Upload folder using huggingface_hub

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a3179b2c4f6785813b0f5065f82cd5a5296e0eb7726884be39b0968b0cd88035
+oid sha256:06971a1c5dc99302c21544d39919da69c140211abb2df19086008fafcdcd40da
 size 1192135096

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:92277ed941097724fe3c18c1ac05922c6bd1c00f6f294722dee5279240f0ba8d
+oid sha256:eae3d5e0b7bf1febbdad1a179b3c1a583dfb50c416c51cda8a0bf6b0fb503018
 size 2384459962

trainer_state.json

@@ -2,933 +2,1850 @@
   "best_global_step": null,
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 2.9993142857142856,
+  "epoch": 2.998857142857143,
   "eval_steps": 500,
-  "global_step": 3279,
+  "global_step": 6561,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.022857142857142857,
-      "grad_norm": 8.125,
-      "learning_rate": 3.6585365853658537e-06,
-      "loss": 1.358,
+      "epoch": 0.011428571428571429,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.8264840182648401e-06,
+      "loss": 1.3625,
       "step": 25
     },
     {
-      "epoch": 0.045714285714285714,
-      "grad_norm": 4.75,
-      "learning_rate": 7.469512195121951e-06,
-      "loss": 1.2495,
+      "epoch": 0.022857142857142857,
+      "grad_norm": 6.46875,
+      "learning_rate": 3.7290715372907152e-06,
+      "loss": 1.3344,
       "step": 50
     },
     {
-      "epoch": 0.06857142857142857,
-      "grad_norm": 3.890625,
-      "learning_rate": 1.128048780487805e-05,
-      "loss": 1.1795,
+      "epoch": 0.03428571428571429,
+      "grad_norm": 5.25,
+      "learning_rate": 5.631659056316591e-06,
+      "loss": 1.2071,
       "step": 75
     },
     {
-      "epoch": 0.09142857142857143,
-      "grad_norm": 4.0625,
-      "learning_rate": 1.5091463414634147e-05,
-      "loss": 1.1428,
+      "epoch": 0.045714285714285714,
+      "grad_norm": 4.6875,
+      "learning_rate": 7.5342465753424655e-06,
+      "loss": 1.1908,
       "step": 100
     },
     {
-      "epoch": 0.11428571428571428,
-      "grad_norm": 4.28125,
-      "learning_rate": 1.8902439024390246e-05,
-      "loss": 1.1436,
+      "epoch": 0.05714285714285714,
+      "grad_norm": 3.875,
+      "learning_rate": 9.436834094368341e-06,
+      "loss": 1.1741,
       "step": 125
     },
     {
-      "epoch": 0.13714285714285715,
-      "grad_norm": 4.03125,
-      "learning_rate": 2.2713414634146344e-05,
-      "loss": 1.1044,
+      "epoch": 0.06857142857142857,
+      "grad_norm": 4.34375,
+      "learning_rate": 1.1339421613394216e-05,
+      "loss": 1.0985,
       "step": 150
     },
     {
-      "epoch": 0.16,
-      "grad_norm": 4.625,
-      "learning_rate": 2.652439024390244e-05,
-      "loss": 1.0569,
+      "epoch": 0.08,
+      "grad_norm": 4.1875,
+      "learning_rate": 1.3242009132420092e-05,
+      "loss": 1.1444,
       "step": 175
     },
     {
-      "epoch": 0.18285714285714286,
-      "grad_norm": 3.9375,
-      "learning_rate": 3.0335365853658536e-05,
-      "loss": 1.0175,
+      "epoch": 0.09142857142857143,
+      "grad_norm": 4.0,
+      "learning_rate": 1.5144596651445967e-05,
+      "loss": 1.0922,
       "step": 200
     },
     {
-      "epoch": 0.2057142857142857,
-      "grad_norm": 4.34375,
-      "learning_rate": 3.414634146341464e-05,
-      "loss": 1.0053,
+      "epoch": 0.10285714285714286,
+      "grad_norm": 3.671875,
+      "learning_rate": 1.7047184170471843e-05,
+      "loss": 1.0543,
       "step": 225
     },
     {
-      "epoch": 0.22857142857142856,
-      "grad_norm": 4.5,
-      "learning_rate": 3.7957317073170734e-05,
-      "loss": 1.0415,
+      "epoch": 0.11428571428571428,
+      "grad_norm": 4.90625,
+      "learning_rate": 1.8949771689497718e-05,
+      "loss": 1.0472,
       "step": 250
     },
     {
-      "epoch": 0.25142857142857145,
-      "grad_norm": 4.375,
-      "learning_rate": 4.176829268292683e-05,
-      "loss": 0.9895,
+      "epoch": 0.12571428571428572,
+      "grad_norm": 4.34375,
+      "learning_rate": 2.0852359208523592e-05,
+      "loss": 1.089,
       "step": 275
     },
     {
-      "epoch": 0.2742857142857143,
-      "grad_norm": 4.0625,
-      "learning_rate": 4.557926829268293e-05,
-      "loss": 0.9795,
+      "epoch": 0.13714285714285715,
+      "grad_norm": 4.21875,
+      "learning_rate": 2.2754946727549467e-05,
+      "loss": 1.0493,
       "step": 300
     },
     {
-      "epoch": 0.29714285714285715,
-      "grad_norm": 3.921875,
-      "learning_rate": 4.9390243902439024e-05,
-      "loss": 0.9568,
+      "epoch": 0.14857142857142858,
+      "grad_norm": 4.34375,
+      "learning_rate": 2.4657534246575342e-05,
+      "loss": 1.0405,
       "step": 325
     },
     {
-      "epoch": 0.32,
-      "grad_norm": 3.9375,
-      "learning_rate": 4.999375270749916e-05,
-      "loss": 1.0014,
+      "epoch": 0.16,
+      "grad_norm": 3.859375,
+      "learning_rate": 2.656012176560122e-05,
+      "loss": 1.0518,
       "step": 350
     },
     {
-      "epoch": 0.34285714285714286,
-      "grad_norm": 3.546875,
-      "learning_rate": 4.997002907071694e-05,
-      "loss": 0.9417,
+      "epoch": 0.17142857142857143,
+      "grad_norm": 3.953125,
+      "learning_rate": 2.846270928462709e-05,
+      "loss": 1.0101,
       "step": 375
     },
     {
-      "epoch": 0.3657142857142857,
+      "epoch": 0.18285714285714286,
       "grad_norm": 4.1875,
-      "learning_rate": 4.992861922480944e-05,
-      "loss": 0.9401,
+      "learning_rate": 3.036529680365297e-05,
+      "loss": 0.99,
       "step": 400
     },
     {
-      "epoch": 0.38857142857142857,
-      "grad_norm": 3.234375,
-      "learning_rate": 4.9869552500266925e-05,
-      "loss": 0.9604,
+      "epoch": 0.19428571428571428,
+      "grad_norm": 4.4375,
+      "learning_rate": 3.226788432267884e-05,
+      "loss": 1.0216,
       "step": 425
     },
     {
-      "epoch": 0.4114285714285714,
-      "grad_norm": 3.40625,
-      "learning_rate": 4.979287073390262e-05,
-      "loss": 0.9284,
+      "epoch": 0.2057142857142857,
+      "grad_norm": 4.34375,
+      "learning_rate": 3.417047184170472e-05,
+      "loss": 0.9716,
       "step": 450
     },
     {
-      "epoch": 0.4342857142857143,
-      "grad_norm": 3.71875,
-      "learning_rate": 4.969862823921981e-05,
-      "loss": 0.9228,
+      "epoch": 0.21714285714285714,
+      "grad_norm": 4.25,
+      "learning_rate": 3.60730593607306e-05,
+      "loss": 0.9779,
       "step": 475
     },
     {
-      "epoch": 0.45714285714285713,
-      "grad_norm": 4.0625,
-      "learning_rate": 4.958689176794172e-05,
-      "loss": 0.9457,
+      "epoch": 0.22857142857142856,
+      "grad_norm": 3.8125,
+      "learning_rate": 3.797564687975647e-05,
+      "loss": 0.95,
       "step": 500
     },
     {
-      "epoch": 0.48,
-      "grad_norm": 3.34375,
-      "learning_rate": 4.945774046273145e-05,
-      "loss": 0.8994,
+      "epoch": 0.24,
+      "grad_norm": 4.15625,
+      "learning_rate": 3.9878234398782346e-05,
+      "loss": 0.9273,
       "step": 525
     },
     {
-      "epoch": 0.5028571428571429,
-      "grad_norm": 3.40625,
-      "learning_rate": 4.9311265801135384e-05,
-      "loss": 0.8787,
+      "epoch": 0.25142857142857145,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.1780821917808224e-05,
+      "loss": 0.9343,
       "step": 550
     },
     {
-      "epoch": 0.5257142857142857,
-      "grad_norm": 4.1875,
-      "learning_rate": 4.9147571530789894e-05,
-      "loss": 0.908,
+      "epoch": 0.26285714285714284,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.3683409436834095e-05,
+      "loss": 0.9279,
       "step": 575
     },
     {
-      "epoch": 0.5485714285714286,
-      "grad_norm": 2.984375,
-      "learning_rate": 4.896677359593718e-05,
-      "loss": 0.865,
+      "epoch": 0.2742857142857143,
+      "grad_norm": 6.84375,
+      "learning_rate": 4.5585996955859973e-05,
+      "loss": 0.9683,
       "step": 600
     },
     {
-      "epoch": 0.5714285714285714,
-      "grad_norm": 3.984375,
-      "learning_rate": 4.8769000055302205e-05,
-      "loss": 0.8685,
+      "epoch": 0.2857142857142857,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.7488584474885845e-05,
+      "loss": 0.904,
       "step": 625
     },
     {
-      "epoch": 0.5942857142857143,
-      "grad_norm": 2.546875,
-      "learning_rate": 4.855439099138905e-05,
-      "loss": 0.8979,
+      "epoch": 0.29714285714285715,
+      "grad_norm": 4.71875,
+      "learning_rate": 4.939117199391172e-05,
+      "loss": 0.9745,
       "step": 650
     },
     {
-      "epoch": 0.6171428571428571,
-      "grad_norm": 3.875,
-      "learning_rate": 4.83230984112608e-05,
-      "loss": 0.9231,
+      "epoch": 0.30857142857142855,
+      "grad_norm": 4.25,
+      "learning_rate": 4.9998977150017555e-05,
+      "loss": 0.8935,
       "step": 675
     },
     {
-      "epoch": 0.64,
-      "grad_norm": 4.21875,
-      "learning_rate": 4.807528613887338e-05,
-      "loss": 0.8809,
+      "epoch": 0.32,
+      "grad_norm": 3.890625,
+      "learning_rate": 4.999375693918911e-05,
+      "loss": 0.8999,
       "step": 700
     },
     {
-      "epoch": 0.6628571428571428,
-      "grad_norm": 3.78125,
-      "learning_rate": 4.781112969903937e-05,
-      "loss": 0.8579,
+      "epoch": 0.3314285714285714,
+      "grad_norm": 4.875,
+      "learning_rate": 4.9984113776601504e-05,
+      "loss": 0.9305,
       "step": 725
     },
     {
-      "epoch": 0.6857142857142857,
-      "grad_norm": 3.390625,
-      "learning_rate": 4.75308161931043e-05,
-      "loss": 0.873,
+      "epoch": 0.34285714285714286,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.9970049368730605e-05,
+      "loss": 0.933,
       "step": 750
     },
     {
-      "epoch": 0.7085714285714285,
-      "grad_norm": 3.5625,
-      "learning_rate": 4.723454416642317e-05,
-      "loss": 0.8187,
+      "epoch": 0.35428571428571426,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.9951566204445834e-05,
+      "loss": 0.932,
       "step": 775
     },
     {
-      "epoch": 0.7314285714285714,
-      "grad_norm": 3.359375,
-      "learning_rate": 4.692252346773133e-05,
-      "loss": 0.8335,
+      "epoch": 0.3657142857142857,
+      "grad_norm": 3.890625,
+      "learning_rate": 4.992866755456975e-05,
+      "loss": 0.8781,
       "step": 800
     },
     {
-      "epoch": 0.7542857142857143,
-      "grad_norm": 3.5,
-      "learning_rate": 4.6594975100509094e-05,
-      "loss": 0.8578,
+      "epoch": 0.37714285714285717,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.990135747129923e-05,
+      "loss": 0.8927,
       "step": 825
     },
     {
-      "epoch": 0.7771428571428571,
-      "grad_norm": 3.65625,
-      "learning_rate": 4.625213106644558e-05,
-      "loss": 0.843,
+      "epoch": 0.38857142857142857,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.986964078748837e-05,
+      "loss": 0.8595,
       "step": 850
     },
     {
-      "epoch": 0.8,
-      "grad_norm": 3.6875,
-      "learning_rate": 4.589423420111244e-05,
-      "loss": 0.8211,
+      "epoch": 0.4,
+      "grad_norm": 2.84375,
+      "learning_rate": 4.983352311579329e-05,
+      "loss": 0.8953,
       "step": 875
     },
     {
-      "epoch": 0.8228571428571428,
-      "grad_norm": 3.328125,
-      "learning_rate": 4.552153800196407e-05,
-      "loss": 0.8575,
+      "epoch": 0.4114285714285714,
+      "grad_norm": 3.71875,
+      "learning_rate": 4.979301084767886e-05,
+      "loss": 0.8421,
       "step": 900
     },
     {
-      "epoch": 0.8457142857142858,
-      "grad_norm": 3.65625,
-      "learning_rate": 4.513430644878598e-05,
-      "loss": 0.813,
+      "epoch": 0.4228571428571429,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.974811115228767e-05,
+      "loss": 0.8636,
       "step": 925
     },
     {
-      "epoch": 0.8685714285714285,
-      "grad_norm": 3.53125,
-      "learning_rate": 4.4732813816718595e-05,
-      "loss": 0.7998,
+      "epoch": 0.4342857142857143,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.969883197517137e-05,
+      "loss": 0.8655,
       "step": 950
     },
     {
-      "epoch": 0.8914285714285715,
-      "grad_norm": 3.328125,
-      "learning_rate": 4.43173444819888e-05,
-      "loss": 0.8045,
+      "epoch": 0.44571428571428573,
+      "grad_norm": 4.21875,
+      "learning_rate": 4.964518203688462e-05,
+      "loss": 0.8487,
       "step": 975
     },
     {
-      "epoch": 0.9142857142857143,
-      "grad_norm": 4.125,
-      "learning_rate": 4.388819272048707e-05,
-      "loss": 0.7961,
+      "epoch": 0.45714285714285713,
+      "grad_norm": 4.0,
+      "learning_rate": 4.958717083144182e-05,
+      "loss": 0.8706,
       "step": 1000
     },
     {
-      "epoch": 0.9371428571428572,
-      "grad_norm": 3.53125,
-      "learning_rate": 4.3445662499332474e-05,
-      "loss": 0.8105,
+      "epoch": 0.4685714285714286,
+      "grad_norm": 3.25,
+      "learning_rate": 4.952480862463712e-05,
+      "loss": 0.8521,
       "step": 1025
     },
     {
-      "epoch": 0.96,
-      "grad_norm": 3.671875,
-      "learning_rate": 4.299006726157363e-05,
-      "loss": 0.8175,
+      "epoch": 0.48,
+      "grad_norm": 3.796875,
+      "learning_rate": 4.945810645222767e-05,
+      "loss": 0.8352,
       "step": 1050
     },
     {
-      "epoch": 0.9828571428571429,
-      "grad_norm": 2.828125,
-      "learning_rate": 4.2521729704177706e-05,
-      "loss": 0.7671,
+      "epoch": 0.49142857142857144,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.938707611798078e-05,
+      "loss": 0.8311,
       "step": 1075
     },
     {
-      "epoch": 1.0064,
-      "grad_norm": 4.59375,
-      "learning_rate": 4.2040981549465015e-05,
-      "loss": 0.8307,
+      "epoch": 0.5028571428571429,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.931173019158504e-05,
+      "loss": 0.8434,
       "step": 1100
     },
     {
-      "epoch": 1.0292571428571429,
-      "grad_norm": 2.8125,
-      "learning_rate": 4.154816331015088e-05,
-      "loss": 0.6611,
+      "epoch": 0.5142857142857142,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.923208200642602e-05,
+      "loss": 0.8155,
       "step": 1125
     },
     {
-      "epoch": 1.0521142857142858,
-      "grad_norm": 3.796875,
-      "learning_rate": 4.104362404816138e-05,
-      "loss": 0.6466,
+      "epoch": 0.5257142857142857,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.914814565722671e-05,
+      "loss": 0.8337,
       "step": 1150
     },
     {
-      "epoch": 1.0749714285714287,
-      "grad_norm": 4.03125,
-      "learning_rate": 4.052772112739359e-05,
-      "loss": 0.6549,
+      "epoch": 0.5371428571428571,
+      "grad_norm": 3.796875,
+      "learning_rate": 4.905993599755331e-05,
+      "loss": 0.838,
       "step": 1175
     },
     {
-      "epoch": 1.0978285714285714,
-      "grad_norm": 3.921875,
-      "learning_rate": 4.000081996059573e-05,
-      "loss": 0.653,
+      "epoch": 0.5485714285714286,
+      "grad_norm": 4.21875,
+      "learning_rate": 4.896746863718671e-05,
+      "loss": 0.8077,
       "step": 1200
     },
     {
-      "epoch": 1.1206857142857143,
-      "grad_norm": 3.125,
-      "learning_rate": 3.94632937505461e-05,
-      "loss": 0.6748,
+      "epoch": 0.56,
+      "grad_norm": 3.265625,
+      "learning_rate": 4.8870759939360136e-05,
+      "loss": 0.8583,
       "step": 1225
     },
     {
-      "epoch": 1.1435428571428572,
-      "grad_norm": 4.15625,
-      "learning_rate": 3.8915523225714636e-05,
-      "loss": 0.6696,
+      "epoch": 0.5714285714285714,
+      "grad_norm": 4.125,
+      "learning_rate": 4.8769827017863514e-05,
+      "loss": 0.8133,
       "step": 1250
     },
     {
-      "epoch": 1.1663999999999999,
+      "epoch": 0.5828571428571429,
       "grad_norm": 4.125,
-      "learning_rate": 3.835789637059372e-05,
-      "loss": 0.6514,
+      "learning_rate": 4.86646877340149e-05,
+      "loss": 0.8224,
       "step": 1275
     },
     {
-      "epoch": 1.1892571428571428,
-      "grad_norm": 3.96875,
-      "learning_rate": 3.7790808150889886e-05,
-      "loss": 0.6715,
+      "epoch": 0.5942857142857143,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.8555360693499786e-05,
+      "loss": 0.8146,
       "step": 1300
     },
     {
-      "epoch": 1.2121142857142857,
-      "grad_norm": 3.796875,
-      "learning_rate": 3.721466023377052e-05,
-      "loss": 0.6573,
+      "epoch": 0.6057142857142858,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.844186524307853e-05,
+      "loss": 0.8039,
       "step": 1325
     },
     {
-      "epoch": 1.2349714285714286,
-      "grad_norm": 3.65625,
-      "learning_rate": 3.662986070336399e-05,
-      "loss": 0.6389,
+      "epoch": 0.6171428571428571,
+      "grad_norm": 3.828125,
+      "learning_rate": 4.832422146716275e-05,
+      "loss": 0.8203,
       "step": 1350
     },
     {
-      "epoch": 1.2578285714285715,
-      "grad_norm": 3.84375,
-      "learning_rate": 3.603682377171476e-05,
-      "loss": 0.6474,
+      "epoch": 0.6285714285714286,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.8202450184261116e-05,
+      "loss": 0.7826,
       "step": 1375
     },
     {
-      "epoch": 1.2806857142857142,
-      "grad_norm": 4.03125,
-      "learning_rate": 3.5435969485397956e-05,
-      "loss": 0.6636,
+      "epoch": 0.64,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.807657294329527e-05,
+      "loss": 0.7926,
       "step": 1400
     },
     {
-      "epoch": 1.3035428571428571,
-      "grad_norm": 3.453125,
-      "learning_rate": 3.482772342800146e-05,
-      "loss": 0.647,
+      "epoch": 0.6514285714285715,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.79466120197865e-05,
+      "loss": 0.765,
       "step": 1425
     },
     {
-      "epoch": 1.3264,
-      "grad_norm": 6.1875,
-      "learning_rate": 3.421251641868613e-05,
-      "loss": 0.6186,
+      "epoch": 0.6628571428571428,
+      "grad_norm": 4.375,
+      "learning_rate": 4.781259041191375e-05,
+      "loss": 0.7731,
       "step": 1450
     },
     {
-      "epoch": 1.349257142857143,
-      "grad_norm": 3.5,
-      "learning_rate": 3.359078420703766e-05,
-      "loss": 0.6282,
+      "epoch": 0.6742857142857143,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.767453183644391e-05,
+      "loss": 0.8442,
       "step": 1475
     },
     {
-      "epoch": 1.3721142857142858,
-      "grad_norm": 4.0625,
-      "learning_rate": 3.2962967164426225e-05,
-      "loss": 0.6674,
+      "epoch": 0.6857142857142857,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.753246072453474e-05,
+      "loss": 0.736,
       "step": 1500
     },
     {
-      "epoch": 1.3949714285714285,
-      "grad_norm": 3.578125,
-      "learning_rate": 3.2329509972092524e-05,
-      "loss": 0.6192,
+      "epoch": 0.6971428571428572,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.7386402217411555e-05,
+      "loss": 0.8104,
       "step": 1525
     },
     {
-      "epoch": 1.4178285714285714,
-      "grad_norm": 3.5,
-      "learning_rate": 3.169086130618124e-05,
-      "loss": 0.6195,
+      "epoch": 0.7085714285714285,
+      "grad_norm": 3.671875,
+      "learning_rate": 4.723638216191811e-05,
+      "loss": 0.7847,
       "step": 1550
     },
     {
-      "epoch": 1.4406857142857143,
-      "grad_norm": 3.75,
-      "learning_rate": 3.104747351994476e-05,
-      "loss": 0.6093,
+      "epoch": 0.72,
+      "grad_norm": 3.625,
+      "learning_rate": 4.708242710594277e-05,
+      "loss": 0.7486,
       "step": 1575
     },
     {
-      "epoch": 1.463542857142857,
-      "grad_norm": 4.25,
-      "learning_rate": 3.0399802323342595e-05,
-      "loss": 0.6033,
+      "epoch": 0.7314285714285714,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.6924564293720434e-05,
+      "loss": 0.7868,
       "step": 1600
     },
     {
-      "epoch": 1.4864,
-      "grad_norm": 4.0,
-      "learning_rate": 2.9748306460263058e-05,
-      "loss": 0.5733,
+      "epoch": 0.7428571428571429,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.676282166101142e-05,
+      "loss": 0.7314,
       "step": 1625
     },
     {
-      "epoch": 1.5092571428571429,
-      "grad_norm": 4.5625,
-      "learning_rate": 2.9093447383596174e-05,
-      "loss": 0.6684,
+      "epoch": 0.7542857142857143,
+      "grad_norm": 3.625,
+      "learning_rate": 4.659722783015785e-05,
+      "loss": 0.7461,
       "step": 1650
     },
     {
-      "epoch": 1.5321142857142858,
-      "grad_norm": 5.34375,
-      "learning_rate": 2.8435688928387728e-05,
-      "loss": 0.6434,
+      "epoch": 0.7657142857142857,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.6427812105018576e-05,
+      "loss": 0.7334,
       "step": 1675
     },
     {
-      "epoch": 1.5549714285714287,
-      "grad_norm": 4.5,
-      "learning_rate": 2.7775496983306082e-05,
-      "loss": 0.6325,
+      "epoch": 0.7771428571428571,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.625460446578348e-05,
+      "loss": 0.7279,
       "step": 1700
     },
     {
-      "epoch": 1.5778285714285714,
-      "grad_norm": 4.375,
-      "learning_rate": 2.7113339160654405e-05,
-      "loss": 0.6183,
+      "epoch": 0.7885714285714286,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.6077635563668195e-05,
+      "loss": 0.7701,
       "step": 1725
     },
     {
-      "epoch": 1.6006857142857143,
-      "grad_norm": 4.40625,
-      "learning_rate": 2.6449684465162012e-05,
-      "loss": 0.6416,
+      "epoch": 0.8,
+      "grad_norm": 4.375,
+      "learning_rate": 4.5896936715489885e-05,
+      "loss": 0.7034,
       "step": 1750
     },
     {
-      "epoch": 1.6235428571428572,
-      "grad_norm": 4.375,
-      "learning_rate": 2.5785002961789496e-05,
-      "loss": 0.5869,
+      "epoch": 0.8114285714285714,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.571253989812545e-05,
+      "loss": 0.7281,
       "step": 1775
     },
     {
-      "epoch": 1.6463999999999999,
-      "grad_norm": 4.34375,
-      "learning_rate": 2.5119765442782926e-05,
-      "loss": 0.5745,
+      "epoch": 0.8228571428571428,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.5524477742852745e-05,
+      "loss": 0.7474,
       "step": 1800
     },
     {
-      "epoch": 1.669257142857143,
-      "grad_norm": 5.5625,
-      "learning_rate": 2.445444309421284e-05,
-      "loss": 0.6226,
+      "epoch": 0.8342857142857143,
+      "grad_norm": 4.40625,
+      "learning_rate": 4.5332783529576146e-05,
+      "loss": 0.739,
       "step": 1825
     },
     {
-      "epoch": 1.6921142857142857,
-      "grad_norm": 4.28125,
-      "learning_rate": 2.3789507162234347e-05,
-      "loss": 0.6089,
+      "epoch": 0.8457142857142858,
+      "grad_norm": 3.890625,
+      "learning_rate": 4.5137491180937196e-05,
+      "loss": 0.7537,
       "step": 1850
     },
     {
-      "epoch": 1.7149714285714286,
-      "grad_norm": 4.5625,
-      "learning_rate": 2.312542861930465e-05,
-      "loss": 0.5715,
+      "epoch": 0.8571428571428571,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.4938635256311634e-05,
+      "loss": 0.695,
       "step": 1875
     },
     {
-      "epoch": 1.7378285714285715,
-      "grad_norm": 4.25,
-      "learning_rate": 2.2462677830594438e-05,
-      "loss": 0.6236,
+      "epoch": 0.8685714285714285,
+      "grad_norm": 3.875,
+      "learning_rate": 4.4736250945693655e-05,
+      "loss": 0.7546,
       "step": 1900
     },
     {
-      "epoch": 1.7606857142857142,
-      "grad_norm": 4.34375,
-      "learning_rate": 2.1801724220829405e-05,
-      "loss": 0.6049,
+      "epoch": 0.88,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.453037406346862e-05,
+      "loss": 0.7416,
       "step": 1925
     },
     {
-      "epoch": 1.783542857142857,
-      "grad_norm": 3.671875,
-      "learning_rate": 2.1143035941797885e-05,
-      "loss": 0.6148,
+      "epoch": 0.8914285714285715,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.4321041042075254e-05,
+      "loss": 0.7364,
       "step": 1950
     },
     {
-      "epoch": 1.8064,
-      "grad_norm": 4.125,
-      "learning_rate": 2.048707954076014e-05,
-      "loss": 0.6017,
+      "epoch": 0.9028571428571428,
+      "grad_norm": 4.5625,
+      "learning_rate": 4.4108288925558505e-05,
+      "loss": 0.6852,
       "step": 1975
     },
     {
-      "epoch": 1.8292571428571427,
-      "grad_norm": 4.03125,
-      "learning_rate": 1.9834319629994052e-05,
-      "loss": 0.6137,
+      "epoch": 0.9142857142857143,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.389215536301412e-05,
+      "loss": 0.6998,
       "step": 2000
     },
     {
-      "epoch": 1.8521142857142858,
-      "grad_norm": 3.796875,
-      "learning_rate": 1.9185218557711483e-05,
-      "loss": 0.6015,
+      "epoch": 0.9257142857142857,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.3672678601926176e-05,
+      "loss": 0.756,
       "step": 2025
     },
     {
-      "epoch": 1.8749714285714285,
-      "grad_norm": 3.96875,
-      "learning_rate": 1.8540236080578116e-05,
-      "loss": 0.6043,
+      "epoch": 0.9371428571428572,
+      "grad_norm": 4.40625,
+      "learning_rate": 4.344989748139873e-05,
+      "loss": 0.7293,
       "step": 2050
     },
     {
-      "epoch": 1.8978285714285714,
-      "grad_norm": 5.03125,
-      "learning_rate": 1.7899829038069012e-05,
-      "loss": 0.6104,
+      "epoch": 0.9485714285714286,
+      "grad_norm": 3.84375,
+      "learning_rate": 4.3223851425282765e-05,
+      "loss": 0.6869,
       "step": 2075
     },
     {
-      "epoch": 1.9206857142857143,
-      "grad_norm": 4.15625,
-      "learning_rate": 1.7264451028890318e-05,
-      "loss": 0.6057,
+      "epoch": 0.96,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.299458043519964e-05,
+      "loss": 0.7269,
       "step": 2100
     },
     {
-      "epoch": 1.943542857142857,
-      "grad_norm": 4.25,
-      "learning_rate": 1.663455208969646e-05,
-      "loss": 0.6425,
+      "epoch": 0.9714285714285714,
+      "grad_norm": 4.0,
+      "learning_rate": 4.276212508346232e-05,
+      "loss": 0.6421,
       "step": 2125
     },
     {
-      "epoch": 1.9664000000000001,
-      "grad_norm": 3.4375,
-      "learning_rate": 1.6010578376330303e-05,
-      "loss": 0.565,
+      "epoch": 0.9828571428571429,
+      "grad_norm": 4.75,
+      "learning_rate": 4.252652650589563e-05,
+      "loss": 0.6815,
       "step": 2150
     },
     {
-      "epoch": 1.9892571428571428,
-      "grad_norm": 4.03125,
-      "learning_rate": 1.5392971847812086e-05,
-      "loss": 0.5628,
+      "epoch": 0.9942857142857143,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.228782639455674e-05,
+      "loss": 0.7155,
       "step": 2175
     },
     {
-      "epoch": 2.0128,
-      "grad_norm": 4.3125,
-      "learning_rate": 1.4782169953300934e-05,
-      "loss": 0.5853,
+      "epoch": 1.0054857142857143,
+      "grad_norm": 3.359375,
+      "learning_rate": 4.2046066990357235e-05,
+      "loss": 0.6069,
       "step": 2200
     },
     {
-      "epoch": 2.035657142857143,
-      "grad_norm": 3.1875,
-      "learning_rate": 1.4178605322250743e-05,
-      "loss": 0.4976,
+      "epoch": 1.0169142857142857,
+      "grad_norm": 4.65625,
+      "learning_rate": 4.18012910755881e-05,
+      "loss": 0.5746,
       "step": 2225
     },
     {
-      "epoch": 2.0585142857142857,
-      "grad_norm": 4.3125,
-      "learning_rate": 1.3582705457979773e-05,
-      "loss": 0.4977,
+      "epoch": 1.0283428571428572,
+      "grad_norm": 4.25,
+      "learning_rate": 4.155354196634886e-05,
+      "loss": 0.5889,
       "step": 2250
     },
     {
-      "epoch": 2.0813714285714284,
-      "grad_norm": 4.65625,
-      "learning_rate": 1.2994892434871209e-05,
-      "loss": 0.4918,
+      "epoch": 1.0397714285714286,
+      "grad_norm": 4.21875,
+      "learning_rate": 4.130286350488224e-05,
+      "loss": 0.5856,
       "step": 2275
     },
     {
-      "epoch": 2.1042285714285716,
-      "grad_norm": 4.375,
-      "learning_rate": 1.2415582599418812e-05,
-      "loss": 0.5095,
+      "epoch": 1.0512,
+      "grad_norm": 4.875,
+      "learning_rate": 4.1049300051815763e-05,
+      "loss": 0.5924,
       "step": 2300
     },
     {
-      "epoch": 2.1270857142857142,
-      "grad_norm": 3.765625,
-      "learning_rate": 1.1845186275329833e-05,
-      "loss": 0.4722,
+      "epoch": 1.0626285714285715,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.0792896478311614e-05,
+      "loss": 0.5592,
       "step": 2325
     },
     {
-      "epoch": 2.1499428571428574,
-      "grad_norm": 3.3125,
-      "learning_rate": 1.1284107472893707e-05,
-      "loss": 0.4935,
+      "epoch": 1.0740571428571428,
+      "grad_norm": 4.25,
+      "learning_rate": 4.0533698158126085e-05,
+      "loss": 0.5558,
       "step": 2350
     },
     {
-      "epoch": 2.1728,
-      "grad_norm": 3.921875,
-      "learning_rate": 1.0732743602822575e-05,
-      "loss": 0.5073,
+      "epoch": 1.0854857142857144,
+      "grad_norm": 4.40625,
+      "learning_rate": 4.0271750959580166e-05,
+      "loss": 0.5913,
       "step": 2375
     },
     {
-      "epoch": 2.1956571428571428,
-      "grad_norm": 3.984375,
-      "learning_rate": 1.0191485194766334e-05,
-      "loss": 0.5034,
+      "epoch": 1.0969142857142857,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.00071012374426e-05,
+      "loss": 0.5583,
       "step": 2400
     },
     {
-      "epoch": 2.218514285714286,
-      "grad_norm": 4.53125,
-      "learning_rate": 9.660715620701363e-06,
-      "loss": 0.5113,
+      "epoch": 1.108342857142857,
+      "grad_norm": 4.3125,
+      "learning_rate": 3.9739795824726804e-05,
+      "loss": 0.5687,
       "step": 2425
     },
     {
-      "epoch": 2.2413714285714286,
-      "grad_norm": 4.8125,
-      "learning_rate": 9.14081082338921e-06,
-      "loss": 0.51,
+      "epoch": 1.1197714285714286,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.946988202440321e-05,
+      "loss": 0.6084,
       "step": 2450
     },
     {
-      "epoch": 2.2642285714285713,
-      "grad_norm": 4.46875,
-      "learning_rate": 8.632139050097188e-06,
-      "loss": 0.4789,
+      "epoch": 1.1312,
+      "grad_norm": 5.125,
+      "learning_rate": 3.919740760102841e-05,
+      "loss": 0.5739,
       "step": 2475
     },
     {
-      "epoch": 2.2870857142857144,
-      "grad_norm": 4.25,
-      "learning_rate": 8.135060591769824e-06,
-      "loss": 0.5074,
+      "epoch": 1.1426285714285713,
+      "grad_norm": 3.96875,
+      "learning_rate": 3.8922420772292644e-05,
+      "loss": 0.5585,
       "step": 2500
     },
     {
-      "epoch": 2.309942857142857,
-      "grad_norm": 4.34375,
-      "learning_rate": 7.649927527835666e-06,
-      "loss": 0.4813,
+      "epoch": 1.154057142857143,
+      "grad_norm": 4.5625,
+      "learning_rate": 3.864497020048712e-05,
+      "loss": 0.5833,
       "step": 2525
     },
     {
-      "epoch": 2.3327999999999998,
-      "grad_norm": 3.578125,
-      "learning_rate": 7.177083476830254e-06,
-      "loss": 0.4911,
+      "epoch": 1.1654857142857142,
+      "grad_norm": 5.125,
+      "learning_rate": 3.8365104983892533e-05,
+      "loss": 0.5939,
       "step": 2550
     },
     {
-      "epoch": 2.355657142857143,
-      "grad_norm": 4.28125,
-      "learning_rate": 6.716863353012062e-06,
-      "loss": 0.4804,
+      "epoch": 1.1769142857142858,
+      "grad_norm": 4.25,
+      "learning_rate": 3.808287464809063e-05,
+      "loss": 0.5721,
       "step": 2575
     },
     {
-      "epoch": 2.3785142857142856,
-      "grad_norm": 4.03125,
-      "learning_rate": 6.269593129143478e-06,
-      "loss": 0.4875,
+      "epoch": 1.1883428571428571,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.77983291372e-05,
+      "loss": 0.5348,
       "step": 2600
     },
     {
-      "epoch": 2.4013714285714287,
-      "grad_norm": 4.15625,
-      "learning_rate": 5.83558960560518e-06,
-      "loss": 0.5143,
+      "epoch": 1.1997714285714285,
+      "grad_norm": 4.34375,
+      "learning_rate": 3.751151880503782e-05,
+      "loss": 0.6,
       "step": 2625
     },
     {
-      "epoch": 2.4242285714285714,
-      "grad_norm": 4.40625,
-      "learning_rate": 5.415160186007176e-06,
-      "loss": 0.4907,
+      "epoch": 1.2112,
+      "grad_norm": 4.25,
+      "learning_rate": 3.722249440620917e-05,
+      "loss": 0.538,
       "step": 2650
     },
     {
-      "epoch": 2.4470857142857145,
-      "grad_norm": 4.15625,
-      "learning_rate": 5.008602659455602e-06,
-      "loss": 0.4966,
+      "epoch": 1.2226285714285714,
+      "grad_norm": 5.0,
+      "learning_rate": 3.6931307087125425e-05,
+      "loss": 0.5264,
       "step": 2675
     },
     {
-      "epoch": 2.469942857142857,
-      "grad_norm": 4.28125,
-      "learning_rate": 4.61620498962948e-06,
-      "loss": 0.4712,
+      "epoch": 1.234057142857143,
+      "grad_norm": 3.609375,
+      "learning_rate": 3.663800837695315e-05,
+      "loss": 0.5233,
       "step": 2700
     },
     {
-      "epoch": 2.4928,
-      "grad_norm": 4.15625,
-      "learning_rate": 4.238245110816702e-06,
-      "loss": 0.5032,
+      "epoch": 1.2454857142857143,
+      "grad_norm": 3.953125,
+      "learning_rate": 3.634265017849549e-05,
+      "loss": 0.5722,
       "step": 2725
     },
     {
-      "epoch": 2.515657142857143,
-      "grad_norm": 3.734375,
-      "learning_rate": 3.8749907310539345e-06,
-      "loss": 0.4903,
+      "epoch": 1.2569142857142857,
+      "grad_norm": 2.921875,
+      "learning_rate": 3.604528475900729e-05,
+      "loss": 0.5632,
       "step": 2750
     },
     {
-      "epoch": 2.5385142857142857,
-      "grad_norm": 4.40625,
-      "learning_rate": 3.5266991425096234e-06,
-      "loss": 0.5226,
+      "epoch": 1.2683428571428572,
+      "grad_norm": 5.21875,
+      "learning_rate": 3.574596474094578e-05,
+      "loss": 0.5267,
       "step": 2775
     },
     {
-      "epoch": 2.5613714285714284,
-      "grad_norm": 3.6875,
-      "learning_rate": 3.1936170392446535e-06,
-      "loss": 0.472,
+      "epoch": 1.2797714285714286,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.544474309265834e-05,
+      "loss": 0.5508,
       "step": 2800
     },
     {
-      "epoch": 2.5842285714285715,
-      "grad_norm": 4.09375,
-      "learning_rate": 2.875980342479523e-06,
-      "loss": 0.4923,
+      "epoch": 1.2912,
+      "grad_norm": 4.3125,
+      "learning_rate": 3.514167311900917e-05,
+      "loss": 0.5592,
       "step": 2825
     },
     {
-      "epoch": 2.6070857142857142,
-      "grad_norm": 4.03125,
-      "learning_rate": 2.5740140334919337e-06,
-      "loss": 0.4965,
+      "epoch": 1.3026285714285715,
+      "grad_norm": 3.96875,
+      "learning_rate": 3.483680845194629e-05,
+      "loss": 0.5389,
       "step": 2850
     },
     {
-      "epoch": 2.629942857142857,
-      "grad_norm": 4.75,
-      "learning_rate": 2.287931994263151e-06,
-      "loss": 0.4886,
+      "epoch": 1.3140571428571428,
+      "grad_norm": 5.15625,
+      "learning_rate": 3.4530203041010745e-05,
+      "loss": 0.5379,
       "step": 2875
     },
     {
-      "epoch": 2.6528,
-      "grad_norm": 3.84375,
-      "learning_rate": 2.01793685598588e-06,
-      "loss": 0.4877,
+      "epoch": 1.3254857142857142,
+      "grad_norm": 4.25,
+      "learning_rate": 3.422191114378952e-05,
+      "loss": 0.5792,
       "step": 2900
     },
     {
-      "epoch": 2.6756571428571427,
-      "grad_norm": 3.296875,
-      "learning_rate": 1.7642198555411466e-06,
-      "loss": 0.5027,
+      "epoch": 1.3369142857142857,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.3911987316314054e-05,
+      "loss": 0.5165,
       "step": 2925
     },
     {
-      "epoch": 2.698514285714286,
-      "grad_norm": 4.25,
-      "learning_rate": 1.5269607000456543e-06,
-      "loss": 0.5272,
+      "epoch": 1.348342857142857,
+      "grad_norm": 5.40625,
+      "learning_rate": 3.360048640340585e-05,
+      "loss": 0.4876,
       "step": 2950
     },
     {
-      "epoch": 2.7213714285714286,
-      "grad_norm": 4.46875,
-      "learning_rate": 1.3063274395657187e-06,
-      "loss": 0.5149,
+      "epoch": 1.3597714285714286,
+      "grad_norm": 5.03125,
+      "learning_rate": 3.3287463528971e-05,
+      "loss": 0.5796,
       "step": 2975
     },
     {
-      "epoch": 2.7442285714285717,
-      "grad_norm": 4.125,
-      "learning_rate": 1.1024763480877886e-06,
-      "loss": 0.5165,
+      "epoch": 1.3712,
+      "grad_norm": 6.375,
+      "learning_rate": 3.2972974086245376e-05,
+      "loss": 0.5073,
       "step": 3000
     },
     {
-      "epoch": 2.7670857142857144,
-      "grad_norm": 3.421875,
-      "learning_rate": 9.155518128299983e-07,
-      "loss": 0.4632,
+      "epoch": 1.3826285714285715,
+      "grad_norm": 4.90625,
+      "learning_rate": 3.265707372799208e-05,
+      "loss": 0.5306,
       "step": 3025
     },
     {
-      "epoch": 2.789942857142857,
-      "grad_norm": 3.921875,
-      "learning_rate": 7.456862319730667e-07,
-      "loss": 0.5067,
+      "epoch": 1.3940571428571429,
+      "grad_norm": 4.46875,
+      "learning_rate": 3.233981835665301e-05,
+      "loss": 0.5312,
       "step": 3050
     },
     {
-      "epoch": 2.8128,
-      "grad_norm": 4.28125,
-      "learning_rate": 5.929999208829667e-07,
-      "loss": 0.4962,
+      "epoch": 1.4054857142857142,
+      "grad_norm": 4.78125,
+      "learning_rate": 3.202126411445624e-05,
+      "loss": 0.5244,
       "step": 3075
     },
     {
-      "epoch": 2.835657142857143,
-      "grad_norm": 3.90625,
-      "learning_rate": 4.576010268918879e-07,
-      "loss": 0.4745,
+      "epoch": 1.4169142857142858,
+      "grad_norm": 4.03125,
+      "learning_rate": 3.1701467373480995e-05,
+      "loss": 0.5403,
       "step": 3100
     },
     {
-      "epoch": 2.8585142857142856,
-      "grad_norm": 4.21875,
-      "learning_rate": 3.3958545269772434e-07,
-      "loss": 0.5049,
+      "epoch": 1.4283428571428571,
+      "grad_norm": 4.96875,
+      "learning_rate": 3.1380484725681876e-05,
+      "loss": 0.5161,
       "step": 3125
     },
     {
-      "epoch": 2.8813714285714287,
-      "grad_norm": 5.15625,
-      "learning_rate": 2.3903678843644473e-07,
-      "loss": 0.4917,
+      "epoch": 1.4397714285714285,
+      "grad_norm": 4.375,
+      "learning_rate": 3.105837297287423e-05,
+      "loss": 0.5414,
       "step": 3150
     },
     {
-      "epoch": 2.9042285714285714,
-      "grad_norm": 4.375,
-      "learning_rate": 1.5602625247541358e-07,
-      "loss": 0.4907,
+      "epoch": 1.4512,
+      "grad_norm": 4.46875,
+      "learning_rate": 3.0735189116682414e-05,
+      "loss": 0.5432,
       "step": 3175
     },
     {
-      "epoch": 2.927085714285714,
-      "grad_norm": 4.375,
-      "learning_rate": 9.061264096958955e-08,
-      "loss": 0.5023,
+      "epoch": 1.4626285714285714,
+      "grad_norm": 4.78125,
+      "learning_rate": 3.0410990348452573e-05,
+      "loss": 0.5056,
       "step": 3200
     },
     {
-      "epoch": 2.949942857142857,
-      "grad_norm": 4.59375,
-      "learning_rate": 4.2842286216368544e-08,
-      "loss": 0.4629,
+      "epoch": 1.4740571428571427,
+      "grad_norm": 5.21875,
+      "learning_rate": 3.0085834039132033e-05,
+      "loss": 0.4907,
       "step": 3225
     },
     {
-      "epoch": 2.9728,
-      "grad_norm": 3.8125,
-      "learning_rate": 1.2749023838509732e-08,
-      "loss": 0.4799,
+      "epoch": 1.4854857142857143,
+      "grad_norm": 3.828125,
+      "learning_rate": 2.975977772911671e-05,
+      "loss": 0.4958,
       "step": 3250
     },
     {
-      "epoch": 2.995657142857143,
-      "grad_norm": 4.53125,
-      "learning_rate": 3.5416881848104875e-10,
-      "loss": 0.513,
+      "epoch": 1.4969142857142856,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.9432879118068722e-05,
+      "loss": 0.545,
       "step": 3275
+    },
+    {
+      "epoch": 1.508342857142857,
+      "grad_norm": 4.03125,
+      "learning_rate": 2.9105196054705663e-05,
+      "loss": 0.4967,
+      "step": 3300
+    },
+    {
+      "epoch": 1.5197714285714286,
+      "grad_norm": 4.25,
+      "learning_rate": 2.8776786526563575e-05,
+      "loss": 0.5128,
+      "step": 3325
+    },
+    {
+      "epoch": 1.5312000000000001,
+      "grad_norm": 4.53125,
+      "learning_rate": 2.8447708649735356e-05,
+      "loss": 0.529,
+      "step": 3350
+    },
+    {
+      "epoch": 1.5426285714285715,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.8118020658586415e-05,
+      "loss": 0.5453,
+      "step": 3375
+    },
+    {
+      "epoch": 1.5540571428571428,
+      "grad_norm": 4.9375,
+      "learning_rate": 2.778778089544935e-05,
+      "loss": 0.5323,
+      "step": 3400
+    },
+    {
+      "epoch": 1.5654857142857144,
+      "grad_norm": 5.03125,
+      "learning_rate": 2.7457047800299585e-05,
+      "loss": 0.4782,
+      "step": 3425
+    },
+    {
+      "epoch": 1.5769142857142857,
+      "grad_norm": 5.90625,
+      "learning_rate": 2.71258799004137e-05,
+      "loss": 0.5319,
+      "step": 3450
+    },
+    {
+      "epoch": 1.588342857142857,
+      "grad_norm": 3.84375,
+      "learning_rate": 2.6794335800012293e-05,
+      "loss": 0.5254,
+      "step": 3475
+    },
+    {
+      "epoch": 1.5997714285714286,
+      "grad_norm": 5.65625,
+      "learning_rate": 2.646247416988926e-05,
+      "loss": 0.5265,
+      "step": 3500
+    },
+    {
+      "epoch": 1.6112,
+      "grad_norm": 4.5,
+      "learning_rate": 2.6130353737029284e-05,
+      "loss": 0.5077,
+      "step": 3525
+    },
+    {
+      "epoch": 1.6226285714285713,
+      "grad_norm": 6.65625,
+      "learning_rate": 2.579803327421536e-05,
+      "loss": 0.5087,
+      "step": 3550
+    },
+    {
+      "epoch": 1.6340571428571429,
+      "grad_norm": 5.0,
+      "learning_rate": 2.5465571589628223e-05,
+      "loss": 0.4842,
+      "step": 3575
+    },
+    {
+      "epoch": 1.6454857142857144,
+      "grad_norm": 5.78125,
+      "learning_rate": 2.5133027516439554e-05,
+      "loss": 0.5079,
+      "step": 3600
+    },
+    {
+      "epoch": 1.6569142857142856,
+      "grad_norm": 4.3125,
+      "learning_rate": 2.4800459902400684e-05,
+      "loss": 0.5259,
+      "step": 3625
+    },
+    {
+      "epoch": 1.6683428571428571,
+      "grad_norm": 5.59375,
+      "learning_rate": 2.446792759942882e-05,
+      "loss": 0.4971,
+      "step": 3650
+    },
+    {
+      "epoch": 1.6797714285714287,
+      "grad_norm": 5.5,
+      "learning_rate": 2.413548945319239e-05,
+      "loss": 0.4991,
+      "step": 3675
+    },
+    {
+      "epoch": 1.6912,
+      "grad_norm": 4.9375,
+      "learning_rate": 2.3803204292697704e-05,
+      "loss": 0.5282,
+      "step": 3700
+    },
+    {
+      "epoch": 1.7026285714285714,
+      "grad_norm": 5.78125,
+      "learning_rate": 2.3471130919878304e-05,
+      "loss": 0.5017,
+      "step": 3725
+    },
+    {
+      "epoch": 1.714057142857143,
+      "grad_norm": 4.96875,
+      "learning_rate": 2.3139328099189334e-05,
+      "loss": 0.5094,
+      "step": 3750
+    },
+    {
+      "epoch": 1.7254857142857143,
+      "grad_norm": 5.3125,
+      "learning_rate": 2.280785454720843e-05,
+      "loss": 0.4696,
+      "step": 3775
+    },
+    {
+      "epoch": 1.7369142857142856,
+      "grad_norm": 4.1875,
+      "learning_rate": 2.2476768922245072e-05,
+      "loss": 0.4994,
+      "step": 3800
+    },
+    {
+      "epoch": 1.7483428571428572,
+      "grad_norm": 4.0625,
+      "learning_rate": 2.21461298139603e-05,
+      "loss": 0.5082,
+      "step": 3825
+    },
+    {
+      "epoch": 1.7597714285714285,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.1815995732998584e-05,
+      "loss": 0.5247,
+      "step": 3850
+    },
+    {
+      "epoch": 1.7711999999999999,
+      "grad_norm": 3.671875,
+      "learning_rate": 2.1486425100633575e-05,
+      "loss": 0.4679,
+      "step": 3875
+    },
+    {
+      "epoch": 1.7826285714285715,
+      "grad_norm": 4.8125,
+      "learning_rate": 2.1157476238429803e-05,
+      "loss": 0.4905,
+      "step": 3900
+    },
+    {
+      "epoch": 1.794057142857143,
+      "grad_norm": 5.53125,
+      "learning_rate": 2.082920735792195e-05,
+      "loss": 0.4998,
+      "step": 3925
+    },
+    {
+      "epoch": 1.8054857142857141,
+      "grad_norm": 4.34375,
+      "learning_rate": 2.0501676550313594e-05,
+      "loss": 0.4987,
+      "step": 3950
+    },
+    {
+      "epoch": 1.8169142857142857,
+      "grad_norm": 4.96875,
+      "learning_rate": 2.017494177619725e-05,
+      "loss": 0.4986,
+      "step": 3975
+    },
+    {
+      "epoch": 1.8283428571428573,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.984906085529758e-05,
+      "loss": 0.4614,
+      "step": 4000
+    },
+    {
+      "epoch": 1.8397714285714286,
+      "grad_norm": 5.65625,
+      "learning_rate": 1.9524091456239417e-05,
+      "loss": 0.4805,
+      "step": 4025
+    },
+    {
+      "epoch": 1.8512,
+      "grad_norm": 4.46875,
+      "learning_rate": 1.9200091086342634e-05,
+      "loss": 0.5164,
+      "step": 4050
+    },
+    {
+      "epoch": 1.8626285714285715,
+      "grad_norm": 5.40625,
+      "learning_rate": 1.8877117081445524e-05,
+      "loss": 0.4836,
+      "step": 4075
+    },
+    {
+      "epoch": 1.8740571428571429,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.8555226595758485e-05,
+      "loss": 0.4719,
+      "step": 4100
+    },
+    {
+      "epoch": 1.8854857142857142,
+      "grad_norm": 4.90625,
+      "learning_rate": 1.8234476591749943e-05,
+      "loss": 0.4785,
+      "step": 4125
+    },
+    {
+      "epoch": 1.8969142857142858,
+      "grad_norm": 5.15625,
+      "learning_rate": 1.7914923830066074e-05,
+      "loss": 0.506,
+      "step": 4150
+    },
+    {
+      "epoch": 1.9083428571428571,
+      "grad_norm": 4.65625,
+      "learning_rate": 1.7596624859486347e-05,
+      "loss": 0.5037,
+      "step": 4175
+    },
+    {
+      "epoch": 1.9197714285714285,
+      "grad_norm": 5.03125,
+      "learning_rate": 1.7279636006916537e-05,
+      "loss": 0.5128,
+      "step": 4200
+    },
+    {
+      "epoch": 1.9312,
+      "grad_norm": 4.625,
+      "learning_rate": 1.6964013367420966e-05,
+      "loss": 0.4839,
+      "step": 4225
+    },
+    {
+      "epoch": 1.9426285714285716,
+      "grad_norm": 5.5625,
+      "learning_rate": 1.6649812794295782e-05,
+      "loss": 0.4991,
+      "step": 4250
+    },
+    {
+      "epoch": 1.9540571428571427,
+      "grad_norm": 4.125,
+      "learning_rate": 1.633708988918509e-05,
+      "loss": 0.468,
+      "step": 4275
+    },
+    {
+      "epoch": 1.9654857142857143,
+      "grad_norm": 4.53125,
+      "learning_rate": 1.602589999224149e-05,
+      "loss": 0.4796,
+      "step": 4300
+    },
+    {
+      "epoch": 1.9769142857142858,
+      "grad_norm": 4.6875,
+      "learning_rate": 1.5716298172332995e-05,
+      "loss": 0.4857,
+      "step": 4325
+    },
+    {
+      "epoch": 1.9883428571428572,
+      "grad_norm": 5.65625,
+      "learning_rate": 1.540833921729792e-05,
+      "loss": 0.4827,
+      "step": 4350
+    },
+    {
+      "epoch": 1.9997714285714285,
+      "grad_norm": 5.125,
+      "learning_rate": 1.5102077624249497e-05,
+      "loss": 0.4665,
+      "step": 4375
+    },
+    {
+      "epoch": 2.0109714285714286,
+      "grad_norm": 4.53125,
+      "learning_rate": 1.4797567589931924e-05,
+      "loss": 0.4299,
+      "step": 4400
+    },
+    {
+      "epoch": 2.0224,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.4494863001129638e-05,
+      "loss": 0.4671,
+      "step": 4425
+    },
+    {
+      "epoch": 2.0338285714285713,
+      "grad_norm": 4.90625,
+      "learning_rate": 1.4194017425131323e-05,
+      "loss": 0.387,
+      "step": 4450
+    },
+    {
+      "epoch": 2.045257142857143,
+      "grad_norm": 4.125,
+      "learning_rate": 1.3895084100250569e-05,
+      "loss": 0.3808,
+      "step": 4475
+    },
+    {
+      "epoch": 2.0566857142857145,
+      "grad_norm": 4.0625,
+      "learning_rate": 1.3598115926404683e-05,
+      "loss": 0.3955,
+      "step": 4500
+    },
+    {
+      "epoch": 2.0681142857142856,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.330316545575338e-05,
+      "loss": 0.3914,
+      "step": 4525
+    },
+    {
+      "epoch": 2.079542857142857,
+      "grad_norm": 4.53125,
+      "learning_rate": 1.3010284883399076e-05,
+      "loss": 0.3912,
+      "step": 4550
+    },
+    {
+      "epoch": 2.0909714285714287,
+      "grad_norm": 4.78125,
+      "learning_rate": 1.2719526038150308e-05,
+      "loss": 0.4035,
+      "step": 4575
+    },
+    {
+      "epoch": 2.1024,
+      "grad_norm": 4.34375,
+      "learning_rate": 1.2430940373349945e-05,
+      "loss": 0.4366,
+      "step": 4600
+    },
+    {
+      "epoch": 2.1138285714285714,
+      "grad_norm": 4.625,
+      "learning_rate": 1.2144578957769995e-05,
+      "loss": 0.4067,
+      "step": 4625
+    },
+    {
+      "epoch": 2.125257142857143,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.1860492466574222e-05,
+      "loss": 0.4404,
+      "step": 4650
+    },
+    {
+      "epoch": 2.136685714285714,
+      "grad_norm": 3.921875,
+      "learning_rate": 1.157873117235067e-05,
+      "loss": 0.4218,
+      "step": 4675
+    },
+    {
+      "epoch": 2.1481142857142856,
+      "grad_norm": 5.28125,
+      "learning_rate": 1.129934493621527e-05,
+      "loss": 0.4129,
+      "step": 4700
+    },
+    {
+      "epoch": 2.159542857142857,
+      "grad_norm": 3.984375,
+      "learning_rate": 1.1022383198988265e-05,
+      "loss": 0.4041,
+      "step": 4725
+    },
+    {
+      "epoch": 2.1709714285714288,
+      "grad_norm": 4.40625,
+      "learning_rate": 1.074789497244512e-05,
+      "loss": 0.4286,
+      "step": 4750
+    },
+    {
+      "epoch": 2.1824,
+      "grad_norm": 4.5625,
+      "learning_rate": 1.0475928830643245e-05,
+      "loss": 0.3916,
+      "step": 4775
+    },
+    {
+      "epoch": 2.1938285714285715,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.020653290132624e-05,
+      "loss": 0.4289,
+      "step": 4800
+    },
+    {
+      "epoch": 2.205257142857143,
+      "grad_norm": 3.78125,
+      "learning_rate": 9.939754857407063e-06,
+      "loss": 0.4146,
+      "step": 4825
+    },
+    {
+      "epoch": 2.216685714285714,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.675641908531774e-06,
+      "loss": 0.425,
+      "step": 4850
+    },
+    {
+      "epoch": 2.2281142857142857,
+      "grad_norm": 4.625,
+      "learning_rate": 9.414240792725179e-06,
+      "loss": 0.432,
+      "step": 4875
+    },
+    {
+      "epoch": 2.2395428571428573,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.155597768119978e-06,
+      "loss": 0.4114,
+      "step": 4900
+    },
+    {
+      "epoch": 2.2509714285714284,
+      "grad_norm": 4.28125,
+      "learning_rate": 8.899758604770809e-06,
+      "loss": 0.4102,
+      "step": 4925
+    },
+    {
+      "epoch": 2.2624,
+      "grad_norm": 5.40625,
+      "learning_rate": 8.646768576554665e-06,
+      "loss": 0.4174,
+      "step": 4950
+    },
+    {
+      "epoch": 2.2738285714285715,
+      "grad_norm": 3.984375,
+      "learning_rate": 8.396672453159163e-06,
+      "loss": 0.4372,
+      "step": 4975
+    },
+    {
+      "epoch": 2.2852571428571427,
+      "grad_norm": 5.75,
+      "learning_rate": 8.149514492159978e-06,
+      "loss": 0.3993,
+      "step": 5000
+    },
+    {
+      "epoch": 2.296685714285714,
+      "grad_norm": 5.15625,
+      "learning_rate": 7.90533843118889e-06,
+      "loss": 0.4007,
+      "step": 5025
+    },
+    {
+      "epoch": 2.308114285714286,
+      "grad_norm": 4.375,
+      "learning_rate": 7.66418748019396e-06,
+      "loss": 0.4236,
+      "step": 5050
+    },
+    {
+      "epoch": 2.3195428571428574,
+      "grad_norm": 4.5,
+      "learning_rate": 7.4261043137928896e-06,
+      "loss": 0.4146,
+      "step": 5075
+    },
+    {
+      "epoch": 2.3309714285714285,
+      "grad_norm": 5.375,
+      "learning_rate": 7.191131063721316e-06,
+      "loss": 0.4369,
+      "step": 5100
+    },
+    {
+      "epoch": 2.3424,
+      "grad_norm": 4.15625,
+      "learning_rate": 6.959309311377038e-06,
+      "loss": 0.4233,
+      "step": 5125
+    },
+    {
+      "epoch": 2.3538285714285716,
+      "grad_norm": 4.375,
+      "learning_rate": 6.7306800804616824e-06,
+      "loss": 0.4191,
+      "step": 5150
+    },
+    {
+      "epoch": 2.3652571428571427,
+      "grad_norm": 5.625,
+      "learning_rate": 6.505283829721068e-06,
+      "loss": 0.4135,
+      "step": 5175
+    },
+    {
+      "epoch": 2.3766857142857143,
+      "grad_norm": 4.0625,
+      "learning_rate": 6.283160445785532e-06,
+      "loss": 0.3916,
+      "step": 5200
+    },
+    {
+      "epoch": 2.388114285714286,
+      "grad_norm": 4.5625,
+      "learning_rate": 6.0643492361115026e-06,
+      "loss": 0.3923,
+      "step": 5225
+    },
+    {
+      "epoch": 2.399542857142857,
+      "grad_norm": 4.84375,
+      "learning_rate": 5.848888922025553e-06,
+      "loss": 0.3973,
+      "step": 5250
+    },
+    {
+      "epoch": 2.4109714285714285,
+      "grad_norm": 5.03125,
+      "learning_rate": 5.636817631872185e-06,
+      "loss": 0.4032,
+      "step": 5275
+    },
+    {
+      "epoch": 2.4224,
+      "grad_norm": 5.0625,
+      "learning_rate": 5.428172894266576e-06,
+      "loss": 0.4206,
+      "step": 5300
+    },
+    {
+      "epoch": 2.4338285714285712,
+      "grad_norm": 4.59375,
+      "learning_rate": 5.222991631453405e-06,
+      "loss": 0.436,
+      "step": 5325
+    },
+    {
+      "epoch": 2.445257142857143,
+      "grad_norm": 4.78125,
+      "learning_rate": 5.0213101527730345e-06,
+      "loss": 0.4113,
+      "step": 5350
+    },
+    {
+      "epoch": 2.4566857142857144,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.823164148236078e-06,
+      "loss": 0.3966,
+      "step": 5375
+    },
+    {
+      "epoch": 2.468114285714286,
+      "grad_norm": 3.78125,
+      "learning_rate": 4.628588682207677e-06,
+      "loss": 0.3977,
+      "step": 5400
+    },
+    {
+      "epoch": 2.479542857142857,
+      "grad_norm": 4.09375,
+      "learning_rate": 4.4376181872024e-06,
+      "loss": 0.4125,
+      "step": 5425
+    },
+    {
+      "epoch": 2.4909714285714286,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.250286457790961e-06,
+      "loss": 0.4254,
+      "step": 5450
+    },
+    {
+      "epoch": 2.5023999999999997,
+      "grad_norm": 4.34375,
+      "learning_rate": 4.066626644619934e-06,
+      "loss": 0.4207,
+      "step": 5475
+    },
+    {
+      "epoch": 2.5138285714285713,
+      "grad_norm": 4.75,
+      "learning_rate": 3.886671248545243e-06,
+      "loss": 0.4104,
+      "step": 5500
+    },
+    {
+      "epoch": 2.525257142857143,
+      "grad_norm": 5.0,
+      "learning_rate": 3.7104521148808054e-06,
+      "loss": 0.4128,
+      "step": 5525
+    },
+    {
+      "epoch": 2.5366857142857144,
+      "grad_norm": 4.21875,
+      "learning_rate": 3.5380004277630994e-06,
+      "loss": 0.417,
+      "step": 5550
+    },
+    {
+      "epoch": 2.5481142857142856,
+      "grad_norm": 3.421875,
+      "learning_rate": 3.3693467046327117e-06,
+      "loss": 0.4066,
+      "step": 5575
+    },
+    {
+      "epoch": 2.559542857142857,
+      "grad_norm": 4.65625,
+      "learning_rate": 3.2045207908339702e-06,
+      "loss": 0.4027,
+      "step": 5600
+    },
+    {
+      "epoch": 2.5709714285714287,
+      "grad_norm": 5.1875,
+      "learning_rate": 3.0435518543333765e-06,
+      "loss": 0.4211,
+      "step": 5625
+    },
+    {
+      "epoch": 2.5824,
+      "grad_norm": 3.1875,
+      "learning_rate": 2.8864683805580133e-06,
+      "loss": 0.425,
+      "step": 5650
+    },
+    {
+      "epoch": 2.5938285714285714,
+      "grad_norm": 4.28125,
+      "learning_rate": 2.733298167354703e-06,
+      "loss": 0.4216,
+      "step": 5675
+    },
+    {
+      "epoch": 2.605257142857143,
+      "grad_norm": 4.53125,
+      "learning_rate": 2.5840683200708067e-06,
+      "loss": 0.4378,
+      "step": 5700
+    },
+    {
+      "epoch": 2.6166857142857145,
+      "grad_norm": 4.3125,
+      "learning_rate": 2.4388052467576308e-06,
+      "loss": 0.4013,
+      "step": 5725
+    },
+    {
+      "epoch": 2.6281142857142856,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.297534653497191e-06,
+      "loss": 0.435,
+      "step": 5750
+    },
+    {
+      "epoch": 2.639542857142857,
+      "grad_norm": 4.3125,
+      "learning_rate": 2.160281539853201e-06,
+      "loss": 0.4214,
+      "step": 5775
+    },
+    {
+      "epoch": 2.6509714285714283,
+      "grad_norm": 4.75,
+      "learning_rate": 2.027070194447081e-06,
+      "loss": 0.3948,
+      "step": 5800
+    },
+    {
+      "epoch": 2.6624,
+      "grad_norm": 4.03125,
+      "learning_rate": 1.897924190659822e-06,
+      "loss": 0.3921,
+      "step": 5825
+    },
+    {
+      "epoch": 2.6738285714285714,
+      "grad_norm": 4.0625,
+      "learning_rate": 1.7728663824603586e-06,
+      "loss": 0.3567,
+      "step": 5850
+    },
+    {
+      "epoch": 2.685257142857143,
+      "grad_norm": 4.03125,
+      "learning_rate": 1.6519189003612767e-06,
+      "loss": 0.3819,
+      "step": 5875
+    },
+    {
+      "epoch": 2.696685714285714,
+      "grad_norm": 4.59375,
+      "learning_rate": 1.535103147502584e-06,
+      "loss": 0.3772,
+      "step": 5900
+    },
+    {
+      "epoch": 2.7081142857142857,
+      "grad_norm": 4.28125,
+      "learning_rate": 1.422439795864114e-06,
+      "loss": 0.4203,
+      "step": 5925
+    },
+    {
+      "epoch": 2.7195428571428573,
+      "grad_norm": 4.125,
+      "learning_rate": 1.3139487826073937e-06,
+      "loss": 0.3637,
+      "step": 5950
+    },
+    {
+      "epoch": 2.7309714285714284,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.2096493065475223e-06,
+      "loss": 0.4062,
+      "step": 5975
+    },
+    {
+      "epoch": 2.7424,
+      "grad_norm": 6.4375,
+      "learning_rate": 1.1095598247556793e-06,
+      "loss": 0.3903,
+      "step": 6000
+    },
+    {
+      "epoch": 2.7538285714285715,
+      "grad_norm": 3.890625,
+      "learning_rate": 1.0136980492929605e-06,
+      "loss": 0.388,
+      "step": 6025
+    },
+    {
+      "epoch": 2.765257142857143,
+      "grad_norm": 4.09375,
+      "learning_rate": 9.220809440759592e-07,
+      "loss": 0.3986,
+      "step": 6050
+    },
+    {
+      "epoch": 2.776685714285714,
+      "grad_norm": 3.46875,
+      "learning_rate": 8.347247218748411e-07,
+      "loss": 0.393,
+      "step": 6075
+    },
+    {
+      "epoch": 2.7881142857142858,
+      "grad_norm": 4.90625,
+      "learning_rate": 7.516448414442739e-07,
+      "loss": 0.3973,
+      "step": 6100
+    },
+    {
+      "epoch": 2.799542857142857,
+      "grad_norm": 4.5,
+      "learning_rate": 6.728560047878058e-07,
+      "loss": 0.3999,
+      "step": 6125
+    },
+    {
+      "epoch": 2.8109714285714285,
+      "grad_norm": 3.921875,
+      "learning_rate": 5.98372154556176e-07,
+      "loss": 0.3963,
+      "step": 6150
+    },
+    {
+      "epoch": 2.8224,
+      "grad_norm": 5.78125,
+      "learning_rate": 5.282064715799895e-07,
+      "loss": 0.4023,
+      "step": 6175
+    },
+    {
+      "epoch": 2.8338285714285716,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.6237137253721527e-07,
+      "loss": 0.381,
+      "step": 6200
+    },
+    {
+      "epoch": 2.8452571428571427,
+      "grad_norm": 4.0625,
+      "learning_rate": 4.008785077558752e-07,
+      "loss": 0.4355,
+      "step": 6225
+    },
+    {
+      "epoch": 2.8566857142857143,
+      "grad_norm": 4.59375,
+      "learning_rate": 3.4373875915241493e-07,
+      "loss": 0.4114,
+      "step": 6250
+    },
+    {
+      "epoch": 2.868114285714286,
+      "grad_norm": 5.96875,
+      "learning_rate": 2.909622383059835e-07,
+      "loss": 0.397,
+      "step": 6275
+    },
+    {
+      "epoch": 2.879542857142857,
+      "grad_norm": 5.34375,
+      "learning_rate": 2.4255828466907825e-07,
+      "loss": 0.3944,
+      "step": 6300
+    },
+    {
+      "epoch": 2.8909714285714285,
+      "grad_norm": 3.640625,
+      "learning_rate": 1.985354639148229e-07,
+      "loss": 0.373,
+      "step": 6325
+    },
+    {
+      "epoch": 2.9024,
+      "grad_norm": 4.5,
+      "learning_rate": 1.5890156642114075e-07,
+      "loss": 0.425,
+      "step": 6350
+    },
+    {
+      "epoch": 2.9138285714285717,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.2366360589217974e-07,
+      "loss": 0.4118,
+      "step": 6375
+    },
+    {
+      "epoch": 2.925257142857143,
+      "grad_norm": 5.0625,
+      "learning_rate": 9.282781811714159e-08,
+      "loss": 0.3736,
+      "step": 6400
+    },
+    {
+      "epoch": 2.9366857142857143,
+      "grad_norm": 4.71875,
+      "learning_rate": 6.63996598667671e-08,
+      "loss": 0.4139,
+      "step": 6425
+    },
+    {
+      "epoch": 2.9481142857142855,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.438380792772534e-08,
+      "loss": 0.4324,
+      "step": 6450
+    },
+    {
+      "epoch": 2.959542857142857,
+      "grad_norm": 4.71875,
+      "learning_rate": 2.6784158274964498e-08,
+      "loss": 0.4182,
+      "step": 6475
+    },
+    {
+      "epoch": 2.9709714285714286,
+      "grad_norm": 4.53125,
+      "learning_rate": 1.3603825382293988e-08,
+      "loss": 0.4026,
+      "step": 6500
+    },
+    {
+      "epoch": 2.9824,
+      "grad_norm": 4.0,
+      "learning_rate": 4.845141671239173e-09,
+      "loss": 0.3775,
+      "step": 6525
+    },
+    {
+      "epoch": 2.9938285714285713,
+      "grad_norm": 4.25,
+      "learning_rate": 5.096570982743298e-10,
+      "loss": 0.4082,
+      "step": 6550
     }
   ],
   "logging_steps": 25,
-  "max_steps": 3279,
+  "max_steps": 6561,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 3,
   "save_steps": 500,
@@ -944,7 +1861,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 3.782596152490721e+17,
+  "total_flos": 1.6747303870715658e+18,
   "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null