "step": 97
+    },
+    {
+      "epoch": 0.196,
+      "gl_detached_lm_loss": 0.49130597710609436,
+      "gl_distil_loss": 0.005308451130986214,
+      "gl_dms_closed_frac": 0.755087673664093,
+      "gl_dms_cr": 4.086455345153809,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.8529412746429443,
+      "gl_dms_target_frac": 0.7404580116271973,
+      "gl_eos_tokens": 12.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005308451130986214,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.75,
+      "gl_positions_without_loss_calculation": 12.25,
+      "grad_norm": 0.8162508606910706,
+      "learning_rate": 3e-05,
+      "loss": 0.0053,
+      "step": 98
+    },
+    {
+      "epoch": 0.198,
+      "gl_detached_lm_loss": 0.4728451669216156,
+      "gl_distil_loss": 0.004922911990433931,
+      "gl_dms_closed_frac": 0.7501329779624939,
+      "gl_dms_cr": 4.0030927658081055,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 3.882352828979492,
+      "gl_dms_target_frac": 0.7424242496490479,
+      "gl_eos_tokens": 12.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.004922911990433931,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.25,
+      "gl_positions_without_loss_calculation": 12.75,
+      "grad_norm": 0.8720424771308899,
+      "learning_rate": 3e-05,
+      "loss": 0.0049,
+      "step": 99
+    },
+    {
+      "epoch": 0.2,
+      "gl_detached_lm_loss": 0.46046435832977295,
+      "gl_distil_loss": 0.004833785351365805,
+      "gl_dms_closed_frac": 0.7439101934432983,
+      "gl_dms_cr": 3.9068267345428467,
+      "gl_dms_loss": 0.0024033188819885254,
+      "gl_dms_target_cr": 3.91176438331604,
+      "gl_dms_target_frac": 0.7443609237670898,
+      "gl_eos_tokens": 10.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0072371033020317554,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.5,
+      "gl_positions_without_loss_calculation": 10.5,
+      "grad_norm": 8.47694206237793,
+      "learning_rate": 3e-05,
+      "loss": 0.0072,
+      "step": 100
+    },
+    {
+      "epoch": 0.202,
+      "gl_detached_lm_loss": 0.49140673875808716,
+      "gl_distil_loss": 0.004810246638953686,
+      "gl_dms_closed_frac": 0.7399882078170776,
+      "gl_dms_cr": 3.8466882705688477,
+      "gl_dms_loss": 0.006481736898422241,
+      "gl_dms_target_cr": 3.9411768913269043,
+      "gl_dms_target_frac": 0.7462686896324158,
+      "gl_eos_tokens": 12.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.011291983537375927,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.0,
+      "gl_positions_without_loss_calculation": 12.0,
+      "grad_norm": 14.86678409576416,
+      "learning_rate": 3e-05,
+      "loss": 0.0113,
+      "step": 101
+    },
+    {
+      "epoch": 0.204,
+      "gl_detached_lm_loss": 0.4723597764968872,
+      "gl_distil_loss": 0.004680398851633072,
+      "gl_dms_closed_frac": 0.7438403964042664,
+      "gl_dms_cr": 3.9060354232788086,
+      "gl_dms_loss": 0.005097024142742157,
+      "gl_dms_target_cr": 3.970588445663452,
+      "gl_dms_target_frac": 0.7481480836868286,
+      "gl_eos_tokens": 12.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.009777422994375229,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.75,
+      "gl_positions_without_loss_calculation": 12.25,
+      "grad_norm": 12.764368057250977,
+      "learning_rate": 3e-05,
+      "loss": 0.0098,
+      "step": 102
+    },
+    {
+      "epoch": 0.206,
+      "gl_detached_lm_loss": 0.45221349596977234,
+      "gl_distil_loss": 0.00520294951274991,
+      "gl_dms_closed_frac": 0.7546575665473938,
+      "gl_dms_cr": 4.077920436859131,
+      "gl_dms_loss": 0.0010900869965553284,
+      "gl_dms_target_cr": 4.0,
+      "gl_dms_target_frac": 0.75,
+      "gl_eos_tokens": 12.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.006293036509305239,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.5,
+      "gl_positions_without_loss_calculation": 12.5,
+      "grad_norm": 4.082666873931885,
+      "learning_rate": 3e-05,
+      "loss": 0.0063,
+      "step": 103
+    },
+    {
+      "epoch": 0.208,
+      "gl_detached_lm_loss": 0.4825055003166199,
+      "gl_distil_loss": 0.0058518978767097,
+      "gl_dms_closed_frac": 0.7645120620727539,
+      "gl_dms_cr": 4.248115539550781,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.029411792755127,
+      "gl_dms_target_frac": 0.7518247961997986,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0058518978767097,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 1.0137183666229248,
+      "learning_rate": 3e-05,
+      "loss": 0.0059,
+      "step": 104
+    },
+    {
+      "epoch": 0.21,
+      "gl_detached_lm_loss": 0.47762972116470337,
+      "gl_distil_loss": 0.005726248025894165,
+      "gl_dms_closed_frac": 0.770868182182312,
+      "gl_dms_cr": 4.367452621459961,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.058823108673096,
+      "gl_dms_target_frac": 0.7536231279373169,
+      "gl_eos_tokens": 10.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005726248025894165,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.75,
+      "gl_positions_without_loss_calculation": 10.25,
+      "grad_norm": 1.5721111297607422,
+      "learning_rate": 3e-05,
+      "loss": 0.0057,
+      "step": 105
+    },
+    {
+      "epoch": 0.212,
+      "gl_detached_lm_loss": 0.4453291594982147,
+      "gl_distil_loss": 0.005907304584980011,
+      "gl_dms_closed_frac": 0.7767513990402222,
+      "gl_dms_cr": 4.480729103088379,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.088235378265381,
+      "gl_dms_target_frac": 0.755395770072937,
+      "gl_eos_tokens": 13.375,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005907304584980011,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32754.625,
+      "gl_positions_without_loss_calculation": 13.375,
+      "grad_norm": 1.3633668422698975,
+      "learning_rate": 3e-05,
+      "loss": 0.0059,
+      "step": 106
+    },
+    {
+      "epoch": 0.214,
+      "gl_detached_lm_loss": 0.4815954267978668,
+      "gl_distil_loss": 0.005865244194865227,
+      "gl_dms_closed_frac": 0.7773128747940063,
+      "gl_dms_cr": 4.4921698570251465,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.117647171020508,
+      "gl_dms_target_frac": 0.7571429014205933,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005865244194865227,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 1.251830816268921,
+      "learning_rate": 3e-05,
+      "loss": 0.0059,
+      "step": 107
+    },
+    {
+      "epoch": 0.216,
+      "gl_detached_lm_loss": 0.473334401845932,
+      "gl_distil_loss": 0.005540169775485992,
+      "gl_dms_closed_frac": 0.7712863683700562,
+      "gl_dms_cr": 4.375051498413086,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.147058963775635,
+      "gl_dms_target_frac": 0.7588652968406677,
+      "gl_eos_tokens": 12.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005540169775485992,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.75,
+      "gl_positions_without_loss_calculation": 12.25,
+      "grad_norm": 0.9412449598312378,
+      "learning_rate": 3e-05,
+      "loss": 0.0055,
+      "step": 108
+    },
+    {
+      "epoch": 0.218,
+      "gl_detached_lm_loss": 0.4723076820373535,
+      "gl_distil_loss": 0.005424611270427704,
+      "gl_dms_closed_frac": 0.7736709713935852,
+      "gl_dms_cr": 4.420588493347168,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.17647123336792,
+      "gl_dms_target_frac": 0.7605633735656738,
+      "gl_eos_tokens": 11.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.005424611270427704,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.75,
+      "gl_positions_without_loss_calculation": 11.25,
+      "grad_norm": 1.0735340118408203,
+      "learning_rate": 3e-05,
+      "loss": 0.0054,
+      "step": 109
+    },
+    {
+      "epoch": 0.22,
+      "gl_detached_lm_loss": 0.48178762197494507,
+      "gl_distil_loss": 0.005986446980386972,
+      "gl_dms_closed_frac": 0.764805793762207,
+      "gl_dms_cr": 4.252771854400635,
+      "gl_dms_loss": 0.000624343752861023,
+      "gl_dms_target_cr": 4.205882549285889,
+      "gl_dms_target_frac": 0.7622377872467041,
+      "gl_eos_tokens": 9.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.006610790733247995,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32759.0,
+      "gl_positions_without_loss_calculation": 9.0,
+      "grad_norm": 3.545208215713501,
+      "learning_rate": 3e-05,
+      "loss": 0.0066,
+      "step": 110
+    },
+    {
+      "epoch": 0.222,
+      "gl_detached_lm_loss": 0.4608258605003357,
+      "gl_distil_loss": 0.00500837666913867,
+      "gl_dms_closed_frac": 0.763525128364563,
+      "gl_dms_cr": 4.232396125793457,
+      "gl_dms_loss": 0.003266759216785431,
+      "gl_dms_target_cr": 4.235294342041016,
+      "gl_dms_target_frac": 0.7638888955116272,
+      "gl_eos_tokens": 11.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.008275135420262814,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.0,
+      "gl_positions_without_loss_calculation": 11.0,
+      "grad_norm": 7.840135097503662,
+      "learning_rate": 3e-05,
+      "loss": 0.0083,
+      "step": 111
+    },
+    {
+      "epoch": 0.224,
+      "gl_detached_lm_loss": 0.46065497398376465,
+      "gl_distil_loss": 0.005225225817412138,
+      "gl_dms_closed_frac": 0.7623160481452942,
+      "gl_dms_cr": 4.2096405029296875,
+      "gl_dms_loss": 0.004034481942653656,
+      "gl_dms_target_cr": 4.264705657958984,
+      "gl_dms_target_frac": 0.7655172348022461,
+      "gl_eos_tokens": 11.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.009259707294404507,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.25,
+      "gl_positions_without_loss_calculation": 11.75,
+      "grad_norm": 11.423973083496094,
+      "learning_rate": 3e-05,
+      "loss": 0.0093,
+      "step": 112
+    },
+    {
+      "epoch": 0.226,
+      "gl_detached_lm_loss": 0.47371160984039307,
+      "gl_distil_loss": 0.005666877143085003,
+      "gl_dms_closed_frac": 0.774148166179657,
+      "gl_dms_cr": 4.4307637214660645,
+      "gl_dms_loss": 0.0007906854152679443,
+      "gl_dms_target_cr": 4.294117450714111,
+      "gl_dms_target_frac": 0.767123281955719,
+      "gl_eos_tokens": 11.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.006457562558352947,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.75,
+      "gl_positions_without_loss_calculation": 11.25,
+      "grad_norm": 1.7580599784851074,
+      "learning_rate": 3e-05,
+      "loss": 0.0065,
+      "step": 113
+    },
+    {
+      "epoch": 0.228,
+      "gl_detached_lm_loss": 0.4567711353302002,
+      "gl_distil_loss": 0.00592584116384387,
+      "gl_dms_closed_frac": 0.7843614220619202,
+      "gl_dms_cr": 4.637998580932617,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.32352876663208,
+      "gl_dms_target_frac": 0.7687073945999146,
+      "gl_eos_tokens": 11.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.00592584116384387,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.0,
+      "gl_positions_without_loss_calculation": 11.0,
+      "grad_norm": 0.8876529932022095,
+      "learning_rate": 3e-05,
+      "loss": 0.0059,
+      "step": 114
+    },
+    {
+      "epoch": 0.23,
+      "gl_detached_lm_loss": 0.4576658606529236,
+      "gl_distil_loss": 0.0065962583757936954,
+      "gl_dms_closed_frac": 0.7914761900901794,
+      "gl_dms_cr": 4.798535346984863,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.352941036224365,
+      "gl_dms_target_frac": 0.7702702879905701,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0065962583757936954,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 1.1149492263793945,
+      "learning_rate": 3e-05,
+      "loss": 0.0066,
+      "step": 115
+    },
+    {
+      "epoch": 0.232,
+      "gl_detached_lm_loss": 0.4753264784812927,
+      "gl_distil_loss": 0.006386497989296913,
+      "gl_dms_closed_frac": 0.7959333658218384,
+      "gl_dms_cr": 4.9022216796875,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.382352828979492,
+      "gl_dms_target_frac": 0.7718120217323303,
+      "gl_eos_tokens": 10.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.006386497989296913,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.25,
+      "gl_positions_without_loss_calculation": 10.75,
+      "grad_norm": 1.1145360469818115,
+      "learning_rate": 3e-05,
+      "loss": 0.0064,
+      "step": 116
+    },
+    {
+      "epoch": 0.234,
+      "gl_detached_lm_loss": 0.4654878079891205,
+      "gl_distil_loss": 0.006225732620805502,
+      "gl_dms_closed_frac": 0.7925079464912415,
+      "gl_dms_cr": 4.8202338218688965,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.411764621734619,
+      "gl_dms_target_frac": 0.7733333110809326,
+      "gl_eos_tokens": 12.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.006225732620805502,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.25,
+      "gl_positions_without_loss_calculation": 12.75,
+      "grad_norm": 1.121919870376587,
+      "learning_rate": 3e-05,
+      "loss": 0.0062,
+      "step": 117
+    },
+    {
+      "epoch": 0.236,
+      "gl_detached_lm_loss": 0.46314623951911926,
+      "gl_distil_loss": 0.0068687875755131245,
+      "gl_dms_closed_frac": 0.7923510074615479,
+      "gl_dms_cr": 4.819302558898926,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.441176891326904,
+      "gl_dms_target_frac": 0.7748345136642456,
+      "gl_eos_tokens": 10.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0068687875755131245,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.0,
+      "gl_positions_without_loss_calculation": 10.0,
+      "grad_norm": 2.76143741607666,
+      "learning_rate": 3e-05,
+      "loss": 0.0069,
+      "step": 118
+    },
+    {
+      "epoch": 0.238,
+      "gl_detached_lm_loss": 0.4607497453689575,
+      "gl_distil_loss": 0.006361807230859995,
+      "gl_dms_closed_frac": 0.786472499370575,
+      "gl_dms_cr": 4.685853004455566,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.470588207244873,
+      "gl_dms_target_frac": 0.7763157486915588,
+      "gl_eos_tokens": 9.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.006361807230859995,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.25,
+      "gl_positions_without_loss_calculation": 9.75,
+      "grad_norm": 0.921043872833252,
+      "learning_rate": 3e-05,
+      "loss": 0.0064,
+      "step": 119
+    },
+    {
+      "epoch": 0.24,
+      "gl_detached_lm_loss": 0.47533145546913147,
+      "gl_distil_loss": 0.005683056078851223,
+      "gl_dms_closed_frac": 0.7742487788200378,
+      "gl_dms_cr": 4.4305853843688965,
+      "gl_dms_loss": 0.0037161409854888916,
+      "gl_dms_target_cr": 4.5,
+      "gl_dms_target_frac": 0.7777777314186096,
+      "gl_eos_tokens": 11.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.009399197064340115,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.25,
+      "gl_positions_without_loss_calculation": 11.75,
+      "grad_norm": 13.05197811126709,
+      "learning_rate": 3e-05,
+      "loss": 0.0094,
+      "step": 120
+    },
+    {
+      "epoch": 0.242,
+      "gl_detached_lm_loss": 0.4673323631286621,
+      "gl_distil_loss": 0.005364595912396908,
+      "gl_dms_closed_frac": 0.774548351764679,
+      "gl_dms_cr": 4.43836784362793,
+      "gl_dms_loss": 0.005432605743408203,
+      "gl_dms_target_cr": 4.529411792755127,
+      "gl_dms_target_frac": 0.779220700263977,
+      "gl_eos_tokens": 12.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.010797200724482536,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.0,
+      "gl_positions_without_loss_calculation": 12.0,
+      "grad_norm": 11.362590789794922,
+      "learning_rate": 3e-05,
+      "loss": 0.0108,
+      "step": 121
+    },
+    {
+      "epoch": 0.244,
+      "gl_detached_lm_loss": 0.46205538511276245,
+      "gl_distil_loss": 0.005460300482809544,
+      "gl_dms_closed_frac": 0.7765141725540161,
+      "gl_dms_cr": 4.475861072540283,
+      "gl_dms_loss": 0.00442194938659668,
+      "gl_dms_target_cr": 4.558823108673096,
+      "gl_dms_target_frac": 0.7806451320648193,
+      "gl_eos_tokens": 12.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.009882249869406223,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.25,
+      "gl_positions_without_loss_calculation": 12.75,
+      "grad_norm": 11.339438438415527,
+      "learning_rate": 3e-05,
+      "loss": 0.0099,
+      "step": 122
+    },
+    {
+      "epoch": 0.246,
+      "gl_detached_lm_loss": 0.4668773412704468,
+      "gl_distil_loss": 0.006131088826805353,
+      "gl_dms_closed_frac": 0.7835890650749207,
+      "gl_dms_cr": 4.622951030731201,
+      "gl_dms_loss": 0.0013428032398223877,
+      "gl_dms_target_cr": 4.588235378265381,
+      "gl_dms_target_frac": 0.7820512056350708,
+      "gl_eos_tokens": 10.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.007473892066627741,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.0,
+      "gl_positions_without_loss_calculation": 10.0,
+      "grad_norm": 4.741786479949951,
+      "learning_rate": 3e-05,
+      "loss": 0.0075,
+      "step": 123
+    },
+    {
+      "epoch": 0.248,
+      "gl_detached_lm_loss": 0.48726099729537964,
+      "gl_distil_loss": 0.006155762821435928,
+      "gl_dms_closed_frac": 0.7912291884422302,
+      "gl_dms_cr": 4.793363094329834,
+      "gl_dms_loss": 0.00044589489698410034,
+      "gl_dms_target_cr": 4.617647171020508,
+      "gl_dms_target_frac": 0.783439576625824,
+      "gl_eos_tokens": 9.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.006601657718420029,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.25,
+      "gl_positions_without_loss_calculation": 9.75,
+      "grad_norm": 1.6124783754348755,
+      "learning_rate": 3e-05,
+      "loss": 0.0066,
+      "step": 124
+    },
+    {
+      "epoch": 0.25,
+      "gl_detached_lm_loss": 0.4655800461769104,
+      "gl_distil_loss": 0.006976068951189518,
+      "gl_dms_closed_frac": 0.8033832907676697,
+      "gl_dms_cr": 5.088557243347168,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.647058963775635,
+      "gl_dms_target_frac": 0.7848101258277893,
+      "gl_eos_tokens": 11.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.006976068951189518,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.75,
+      "gl_positions_without_loss_calculation": 11.25,
+      "grad_norm": 1.4380581378936768,
+      "learning_rate": 3e-05,
+      "loss": 0.007,
+      "step": 125
+    },
+    {
+      "epoch": 0.252,
+      "gl_detached_lm_loss": 0.4725133776664734,
+      "gl_distil_loss": 0.007554593961685896,
+      "gl_dms_closed_frac": 0.8122463822364807,
+      "gl_dms_cr": 5.328559398651123,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.67647123336792,
+      "gl_dms_target_frac": 0.7861634492874146,
+      "gl_eos_tokens": 10.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.007554593961685896,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.25,
+      "gl_positions_without_loss_calculation": 10.75,
+      "grad_norm": 1.0721145868301392,
+      "learning_rate": 3e-05,
+      "loss": 0.0076,
+      "step": 126
+    },
+    {
+      "epoch": 0.254,
+      "gl_detached_lm_loss": 0.470710426568985,
+      "gl_distil_loss": 0.007993207313120365,
+      "gl_dms_closed_frac": 0.8147773742675781,
+      "gl_dms_cr": 5.401104927062988,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.705882549285889,
+      "gl_dms_target_frac": 0.7874999642372131,
+      "gl_eos_tokens": 10.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.007993207313120365,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.25,
+      "gl_positions_without_loss_calculation": 10.75,
+      "grad_norm": 1.3880622386932373,
+      "learning_rate": 3e-05,
+      "loss": 0.008,
+      "step": 127
+    },
+    {
+      "epoch": 0.256,
+      "gl_detached_lm_loss": 0.4928918182849884,
+      "gl_distil_loss": 0.007520680315792561,
+      "gl_dms_closed_frac": 0.8154762983322144,
+      "gl_dms_cr": 5.420012950897217,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.735294342041016,
+      "gl_dms_target_frac": 0.7888198494911194,
+      "gl_eos_tokens": 12.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.007520680315792561,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.0,
+      "gl_positions_without_loss_calculation": 12.0,
+      "grad_norm": 1.0682696104049683,
+      "learning_rate": 3e-05,
+      "loss": 0.0075,
+      "step": 128
+    },
+    {
+      "epoch": 0.258,
+      "gl_detached_lm_loss": 0.4680056571960449,
+      "gl_distil_loss": 0.00793403573334217,
+      "gl_dms_closed_frac": 0.8118024468421936,
+      "gl_dms_cr": 5.31453800201416,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.764705657958984,
+      "gl_dms_target_frac": 0.790123462677002,
+      "gl_eos_tokens": 10.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.00793403573334217,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.0,
+      "gl_positions_without_loss_calculation": 10.0,
+      "grad_norm": 1.0643372535705566,
+      "learning_rate": 3e-05,
+      "loss": 0.0079,
+      "step": 129
+    },
+    {
+      "epoch": 0.26,
+      "gl_detached_lm_loss": 0.47951599955558777,
+      "gl_distil_loss": 0.007271426264196634,
+      "gl_dms_closed_frac": 0.8056186437606812,
+      "gl_dms_cr": 5.148845672607422,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.794117450714111,
+      "gl_dms_target_frac": 0.7914109826087952,
+      "gl_eos_tokens": 11.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.007271426264196634,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.75,
+      "gl_positions_without_loss_calculation": 11.25,
+      "grad_norm": 2.485182046890259,
+      "learning_rate": 3e-05,
+      "loss": 0.0073,
+      "step": 130
+    },
+    {
+      "epoch": 0.262,
+      "gl_detached_lm_loss": 0.4433704912662506,
+      "gl_distil_loss": 0.006428849883377552,
+      "gl_dms_closed_frac": 0.8004820942878723,
+      "gl_dms_cr": 5.012540817260742,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.82352876663208,
+      "gl_dms_target_frac": 0.7926830053329468,
+      "gl_eos_tokens": 9.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.006428849883377552,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.5,
+      "gl_positions_without_loss_calculation": 9.5,
+      "grad_norm": 1.4084274768829346,
+      "learning_rate": 3e-05,
+      "loss": 0.0064,
+      "step": 131
+    },
+    {
+      "epoch": 0.264,
+      "gl_detached_lm_loss": 0.4866332709789276,
+      "gl_distil_loss": 0.006288310047239065,
+      "gl_dms_closed_frac": 0.78985196352005,
+      "gl_dms_cr": 4.759027481079102,
+      "gl_dms_loss": 0.004087455570697784,
+      "gl_dms_target_cr": 4.852941036224365,
+      "gl_dms_target_frac": 0.793939471244812,
+      "gl_eos_tokens": 11.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.010375766083598137,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.75,
+      "gl_positions_without_loss_calculation": 11.25,
+      "grad_norm": 13.989530563354492,
+      "learning_rate": 3e-05,
+      "loss": 0.0104,
+      "step": 132
+    },
+    {
+      "epoch": 0.266,
+      "gl_detached_lm_loss": 0.4728991985321045,
+      "gl_distil_loss": 0.00651888782158494,
+      "gl_dms_closed_frac": 0.7894306182861328,
+      "gl_dms_cr": 4.750518321990967,
+      "gl_dms_loss": 0.005873598158359528,
+      "gl_dms_target_cr": 4.882352828979492,
+      "gl_dms_target_frac": 0.7951807379722595,
+      "gl_eos_tokens": 10.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.01239248737692833,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.75,
+      "gl_positions_without_loss_calculation": 10.25,
+      "grad_norm": 12.475295066833496,
+      "learning_rate": 3e-05,
+      "loss": 0.0124,
+      "step": 133
+    },
+    {
+      "epoch": 0.268,
+      "gl_detached_lm_loss": 0.43680471181869507,
+      "gl_distil_loss": 0.005585328210145235,
+      "gl_dms_closed_frac": 0.7920165657997131,
+      "gl_dms_cr": 4.809713363647461,
+      "gl_dms_loss": 0.004696056246757507,
+      "gl_dms_target_cr": 4.911764621734619,
+      "gl_dms_target_frac": 0.796407163143158,
+      "gl_eos_tokens": 12.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.010281385853886604,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.75,
+      "gl_positions_without_loss_calculation": 12.25,
+      "grad_norm": 11.904838562011719,
+      "learning_rate": 3e-05,
+      "loss": 0.0103,
+      "step": 134
+    },
+    {
+      "epoch": 0.27,
+      "gl_detached_lm_loss": 0.49193722009658813,
+      "gl_distil_loss": 0.006255386397242546,
+      "gl_dms_closed_frac": 0.794114351272583,
+      "gl_dms_cr": 4.860118865966797,
+      "gl_dms_loss": 0.004544064402580261,
+      "gl_dms_target_cr": 4.941176891326904,
+      "gl_dms_target_frac": 0.7976189851760864,
+      "gl_eos_tokens": 10.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.010799449868500233,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.75,
+      "gl_positions_without_loss_calculation": 10.25,
+      "grad_norm": 10.285233497619629,
+      "learning_rate": 3e-05,
+      "loss": 0.0108,
+      "step": 135
+    },
+    {
+      "epoch": 0.272,
+      "gl_detached_lm_loss": 0.454690158367157,
+      "gl_distil_loss": 0.006719416473060846,
+      "gl_dms_closed_frac": 0.8077540397644043,
+      "gl_dms_cr": 5.204880237579346,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 4.970588207244873,
+      "gl_dms_target_frac": 0.7988166213035583,
+      "gl_eos_tokens": 11.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.006719416473060846,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.25,
+      "gl_positions_without_loss_calculation": 11.75,
+      "grad_norm": 8.923818588256836,
+      "learning_rate": 3e-05,
+      "loss": 0.0067,
+      "step": 136
+    },
+    {
+      "epoch": 0.274,
+      "gl_detached_lm_loss": 0.45901381969451904,
+      "gl_distil_loss": 0.007275748066604137,
+      "gl_dms_closed_frac": 0.8141677975654602,
+      "gl_dms_cr": 5.384031295776367,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.0,
+      "gl_dms_target_frac": 0.8000000715255737,
+      "gl_eos_tokens": 11.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.007275748066604137,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.0,
+      "gl_positions_without_loss_calculation": 11.0,
+      "grad_norm": 3.378786325454712,
+      "learning_rate": 3e-05,
+      "loss": 0.0073,
+      "step": 137
+    },
+    {
+      "epoch": 0.276,
+      "gl_detached_lm_loss": 0.4887048006057739,
+      "gl_distil_loss": 0.007861443795263767,
+      "gl_dms_closed_frac": 0.8187439441680908,
+      "gl_dms_cr": 5.519390106201172,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.029411792755127,
+      "gl_dms_target_frac": 0.8011695146560669,
+      "gl_eos_tokens": 11.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.007861443795263767,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.25,
+      "gl_positions_without_loss_calculation": 11.75,
+      "grad_norm": 1.7097647190093994,
+      "learning_rate": 3e-05,
+      "loss": 0.0079,
+      "step": 138
+    },
+    {
+      "epoch": 0.278,
+      "gl_detached_lm_loss": 0.4592478573322296,
+      "gl_distil_loss": 0.0075279176235198975,
+      "gl_dms_closed_frac": 0.8244554400444031,
+      "gl_dms_cr": 5.699601650238037,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.058823108673096,
+      "gl_dms_target_frac": 0.8023256659507751,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0075279176235198975,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 1.3076081275939941,
+      "learning_rate": 3e-05,
+      "loss": 0.0075,
+      "step": 139
+    },
+    {
+      "epoch": 0.28,
+      "gl_detached_lm_loss": 0.46759268641471863,
+      "gl_distil_loss": 0.008083690889179707,
+      "gl_dms_closed_frac": 0.8244903087615967,
+      "gl_dms_cr": 5.70239782333374,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.088235378265381,
+      "gl_dms_target_frac": 0.8034682273864746,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.008083690889179707,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 2.303365468978882,
+      "learning_rate": 3e-05,
+      "loss": 0.0081,
+      "step": 140
+    },
+    {
+      "epoch": 0.282,
+      "gl_detached_lm_loss": 0.4478227496147156,
+      "gl_distil_loss": 0.0072470554150640965,
+      "gl_dms_closed_frac": 0.8226778507232666,
+      "gl_dms_cr": 5.6427435874938965,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.117647171020508,
+      "gl_dms_target_frac": 0.8045977354049683,
+      "gl_eos_tokens": 11.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0072470554150640965,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.25,
+      "gl_positions_without_loss_calculation": 11.75,
+      "grad_norm": 2.6740522384643555,
+      "learning_rate": 3e-05,
+      "loss": 0.0072,
+      "step": 141
+    },
+    {
+      "epoch": 0.284,
+      "gl_detached_lm_loss": 0.4702463150024414,
+      "gl_distil_loss": 0.0071958452463150024,
+      "gl_dms_closed_frac": 0.8195258975028992,
+      "gl_dms_cr": 5.542600631713867,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.147058963775635,
+      "gl_dms_target_frac": 0.8057142496109009,
+      "gl_eos_tokens": 10.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0071958452463150024,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.25,
+      "gl_positions_without_loss_calculation": 10.75,
+      "grad_norm": 3.9141881465911865,
+      "learning_rate": 3e-05,
+      "loss": 0.0072,
+      "step": 142
+    },
+    {
+      "epoch": 0.286,
+      "gl_detached_lm_loss": 0.4721701443195343,
+      "gl_distil_loss": 0.006920252926647663,
+      "gl_dms_closed_frac": 0.8142572045326233,
+      "gl_dms_cr": 5.386926651000977,
+      "gl_dms_loss": 5.720555782318115e-05,
+      "gl_dms_target_cr": 5.17647123336792,
+      "gl_dms_target_frac": 0.8068181276321411,
+      "gl_eos_tokens": 13.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.006977458484470844,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32754.75,
+      "gl_positions_without_loss_calculation": 13.25,
+      "grad_norm": 1.4146333932876587,
+      "learning_rate": 3e-05,
+      "loss": 0.007,
+      "step": 143
+    },
+    {
+      "epoch": 0.288,
+      "gl_detached_lm_loss": 0.4847974479198456,
+      "gl_distil_loss": 0.0072076404467225075,
+      "gl_dms_closed_frac": 0.813251793384552,
+      "gl_dms_cr": 5.357405185699463,
+      "gl_dms_loss": 7.994472980499268e-05,
+      "gl_dms_target_cr": 5.205882549285889,
+      "gl_dms_target_frac": 0.8079095482826233,
+      "gl_eos_tokens": 12.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0072875851765275,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.75,
+      "gl_positions_without_loss_calculation": 12.25,
+      "grad_norm": 1.362390160560608,
+      "learning_rate": 3e-05,
+      "loss": 0.0073,
+      "step": 144
+    },
+    {
+      "epoch": 0.29,
+      "gl_detached_lm_loss": 0.46022504568099976,
+      "gl_distil_loss": 0.00686027854681015,
+      "gl_dms_closed_frac": 0.8179160356521606,
+      "gl_dms_cr": 5.494892120361328,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.235294342041016,
+      "gl_dms_target_frac": 0.8089886903762817,
+      "gl_eos_tokens": 11.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.00686027854681015,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.25,
+      "gl_positions_without_loss_calculation": 11.75,
+      "grad_norm": 1.7687309980392456,
+      "learning_rate": 3e-05,
+      "loss": 0.0069,
+      "step": 145
+    },
+    {
+      "epoch": 0.292,
+      "gl_detached_lm_loss": 0.46016189455986023,
+      "gl_distil_loss": 0.0068475911393761635,
+      "gl_dms_closed_frac": 0.8167369365692139,
+      "gl_dms_cr": 5.463459014892578,
+      "gl_dms_loss": 0.00013460218906402588,
+      "gl_dms_target_cr": 5.264705657958984,
+      "gl_dms_target_frac": 0.8100557923316956,
+      "gl_eos_tokens": 12.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.006982193328440189,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.0,
+      "gl_positions_without_loss_calculation": 12.0,
+      "grad_norm": 2.82254695892334,
+      "learning_rate": 3e-05,
+      "loss": 0.007,
+      "step": 146
+    },
+    {
+      "epoch": 0.294,
+      "gl_detached_lm_loss": 0.46145033836364746,
+      "gl_distil_loss": 0.00709201954305172,
+      "gl_dms_closed_frac": 0.8229960799217224,
+      "gl_dms_cr": 5.653026103973389,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.294117450714111,
+      "gl_dms_target_frac": 0.8111110329627991,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.00709201954305172,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 1.2006562948226929,
+      "learning_rate": 3e-05,
+      "loss": 0.0071,
+      "step": 147
+    },
+    {
+      "epoch": 0.296,
+      "gl_detached_lm_loss": 0.4399712383747101,
+      "gl_distil_loss": 0.007063089869916439,
+      "gl_dms_closed_frac": 0.8253207802772522,
+      "gl_dms_cr": 5.728308200836182,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.32352876663208,
+      "gl_dms_target_frac": 0.8121547102928162,
+      "gl_eos_tokens": 10.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.007063089869916439,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.0,
+      "gl_positions_without_loss_calculation": 10.0,
+      "grad_norm": 2.0583231449127197,
+      "learning_rate": 3e-05,
+      "loss": 0.0071,
+      "step": 148
+    },
+    {
+      "epoch": 0.298,
+      "gl_detached_lm_loss": 0.4723561108112335,
+      "gl_distil_loss": 0.007901018485426903,
+      "gl_dms_closed_frac": 0.8258881568908691,
+      "gl_dms_cr": 5.744671821594238,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.352940559387207,
+      "gl_dms_target_frac": 0.813186764717102,
+      "gl_eos_tokens": 10.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.007901018485426903,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.25,
+      "gl_positions_without_loss_calculation": 10.75,
+      "grad_norm": 1.7985323667526245,
+      "learning_rate": 3e-05,
+      "loss": 0.0079,
+      "step": 149
+    },
+    {
+      "epoch": 0.3,
+      "gl_detached_lm_loss": 0.4717611074447632,
+      "gl_distil_loss": 0.008037258870899677,
+      "gl_dms_closed_frac": 0.8242531418800354,
+      "gl_dms_cr": 5.693231105804443,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.382352828979492,
+      "gl_dms_target_frac": 0.8142076134681702,
+      "gl_eos_tokens": 10.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.008037258870899677,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.25,
+      "gl_positions_without_loss_calculation": 10.75,
+      "grad_norm": 1.1288707256317139,
+      "learning_rate": 3e-05,
+      "loss": 0.008,
+      "step": 150
+    },
+    {
+      "epoch": 0.302,
+      "gl_detached_lm_loss": 0.46595779061317444,
+      "gl_distil_loss": 0.006495564244687557,
+      "gl_dms_closed_frac": 0.8167760968208313,
+      "gl_dms_cr": 5.466947555541992,
+      "gl_dms_loss": 0.0025366097688674927,
+      "gl_dms_target_cr": 5.411765098571777,
+      "gl_dms_target_frac": 0.8152174353599548,
+      "gl_eos_tokens": 13.875,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.00903217401355505,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32754.125,
+      "gl_positions_without_loss_calculation": 13.875,
+      "grad_norm": 4.6393632888793945,
+      "learning_rate": 3e-05,
+      "loss": 0.009,
+      "step": 151
+    },
+    {
+      "epoch": 0.304,
+      "gl_detached_lm_loss": 0.4414372742176056,
+      "gl_distil_loss": 0.006566391792148352,
+      "gl_dms_closed_frac": 0.8200860023498535,
+      "gl_dms_cr": 5.560153484344482,
+      "gl_dms_loss": 0.000374525785446167,
+      "gl_dms_target_cr": 5.441176891326904,
+      "gl_dms_target_frac": 0.8162163496017456,
+      "gl_eos_tokens": 12.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.006940917111933231,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.25,
+      "gl_positions_without_loss_calculation": 12.75,
+      "grad_norm": 3.3325436115264893,
+      "learning_rate": 3e-05,
+      "loss": 0.0069,
+      "step": 152
+    },
+    {
+      "epoch": 0.306,
+      "gl_detached_lm_loss": 0.4721607267856598,
+      "gl_distil_loss": 0.007050786633044481,
+      "gl_dms_closed_frac": 0.8164743185043335,
+      "gl_dms_cr": 5.452330112457275,
+      "gl_dms_loss": 0.0020945072174072266,
+      "gl_dms_target_cr": 5.470588207244873,
+      "gl_dms_target_frac": 0.8172043561935425,
+      "gl_eos_tokens": 10.625,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.009145294316112995,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.375,
+      "gl_positions_without_loss_calculation": 10.625,
+      "grad_norm": 4.4689860343933105,
+      "learning_rate": 3e-05,
+      "loss": 0.0091,
+      "step": 153
+    },
+    {
+      "epoch": 0.308,
+      "gl_detached_lm_loss": 0.49011290073394775,
+      "gl_distil_loss": 0.007768554147332907,
+      "gl_dms_closed_frac": 0.8201054930686951,
+      "gl_dms_cr": 5.564466953277588,
+      "gl_dms_loss": 0.0018368959426879883,
+      "gl_dms_target_cr": 5.5,
+      "gl_dms_target_frac": 0.8181818723678589,
+      "gl_eos_tokens": 9.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.009605450555682182,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.5,
+      "gl_positions_without_loss_calculation": 9.5,
+      "grad_norm": 4.584949493408203,
+      "learning_rate": 3e-05,
+      "loss": 0.0096,
+      "step": 154
+    },
+    {
+      "epoch": 0.31,
+      "gl_detached_lm_loss": 0.4801517724990845,
+      "gl_distil_loss": 0.0074509927071630955,
+      "gl_dms_closed_frac": 0.829155683517456,
+      "gl_dms_cr": 5.857370376586914,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.529411792755127,
+      "gl_dms_target_frac": 0.8191489577293396,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0074509927071630955,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 0.992720365524292,
+      "learning_rate": 3e-05,
+      "loss": 0.0075,
+      "step": 155
+    },
+    {
+      "epoch": 0.312,
+      "gl_detached_lm_loss": 0.47425395250320435,
+      "gl_distil_loss": 0.008633583784103394,
+      "gl_dms_closed_frac": 0.8352111577987671,
+      "gl_dms_cr": 6.070135593414307,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.558823108673096,
+      "gl_dms_target_frac": 0.8201056718826294,
+      "gl_eos_tokens": 10.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.008633583784103394,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.5,
+      "gl_positions_without_loss_calculation": 10.5,
+      "grad_norm": 1.5924896001815796,
+      "learning_rate": 3e-05,
+      "loss": 0.0086,
+      "step": 156
+    },
+    {
+      "epoch": 0.314,
+      "gl_detached_lm_loss": 0.4853774905204773,
+      "gl_distil_loss": 0.008461525663733482,
+      "gl_dms_closed_frac": 0.8332453370094299,
+      "gl_dms_cr": 6.001097679138184,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.588234901428223,
+      "gl_dms_target_frac": 0.821052610874176,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.008461525663733482,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 1.1708149909973145,
+      "learning_rate": 3e-05,
+      "loss": 0.0085,
+      "step": 157
+    },
+    {
+      "epoch": 0.316,
+      "gl_detached_lm_loss": 0.5149613618850708,
+      "gl_distil_loss": 0.008489280007779598,
+      "gl_dms_closed_frac": 0.8285330533981323,
+      "gl_dms_cr": 5.833188056945801,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.617647171020508,
+      "gl_dms_target_frac": 0.8219895362854004,
+      "gl_eos_tokens": 11.5,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.008489280007779598,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.5,
+      "gl_positions_without_loss_calculation": 11.5,
+      "grad_norm": 1.376883864402771,
+      "learning_rate": 3e-05,
+      "loss": 0.0085,
+      "step": 158
+    },
+    {
+      "epoch": 0.318,
+      "gl_detached_lm_loss": 0.4815997779369354,
+      "gl_distil_loss": 0.007431944832205772,
+      "gl_dms_closed_frac": 0.8258528709411621,
+      "gl_dms_cr": 5.750091552734375,
+      "gl_dms_loss": 0.0016558989882469177,
+      "gl_dms_target_cr": 5.647059440612793,
+      "gl_dms_target_frac": 0.8229166269302368,
+      "gl_eos_tokens": 10.875,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.00908784382045269,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.125,
+      "gl_positions_without_loss_calculation": 10.875,
+      "grad_norm": 2.853998899459839,
+      "learning_rate": 3e-05,
+      "loss": 0.0091,
+      "step": 159
+    },
+    {
+      "epoch": 0.32,
+      "gl_detached_lm_loss": 0.4931546449661255,
+      "gl_distil_loss": 0.0077752345241606236,
+      "gl_dms_closed_frac": 0.8248758912086487,
+      "gl_dms_cr": 5.713653087615967,
+      "gl_dms_loss": 0.001398041844367981,
+      "gl_dms_target_cr": 5.67647123336792,
+      "gl_dms_target_frac": 0.8238343000411987,
+      "gl_eos_tokens": 9.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.009173276834189892,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32758.25,
+      "gl_positions_without_loss_calculation": 9.75,
+      "grad_norm": 5.879312515258789,
+      "learning_rate": 3e-05,
+      "loss": 0.0092,
+      "step": 160
+    },
+    {
+      "epoch": 0.322,
+      "gl_detached_lm_loss": 0.49633410573005676,
+      "gl_distil_loss": 0.007922427728772163,
+      "gl_dms_closed_frac": 0.8279393911361694,
+      "gl_dms_cr": 5.816441535949707,
+      "gl_dms_loss": 0.0007596909999847412,
+      "gl_dms_target_cr": 5.705882549285889,
+      "gl_dms_target_frac": 0.8247422575950623,
+      "gl_eos_tokens": 11.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.008682118728756905,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.25,
+      "gl_positions_without_loss_calculation": 11.75,
+      "grad_norm": 4.280725002288818,
+      "learning_rate": 3e-05,
+      "loss": 0.0087,
+      "step": 161
+    },
+    {
+      "epoch": 0.324,
+      "gl_detached_lm_loss": 0.46856874227523804,
+      "gl_distil_loss": 0.0076766908168792725,
+      "gl_dms_closed_frac": 0.8342061042785645,
+      "gl_dms_cr": 6.03446102142334,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.735294342041016,
+      "gl_dms_target_frac": 0.8256410956382751,
+      "gl_eos_tokens": 12.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0076766908168792725,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.25,
+      "gl_positions_without_loss_calculation": 12.75,
+      "grad_norm": 1.2406954765319824,
+      "learning_rate": 3e-05,
+      "loss": 0.0077,
+      "step": 162
+    },
+    {
+      "epoch": 0.326,
+      "gl_detached_lm_loss": 0.4626222252845764,
+      "gl_distil_loss": 0.007656690198928118,
+      "gl_dms_closed_frac": 0.8375633358955383,
+      "gl_dms_cr": 6.158156394958496,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.764705657958984,
+      "gl_dms_target_frac": 0.8265305757522583,
+      "gl_eos_tokens": 11.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.007656690198928118,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.25,
+      "gl_positions_without_loss_calculation": 11.75,
+      "grad_norm": 1.0347411632537842,
+      "learning_rate": 3e-05,
+      "loss": 0.0077,
+      "step": 163
+    },
+    {
+      "epoch": 0.328,
+      "gl_detached_lm_loss": 0.465984582901001,
+      "gl_distil_loss": 0.008057629689574242,
+      "gl_dms_closed_frac": 0.8382847905158997,
+      "gl_dms_cr": 6.185460090637207,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.794117450714111,
+      "gl_dms_target_frac": 0.8274111747741699,
+      "gl_eos_tokens": 13.125,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.008057629689574242,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32754.875,
+      "gl_positions_without_loss_calculation": 13.125,
+      "grad_norm": 1.1180733442306519,
+      "learning_rate": 3e-05,
+      "loss": 0.0081,
+      "step": 164
+    },
+    {
+      "epoch": 0.33,
+      "gl_detached_lm_loss": 0.4694860577583313,
+      "gl_distil_loss": 0.007432657293975353,
+      "gl_dms_closed_frac": 0.8331831097602844,
+      "gl_dms_cr": 5.996460914611816,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.82352876663208,
+      "gl_dms_target_frac": 0.8282828330993652,
+      "gl_eos_tokens": 11.625,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.007432657293975353,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.375,
+      "gl_positions_without_loss_calculation": 11.625,
+      "grad_norm": 1.125787615776062,
+      "learning_rate": 3e-05,
+      "loss": 0.0074,
+      "step": 165
+    },
+    {
+      "epoch": 0.332,
+      "gl_detached_lm_loss": 0.4963419437408447,
+      "gl_distil_loss": 0.008539569564163685,
+      "gl_dms_closed_frac": 0.8266037702560425,
+      "gl_dms_cr": 5.772175312042236,
+      "gl_dms_loss": 0.003481827676296234,
+      "gl_dms_target_cr": 5.852940559387207,
+      "gl_dms_target_frac": 0.8291457891464233,
+      "gl_eos_tokens": 10.125,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.012021397240459919,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.875,
+      "gl_positions_without_loss_calculation": 10.125,
+      "grad_norm": 5.81721830368042,
+      "learning_rate": 3e-05,
+      "loss": 0.012,
+      "step": 166
+    },
+    {
+      "epoch": 0.334,
+      "gl_detached_lm_loss": 0.47259899973869324,
+      "gl_distil_loss": 0.007361316587775946,
+      "gl_dms_closed_frac": 0.8257991075515747,
+      "gl_dms_cr": 5.7416534423828125,
+      "gl_dms_loss": 0.004307880997657776,
+      "gl_dms_target_cr": 5.882352828979492,
+      "gl_dms_target_frac": 0.8299999833106995,
+      "gl_eos_tokens": 10.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.011669197119772434,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32757.25,
+      "gl_positions_without_loss_calculation": 10.75,
+      "grad_norm": 10.337682723999023,
+      "learning_rate": 3e-05,
+      "loss": 0.0117,
+      "step": 167
+    },
+    {
+      "epoch": 0.336,
+      "gl_detached_lm_loss": 0.4550955891609192,
+      "gl_distil_loss": 0.0070419758558273315,
+      "gl_dms_closed_frac": 0.8298782110214233,
+      "gl_dms_cr": 5.879843711853027,
+      "gl_dms_loss": 0.0018124282360076904,
+      "gl_dms_target_cr": 5.911765098571777,
+      "gl_dms_target_frac": 0.8308457732200623,
+      "gl_eos_tokens": 12.25,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.008854404091835022,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32755.75,
+      "gl_positions_without_loss_calculation": 12.25,
+      "grad_norm": 5.435131072998047,
+      "learning_rate": 3e-05,
+      "loss": 0.0089,
+      "step": 168
+    },
+    {
+      "epoch": 0.338,
+      "gl_detached_lm_loss": 0.4601227045059204,
+      "gl_distil_loss": 0.007701840251684189,
+      "gl_dms_closed_frac": 0.835907518863678,
+      "gl_dms_cr": 6.098821640014648,
+      "gl_dms_loss": 0.0005731061100959778,
+      "gl_dms_target_cr": 5.941176891326904,
+      "gl_dms_target_frac": 0.8316831588745117,
+      "gl_eos_tokens": 12.0,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.008274946361780167,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.0,
+      "gl_positions_without_loss_calculation": 12.0,
+      "grad_norm": 0.9658327698707581,
+      "learning_rate": 3e-05,
+      "loss": 0.0083,
+      "step": 169
+    },
+    {
+      "epoch": 0.34,
+      "gl_detached_lm_loss": 0.4400879442691803,
+      "gl_distil_loss": 0.0074328347109258175,
+      "gl_dms_closed_frac": 0.8415706753730774,
+      "gl_dms_cr": 6.313007354736328,
+      "gl_dms_loss": 0.0,
+      "gl_dms_target_cr": 5.970588207244873,
+      "gl_dms_target_frac": 0.8325123190879822,
+      "gl_eos_tokens": 11.75,
+      "gl_input_tokens": 32768.0,
+      "gl_loss": 0.0074328347109258175,
+      "gl_masked_tokens": 0.0,
+      "gl_positions_for_loss_calculation": 32756.25,
+      "gl_positions_without_loss_calculation": 11.75,
+      "grad_norm": 1.949934720993042,
+      "learning_rate": 3e-05,
+      "loss": 0.0074,
+      "step": 170
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 238,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 34,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}